Cognitive Mechanisms Underlying the
Origin and Evolution of Culture
DOCTORAL THESIS
Liane Gabora
Center Leo Apostel for Interdisciplinary Studies
Vrije Universiteit Brussel
Krijgskundestraat 33,
B1160 Brussels, Belgium
http://www.vub.ac.be/CLEA/liane/
[email protected]
Promotor/Doctoral Advisor: Prof. Dr. Diederik Aerts
Faculty of Science
Free University of Brussels, 2001
THE
Liane Gabora
TABLE OF CONTENTS
Table of Contents............................................................................................................................................iii
Detailed Table of Contents ............................................................................................................................. iv
Dankwoord....................................................................................................................................................... x
Preface.............................................................................................................................................................xi
Bijstelling......................................................................................................................................................xiii
1 The Need for a Theory of Cultural Evolution .............................................................................................. 1
2 Current Evolutionary Approaches to Culture............................................................................................. 13
3 Unresolved Issues and Potentially Fruitful Directions............................................................................... 29
4 A Computer Model of Cultural Evolution ................................................................................................. 47
5 Mind: The Culture Evolving Architecture ................................................................................................. 63
6 Creativity and Cultural Novelty ................................................................................................................. 83
7 What Sparked the Origin of Culture?......................................................................................................... 97
8 Autocatalytic Closure in a Cognitive System .......................................................................................... 109
9 Embryology of One and Many Worldviews ............................................................................................ 135
10 What is Missing in Current Evolutionary Theory ................................................................................ 145
11 Potentiality, Context, and Change of State........................................................................................... 153
12 Toward a General Theory of Evolution................................................................................................ 181
13 Contextualizing Theories of Concepts and Culture.............................................................................. 191
14 Summary and Synthesis of Main Points................................................................................................211
References.....................................................................................................................................................233
ii
TOC, Dankwoord, Preface, Bijstelling
DETAILED TABLE OF CONTENTS
1 The Need for a Theory of Cultural Evolution ......................................................................................... 1
1.1 Terminology ........................................................................................................................................ 2
1.2 Aim and Objectives ............................................................................................................................. 4
1.3 Culture is Not just a Natural Extension of Biological Evolution ........................................................ 5
1.4 Analogy or Another Form of Evolution? ............................................................................................ 7
1.5 Chapter by Chapter Outline................................................................................................................. 8
2 Current Evolutionary Approaches to Culture ...................................................................................... 13
2.1 Darwinian Approaches to Culture..................................................................................................... 13
2.1.1 Diffusion and Coevolutionary Studies......................................................................................... 14
2.1.1.1 An Example: Individuation in Families and Social Groups ................................................. 15
2.2 Neo-Darwinian Aproaches to Culture............................................................................................... 15
2.2.1 Approaches Inspired by Population Genetics .............................................................................. 16
2.2.2 Epidemiological Models of Culture............................................................................................. 17
2.2.2.1 A First Example: Conceptual Linkage Disequilibrium ........................................................ 17
2.2.2.2 A Second Example: A Cultural Analog to Genetic Hitchhiking.......................................... 18
2.2.2.3 A Third Example: A Possible Explanation for Cognitive Redundancy ............................... 20
2.2.3 Universal Darwinism, Second Replicators, and Memes.............................................................. 20
2.2.3.1 A First Example: A Potential Explanation for Human Altruism.......................................... 21
2.2.3.2 A Second Example: Runaway Cultural Selection ................................................................ 22
2.2.4 Interactors and Lineages .............................................................................................................. 23
2.3 Complexity Theory and the Genetic Algorithm................................................................................ 23
2.3.1 A First Example: Emergence of an Interconnected, Culture-evolving Worldview..................... 25
2.3.2 A Second Example: A Cultural Analog to the Pre-Cambrian Explosion .................................... 25
2.4 Selectionism and Evolutionary Epistomology .................................................................................. 25
2.4.1 An Example: Vicarious Selection................................................................................................ 27
2.5 Summary ........................................................................................................................................... 27
3 Unresolved Issues and Potentially Fruitful Directions ......................................................................... 29
3.1 Evolution without Replicators........................................................................................................... 29
3.1.1 Self Description versus Instructions for Self Replication............................................................ 29
3.1.2 Representation versus Expression of Cultural Entities................................................................ 30
3.1.3 Non-replicator Transmission Results in Inheritance of Acquired Characteristics....................... 30
3.1.4 Replacing Replication with Retention ......................................................................................... 31
3.1.5 Substrate Neutrality ..................................................................................................................... 32
3.2 Culture is not Limited to Imitated Entities........................................................................................ 32
3.2.1 Consequences of Excluding Individual Learning ........................................................................ 32
3.2.2 Consequences of Excluding Non-imitative Social Exchange...................................................... 34
3.2.3 Humans are not Passive ‘Meme Hosts’ ....................................................................................... 34
3.3 The Need to Merge Transmission Studies with Studies of Creativity .............................................. 35
3.4 Blending versus Particulate Inheritance ............................................................................................ 36
3.5 The Generation of Cultural Novelty.................................................................................................. 37
3.5.1 Can Variation in Evolution be Nonrandom? ............................................................................... 37
3.5.2 Parallel versus Heuristic Search................................................................................................... 37
3.5.3 What Enables Novelty to be Strategically Generated? ................................................................ 38
3.5.4 Does Creativity Yield to Mathematical Description?.................................................................. 38
iii
Liane Gabora
3.6 Is Thought a Darwinian Process?...................................................................................................... 40
3.6.1 Selection Theory works with Distinct, Actualized States............................................................ 43
3.6.2 Selection Theory Cannot Describe Intrinsic Contextuality ......................................................... 44
3.7 Summary ........................................................................................................................................... 45
4 A Computer Model of Cultural Evolution............................................................................................. 47
4.1 The Model ......................................................................................................................................... 48
4.1.1 The Domain ................................................................................................................................. 48
4.1.2 The Neural Network .................................................................................................................... 48
4.1.3 The Embodiment.......................................................................................................................... 50
4.1.4 The Fitness Function.................................................................................................................... 50
4.1.5 Using Experience to Bias the Generation of Novelty.................................................................. 51
4.2 Protocol ............................................................................................................................................. 52
4.3 Results ............................................................................................................................................... 53
4.3.1 Outline of a Run: Culture Evolves............................................................................................... 53
4.3.2 Trade-off Between Diversity and Global Optimization............................................................... 54
4.3.3 Frequency of Change Must be Intermediate................................................................................ 55
4.3.4 Epistasis Decreases Rate of Fitness Increase............................................................................... 56
4.3.5 Cultural Drift................................................................................................................................ 57
4.3.6 Effect of Knowledge-based Operators, Imitation, and Mental Simulation ................................. 58
4.3.7 Fittest Society with Creation to Imitation Ratio of 2:1................................................................ 59
4.4 Comparison with Other Approaches ................................................................................................. 60
4.5 Summary ........................................................................................................................................... 61
5 Mind: The Culture Evolving Architecture ............................................................................................ 63
5.1 Elucidating What Evolves in Mind and Culture................................................................................ 63
5.1.1 Episodic, Semantic, and Procedural Memories ........................................................................... 63
5.1.2 Classical, Prototype, and Exemplar Theories of Concepts .......................................................... 64
5.2 The Structure of Mental Entities ....................................................................................................... 64
5.2.1 Properties, Features, and Dimensions.......................................................................................... 65
5.2.2 Chunking and Categorization ...................................................................................................... 66
5.2.3 Hierarchical and Dynamical Structure......................................................................................... 66
5.3 Conceptual Space .............................................................................................................................. 67
5.3.1 Sparse........................................................................................................................................... 67
5.3.2 Constrained Distribution of Memories and the Edge of Chaos ................................................... 69
5.3.2.1 Distributed Storage ............................................................................................................... 69
5.3.2.2 Constraining the Distribution ............................................................................................... 70
5.3.2.3 The Edge of Chaos................................................................................................................ 73
5.4 The Architecture that Realizes Conceptual Space ............................................................................ 74
5.4.1 Integrating Abstractions, Memories, Stimuli, and Drives ........................................................... 74
5.4.2 Content Addressability................................................................................................................. 75
5.4.3 Organized Modularity.................................................................................................................. 76
5.4.4 Habituation................................................................................................................................... 76
5.5 Generating a Stream of Thought and Culturally Expressing it ......................................................... 77
5.5.1 An Instant of Experience ............................................................................................................. 77
5.5.2 The Continuity of a Stream of Experience .................................................................................. 77
5.5.3 Retrieval as Reconstruction ......................................................................................................... 78
5.5.4 A Stream of Thought as Representational Redescription ............................................................ 79
5.5.5 Symbol Manipulation................................................................................................................... 79
5.5.6 Embodiment and Cultural Expression ......................................................................................... 80
5.6 Summary ........................................................................................................................................... 80
6 Creativity and Cultural Novelty ............................................................................................................. 83
6.1 Attributes of the Creative Mind......................................................................................................... 83
6.1.1 Defocused Attention, Sensitivity, and Flat Association Hierarchies........................................... 83
6.1.2 Relationship of Creativity Attributes to Activation Threshold.................................................... 84
iv
TOC, Dankwoord, Preface, Bijstelling
6.1.3 Conceptual Fluidity and Alertness versus Depth of Processing .................................................. 85
6.1.4 Variable Fluidity as the Crux of Creative Potential..................................................................... 86
6.1.5 Effect of Density of Abstractions on Creative Potential.............................................................. 87
6.2 The Birth of a Creative Idea .............................................................................................................. 87
6.2.1 Relevant Possibilities Become Activated .................................................................................... 87
6.2.2 Analogy........................................................................................................................................ 89
6.2.3 Retrieving and Reconstructing..................................................................................................... 89
6.3 Refinement and Expression of an Idea.............................................................................................. 90
6.3.1 Merging Simultaneously Evoked Concepts................................................................................. 90
6.3.2 Honing in on an Idea.................................................................................................................... 91
6.4 Selection and the Cultural Fitness Landscape................................................................................... 91
6.4.1 Mental Simulation........................................................................................................................ 92
6.4.2 Interpersonal Prompting............................................................................................................... 92
6.4.3 Brains Select Ideas that Satisfy Needs......................................................................................... 92
6.4.4 Cultural Fitness Landscape is Sculpted by Need for Worldview Cohesion ................................ 93
6.4.5 How Variation Feeds Back on the Fitness Landscape................................................................. 94
6.5 We Still Cannot Mathematically Describe Impossibilist Creativity ................................................. 94
6.6 Summary ........................................................................................................................................... 95
7 What Sparked the Origin of Culture?.................................................................................................... 97
7.1 Was it Imitation or Creativity?.......................................................................................................... 97
7.2 What Differentiates Humans from Animals?.................................................................................... 98
7.3 Which Hypothesis does the Archeological Evidence Support? ...................................................... 100
7.3.1 An Archeological Transition...................................................................................................... 100
7.3.2 A Second Archeological Transition........................................................................................... 101
7.4 Two Paradoxes: The Origin of Life and the Origin of Culture ....................................................... 102
7.5 A Psychological Perspective on the Origin of Culture.................................................................... 104
7.5.1 Underlying an Archeological Transition is a Cognitive Transition........................................... 104
7.5.2 The Origin of Culture Paradox: A Concrete Example............................................................... 105
7.6 Summary ......................................................................................................................................... 106
8 Autocatalytic Closure in a Cognitive System....................................................................................... 109
8.1 A Return to the Origin of Life (OOL) Paradox............................................................................... 109
8.1.1 A First OOL Hypothesis: Prebiotic Soup .................................................................................. 110
8.1.2 A Second OOL Hypothesis: Ribozymes.................................................................................... 110
8.1.3 Third OOL Hypothesis: Autocatalytic Closure of a Set of Catalytic Molecules....................... 111
8.1.3.1 Weaving Catalytic Molecules into an Interconnected Web ............................................... 112
8.1.3.2 Replication of the Primitive Organic Closure System........................................................ 114
8.1.3.3 Theoretical and Experimental Support for Autocatalytic Origin of Life Theory............... 115
8.1.3.4 Mutual Decoding of Parts of a ‘Collective Self’ ................................................................ 116
8.2 Abstract Structure of the Two Problems of Origins........................................................................ 117
8.3 Self-organizing Memories into an Autocatalytically Unified Whole ............................................. 118
8.3.1 What is the Associative Mind Lacking? .................................................................................... 118
8.3.2 What if Oga’s Activation Threshold is Lower?......................................................................... 120
8.3.3 Reminding Events...................................................................................................................... 120
8.3.4 An Abstraction Emerges ............................................................................................................ 121
8.3.5 Establishing a Stream of Thought.............................................................................................. 123
8.3.6 Conceptual Closure in Oga’s Mind............................................................................................ 123
8.3.7 Hierarchical Levels of Conceptual Closure ............................................................................... 124
8.4 Social Interaction and Replication of the Conceptual Closure System........................................... 126
8.5 Under What Conditions does Conceptual Closure Occur? ............................................................. 127
8.6 Implications and Evolutionary Considerations ............................................................................... 130
8.6.1 Slowly Obviating the Need for Hardwiring............................................................................... 130
8.6.2 The Proposed Transition Might Not Leave Archeological Footprints ...................................... 130
8.6.3 Worldviews are Primitive Replicators....................................................................................... 130
8.6.4 Autocatalysis as an Explanation of Evolutionary Origins ......................................................... 131
v
Liane Gabora
8.7
Summary ......................................................................................................................................... 131
9 Embryology of One and Many Worldviews......................................................................................... 135
9.1 Creativity Initially Hinders Conceptual Closure ............................................................................. 135
9.2 Weaving and Reweaving a Worldview ........................................................................................... 136
9.2.1 Pros and Cons of Thorough Processing and Variable Fluidity.................................................. 136
9.2.2 Subcritical versus Supracritical Conceptual Closure................................................................. 137
9.2.3 Annealing on a New Worldview................................................................................................ 138
9.2.4 Self-organized Criticality........................................................................................................... 139
9.3 Fragmenting the Worldview............................................................................................................ 139
9.3.1 Censorship and Repression ........................................................................................................ 139
9.3.2 How Deception Invites Worldview Distortion .......................................................................... 141
9.4 Co-evolution of Mental Entities and Multiple Worldviews............................................................ 142
9.5 One and Many Worldviews............................................................................................................. 142
9.6 A Possible Explanation for Consciousness ..................................................................................... 143
9.7 Summary ......................................................................................................................................... 144
10 What is Missing in Current Evolutionary Theory........................................................................... 145
10.1 Review of Neo-Darwinian Theory of Evolution............................................................................. 145
10.2 The Need to Incorporate Potentiality and Actualization ................................................................. 146
10.2.1Evolution as a Means of Preserving and Enhancing Potentiality .............................................. 148
10.3 The Need to Incorporate Context and Contextuality....................................................................... 149
10.3.1Deterministic versus Nondeterministic Contextuality ............................................................... 150
10.4 The Problem of Incomplete Knowledge ......................................................................................... 151
10.5 Summary ......................................................................................................................................... 152
11 Potentiality, Context, and Change of State ...................................................................................... 153
11.1 General Scheme for Change of an Entity under the Influence of a Context ................................... 153
11.1.1Representation of the Trajectory of an Entity............................................................................ 153
11.1.2The Concept of a Potentiality State ........................................................................................... 155
11.2 How Standard Quantum Mechanics Describes Change.................................................................. 156
11.2.1Dynamical Evolution ................................................................................................................. 156
11.2.2Change of State Due to a Measurement..................................................................................... 156
11.2.2.1 Eigenstates and Superposition States.................................................................................. 157
11.2.2.2 Quantum Collapse .............................................................................................................. 157
11.2.2.3 The Role of Lack of Knowledge ........................................................................................ 158
11.2.3The Two Evolutions of Quantum Mechanics United Under the Context View ........................ 158
11.2.4Entanglement and the Generation of New States with New Properties..................................... 159
11.3 Classical Mechanics Cannot Describe Indeterminism Due To Context ......................................... 160
11.3.1Measurements in Classical Physics ........................................................................................... 161
11.3.2The Unstable Equilibrium as a First Order Quantum Situation................................................. 161
11.3.3Chaos and Complexity Theories Cannot Describe Indeterminism Due to Context .................. 162
11.4 Why We Need the Generalized Quantum Formalisms ................................................................... 162
11.4.1Overcoming the Limitation of the Linear State Space............................................................... 163
11.4.2Describing Situations Between Quantum and Classical............................................................ 163
11.4.3Classical and Pure Quantum as Special Cases in the Generalized Formalisms ........................ 164
11.5 Evolution of Potentiality State under the Influence of Context ...................................................... 165
11.5.1The Water Example ................................................................................................................... 165
11.5.2The Epsilon Model Example ..................................................................................................... 166
11.6 Identifying Quantum Structure through Violation of Bell Inequalities .......................................... 170
11.6.1The Original Violation of Bell Inequalities in the Microworld ................................................. 172
11.6.2The Violation of Bell Inequalities by the Potentiality State of Water ....................................... 174
11.7 Two Types of Indeterminism .......................................................................................................... 175
11.7.1Pure States versus Mixed States in Stochastic Processes .......................................................... 175
11.7.2Can Stochastic Processes Describe Context-driven Indeterminism?......................................... 176
11.7.3The Water Example in a Classical Statistical Description......................................................... 177
vi
TOC, Dankwoord, Preface, Bijstelling
11.8 Summary ......................................................................................................................................... 179
12 Toward a General Theory of Evolution............................................................................................ 181
12.1 On the Re-application of Concepts from Physics ........................................................................... 181
12.1.1Collapse...................................................................................................................................... 181
12.1.2Superposition ............................................................................................................................. 182
12.2 Generalizing the Process of Evolution ............................................................................................ 182
12.2.1The Basic Idea............................................................................................................................ 182
12.2.2Quantum Structure in Evolutionary Systems............................................................................. 183
12.3 Distinguishing Different Evolutionary Systems.............................................................................. 185
12.3.1Degree to which Potential is Retained ....................................................................................... 185
12.3.2Degree of Determinism.............................................................................................................. 186
12.3.3Degree of Sensitivity to Context................................................................................................ 186
12.3.4Context Dependence and Context Independence ...................................................................... 186
12.4 Different Ways Reality has Found of Actualizing Potential........................................................... 187
12.4.1Nondeterministic Collapse of a Quantum Entity as a Kind of CAP.......................................... 187
12.4.2Deterministic Evolution of Quantum or Classical Entities as a Second Kind of CAP.............. 187
12.4.3Biological Evolution as a Third Kind of CAP ........................................................................... 188
12.4.4Stream of Thought as a Fourth Kind of CAP ............................................................................ 189
12.4.5Cultural Evolution as a Fifth Kind of CAP................................................................................ 189
12.5 Summary ......................................................................................................................................... 190
13 Contextualizing Theories of Concepts and Culture ........................................................................ 191
13.1 Quantum Structure in the Mind....................................................................................................... 191
13.1.1The Violation of Bell Inequalities in Cognition ........................................................................ 191
13.1.2Implications................................................................................................................................ 194
13.2 Describing States of Mind that are Unfocused or Undecided ......................................................... 195
13.3 How the Pet Fish Problem is Solved by Contextualizing Concepts ............................................... 196
13.4 Using the CAP Approach to Model Creativity ............................................................................... 198
13.4.1An Example: Oga Invents the Torch ......................................................................................... 199
13.4.2Why Oga’s Stream of Thought Could Not be Described Classically........................................ 203
13.4.3Why a Pure Quantum Formalism Would Not Work Either....................................................... 204
13.4.4Why Generalized Quantum Formalisms can Describe Impossibilist Creativity ....................... 204
13.5 Contextualizing a Theory of Cultural Interaction ........................................................................... 205
13.5.1The Final Appearance of Og and Oga ....................................................................................... 206
13.5.2Another Example: Mike and Abby and the Four Mad Scientists .............................................. 207
13.6 Summary ......................................................................................................................................... 209
14 Summary and Synthesis of Main Points........................................................................................... 211
14.1 Comparison of Cultural and Biological Evolution.......................................................................... 211
14.1.1Relation Between Biological and Cultural Evolution................................................................ 212
14.1.1.1 Culture is not Simply an Extension of Biology .................................................................. 212
14.1.1.2 Biological Needs Constrain but don’t Control the Evolution of Culture ........................... 213
14.1.2Similarities Between Biological and Cultural Evolution........................................................... 213
14.1.2.1 Particulate versus Blending Inheritance ............................................................................. 213
14.1.2.2 Conceptual Linkage Disequilibrium................................................................................... 214
14.1.2.3 A Cultural Analog of Genetic Hitchhiking......................................................................... 214
14.1.2.4 Individuation in Families and Social Groups ..................................................................... 215
14.1.2.5 Evolution Does Take Place in a Model of Culture ............................................................. 215
14.1.2.6 Cultural Drift ...................................................................................................................... 216
14.1.2.7 Frequency of Change Must be Intermediate....................................................................... 216
14.1.2.8 Epistasis Increases Time to Fixation .................................................................................. 216
14.1.3Differences Between Biological and Cultural Evolution........................................................... 216
14.1.3.1 Cultural Entities are not Replicators................................................................................... 217
14.1.3.2 Cultural Novelty is Generated not Randomly but Strategically and Contextually............. 217
14.1.3.3 Mental Simulation .............................................................................................................. 218
vii
Liane Gabora
14.1.3.4 Culture Works with not just Actual but Potential Entities ................................................. 218
14.1.3.5 Tradeoff Between Innovation and Imitation....................................................................... 219
14.1.4Conclusion Regarding Similarities and Differences.................................................................. 219
14.2 How Did Culture Begin to Evolve?................................................................................................. 220
14.2.1Arguments that the Bottleneck was Creativity rather than Imitation ........................................ 220
14.2.2Creativity Arises through Variability of the Activation Threshold ........................................... 220
14.2.3Creativity Requires a Relationally Structured Worldview ........................................................ 221
14.2.4The Existence of a Worldview Generates a Chicken and Egg Paradox .................................... 222
14.2.5Origin of Culture through Conceptual Closure to Yield a Worldview...................................... 222
14.2.6Conceptual Closure Recurs in the Mind of Every Encultured Child......................................... 223
14.2.7Worldviews are Primitive Replicators ....................................................................................... 223
14.2.8Conceptual Networks Contain Quantum-like Structure ............................................................ 224
14.3 Toward a General, Trans-disciplinary Theory of Evolution ........................................................... 224
14.3.1The Inadequacy of Neo-Darwinian Theory ............................................................................... 224
14.3.2Context, Potential, Actualization, and Lack of Knowledge....................................................... 225
14.3.3Describing Change of State when One Lacks Knowledge about the State ............................... 226
14.3.4Describing Contextuality (Lack of Knowledge about State-Context Interaction)..................... 226
14.3.5Intermediate Contextuality and the Generalized Quantum Formalisms.................................... 227
14.3.6Evolution as Context-driven Actualization of Potential (CAP)................................................. 227
14.3.6.1 Nondeterministic Collapse of a Quantum Entity................................................................ 228
14.3.6.2 Deterministic Evolution of Quantum or Classical Entity ................................................... 228
14.3.6.3 Biological Evolution........................................................................................................... 228
14.3.6.4 A Stream of Thought .......................................................................................................... 229
14.3.6.5 Cultural Evolution .............................................................................................................. 230
14.4 Conclusion....................................................................................................................................... 230
References ................................................................................. ..............................................................233
viii
TOC, Dankwoord, Preface, Bijstelling
DANKWOORD1
The first person to read this thesis was my mother, Beverlee Moore, and I am grateful to her for her helpful
comments and for believing in me.
Two books that had a big impact on how my thoughts took shape are ‘Origins of the Modern Mind’
by Merlin Donald and ‘Origins of Order’ by Stuart Kauffman, both of whom took the time to talk to and
encourage me on several occasions. I also thank Merlin Donald, as well as Nicole Degrande, Francis
Heylighen, Mark Lake, Luc Steels, Philip Polk, and Jean Paul Van Bendegem for participating on my
thesis committee. I am grateful to Philip Polk and Francis Heylighen for useful discussion. I may not see
eye to eye with Francis on every matter, but his insights and knowledge of the relevant literature definitely
made this thesis better. And in fact, were it not for him it would not exist, since he was the catalyst that led
me to the Center Leo Apostel for interdisciplinary studies (CLEA). Thanks also to Alex Riegler for two of
the figures and comments on chapter eight, and in fact to all the kind and interesting people at CLEA,
where the thesis was written.
Many others provided inspiration and encouragement, including David Chalmers, Nita Chaudhuri,
Alan Combs, Bethany Isenberg, Michelle Hart, Pia Henriques, Dale Kreutzer, Stan Lathers, Charles
Lumsden Bill Macready, Franco Orsucci, Liz Shapiro, Anna Maria Van Hoye, Inkling and Glimmer, and
my sister, brother, and father: Natalie, Michael, and Herbert Gabora. You are the people of my little world
who inspired me to think about how ideas are born and how they evolve when people interact. The Global
Underground series (especially Sasha and BT) provided musical inspiration during the final writing stage.
But most of all I thank my brilliant and imaginative promoter, Diederik Aerts. In what has been the
most challenging yet fascinating scientific collaboration I could imagine, he devoted countless hours to
unraveling the mysteries of quantum mechanics and other subjects for me. He recently said that he, and
now I, am part of a ‘cultural lineage’ that goes back three generations, apprentice to teacher, to Niels Bohr,
Louis De Broglie, Wolfgang Pauli, and Henri Poincaré 2. Certainly from the instant I entered his sphere, I
felt that I had stepped into a realm where it the most natural and exhilarating thing to be in a state of
devotion to the fruits of the mind, to give oneself completely to the task of seeing deeper into reality.
It is the custom at the VUB that a thesis begins with a page titled Dankwoord, which is Flemish for ‘word of thanks’.
The details of this cultural lineage: Diederik Aerts had two thesis advisors. The first, Jean Reignier was a student of
Jules Geheniau and Leon Rosenfeld. Jules Geheniau was a student of Louis de Broglie while Leon Rosenfeld was a
student of Niels Bohr, and also worked with and was very close to Wolfgang Pauli. The second, Constantin Piron, was
a student of Jozef Jauch, and was as a child instead of school educated by his mother, who was a student of Theophile
De Donder. Jozef Jauch was a student of Niels Bohr, while Theophile De Donder was a student of Henri Poincaré.
(Another interesting cultural lineage is that Leo Apostel, who was the benefactor of the center where I wrote this thesis,
was a student of Jean Piaget.)
1
2
ix
Liane Gabora
BIJSTELLING3
The cognitive scientist’s analysis of how abstract concepts are evoked to give meaning to
situations can also be used to show how fiction archetypes are evoked to create powerful stories.
3
At the VUB, it is the custom to have a short second thesis, referred to as the ‘bijstelling’, which is presented orally
during the thesis defense.
x
TOC, Dankwoord, Preface, Bijstelling
Nothingthe is written here
xi
Chapter One
1
The Need for a Theory of Cultural Evolution
This thesis aims to clarify how the invention and development of cultural entities—such as ideas,
artifacts, mannerisms, and attitudes—constitutes a form of evolution distinct from, yet
intertwined with, biological evolution, and to propose how this evolutionary process may have
begun. This is clearly an ambitious goal, so at best this thesis will comprise a few small steps
toward its realization.
The application of evolution to culture is far from new. Clearly things like tools and
languages do evolve in the general sense of incremental change that reflects the constraints and
affordances of an environment. Ideally we will be able to flesh out a theoretical framework for
this process that unifies the psychological and social sciences as the theory of Darwinian
evolution did for the biological sciences; although much was known about plants and animals
before Darwin, his theory of how life evolves through natural selection united previously
disparate phenomena and paved the way for further inquiry. However, so far the endeavor to
frame culture in terms of evolution has not been very fruitful.
Such efforts have always engendered a heated mixture of enthusiasm and hostility, perhaps
because of their implications for issues such as individuality and free will. But what may be the
greater impediment is a tendency to slide into either one of two extreme positions. Supporters of
the notion that culture evolves tend to make strong claims to the effect that natural selection is a
‘universal acid’, and to ignore or downplay the differences between biological and cultural
evolution. They are particularly apt to ignore the strategic, creative, contextual manner in which
cultural novelty is generated and assimilated. Arguments against a theory of cultural evolution, on
the other hand, generally consist merely of statements as to how the cultural situation differs from
that of biology [e.g. Gould 1991; Hallpike 1986; Jeffreys 2000; Pinker 1997; Sperber 1994;
Thagard 1980]. But such arguments do not a priori constitute a viable reason not to seek an
1
Liane Gabora
evolutionary framework for culture. It is not the case that if culture does not evolve the way
biological organisms evolve that it does not evolve at all. By being extremely clear about both the
similarities and differences between biological and cultural change, we can determine objectively
whether evolutionary theory as it stands can shed light on culture. Only then can we hope to
encompass the richness of the cultural dynamic in the simple beauty of a theory of evolution. And
in so doing, we are led toward a more general theory of how something could evolve.
This thesis outlines how the concept of evolution can map onto culture, and illustrates how
an evolutionary perspective provides not only a foundation for research into the dynamics of
ideas and artifacts at the cultural level, but a synthetic framework for understanding the cognitive
transformation by which an individual becomes a cog in the culture evolving machinery.
1.1
TERMINOLOGY
It is useful to begin by clarify the meanings of terms that will be commonly used. An idea, artifact
(e.g. a painting or gadget) or overt action (e.g. a gymnastics move, or smile of approval) that is or
could potentially be expressed or manifested in the social world is referred to as a cultural entity4.
These sorts of patterns in physical space begin as patterns in a state space that is commonly
referred to as conceptual space since the states it deals with are states of mind. Thus, we will also
be dealing with the mental entities that populate this inner space. When there is no need to
distinguish between a cultural entity expressed in physical space, and a mental entity in
conceptual space, or when an entity is present in both, for simplicity the word idea is sometimes
used. When I use the word culture I refer to the totality of these cultural entities.
We can distinguish different categories of mental entities (although it should be kept in mind
that these distinctions are blurry, and it may be that most mental entities straddle multiple
categories.) Hardwired stimulus-response patterns are referred to as instincts. Experiences stored
in memory are referred to simply as memories. Mental entities that are derived from multiple such
experiences, or that are the product of reasoning or creative thought—such as concepts, stories,
plans and attitudes—are referred to as abstractions. Such constructions are seen to come into
being through interaction between the world, and an ever-changing mental model of that world, or
worldview. (The centerpiece of this thesis is, in fact, a proposal for how this worldview emerges,
and it is suggested that the origin of this kind of relationally structured worldview was the
cornerstone of cultural evolution.)
4
2
The use of and rationale for the term ‘entity’ is introduced in [Aerts 1981].
Chapter 1: The Need for a Theory of Cultural Evolution
Although I have sometimes in the past used the term meme as a concise way of referring to
an evolving unit of cultural information, here it will generally be avoided. One problem with the
word is that it is often used as a catch-all; there is no agreed upon way of distinguishing between
cultural information as mental representation, and cultural information as implemented behavior
or artifact. For example, Durham [1991] defines a meme as “any kind, amount, and configuration
of information in culture that shows both variation and coherent transmission.” It seems useful to,
on at least some occasions, be clear about whether one is dealing with things in the head or things
in the physical world. This problem could be remedied, and if this were the only problem such a
remedy would be worthwhile. However, since the assumptions underlying the way the word came
into being, and the way it is most commonly used, are problematic, as outlined in subsequent
chapters, it seems better to avoid it.
The term mental representation will also be avoided, in keeping with the idea that sensed
information about the world simply changes to a form that is perceivable, so there is no need for a
re-presentation of it to be made in the mind [Skarda 1999]. Another reason to avoid the term is
that it implies that abstract concepts are fixed in form as opposed to changing dynamically
according to context. The approach taken here is compatible with Rosch’s [1999] view that
concepts don’t just identify something meaningful about a stimulus or situation, but they
participate in the construction of this meaning. [For more on anti-representationalism and
constructivist approaches to meaning, see Brooks 1991; Clancey 1993; Freeman & Skarda 1990;
Maturana & Varela 1987; Riegler et al. 1999; Riegler in press; Wheeler 1994].
By emergent I mean a property of an entity that cannot be reduced to its parts because of
nonlinear interactions amongst them. In other words, an emergent property exists at one level of
structure but cannot be fully explained in terms of structure at a lower level.
By context I mean those aspects of reality external to the entity of interest which could
potentially affect the change of state of that entity. In quantum mechanics, the measurement is
what constitutes the context. In psychology and biology, a stimulus constitutes a context. (The
concept of context will be explained in more detail further on.)
Finally, a comment on the term potential. I do not use it in the sense of something waiting to
be actualized in a prefabricated, deterministic sort of way, as biologists sometimes use it. I mean
potential in the sense of any and all states that are possible given the current state. Thus I do not
even mean something as tangible as thoughts or plans or fantasies, which although they are not
visible to us in the everyday world, they are actualized states in conceptual space. Rather I mean
all those mental entities that could, in some possible circumstance, given what one is thinking at
one instant, become a thought or plan or fantasy in the next instant.
3
Liane Gabora
1.2
AIM AND OBJECTIVES
The overall aim of this thesis is to establish culture as an evolutionary process. More specific
objectives are (1) to show how this process is similar to and different from biological evolution,
(2) where it is different, to uncover the mechanisms by which evolutionary change takes place,
and (3) to present a plausible hypothesis for how cultural evolution began.
Two developments arose en route to the fulfillment of this aim that shaped and even came to
determine the content of a substantial part of this body of work. A first development was that I
became increasingly convinced that a theory of cultural evolution must incorporate not just how
ideas are transmitted, but also how they are generated and assimilated in the mind. Thus much of
this thesis is devoted to the mechanisms underlying cognition. One could argue that Darwin came
up with the theory of natural selection, and Mendel the foundations of genetics, before the
discovery of genes. However, I posit that the crucial difference that makes the one possible
without knowledge of the underlying substrate, and the other not, is the following. In biology it
has been possible to go far considering only Darwinian processes acting on entities whose states
and properties or traits are actualized in physical space and directly perceivable in the everyday
world. However, a description of culture must additionally consider non-Darwinian processes
acting on mental entities that are only potential, not actualized, and not directly perceivable in the
everyday world.
This brings me to the second development, which has been the unfolding of a more general,
trans-disciplinary theory of evolution. The gradual clarification of this vision of a unified theory
of evolution came about originally through the desire for a united description of biological and
cultural change. Upon coming to Brussels and learning of the work by FUND5, I came
increasingly to see that the greatest obstacles to a theory of cultural evolution might be solvable
through an approach they were taking to analogous problems concerning potentiality and
contextuality in the micro-world. In this way, Diederik Aerts and I were led to explore a very
tentative but fascinating theory that encompasses all these kinds of change. In this theory,
evolution is viewed as incremental adaptation in response to environmental constraints and
affordances. The concept of ‘environment’ is further generalized to context, and the concept
‘adaptation’ to context-driven actualization of potential. The process of actualizing feeds back on
the state of the entity, and changes its potential. Thus, evolution is viewed as a process of
adaptation through recursive, context-driven actualization of potential. I will show how this
general theory encompasses the following kinds of evolutionary change: (1) dynamical evolution
4
Chapter 1: The Need for a Theory of Cultural Evolution
of a micro-particle as per the Schrödinger equation, (2) measurement-induced collapse of a
micro-particle in a superposition or entangled state to an eigenstate of that measurement, (3)
change of state of classical entities, (4) biological evolution through replication with variation and
selection of actualized organisms, (5) the strategic (or perhaps intuitive) transformation of mental
entities in a stream of thought, and (6) the variation and selective retention of ideas artifacts in
cultural evolution. (It is tempting to refer to a stream of thought as a form of evolution, which we
might call cognitive evolution, except that were it not for culture the process would terminate;
that is, the evolving mental entities would cease to exist when the cognitive individual died.)
1.3
CULTURE IS NOT JUST A NATURAL EXTENSION OF BIOLOGICAL EVOLUTION
A first point to make clear is that culture is not simply a predictable extension of biological
evolution. Let us consider the world as a vast space that comes into being through the creation,
transformation, and destruction of information. This information often exhibits pattern, or
statistical regularity that can be expressed mathematically. After seeing many shadows cast by the
same object we can develop an internal model of what that object looks like without having seen
it, and if more than one object is casting shadows we can learn to tell which object is casting any
particular shadow. Similarly, by viewing every pattern we encounter as a shadow or footprint of
one or more broad causal principles6, we can gain insight into the causal principles that manifest
that pattern.
To examine these causal principles, then, we will take a trip backward in time. If you were to
go back to some time during the first billion years of Earth’s history, you would not find plants or
animals, but you would find pattern. You would find pattern in the trajectories of lava and water,
the weather, and the molecular bonding in rocks. The only causal principle you would need to
invoke to explain pattern in the information present (with the exception of yourself) would be the
physical constraints and self-organizing properties of matter.
If you were to go back to some time after the origin of life, approximately three billion years
ago, this would no longer be the case. Not that life doesn’t exhibit the properties of matter. But it
would be virtually impossible for, say, a giraffe to appear in an information space not acted upon
by natural selection. Another causal principle—biological evolution—would have to be invoked
from this point on.
5
FUND is a research group led by Diederik Aerts at the Free University of Brussels that works on the
mathematical foundations of quantum structure.
6
By ‘causal principle’ I mean something that generates useful descriptions, rather than a ‘law’.
5
Liane Gabora
Today the Earth is embedded with things like paintings, circuses, and computer networks
that cannot be accounted for by appeal to either the properties of matter or biological evolution.
That is, biological evolution does not provide us with adequate explanatory power to account for
the existence of computers any more than the properties of matter can explain the existence of
giraffes. Computers are manifestations of yet another causal principle: the evolution of culture.
Thus pattern in the structure and dynamics of information we encounter in the everyday
world can be traced to three broad causal principles—the physical constraints and self-organizing
properties of matter, biological evolution, and cultural evolution. This classification scheme, like
all classification schemes, is somewhat arbitrary. There may be subclasses of these principles that
deserve to be considered principles unto themselves7, or one could argue that evolution is a selforganizing property of matter, albeit a spectacular one8. The point is: culture is the only process
that has arisen since the origin of life that relentlessly exploits the combinatorial potential of
information.
It is true that brains evolved just like cell membranes or eyes, and thus culture is grounded in
biology (like biology is grounded in the physical constraints and self-organizing properties of
matter). But just because culture is grounded in biological hardware doesn’t mean it can be
dismissed as a natural outgrowth of biology, any more than biology can be dismissed as a
predictable outgrowth of physics or chemistry. The probability of computers arising
spontaneously in an information space not acted upon by cultural evolution (like the probability
of giraffes arising spontaneously in an information space not acted upon by biological evolution)
is vanishingly small. Culture is vastly different from anything else biology has ever given rise to.
In fact it now seems to be outrunning biology (and threatens to annihilate much of it). Writers,
artists, and movie makers invent new life-forms faster than nature itself ever has... new cults and
religions, scientific theories, products and services, not to mention advertisements for products
and services, are springing up faster than new species of trees. With the advent of breeding,
cloning, and genetic engineering, culture has even begun to extend its claws into the very
7
Though viruses are unique in the biological world in that they rely on hosts to replicate, we will consider
viral evolution an anomalous offshoot of biological evolution, because: (1) the evolving patterns of
information are encoded as sequences of nucleotides, (2) variation is through mutation and recombination,
and (3) transmission and selection are mediated through genotype.
8
Or one could argue that the selection of matter over antimatter, and its subsequent amplification and
variation, or cosmic evolution by the natural selection of black-hole universes [Smolin 1999], constitutes
yet another form of evolution.
6
Chapter 1: The Need for a Theory of Cultural Evolution
structure of biological information and manipulate it to suit its needs. It is helping its slow
forefather keep up the pace. Thus it is inappropriate to dismiss culture as a predictable extension
of biological evolution. It is qualitatively different from anything else biology has produced.
A consequence of the fact that the machinery that renders cultural evolution—the human
brain—was originally a product of biological evolution, is that much of what is ‘out there’ cannot
be cleanly traced to a biological or cultural origin. Biology constrains culture through the
preferential spread of ideas that satisfy biologically-derived needs. It goes the other way too;
culture not only affects biological fitness through its effect on behavior9, but it dramatically
modifies the biological world. Some of the ways in which biological information gets tainted with
cultural information seem relatively inconsequential, such as the trimming of hedges, whereas
others, such as dog-breeding, have a long-lasting effect. (In fact, one could view dogs as the
consequence of a cultural trajectory that was launched by the need to protect property.)
Nevertheless, much as it is not imperative to address the role of physical constraints like gravity
in studies of, say, embryonic development or squirrel foraging behavior, much can be said about
culture without addressing the role of biological constraints.
1.4
ANALOGY OR ANOTHER FORM OF EVOLUTION?
Is culture just analogous to biological evolution? Or is it really another form of evolution? And if
it is simply a matter of perspective, is one perspective more fruitful than the other?
The rationale behind the view that they are just analogies is that this helps us avoid
misleading pitfalls [Blackmore 1999]. However, it also discourages potentially fruitful directions.
Consider what happens when someone familiar with the concept of snow skiing first hears of
water skiing. She can continue to include ‘snow’ as vital to the concept of ‘skiing’, and view
water skiing as analogous to skiing. Or she might generalize the definition of ‘skiing’ to include
both snow skiing and water skiing. Neither is objectively more correct than the other, but the
generalized concept of skiing more readily invites the application of knowledge gained through
the study of snow skiing to water skiing. This is not equivalent to saying that water skiing is snow
skiing; it’s just a way of organizing knowledge that is less likely to foster wasting time by
reinventing the wheel. Let us suppose that, a few years before water skis hit the market, you had
been a venture capitalist, and were approached by two inventors who had both come up with the
idea of water skis. One inventor had spent the last few years working for a company that makes
snow skis, and talked at length about how the length and width of a ski affect glide and balance,
7
Liane Gabora
why skis curl up at the tip, etc. The other knew little about snow skis and argued this didn't matter
because snow and water are not the same thing, it’s “just an analogy”. Which inventor would you
have been more likely to fund?
Similarly, we can either: (1) say that DNA, membranes, and so forth (in the context of their
natural milieu) are vital components of evolution, in which case cultural change is just analogous
to evolution. Or we can (2) say that the physical medium is not important, but what makes a
process an evolutionary process is that it involves replication, variation, and selection. Although
some scientists would then say that culture evolves, this is far from proven or universally
accepted. Or we can (3) define evolution more generally as incremental adaptation to
environmental constraint and affordances, and view cultural change as definitely a form of
evolution. We then explore to what extent culture evolves through replication, variation, and
selection, and to what extent other mechanisms are involved. It is this third approach that is taken
here. In so doing, we do not commit ourselves to the applicability of any particular aspect of
biology to culture. However, we find that concepts such as fitness, epistasis, drift, and so forth are
invaluable tools for investigating how ideas unfold as one individual after another assimilates
them and gives them their own unique slant. Knowledge of how evolution works under one set of
circumstances can then be used as a scaffold to direct the study of how it works under a different
set of circumstances.
The issue of ‘analogies’ also comes up with respect to two other aspects of this thesis. First,
regarding the theory of conceptual closure in chapter eight. Like Kauffman’s theory of the origin
of life, it is derived from the work of graph theorists Erdos and Renyi on closure. It is presented
as not as just an ‘analog’ to Kauffman’s theory, but rather as another manifestation of a closure
structure. The same goes for the application of approaches to dealing with potentiality and
nondeterministic effect of context in chapters eleven to thirteen, which were first used in quantum
mechanics. They are not merely analogous to the sort of structure that appears in quantum
mechanics. They are genuinely another manifestation of this kind of underlying structure.
1.5
CHAPTER BY CHAPTER OUTLINE
Here I summarize the contents chapter by chapter. (Note that if you do not have time to read the
whole thesis, the most important chapters are three, eight, twelve, thirteen, and fourteen.)
In chapter one it was argued that culture is not simply an extension of biology. Biological
evolution offers about as complete an explanation for the existence of dishwashers as the physical
9
8
The phenomenon wherein behavior affects biological fitness is known as the Baldwin Effect.
Chapter 1: The Need for a Theory of Cultural Evolution
constraints and self-organizing properties of matter offers for the existence of giraffes. Moreover,
culture need not be viewed as just analogous to biological evolution, but rather as a genuine new
form of it.
Chapter two summarizes how new perspectives on biological evolution, such as neoDarwinism, universal Darwinism, and complexity theory, have spawned new approaches to
cultural evolution, such as diffusion studies, evolutionary epistomology, and memetics. These
approaches have attained some measure of success applying concepts from biology—e.g. linkage,
altruism, and hitchhiking—to culture. However each of these approaches has limitations, and the
concerns voiced by skeptics need to be taken seriously, for their critiques are invaluable for
making real progress toward an understanding of cultural change.
Chapter three examines the current unresolved issues and problems that stand in the way of
viewing culture as an evolutionary process. I argue that the second replicator notion is misleading
because although cultural entities may constitute a self-description, they lack instructions for how
to self-replicate. It is we who do the replicating, not the entities themselves, and in so doing we
change them however we wish; thus acquired characteristics can be inherited. The replicator idea
misleadingly led many to restrict that which can evolve through culture to entities acquired
through imitation. Current evolutionary theories do not address the strategic, intuitive, contextual
manner in which cultural novelty is generated and assimilated. This is examined in the context of
breadth first versus heuristic search. Finally, current evolutionary theories of culture cannot cope
with potentiality. This is examined with respect to internal selection, indecision states, and
blending versus particulate inheritance.
Chapter four presents a computational model of the process by which culture evolves in a
society of interacting individuals. The program consists of an artificial society of interacting
neural network based agents which do not have genomes, and neither die nor have offspring, but
which invent, implement, and imitate ideas. Every iteration, each agent has the opportunity to
acquire a new idea, either through (1) innovation, by mutating a previously learned idea, or (2)
imitation, by copying a idea implemented by a neighbor. This minimal model enables us to
specify precisely some of the mechanisms that differentiate cultural evolution from biological
evolution. We find that these different mechanisms nevertheless manifest similar phenomena.
Since culture evolves when an individual makes contact with either (1) something or some
one in the world, or (2) a memory or abstraction it has stored within it, an understanding of
cognition is clearly important here. Chapter five examines the structure of memories and
abstractions. It also investigates how they are stored in a sparse, distributed, content-addressable
memory, and later on called back into awareness—that is, evoked or reconstructed—in the course
9
Liane Gabora
of a stream of thought. This chapter is largely to provide background material from cognitive
science and psychology for what comes next.
Chapter six discusses the mechanisms underlying both the divergent, inspirational phase and
the convergent, refinement phase of the creative process. It presents a hypothesis regarding the
cognitive basis of conceptual fluidity, and discusses this in relation to traits of highly creative
individuals: defocused attention, heightened sensitivity, and flat associative hierarchies. It also
discusses how actualized cultural novelty undergoes a Darwinian sort of selective process.
It is often assumed that imitation is what makes us unique; the idea being that humans can
imitate, and animals cannot, and that is why we alone evolved culture. Chapter seven presents
evidence from animal research and archeology that point toward an alternative hypothesis: that it
is the strategic, contextual manner humans generate novelty—in other words, creativity—that
unleashed culture. This chapter also presents a paradox that arises when one considers how the
sort of creative mind that could constitute the hub of cultural evolution could have arisen: until a
mind incorporates relationships amongst memories, how can one evoke another in a stream of
thought? And until one can evoke another, how are relationships established amongst memories
so that they become connected into a worldview? The origin of life presents an analogous
paradox: if living organisms come into existence when other organisms give birth to them, how
did the first organism arise? How did something able to reproduce itself come to be?
Chapter eight shows how a promising solution to the origin of life problem—namely, the
idea that life began through the self-organization of an autocatalytic network of
molecules—suggests an avenue for tackling the origin of culture problem. If one considers
memories to play the role of molecules, and reminding events to play the role of catalysis
(chemical reactions amongst these molecules) this basic idea translates nicely into a plausible
account of how memories become woven into a relationally structured model of the world, or
worldview.
Chapter nine explores the possibility that this same process re-occurs in the mind of every
young human in the process of cognitive development; it sketches out some thoughts concerning
how an infant develops an internal model of the world that both structures, and is structured by,
self-triggered streams of potentially-creative thought.
Having examined the underlying cognitive mechanisms, chapter ten returns to the problem
of tying together a coherent theory of cultural evolution. In so doing, we are led to the attempt to
reformulate the concept of evolution itself more precisely. Neo-Darwinian evolution is reviewed
and seen to be lacking with respect to its capacity to incorporate potentiality and contextuality, or
10
Chapter 1: The Need for a Theory of Cultural Evolution
more precisely situations where we do not have complete knowledge regarding how potential gets
actualized under the influence of a context.
Chapter eleven shows how it is possible to incorporate the aspects of evolution shown to be
lacking in the previous chapter using mathematical formalisms developed for the purpose of
generalizing quantum mechanics to describe entities that are pure quantum, or pure classical, or
anything between. This formalism is used, not because of any relation between culture and events
at the quantum level, but because it happens to be the most serious attempt to date to describe
situations involving the appearance of genuine novelty, and intermediate degrees of potentiality
and contextuality.
Drawing on the ideas of the previous chapter, in chapter twelve, a tentative, new, transdisciplinary description of evolution is proposed, wherein the process is viewed as the recursive
context-driven actualization of potential (or CAP). It shows how physical, biological, and cultural
evolution fall out of this general theory of evolution. These forms of evolution differ in the degree
to which potential is retained, degree of determinism, degree of sensitivity to context, and
whether they are context dependent or context independent.
Chapter thirteen provides preliminary evidence that the general theory of evolution
developed in the previous chapter is of explanatory value for culture and its underlying cognitive
mechanisms. It demonstrates the presence of quantum structure in the cognitive architecture by
proving that Bell inequalities are violated using a simple example involving a concept and
instances of that concept. It shows that this model of concepts overcomes problems other models
have explaining how the mind deals with conjunctions of concepts using the well-known ‘pet fish
problem’.
Finally, chapter fourteen provides a synthesis of the previous chapters, and summarizes
what this thesis tells us about how adaptative change works in culture and other processes of
evolution.
11
Liane Gabora
theThis says nothing
12
Chapter Two
2
Current Evolutionary Approaches to Culture
In the introduction, evolution was loosely described as a process of incremental change that
reflects the constraints and affordances of an environment. We now examine more precisely what
evolution is thought to entail. In fact, the concept of evolution has itself evolved. This chapter
briefly investigates some of the stages in its evolution, and the slow but steady trickle of efforts
they have engendered to shed light on the dynamics of culture. Each section describes a stage in
evolutionary theory and is accompanied by examples of how it can be applied to culture. Unless
stated otherwise the examples are my own (and for others the reader is referred to the literature
cited). Not all of them may, in the end, prove to be fruitful; the hope is merely that they give some
idea of what is possible.
2.1
DARWINIAN APPROACHES TO CULTURE
First according to a strict Darwinian perspective, evolution requires:
•
An organism that replicates through sexual or asexual reproduction.
•
Replication is imperfect such that random variation is introduced.
•
An environment in which the variant organisms compete to survive and reproduce.
Thus, organisms reproduce in an environment, some thrive better in this environment than others
and have more offspring, and their traits tend to be more widely represented in the next
generation. In this way, successive generations of organisms tend to become better able to meet
the challenges and capitalize on the opportunities afforded by the circumstances they are faced
with. The situation is complicated by the fact that the evolving entity’s environment often consists
13
Liane Gabora
largely of other entities which are themselves evolving, so symbiotic alliances and coevolutionary
arms races become established.
2.1.1
Diffusion and Coevolutionary Studies
Steps toward a Darwinian perspective on culture have been taken in the social sciences. The
evolving entities are ideas, attitudes, values, artifacts, stories, and so forth. They are said to
replicate when transmitted to others through social learning processes such as teaching or
imitation [e.g. Tomasello et al. 1993]. Cultural variation—which as often noted is not random—is
generated by strategically and creatively combining and transforming these entities. They are
selected on the basis of how well they satisfy current biological drives, goals, desires, aesthetic
preferences, and so forth. The environment in which they compete could be said to consist of both
the worldviews of interacting individuals, and the physical milieu in which they manifest as
actions or artifacts.
Social diffusion studies [Abrahamson 1991; Abrahamson & Rosenkopf 1993, 1997; Brown
1981; Rogers 1962] examine the spread of new ideas and technological advancements through a
society. The assimilation and diffusion of cultural novelty alters the selective pressures and
constraints it exerts on the individuals embedded in it, which in turn alters the generation and
proliferation of future ideas. The ideas any one individual produces build on the ideas of others, a
phenomenon known as the ratchet effect [Boesch & Tomasello 1998].
A related project, coevolutionary studies, [Durham 1991; see also Lumsden & Wilson 1981],
focuses on interactions between biological and cultural evolution. They investigate how such
cultural entities as marriage customs affect distributions of genes, and how this in turn feeds back
on and affects the distribution of marriage customs. Viewing the human mind as a player in the
process of cultural evolution is not incompatible with approaches that stress the role of innate
mechanisms [e.g. Pinker 1995]. Rather, as Lumsden and Wilson point out, it builds on this
framework, adding that the study of cognition will flounder until we admit that the role of nongenetic processes is equally undeniable.
These studies provide an interesting new perspective on how ideas are transmitted amongst
individuals, and the global patterns that can result. However, they necessarily sidestep the
complex processes through which cultural novelty is generated and assimilated; for instance, the
subtle changes cultural entities undergo as they are mulled over in one mind and then another.
There is some justification for this approach; Mendel went far toward an understanding of
biological change without any knowledge of DNA. However, the assumption that the underlying
mechanisms can be swept under the rug is less valid for culture than for biology because cultural
14
Chapter 2: Current Evolutionary Approaches to Culture
(unlike biological) novelty is generated and assimilated strategically and intuitively, on the fly,
and the associative organization of the memories and abstractions that together constitute a
mental model of the world constrains how one idea evokes another.
2.1.1.1
An Example: Individuation in Families and Social Groups
Different species result when populations of organisms become too different from one another to
be able to mate, and which of course leads them to become even more different from one another.
This process is referred to as speciation. The concept of speciation can be applied to the
individuation and division of labor in a family or society. In both the biological case and its
cultural analog, small differences are amplified through positive feedback leading to formation
and transformation of viable niches. This could provide answers to questions such as why siblings
are often so different from one another. As one sibling becomes increasingly identified as the
athlete of the family, other siblings, unable to compete in this particular domain, veer toward
other domains such as academics or music. As another example, it might shed light on how
sometimes researchers in the same academic discipline, or employees of the same company,
become so specialized in their work that they can no longer ‘conceptually mate’, or speak
meaningfully to one another such that fruitful new ideas result.
2.2
NEO-DARWINIAN APROACHES TO CULTURE
With the discovery of DNA came the advent of neo-Darwinism. It became possible to specify
precisely what gets transmitted from one generation to the next (which sometimes led to a
reductionist tendency to over-estimate the degree to which genes determine biological form). It
was learned that all cells contain a number of chromosomes consisting of genes, encoded as
sequences of nucleotides, that affect the expression of certain traits. Genes come in variant forms
called alleles. For example, the allele of the gene for eye color on one member of a chromosome
pair might code for brown eyes, while the allele on the other might code for blue eyes. Variations
arise through mutation and recombination through crossover (intermixing) and natural selection
weeds out those that are maladaptive.
The combination of genes that make up an organism is referred to as the genotype, and the
combination of physical traits and behavior that result from the combined influence of the
genotype and the environment is referred to as the phenotype. The fitness of an organism is a
measure of how well it survives relative to others of the same species, often assessed as the
number of offspring it has. One can consider the relative fitnesses of all possible genotypes for a
species, where the distance between one genotype and another is determined by how many single
15
Liane Gabora
point (that is, one gene) mutations they differ. The result is referred to as a fitness landscape. By
continuously generating variety (divergence) and selectively replicating variants that manage to
survive for some time (convergence) the possibilities of the fitness landscape are explored.
Thus the newfound understanding of the molecular mechanisms underlying natural selection
caused the focus to shift from the organism to, on the one hand, the genome—the set of all genes
for an individual—and on the other hand, the population—the group of organisms that interbreed
with one another, and therefore whose genomes interact. It also inspired a minuscule (by
comparison) flourishing of applications to culture. Some attempts to put neo-Darwinism to use to
gain insight into cognitive and cultural phenomena are summarized below.
2.2.1
Approaches Inspired by Population Genetics
Many have drawn from mathematical models of population genetics, developed originally by
Fisher, Haldane, and Wright, to model the transmission of ideas [Cavalli-Sforza & Feldman 1981;
Schuster & Sigmund 1983; Boyd & Richerson 1985; Hofbauer & Sigmund 1988]. These models
bravely demonstrate how far one can go toward a formal description of cultural change assuming
underlying mechanisms virtually identical to those of biology. They unearthed cultural analogs of
a number of biological phenomena. The network of mental entities that together constitute a
mental model of the world could now be viewed as the cultural analog of the genotype. The way
they get implemented or communicated, through facial expressions, gestures, actions,
vocalizations, or artifacts could be viewed as the analog of phenotype. They explore the
mathematical consequences of biased transmission of cultural information, such that some
individuals are imitated more often than others, or of allowing for cultural entities with more than
two ‘parents’ [Boyd & Richerson 1985]. Cavalli-Sforza & Feldman [1981] demonstrated a
cultural analog of genetic drift—changes in the relative frequencies of different alleles due to
random sampling processes in a finite population [Wright 1969]. However, these models do not
begin to come to terms with the open-ended diversity of culturally-derived information. Whereas
Darwinian models ignore the underlying mechanisms of culture, here they are incorporated, but
the cultural mechanisms they work with are biologically-inspired caricatures of the real thing. For
example, cultural novelty comes about through trial and error learning or transmission error. One
worries that by isolating and manipulating the easiest part of the problem, and assuming
equivalence with the biology when things get messy, they present the illusion of precision and
clarity while throwing the baby out with the bath water.
16
Chapter 2: Current Evolutionary Approaches to Culture
2.2.2
Epidemiological Models of Culture
Epidemiology, the study of the origin and spread of disease, has also had cultural offshoots, such
as investigations into social epidemics [e.g. Mackintosh & Stewart 1979] and social contagion
[Burt 1987; Rodgers & Rowe 1993]. These studies are largely aimed at discovering the factors
that influence the spread of new artifacts, or attitudes and behaviors. They have potentially
valuable practical applications, such as curbing the spread of addictive habits like cigarette
smoking, and enhancing the spread of beneficial ones like recycling.
2.2.2.1
A First Example: Conceptual Linkage Disequilibrium
As a first example of the application of neo-Darwinism to culture, recall from chapter one how
time and again it is argued that a theory of cultural evolution is doomed simply because it would
have to work through different mechanisms from those of biological evolution. Ironically this
situation in itself provides us with a nice example of how knowledge of evolution acquired in the
realm of biology can help unravel analogous situations in the realm of culture. An important
theme in population genetics concerns how the present is biased by accidents or random
associations of the past. The closer together two genes are on a chromosome, the greater the
degree to which they are linked. Linkage equilibrium is defined as random association amongst
alleles of linked genes. Consider, for example, the following situation:
•
A and a are equally common alleles of Gene 1.
•
B and b are equally common alleles of Gene 2.
•
Genes 1 and 2 are linked (nearby on same chromosome).
There are four possible combinations of genes 1 and 2: AB, Ab, aB, and ab. If these occur with
equal frequency, the system is in a state of linkage equilibrium. If not, it is in a state of linkage
disequilibrium. If alleles of linked genes have tended to show up together on the same
chromosome in the past, it can take many generations before you ever see one without the other,
although each generation there is some chance that they will get separated. For example, the allele
for red hair has historically been linked to the allele for freckles, and even today, people tend to
either have both red hair and freckles, or neither.
One can measure the number of generations necessary for these genes to achieve a state of
random association, or linkage equilibrium, so that they have only a chance probability of
appearing on the same chromosome. (This process can also be modeled computationally.)
Disequilibrium starts out high, but tends to decrease over time because mutation and
17
Liane Gabora
recombination break down arbitrary associations between pairs of linked alleles. However, at loci
where this does not happen, one can infer that some combinations are fitter, or more adapted to
the constraints of the environment, than others. Thus when disequilibrium does not go away, it
reflects some structure, regularity, or pattern in the world.
This kind of history-dependent association also plays a role in the workings of the mind.
Like genes, the features of memories and abstractions are connected through arbitrary
associations as well as meaningful ones. We often have difficulty applying an idea or problemsolving technique to situations other than the one where it was originally encountered, and
conversely, exposure to one problem-solving technique interferes with ability to solve a problem
using another technique [Luchins 1942]. This phenomenon, referred to as mental set, plays a role
in cultural evolution that is reminiscent of linkage in biological evolution. To incorporate more
subtlety into the way we carve up reality, we must first melt away arbitrary linkages amongst the
discernable features of memories and concepts, thereby increasing the degree of equilibrium. As
we destroy patterns of association that exist because of the historical contingencies of a particular
domain, we pave the way for the forging of associations that reflect genuine structure in the world
of human experience which may manifest in several or perhaps all domains. This needn’t be an
intellectual process. For example, one might have a sudden glimmer of insight into how the
feeling of a particularly emotional experience could be extricated from the specifics of that
experience, and re-manifest itself as, say, a piece of music.
We can view mental set as a state of conceptual linkage disequilibrium. Blindly insisting that
the concept of evolution is only useful for understanding biological organisms is a good example.
In this case, achieving conceptual linkage equilibrium is a matter of abstracting the basic concept
of evolution from its biological manifestation so that it can be applied with ease to the case of
culture. One could argue that it would make sense for cultural evolution to be the default form of
evolution outside of biology, just as the default form of skiing in tropical climates is water-skiing
rather than snow-skiing.
As in biology, the biasing effect of historical association gradually decreases over time. For
example, as Hodder [1998] points out, although the first motorized vacuum cleaners and washing
machines looked like their pre-motorized predecessors, gradually they lost the aspects that had
been imposed by constraints that no longer existed, and took on their modern forms.
2.2.2.2
A Second Example: A Cultural Analog to Genetic Hitchhiking
The closer together genes are on a chromosome the less likely they will be separated by
crossover, so the more tightly linked they are said to be. Hitchhiker alleles confer no fitness
18
Chapter 2: Current Evolutionary Approaches to Culture
advantage, but endure because they are tightly linked to alleles that are important for survival
[Kojima & Schaeffer 1967; Maynard Smith & Haigh 1974; Kaplan et al. 1989]. (The concept of
hitchhiking is related to that of exaptation [Gould 1991]—the evolution of organs or traits not
evolved through natural selection for their current use.)
In an influential paper on the relationship between DNA polymorphism and recombination
rates, Begun and Aquadro [1992] suggested that genetic hitchhiking may have a more significant
evolutionary impact than had previously been supposed:
This correlation suggests that levels of neutral variation in many of the gene regions for which
variation has been measured have been reduced by one or more hitchhiking events. Provided
that a new selectively favored mutation goes to fixation before another advantageous mutation
arises close to it, each fixation will be surrounded by a ‘window’ of reduced polymorphism, the
relative size of which is proportional to the rate of recombination for that region of the genome.
The general idea here translates nicely to culture. If an idea goes to fixation in a society due to
some advantage conferred by it, the region of conceptual space containing ideas related to it will
also exhibit a ‘window’ of reduced polymorphism, the size of which may vary according to how
closely related or dependent they are upon the fixated idea. For example, once midi became the
standard for electronic music keyboards, other electronic music gadgetry became midi
compatible. Conceptual linkage equilibrium is achieved when all instances of hitchhiking have
been obliterated. In both genetic hitchhiking and its cultural analog there is indirect selection for
useless (or even detrimental) patterns through their association with beneficial ones. One could
argue that play, recreation, and other creative endeavors are the re-creation of information
patterns in different domains from the ones in which they were originally encountered, thereby
filtering out conceptual prejudices that reflect nothing more than mechanical constraints or
historical legacies of the original domain. Play is then an algorithm for achieving a state of
conceptual linkage equilibrium through mental operations that, like genetic recombination,
increase polymorphism by reducing fixation through hitchhiking.
It may be useful to decompose cultural information into: (1) core features, which contribute
directly to fitness, (2) enabler features, which enable or facilitate the implementation of core
features, and (3) hitchhiker features, which are present due to arbitrary or accidental historical
associations to features of the first two kinds. Core features tend to convey semantic information,
and enabler features syntactic information, though one can think of situations in which some
semantic information serves simply to facilitate expression of other semantic information i.e.
19
Liane Gabora
functions as an enabler. The first two categories are vaguely analogous to the categorization of
genes as structural or regulatory, and the last category is inspired by the phenomenon of genetic
hitchhiking.
2.2.2.3
A Third Example: A Possible Explanation for Cognitive Redundancy
The question of why there is so much redundancy in the genetic code has generated much
discussion which may also apply to the question of why there are redundant mental maps in the
brain; both may reflect the need for reliable fault tolerance in an evolving (and thus everchanging) information code [Calvin 1996; Edelman 1989, 1992; Gabora 1996, 1997]. Note that
this issue is indirectly related to the issues of linkage disequilibrium and hitchhiking discussed
above. One might expect less redundancy, for example, in hitchhiking material; since it is
possible to get along without it, it doesn’t matter so much whether it changes or not.
2.2.3
Universal Darwinism, Second Replicators, and Memes
In The Selfish Gene, Dawkins [1982] construes of genes as replicators. A replicator is a physical
entity that makes copies of itself, and which has the following properties:
•
Longevity—it survives long enough to replicate.
•
Fecundity—at least one version of the replicator can make multiple copies of itself.
•
Fidelity—even after the replicator has undergone several generations of replication, it is still
almost identical to the original.
Dawkins espoused the notion of universal Darwinism, the idea that Darwinian evolution is not
entirely dependent on the physical structure of organic life; in other words, that it is ‘substrate
neutral’. Thus, he suggests, the evolution of species is not the only Darwinian process; others
include intra-individual biological processes (such as the functioning of the immune system), life
elsewhere in the universe (if it exists), and culture. Accordingly, he thought of cultural entities as
replicators of cultural information analogous to the gene, and coined the word meme to refer to
them. The notion of the replicator had an impact on culture to the extent that the view that culture
constitutes an evolutionary process is sometimes assumed to be synonymous with commitment to
the notion of a ‘second replicator’. He elaborates: “Just as genes propagate themselves in the gene
pool by leaping from body to body via sperm or eggs, so memes propagate themselves in the
meme pool by leaping from brain to brain.”
20
Chapter 2: Current Evolutionary Approaches to Culture
The meme concept has been taken up and expanded upon by many [Best 1997, 1998, 2000;
Blackmore 1999; Brodie 1996; Calvin 1997; Dennett 1991, 1995; Durham 1991; Gabora 1995,
1996, 1997; Heylighen 1992; Lake 1998; Lynch 1996; Speel 1997]. It even gave rise to a
fledgling discipline, appropriately termed memetics. However, the meme idea has also been
sharply criticized as a “meaningless metaphor” [Gould 1991], and “cocktail party science” [Orr
1996]. Many have argued over the extent to which cultural entities exhibit fidelity, fecundity, and
longevity. However, a more fundamental problem is that it is not clear that anything in culture
makes copies of itself in the first place. The process of meiosis by which this happens in biology
is extremely complex, and precisely orchestrated, according to instructions in the DNA.
Chromosomes line up in pairs along the equator of a cell where their genetic information
replicates, and recombines. They then pull apart on tightrope-like spindles such that one copy of
each ends up in each of four daughter cells. The process of transmitting an idea to others seems to
have little in common with meiosis; it seems more akin to a radio signal picked up by multiple
radios, or even the light of distant stars reaching planets such as earth. It is one thing for an entity
to continue to exist, and to undergo transformations as it moves through time and space, and to
leave imprints on the various physical media it encounters along the way. It is quite something
else for an entity to explicitly contain and carry out instructions for how to make copies of itself.
We will return to this issue in the next chapter.
2.2.3.1
A First Example: A Potential Explanation for Human Altruism
Perhaps the most valuable contribution of the approach is that it put the emphasis at the level of,
not just the individuals who cultivate and express ideas, but at the level of the ideas themselves.
Much can be gained by looking at how they evolve as they pass from one mind to another. For
example, from within the memetic framework arose an intriguing possible explanation for human
altruism. The fact that new ideas regularly take root and die within a single human generation
suggests that selective pressures operating at the level of their bearers’ survival are not the only
selective pressures at play. Heylighen [1992] suggests that memes, like genes, exert pressure on
individuals to behave altruistically toward others who share them (the cultural equivalent of kin
selection). Intuitively this makes sense; we are more inclined to be nice to those who share our
ideas, attitudes, tastes, and so forth, than those who share our eye color or blood type. Thus
reciprocal interactions between like-minded individuals—such as the telling of stories, rumors,
and jokes—would be a form of cultural altruism analogous to the biological altruism that occurs
between genetically-similar individuals. Ongoing reciprocal interactions of this kind may result in
the emergence of a culturally-derived social structure, wherein individuals who regularly generate
21
Liane Gabora
pleasurable or powerful ideas come to be observed carefully and imitated frequently, while other
individuals are ignored [Gabora 1996, 1997]. These outcasts may be excluded from society and
come to exhibit a cultural version of the Founder Effect [Holgate 1966]—reduced variation due to
drift (which we recall refers to random statistical bias due to sampling error).
Note that this provocative explanation of altruisum goes deeply against the grain of
biological thought. Laland et al. [1999], who even explicitly support the general notion of mental
entities, explain behavior exclusively in terms of biological pressures, i.e. natural selection. They
suggest that in human altruism “it is not genes that are selected for, but rather groups of
individuals expressing a particular culturally transmitted idea.” However, it seems simpler to
posit that selection is taking place, not at the level of groups of individuals, but at the level of
ideas themselves. If natural selection can have a huge impact on behavior, why not cultural
selection?
2.2.3.2
A Second Example: Runaway Cultural Selection
Runaway sexual selection occurs when females prefer to mate with males that are exaggerated
with respect to a particular trait. The more extreme the trait, the greater the fitness of the male;
that is, the more offspring he has. This leads through positive feedback to the evolution of
extremities such as the spectacular plumage of the peacock. Note that the original reason for
female preference can be overshadowed by the fact that the exaggerated form of the trait comes to
indicate male fitness; thus so long as females prefer it, it may be selected for even if it impedes
the male’s everyday functioning.
Similarly, ideas—despite being derived directly or indirectly from human need—sometimes
work against our own survival [Greene 1978; Alexander 1980]. Much like runaway sexual
selection in biology, an idea can evolve out of the orbit of the need that originated it. This
phenomenon has been referred to as cultural momentum [Gabora 1997], or memetic driving
[Blackmore 1999]. We cannot help but engage in a stream of thought, spontaneously generating
new ideas like “if only such and such had been different...”, any more than biological evolution
can help but generate new species. This could explain why, despite the intuition that individuals
control their streams of thought, creators often express surprise at the sudden appearance of an
idea, and deny active effort in its immediate creation [Bowers et al. 1990; Guilford 1979; Kubose
& Umenoto 1980; Wallas 1926]. We seem to control the birth of ‘our’ ideas only to the extent
that we provide a fertile ground for them to be fruitful and multiply—by internalizing relevant
background knowledge, identifying new needs, and exposing ourselves to stimuli that help trigger
ideas that fulfill those needs. (So if you don’t like this idea, don’t blame me. :-)
22
Chapter 2: Current Evolutionary Approaches to Culture
2.2.4
Interactors and Lineages
The concept of replicator is related to the idea of genotype. Dawkins also presented a somewhat
new perspective on the concept of phenotype; he introduced the term vehicle to refer to the
physical body that houses the replicator and gets it from place to place. Hull [1988] finds the
notion of vehicle too passive, and uses instead the term interactor. An interactor is “an entity that
interacts as a cohesive whole with its environment in such way that this interaction causes
replication to be differential.” Thus Hull stresses the contextual relation between an entity and its
environment. This is seen also in his writings on culture, where he hints to a perspective wherein
cultural entities are not explicitly replicated structures, but rather structures that take on different
forms as they encounter different media, including different minds. Hull also prefers to think in
terms of lineages rather than species.
The perspective taken here follows through on Hull’s basic insight, by viewing humans as
not just passive receptors and transmitters of ideas, but active evolvers of cultural information,
and to view minute-to-minute experience as a continuously creative process. We don’t assimilate
a new idea unless it either logically fits into our worldview, or rings true intuitively, and we use
this continually-updated worldview to guide innovation. These events, operating in parallel at the
level of the individual, give rise to a higher level structure at the level of society.
2.3
COMPLEXITY THEORY AND THE GENETIC ALGORITHM
The neo-Darwinian concept of evolution has been abstracted by computer scientists in the form of
a search technique referred to as the genetic algorithm, or GA [Goldberg 1989; Holland 1975].
GAs have in fact been used to solve a broad class of difficult problems ranging from electronic
circuit design to factory scheduling to automated programming. The basic idea is: (1) generate a
set of possible solutions to the problem which are coded as bitstring ‘artificial chromosomes’, (2)
rate them according to how well they solve the problem, (3) select the best ones to undergo
mutation and/or recombination and/or other operators, thereby producing a new generation, and
(4) repeat steps 1 to 3 until a satisfactory solution is found.
Thus arose a new perspective on evolution as a parallel, or more specifically, populationbased approach to problem solving which requires:
•
A pattern of information (an actualized state within a space of possible states).
•
A means of randomly varying the pattern (exploring the space).
23
Liane Gabora
•
A rationale for selecting variations that are adaptive, i.e. that tend to give better performance
than their predecessors in the context of some problem or set of constraints (a fitness
landscape applied to the space).
The GA is generally considered to fall under the umbrella of complexity theory. However, also
largely under the umbrella of complexity theory, there has arisen several sources of evidence that
natural selection, powerful though it is, cannot account for all, or perhaps even most, evolutionary
change [e.g. Boone et. al. 1998; Kauffman 1999; Wesson 1994]. As Kauffman writes:
Darwin was right about descent with modification and natural selection. However, these
concepts do not answer the question of how forms, morphologies, phenotypes, and behaviors
arise in the first place.
...the concept of meme, and its descent with modification, is taken as a (or perhaps the)
central conceptual contribution to the evolution of human culture. But the conceptual framework
is so limited as to be nearly trivial. Like NeoDarwinism, it suffers from the inability to account
for the source of new forms, new memes. Moreover, mere descent with modification is a vastly
oversimplified image and understanding of how in cultural and technological evolution, new
concepts, artifacts, legal systems, modes of governance, and modes of coevolving organizations
at different level have come into existence in the past three million years, of how culture
continues to transform today.
Non-neo-Darwinian processes—such as autopoiesis [Varela 1979] emergence [e.g. Chandler &
Van de Vijver 1999; Kampis 1991; Kauffman 1993; Rosen 1985], symbiosis [Margulis & Fester
1991], cooperation [Maynard Smith & Szanthmary 1995], hypercycles [Eigen & Schuster 1978],
punctuated equilibrium [Eldridge & Gould 1973], and epigenetic mechanisms [Newman &
Muller 1999]—impose constraints on the structural stability and change of biological form.
Furthermore, the generation of variation is not completely random; convergent pressures are
already at work prior to the physical realization of organisms. First, mating is often
assortative—mates are chosen on the basis of traits they possess or lack, rather than at
random—and relatives are avoided as mates. Second, since Cairns [1988] initial report, there is
increasing evidence and acceptance of directed mutation, where the frequency of beneficial
mutations is much higher than chance, particularly in environments to which an organism is not
well adapted. Thus, there is more going on in evolution than random variation and natural
selection. Furthermore, neo-Darwinian theory says nothing about the highly contextual
24
Chapter 2: Current Evolutionary Approaches to Culture
circumstances that culminate in the conception of particular individuals (for example, why you
exist rather than some other individual your parents could have conceived).
Complexity theory also has been applied to culture [e.g. Axelrod 1997; Bak 1996; Bhargava
et al. 1993; Cohen & Stewart 1994, 1997], although generally not with the purpose of explaining
the evolution of cultural entities. Below I give a few examples.
2.3.1
A First Example: Emergence of an Interconnected, Culture-evolving Worldview
In chapter seven I present in detail an autocatalytic model for how an individual develops a
relationally structured worldview and thereby becomes a hub in the culture evolving machinery.
The basic idea is that reminding events trigger the emergence of abstractions, and as the number
of abstractions increases, the probability that they ‘crystallize’ into an associative network
increases exponentially. When this happens, for each mental entity there exists a possible
pathway of reminding events through which it could be retrieved, and new stimuli and ideas are
continuously woven into this interconnected structure.
2.3.2
A Second Example: A Cultural Analog to the Pre-Cambrian Explosion
The last century has witnessed a sudden proliferation of gadgets, electronic devices and the like.
This cultural explosion is not unlike the explosion of new species in the Pre-Cambrian era. In
both cases, a period of evolutionary stasis was followed by a period of rapid change and abundant
novelty. It may be that research into the factors the precipitated one transition may shed light on
the factors that precipitated the other. For example, it has been suggested that many complex
systems have critical transitions of varying scales or sizes, the distribution of which follows a
power law [Bak 1996, Bak & Chen 1991]. This has been referred to as self-organized criticality
because it comes about through the intrinsic dynamics of the system, independent of the value of
any control parameter. The paradigmatic example of a self-organized critical system is a sand
pile. Sand is dropped one grain at a time, forming a pile. As the pile grows, avalanches occur
which carry sand from the top to the bottom of the pile, and it is the distribution of sizes of such
avalanches that follows a power law. It may be that the Pre-Cambrian explosion of species and
the later century explosion of artifacts are, like particularly large avalanches, instances of selforganized criticality.
2.4
SELECTIONISM AND EVOLUTIONARY EPISTOMOLOGY
It has been suggested that many or all of the phenomena claimed by complexity theorists and
others to diminish the explanatory power of neo-Darwinism can be subsumed by it through a
25
Liane Gabora
change of perspective, i.e. by shifting the focus from the level of the entity of interest to another
level of the hierarchical structure in which it is embedded [Bickhard & Campbell, unpublished
ms; Campbell 1990; Cziko 1997; Popper 1963]. What this generally means is looking at selective
pressures not just external to, but internal to, the entity. For instance, the abortion of organisms
with lethal mutations is construed as internal selection. This perspective will be referred to as
selectionism, since phenomena that others construe as flying in the face of the Darwinian
paradigm, they construe as exotic forms of selection. Selectionism differs from universal
Darwinism in that it stresses that explicit replication is not necessary so long as there is a retention
or preservation of information.
Interestingly, the intellectual milieu in which selectionism arose was concerned with
applications to culture, particularly the development of scientific thought. Seeking to unite
philosophy and biology, Popper [1963], Lorenz [1971], Campbell [1974, 1987] and others alerted
us to the evolutionary flavor of epistemology, describing the growth of knowledge and rationality
as a Darwinian process wherein conjectures must be refutable, i.e. able to be selected against.
Evolutionary epistomology also stresses that since the original function of knowledge is enhanced
survival of self and offspring, its evolution is affected by the survival value it has for its carriers.
In my view, the selectionist perspective is valuable in many cases. Some processes that have
been labeled emergent do seem to involve a ‘survival of the fittest’, or selective weeding out of
unfit variants. For example, the circuitry of the brain comes about through a process of selective
pruning, wherein neurons that get activated by one another thrive, while those that do not die off
[Edelman 1987]. This process is in fact referred to as neural Darwinism. Darwinism also is of
explanatory value in explaining the functioning of the immune system [Burnet 1959; Jerne 1967]
as mentioned previously. However, the selectionist perspective runs into some difficulties when
the attempt is made to stretch it to cope with not just actual but potential states; for example,
when self-organization is described as a situation where the internal constraints of an entity cause
one organization to get selected from amongst the organizations that were possible for it, or when
it is applied to the focusing of an idea. The mathematics of biological selection theory predicts
change only when there exists a variety of actualized states. It does not provide a means of
selecting amongst potential future states of a single entity. (In fact, it does not even predict change
when there are multiple entities unless there are differences amongst them.) This issue will be
examined in detail in the next chapter. For now, though, note that the critique merely refers to
selection theory as biologists have developed it. Clearly something selection-like goes on during
the refinement of an idea, in the sense that some aspects and implications are culled out and
26
Chapter 2: Current Evolutionary Approaches to Culture
developed, while others not. Campbell in particular emphasizes this, describing our categories of
thought and perception as a nested hierarchy of selective-retention processes.
2.4.1
An Example: Vicarious Selection
Hand in hand with the concept of internal selection arose the closely related concept of vicarious
selection [Campbell 1965, 1974, 1987]. The basic idea is that constraints, pressures, or
preferences operating on one level, domain, or individual, vicariously select for fitter outcomes at
another level, domain, or individual. For example, the bearer of a detrimental idea need not die in
order for the idea to be selected against; it can suffice to witness or hear from someone else the
consequences of believing in or implementing the idea. As another example, the preference for
sugar is said to be a vicarious selector for the propensity to adequately nourish oneself. Since
vicarious selection is indirect, it cannot invoke an appropriate response in an environment where,
for example, sugar is plentiful and robbed of nutrients. Thus instincts and acquired knowledge
vicariously anticipate selection by the environment.
2.5
SUMMARY
This chapter summarized developments of evolution theory that have had an impact on theories
of culture. Although cultural evolution operates through different mechanisms from those of
biology, similar phenomena appear such as linkage and runaway selection, which suggests that
knowledge of biological evolution can be put to use to gain insight into cognitive and cultural
phenomena.
It was in fact culture that inspired the Darwinian theory of how organisms evolve through
natural selection. With the discovery of DNA came the advent of neo-Darwinian, which inspired
population-based and epidemiological models of the spread of ideas. Replicator and interactor
theories gave rise to memetics. Complexity theory pointed to the limitations of natural selection,
and inspired the application of concepts like emergence and phase transitions. Selectionist
theories attempt to revive the idea that variation and selective retention can explain all (or most)
biological and cultural change by positing the notion of internal selection. Each of these
approaches has had its strengths and weaknesses. In all cases, the debate they engendered served
the purpose of clarifying the issues and moving us closer to an answer to the basic question: how
does culture evolve? However, none has generated conclusive answers. In the next chapter we
look at the problems and issues that remain unresolved.
27
Liane Gabora
the
28
Chapter Three
3
Unresolved Issues and Potentially Fruitful Directions
Having examined current evolutionary theories of culture it is appropriate to ask: where do we
stand now? This section outlines what I believe to be the most serious unresolved issues and
outstanding problems that stand in the way of a full description of culture as an evolutionary
process.
3.1
EVOLUTION WITHOUT REPLICATORS
Recall from the last chapter that a replicator is an entity that self-replicates, or makes copies of
itself. In the last chapter I suggest that the concept of replicator is not appropriate for cultural
entities because they do not make copies of themselves. Let us examine this issue more closely.
3.1.1
Self Description versus Instructions for Self Replication
Clearly replication is an integral component of biological evolution, and the transmission of ideas
through social processes such as teaching and imitation is likewise important to culture. But there
is an important distinction between self-replication and other retentive processes, as was first
clearly enunciated by Von Neumann [1966]. He postulated that a self-replicating system consists
of two parts: (1) interpreted information—coded instructions for how to produce offspring (2)
uninterpreted information—a self-description that is passed on to offspring. This is of course the
case for biological life; there are genes that generate a body that acts to sustain itself, and genes
that contain instructions for how (with the help of someone of the opposite sex, for sexual
reproducing species) to create a child. But unlike genes, ideas do not come prepackaged with
instructions for their reproduction. They rely on the machinery of our brains to remember and,
when appropriate, express or embody them. As noted previously, the transmission of an idea is
perhaps more aptly compared to, not biological reproduction, but the transmission of a radio
29
Liane Gabora
signal and its reception by one or more radios. Neither an idea nor a radio signal self-replicates in
the biological sense elucidated by Von Neumann involving not just uninterpreted but also
interpreted information.
The distinction between self-replication instructions and self description, or interpreted and
uninterpreted information, is reminiscent of the distinction between replicator and vehicle or
interactor, mentioned in the last chapter. However, both interpreted and uninterpreted information
are replicators. (The uninterpreted information can be thought of as replicators that say ‘here is
how you make a body’, and the interpreted information as replicators that say ‘and here is how
you make a body that can itself make a body’, whereas the vehicle or interactor is the resulting
body that can itself make a body.) Note that it is only in virtue of the fact that they come
packaged together that genes can function as replicators. Cultural entities also sometimes come
packaged together as parts of larger systems, such as minds or societies, that enable them, in a
sense, to replicate. But this happens merely through happenstance interactions amongst
individuals; they do self-replicate in the strictly biological sense of interpreting symbolically
coded instructions for how to make copies of itself.
3.1.2
Representation versus Expression of Cultural Entities
As Lake [1998] points out, symbolically coded cultural information is a representation as
opposed to an expression. For example, whereas singing a song is an expression of a musical
concept—analogous to an interactor—the symbolically coded score is a representation of
it—analogous to the uninterpreted part of a replicator. As another example, the spontaneous
verbal explanation of an idea is an expression, whereas the text version of it is a representation.
Lake comments that some cultural entities, such as village plans, constitute both a representation
of a symbolic plan, and an expression of that plan, because they are both expressed by and
transmitted through the same material form.
3.1.3
Non-replicator Transmission Results in Inheritance of Acquired Characteristics
It must be noted that neither in expression nor representation does any part of the entity act as the
interpreted part of a replicator. A musical score cannot, on its own, produce lots of little copies of
this musical score, a text cannot independently generate ‘offspring texts’, and so forth. One can
imagine a sort of molecular construction that could be understood by a musician as a musical
score and played accordingly, and that had, encoded in that score, instructions for how to piece
together molecules in the surrounding medium to generate identical or similar molecular musical
scores. One could change the ‘parent’ score, and this would accordingly change how it was
30
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
played. But unless you changed that particular part of the score that dealt with how molecules in
the surrounding medium are to get pieced together to generate a new score—in other words, the
interpreted part of the code—the next generation of musical scores would not be affected. (And
this part of the code would likely be so highly constrained that any change to it would destroy its
self-replication capacity.) Thus changes acquired during any particular generation of this score
could not be passed on; its evolution would not be Lamarckian. But this is not the case for the
musical scores with which we are familiar. In this case, it is we who do the replicating. And since
we replicate either the expression or representation (or both) of an idea, modifying it
spontaneously according to our wishes, needs and desires, there is nothing to prohibit the
Lamarckian inheritance of acquired characteristics.
Some, such as Sperber, claim that “cultural information is transformed every time it is
transmitted to such an extent that an analogy with biological reproduction or replication is
inappropriate” [Sperber 1994]. But this implies that the comparison between biology to culture
was pulled out of thin air, when as we saw in the last chapter, there are many compelling reasons
for it. Unless the transformation of cultural information during transmission hinders its
evolvability, there is no reason not to go forward with a theory that culture evolves, and to look to
biology, another evolutionary system, for guidance. (The next chapter looks at two possible
reasons why cultural information can withstand a larger degree of transformation than biological
information.) Thus the possibility that the two have enough in common that the former can pave
the way for the latter seems at least worth pursuing. Nevertheless, it is important to take seriously
Sperber’s point that cultural entities get transformed when they are expressed and assimilated.
3.1.4
Replacing Replication with Retention
In moving to the selectionist perspective, ‘replication’ was replaced with ‘retention’. The
rationale is that a stream of information can evolve by retaining pattern without explicitly
replicating it. For example, in a train of thought, pattern is retained in the sense that each thought
is a statistically similar variant of the one that preceded and prompted it. The variation can be
sequential, spread out in time rather than space, as when you try one way of solving a problem,
and then another [Heylighen 1991, 1999]. Sometimes variation is never explicitly manifested in
the physical space of the everyday world, but it is manifested in conceptual space, as when you
simulate a series of possible solutions to a problem in your mind. Still other times, it seems to be
a matter of strategically or intuitively honing in on a promising solution. In each of these cases,
some aspects of the previous instant of thought are retained, while other aspects are lost or varied.
The same is true when an idea takes shape by being bounced back and forth between two or more
31
Liane Gabora
minds: some aspects are retained, and others lost or varied. Thus there does not appear to be good
reason to bring the concept of replicators or replication into a theory of culture. Not only is there
no point in the transmission process where the idea stops and interprets instructions that are
internal to it to make a copy of itself, but mere retention accomplishes all that is needed.
3.1.5
Substrate Neutrality
The previous chapter discussed the idea that evolution is substrate neutral; that it is ‘algorithmic’,
and does not depend on the physical structure of organic life, but could work with other
underlying substrates. Given the existence of genetic algorithms, this seems to be the case.
However, it is sometimes assumed those who deny that cultural entities are replicators simply
don’t follow the substrate neutrality argument [e.g. Dennett 1991]. From the line of argument
presented above, it should be clear that this is not the case. It is possible that evolution is substrate
neutral, and that mental entities evolve, and that this is further evidence of the substrate neutrality
of evolution. But it still does not follow that mental entities are replicators. The claim for
substrate neutrality is reasonable; the claim that cultural entities are replicators is much stronger,
and unsupported.
3.2
CULTURE IS NOT LIMITED TO IMITATED ENTITIES
The concept of ‘second replicator’ has led to the additional problem of causing some to assume
that only information that replicates via imitation can evolve through culture. Here we examine
the consequences of this restriction.
3.2.1
Consequences of Excluding Individual Learning
It is commonly assumed that individual learning—learning that occurs alone through experience
in the non-social world—does not play a role in the evolution of culture [e.g. Blackmore 1998,
1999; Boyd & Richerson 1985; Laland et al. 1999]. As I and some others [e.g. Heyes & Plotkin
1989; Lake 1998] see it, the consequences of excluding individual learning from the cultural
dynamic are extremely problematic. For example, although Blackmore counts information
obtained through reading as imitated—and therefore evolvable—she claims that spatial
information is not imitated, and therefore does not evolve through culture. What do we then do
about spatial information obtained through reading, e.g. from a map?
As another example, according to Blackmore, if a child learns to peel a banana by watching
her mother, cultural transmission has taken place. But if the child learns this skill from a cartoon
character, cultural transmission has not taken place. And if the child gets the idea for how to peel
32
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
a banana by watching the petals of a flower unfold then, following through on Blackmore’s
perspective, this flower-inspired method of peeling a banana is likewise not transmittable. But
when this child shows her friend how to peel a banana, does it matter whether she acquired this
knowledge from her mother, or a cartoon character, or a flower? Why should human-imitated
knowledge be so privileged that it alone participates in the cultural dynamic? There is no reason
to believe that the jingle we learned from a friend is going to have a bigger impact on ones’
contribution to culture than the feeling of peacefulness or sense of organic artistry assimilated
during a solitary hike in the woods. Is it not true that a hike in the woods can inspire a painting or
haiku? Surely individual learning is as vital to culture as social learning. In fact there is evidence
that quite the contrary to being alternative forms of learning, the underlying mechanisms are very
similar [Bandura 1977]. Individual learning is the wellspring of cultural variation; as we will see
in the computer model in the next chapter, there has to be something worth transmitting before
social learning will even manifest itself.
In fact, strictly speaking, in the limit case of one individual, or individuals that don’t interact,
an evolution takes place through individual processes only, with no imitation. If, for example, you
were the only human left on the planet, but you could live forever, would cultural evolution grind
to a halt? If you found an ingenious way to scale a mountain, you would still have come up with
something new, something more adapted to the environment, something which you might go on
to modify and perfectto evolve. Admittedly, the ‘culture’ of a single individual would be
impoverished, to say the least. Cultural variety increases exponentially as a function of the
number of creative, interacting individuals. As a simple example, a single individual who invents
ten new words is stuck with just those ten. A society of ten interacting individuals, only one of
whom is creative, is no better off; there are still just ten words. In a society of ten nonsocial
individuals, each of whom invents ten words but does not share them, each individual still has
only ten words. In a society where the ten individuals invent ten words and share them all,
everyone ends up with a hundred words. But in a society where the ten individuals invent ten
words and not only share them all but combine them together in all sorts of new ways, the number
is much greater. The bottom line is: culture as we know it, with its explosive array of meaningful
gestures, languages, and artifacts, requires the kind of parallel processing that imitation and other
forms of social transmission provide. However, the reason imitation is important is that it plays
the role of preserving pattern, a role that as we saw in the previous section is also carried out
through the retention of information both in memory, in artifacts, and in the course of stream of
thought. Thus imitation, though important, does not play this role exclusively.
33
Liane Gabora
3.2.2
Consequences of Excluding Non-imitative Social Exchange
Blackmore also excludes non-imitative forms of social learning, such local enhancement, wherein
the activity of one individual increases the chances that another individual will learn the behavior
on its own. She also excludes contagious yawning and laughter from the cultural dynamic
because they are considered to be instances of social facilitation rather than imitation. So even if
your laughter is a husky hee haw, and your friend’s a staccato trill, and after hanging out with you
your friend takes on your husky hee haw, according to Blackmore no cultural happening has
occurred. This looks arbitrary. What if your friend goes on to invent a doll that laughs just like
you? Hasn’t the laughter that occurred in the interaction between you and your friend then had an
impact on culture? In The Meme Machine, Blackmore also claims that “perceptions and emotions
are not memes because they are ours alone and we may never pass them on.” It follows from this
view that the feeling evoked by a stormy piece of music has no relationship to what the artist was
feeling at the time... that a teacher’s attitude of compassion has no impact on the cultural
dynamics of the classroom.
Cultural transmission often occurs through imitation of conspecifics [Smith 1977; Bonner
1980; Robert 1990], or guided instruction [Vygotsky 1978; Tomasello et al. 1993] but not
necessarily. Some cultural entities, such as goulash recipes, are straightforwardly transmitted
through imitation. Others, such as, say, an attitude of racial prejudice, seem to be culturally
transmitted, but it may be difficult to point to any specific phrase or gesture through which this
transmission is mediated. Still others partake in the cultural dynamic in even subtler ways, as
when a composer releases the painful experience of his daughter’s death in a piece of music. If
you restrict a theory of cultural evolution to only those entities that are passed on through
imitation from one person to another relatively intact, such as like eye-catching fashions, or belief
in God, you run into problems. The instant you try to rope off that portion of the mind that is
culturally transmittable, you find that everything in there is fair game. Any memory or abstraction
that is food for thought is food for culture. If it can be conceived, it can find its way into the
stories, designs, and other flights of fancy that are culturally transmitted.
3.2.3
Humans are not Passive ‘Meme Hosts’
The imitation-based view of culture leads many to claim that we are mere ‘meme hosts’, passive
imitators and transmitters of memes (and some authors capitalize on the shock value of the
ensuing dismal view of what it means to be human). Memes are spoken of as if they were
retrieved from memory like a hat from a box, and thus virtually identical each time they are
expressed or assimilated. The perspective ignores the strategic, intuitive, creative, contextual
34
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
manner in which cultural novelty actually manifests. It leads one to equate the assimilation of a
meme with the ability to use it effectively. For example, Blackmore writes (arguing against a
definition of meme that includes more than explicitly imitated entities): “According to this
broadest definition a garden frog would have a mass of memes (even though it is totally incapable
of imitation or any kind of culture) because it has perceptions and emotions, and is capable of
many kinds of learning.” I agree that the garden frog does not have culture. However, this is
unlikely to have anything to do with its ability to imitate (or lack thereof). The garden frog cannot
strategically invent and manipulate ideas. Moreover, there is little selective pressure toward
evolving this capacity, because the frog has neither the vocal apparatus nor the manual dexterity
and freedom of upper limbs to implement creative ideas once they have been invented. No matter
how brilliant its ideas were, it would have difficult doing anything useful with them.
3.3
THE NEED TO MERGE TRANSMISSION STUDIES WITH STUDIES OF CREATIVITY
Perhaps the most salient shortcoming of existing models of cultural evolution is that although
they tell us something about the transmission or characteristics of ‘catchy memes’, they fail to
address the strategic, contextual manner in which cultural novelty is generated [Czikszentmihalyi
1990, 1993, 1999; Gabora 1996, 1997]. For example, although Laland et al. [1999] speak not of
cultural evolution, but of cultural change, they do not address the process by which culture
changes, just the process by which a change, once in existence, spreads to others. If an
evolutionary perspective on culture is to be of significant theoretical or predictive value, we must
give serious consideration to how experience in the world creatively transforms into new ideas in
our brains.
Studies of creativity, on the other hand, have focused almost exclusively on the individual
[e.g. Boden 1991; Lenat 1974; Schank & Cleary 1995; Schank & Leake 1989] obscuring the fact
that creativity is a collective affair. Although this research may not explicitly attempt to address
group processes it typically focuses not on the sorts of simple inferences and creative acts that a
person raised alone in the wild would be capable of, but on complex acts such as story
comprehension, that might be unlikely to develop in isolation. The ideas and inventions an
individual produces build on the ideas and inventions of others. Which ones spread and which die
out reflects the dynamics of the entire society of individuals bearing them.
Thus although at a sufficiently abstract level the notion that culture evolves is obvious, we
lack a theoretical framework that bridges transmission studies with studies of creativity, and
spells out explicitly how cultural change takes place. To accomplish this, we need more than a
quick and dirty list of what makes for a ‘catchy meme’. We have to take into account the extent to
35
Liane Gabora
which a given unit of cultural information resonates with the complex web of assumptions,
beliefs, motives, and attitudes of each person who encounters it, and what might cause it to be
evoked and expressed in the context of some situation and some need or desire. In short, we must
give serious consideration to the cognitive apparatus of the individuals that make up a society, the
contextual interactions through which social structure emerges, and the environment in which
they are trying to make their way.
3.4
BLENDING VERSUS PARTICULATE INHERITANCE
Orr [1996] voices the concern that it is pointless to speculate about a theory of cultural evolution
unless we can determine that the mode of cultural inheritance is particulate rather than blending:
Evolution would quickly grind to a halt, for instance, if inheritance were blending, not
particulate. With blending inheritance, the genetic material from two parents seamlessly blends
together like different colored paints. With particulate Mendelian inheritance, genes from Mom
and Dad remain forever distinct in Junior. This substrate problem was so acute that turn-of-thecentury biologists—all fans of blending inheritance—concluded that Darwinism just can't work.
Modern evolutionary genetics was born in 1930 when Sir Ronald Fisher cracked this problem:
Population genetics shows that particulate Mendelian inheritance saves the day. It is just the
kind of substrate needed for evolution by natural selection to work.
What, then, about Dennett’s memes—all those “tunes, ideas, catch-phrases, clothesfashions, ways of making pots or of building arches.” Do they show particulate or blending
inheritance? Do street fashion and high fashion segregate like good genes, or do they first mix
before replicating in magazines or storefronts? Does postmodern architecture reflect a blending
of the modernist and classical or the inheritance of distinct LeCorbusier and Vitruvius genes? I
do not know the answers to these questions. And neither does Dennett. And neither does anyone
else.
It is difficult to believe that Orr really means what he says here, because if cognition were
blending rather than particulate in nature, then for example the instant he conceived of the phrase
“street fashion” he would never again be able to conceive separately of the unblended concepts of
which it is composed, “street” and “fashion”. In fact, what particulate inheritance does is preserve
potential. To continue with Orr’s analogy, once white paint starts blending with black paint, it
loses the potential to ever again be completely white, and vice versa. Sooner or later all the paint
is just various shades of gray. Fortunately, although co-dominant alleles and polygenes give the
phenotypic appearance of blending, this is not in fact how genetic material is inherited.
36
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
Alternative forms, or alleles, of a gene, rather than blending together to form a new kind of allele,
retain their original structure through successive generations, as Mendel predicted. For example, a
cross of AA x aa will yield all Aa individuals, each of whom has the potential to have offspring
that are either AA, aa, or Aa, depending on whom they mate. It is as if the gray paint had the
potential to once again give rise to paint that is as purely black or as purely white as the paints of
which it was originally composed. Fortunately, as we will see in the chapters to come, the mind is
also able to preserve potential in this way. In fact, a memory is the paradigmatic example of a
potential-preserving structure. So Orr’s concern does not pose a serious problem. But he is correct
that the mode of conceptual inheritance is an issue that deserves attention.
3.5
THE GENERATION OF CULTURAL NOVELTY
We now turn to issues concerning the creative generation of novelty in culture.
3.5.1
Can Variation in Evolution be Nonrandom?
Pinker [1997] scoffs at the idea of treating culture as an evolutionary system, arguing “a complex
meme does not arise from the retention of copying errors... The value added with each iteration
comes from focusing brainpower on improving the product, not from retelling or recopying it
hundreds of thousands of times in the hope that some of the malaprops or typos will be useful.”
This is clearly true, but it is not problematic to a theory of cultural evolution unless randomness is
vital to evolvability. In the next chapter we will look at a computer model that demonstrates in
fact the opposite is true: evolution proceeds faster when variation is strategic rather than random.
Thus Pinker’s critique is not sufficient grounds to claim that culture does not evolve; it just means
that in culture randomness plays less of a role.
3.5.2
Parallel versus Heuristic Search
The difference between random and nonrandom modes of generating novelty can be
characterized in terms of parallel versus heuristic search. A biological organism produces dozens
to billions of gametes. Most don’t survive, but a few do. The occasional one is better than the
average, and over generations it tends to increase in the population. This kind of
approach—search the entire space of possibilities without devoting much effort to any one
possibility, and probably at least one of them will be better than what exists now—is referred to
in computer science as parallel search.
Human creativity, on the other hand, is highly non-random. We generate novelty
strategically, using an internal model of the relationships amongst the various elements of the
37
Liane Gabora
problem domain, and contextually, responding to the specifics of how the present situation differs
from previously encountered ones. This kind of approach—explore few possibilities, but choose
them wisely and explore them well—is referred to as heuristic search.
Effective heuristic search requires some ‘smarts’ about what sorts of transformations of the
given entity stand a good chance of being beneficial. Of course, even a relatively parallel system
like biological evolution operates with some degree of smarts, though clearly it is not strategic
and contextual the way a stream of thought is. So the distinction is just a matter of degree. The
bottom line is that a heuristic system is just as capable as a parallel system of incrementally
adapting to its environment. In a sense, culture embodies the best of both worlds; that is, each
individual’s heuristic stream of thought is embedded in a highly parallel, relatively parallel social
matrix which provides a second, outer tier of convergent pressure. Heritable variation is generated
both when an individual mulls over an idea alone, and when the individual talks it over with
others.
3.5.3
What Enables Novelty to be Strategically Generated?
The reason that we are able to generate novelty in a strategic or depth first manner is that items in
memory are woven together into an associative network, which from the first person point of
view constitutes a mental model of the world, or worldview. This associative network will be
examined in chapter five, but for now what is important is that retrieving an item from memory is
a reconstruction process—the remembered item is constructed anew from all its parts—such that
the form it takes reflects the context of the current experience that evoked it, including ones’ goals
and desires. This sort of reconstruction happens not just during reminding events but when we use
knowledge and experience to refine an idea or even just interpret the scene or situation before us.
Consequentially, concepts, ideas, and so forth, are never retrieved exactly as they were originally
assimilated. They are colored, however subtly, by what we have experienced in the meantime,
and spontaneously and creatively re-assembled in a way that relates meaningfully to the task at
hand. In this sense the idea of a cultural entity having ‘multiple parents’ seems far too simplistic,
because these ‘parents’ are themselves not discrete entities, but rather contextually elicited
aspects of an interconnected conceptual network.
3.5.4
Does Creativity Yield to Mathematical Description?
In a mathematical model, we cut out a piece of reality and say this is the entity of interest, and
these are its properties. Since it is impossible to do anything with an entity that can only exist in
one state, we also define a state space, which delineates what are the possible states given how the
38
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
properties can change, using our goals and intuitions regarding which possible changes are
important. Note that the decision concerning what are the relevant properties, and the decision
concerning how these properties can change, are as much a process of ‘cutting a piece out of
reality’ as was the decision about what to take as the entity of interest. For example, if the entity
of interest is a door, we might consider as properties the position of its centre of mass, and the
angular momentum with which it moves. Our state space then includes the various possible
combinations of positions it could occupy and momentums it could have. But it does not include
the possibility of a door that is smashed to pieces, or painted red, or being opened by Michael
Jackson.
Now, given the goal of developing a theory of cultural evolution, we need at the very least to
be able to describe the process by which a mind conceives of a new idea. For example, let us say
the state of a mind undergoes a change from imagining a door that is closed to imagining it partly
open. According to Boden’s [1990] classification of creativity, this counts as an instance of
improbabilist creativity because it explores a state space (although the examples she gives are
generally more impressive). However, during the creative process, the mind spontaneously
generates new states with new properties. The state of mind that entails thinking about a door has
the potential to change to the state of thinking about a door that is smashed to pieces, or painted
red, or being opened by Michael Jackson, or even the concept ‘door bell’. If we have never
conceived of one of these things before then, according to Boden, the act of thinking of it would
constitute an instance of impossibilist creativity because it is not just an exploration but a
transformation of the state space (although once again the examples she gives are generally more
impressive). It changes not just the state but the state space itself.
But actually, there is another state space we must consider here. Whereas conceptual space
is the state space that includes all the sensorimotor associations, memories and abstract thoughts
and ideas stored in the mind, the second state space includes only the states of mind that are
immediately accessible given the current thought. Thus it includes the current state of
mind—which is the state in which we are aware of various aspects or properties of a real or
imagined situation such as the position and colour of a door—and it also includes states that
would result from a certain number of conceivable transformations of these properties, such as the
state of considering a door of a different colour. We call this second state space working memory.
The number of properties a mind is able to consider at any given time has been shown to be
approximately seven [Miller 1956]. Thus if one is familiar with numbers, and therefore doesn’t
have to take into account the properties of numbers themselves as the relevant properties, one can
take as properties each of the slots of a seven digit phone number, and hold it in working memory
39
Liane Gabora
long enough to make a phone call. Or one can hold in working memory approximately seven
different aspects of a door, or of how a door could relate to other things. This means that the state
space of working memory concerns changes of state of seven different properties. Thus a
mathematical model of the creative process must start with a state space that is not infinitely
large, nor even very large, but limited to approximately seven properties. But then what sort of
formalism can allow in advance for the possibility that a mind in the state of considering a door
will suddenly find Michael Jackson or the concept ‘bell’ to be relevant? What sort of formalism
could describe the change of state of a mind that is in a state of worry about not being able to
knock hard enough on a door to the state of mind of inventing a door bell?
Classical physics has developed elaborate techniques for mathematically describing changes
of state. But that doesn’t help us here. In classical physics, the state of a composite or joint entity
can only be described as a product state of the states of the two subentities. If X1 is the state space
of the first subentity, and X2 the state space of the second subentity, the product state space is X1 ×
X2. Thus if the first subentity is ‘door’ and the second is ‘bell’, one can give a description of the
two at once, but they are still two, with separate properties. Classical physics cannot describe the
situation wherein the two entities give you a new entity that has all the properties of its
subentities, let alone a new entity that has only certain properties of one subentity and certain of
the properties of the other. The problem can be solved ad hoc by starting all over again with a
new state space every time an episode of impossibilist creativity occurs. But this defeats the
purpose. A complete theory of cultural evolution requires a description of, not just the exploration
of a state space that occurs in a stream of thought, but exactly those creative episodes wherein the
state space is transformed. By giving up and starting over again with a new state space every time
there appears a state that was not possible given the previous state space, we fail to include
exactly those changes of state that are most relevant to us.
Thus the mathematics that is sufficient for the description of classical entities appears to be
insufficient for a description of the change of state of mind that occurs during the creative
process. But virtually all mathematical theories in biology, cognitive science, and cultural studies
are classical theories. How can this problem be overcome? The question is vital to a theory of
culture since creativity is the process through which cultural novelty comes into being.
3.6
IS THOUGHT A DARWINIAN PROCESS?
Many of those behind the evolutionary approaches to culture discussed in the previous chapter
imply—and sometimes claim outright—that all aspects of culture (and indeed just about
anything) can be explained in terms of natural selection. Dennett [1991], for example claims
40
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
“Darwin’s dangerous idea is reductionism incarnate, promising to unite and explain just about
everything in one magnificent vision.” Richardson and Boyd [2000] are also amongst the most
optimistic: “Darwinians aim for a comprehensive theory of organic and cultural change. Our
boast is that we can model and investigate empirically any cogent proposed mechanism of
change.”
They take their cue from Campbell [1965] who believed natural selection to be the ultimate
causal process in both biological and cultural evolution. Campbell thought deeply about this
issue, and took it as far as possible given the conceptual tools that were widely available at the
time. And indeed, natural selection does seem to be a useful way of looking the more externalized
aspects of culture. It is no doubt of explanatory value when it comes to the process wherein, for
example, various sorts of peanut butter compete in the marketplace for the ‘peanut butter’ niche,
and the tastiest of them are more likely to ‘survive’. But can natural selection describe the
cognitive process by which someone came up with the idea of turning peanuts into a spreadable
substance in the first place?
Some say yes. Campbell viewed thought as a series of tiny selections. In other words, he
believed that when one hones in on a particular idea, the process of varying the idea entails a
selection amongst the possible ways of varying it. One compelling argument against the view of
thought as selections is simply that there are times when one is aware of selecting amongst
alternate possibilities, which suggests that if it happens, one is aware of it. But there are other
times when it just does not feel like one is selecting amongst alternate possibilities. It just feels
like one is honing in on or refining an idea, getting closer to the core of the thing.
Nevertheless, it is possible that we are selecting amongst alternatives that have been
actualized only at a pre-conscious level, such that we are not aware of them (Heylighen, pers.
com.). So let us examine the view more closely. The idea is that the mind is in a certain current
state, and it selects from amongst the possible successor states—which are variants of the current
state—to generate an actual successor state, and repeats this process over and over. We call p(t0)
the state of the mind at t0. Let us refer to the set of possible successor states at t1 as P(t1), and the
actual state at t1 we call p(t1). Thus p(t1) ∈ P(t1). In order to select amongst the P(t1) possible
successor states, it must be possible to distinguish them from one another, because selection
theory, as it has been formulated by biologists, works on distinct entities with different traits. This
is true no matter how miniscule the transformation that causes the change of state from p(t0) to
p(t1). To be able to distinguish them, it is necessary to carry out the transformation from p(t0) to
the first element of P(t1), followed by the inverse transformation back to p(t0) so that the process
41
Liane Gabora
can be repeated for the second element of P(t1), and so forth. Thus one simulates or vicariously
experiences various possibilities before committing to any one of them.
Note that it isn’t possible to carry out all these transformations because P(t1) is infinitely
large, not just in the sense of the relevant properties being continuous variables, but in the sense
of there being an infinite number of them. More concretely, let us say that p(t0) is the thought ‘It
is raining’. P(t1) then includes everything from ‘I should take my umbrella to work’ to ‘I like the
sound of rain falling on the roof’ to ‘I wonder why earthworms come out in the rain’. There are
infinitely many ways of transforming any given thought.
Thus Campbell says that one tries out a few blindly chosen elements of P(t1), and then
selects one of them to be p(t1). But let us say that a mind is in state p(t0) which corresponds to the
thought ‘It is raining’, and then tries out just the three possibilities for p(t1) enumerated above
starting with ‘I should take my umbrella to work’. The state of the mind after considering this
first possibility cannot return to a state that is identical to p(t0). The closest the state can come is
the thought ‘It is raining’ in the context of having considered as a possible subsequent thought ‘I
should take my umbrella to work’. Neither can the selective pressure return to what it was. The
closest the selective pressure can come is ‘what is the best though to have next having considered
as a possibility ‘I should take my umbrella to work’? So now we face a problem: only the first
blindly generated possibility is truly an element of P(t1), and more importantly, only it underwent
the original selection criterion. In other words, the selective pressure itself is changed by states of
mind it has an impact on. Thus, as far as selection theory is concerned, there is never more than
one possibility to select from.
Another way variation and selection could be applied to a stream of thought is to say that
one generates a variant of the current thought and then selects from only two possibilities: keep it
or discard it. This is an idiosyncratic use of the term ‘selection’, but let’s see what happens. Say
the state of your mind is p(t0) and you generate a variant of p(t0) which we call p(t1). You decide
that the variant thought p(t1) is no good and you want to discard it. According to this version of
selection theory, you go back to the previous thought, p(t0), and start again. But as above, you
cannot truly go back to the same state p(t0), nor can you truly go back to the same selection
criterion. (And even if you could, then given the state of mind p(t0), the next thing one does is go
to p(t1), which would take you back to p(t0)... You are thus trapped in an endless loop.)
The problem is that you don’t immediately discard all aspects of a thought, and neither do
you keep all aspects. Now this is also true in biology—never do all traits die out (until organic life
comes to an end) and never are all traits retained, yet selection theory does fine there. But the
crucial difference is that there the selection process is happening in parallel on actualized entities
42
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
rather than serially on potential ones, so a selection event that affects one entity generally does
not have to be taken into account when considering a selection event that affects another entity.
The selection pressure can be assumed to be the same.
3.6.1
Selection Theory works with Distinct, Actualized States
Thus the problem here boils down to the fact that selection theory—both as originally formulated
by biologists and as modified for culture—is only capable of dealing with distinct, separate,
actualized states. It cannot cope with potential states. Potentiality is not a concept that comes to
mind so readily when one contemplates biological evolution, because selection appears to act
primarily upon entities that have already been explicitly manifested, or reproduced. One is not
immediately forced to think in terms of which of the potential offspring a parent is capable of
having actually do get born, although it is becoming increasingly clear that selection does also act
at this level. In culture, however, one is more readily forced to think in terms of potentiality. The
hub of culture, the place where cultural novelty is generated and assimilated is, of course, the
human mind, and the human mind regularly cultivates and weeds out ideas that have never been
explicitly materialized, and even acts on the basis of fears and intuitions that have never been
explicitly thought about.
All this is just an attempt to clarify what is left to do, rather than to downplay the importance
of Darwinian selection, which cannot be underestimated. As noted earlier in this chapter, there are
probably occasions where one even consciously considers one possible solution to a problem after
another, and then chooses what seems to be the best solution. In this particular situation, since
they were actualized in conceptual space, it may be possible to consider them all at once
simultaneously as discrete entities, not in their details but in symbolic form, and evaluate them
simultaneously. In this case, selection theory may generate an approximate description. But
selection theory can only describe this special situation. It is particularly unsuited to the
description of states of mind that are unfocused, or context-dependent, or where the various
possible subsequent states are inter-mixed, not yet cleanly distinguishable one from the other. As
far as selection theory goes, it is only possible to predict how the state of an entity will change by
enumerating possible future states—stating in advance before they have actualized how they
would actualize—and choosing amongst them. In chapter eleven we will look at a formalism that
overcomes this limitation.
43
Liane Gabora
3.6.2
Selection Theory Cannot Describe Intrinsic Contextuality
There is an even subtler problem with the Darwinian approach when states of potentiality are
involved. The problem is that enumerating some blindly chosen possible future states of an entity
and saying that p(t1) is one or another of them, does not give us an ontologically accurate
description of its state if there is any degree of contextuality present. To many this would not
matter if the Darwinian approach can nevertheless predict future states as quickly and accurately
as any alternative approach. However, this is not the case.
Take the following example that Diederik Aerts and I came up with in a conversation of
these matters. Natalie is driving to the election booth, in a state of indecision as to whom she will
vote for. We refer to the state of her mind as p(t0). In a first scenario, a reckless driver almost runs
into her, and her state of mind collapses to the decision to vote for Herbert, the candidate who has
taken a strong stance to make the roads safer. In a second scenario, she is stopped and given a
speeding ticket, and her state of mind collapses to the decision to vote for the alternative
candidate, Beverlee, who has argued that too much money is spent on the police force. Now we
cannot say that the state p(t0) her mind was in when she began the trip to the election booth had as
a true property ‘I will vote for Herbert’. However, we also cannot say that it had as a true property
‘I will not vote for Herbert’. Because of the contextual nature of the situation, both were
potentially true, and both were potentially false.
As another example if someone were to ask you in a hostile tone of voice ‘Are you angry?’
you might answer (and believe) ‘yes!’, whereas if asked the same question in a sympathetic voice
you might answer (and believe) ‘no’. Again, before you were asked the question, both were
potentially true, and both were potentially false; it isn’t a ‘one or the other is true’ situation. The
opinion actualized in the process of being measured.
The existence of such contextual states of mind must be treated as an indication that in order
to accurately describe cognition and the cultural phenomena it gives rise to we will need to look
further afield than the classical descriptions that have until now been used to explain these
phenomena. For situations described by classical physics, for a given proposition A it is always
the case that either A is true or NOT A is true. Either one or the other must be true. However, it is
interesting (and further on will become relevant) to note that in situations described by quantum
mechanics, it is possible for there to be propositions where A could be true, and NOT A could be
true, and also it could be that neither one of them is true. This is, in fact, the essential difference
between situations that can be described by classical logic and situations where quantum logic is
needed [for further analysis see Aerts et. al. 2000b]. One might think that such strange contextdependent states are restricted to the mysterious micro world, but the above examples are
44
Chapter 3: Unresolved Issues and Potentially Fruitful Directions
amongst an increasing wealth of evidence (discussed in chapters eleven and thirteen) that this is
not so. The highly contextual nature of human experience suggests that quantum structure is not a
mysterious thing confined to the micro world, but rather, part of ordinary, everyday life. Music,
for example, can be viewed as a context that sees whether you have the potential within you to be
moved emotionally by a pattern with which the composer expressed something he or she was
moved by.
3.7
SUMMARY
This chapter examines the current unresolved issues and problems that stand in the way of
viewing culture as an evolutionary process. Although all are deserving of attention, in no case is
the problem insurmountable.
The concept of ‘second replicator’ is misleading because cultural information does not
interpret instructions that constitute part of itself to create new copies of itself; it just impinges on
minds like a radio signal received by multiple radios. That is, although cultural entities may be
seen to constitute a self-description—uninterpreted code—they lack instructions for how to selfreplicate—interpreted code (in the von Neurmann sense). It is we who do the interpreting, not the
entities themselves, and in so doing, we can change them however we wish. This is why acquired
characteristics can be inherited. The ‘second replicator’ notion has had the unfortunate
consequence of leading some to assume that only ideas transmitted through imitation participate
in culture. By excluding emotions, individual learning, and information acquired through other
forms of social exchange, one is forced to some strange conclusions, such as that a haiku inspired
by a walk in the woods does not enter into culture. Rather than viewing cultural entities as
replicators, one can say that in interactions amongst individuals, and as well in a stream of
thought, some aspects are retained or preserved, and others lost or varied. It is because a memory
is an excellent potential-preserving device that it, like biology, can be characterized as a form of
inheritance that is particulate rather than blending. The structure of human memory is particularly
apt in this regard because its contents are associatively and hierarchically connected into an
internal model of the world.
Heritable cultural novelty is generated and assimilated spontaneously in a strategic, intuitive
manner, as opposed to the parallel, random manner in which biological novelty is generated.
Although this does not interfere with the ‘evolvability’ of culture, it has not been seriously
addressed by models of cultural evolution. This is partly because of the difficulty of
mathematically describing what has been referred to as impossibilist creativity, which involves a
45
Liane Gabora
change of state that doesn’t just explore but transforms the given state space. However, for a full
theory of how culture evolves, we need a description of exactly this.
One can attempt to describe cognition in biological terms as the internal variation and
selection of mental entities. However, this only works if all potential mental entities can be
uniquely distinguished prior to their being actualized (otherwise there is no basis for them to be
selected amongst). Sometimes this is the case; we consider different ways of seeing or doing
something one after another and then select the best. But often we just refine an idea, taking it
further and further, and here Darwinian approaches do not work. They are appropriate for entities
that are actualized and distinguishable one from another, but not appropriate to the description of
a change of the state of the mind when the mental entity that constitutes the subject of thought is
unfocused, inter-mixed, contextual, or potential rather than actual. In fact, no formalism rooted in
classical mechanics can explain these phenomena. Unfortunately virtually all theories in the life
sciences are of this type, so we are left with a challenge.
46
Chapter Four
4
A Computer Model of Cultural Evolution
Holland’s [1975] genetic algorithm is a minimal computer model of natural selection that made it
possible to investigate the effect of manipulating specific parameters on the evolutionary process.
If culture is, like biology, a form of evolution, it should be possible to similarly abstract the
underlying skeleton of the process and develop a minimal model of it. Meme and Variations, or
MAV, is a computational model, inspired by the genetic algorithm, of how ideas evolve in a
society of interacting individuals [Gabora 1995]. The name is a pun on the classical music form
‘theme and variations’, because it is based on the premise that novel ideas are variations of old
ones; they result from tweaking or combining existing ideas in new ways [Holland et al. 1981].
MAV explores the impact of several phenomena that are unique to culture. These are introduced
briefly here, and the technical details of how they are implemented will be presented shortly.
The first is knowledge-based operators. Brains detect regularity and build schemas with
which they adapt the mental equivalents of mutation and recombination to the constraints of their
world, and the situation at hand. Thus they generate novelty strategically, on the basis of past
experience. Knowledge-based operators are a crude attempt to incorporate this into the model.
The second crudely implemented cultural phenomenon is imitation. Ideas for how to perform
actions spread when agents copy what one another is doing. This enables them to share solutions
to the problems they face.
The third is mental simulation; agents can ‘imagine’, or guess, how successful an idea would
be if it were implemented before they actually commit to implementing it. This provides them
with a rudimentary form of selection before the phenotypic expression of an idea. Once again, the
way this phenomenon is implemented is not terribly life-like, but the goal here was to abstract the
essence of the phenomenon and see how it affects the evolution of culture.
47
Liane Gabora
Every iteration, each neural-network based agent in an artificial society has the opportunity
to acquire a new idea, either through 1) innovation, by changing a previously learned idea, or 2)
imitation, by copying what a neighbor is doing. Thus, variation occurs through mutation of preexisting ideas, selection occurs through choice of which pre-existing idea to mutate, and how to
mutate it, and ideas spread through imitation.
4.1
THE MODEL
Since the model is artificially limited with respect to the number and kinds of features an idea can
have, it does not hurt in this case to adopt the terminology of biology. Thus the features or
components of an idea are referred to as loci, and alternative forms of a locus are referred to as
alleles. The processes that generate variation are referred to as mutation operators. Forward
mutation is mutation away from the original (or, as biologists refer to it, the wild type) allele, and
backmutation is mutation from an alternative form back to the original. An individual is referred
to as an agent, and the set of all agents is referred to as the society.
4.1.1
The Domain
Donald [1991] has provided substantial evidence that the earliest culture took the form of
physical actions, such as displays of aggression or submission. The ideas in MAV can be thought
of as mating displays. An idea is a pattern consisting of six loci that dictate the degree of
movement for six body parts: left arm, right arm, left leg, right leg, head, and tail. Each locus has
a floating point activation between -0.5 and 0.5 which determines the amount of movement (angle
of rotation from rest position) of the corresponding body part when the idea is implemented. A
value of 0.0 corresponds to rest position; values above 0.0 correspond to upward movement, and
values below 0.0 correspond to downward movement. Floating point loci activations produce
graded limb movement. However, for the purpose of mutation, loci are treated as if there are only
three possible alleles at each locus: stationary, up, and down. Six loci with three possible alleles
each gives a total of 729 possible ideas.
4.1.2
The Neural Network
The neural network is an autoassociator; it learns the identity function between input and output
patterns. It has six input/output units numbered 1 through 6, corresponding to the six body parts.
It has six hidden units numbered 7 through 12, corresponding to the general concepts, “arms”,
“legs”, “left”, “right”, “movement”, and “symmetry” (Figure 4.1).
48
Chapter 4: A Computer Model of Cultural Evolution
Figure 4.1 The neural network. Arrows represent connections with positive weights. For clarity, negative connections and
connections to the symmetry unit are not shown.
Hidden units are linked with positive weights to input/output units that are positive instances of
the concepts they represent, and linked with negative weights to input/output units that represent
negative instances of the ideas they represent (thus “left arm” is positively linked to “left” and
negatively linked to “right”). Hidden units that represent opposite concepts have negative
connections between them. The hidden units enable the network to encode the semantic structure
of an idea, and their activations are used to bias the generation of variation.
The neural network starts with small random weights between input/output nodes. Weights
between hidden nodes, and weights between hidden nodes and input/output nodes, are fixed at +/1.0. Patterns (representing ideas) are learned by training for 50 iterations using the generalized
delta rule [Rumelhart et. al. 1986] with a sigmoid activation function [Hinton 1981]. The basic
idea is that patterns are repeatedly presented to the network, and each time, the network’s actual
output is compared to the desired output. Since this neural network is an autoassociator, we desire
simply that the output be identical to the input. The relevant variables are:
aj = activation of unit j
49
Liane Gabora
ai = activation of unit i that contributes to activation of unit j
tj = the jth component of the target pattern (the external input)
wij = weight on link from unit i to unit j
= 0.15
= 0.5
1
_________________________________
1+e-
aj =
[∑i wij ai+
]
(4.1)
For the movement node, we use the absolute value of ai (since negative movement is not possible;
the least you can move is to not move at all).
The comparison between input and output involves computing an error term, which is used
to modify the pattern of connectivity in the network such that its responses become more correct.
The error signal, δj, is calculated such that the more intermediate (and thus ‘uncommitted’) the
activation of the unit, the more it is affected. For input/output units it is computed as follows:
j=
(t j - a j)a j (1 - a j)
(4.2)
For hidden units we do the same thing by determining how much each unit is contributing to the
error as follows:
i=
a j (1 – a j )∑ i
4.1.3
j wij
(4.3)
The Embodiment
The embodiment is a six-digit array specifying the behavior of the six body parts. While the
output of the neural network represents what the agent is thinking about, the embodiment
represents what it is actually doing. An idea cannot be observed and imitated by other agents until
it has been copied from the neural network to the embodiment and is implemented as an action.
4.1.4
The Fitness Function
An optimal action is one where all body parts except the head are moving, and limb movement is
anti-symmetrical. (Thus if the left arm is moving up, the right arm is moving down, and vice
versa.) This is implemented as follows:
50
Chapter 4: A Computer Model of Cultural Evolution
F = fitness
= 2.5
am = activation of movement hidden node
as = activation of symmetry hidden node
ah = activation of head node
i = 1 if ah = 0.0, otherwise i = 0
F=
a m + 2 as +
i
This fitness function corresponds to a relatively realistic display, but it also has some interesting
properties. An agent that develops the general rule “movement improves fitness” risks
overgeneralization since head stability contributes as much to fitness as movement at every other
limb. This creates a situation that is the cultural analog of overdominance in genetics; the optimal
value of this locus lies midway between the two extremes. We also have a situation analogous to
bidirectional selection or underdominance; the optimal value of the tail locus lies at either of the
two extremes. (The desire to investigate underdominance was the reason for giving the agents
tails). There is a cultural analog of epistasis—where the fitness at one locus depends on which
allele is present at another locus. Epistasis is present because the value of what one limb is doing
depends on what its counterpart is doing; for example, if the left leg is moving backward the right
should be moving forward, and vice versa. Finally, since there is one optimal allele for the head,
two optimal alleles for the tail, two optimal arm combinations, and two optimal leg combinations,
we have a total of eight different optimal actions. This enables us to perform a comparative
analysis of diversity under different ratios of creation to imitation.
4.1.5
Using Experience to Bias the Generation of Novelty
The idea here is to translate knowledge acquired during evaluation of an action into educated
guesses about what increases fitness. Each locus starts out with the allele for no movement, and
with an equal probability of mutating to each of the other two alleles (those for upward and
downward movement). A new action is not learned unless it is fitter than the currentlyimplemented action, so we use the difference between the two to bias the direction of mutation.
51
Liane Gabora
Two rules of thumb are used. The first rule is: if the fitter action codes for more movement,
increase the probability of forward mutation and decrease the probability of back mutation. Do
the opposite if the fitter action codes for less movement. This rule of thumb is based on the
assumption that movement in general (regardless of which particular body part is moving) can be
beneficial or detrimental. This seems like a useful generalization since movement of any body
part uses energy and increases the likelihood of being detected. It is implemented as follows:
am1 = activation of movement unit for currently-implemented action
am2 = activation of movement unit for new action
p(fmut)i = probability of forward mutation at allele i (increased movement)
p(bmut)i = probability of backward mutation at allele i (decreased movement)
IF (am2 > am1)
THEN p(fmut)i = MAX(1.0, p(fmut)i + 0.1)
ELSE IF (am2 < am1)
THEN p(fmut)i = MIN(0.0, p(fmut)i - 0.1)
p(bmut)i = 1 - p(fmut)i
The second rule of thumb biases the agent either toward or away from symmetrical limb
movement. It has two parts. First, if in the fitter action both members of one pair of limbs are
moving either up or down, increase the probability that you will do the same with the other pair of
limbs. Second, if in the fitter action, one member of a pair of limbs is moving in one direction and
its counterpart is moving in the opposite direction, increase the probability that you will do the
same with the other pair of limbs. This generalization is also biologically useful, since many
beneficial behaviors (walking, etc.) entail movement of limbs in opposite directions, while others
(galloping, etc.) entail movement of limbs in the same direction. The implementation of this rule
is analogous to that of the first rule.
In summary, each action is associated with a measure of its effectiveness, and
generalizations about what seems to work and what does not are translated into guidelines that
specify the behavior of the cultural algorithm.
52
Chapter 4: A Computer Model of Cultural Evolution
4.2
PROTOCOL
Agents are in a two-dimensional wrap-around 10x10 grid-cell world, one agent per cell. Each
iteration, every agent has the opportunity to (1) acquire an idea for a new action, either by
imitating a neighbor, or creating one anew (2) update the mutation operator, and (3) implement
the new action.
Agents have an equal probability of creating and imitating. To invent or create a new idea,
the cultural algorithm is applied to the idea currently represented on the input/output layer of the
neural network. For each locus, the agent decides whether mutation will take place. The
probability of mutation is specified globally at the beginning of a run. If it decides to mutate, the
direction of mutation is stochastically determined. If the new idea has a higher fitness than the
currently-implemented idea, the agent learns and implements the action specified by that idea.
To acquire an idea through imitation, an agent randomly chooses one of its eight neighbors
and evaluates the fitness of the action the neighbor is implementing. If its own action is fitter than
that of the neighbor, it chooses another neighbor, until it has either observed all eight neighbors,
or found one with a fitter action. If no fitter action is found, the agent does nothing. Otherwise,
the neighbor’s action is copied to the input/output layer, and it is learned and implemented.
Since in both creation and imitation, a new idea is not acquired unless it is fitter than the
currently implemented action, the new idea provides information that is used by the cultural
algorithm. For example, since we arbitrarily chose a fitness function in which movement is
generally beneficial, if the new action does code for more movement than the old one, the
probability of forward mutation will almost always increase.
No matter how the new idea has been acquired, it gets implemented as an action when it is
copied from the neural network to the embodiment. In the ‘no mental simulation’ condition,
whether the new idea was acquired through creation or imitation, it must be implemented as an
action for at least one iteration before its fitness can be assessed. In this case, mutation operators
are updated the following iteration.
4.3
RESULTS
The following experiments were conducted using a mutation rate of 0.17 per locus, a 1:1 creation
to imitation ratio, and all cultural evolution strategies operative, unless otherwise indicated.
4.3.1
Outline of a Run: Culture Evolves
Initially all agents were immobile, thus the number of different actions implemented was zero, as
shown in Figure 4.2. The immobility idea quickly mutated to a new idea that coded for movement
53
Liane Gabora
of a body part. This new idea had a higher fitness and was preferentially implemented. As ideas
continued to be created, get implemented as actions, and spread through imitation, the society
evolved toward increasingly fit actions.
Figure 4.2. Effect of varying p(create) to p(imitate) ratio on average fitness of an idea. Result shown is for one run at each
p(create) to p(imitate) ratio. The diversity of actions increases rapidly, and then decreases as the society stabilizes on
fitter actions. The greater the ratio of p(create) to p(imitate), the greater the diversity throughout the run. These results
were robust across runs.
4.3.2
Trade-off Between Diversity and Global Optimization
Figure 4.2 shows how diversity peaked when the first maximally fit idea was found, and
decreased as the society converged on maximally-fit ideas. As in a genetic algorithm, increasing
the frequency of variation-inducing operations—in this case, the creativity to imitation
ratio—increased diversity. This was true both as the society was evolving, and when it finally
stabilized. An interesting result can be seen if one looks more closely at how diversity varied with
the creation to imitation ratio. Diversity ranged from 1-2 actions when p(create) = 0.25, to 10-11
actions when p(create) = 1.0. We know that when p(create) = 1.0, some agents did not find an
54
Chapter 4: A Computer Model of Cultural Evolution
optimal action, because the diversity is greater than 8, the number of optimal actions. However,
when p(create) = 0.75, the society converged on 7-8 actions. Thus it found all (or nearly all) of
the fittest actions. A nice balance was thereby struck between the diversifying effect of creation
and the converging effect of imitation.
4.3.3
Frequency of Change Must be Intermediate
Agents could vary with respect to not only just the frequency with which they invented, but with
respect to how much change they introduced when they did. As in a genetic algorithm, evolution
did not occur in the complete absence of mutation. The best performance was obtained with
between 0.07 and 0.22 mutations per locus (Figure 4.3). In biological terms this would constitute
a very high mutation rate; however because there are so few loci it ends up being approximately
one mutation per innovation event, which is very reasonable. There are however two reasons why
it can afford to be high. The first is that mental simulation ensures that unfit ideas are not
implemented as actions. The second is that fit actions are imitated by others, thus never lost from
the society as a whole.
55
Liane Gabora
Figure 4.3 Optimization time decreases sharply, and then increases slowly, as mutation rate increases. This trend holds
true for both the mean fitness of all ideas and the fittest idea that has appeared in a given iteration. Result shown is for
one run (with 100 agents) at each mutation rate, and the result was robust across runs.
Note that this is similar to what happens in a genetic algorithm when the mutation rate is much
above the minimum necessary for evolution. In both MAV and the GA, the frequency of change
must be intermediate between the two extremes.
4.3.4
Epistasis Decreases Rate of Fitness Increase
As in biology, fitness increased more slowly, and stabilization took longer, for epistatically linked
loci than for either over- or underdominant loci Figures 4.4 and 4.5 show this at two different
mutation rates; in fact, it was observed at every mutation rate tested. In figure 4.4, we see that
whereas the over-dominant locus had stabilized by the 100th iteraction, and the under-dominant
by the 150th, the two epistatic loci took 200 iterations to stabilize. In figure 4.5 whereas the
overdominant locus stabilizes by the 70th iteration, and the under-dominant by approximately
iteration 130, the two epistatic loci never do manage to stabilize. The phenomenon can be
attributed to the fact that for epistatically linked loci there are more constraints to be met; what
one arm should be doing, for example, depends on what the other arm is doing. It is also
interesting to note that the epistatic loci to some extent mirror one another, such that if the mean
activation for one locus increases, the mean activation for the epistatically linked locus decreases,
and vice versa.
56
Chapter 4: A Computer Model of Cultural Evolution
Figure 4.4 Mean loci activations with a mutation rate of 0.01 for one run. Over-dominant locus stabilizes most quickly,
followed by under-dominant locus, and then epistatic loci. This result is robust across runs.
4.3.5
Cultural Drift
Since we have eight optimal ideas, there are many stable configurations for the distribution of
ideas. Figures 4.4 and 4.5 reveal amongst equally-fit alleles the presence of drift (which, recall
from chapter two, refers to changes in the relative frequencies of different alleles due to random
sampling processes in a finite population). Drift is indicated by the fact that since we are looking
at mean activation values across the entire society, if the activation value is very high or very low
that means that almost all agents had stabilized on the same value at a particular locus. This is the
case for both the over-dominant and epistatic loci. (For the under-dominant locus it is not possible
to distinguish between all individuals stabilized on the intermediate value, or as many stabilized
above this value as below.) This is in accord with Cavalli-Sforza and Feldman’s [1981]
mathematical model of culture.
57
Liane Gabora
Figure 4.5 Mean loci activations with a mutation rate of 0.67 for one run. Over-dominant locus stabilizes most quickly.
Under-dominant locus takes longer. After 200 iterations, epistatic loci still have not stabilized. Result is robust across runs.
4.3.6
Effect of Knowledge-based Operators, Imitation, and Mental Simulation
The three cultural evolution strategies—knowledge-based operators, imitation, and mental
simulation—were made inoperative one at a time to determine their contribution to optimization
speed and peak mean fitness. These experiments were performed separately, and also
cumulatively, adding each strategy one at a time. Since, the results were comparable for the
separate and cumulative experiments, only the cumulative results are presented here (Figure 4.6).
58
Chapter 4: A Computer Model of Cultural Evolution
Figure 4.6 Cumulative improvement with successively applied cultural strategies. MS refers to mental simulation. ‘Imit’
refers to imitation. ‘Kmut’ refers to knowledge based operators. Result shown is for one run at , but the increase in fitness
with each added cultural strategy was robust across runs.
All three cultural evolution strategies increased the rate of optimization. Mental simulation and
imitation also increased peak mean fitness.
4.3.7
Fittest Society with Creation to Imitation Ratio of 2:1
The highest mean fitness was achieved when both creation and imitation were employed, as
illustrated in Figure 4.7. The best performance was observed when the creativity to imitation ratio
was either 1:1 or 3:1. The society with a 3:1 ratio improved most quickly until the 21st iteration,
at a mean fitness of 8.6. The society with a 1:1 ratio then overtook it and converged entirely on
optimal actions slightly earlier than the 3:1 society (32 iterations as opposed to 37). Thus it can be
said that overall the optimal creativity to imitation ratio is most likely midway between these
values, or approximately 2:1. The 1:3 ratio society took longer to converge, though it did
eventually after 47 iterations. The society that just created and never imitated had the opposite
59
Liane Gabora
problem. It did as well as the 1:1 and 3:1 ratio societies for the first ten iterations or so, at which a
mean fitness of 6.2 was attained. However, after that, its performance dropped compared to the
others, such that even after 100 generations it never did converge on optimal solutions. As might
be expected, when there was no creativity, just imitation, then there was no evolution at all; the
agents simply remained immobile.
Figure 4.7 Effect of varying the creation-to-imitation ratio for one run at each creation-to-imitation ratio. The optimum
seems to be between 0.5 and 0.75. This result is robust across runs.
However, it is interesting to note that the fitness of the fittest idea (Figure 4.8) increased as a
function of the ratio of creation to imitation. Since the agents with the fittest ideas gain nothing by
imitating others, there is a trade-off between average action fitness, and fitness of the fittest
action. (Of course, this result should not be taken too seriously as indicating that smart people
don’t need to imitate, since the agents in MAV only had one problem to solve. Thus, those who
happened to be lucky simply got a head start.)
60
Chapter 4: A Computer Model of Cultural Evolution
Figure 4.8 The fitness of the fittest idea increased as a function of the p(create) versus p(imitate) ratio. Result shown is for
one run for each creation-to-imitation ratio. Result was robust across runs.
4.4
COMPARISON WITH OTHER APPROACHES
The computational approach taken here allows us to look for patterns that arise over time when
the cultural activities of inventing and imitating are carried out in parallel in a society of
interacting individuals. To my knowledge, MAV is the first computational model of the evolution
of culture as opposed to the effect of culture on biological evolution. For example, Hutchins and
Hazelhurst [1992] used a computer model to explore the relationship between environment,
internal representation of the environment, and cultural artifacts that mediate the transmission of
knowledge about environmental regularity vertically across generations. In MAV, on the other
hand, we stick to one generation so that the effects of cultural evolution can be disentangled from
biological evolution. Similarly, computer models of the evolution of creativity [Sims 1991; Todd
& Latham 1992] and cooperation [Axelrod 1985], although they explore a cultural process, they
use a genetic algorithm—a model of biological evolution. Axelrod’s work has inspired others,
61
Liane Gabora
who have taken a more culturally realistic approach [e.g. Conte & Paolucci forthcoming; Hales
forthcoming], but in these studies the space of possible cultural entities is too small to evolve, so
the study is really a study of how particular cooperative strategies diffuse across a population of
agents. Note, however, that although MAV is modeled after cultural evolution, even it is too
simple to explore many cultural phenomena. The space of possible ideas is fixed and small, and
(unlike real life) the fitness function is predetermined and static. It would be particularly
interesting to explore the effects of a dynamically changing fitness function given recent findings
that maintaining diversity in a population is, over the long term, more important than global
fitness in the short term [Hutter 2001]. Another shortcoming of MAV, imitation and innovation
are not as discrete in real life as MAV would suggest. Despite these shortcomings, however,
MAV demonstrates the feasibility of computationally modeling the processes by which creative
ideas spread through a society giving rise to observable patterns of cultural diversity.
4.5
SUMMARY
If culture, like biology, is a form of evolution, it should be possible to abstract the underlying
skeleton of the process and develop a minimal model of it analogous to the genetic algorithm.
MAV is a minimal computer model of the process by which culture evolves. It consists of an
artificial society of neural network-based agents that don’t have genomes, and neither die nor
have offspring, but they invent, imitate, and implement ideas, and thereby their actions gradually
become more fit. Every iteration, each neural-network based agent in an artificial society has the
opportunity to acquire a new idea, either through 1) innovation, by mutating a previously learned
idea, or 2) imitation, by copying an action performed by a neighbor.
The program exhibits features observed in biology such as: (1) drift, (2) epistasis increases
time to reach equilibrium, (3) increasing the frequency of innovative or variation-generating
operations increases diversity, and (4) although in the absence of variation-generating operations,
culture does not evolve, increasing innovation much beyond the minimum necessary for evolution
causes average fitness to decrease.
The model also addresses the evolutionary consequences of phenomena specific to culture.
Imitation, mental simulation (ability to assess the relative fitness of an action before actually
implementing it), and strategic (as opposed to random) generation of variation all increase the rate
at which fitter actions evolve. The higher the ratio of innovation to imitation, the greater the
diversity, and the higher the fitness of the fittest action. Interestingly, however, for the society as
62
Chapter 4: A Computer Model of Cultural Evolution
a whole, the optimal innovation-to-imitation ratio was approximately 2:1 (but diversity is then
compromised). For the agent with the fittest behavior, the less it imitated (i.e. the more
computational effort reserved for innovation), the better.
63
Chapter Five
5
Mind: The Culture Evolving Architecture
Although sufficient to evolve a primitive sort of culture, the ‘minds’ of the agents in Meme and
Variations were clearly not much like our own. An understanding of human cognition is vital to
an understanding of culture, since culture evolves through minds processing their own contents
and making contact with either something, or someone, in the world. This chapter takes a brief
look at the architecture of the mind, and how it generates a stream of thought that may be
culturally expressed through word or action. Thus, the material presented here serves the purpose
of providing a common turf for further developments.
5.1
ELUCIDATING WHAT EVOLVES IN MIND AND CULTURE
Chapter one introduced the distinction between mental entities, which manifest internally in
conceptual space, and cultural entities, which are the external manifestation of mental entities in
physical space. We also distinguished amongst hardwired instincts, memories encoded through
direct experience of the physical world, and more abstract mental entities such as concepts and
attitudes, referred to collectively as abstractions. It is necessary to make a few more distinctions.
5.1.1
Episodic, Semantic, and Procedural Memories
We have differentiated between two kinds of mental entities: memories, which are records of
direct experience in the physical world, and abstractions, which are formed through abstract
thinking, fantasizing, or other sorts of processing of memories or other abstractions. Memories
can be further distinguished as episodic, which are records or concrete episodes, semantic, which
are records of knowledge about the world, and procedural, which are records of how to do
something.
65
Liane Gabora
Some [e.g. Donald 1991] suggest that artifacts function as yet a fourth, externalized or
extrasomatic form of memory. One could take this idea even further and say that if, say, a diary
or tape recording can function as an externalized form of memory, then perhaps we each function
as an externalized form of memory for each other. For example, a child might be tempted to eat a
second chocolate bar, but then the look on his mother’s face reminds him of how sick he felt the
last time he did that.
5.1.2
Classical, Prototype, and Exemplar Theories of Concepts
The term abstraction encompasses not just concepts but also other sorts of higher-order abstract
cognitive entities such as attitudes, prejudices, stories, and so forth. However, in cognitive
science, most of the research has been concentrated on concepts. It was long thought that for each
concept there exists a set of defining features that are singly necessary and jointly sufficient [e.g.,
Sutcliffe, 1993]. This has come to be called the classical theory of concepts. Extensive evidence
has been provided against it [e.g. see Komatsu 1992, and Smith & Medin 1981, for overviews].
Two major alternatives have been put forth (each with many variants). According to the prototype
theory [Rosch, 1975a, 1978, 1983; Rosch & Mervis, 1975], concepts are represented by a set of,
not defining, but characteristic features, which are weighted in the definition of the prototype. A
new item is categorized as an instance of the concept if it is sufficiently similar to this prototype.
According to the exemplar theory, [e.g., Heit & Barsalou 1996, Komatsu 1992, Reed 1972, Rosch
1975, Rosch & Mervis 1975] a concept is represented by, not defining or characteristic features,
but a set of particular instances of it stored in memory. A new item is categorized as an instance
of the concept if it is sufficiently similar to one or more of these previously-encountered
instances. We use the term representational theories to refer to both prototype and exemplar
theories since concepts take the form of fixed representations (as opposed to varying dynamically
according to context). Although the concept B that follows concept A is sometimes referred to as
an instantiation of concept A, it is good to keep in mind that many sorts of relationships between
A and B are possible. For example, not only could B be an instance of A, but A could be an
instance of B, or A could simply remind one of B.
5.2
THE STRUCTURE OF MENTAL ENTITIES
Blackmore worries that in individual learning and non-imitative forms of social learning “the
details of the first behaviour are not transmitted and therefore cannot be built upon and refined by
further selective copying. In this sense, then, there is no true heredity.” We have argued that there
is no copying going on, at least in the sense that there is copying in biological evolution.
66
Chapter 5: Mind: The Culture Evolving Architecture
Nevertheless, her point would still be valid, except that it is only true if we look at what is
happening at a very low resolution. However, culturally transmitted information contains many
levels of structure, one embedded within another. All that is necessary for evolution to take place
is that one level of structure remains intact. So, for example, say Mary directs John’s attention
toward the water fountain, and John takes a drink. This is an example of stimulus enhancement, a
non-imitative form of social learning. The way John drinks the water will indeed not be
influenced by the way Mary drinks water. But when you look at what has happened in the context
of an ongoing stream of experience and transmittable behavior, a cultural event has taken place.
The ‘how to drink from a fountain’ meme may not be evolving, but the schema for ‘what to do
when you walk down a hallway’ may well be. Now that John knows the fountain exists, he may
start sipping water there every time he passes it, and others may well follow suit. The bottom line
is that cultural entities don’t simply replicate or not; some aspects may proliferate, and others not.
Thus, in order to talk about how it is that they evolve, it is necessary to say something about their
structure.
5.2.1
Properties, Features, and Dimensions
Biologists use the term ‘allele’ to capture the notion of alternative heritable versions of a gene,
and Durham [1991] accordingly adopted the term ‘allomeme’ to refer to alternative versions of a
idea. This basic concept was tailored to meet the constraints of biology; we all have the same
number of genes, and two alleles of each gene (one from each parent). Durham’s cultural analog
is perhaps too clumsy to capture the subtle relationships amongst concepts, which are stored in a
distributed fashion, connected through webs of association [Hebb 1949; Quillian 1968; Pribram
1994]. There is not necessarily a definitive rationale for saying where one stops and another
begins, in conceptual space let alone physical space. For example would we consider ‘My mother
looks good in blue’ and ‘My mother looks good under a blue umbrella’ to be allomental entities
of the same idea, or different ones altogether?
This kind of difficulty is circumvented by avoiding the notion of alternate versions
altogether, and thinking instead in terms of properties. The relevant properties in cognition are
commonly referred to as features (for qualitative differences) or dimensions (for quantitative
differences).10 Thus related concepts share features, or the same dimensions are relevant to them,
and they may take on similar values with respect to these dimensions.
10
This use of the word ‘dimension’ follows the psychological literature on concepts [e.g. Smith & Medin
1981].
67
Liane Gabora
5.2.2
Chunking and Categorization
Frequently many concepts get integrated into one through a process referred to in the
psychological literature as chunking [Miller 1956]. Chunking involves forming associations
amongst previously learned concepts and establishing this constellation of associations as a new
entity in long term memory; it is analogous to the formation of coadapted genes, or schemata
[Holland 1975].
Whereas chunking generally refers to the binding of semantically unrelated concepts (as in
the memorization of an arbitrary string of numbers), categorization involves the recognition of
shared properties through the formation of associations. This sometimes results in the emergence
of an abstract concept that consists of these shared properties. The abstract concept thus generally
has fewer properties than its instances. For example, the abstract concept ‘depth’ has fewer
properties than ‘deep blue sea’. ‘Depth’ is deeply woven throughout the matrix of concepts that
constitute one’s worldview. It is latent in concepts as varied as ‘deep swimming pool’, ‘deep-fried
zucchini’, and ‘deeply moving book’. An abstraction is not something that one experienced at a
particular time in the concrete world, but it captures the gist or common essence of many such
concrete experiences.
Abstractions are grounded in perceptual experience [Harnad 1990; Steels 1996], and retain
something of the “perceptual character” of the experiences of which they were derived [Barsalou
1999]. However, to be grounded in experience does not mean to spend eternity underground. Just
as a seed transforms into a sprout which emerges from the ground as a plant, an abstract concept
can originate from, and be grounded in, perceptual experiences, yet turn into something quite
different from anything ever directly perceived. Thus, what was once a constellation of memories
of similar experiences organizes itself into an entity whose structure resides primarily at a level
that was not present in the constituents from which it was derived.
5.2.3
Hierarchical and Dynamical Structure
Mental entities consist not just of sets of relevant features or dimensions, but hierarchically
embedded layers of them, and it is the covariations amongst these features or dimensions—the
way they are related to one another—that gives them their hierarchical structure. This basic
insight has been approached from many directions, and the nested structures have variously
described as schemas [Head 1920; Bartlett 1932; Evans 1967; Evans, Hoffman, Arnoult & Zinser
1968], dependence systems [Rescher & Oppenheim 1955; Latimer & Stevens 1997], schematic
maps [Hochberg 1969], frames [Minsky 1975], or hyperstructures [Richardson 1999].
68
Chapter 5: Mind: The Culture Evolving Architecture
Note that static stimuli elicit no response; unless there is a change, nothing registers
[Mackay, 1986]. From a dynamical perspective, what is of interest is changes in the schemas
(also referred to as dependence systems, or hyperstructures as mentioned in the preceding section)
that carry sensory, cognitive, or cultural information. The importance of taking a dynamical
perspective on cognition is increasingly appreciated [Brooks 1991; Combs 1996; Combs et. al. in
press; Edelman 1992; Freeman & Skarda 1990; Orsucci 1998; Port et. al. 1995; Roszak et. al.
1995; Sheets-Johnstone 1999; Thelen & Smith 1994; van Gelder 1995, 1998; van Gelder & Port
1995; Varela et al. 1991]. (Thus features most likely take the form of nested phase relations, and
a stream of cognitive information is perhaps best viewed as a spatiotemporal wavefront of these
nested phase relations, where different kinds of information are carried by different frequencies,
like a radio broadcast system [Cariani 1995, 1997].) Edmonds [1999] therefore suggests that we
think in terms of, not concepts, but conceptual pathways. For simplicity I will stick to the term
concepts, but adopt his point that it should be kept in mind that they are dynamical entities.
5.3
CONCEPTUAL SPACE
Having briefly explored the structure of mental entities and kinds of mental entities that are
possible, we can begin to explore how they are related to one another, and to the cognitive
architecture that evolves them. To this end, it is useful to consider the properties that differentiate
different mental entities as defining a state space referred to as conceptual space. We will also
consider how conceptual space relates to and is realized in physical space by the architecture of
the mind.
5.3.1
Sparse
The first thing to note is that conceptual space is extremely sparse. The number of properties
along which mental entities can differ is related to the number of distinctions our sensory
apparatus can register. Where n is the number of features the senses can distinguish, N, the
number of mental entities that could potentially be stored = 2n for Boolean variables (and it is
infinitely large for continuous variables). For example, if n =1,000, N = 2 1,000 mental entities (or
even more if we assume that the mind rarely if ever attends all the stimulus properties it is
capable of detecting) 11. Since assuming n is large, N is enormous, so the memory is sparse in that
the number of locations L where mental entities can be stored is only a small fraction of the N
perceivable mental entities. The number of different mental entities actually stored at a given
11
The size of this number is perhaps better appreciated when we realize that it is of the same order of
magnitude as 10300.
69
Liane Gabora
time, s, is constrained by L, as well as by the variety of perceptual experience, and the fact that
retrieval, though distributed at the storage end, is serial at the awareness end. That is, the rate at
which streams of thought reorganize the cognitive network is limited by the fact that everything is
funneled through an awareness/attention mechanism; we can only figure one thing out at a time.
Note that a ‘retrieval’ can be reminding, a redescription of something in light of new contextual
information, or a creative blend or reconstruction of many stored mental entities.
The set of all possible n-dimensional mental entities a mind is capable of storing can be
represented as the set of vertices (if features assume only binary values) or points (if features
assume continuous values) in an n-dimensional hypercube, where the s stored mental entities
occupy some subset of these points. The distance between two points in this space is a measure of
how dissimilar they are, referred to as the Hamming distance. Kanerva [1988] makes some astute
observations about this space (which are mentioned here since they are relevant to how
conceptual space changes when abstractions emerge, which will be the focus of considerable
attention in the forthcoming chapters). The number of mental entities at Hamming distance d
away from any given mental entity is equal to the binomial coefficient of n and d, which is well
approximated by a Gaussian distribution. Thus consider a certain mental entity which we will call
X, and which for simplicity we represent as 111...1. Consider also its antipode, 000...0. We
consider mental entity X and its antipode to be the ‘poles’ of the hypersphere, then approximately
68% of the other mental entities lie within one standard deviation (sqrt[n]) of the ‘equator’ region
between these two extremes (Figure 5.1). As we move through state space away from the equator
toward either Mental entity X or its antipode, the probability of encountering a memory or
abstraction falls off sharply by the proportion sqrt[n]/n.
70
Chapter 5: Mind: The Culture Evolving Architecture
Figure 5.1 Distribution of the Hamming distances12 from address of a given item to addresses of other items in a sparse
memory. The Gaussian distribution arises because there are more ways of sharing an intermediate number of features
than of being extremely similar or different. A computer stores each item in left-most address only. A fully distributed
network stores it throughout. A constrained distributed activation function such as the radial basis function is intermediate
between these extremes. Activation decreases with distance from the ideal address, as indicated by darker shading.
In fact, the space of possibilities is even larger given that the mind rarely if ever attends all the
stimulus properties it is capable of detecting. Therefore, the number of properties paid attention
to, n, is only a subset of the maximum, M; i.e. n ≤ M. Since the memory can now store mental
entities of any length up to M, the number of possible mental entities is:
N = 2M+1 - 2
2M+1
(5.1)
The bottom line is: the memory would probably have to be larger than the number of particles in
the universe to store all the permutations of sensory stimuli it is capable of registering. It is
therefore sparse.
5.3.2
Constrained Distribution of Memories and the Edge of Chaos
Now let us examine the ingenious way in which memories occupy conceptual space.
5.3.2.1
Distributed Storage
If the mind stored each experience in just one memory location as a computer does, and as shown
in Figure 5.2a, then in order for one experience to evoke a reminding of a previous experience, it
would have to be identical to that previous experience. And the memory is so sparse and the
space of possible experiences so vast, that no two ever are exactly identical. So this kind of
organization would be fairly useless.
In connectionist networks, this problem is solved by distributing the storage of a mental
entity across many locations. Likewise, each location participates in the storage of many mental
entities. Input/output nodes correspond roughly to the sensory apparatus, memory locations are
represented as hidden nodes, and their pattern of connectivity as weighted links. An input touches
off a pattern of activation which spreads through the network until it relaxes into a stable
configuration, or achieves the desired input-output mapping using a learning algorithm. The
output vector is then determined through linear summation of weighted inputs.
12
Hamming distance refers to the number of digit positions in which the corresponding digits of two binary
words of the same length are different. (Thus the Hamming distance between 101 and 111 is 1.)
71
Liane Gabora
Memory
Memory
Input
Memory
Input
Input
(a)
(b)
(c)
Figure 5.2 (a) A one-to-one correspondence between input and memory, as in a computer. (b) A distributed memory, as in
some neural networks. (c) A constrained distributed memory, as in neural networks that use a radial basis function for the
activation function. This is closest to how human memory works.
Thus a retrieved mental entity is not activated from a dormant state, but ‘reconstructed’. A
reconstructive approach is necessary if we aim to model cognition at a fine-grained level of
resolution—down to the threshold of human discrimination.13 It enables the connectionist
memory to abstract a prototype, fill in missing features of a noisy or incomplete pattern, or create
a new mental entity on the fly that is more appropriate to the situation than any mental entity it
has actually experienced [Rumelhart & McClelland 1986]. For example, if an autoassociative
network has been fed vectors in which feature one is present whenever feature two is present, and
vice versa, it will respond to an input that lacks information about feature one—which, again for
simplicity, is represented as *101—by generating 1101. It may never have actually encountered
1101 before, but given that in its ‘world’ there exists a correlation between features one and two,
this is an appropriate response. In addition to associations between inputs and outputs of features,
the network has learned a higher-level association between two features. In effect, it contains
more information than has been explicitly fed into it. New features emerged.
5.3.2.2
Constraining the Distribution
Having seen the benefit of distributing memories across many locations rather than storing each
in a separate location, we now consider the other extreme. If, on the other hand, every experience
activates every memory location as in Figure 5.2b, the memory is subject to crosstalk, a
phenomenon wherein nonorthogonal patterns interfere. This can be solved by constraining the
storage region as in Figure 5.2c. This is sometimes done in neural networks using a radial basis
function (RBF) [Hancock et al., 1991; Holden & Niranjan, 1997; Lu et al. 1997; Willshaw &
Dayan, 1990]. Each input activates a hypersphere of memory locations, such that activation is
13
As Philip Polk points out, this threshold has been lowered by the invention of things like microscopes
and telescopes.
72
Chapter 5: Mind: The Culture Evolving Architecture
maximal at the center k of the RBF and tapers off in all directions according to a (usually)
Gaussian distribution of width σ (as in Figure 5.3; this also corresponds to the white zone in
Figure 5.1). This enables one part of the network to be modified without affecting the capacity of
other parts to store other patterns. Where F is the activation, x is an i-dimensional input vector,
and k is the center of the RBF, hidden nodes are activated as follows:
F ( x) = e
(( xi − ki ) / ) 2
−
(5.2)
i
The further a stored concept is from k, the less activation it not only receives from the stimulus
input but in turn contributes to the stimulus output, and the more likely its contribution is
cancelled out by that of other simultaneously evoked locations. A wide σ is said to model the
situation where neurons have a low activation threshold, so more of them fire in response to a
given stimulus. Note though that given the enormous complexity of the brain it is not possible (at
this time, at least) to precisely pinpoint which physiological mechanism underlies an increase or
decrease of the activation threshold in real brains. Suffice it to say that many different
mechanisms could be involved. Thus, in general it is appropriate to think of the activation
threshold as applying to events in conceptual space, and how they are carried out in physical
space is a subject that will require further investigation.
In neural networks, suitable values for k and σ are found during a training phase. In the brain
the requisite tuning of patterns of neuron interconnectivity is probably achieved through selforganizing feedback processes [Edelman, 1987; Pribram 1994].
Memory location at
center k for current
stimulus
A barely activated memory
Location
Memory location not
activated by stimulus
Figure 5.3 Schematic diagram of stimulus activating two dimensions of memory space. Each vertex represents a possible
memory location, and black-ringed dots represent actual locations. If many memories are stored in locations near k, they
blend to generate the next experience.
73
Liane Gabora
Now let us turn to the distributed representation of abstract concepts in conceptual space.
Consider the segment of conceptual space schematically represented as a hypercube14 with four
properties in Figure 5.4. The four properties are ‘ALIVE’, ‘SNOW-COVERED, ‘HAS EYES’
and ‘HAS ROOTS’. For simplicity, they take on the value 1 if present in a given concept, 0 if not
present, and * if irrelevant to the concept. Thus a concept for which all four are relevant will
occupy a vertex of this space, and concepts for which between one and three are relevant will
occupy a more diffuse region of the space. Two concepts are represented as vertices of this
space—‘dead palm tree’ and ‘arctic pine’. The abstraction ‘tree’ is also implicit in it.
x3 HAS
EYES
x2
SNOW-
COVERED
1101 Arctic Pine
**01
Tree
x4 HAS
0001 Dead
ROOTS
Palm Tree
x1 ALIVE
Figure 5.4: A hypercube that schematically represents a segment of conceptual space. ‘ALIVE’, ‘SNOW-COVERED,
‘EYES’ and ‘ROOTS’ lie on x1, x 2, x 3, and x4 axes respectively. Two concepts are stored here: ‘dead palm tree’ and ‘arctic
pine’. Black-ringed dots represent centers of distributed regions where they are stored. Fuzzy white region indicates space
activated by ‘arctic pine’. Emergence of abstract concept ‘tree’ implicit in the conceptual space (darkened square region)
made possible by constrained distribution of activation.
Note that ‘tree’ does not just activate instances like ‘dead palm tree’, arctic pine’, and so forth; it
derives its very existence from them. Likewise, once ‘arctic pine’ has been identified as an
instance of ‘tree’, it is thereafter affected, however subtly, by experiences that activate ‘tree’.
Notice also that whereas each instance occupies merely a point, the abstraction occupies the
whole central square region of the portion of conceptual space shown. Abstractions tend to
occupy more of conceptual space than instances, because they refer to all entities with certain
salient properties, no matter what other properties they might have.
14
A hypercube is simply a cube with more than three dimensions.
74
Chapter 5: Mind: The Culture Evolving Architecture
5.3.2.3
The Edge of Chaos
The degree to which the parts of an information-evolving system are correlated or causally
connected falls squarely in a narrow regime between order and chaos [Langton, 1992]. Another
way of expressing this is to say that the mutual information—that is, the amount of information
that can be gleaned about one component of a system by examining another component—must be
intermediate. Consider how this principle manifests in biological systems. Organic polymers such
as protein and RNA molecules commonly act as ingredients or catalysts15 of some chemical
reactions, and products of others. The more reactions catalyzed per polymer, the more
interconnected the system. If each enzyme catalyzed only one reaction, the specificity of the
system would be so high that once it worked there would be no room for improvement. However,
if each enzyme catalyzed each reaction equally well, the system would be completely unstable.
The actual situation is that, much as different keys sometimes open the same door but require
different amounts of effort, reactions can be catalyzed by many catalysts, with varying degrees of
efficiency. Kauffman [1993] found that living systems, and boolean network models of them,
attain the delicate ‘edge of chaos’ state of intermediate connectedness without top-down control
through spontaneous self-organization.
The implications of the edge of chaos principle for cognition is clear. If memories were
distributed so narrowly (that is, stored in such tiny portions of the mind) that their storage regions
never overlapped, the current experience would have to be identical to a previously-stored one to
evoke it. However, if they were too widely distributed, successive thoughts would not necessarily
be meaningfully related to one another. The free-association of a schizophrenic seems to
correspond to what one might expect of a system like this [Weisberg, 1986]. For the mind to be
capable of evolving a stream of coherent yet potentially creative thought, the degree to which an
experience is distributed must lie between these extremes; that is, the size of the sphere of
activated memory locations must fall within an intermediate range. A given instant of experience
activates not just one location in memory, nor does it activate every location to an equal degree,
but activation is distributed across many memory locations, with degree of activation decreasing
with distance from k, the one that is, at that instant, the most activated one (Figure 5.2c and
Figure 5.3). The result of achieving the delicate edge-of-chaos state is that sequential ‘slices’ of a
stream of experience are self-similar, that is, correlated but non-identical; each is a variation of its
predecessor. Self-similarity is enhanced by the fact that the motivational state generally does not
change unless the present goal is satisfied, or another becomes more pressing, and also by the
15
Catalysts speed up chemical reactions that would otherwise occur very slowly.
75
Liane Gabora
high degree of continuity in our percepts (for instance, if a window is to your left now it is likely
to still be there now).
5.4
THE ARCHITECTURE THAT REALIZES CONCEPTUAL SPACE
Conceptual space is an abstraction, like the use of a pie chart to depict business expenses. In order
to understand how the mind generates a stream of thought, we now need to examine some
attributes of the cognitive architecture that realizes conceptual space.
5.4.1
Integrating Abstractions, Memories, Stimuli, and Drives
Conceptual space is inconceivably vast, but as noted in chapter three, we are only able to hold
approximately seven items in mind at any one time [Miller 1956]. Of course, when we engage in
chunking or categorization (section 5.2.2) we squeeze in more by considering multiple properties
as one. But we are still limited in what we can consider. So, vital to cognition is a means of
integrating sensations, drives, and stored memories and abstractions—including their emotional
aspects—to produce a seamless stream of conscious experience and purposeful motor action16.
Thus there needs to be a way for the demands of the current situation to make contact with those
regions of conceptual space that are potentially most relevant. We will use the terminology of
cognitive science and use the term working memory, introduced in chapter three, to indicate a
temporary storage buffer for information that is the immediate subject of attention. Note,
however, that a similar construct is used in the consciousness literature, where the locus of where
stimuli, drives, and so forth are integrated and conscious experience emerges has been referred to
in various ways, such as Cartesian theatre or global workspace [Baars 1988].
So let us think in terms of a state, which is a state of the mind at any one time, which
consists of approximately seven properties, and the state space of working memory as being
defined by the most obvious ways of changing these properties. Note that any formalism with
which we could describe the dynamics of the state of the mind must allow for the possibility of
not only transcending these most obvious sorts of change, but in fact of changing to states with
different properties altogether.
Donald [1991] has pointed out that, especially in modern human culture, working memory is
virtually always supplemented by, and works in interaction with, material artifacts, particularly
external symbolic media such as computers. This increases the complexity of what we can
mentally cope with or manipulate at one time. So let us say that the number of properties is
16
The place where this information is coordinated need not correspond to a single anatomical structure,
though it is often suggested that the intralaminar nuclei of the thalamus are involved.
76
Chapter 5: Mind: The Culture Evolving Architecture
somewhat greater than seven. It is still, however, exceedingly small compared to the number of
properties of the entire state space that we refer to as conceptual space.
5.4.2
Content Addressability
There is a systematic relationship between the content of an experience (not just as the subject
matter, but the qualitative feel of it) and the memory locations where it gets stored (and from
which material for the next instant of experience is evoked). In a computer, this kind of one-toone correspondence is accomplished by assigning each possible input a unique address in
memory. Retrieval is thus simply a matter of looking at the address in the address register and
fetching the item at the specified location. The distributed nature of human memory prohibits this
kind of one-to-one correspondence, but content addressability is still achievable as follows. The
pattern of features (or phase relations) that constitutes a given experience induces a chain reaction
wherein some neurons are inhibited and others excited. The address of a neuron is therefore the
pattern of excitatory and inhibitory synapses that make it fire. Thus there is a systematic
relationship between the information content and the locations it activates. Since correlated
patterns get stored in overlapping locations, the system appears to retrieve experiences that are
similar, or abstractions that are relevant, to the current experience. As a result, the entire memory
does not have to be searched in order for, for example, one painting to remind you of another.
Thus each memory location is particularly responsive to some stimulus property, and more
likely to become involved in the storage and retrieval of experiences with this property. This
stimulus property may be something we would recognize, such as size or color, or it may not be
something we are likely to ever consciously conceive of, such as, perhaps, the property of being a
word that simultaneously evokes feelings of joy and reminds one of snow. Thus there will be
overlap in the properties of memory traces that get stored in a particular location, though for any
given location, the different memories stored there will also have other properties that are quite
different. So if one specifies the memory location of interest, one cannot specify the particular
memory trace, because many memory traces are stored in each location. However if one specifies
the particular memory trace of interest, one cannot specify the particular memory location
involved, because the memory trace will be stored, in a graded fashion, in all the locations
responsive to its various salient properties.
To make this more concrete, say you hear a meow sound coming from a box but you don’t
know which cat it is, Glimmer the white one, or Inkling the black one. (This in fact happened.) If
you want to specify a particular memory location that gets activated at this instant, say one that is
particularly responsive to meow sounds, you find that there are many different memories stored
77
Liane Gabora
there, some that involve meow sounds made by Glimmer, others that involve meow sounds made
by Inkling, and still others that involve meow sounds made by other cats, or by people pretending
to be cats, and so forth. On the other hand, if you want to specify what memory will be evoked by
this experience then many different locations will be involved. If this experience of the meow
sound coming from the box reminds you of an event wherein you heard a meow sound in the next
room and it turned out to be Glimmer, this particular memory trace is stored in and gets evoked
through the activation of many different memory locations. In other words, there is a kind of
uncertainty relationship between the property of interest and the location in memory. (Of course,
to be completely certain of this, we might want to do a ‘cat’ scan. :- )
5.4.3
Organized Modularity
We mentioned that cross-talk can be avoided by constraining the region in which any experience
is stored. Another way of avoiding cross-talk in a neural network is to induce a division of labor
amongst competing subnetworks; in other words, to make the memory functionally encapsulated,
or modular [Jacobs et al. 1991; Jordan & Jacobs 1995]. There is abundant evidence of modularity
in the brain, and its preservation in phylogenetic history suggests that it is not arbitrary. We
assume that (1) the world we live in is highly patterned and redundant, and (2) this pattern and
redundancy is reflected in the connectivity of the neurons where mental entities are stored. After
birth there is large-scale pruning of neurons. It seems reasonable that the surviving subset of the
M possible inputs to each neural pathway is determined by biological and cultural selective
pressures, instead of at random. These pressures sculpt the pattern of neuron connectivity such
that the L (out of N possible) locations can store most of the mental entities we stand a chance of
encountering. This means that the probability a given stimulus activates a retrieval event is not as
low as the statistics suggest.
5.4.4
Habituation
We do not want an ongoing stimulus, such as the sound of rain, to recursively evoke remindings
of rain. The nervous system avoids this kind of perseveration as follows. First, neurons have a
refractory period during which they cannot fire, or their response is greatly attenuated. Second,
they ‘team play’; the responsibility for producing a response is shared by a cooperative group of
neurons such that when one is refractory another is active. If exactly the same neurons are
stimulated repeatedly, they all become refractory, and there is little or no response. This
phenomenon is referred to as habituation.
78
Chapter 5: Mind: The Culture Evolving Architecture
5.5
GENERATING A STREAM OF THOUGHT AND CULTURALLY EXPRESSING IT
We now lay out how a cognitive architecture that can implement a conceptual space with the
above attributes generates a stream of thought, and expresses it culturally.
5.5.1
An Instant of Experience
Before we finally get to what happens in a stream of thought, let us look briefly at what happens
during a mundane instant of experience. We refer to the information about the world that hits the
senses at a particular instant of time as the stimulus. The integration of stimulus, current drives
and goals, and memories and concepts derived from the worldview, generates an experience—an
impression, observation, interpretation, judgement, emotional reaction, or assessment of some
kind. Thus the specific quality of this experience emerges through an interaction between the
worldview, and the world itself as perceived at that instant. Sometimes this actualizes a thought or
feeling that manifests in conceptual space only. Sometimes, however, a thought or feeling is
expressed behaviorally through the spoken word, song, action, and so forth, and thereby gets
observably actualized in physical space as well. Actions influence culture by affecting other
individuals directly, as in dance or displays of affection or aggression, or indirectly, via artifacts
such as buildings or art.
5.5.2
The Continuity of a Stream of Experience
The current instant of experience touches off a cascade of activation, which leads to the
distributed storage of that experience in the memory locations of particular neurons. This process
of storing to these locations activates retrieval from these locations. Thus whatever else is stored
in those locations merges with salient information from senses and drives to elicit the next instant
of experience. And so forth. This process is just information flowing through a system displaced
from equilibrium. One’s conceptual network is in an ongoing process of responding and adapting
to the world around it, and these changes in turn affect the impact it has back on the world.
Because memory is content addressable—correlated stimuli and ideas get stored in
overlapping locations—what emerges is that the system appears to retrieve experiences that are
similar, or concepts that are relevant, to the current experience. Since sequentially evoked mental
entities tend to be correlated, i.e. they have a greater-than-zero temporal correlation length,
statistical similarity tends to be preserved in a stream of thought. This gives a stream of thought
continuity; it has a unifying effect on memory dynamics that counters the diversifying effect of
sensory novelty. Unlike biology, in a stream of thought, the convergence and divergence
79
Liane Gabora
operations are not spatiotemporally separated but intimately intertwined, one pattern of qualia
fluidly transmuting into the next depending on the specifics of the context in which it arises.
It is because consecutive instants of experience are not exact replicas—each is a variant of
its predecessor—that, as mentioned previously, some have viewed thought as a form of selfreplication. It could be argued that the correlation between consecutive concepts is so low that
this hardly deserves to be called a form of self-replication; one wouldn’t want consecutive
concepts to be identical. Surely Eigen and Schuster’s [1979] error catastrophe argument applies
here; that is, the ‘copying fidelity’ of this process is so low that errors would quickly accumulate
and in no time the ‘lineage’ would ‘die’. But this argument doesn’t apply. The only reason it is a
problem for biological evolution is that copying error tends to impair the capacity to selfreplicate. So long as offspring are as good as their parents at reproducing themselves, and live
long enough to do so, it doesn’t matter how much error is introduced from one generation to the
next. It is only when a generation dies without having reproduced that there is a problem. In the
biological world, once something is dead it can not spring forth life. But in cultural evolution this
isn’t necessarily the case. To show why this is so, say that half way through the train of
consecutive concepts in Einstein’s brain that culminated in the theory of relativity, a tiger burst in
through the window. The correlation between the relativity concept of one instant and the tigerperception-concept of the next instant would be almost zero. Since bodily protection is higher on
the hierarchy of needs at that instant than the need to continue with the theory of relativity, his
momentous conceptual lineage would come to a screeching halt. But would it be lost forever? No.
Sooner or later, once the tiger situation was taken care of, the relativity stream of thought would
inevitably resume itself. Memory (and external artifacts) function as a ‘cultural sperm bank’ of
sorts, allowing a defunct ancestral line to be brought back to life and resume self-replication. The
upshot is that in culture you can get away with a higher error rate than in biology.
5.5.3
Retrieval as Reconstruction
It should be pointed out that memory is put to use even in those streams of experience where it is
not obvious that one is recalling a previous episode, simply to accomplish the interpretation of the
situation, to frame it in terms of what has been experienced before. Experience is thus a
construction that pieces together the present situation with past experiences and instincts.
Even the recollection of a previous episode is a construction event. Heraclitus said ‘You
never step into the same stream twice’, and this applies to streams of experience as well as
streams of water. At a high enough level of resolution, it is never the exact same memory or
concept conjured up time and again; your understanding of it is always colored by, and
80
Chapter 5: Mind: The Culture Evolving Architecture
reinterpreted in the context of, events that have taken place since the last time you thought of it,
and your current goals or desires. For example, right now I am recalling how, last night as I was
working on this, my cat Inkling fell asleep on my lap. Tomorrow I may retrieve the ‘same’
memory. But today it is colored by today’s mood, today’s events; tomorrow it will experienced
slightly differently.
In fact, what is ‘retrieved’ may never have been explicitly stored. Next month I might
wrongly ‘remember’ it as having been my other cat, Glimmer, perhaps because I will blend this
memory with a memory of Glimmer walking on my keyboard. The mind does not pull items from
memory like mitts from a box, but creatively weaves external stimuli with stored experiences and
abstractions relevant to the current motivational state. But though the ‘retrieved’ mental entity is a
reconstruction, something never actually experienced, it can still be said to have been evoked
from memory. (It’s like getting a ‘bite’ on many fishing rods at once, and when you reel them in
you get a fish that is a mixture of the characteristics of the various fish that bit.)
5.5.4
A Stream of Thought as Representational Redescription
In abstract thought, the previous stimulus recursively feeds back on itself and gets re-experienced
in the context of the changes to the conceptual network that it just evoked. The lower the
activation threshold, the higher the probability that the content of successive drafts will be
different from one another. This continual re-experiencing and re-storing of slightly modified
forms of an idea in a stream of thought is referred to by Karmiloff-Smith as representational
redescription [1992]. The net effect is that the idea gets rooted in the network of understandings
that comprise the worldview, and the worldview is perpetually revised as new experiences are
assimilated. In the next chapter we will look at how this process sometimes generates not just
modified versions of an idea, but strikingly new ideas.
5.5.5
Symbol Manipulation
The connectionist methods described in this chapter are examples of the subsymbolic approach to
cognition, which works best for modeling perceptual and low-level cognitive phenomena. These
include detecting, representing, and responding flexibly to patterns of correlation, learning fuzzy
categories, and solving simple constraint satisfaction problems.
Symbolic models of cognition focus on the serial and potentially recursive application of
logical operations on symbols, without attempting to represent their internal structure. They are
particularly good at modeling the high-level cognitive abilities that characterize modern human
minds, such as planning and deductive reasoning. Arguments for a reconstructive view of
81
Liane Gabora
retrieval notwithstanding, highly abstract concepts that have been used millions of times, such as
‘space’ or ‘different’ or ‘is’, would be unlikely to emerge from memory retaining the associations
of any particular usage. Thus it seems reasonable to begin with the working hypothesis that
subsymbolic processing predominates for low-level, parallel, automatically-generated cognitive
phenomena, and that symbolic processing provides a satisfactory approximation for many highlevel, serial, consciously-directed aspects of cognition. Neurophysiological evidence suggests that
the manipulation of abstractions—such as creating new sets of associations which make them
relevant in new contexts—involves the synchronization of features encoded by distributed
cortical cell assemblies [Klimesch 1995].
5.5.6
Embodiment and Cultural Expression
It is increasingly pointed out that cognitive science has focused too exclusively on relatively
abstract aspects of cognition, and ignored that minds are embodied; that they learn and change
through the physical interactions between bodies and the world, and that these bodies are
themselves culturally embedded [Sheets-Johnstone 1999; Thelen & Smith 1994; Varela et al.
1999]. It is worth noting that, much as the study of culture can benefit through increased attention
to cognition, the study of cognition benefits through a cultural perspective by forcing us to keep
in mind that it is only through the realization or outward expression of thoughts and ideas that
they exert an effect.
5.6
SUMMARY
This chapter examined the nature of cognition, since it is the expression of mental entities as
vocalizations or actions of which culture is constituted. Mental entities can be loosely categorized
as memories or abstractions. Memories can be episodic, semantic, or procedural. Most of the
work on abstractions has been done on concepts. The classical theory of concepts—according to
which there exists for each concept a set of defining properties that are singly necessary and
jointly sufficient—has been superceded by two alternatives. According to the prototype theory,
concepts are represented by a set of, not defining, but characteristic properties, which are
weighted in the definition of the prototype. According to the exemplar theory, a concept is
represented by, not defining or characteristic properties, but a set of particular instances of it
stored in memory. All of these approaches can be criticized as ignoring the dynamical and
contextual aspects of how mental entities are activated in response to a given stimulus (and in
chapter thirteen we will explore a new approach that better takes these aspects into account).
82
Chapter 5: Mind: The Culture Evolving Architecture
Their storage in memory is distributed (but not fully distributed), and content-addressable. It can
be useful to think of all distinguishable features and dimensions as defining a conceptual space.
The mind is somewhat modular, habituates to repeated inputs, and brings together memories
and stored abstractions, stimuli, emotions and drives to construct a continuous stream of
experience. Even retrieval of an episode from memory is a construction event. Some experiences
elicit spontaneous behavioral expression; other times the subject of the experience is first
recursively redescribed or manipulated in a stream of thought.
83
Liane Gabora
the
P82
84
Chapter Six
6
Creativity and Cultural Novelty
Whereas the goal of the previous chapter was to present the material necessary to explain how the
mind generates a stream of associative thought, the primary goal of this chapter is to explain how
such a stream of thought generates something new.
6.1
ATTRIBUTES OF THE CREATIVE MIND
This morning while reading the Ottawa Citizen I saw a crafty yet bafflingly incompetent vandal
described as “He’s got a full six pack, but the plastic thingy that holds them together is missing”.
It’s a spin-off of the saying “He’s one can short of a full six pack”, which itself is Canadianized
version of “He’s lost a few marbles” or “He’s not playing with a full deck”. Not only does the
newspaper description beautifully exemplify one of the main issues of this chapter—the interplay
of change and continuity as a creative insight is adapted from one context or circumstance to
another. But content-wise, it’s a pithy summary of another, related issue dealt with here: in order
to adapt an idea to a new context, in order to evolve it in new directions, it must originally have
been stored in memory in a way that implicitly identifies its relationships to other ideas. In other
words, when it comes to creativity, how your ‘beer cans’ are connected together is as important as
how many of them there are.
This section examines the attributes of minds that are particularly creative.
6.1.1
Defocused Attention, Sensitivity, and Flat Association Hierarchies
Martindale [1999] has identified a cluster of attributes associated with creativity which includes:
defocused attention [Dewing & Battye 1971; Dykes & McGhie 1976; Mendelsohn 1976], high
sensitivity [Armstrong 1974; Martindale 1977], flat associative hierarchies [Mednick 1962], and
sensitivity to subliminal impressions; that is, stimuli that are perceived but of which we are not
85
Liane Gabora
conscious of having perceived [Smith & Van de Meer 1994]. The steepness of an individual’s
associative hierarchy is measured experimentally by comparing the number of words that
individual generates in response to stimulus words on a word association test. Those who
generate only a few words in response to the stimulus have a steep associative hierarchy, whereas
those who generate many have a flat associative hierarchy.
6.1.2
Relationship of Creativity Attributes to Activation Threshold
Recall that the storage of memories is distributed across many memory locations, with degree of
activation falling with distance from the most activated one, and that the degree of distribution is
determined by the how high the activation threshold is. It seems sensible that the more stimulus
features one attends, or is sensitive to, the more memory locations the current instant of
experience gets stored to; that is, the lower the activation threshold. Therefore, the more memory
locations from which ‘ingredients’ for the next instant of experience can be drawn. Thus the
greater the likelihood of ‘catching’ a concept (or portion of one) that isn’t usually associated with
the experience that evoked it. So once the subject has run out of the more usual associations (e.g.
‘chair’ in response to ‘table’), unusual ones (e.g. ‘elbow’ in response to ‘table’) come to mind.
Thus new ideas arise through a sort of ‘conceptual meltdown’, in that the meanings of concepts
merge or blend into one another more than usual, such that they are more readily recombined
[Boden 1991, Koestler 1964].
Flatter associative hierarchies are exactly what you would expect to result from a lower
activation threshold, and defocused attention may well be the means by which this is achieved.
Note that in a state of defocused attention or heightened sensitivity to detail, stimulus properties
that are less directly relevant to the current goal get encoded in memory. Since more features of
attended stimuli participate in the process of storing to and evoking from memory, more memory
locations are activated and participate in the encoding of an instant of experience and release of
‘ingredients’ for the next instant. The more memory locations activated, the more they in turn
activate, and so on; thus streams of thought tend to last longer. So if a stimulus does manage to
attract attention, it will tend to be more thoroughly assimilated into the matrix of associations that
constitutes the worldview, and more time is taken to settle into, or collapse on, any particular
interpretation of it. Thus, the less new stimuli can compete with what has been set in motion by
previous stimuli, i.e. the associative network plays a larger role in conscious experience.
The stimulus plays a role analogous to that of a measurement in quantum mechanics in the
sense that it actualizes an experience that was latent in the individual’s worldview, one that it was
potentially capable of given the appropriate context. Thus the stimulus not only provides
86
Chapter 6: Creativity and Cultural Novelty
information about the world, but it tests the workability of some portion of the worldview. One
could say it is like throwing a ball against a wall and observing how it responds. The more
flexible the material the ball is made of, the more it ‘gives’ when it makes contact. Similarly, the
more sensitive the individual is to stimuli, the greater the portion of the worldview that makes
contact with the world at that instant.17 Sometimes a stimulus does not elicit behavior
spontaneously, but triggers a train of thought that in the long term has an effect on the
individual’s cultural expression.
Another interesting consequence of defocused attention or heightened sensitivity is that a
concept that lies in the periphery of the hypersphere of activated memory locations can pull the
content of the next instant of experience quite far from the one that preceded and evoked it. Thus,
consecutive instants are less correlated because a concept from the periphery of the sphere of
activated locations can pull the content of the next instant of experience far from its predecessor.
So the conceptual network is not only penetrated more deeply, but also traversed more quickly. In
other words, there is an increased probability that one thought will lead in a short period of time
to a seemingly unrelated thought.
6.1.3
Conceptual Fluidity and Alertness versus Depth of Processing
There would seem to be a trade-off between processing provocative stimuli thoroughly, and
remaining alert to new stimuli. For example, say that Ann notices and stores in memory the fact
that Wanda’s book is blue and torn, but Ben doesn’t even notice that she has a book. Ann’s mind
will associate the Wanda with other memories of torn things, whereas Ben’s will not. Ann’s
knowledge that the book was torn may turn out to be useful. For instance, she might be more
alert, and ready to respond appropriately, to future indications of Wanda’s sloppiness. She might
say it’s ‘just a hunch’—she may no longer recall ever having seen the torn book—but the fact that
she once registered it as torn reconfigured her conceptual network in a way that made her more
alert to the potential. On the other hand, Ann might be so busy pondering the torn state of the
book that she fails to notice a truck headed her way! If you are still actively processing something
that happened a minute ago, you might not be as attentive to what is happening now.
Over the long haul, a tendency in this direction will yield a memory with regions of finer
granularity. Since more time is taken up with one’s own musings, less is left over to imitate
others. As a result, the individual settles on a more self-made worldview, and will perceive gaps
17
One can even speculate that, much as irregularities in the bounced ball cause its path to deflect,
constrictions (repressed memories) or gaps (inconsistencies) in the ‘collapsed’ portion of the worldview
87
Liane Gabora
in human understanding that others do not see, and that no existing idea, concept, or artifact
seems to fill. And since the self-made worldview contains not just the rule, but the reason behind
it, it can adapt more easily. New stimuli less readily command attention because they must
compete with what has been set in motion by previous stimuli. But if something does manage to
attract attention, it can get thoroughly assimilated into the matrix of abstractions, and thereby
become increasingly decoupled from the stimulus that triggered it.
In sum then, sensory information gets more or less thrown into one big melting pot (or keg,
you might say) which is the conceptual network. You could say that between stimuli, the beer
tops flip open and beer starts to ooze out, and the lower the activation threshold, the greater the
probability that any particular drop of beer will end up in a different can from the one it was in
just previously. A lower activation threshold is hypothesized to manifest experimentally as a
tendency toward defocused attention and flat associative hierarchies. The dynamical counterpart
to (or consequences of) these traits is longer and more frequent streams of abstract thought, with
lower correlation between one thought and the next. We can refer to this as heightened
conceptual fluidity. The high conceptual fluidity individual will settle into a worldview which,
although possibly lacking in some sorts of common knowledge, may be unique and richlydetailed in other respects.
6.1.4
Variable Fluidity as the Crux of Creative Potential
In the long run maintaining a low activation threshold indefinitely would be untenable since the
relationship between one thought and the next could be so remote that a stream of thought lacks
continuity. In the short run though, it is conducive to creativity, since there is a high probability of
‘catching’ new combinations of memories or concepts. But the important factor in creativity
seems to be not how high nor how low but how variable it is.
There is a considerable body of research suggesting that creativity is associated with, not just
increased fluidity, nor just increased control, but both [Barron 1963; Eysenck 1995; Feist 1999;
Fodor 1995; Richards et al. 1988; Russ 1993]. As Feist [1999] puts it: “It is not unbridled
psychoticism that is most strongly associated with creativity, but psychoticism tempered by high
ego strength or ego control. Paradoxically, creative people appear to be simultaneously very labile
and mutable and yet can be rather controlled and stable” (p. 288). He notes that, as Barron [1963]
said over 30 years ago: “The creative genius may be at once naïve and knowledgeable, being at
home equally to primitive symbolism and rigorous logic. He is both more primitive and more
may cause tension and thereby indicate a need for creative release, revision, or reconstruction. (Of course,
so long as the ball doesn’t completely deflate and slide down the wall, you’re doing fine. :-)
88
Chapter 6: Creativity and Cultural Novelty
cultured, more destructive and more constructive, occasionally crazier yet adamantly saner than
the average person” (p. 224). There is also evidence of an association between creativity and
variability in physiological measures of arousal. High creativity has been found to be correlated
with variability in heart rate [Bowers and Keeling 1971], spontaneous galvanic skin response
[Martindale 1977], and EEG alpha amplitude [Martindale and Hasenfus 1978]. Thus creativity
seems to involve the ability to vary the degree of conceptual fluidity in response to the demands
of any given phase of the creative process.
It is interesting to ask: was Einstein a seven-pack? Or was he a regular six pack like most of
us, but one in which the plastic thingy melted and resolidified in a truly exceptional way through
exquisite fine-tuning of the degree of fluidity? Probably both. The greater the number of
concepts, the more levels of hierarchically structured abstraction are possible, but only some of
these capture regularity of the world and therefore have meaning. So once there’s a certain
number of concepts in there, what matters most is how the individual weaves them together. In
fact, up to an IQ of 120, intelligence and creativity are correlated, but past 120 they are not
[Barron 1963]. My guess is that Einstein had a tendency toward defocused attention and high but
variable fluidity, and as a result, his worldview was less a product of what he was taught, more of
a self-made construction from the bottom up (like an agent in MAV when the creativity-toimitation ratio is turned way up).
6.1.5
Effect of Density of Abstractions on Creative Potential
Recall that abstractions generally have fewer properties than their instances, and thus cover more
of conceptual space. Because abstractions cover more of conceptual space, they vastly increase
the density of the space. The more abstract a concept, the more of conceptual space it covers, and
the greater the number of others potentially evoked by it. Thus, the greater the density of
abstractions, the greater the probability of finding relevant concepts.
6.2
THE BIRTH OF A CREATIVE IDEA
This section examines the cognitive mechanisms underlying the fluid, associative thought gives
rise to an initial glimmer of creative insight.
6.2.1
Relevant Possibilities Become Activated
What is the root source of the creative inspiration that leads to timeless music, architecture,
scientific discoveries, or just a new way to bake bread? Specifically, how is it that items in
memory that could potentially be relevant to our situation and surface to awareness?
89
Liane Gabora
Campbell [1988] claims it happens through a process of blind variation:
[The] tremendous gain in knowledge… can only be explained by a continual breakout from the
bounds of what was already known, a breakout for which blind variation provides the only
mechanism known.
He specifically states that by ‘blind’ he does not mean random, but what he is most emphatic
about is that it is not causal. And in fact as discussed in chapter five, the relationship between
associated memories and abstractions is one of correlation rather than causation18. They have
some commonality—though this commonality may something the human mind is unlikely to ever
consider—that caused them to be stored in overlapping memory locations. It is because of this,
and also because of the content addressable nature of memory, that the trajectory of a stream of
thought is constrained by associations between memories and abstractions that are similar or
spatiotemporally related [Schank 1983], which increases the probability that an advantageous
variant is found.
This can be made more concrete with an example. Since we have a beer theme going here,
imagine that you are the graphic designer and chief executive of advertising at Goosehead
Breweries, and it is your job to come up with a name and label for a new papaya-flavoured beer.
You look out your window at the bleak Saskatchewan landscape and see disappointingly little out
there to put you in a tropical frame of mind. You taste the beer itself, and feel even less inspired.
The taste of the beer registers as a pattern of activation that causes some synapses to be
excited (activated) and others inhibited. This pattern of activation flows through your memory
network, which in turn determines the multidimensional hypersphere of locations where this ‘beer
taste’ experience is stored. Since it is an experience of beer, it activates memory locations that
store the concept ‘beer’, as well as related concepts like ‘thirst-quenching’, ‘beverage’, ‘lager’
and ‘ale’. Since it is papaya beer, it simultaneously activates concepts like ‘fruit’, ‘papaya’,
‘juicy’, and ‘tropical’. If you were to spend some time on this problem, ideas that pertain to the
tropics might for you become a highly active region of conceptual space, analogous to the
uncharacteristically high level of activity (and polymorphism) in a small portion of the human
genome known as the major histocompatibility complex (MHC) which deals with immune
response [Hughes & Nei 1988]. The bottom line is that the content of the stimulus probes for
correlated content in the memory.
18
It is interesting to note that the same is true in quantum mechanics.
90
Chapter 6: Creativity and Cultural Novelty
6.2.2
Analogy
Often the correlated items are not single properties but entire structures. Much research on
creativity deals with how the structure or conceptual skeleton underlying one idea gets abstracted
and applied analogically to another [Gentner 1983; Gick & Holyoak 1983]. We can expand on
this by suggesting that many forms of creative expression begin by exploring what would the
pattern of information that encoded this experience of this particular event look like if expressed
through the constraints of that medium? The existence of inherent limitations on how a pattern
could be translated from one domain to another is consistent with the frequent observation that
creativity involves both freedom and aesthetic constraint. Thus all creativity is directly or
indirectly derived from experience in the world. We find that this world has a large degree of
regularity and predictability; the mathematics with which we are able to describe vast segments of
it, the set of all natural functions, is a small subset of all possible functions. Thus the constraints
that guide the generation of novelty are not arbitrary but objective and familiar; for example the
drum beat of a song might echo a heartbeat, when the rhythm and chord progression are
reminiscent of the sound of someone sobbing we feel sad, and we hear the wrong note even if we
have never heard the song before.
6.2.3
Retrieving and Reconstructing
Recall also how retrieval is a reconstructive event. In the process of reconstructing a memory,
errors may be introduced. In this sense, a distributed memory is less efficient than an expert
system, which stores items separately (like hats put in different boxes). However, although
reconstruction is a source of inaccuracy, it enables the emergence of novelty.
To return to our ‘papaya beer’ example, the process of etching this experience to certain
locations triggers release from them of whatever else has been stored in them. In other words, the
locations that store one instant of experience go on to provide the constituents of the next instant
of experience. Of course, nothing is drawn from them if, at that instant, some stimulus or
biological drive becomes pressing, such as a phone call, or the need to go to the bathroom. But to
the extent that memory contributes to the next instant of awareness, storage of the ‘papaya beer’
experience activates retrieval of not only the ‘papaya beer’ experience itself, but all other
memories stored in the same locations, and the next experience can be calculated by determining
the contributions of these memories feature-by-feature (where features probably take the form of
hierarchically nested phase relations as discussed in the previous chapter). The other evoked
memories contribute less to the next instant of experience than the ‘papaya beer’ experience you
just had, not just because they are less recent, but because they do not lie at the center of the
91
Liane Gabora
hypersphere of activated locations, and so are statistically likely to cancel one another out. In this
instance, the various activated concepts do not completely cancel out, at least not with respect to
their sound. Since the words ‘tropical’ and ‘ale’ sound alike, the ‘al’ sound emerges as a
significant property. (Dynamically speaking, the ‘al’ sound in ‘tropical’ constructively interferes
and resonates with the ‘al’ sound in ‘ale’.)
6.3
REFINEMENT AND EXPRESSION OF AN IDEA
We have looked at how an insight surfaces to awareness through the defocusing of attention and
subsequent entry into a state of conceptual fluidity wherein unusual connections are gleaned.
Now let us examine how an initial flicker of inspiration solidifies into a workable idea as it gets
mulled over in light of the various constraints and affordances of the world into which it is
eventually born.
6.3.1
Merging Simultaneously Evoked Concepts
Even the identification of an abstraction is a creative act. It is a transformation of a conceptual
space—and thus an example of what we have been calling impossibilist creativity—that occurs
through distilling out what different mental entities have in common generally by ignoring one or
more properties. But when we think of creativity we most often think of merging different
properties of concepts. A ‘snowman’, for example, blends some aspects of ‘snow’ with some
aspects of ‘man’. The distributed nature of human memory is the wellspring of both sorts of
creativity, and in fact many or perhaps most instances of creativity involve some of each. The
greater the density and degree of overlap of abstractions, the more different ways of funneling
experience through the conceptual network, abstracting something new out of it, and manifesting
the essence, or feel of it, through the constraints of a new medium or domain [Gabora 2000].
The emergence of the ‘al’ sound as a significant property is an artifact, since the ‘al’ sound
has seemingly nothing to do with ‘papaya beer’. However, since you know that catchy product
names sometimes capitalize on this sort of artifact, your memory is probed again with an instant
of experience that is the same as the previous one except for heightened activation of the ‘al’
sound feature. The first probing, the one that collapsed on an interpretation of the ‘papaya beer’
taste experience, was not a waste of time; it got you closer in conceptual space to where you
needed to be. The second probing is even more successful. It evokes the new construct,
‘TropicAle’.
The point is that it is because at some time in the past you had stored the concept ‘ale’ within
reach of ‘beer’, and ‘tropical’ within reach of ‘papaya’, that you were able to make the first step
92
Chapter 6: Creativity and Cultural Novelty
toward a brand name that creatively combined them. And it is because ‘tropical’ and ‘ale’ had
both, at some time in the past, been etched into overlapping memory locations—those most
attuned to the ‘al’ sound—that the idea of ‘TropicAle’ was possible. Though it is a reconstructed
blend, something you have never actually experienced, it can still be said to have been evoked
from memory. It is certainly not a profound idea. But it is an idea nonetheless, and good enough
to let you to hold onto your job at Goosehead Breweries.
6.3.2
Honing in on an Idea
A large part of the process of honing in on an idea involves distilling a salient insight from the
specifics of the memories or concepts originally involved in its conception, and making it
applicable to the goals and modes of expression available for its manifestation. For example, the
essence of the newly conceived ‘TropicAle’ construct is gleaned by reconsidering it in light of its
various properties (such as ‘fruity’), non-properties (‘bitter’, say), relationships (such as, say, your
visit to Venezuela), and instances (such as your imagined preconception of bottles of ‘TropicAle’
lined up on the grocery shelf), and then funneled through the constraints inherent to the domain of
logo design. The degrees of freedom—that is, the variety of directions the project can
take—gradually decrease. For example, as you zero in on the idea of drawing the ‘T’ in ‘Tropic’
in the shape of a palm tree, alternative ‘T’ fonts are ruled out.
When an idea has closed a gap at the level of one’s own worldview, it is ready to be nurtured
to a state where it will close the corresponding gap at the cultural level. As the idea is refined, the
creator’s associative hierarchy can afford to narrow, and in fact must narrow, if is to be received
by the world. Much as a beam of light diverges when it passes through a concave surface, a
creative idea seems to lose clarity when expressed to others. So the existence of this second level
of acceptance prompts the need for further refinement. In the case of your Goosehead Breweries
assignment, this involves progressively honing in on your most potent final draft of the
‘TropicAle’ logo, surrounding it with a flourish of fruit slices (perhaps, as a final added touch,
making them vaguely genitalia-like so as to invite subliminal inklings in the brains of the buying
public), and placing the finished product on your boss’s desk.
6.4
SELECTION AND THE CULTURAL FITNESS LANDSCAPE
We now examine factors pertaining to the selection of idea amongst individuals, and the topology
of the cultural fitness landscape.
93
Liane Gabora
6.4.1
Mental Simulation
In mental simulation [Nersessian 1993], instead of actually seeing what happens if an idea is
implemented in the real world, one’s internal model of the world simulates it vicariously, and the
idea may or may not be implemented in the real world as a result. The success of mental
simulation varies with the accuracy of ones’ internalized model of the world, but it provides
selection of at least an indirect (or as Campbell would say, vicarious) sort. Even the simple
implementation of mental simulation in MAV proved effective. In some situations, agents
simulated what would happen if they were to implement an idea they invented; in other cases they
simulated what would happen if they were to imitate a neighbor.
6.4.2
Interpersonal Prompting
When one individual exhibits an idea that another observing individual values, the observing
individual often displays reinforcing body language and emits words of encouragement.
Likewise, if the idea is threatening or inconsistent with valued ideas, the observing individual’s
words and behavior tend to be discouraging. Our need for social acceptance makes us more likely
to express ideas that have been reinforced and less likely to exhibit those that have been
discouraged. This almost looks like a subtle strategy by which an idea in the mind of one
individual coaxes implementation of itself in other individuals.
6.4.3
Brains Select Ideas that Satisfy Needs
Since many of our needs have a biological basis—e.g. the need for food, shelter, and so
forth—the generation of cultural novelty is largely constrained by our heritage as products of
biological evolution. Thus the topology of the cultural fitness landscape largely echoes that of the
biological fitness landscape to which we have been adapting since life began. Our perceptual and
cognitive systems are wired up such that they are primed to focus on and highlight those aspects
of external reality that are relevant to our survival (or were in the past), and the mental images we
form in our minds reflect that bias [e.g. Hubel & Wiesel 1979; Marr 1982]. Second, the
associative organization of memory constrains variation-generating operations. So convergence
factors are built right into our hardware.
In the short term, the biological fitness landscape, and thus the cultural fitness landscape,
fluctuates continuously as one need is satisfied and others take precedence [Hull 1943; McFarland
& Sibly 1975; Gabora & Colgan 1990]. For example, after eating, ideas that pertain to finding
food are less likely. However over the lifetime of an individual the set of biologically-based needs
remains relatively constant. The trajectory of survival-motivated thought can be described as a
94
Chapter 6: Creativity and Cultural Novelty
limit cycle (periodic attractor) that moves through the set of stable concepts whose
implementations satisfy the various biological needs. Since we preferentially spread ideas that
satisfy needs, our needs define viable niches for ideas to evolve toward. As infants we might cry
and kick no matter what need is most pressing, but as children we acquire and continually refine a
repertoire of ideas that, when implemented, satisfy various needs. We learn that reaching into the
cookie jar satisfies one need, shouting ‘help’ satisfies another, et cetera. Our ideas, and the
behavior they elicit, slide into need-defined attractors (regions of stability) in the cultural fitness
landscape. The landscape may in fact be better described as ‘basins of attraction’ that become
connected by ‘portals’ to basins of higher fitness [Crutchfield 2001].
6.4.4
Cultural Fitness Landscape is Sculpted by Need for Worldview Cohesion
The fact that ideas are not independently self-replicating like genes does not prevent them from
proliferating. In fact it may ironically work in their favor, because the cognitive machinery they
depend upon not only actively manipulates them to produce ‘offspring-ideas’, but organizes them
into a model of the world, or worldview, which it can use to figure out what to do whenever a
situation is too complicated for its hardwired instincts. The worldview provides a framework
from within which a new idea will make sense, and a need, or niche, for it. Of course, any
relevant precursor ideas must first be assimilated [Wallas 1926]. This constraint amounts to a
malleable, or plastic, form of selection on which ideas one exposes oneself to and assimilates.
The worldview orchestrates behavior such that an idea gets implemented right when it is
likely to be useful, which increases the probability that other individuals will consider it worthy
of imitating. A need that seems to surface to attention when other needs are not pressing is the
need to connect fragments of experience into a relatively consistent mental model of reality, or
worldview. The more unstable the environment, the more our ability to make predictions and
evaluate possible plans of action hangs on the accuracy of this worldview. The survival value of
the tendency to weave experiences together into a relational whole is clear; thus, much attention
will be devoted to how such a worldview emerges.
Our worldviews overlap to the extent that similar experiences and genetic make-ups cause
our brains to process information similarly. But they don’t overlap perfectly; each individual’s
train of thought traces out a unique trajectory through conceptual space. It can be useful to think
in terms of not only the worldview of an individual, but also the worldview of a group or even
human society at large, wherein all frontiers of human endeavor are incorporated.
95
Liane Gabora
6.4.5
How Variation Feeds Back on the Fitness Landscape
Impossibilist creativity transforms conceptual space and thus affects how a fitness function
applies to the space; that is, changes the cultural fitness landscape. Much as the evolution of
rabbits created ecological niches for species that eat them and parasitize them, the invention of
cars created cultural niches for gas stations, seat belts, and garage door openers. As one
progresses from infanthood to maturity, and simple needs give way to increasingly complex
needs, the stream of thought acquires the properties of a chaotic or strange attractor, which can be
viewed as the formation of crevices in the original limit cycle. The landscape is fractal (i.e. there
is statistical similarity under change of scale) in that the satisfaction of one need creates other
needs—every crevice when examined closely reveals more crevices. This is analogous to the
fractal distributions of species and vegetation patterns described by ecologists [Mandelbrot 1982;
Palmer 1992; Scheuring & Riedi 1994]. An endpoint of a cultural evolution trajectory turns out to
be not just a point in multidimensional space, but a set of points with their own fitness
function—a micro-landscape in its own right. So although the cultural fitness landscape loosely
follows the biological fitness landscape, there are places where it deviates, and this effect
undoubtedly becomes more pronounced throughout an individual’s lifetime. This means that the
potential for conceptual diversity, though constrained by need, is open-ended.
6.5
WE STILL CANNOT MATHEMATICALLY DESCRIBE IMPOSSIBILIST CREATIVITY
Let us return to the problem of mathematically describing the appearance of new states with new
properties that occurs during episodes of impossibilist creativity, which was introduced in section
3.5.4. Whether the ‘TropicAle’ example constitutes a case of impossibilist or just improbabilist
creativity depends upon what you as the mathematical modeler of this process assumed to be the
relevant properties given the initial state of considering the advertising of papaya flavored beer. It
also depends on whether you are thinking about the conceptual space of working memory, or the
space of all concepts. In any case, if we were to continue to follow the evolution of this stream of
thought, by which we just mean its dynamical change of state, sooner or later we would encounter
a state of the mind wherein the subject of thought has properties that are different from those one
put not only in the original working memory state space, but in the entire conceptual space; the
spontaneous generation of a new state with not, just even new properties, but relevant ones. So if
one wants a description of the change of state in a stream of thought—as is the case if one wants
to describe the generation of new ideas in culture—there a difficult problem.
96
Chapter 6: Creativity and Cultural Novelty
6.6
SUMMARY
It was hypothesized that the creative generation of cultural variety results from the simultaneous
retrieval of items that were stored in overlapping memory locations but never retrieved before
simultaneously; the lower the activation threshold, the greater the conceptual fluidity, and the
more likely this is to happen. However, it is not just a lower activation threshold that allows
greater creativity, but the possibility for the threshold to increase or decrease depending on the
nature of the situation. Thus, in the refinement of an idea, the threshold should gradually increase
to enable irrelevant or misleading aspects of the idea to be weeded out. This hypothesis is
discussed in relation with common attributes of creative individuals: defocused attention,
heightened sensitivity, and flat association hierarchies.
Some of the factors that shape cultural information stem directly or are derived from
survival needs, and some stem from the value of a more or less coherent, relationally structured
worldview. Later we will explore how such a relationally structured worldview could come about.
To return to the Bob and Doug MacKenzie parlance, culture may have begun with the emergence
of a ‘plastic thingy’—a hierarchical network of abstractions that unifies memories into an
interconnected web.
97
Liane Gabora
the
98
Chapter Seven
7
What Sparked the Origin of Culture?
We have seen that, whether you call culture a second form of evolution, or say that it is merely
analogous to biological evolution, it does exhibit adaptive change in response to environmental
constraints, and many of the same phenomena emerge. But now we have another mystery before
us. How did the evolution of culture begin?
7.1
WAS IT IMITATION OR CREATIVITY?
Many are inclined to limit culture to only those mental entities acquired through imitation. There
is also a substantial literature that equates the origin of culture with the onset of the capacity for
imitation, and to a lesser extent other forms of social learning [Blackmore 1999; Richerson and
Boyd 1998; Dugatkin in press; Kendal & Laland 2000]. The assumption is that imitation is the
missing ingredient that prevents the emergence of complex culture in animals. Blackmore goes so
far as to say that imitation is what makes us human, the source of our uniqueness: “The thesis of
this book is that what makes us different is our ability to imitate”.
However, the idea that the ability to imitate is the defining element of what makes us human
certainly seems counterintuitive. When we feel proud to belong to the human race, we think of
the great pyramids, beautiful music, the airplane… in short, the fruits of creativity. The word
‘imitation’ is, in fact, often used to denote inferiority. Could it really be that conformity, not
creativity, is the hallmark of the human condition, that the capacity to copy one another is what
brought about the explosion of artifacts, language, art, science, religion and so forth that
constitute human culture? Are our intuitions about what makes us special actually so misguided?
This chapter presents an alternative hypothesis: that it is not imitation but creativity—or
rather, the strategic, contextual manner in which the mind is creative—that unleashed culture and
that makes us human [Gabora 1999, 2000].
99
Liane Gabora
7.2
WHAT DIFFERENTIATES HUMANS FROM ANIMALS?
We begin with a comparison of minds that are and are not able to sustain cultural evolution. The
line of reasoning behind the ‘imitation drives culture’ argument basically goes as follows:
imitation is natural and ubiquitous in humans, but almost nonexistent in animals, and that is why
we alone have evolved culture. Though the first part of this argument is clearly true, the second is
highly controversial. A strict definition of imitation involves copying not just the form but also
the goal of an action, and this is difficult to conclude demonstrate conclusively in animals.
However imitation of at least the form of an action has been documented in budgerigars [Galef et
al. 1986], quail [Akins & Zentall 1998], cowbirds [King & West 1989], rats [Heyes & Dawson
1990; Heyes et al. 1992], monkeys [Beck 1976; Hauser 1988; Nishida 1986; Westergaard 1988],
orangutans [Russon & Galdikas 1993] and chimpanzees [Goodall 1986; Mignault 1985; Sumita et
al. 1985; Terrace, Petitto, Sanders & Bever 1979; Waal 1982; Whiten 1998]. Reviews by Bonner
[1980], Robert [1990], and Smith [1977], and more recently, Byrne and Russon [1998, see also
the accompanying commentary] lead one to conclude that the capacity for imitation and related
phenomena in mammals is fairly widespread. Nevertheless, as many authors have pointed out
[e.g. Darwin 1871; Plotkin 1988], although imitation is commonplace in the animal kingdom, no
other species has anything remotely approaching the complexity of human culture. It has been
suggested that the reason is that they are limited by the lack of a ‘theory of mind’ [Cheney &
Seyfarth 1990; Premak & Woodruff 1978], by which they mean an inability to interpret the acts
of another as implying something about the other’s intentions.
Some are tempted to say that the ability of animals to respond appropriately to salient
stimuli, and even learn arbitrary sensorimotor associations, indicates some capacity for symbolic
thought. However, animals’ learned behavior is stereotyped and brittle—it is not readily adapted
to new contexts—which suggests that they use symbols only in an iconic sense. We have no
evidence that they, for example, engage in streams of thought that reorganize mental entities in
ways that make their similarities and differences more explicit.
The lack of culture in animals despite evidence that, when put to the test, they can imitate, is
consistent with the alternative proposal, that it was the advent of the ability to be creative that
brought about the origin of culture. In fact, even if animals could imitate as well as we can, one
should not expect to see them imitating a great deal, because imitative capacity requires variation
for it to work on; it remains latent, hidden from view, until such variation is present. The reason
for this is straightforward. Recall how in the Meme and Variations computer model of cultural
evolution, one can vary the ratio of imitation to invention. We saw that when the agents’ ability to
imitate is set full strength, and their ability to invent is turned off, what happened is... nothing. In
100
Chapter 7: What Sparked the Origin of Culture?
retrospect this should not be surprising. There has to be something worth imitating before the
ability to imitate can come in handy, or even manifest itself. Novelty can then breed more
novelty. Or as one choreographer (whose name I forget) put it: “If we don’t do what our
predecessors did, we’re doing what our predecessors did.”
Since a social group wherein individuals can imitate but not invent is fairly stagnant, the
suggestion that it was the appearance of imitation that brought about the onset of culture does not
look promising. A more reasonable hypothesis, it seems to me, given the animal behavior data, is
that the capacity for imitation has been around for some time, and that it is the capacity to be
strategically creative that came along and unleashed the tidal wave of art, language, religion, and
so forth, that is with us today. Note though that even given a lack of strategic creativity, some sort
of cultural evolution could take place through the random generation of entities worthy of
imitation, and the stringing together of imitated cultural entities in a trial and error (as opposed to
truly creative) manner. However, in order to be worthy of imitation, a cultural entity must on its
own perform some useful function, before it has been strung together with other cultural entities.
And it goes without saying that not only the imitated components, but the trial and error
combination of them, must have some useful function. Thus the possibilities for ending up with
something useful are limited, given both such a random means of exploring the infinitely vast
space of what can be concieved, and the constraint that randomly generated novelty must consist
of parts that themselves are not only actualized in the physical world, but so valuable in this
actualized form that others imitate them. On the other hand, creatively shaping an idea in accord
with a goal, does not have either of these of these limitations. The first it does not have by
definition, and the second it does not have because one can incorporate into the creative product
all sorts of ideas or impressions in a meaningful way which have not had to pass the test of being
useful enough to be imitated. So the onset of strategic creativity seems to be a much more
plausible explanation for a cultural explosion than the onset of imitation.
Yet another proposal is that it is niche construction that differentiates animals from humans
[Laland et al. 1999]. Niche construction simply refers to the manipulation of ones’ environment
in useful ways. It is true that animals modify their environments in ways that help them thrive.
But this sort of phenomenon is even more widespread and doesn’t even require conscious
intention; a river carves out an ever-wider river bed, which in turn ‘gives it more room’ to flow.
So since niche construction is come by rather easily, it is probably not an ideal litmus test for the
ability to generate the complex entities that comprise human culture. In support of their view,
Laland et al. claim that the capacity for culture in animals and humans differs by degree only:
“Modern culture did not suddenly emerge from some pre-cultural Hominid ancestor. The
101
Liane Gabora
psychological processes and abilities that underlie culture have evolved over millions of years,
and can often be found in rudimentary form in animals.” It is clear, however, that a vast gulf
separates the cognitive abilities of humans from other animals [Donald 1991, 1993; Plotkin
1988].
7.3
WHICH HYPOTHESIS DOES THE ARCHEOLOGICAL EVIDENCE SUPPORT?
Nevertheless, for the sake of completeness, let us consider the possibility that the capacity for
strategic creativity was around for some time before the capacity for imitation appeared. What
would have happened then? Consider the opposite situation in the Meme and Variations computer
model to that considered above. When innovation was set full strength, but imitation is turned
off—no means for useful innovations to spread through the society of agents—culture did evolve,
albeit slowly. This makes intuitive sense; there is no way of building on the inventions of others,
so the wheel is constantly being reinvented, but at least it is being invented. Were this situation to
have been historically true, then prior to the origin of imitation, there would have been variation
everywhere, and the onset of imitation would have funneled this variation in a few of the most
useful directions (which would then be further improved and embellished...) Thus, in a creative
society, the onset of imitation would have dramatically accelerated the pace at which culture
evolves by weeding out inferior cultural entities and replacing them with superior ones. However,
this hypothesis is not consistent with the archeological record.
7.3.1
An Archeological Transition
By dating artifacts such as various sorts of tools, we learn approximately when humans acquired
the ability to make and use those tools. Human cognitive abilities seem to have undergone a
transition following the arrival of Homo erectus approximately 1.7 million years ago. This period
more or less marks the appearance of sophisticated stone tools and habitats, use of fire, longdistance hunting strategies and much increased consumption of meat, and migration out of Africa,
as well as a rapid increase in brain size [Bickerton 1990; Chase 1990; Chase & Dibble 1987,
1992; Corballis 1991; Donald 1991]. Blackmore [1999] cites the sudden increase in tool variety
to support her thesis that the onset of the capacity to imitate is what gave rise at this time to the
origin of culture.19 It is at this point that hopes of pulling off the ‘imitation drives culture’ idea
takes a nosedive. This increase in variety is the opposite of what one would expect if humans
19
In fact, it is actually debatable whether the variety of tools did increase at this time, although their
complexity certainly did, and the variety of means of successfully coping with new problems is evidenced
by migration to vastly differing environments at this time.
102
Chapter 7: What Sparked the Origin of Culture?
suddenly acquired the ability to imitate. If imitative capacity were to have suddenly arisen at this
time, then just prior to it there would have been a great variety of tools, and the onset of imitation
would have manifested archeologically as sudden wide-spread adoption of the best. If Og the
caveman and Oga the cave girl each have their own way of making a tool, and Oga notices that
Og’s way works better and therefore imitates it, then the variety of ways that tools get made
decreases. Thus, to bolster her thesis with archeological evidence, Blackmore would have had to
find a period of time where tool variety sharply decreased. The evidence she cites is, in fact,
consistent with the thesis that strategic creativity, rather than imitation, was the bottleneck to
culture.
7.3.2
A Second Archeological Transition
Human culture underwent another perhaps even more explosive transition between 60,000 and
30,000 years ago. [Mellars 1973, 1989a, b; Leakey 1984; Mithen 1996; White 1982, 1993].
Mithen refers to this period as the ‘big bang’ of human culture, claiming that it shows more
innovation than the previous six million years of human evolution. Leakey writes:
Unlike previous eras, when stasis dominated, innovation is now the essence of culture, with change
being measured in millennia rather than hundreds of millennia. Known as the during the Upper
Paleolithic Revolution, this collective archeological signal is unmistakable evidence of the modern
human mind at work.
The Middle / Upper Paleolithic marks the colonization of Australia, the replacement of
Levallois20 tool technology by blade cores in the Near East, and the first appearance of art in
Europe, including naturalistic cave paintings of animals, bone and antler tools with elaborate
engraved designs, ivory statues of animals and sea shells, and personal decoration such as beads,
pendants, and perforated animal teeth. Some of these items are thought to be associated with the
beginnings of ritualized religion, as is the appearance of elaborate burial sites. It is also the
beginning of a much more strategic style of hunting, involving specific animals at specific sites.
Perhaps most importantly, it is thought to mark the beginnings of complex language. Once again,
this period is marked by a transition to increased divergence of cultural entities—as would be
made feasible by enhanced creativity—rather than convergence on highly effective cultural
entities—as would be made feasible by enhanced imitation. (More specifically, enhanced
imitation could cause divergence between different communities, due of the Founder
20
Levallois technology is a method for generating carefully shaped blades out of stone.
103
Liane Gabora
Effect—reduced variation due to drift, as discussed in chapter two—but not within the same
community.)
The debate here about whether imitation or creativity was the bottleneck to culture may be
more polarized than necessary, in the sense that to do either one really well may require the other.
Imitation is rarely perfect the first time, in apes as well as humans, and it often builds on
previously learned behaviors, suggesting that a certain amount of experimentation and
construction is involved [Custance et al. 1995]. Similarly, the more one has assimilated from the
social milieu—the others with whom one interacts—the more one has to work with creatively.
The same underlying cognitive mechanisms may be useful to both. This may explain the
generation of novelty by animals. Nevertheless, the two periods of human history that have been
proclaimed to be the beginnings of culture are times of increased variety, which indicates a higher
level of creativity. We saw that the animal behavior data supports this by showing that animals
can imitate; it seems unlikely, therefore, that difficulty imitating is what holds them back. We
tend to think of replication as the bottleneck to evolution because in the biological case, where the
‘smarts’ went into the complex means by which organisms reproduce, and variation is largely
random, replication certainly got the ball rolling. But in a process like culture, where the complex
part is the strategic, contextual means by which variation is generated, it is much more plausible
that creativity is the bottleneck.
7.4
TWO PARADOXES: THE ORIGIN OF LIFE AND THE ORIGIN OF CULTURE
Whereas the animal mind stores memories and stimulus-response associations independently, the
human mind organizes them in a way that reflects abstract relationships. But the existence of this
uniquely human kind of mind leaves us with a nontrivial question of origins. What sort of
functional reorganization was behind it? For a stream of creative thought to unfold, related
memories must become accessible to one another; they must be woven into an interconnected
conceptual web, or worldview. However until a mind incorporates relationships between
memories through the formation of abstractions, how can one thought evoke another? And until
one can evoke another, how are relationships established amongst memories so that they become
an interconnected worldview? In other words, if you need a relationally structured worldview to
generate a stream of thought, and streams of thought are necessary to generate the abstractions
that connect memories into a worldview, how could one have come into existence without the
other? We have a chicken and egg problem.
Jeffreys, in his own way, makes the same point in response to the following statement by
Dennett concerning ‘invader’ memes and their human ‘hosts’:
104
Chapter 7: What Sparked the Origin of Culture?
It is not surprising that the invaders were well adapted for finding homes in their hosts, since
they were themselves created by their hosts, in much the way spiders create webs and birds
create nests.
Jeffreys reacts:
Now, if memes are made by brains as webs are made by spiders, how could memes also be
analogous to a separate, parasitic species, such as the hypothesized ancestors of our
mitochondria, as well as to the genes that make all the mitochondria and spiders and brains?
This yoking of metaphors hardly seems a sure foundation for the rigorous “science of memetics”
that Dennett hopes will emerge.
What Jeffreys forgot to write is that the same sort of paradox lies at the heart of his own field of
biology. The paradox he points to is reminiscent of the paradox of the origin of life. The paradox
of life can be stated simply: if living things come into existence when other living things give
birth to them, how did the first living thing arise? That is, how did something able to reproduce
itself come to be? In biology, self-replication is orchestrated through an intricate network of
interactions between DNA, RNA, and proteins. DNA is the genetic code; it contains instructions
for how to construct various proteins. Proteins, in turn, both catalyze reactions that orchestrate the
decoding of DNA by RNA, and are used to construct a body to house and protect all this selfreplication machinery. Once again, we have a chicken-and-egg problem. If proteins are made by
decoding DNA, and DNA requires the catalytic action of proteins to be decoded, which came
first? How could a system composed of complex, mutually dependent parts come into existence?
The origin of life and the origin of the cognitive dynamics underlying culture might appear
at first glance to be very different problems. However, deep down, they both amount to the same
thing: the bootstrapping of a system through which information patterns can be generated and regenerated, and the selective proliferation of some variants of these patterns over others. Perhaps it
is not so far fetched to hope that a solution to the one might help provide a solution to the other.
7.5
A PSYCHOLOGICAL PERSPECTIVE ON THE ORIGIN OF CULTURE
Let us move from an archeological to a psychological perspective on the origin of culture.
105
Liane Gabora
7.5.1
Underlying an Archeological Transition is a Cognitive Transition
Merlin Donald [1991] proposes in Origins of the Modern Mind that it was one or more cognitive
transitions that made possible the characteristic complexity and ingenuity of human culture [see
also Barkow et al. 1992; Donald 1993a, 1993b; Steels 1995; Tomasello et al. 1993; Tooby &
Cosmides 1989]. Donald argues convincingly that before the arrival of Homo erectus, the human
memory system was like that of a primate21, limited to the storage and cued retrieval of specific
episodes; it cannot form abstractions [see also Heyes 1998]. Accordingly, he uses the term
episodic to designate such a mind. We noted in chapter five that in addition to episodic memories,
a mind can also store declarative and procedural memories. Since all three kinds of memories
involve the formation of associations amongst spatiotemporally or semantically correlated
features, at the suggestion of Francis Heylighen (pers. com.) we will use the term associative
mind to make sure there is no confusion.
The awareness of an associative mind is dominated by the events of the present moment.
Occasionally it encounters a stimulus that is similar enough to some stored memory to evoke a
retrieval or reminding event. Occasionally such a stimulus evokes a reflexive, or (with much
training) learned response (such as pointing at a token of a certain shape to obtain food).
However, it has great difficulty accessing memories independent of environmental cues. It cannot
manipulate symbols and abstractions, nor invent them on its own, and is unable to improve skills
through self-cued rehearsal. It is capable of social attribution, insight and deception, and is often
sensitive to the significance of events. This suggests that it may have some sort of interconnected
model of the world it lives in, but that such a model, if it exists, is relatively rigid and not
interconnected and continually restructured through hierarchical layers of abstractions.
Donald claims that the cultural transition following the appearance of Homo erectus was
made possible by a transition to a cognitive architecture that encodes relationships between
episodes by way of abstractions, and relates abstractions to one another by way of higher-order
abstractions. For example, we know that the experience of seeing Rover was similar to the
experience of seeing Lassie because Rover and Lassie are both instances of ‘dog’, and we know
that dogs are animals. As we have seen, the conceptual mind can also retrieve and recursively
operate on memories independent of environmental cues through representational redescription.
By redescribing an episode in terms of what is already known, it gets rooted in the network of
understandings that comprise the worldview, and in turn, the worldview is perpetually revised as
new experiences are assimilated and new symbols and abstract concepts invented as needed.
21
Although this may not be completely accurate; see [Donald 1993] and accompanying commentary.
106
Chapter 7: What Sparked the Origin of Culture?
The capacity for a self-sustained stream of thought that both structures and is structured by
an internal model of the world enables us to plan and predict, to generate novelty, to tailor
behavior according to context, to relate the intentions of self and others to appropriate actions,
and to exert precise control over communication. The conceptual mind is able to rehearse and
refine skills, and therefore exhibits enhanced behavioral flexibility, and more precise control over
intentional communication. The upshot is cultural novelty. Mime, play, games, and toolmaking,
Donald says, are manifestations of a new system, built upon the mind’s associative foundations,
and consisting of a “multimodal modeling system with a self-triggered rehearsal loop”. He claims
that it is not clear that this cognitive reorganization must be localized to any single anatomical
location, but that it must have functional unity. What sort of cognitive transformation could it
have been to get the ball rolling, to enable the process of cultural evolution to take hold? As
Steels [1996] put it:
We need to find the major transitions through which neural networks (e.g. brains), which were
initially special-purpose and hence the subject of evolution by natural selection, have become
general-purpose and moldable by developmental and learning processes. There is so far no
theory to explain this, partly because there is not yet an adequate theory that explains the
plasticity of neural networks as such.
There is indeed strong evidence that at least some domains of human skill have an innate
component [e.g. Chomsky 1957, Lumsden & Wilson 1981; Pinker 1994; see also Elman et. al.
1996 for a cognitive science perspective], and many authors hint that the connecting together of
domain-specific brain modules was a crucial step toward complex intelligence [e.g. Fodor 1983;
Gardner 1983, 1993; Karmiloff-Smith 1992; Rozin 1976; Sperber 1994]. Mithen [1996], stressing
the importance of connected domains for analogical thought, suggests that this is what lay behind
the cultural explosion of the Middle/Upper Paleolithic.
7.5.2
The Origin of Culture Paradox: A Concrete Example
Steels emphasizes that the origin of human intelligence had little to do with the complex abilities
we tend to associate with intelligence, such as chess playing ability, but rather with enhanced
ability to perform tasks associated with survival and well being. So let us look at a concrete
example of how cognitive plasticity could be useful in the solution of a specific task. We know
that the brains of an ancestral tribe somehow turned into instruments for the generation,
refinement, and selective assimilation of ideas. Consider Oga, a member of this ancestral tribe.
When Oga had her very first experience, there were no previously-stored memories to be
107
Liane Gabora
reminded of, just hardwired associations, and external and internal stimuli (such as hunger). As
associations accumulated, occasionally it happened that an instant of experience was so similar to
some stored episode that a retrieval event occurred, and she was reminded of that past experience.
Perhaps the retrieval elicited a learned response. For example, the sight of a gourd might have
reminded her of a similar gourd that her brother once scraped the seeds out of and used to carry
water. Embedded in this recollection was the refreshing taste of the water he shared with her. This
might have led her to use the gourd to carry water. But since she just has stored memories, no
abstractions, this is the only kind of influence it could exert. It could not set off a chain reaction of
abstract thought regarding, say, whether other things besides water could be carried in a gourd, or
how the gods might like it if this gourd were buried with a clan member who recently died. Her
awareness was dominated by the stimulus of the present moment.
At some point in her life, however, she managed to direct her attention, not to a particular
stimulus, nor to the satisfaction of a biological drive, but to a stream of abstract thought. She kept
this stream of thought going long enough to refine a concept or perspective, or invent a novel
artifact. But in the absence of representational redescription, how do relationships get established;
how do memories become connected to one another through a hierarchically structured web of
abstractions? And conversely, until a mind incorporates abstract relationships, how can one
thought evoke another, which evokes another, et cetera, in a chain of representational
redescription? How does a mind come to assume a self-sustained stream of thought that
progressively shapes, and is shaped by, a highly-structured model of the world?
7.6
SUMMARY
This chapter weighs two competing hypotheses regarding what gave rise to the origin of culture.
One is that it was the onset of the capacity for imitation, and the other is that it was the capacity
for strategic creativity. The intuition that imitation was the bottleneck to culture probably stems
from a biological mindset; in biology, the advent of self-replication through reproduction was the
critical milestone. However, it does not follow that cultural evolution began with the onset of
imitation, even if imitation plays the analogous role of replication, because the correlatedness of
consecutive instants in a stream of experience also plays this role. Biology, a parallel strategy,
puts its energy into an elaborate system for replicating; culture, a heuristic strategy, puts its
energy into an elaborate system for generating novelty.
There are in fact several sources of evidence that the ‘we are unique because we imitate’
thesis is wrong. The first comes from findings that animals imitate. The second comes from
archeological evidence that the origin of culture is associated with increased variety, which
108
Chapter 7: What Sparked the Origin of Culture?
results not from imitation but strategic creativity. Since strategic creativity is sufficient to evolve
ideas without imitation (albeit with much reinventing the wheel), it is possible that a lack of
culture (as in animals) could be due to the incapacity for strategic creativity; a society that can
imitate but not invent cannot evolve culture. However the converse is not true; a society that can
invent but not imitate can still evolve culture. Thus the evidence supports the hypothesis that it
was the advent of strategic creativity that gave rise to culture.
But now we come to a paradox not unlike the paradox of the origin of life. How does a mind
come to assume a self-sustained stream of thought that progressively shapes and is shaped by a
worldview? A stream of thought requires each mental entity that enters awareness to activate one
or more memories to evoke retrieval. Analogically speaking, the mind must be traversed with
passageways that connect related concepts like an apple crisscrossed with wormholes. However,
representational redescription is the process that puts related mental entities within reach of one
another; it is what recognizes abstract similarities and restructures the memory to take them into
account. How do you get the wormholes without the worms?
the
109
Liane Gabora
Hello.
110
Chapter Eight
8
Autocatalytic Closure in a Cognitive System
So far we have not strayed very far from considerations of culture in terms of the neo-Darwinian
perspective of evolution as natural selection through replication with variation. However, it was
mentioned earlier that some consider natural selection to be not the only, nor even necessarily the
primary, causal factor in an evolutionary pocess. This chapter departs from a neo-Darwinian
perspective, using concepts from complexity theory—emergence, self-organization, and phase
transitions—to postulate a possible solution to the paradox of the origin of culture.
Since the last chapter hinted that a solution to the question of how culture began may be
found in a solution to how life began, we begin by returning to the paradox of the origin of
biological life. From here on, the origin of life and the origin of culture will be referred to as OOL
and OOC respectively.
8.1
A RETURN TO THE ORIGIN OF LIFE (OOL) PARADOX
Recall the paradox of the OOL: if living things come into existence when other living things give
birth to them, how did the first living thing arise? Reproduction is orchestrated through an
intricate network of interactions between DNA, RNA, and proteins. DNA is the genetic code; it
contains instructions for how to construct various proteins. Proteins, in turn, both catalyze
reactions that orchestrate the decoding of DNA by RNA, and are used to construct a body to
house and protect all this self-replication machinery. However, if proteins are made by decoding
DNA, and DNA requires the catalytic action of proteins to be decoded, which came first? How
could a system composed of complex, mutually dependent parts come into being? It is far from
obvious how the chain of self-replicating systems that eventually evolved into you and I got
started.
111
Liane Gabora
8.1.1
A First OOL Hypothesis: Prebiotic Soup
The most straightforward explanation is that life originated in a ‘prebiotic soup’ where, with
enough time, the right molecules collided into one another at the same time and reacted in exactly
the right ways to create the DNA-RNA-protein amalgam that is the crux of life as we know it.
Proponents argue that the improbability of this happening does not invalidate the theory because
it only had to happen once; as soon as there was one self-replicating molecule, the rest could be
copied from this template. Miller [1955] increased the plausibility of this hypothesis by showing
that amino acids, from which proteins are made, form spontaneously when a reducing22 mixture
of oxygen, hydrogen, carbon, nitrogen, water, and ammonia is subjected to high energy. These
molecules were all likely to have been present on the primitive earth, and energy could have come
in the form of electric discharges from thunderstorms, ultraviolet light, or high temperatures
generated by volcanoes. Other experiments have shown that the molecular constituents of DNA
and RNA, as well as the fatty acids from which membranes are constructed, can be formed the
same way.
Unfortunately, the complexity of the DNA-RNA-protein structure is so great, and in the
earth’s early atmosphere the concentrations of the necessary molecules were so dilute, that the
probability of life originating this way is infinitesimally low. Hoyle and Wickramasinghe [1981]
likened it to the probability that a tornado sweeping through a junkyard would spontaneously
assemble a Boeing 747.
8.1.2
A Second OOL Hypothesis: Ribozymes
The discovery of ribozymes—RNA molecules that, like proteins, are capable of catalyzing
chemical reactions—brought hope that the first living molecule had been found. With ribozymes
you wouldn’t need DNA or proteins to establish a self-replicating lineage; these RNA molecules
would do the job of both. The less complex something is, the more feasible its spontaneous
generation.
In practice, however, the self-replication of RNA is fraught with difficulties. It tends to fold
back on itself creating an inert, tangled mess [Joyce, 1987]. Furthermore, the probability of a
ribozyme assembling spontaneously from its components is remote [Orgel 1987], and even if it
managed to come into existence, in the absence of the error-detecting proteins found in all
modern-day organisms, error catastrophe would occur; that is, its self-replication capacity would
22
A reducing atmosphere is one where there is no free oxygen present. The presence of ferrous (FeO)
rather ferric (Fe2O3) iron in primitive rock leads us to believe that the earth’s atmosphere was reducing
when life began. (It isn’t now.)
112
Chapter 8: Autocatalyic Closure in a Cognitive System
inevitably break down in the face of accumulated error over successive generations [Eigen and
Schuster 1979].
8.1.3
Third OOL Hypothesis: Autocatalytic Closure of a Set of Catalytic Molecules
Kauffman [1993] suggested that knowing as much as we do about what life is like now may get
in the way of determining how it began. Accordingly, he decided to focus on how to get from no
life at all to any kind of primitive self-replicating system, and hand the problem of getting from
there to DNA-based life, over to natural selection and processes such as self-organization. Given
the conditions present on earth at the time life began, how might some sort of self-replicating
system have arisen?
Kauffman’s answer is that life may have begun, not with a single molecule capable of
replicating itself, but with an autocatalyticaly closed set of collectively self-replicating molecules.
An autocatalytic set of molecules is one that, as a group, catalyze their own replication. None of
the molecules can replicate itself, but each molecule can catalyze23 the replication of some other
molecule in the set, and likewise, its own replication is induced by some other member of the set.
A set of polymers wherein each molecule’s formation is catalyzed by some other polymer is said
to exhibit closure. (Note that it is not closed in the sense that new molecules cannot be
incorporated into the set.) This kind of dual role as both ingredient (or stimulant) and product of
different chemical reactions is not uncommon for polymers such as protein and RNA molecules.
The idea of autocatalysis is related to the concept of autopoiesis. An autopoietic system is one
that constantly regenerates itself through the interactions and transformations of its parts, thereby
establishing a boundary between itself and its environment, thus counteracting disintegration due
to entropy [Maturana & Varela 1980, 1987; Rosen 1985, 1991; Varela, 1979]. Kauffman’s
proposal combines the concept of organizational closure [Lee et. al. 1996, 1997; Rosen 1985,
1991; Varela 1979] with insights from random graph theory [Erdos and Renyi 1959, 1960]. This
section summarizes the basic idea.
Let us begin by considering the simplest possible autocatalytic set. Thus if A catalyzes the
conversion of X to B, and B catalyzes the conversion of Y to A, then A + B comprise an
autocatalytic set (Figure 8.1). In an environment rich in X and Y, A + B can self-replicate.
23
A catalyst is a molecule that speeds up a chemical reaction that would otherwise occur very slowly. The
process by which a catalyst speeds up a reaction is referred to as catalysis.
113
Liane Gabora
A
B
X
Y
Figure 8.1 An autocatalytic set: A catalyses the formation of B, and B catalyses the formation of A. Thick arrows represent
catalyzed reactions. Thin arrows represent catalysis.
It is of course highly unlikely that two polymers A and B that just happened to bump into one
another would happen to catalyze each other. However, this is more likely than the existence of a
single molecule catalyzing its own replication. Furthermore, Kauffman shows that when polymers
interact, their diversity increases, and so does the probability that some subset of the total reaches
a critical point where there is a catalytic pathway to every member.
8.1.3.1
Weaving Catalytic Molecules into an Interconnected Web
Kauffman conveys the basic idea intuitively as follows (Figure 8.2). Spill some buttons on the
floor. Tie two randomly chosen buttons together with a thread. Repeat this again and again. Every
once in a while, lift a button and see how many connected buttons get lifted. After a while,
clusters emerge. The clusters get larger. Eventually they join together forming one giant cluster
that contains most of the buttons.
(a)
(b)
(c)
(d)
Figure 8.2 (a) A set of loose buttons. (b) Tie two randomly chosen buttons together with a thread. (c) Repeat over and
over. Occasionally lift a button and see how many connected buttons get lifted. (d) Increasingly large clusters emerge, and
eventually reach a point where they form one giant cluster containing most of the buttons.
Kauffman adapts this basic idea as follows. Say the nodes (buttons) in our system are various
different catalytic polymers such as those that would have been around at the time of the origin of
life. Say the edges (strings) are catalyzed reactions. The idea then is that some subset of the total
reaches a critical point where there is a catalytic pathway to every member. To show that this is
114
Chapter 8: Autocatalyic Closure in a Cognitive System
true, we must show that R, the number of reactions by which they can interconvert, increases
faster than N, their total number. Given polymers made up of, say, two different kinds of
monomers (which can be referred to as 0 and 1), of up to a maximum length of M monomers
each, then N, the number of polymers, is 2M+1 as per equation 5.1 in chapter five. Thus as M
increases—which it obviously does, since two of the longest polymers can always join to form a
longer one—the number of polymers increases exponentially.
Now we need to show that the number of reactions between them increases even faster. We
will be conservative and consider only cleavage (for example, a polymer of length three splits
into one of length two and one of length one, e.g. 110 Ý 1 and 10) and ligation (for example, two
polymers of length two and length one join to form a polymer of length three splits, e.g. 1 and 10
Ý 110) reactions on oriented polymers (such as protein and RNA fragments). The number of
possible reactions R is the product of the number of polymers of a certain length times the number
of bonds, summed across all possible lengths:
R = 2 M ( M − 1) + 2 M −1 ( M − 2) + ... + 2 M −( M − 2 ) ( M − ( M − 1))
=
M
2 n (n − 1)
(8.1)
n=2
Dividing equation 8.1 by equation 5.1, Kauffman found that as M increases, the ratio of reactions
to polymers increases by a factor of M-2. This means that if each reaction has some probability of
getting carried out, the system eventually undergoes a phase transition to a state where there is a
catalytic pathway to each polymer present. The probability of this happening shifts abruptly from
highly unlikely to highly likely as R/N increases.
This kind of sharp phase transition is a statistical property of random graphs and related
systems such as this one. Random graphs consist of dots, or ‘nodes’, connected to each other by
lines or ‘edges’. As the ratio of edges to nodes increases, the probability that any one node is part
of a chain of connected nodes increases, and chains of connected nodes become longer. When this
ratio reaches approximately 0.5, the percolation threshold, the probability that one giant cluster
emerges goes from extremely unlikely to almost inevitable (Figure 8.3). The larger the number of
nodes, the steeper the vertical portion of the resulting sigmoidal curve. Thus some subset of the
total inevitably reaches a critical point where there exists a catalytic pathway to every member.
115
Liane Gabora
Size of
Largest
Cluster
0
0.5
1.0
1.5
Ratio of Edges to Nodes
Figure 8.3 When the ratio of edges to nodes reaches approximately 0.5, short segments of connected nodes join to form a
large cluster that encompasses the vast majority of nodes.
8.1.3.2
Replication of the Primitive Organic Closure System
Now the question is: supposing an autocatalytic set did emerge, how would it replicate and
evolve? The answer is fairly straightforward. It is commonly believed that the primitive selfreplicating system was enclosed in a small volume (such as a coascervate or liposome) to permit
the necessary concentration of reactions [Oparin 1971; Morowitz 1992; Cemin & Smolin, in
press]. Since each molecule is getting duplicated somewhere in the set, eventually multiple copies
of all molecules exist. The abundance of new molecules exerts pressure on the vesicle walls. This
often causes such vesicles to engage in a process called budding, where it pinches off and divides
into two ‘twins’. So long as each twin contains at least one copy of each kind of molecule, the set
can continue to self-replicate indefinitely. Replication is far from perfect, so an ‘offspring’ is
unlikely to be identical to its ‘parent’. Different chance encounters of molecules, or differences in
their relative concentrations, or the arrival of new ‘food’ molecules, could all result in different
catalysts catalyzing a given reaction, which in turn alters the set of reactions to be catalyzed. So
there is plenty of room for heritable variation.
Selective pressure is provided by the constraints and affordances of the environment, and by
other parts of the system. For example, say an autocatalytic set of RNA-like polymers arose.
Some of its offspring might have a tendency to attach small molecules such as amino acids (the
building blocks from which proteins are made) to their surfaces. Some of these attachments
inhibit replication and are selected against, while others favor it and are selected for. We now
have the beginnings of the kind of genotype-phenotype distinction seen in present-day life. That
is, we have our first indication of a division of labor between the part that interacts with the
environment (the proteins), and the part of the organism concerned with replication (in this case,
an RNA-based code). The advent of this code is significant because it enables informative acts to
be carried out recursively, hierarchically, and with greater precision. Another significant
116
Chapter 8: Autocatalyic Closure in a Cognitive System
development is the formation of hypercycles, wherein one autocatalytic cycle increases the selfreproduction rate of another, thus paving the way for more complex ensembles [Eigen & Schuster
1977, 1978, 1979].
8.1.3.3
Theoretical and Experimental Support for Autocatalytic Origin of Life Theory
Of course, even if catalytic closure is theoretically possible, we are still a long way from knowing
that it is the correct explanation for the origin of life. How likely is it that an autocatalytic set
would have emerged given the particular concentrations of chemicals and atmospheric conditions
present at the time life began? In particular, some subset of the R theoretically possible reactions
may be physically impossible; how can we be sure that every step in the synthesis of each
member of an autocatalytic set will actually get catalyzed?
Kauffman’s response is: if we can show that autocatalytic sets emerge for a wide range of
hypothetical chemistries—i.e., different collections of catalytic molecules—then the particular
details of the chemistry that produced life do not matter so long as it falls within this range. We
begin by noting that, much as several different keys sometimes open the same door, each reaction
can be catalyzed by, not a single catalyst, but by many different catalytic molecules, with varying
degrees of efficiency. So we assign each polymer an extremely low a priori random probability P
of catalyzing each reaction. The lower the value of P, the greater M must be, and vice versa.
Kauffman shows that the values for M and P necessary to achieve catalytic closure with a
probability of > 0.999 are highly plausible given the conditions of early earth. Error catastrophe is
unlikely because since each reaction can be catalyzed not by a single catalyst but many, an error
in one reaction does not have much effect on the set at large24.
Furthermore, experimental evidence for this theory using real chemistries [Lee et al. 1996,
1997; Severin et al. 1997], and computer simulations [Farmer et al. 1986] have been
unequivocally supportive. Farmer et al. showed that in an ‘artificial soup’ of information strings
capable of cleavage and ligation reactions, autocatalytic sets do indeed arise for a wide range of
values of M and P. Figure 8.4 shows an example of one of the simplest autocatalytic sets the
simulation produced. The original polymers from which an autocatalytic set emerges is referred
to as the ‘food set’. In this case it consists of 0, 00, 1, and 11. As it happens, the autocatalytic set
that eventually emerges contains all members of the original food set. This isn’t always the case.
24
See [Kauffman 1993] for an interesting discussion of why error catastrophe becomes a serious problem
as the parts of the system becomes more co-adapted.
117
Liane Gabora
0010011
1
00100
1
010
1
100
1
100
0
001
0
0010
0
101
1
1
00
000010
0
011
01
11
0
100
10010
1
10
101
1001110
1
1001
1
Figure 8.4 A typical example of a small autocatalytic set in Farmer et al.’s simulation. Reactions are represented by thin
red dots on black lines connecting ligated polymers to their cleavage products. Green arrows indicate catalysis. Dark
ovals represent food set.
Here we see clearly how autocatalysis generates not just novel informative agents, but exactly
those whose information-providing potential can be exploited by what already exists to turn the
system into a unified whole! Notice that the original ‘food set’ polymers are the simplest ones in
the set, and that the direction of novelty is generally outward; that is, simpler polymers ligating to
form more complex ones.
An interesting question explored in this simulation is: once a set of polymers has achieved
autocatalytic closure, does that set remain fixed, or is it able to incorporate new polymer species?
They found that some sets were subcritical—unable to incorporate new polymers—and others
were supracritical—incorporated new polymers with each round of replication. Which of these
two regimes a particular set fell into depended on both P and the maximum length of the food set
polymers.
8.1.3.4
Mutual Decoding of Parts of a ‘Collective Self’
Kauffman’s proposal solves the problem of how something able to sustain and replicate itself
could spontaneously emerge because an autocatalytic system creates what it needs starting from a
food set of pre-existing polymers. It circumvents the ‘chicken-and-egg’ problem by positing that
the same collective entity is both code and decoder. This entity doesn’t look like a code in the
traditional sense because it is a code not by design but by default. The code is embodied in the
physical structures of the molecules; their shapes and charges endow them with propensities to
react with or ‘mutually decode’ one another such that they manifest external structure, in this case
a copy of its ‘collective self’. Since autocatalytic sets appear to be a predictable, emergent
118
Chapter 8: Autocatalyic Closure in a Cognitive System
outcome in any sufficiently complex set of polymers, the theory suggests that life is an expected
outcome rather than a lucky long-shot.
8.2
ABSTRACT STRUCTURE OF THE TWO PROBLEMS OF ORIGINS
Let us pause to reflect on the intriguing parallels between the two problems of origins that we
have been considering. In each case we have a system composed of complex, mutually
interdependent parts, and since it is not obvious how either part could have arisen without the
other, it is an enigma how the system came to exist. In both cases, one part is a storehouse of
encoded information about a self in the context of an environment. In the OOL, DNA encodes
instructions for the construction of a body that is likely to survive in an environment like that its
ancestors survived. In the OOC, an internal model of the world encodes information about the
self, the environment, and the relationships between them. In both cases, decoding a segment of
this storehouse generates another class of information unit that coordinates how the storehouse
itself gets decoded. Decoding DNA generates proteins that, in turn, orchestrate the decoding of
DNA. Retrieving a memory or concept from the worldview and bringing it into awareness
generates an instant of experience, which in turn determines which are the relevant portion(s) of
the worldview to be retrieved to generate the next instant of experience.
In both the OOL and the OOC, it is useful to think of the relevant class of information units
as states in a state space, each of which can act on a hypersphere of other states with varying
degrees of efficiency. Recall from chapter five that the information-carrying capacity of a system
is highest when the degree to which its parts are correlated or causally connected falls within a
narrow regime between order and chaos, and that evolutionary systems such as life fall squarely
in this regime [Kauffman, 1993; Langton, 1992]. In the autocatalytic OOL model, the requisite
intermediate degree of connectedness arises naturally as a consequence of the fact that the shapes
and charges of polymers endows them with the ability to catalyze some reactions and not others;
thus the distribution of reactions any given polymer can catalyze is approximately Gaussian. Each
polymer had a small, random probability P of catalyzing each reaction. We have seen that
something similar happens in the mind. Here the requisite intermediate degree of connectedness
arises as a consequence of the fact that the particular constellation of stimulus properties
perceived at any instant causes some memory locations to be activated and not others; the lower
the activation threshold, the more memory locations activated, and the greater the cognitive
fluidity. In neural network models of cognition, the radial basis function was used to create a
situation where the distribution of memory locations any given stimulus can activate is, again,
approximately Gaussian; in other words, σ plays a roughly analogous role to P.
119
Liane Gabora
8.3
SELF-ORGANIZING MEMORIES INTO AN AUTOCATALYTICALLY UNIFIED WHOLE
Let us now see if Kauffman’s solution to the OOL can help solve the riddle of the psychological
mechanisms underlying the OOC. We have argued that the existence of the sort of cultural
variation we see around us required the capacity for streams of self-triggered thought that shape
and are shaped by an interconnected web of relationally structured memories and abstractions that
together comprise a model of reality, or worldview. We want to determine how such a complex
entity could come to be. Donald claims that the transition to a mind capable of abstract thinking
“would have required a fundamental change in the way the brain operates.” Drawing from
Kauffman’s OOL scenario, we posit that cultural evolution began, like biological evolution, with
the emergence of a collective autocatalytically closed structure that acts as both code and decoder.
(Note that although this account will focus on the integration of a worldview through the
emergence of deeper, more general concepts, in a pschological context it applies equally to
integration of the psyche through the purification of intentions and emotions.)
In the OOL case, Kauffman began by asking: what was lying around on the primitive earth
with the potential to act as the ‘food set’ of a primitive self-replicating system? The most
promising candidate is catalytic polymers, the molecular constituents of either protein or RNA.
Here we ask: what sort of information unit does the associative mind have at its disposal? It has
episodic, procedural, and semantic memories. Memories then constitute the food set of our
system. Thus, memories of past experiences play the role of the nodes (buttons), and the role of
the edges (strings) is played by associations that enable one memory to evoke a reminding of
another.
Next, Kauffman asked: what happens to the food set to turn it into a self-replicating system?
In the OOL case, food set molecules catalyzed reactions on each other that increased their joint
complexity, eventually transforming some subset of themselves into a collective web for which
there existed a catalytic pathway to the formation of each member molecule. Let us see if an
analogous process could transform an associative mind into a culture-evolving one.
8.3.1
What is the Associative Mind Lacking?
We will make the situation more concrete by returning to the example of Oga, the cave girl.
Consider the situation wherein the sight of a rotting, striped, bumpy, red gourd reminds Oga of
the striped, bumpy, yellow gourd that her beau Og, once used to carry water. Oga slashes the top
off the red gourd and scoops water into it. To her dismay, the water leaks out through a soft decay
spot. Just out of sight lies the intestine of a recently killed water buffalo. What sort of cognitive
dynamics would prompt Oga to tie one end of the intestine and use it as a waterbag? It is unlikely
120
Chapter 8: Autocatalyic Closure in a Cognitive System
that the ability to classify ‘gourd’ and ‘knotted intestine’ as potentially-substitutable instances of
the category ‘container’ is hardwired. No one in Oga’s tribe has previously conceived of an
intestine as a container, so social learning is not an option.
This task lies beyond the horizon of what the associative mind can accomplish. It involves a
number of difficult skills including uncued retrieval, redescription, and manual dexterity, and
perhaps most importantly, analogical reasoning requiring knowledge of abstract relationships.
Recall from the previous chapter that although the associative mind—such as that of an
animal—possesses some imitative abilities, it is unable to strategically generate and refine ideas.
It could not retrieve the memory that an intestine is in the cave, much less realize that it is
relevant to the goal of transporting water. What specifically is the associative mind lacking?
Our model of cognition from chapter five suggests a number of possible answers. First, the
resolution of the perceptual apparatus might not be high enough to capture enough features of
salient stimuli (M is too small). Second, there might not be enough memory locations to keep
these distinctions intact during storage (L is too small). Third, the density s/N of stored mental
entities might be too low. In other words, there might not be enough different basins of attraction
for mental entities to slide into, or not enough of these attractors are occupied. Another possibility
is that the activation threshold is too high (and thus σ too narrow), and thus any given stimulus
activates relatively few memory locations. The end result is the same in all cases: the memory
locations activated by a new experience do not overlap much with the memory locations activated
by any other previous experience. In other words, rarely does any stimulus activate enough of the
same memory locations where a mental entity is stored to cause a retrieval or reminding event to
occur. Thus stored mental entities must be virtually identical to a given stimulus to be within
‘retrievable’ distance of it. Since memory is content-addressable (which for a distributed memory
like the brain means that similar or related stimuli or events activate overlapping regions) this
means that only experiences with virtually identical contents can evoke reminding events such
that one causes a retrieval of the other. Experiences that are related in subtle or abstract ways
cannot do this because such relationships are not taken into account in the way they are stored in
memory. Since the memory does not encode these sorts of relationships, rarely can a stream of
interrelated thoughts ensue. In fact, these explanations are interrelated. M limits L, which in turn
limits s. And since if σ = 0, the memory only retrieves mental entities that are identical to the
subject of attention, and therefore cannot form abstractions, σ also limits s.
As noted earlier, the penalty for having too low an activation threshold is very high. Each
thought has little relevance to the one that preceded it; thinking is so garbled that survival tasks
are not accomplished. On the other hand, too high an activation threshold is not life-threatening.
121
Liane Gabora
Attention is virtually always directed toward external stimuli or internal drives, and memory is
reserved for recalling how some goal was accomplished in the past. This may be the situation
present in most brains on this planet, and though not harmful, it has its own drawbacks. A stream
of thought dies out long before it produces something creative. However, this may not be of
practical consequence to other species. The advantages of a stream of thought would largely be
lost on animals because, as mentioned in chapter three, they have neither the vocal apparatus nor
the manual dexterity and freedom of upper limbs to implement creative ideas. (Language, for
instance, drastically increases the degrees of freedom of what can be expressed.) Moreover, in an
evolutionary lineage there is individual variation, so the lower the average activation threshold,
the higher the fraction of individuals for which it is so low that they do not survive.
8.3.2
What if Oga’s Activation Threshold is Lower?
Oga, however, would be able to implement creative ideas if she could conceive of them. She is
capable of a high degree of manual dexterity, which would enable her to generate complex
signals, displays, and artifacts. Moreover, since she walks upright, she would be able to take the
fruits of her labor with her wherever she goes, and to convey information to others of her clan
through signals and displays.
Let us consider what would happen if, due to a genetic mutation, Oga’s activation threshold
were significantly lower than average for her tribe, causing the storage and retrieval of memories
to become more widely distributed. The cariability of mental entities activated in response to any
given experience is larger, and therefore a larger portion of the contents of memory merge and
surface to awareness in the next instant. Thus we assume that Oga’s s mental entities are stored in
her L memory locations in a distributed manner with a Gaussian distribution of width σ, and this
value of σ is greater than it has ever been before in the mind of any previous human. Thus the
probability that one mental entity evokes another is determined by σ, rather than a random
probability P as we saw with the origin of life, but the idea is the same.
8.3.3
Reminding Events
When mental entity X goes fishing in memory for mental entity X', Oga’s large hypersphere is
bound to ‘catch’ another stored mental entity. The smell of one pine tree evokes a memory of
another, the sound of one bird evokes a memory of another, and so forth. To take a slightly more
complex example, let us return to Oga’s gourd. The sight of the red gourd registers as a vector of
features which determines the path of the train of activation through Oga’s memory network,
which in turn determines the hypersphere of memory locations where ‘red gourd’ is stored. The
122
Chapter 8: Autocatalyic Closure in a Cognitive System
process of storing to these memory locations triggers retrieval from these locations of whatever
has been stored in them. Of course, nothing is retrieved from them if, after ‘red gourd’ is stored,
Oga’s attention is directed toward some stimulus or biological drive. But to the extent that
memory contributes to the next instant of awareness, storage of ‘red gourd’ activates retrieval of
not only ‘red gourd’ itself but all other memories stored in the same locations. The next mental
entity to be experienced is found by combining the contributions of all retrieved mental entities in
parallel feature-by-feature as in chapter six. Whereas the retrieved copies of ‘red gourd’ reinforce
one another, the other retrieved mental entities contribute less, and are statistically likely to cancel
one another out. They do not cancel out exactly, however, unless the distribution of stored mental
entities across the activated locations is uniformly dense. In this case it is not, because of the
existence of a mental entity that got stored when she saw her brother carrying water in a yellow
gourd, which we might refer to as ‘yellow gourd container’. (Oga herself does not label it as such
since she cannot speak or understand language.) The result is that the next mental entity ends up
being ‘red gourd container’. Though it is a reconstructed blend, something Oga has never actually
experienced, it can still be said to have been retrieved from memory.
8.3.4
An Abstraction Emerges
In fact, when mental entity X goes fishing in memory for mental entity X', sooner or later the large
hypersphere is bound to catch a stored mental entity that is quite unlike X. The next instant of
experience consists in identifying an abstract relationship between two (or more) experiences.
Thus we see the emergence of a new kind of node emerges—abstract concepts or categories such
as ‘tree’ or ‘deep’, as we saw in chapter six.
Let us make this more concrete using our example. Oga pours water in the red gourd and it
leaks out. Her mental model of the world was in error; not all gourds can transport water.
Stymied, memory is probed again, with knowledge of relationships between objects and attributes
guiding the process. The second probing occurs with intensified activation of the pathway leading
from the ‘concave’ slot of the focus, and inhibition of the ‘permeable’ slot. Let us now zero in on
the portion of Oga’s memory that deals with four discrete features—bumpy, striped, permeable,
and concave (Figure 8.5). ‘Bumpy’, ‘striped’, ‘permeable’ and ‘concave’ lie on the x1, x 2, x 3, and
x4 axes respectively, and a black dot represents the center of a distributed region where a mental
entity is stored. The second probing of memory activates a slightly different set of locations,
which causes her to collapse on the abstraction ‘container’, the class of objects that are concave
and impermeable, and for which the attributes bumpy and striped are irrelevant. ‘Container’ was
implicit in the space; it covered the central square region of the original hypercube. More
123
Liane Gabora
generally, we can view an n-property conceptual space as a set of nested hypercubes, such that
implicit in the outermost hypercube of mental entities with all n properties there exist hypercubes
of mental entities with n-1 properties, n-2 properties, et cetera.
PERMEABLE
1111 Rotten Gourd
STRIPED
1101 Gourd
**01
Container
0001 Knotted Intestine
CONCAVE
BUMPY
Figure 8.5 Hypercube representing a portion of conceptual space. ‘BUMPY’, ‘STRIPED’, ‘PERMEABLE’ and ‘CONCAVE’
lie on x1, x 2, x 3, and x4 axes respectively. Three concepts are stored here: ‘’Knotted Intestine, ‘Gourd’ and ‘Rotten Gourd’.
Black-ringed dots represent centers of distributed regions where they are stored. Dffuse white area shows region
activated by ‘Rotten Gourd’. Oga’s low activation threshold enables emergence of abstract concept ‘container’ implicit in
the conceptual space (central square region).
This evocation of ‘container’ by the gourd episode isn’t much of a stream of thought, and it
doesn’t bring her much closer to an interconnected conceptual web, but it is an important
milestone. It is the first time she ever derived a new mental entity from other mental entities, her
first creative act. Once she has identified this abstraction, and stored it in memory, it can be
manipulated in much the same manner as a concrete episode. Much as catalysis increases the
number of different polymers, which in turn increases the frequency of catalysis, reminding
events increase concept density by triggering abstraction, which in turn increases the frequency of
remindings.
Here one might point out that, whereas in Kauffman’s model, ligation connected two simple
polymers into a more complex, thus more informative one (through what, from an engineering
perspective, amounts to AND gates), abstraction turns two or more representations into a simpler,
thus less informative one (through what amounts to OR gates). Actually, this isn’t the case. First,
the contributing mental entities do not disappear from memory, even if they don’t singlehandedly dominate the next instant of experience. So, in effect, the memory ends up containing
more information than was explicitly fed into it. Second, creatively combining lower dimensional
concepts into a higher dimensional one also takes place. (I am not focusing on that here, since
124
Chapter 8: Autocatalyic Closure in a Cognitive System
abstraction plays a stronger role in connecting experiences into a web.) Furthermore, it is not
information-providing potential per se that is relevant, but the ability to provide information that
can be used by the system. Larger polymers are not necessarily better catalysts, and catalysis is
how they inform. The fact that, being longer, there are more ways it can be broken down into
other polymers, is secondary; it is only relevant to the extent that these breakdown products
catalyze reactions in the system. Similarly, abstractions are informative to the extent that they
participate in the flow of conscious thought. The more abstract a concept, the greater the number
of other concepts that fall within a given Hamming distance of it and therefore are potentially
evoked by it in a collapse. So, while ostensibly less informative per retrieval, abstractions can
participate in exponentially more retrieval events.
8.3.5
Establishing a Stream of Thought
Once ‘container’ has been evoked and stored in memory, the locations involved habituate and
become refractory (so, for instance, ‘container’ does not recursively evoke ‘container’). However
locations storing mental entities that have some of the features of containers, but that were not
involved in the storage of ‘container’, are still active. Thus features might evoke memories of
other container-like entities, such as ‘seed pod’, et cetera, strengthening associations between the
abstract category and its instances. Other abstractions, such as ‘animal’, form in analogous
fashion. As Oga accumulates both episodic memories and abstractions, the probability that any
given attended stimulus is similar enough to some previously-stored mental entity to evoke it
increases. Therefore reminding acts increase in frequency, and eventually become streams of
remindings, which get progressively longer. Oga is now capable of a train of thought. (Little does
she know that, many thousands of years later, the proclivity to engage in a stream of thought will
become so firmly entrenched that it will take devoted yogis years of meditation to even briefly
arrest it.)
8.3.6
Conceptual Closure in Oga’s Mind
As the density of abstractions increases, the number of recall paths amongst them increases
explosively. Thus their joint complexity also increases, as does the probability that they
‘crystallize’ (to use Kauffman’s way of putting it) into an interconnected network. Much as
polymer A brings polymer B into existence by catalyzing its formation, mental entity A brings
mental entity B into conscious awareness by retrieving it from memory. Just as catalytic polymers
reach a critical density where some subset of them undergoes a phase transition to a state where
there is a catalytic pathway to each polymer present, the mental entities reach a critical density
125
Liane Gabora
where some subset of them undergoes a phase transition to a state where each one is retrievable
through a pathway of remindings events or associations. Finally, much as autocatalytic closure
transforms a set of molecules into an interconnected and unified living system, conceptual closure
transforms a set of memories into an interconnected and unified worldview. That is, for any
stimulus that is assimilated and stored in memory, there exists a way of redefining that stimulus in
terms of previously stored stimuli. Memories are now related to one another through a network of
abstract concepts; the more abstract the concept, the greater the number of other concepts that fall
within a given distance of it in conceptual space and therefore are potentially evoked by it. Thus
the brain is no longer just a relay-station for coordinating inputs to appropriate responses; it now
additionally engages in streams of thought that invent new abstractions (through integration), and
refines them to yield responses tailored to the specifics of the situation at hand.
How does an interconnected worldview help Oga manifest the skills that differentiate a
culture-evolving mind from a merely associative one? The capacity to maintain a stream of selftriggered mental entities enables her to plan a course of action, and to refine behavior by
incorporating kinesthetic feedback. The ability to generate abstractions opens up a vast number of
new possibilities for Oga. It allows her to incorporate more of the structure of the world into her
mental model of it. This increases behavioral flexibility by enabling her to define elements of the
world in terms of their substitutable and complementary relationships. For example, several years
after she formed the category ‘container’, she dips into memory again to discover what else
constitutes a member of this category. The closest thing she can come up with is ‘intestine’. She
realizes that the intestine is impermeable and almost concave. Knotted at one end it, ‘intestine’
would constitute another member of the category ‘container’. She could therefore carry water in
it. She runs off to fetch an intestine.
The power of abstraction also enables Oga to express herself artistically by extricating
mental entities from the constraints of their original domain and filtering the resulting pattern
through the constraints of other domains. For example, she can translate the scene before her into
a sequence of motor commands that render it as perhaps a cave painting or stone carving, or
perhaps transform the pattern of information that encodes the sorrow she experienced at her
child’s death into a song.
8.3.7
Hierarchical Levels of Conceptual Closure
In the OOL case, since short, simple molecules are more abundant and readily-formed than long,
complex ones, it made sense to expect that the food set molecules were the shortest and simplest
members of the autocatalytic set that eventually formed. Accordingly, in simulations of this
126
Chapter 8: Autocatalyic Closure in a Cognitive System
process, the ‘direction’ of novelty generation is outward, joining less complex molecules to form
more complex ones through AND operations. In contrast, the cognitive food set molecules are
complex, consisting of all attended features of an episode. In order for them to form an
interconnected web, their interactions tend to move in the opposite direction, starting with
relatively complex mental entities and forming simpler, more abstract ones through OR-like
operations.
The net effect of the two is the same: a network emerges, and joint complexity increases. But
what this means for the OOC is that there are numerous ways of achieving autocatalytic closure,
which convey varying degrees of worldview interconnectedness and consistency on the
individuals who bear them. These ways differ with respect to degree of penetration of the (n-1, n2…)- property nested hypercubes implicit in an n-property conceptual space. Since it is difficult
to visualize the set of nested, multidimensional hypercubes, we will represent this structure as a
set of concentric circles, such that the outer skin of this onion-like structure represents the
hypercube with all n properties, and deeper circles represent hypercubes with fewer properties
(Figure 8.6). Obviously, not all the nested levels can be shown. The centermost location where a
memory or abstraction is stored is shown as a large, black dot.
Increased Abstraction
Object
Rock
Quartz Crystal
Sun
Heavenly Body
Star
Figure 8.6 The role of abstractions in creative thought. For ease of visualization, the set of nested hypercubes
representing the space of possible mental entities is shown as a set of concentric circles, where deeper circles store
deeper layers of abstraction. A black dot represents the centermost storage location for a specific mental entity. ‘Heavenly
body’ is a more general concept than ‘sun’ or ‘star’, and is therefore stored at a deeper layer of abstraction. Grey circle
around each stored mental entity represents hypersphere where the mental entity gets stored and from which the next
mental entity is retrieved.
The outermost shell encodes mental entities in whatever form they are in the first time they are
consciously encountered. This is all the associative mind has to work with. In order for one
127
Liane Gabora
mental entity in this shell to evoke another, they have to be extremely similar at a superficial
level. In a conceptual mind, however, related concepts are within reach of one another because
they are stored in overlapping hyperspheres. ‘Sun’ and ‘star’ might be too far apart in conceptual
space for one to evoke the other directly. However, by attending the abstraction ‘Heavenly Body’,
which ignores the ‘seen at night versus seen during the day’ distinction, the conceptual mind
decreases the apparent Hamming distance between them.
The most primitive level of autocatalytic closure is achieved when stored episodes are
interconnected by way of abstractions just a few ‘onionskin layers’ deep, and streams of thought
zigzag between these superficial layers. A second level occurs when relationships amongst these
abstractions are identified by higher-order abstractions. Et cetera. Reflecting on an idea amounts
to reflecting it back and forth off, considering it and reinterpreting it in the context of concepts of
varying degrees of abstraction (at ‘onionskin layers’ of varying depths). The level of closure
which perhaps has the most profound impact on humanity take place after Oga’s time, with the
advent of language during the Middle / Upper Paleolithic. Language of course dramatically
increases the potential to communicate and manipulate mental entities. Explanations are sought to
fill in blanks or inconsistencies of the worldview, and these sometimes take the form of myths or
religious parables, or are portrayed through art.
8.4
SOCIAL INTERACTION AND REPLICATION OF THE CONCEPTUAL CLOSURE SYSTEM
The above process was described as taking place in a single individual, namely Oga. Very likely
however, as noted by Heylighen (pers. com.), conceptual closure first occurred within a group of
interacting individuals, and only later took place at the level of the individual. The two may coexist. That is, the worldview of the hunter may contain information about the habits of animals
that is not part of the worldview of the midwife, and the worldview of the midwife contain
information about newborn babies that is not part of the worldview of the hunter. Yet because the
members of a social group interact, the knowledge of an entire clan or tribe may form a closure
structure, such that knowledge of any part can become accessible to any individual who belongs
to this group.
At any rate, now that we have an autocatalytic network of mental entities, whether it selfreplicates horizontally through a society, or vertically from one generation to the next, how does it
self-replicate? In the OOL scenario, polymer molecules accumulate one by one until there are at
least two copies of each, and their shell divides through budding to create a second replicant. In
the OOC scenario, Oga and her tribe members share concepts, ideas, stories, and experiences with
each other, spreading their worldviews bit by bit. Categories they had to invent on their own are
128
Chapter 8: Autocatalyic Closure in a Cognitive System
presented to and experienced by others much as any other episode. The others are handed a
shortcut to the category; they don’t have to engage in abstraction to obtain it.
Recall how in Farmer et al.’s OOL simulation the probability of autocatalysis could be
increased by raising either the probability of catalysis or the number of polymers. Something
similar happens here. Even if Oga’s son Bambam has a higher activation threshold than Oga,
once he has assimilated enough of Oga’s abstractions, his mental entities become so densely
packed that a version of Oga’s worldview snaps into place in his mind. Bambam in turn shares
fragments of his worldview with his friend Pebbles, who in turn shares them with others in the
tribe. These different hosts expose their ‘copy’ of these fragments of what was originally Oga’s
worldview to different experiences, different bodily constraints, sculpting them into unique
internal models of the world. Small differences are amplified through positive feedback,
transforming the space of viable worldview niches. Individuals whose activation threshold is too
small to achieve worldview closure are at a reproductive disadvantage, and, over time, eliminated
from the population.
A relationally and hierarchically structured worldview is an invaluable asset when it comes
to biasing the generation of novelty in directions that are likely to be fruitful. But to some degree
it aids imitation too, especially that of actions, vocalizations, or artifacts that are particularly
complex. The two can happen simultaneously. For example, as Bambam imitates a mannerism he
might, on the fly, put his own slant on it. As worldviews become increasingly complex, the
artifacts they manifest in the world become increasingly complex, which necessitates even more
complex world-views, et cetera. Thus a positive feedback cycle sets in.
8.5
UNDER WHAT CONDITIONS DOES CONCEPTUAL CLOSURE OCCUR?
Under what conditions will the transformation from discrete memories to interconnected
conceptual web actually occur? In the OOL case, Kauffman had to show that R, the number of
reactions, increased faster than N, the number of polymers. He found that R/N increased by a
factor of M–2, where M was the maximum number of monomers per polymer. Because of the
highly parallel nature of this system, it was reasonable to equate potential reactions with actual
reactions, and therefore to assume that the new polymers resulting from these reactions actually
exist (and can themselves partake in reactions). Thus for a large range of values of R and N, the
system inevitably reaches a phase transition to a critical state wherein for some subset of mental
entities there exists a retrieval pathway to each mental entity in the subset.
How do we know that streams of thought will do the same thing? We need to show that
some subset of the mental entities stored in an individual’s mind inevitably reach a critical point
129
Liane Gabora
where there exists a retrieval pathway by which every mental entity in that subset can get evoked.
But here, it is not reasonable to assume that all N perceivable mental entities actually exist (and
can therefore partake in retrieval operations). Recall from chapter five that not everything stored
in memory is attended to and experienced at any given time, and that working memory is in fact
very limited. This presents a bottleneck that has no analog in the OOL scenario. As a result,
whereas OOL polymers underwent a sharp transition to a state of autocatalytic closure, the
transition in conceptual inter-relatedness is expected to take place gradually. So we need to show
that R, the diversity of ways one mental entity can evoke another, increases faster than not N but
s, the number of mental entities that have made it through this bottleneck. That is, as the memory
assimilates mental entities, it comes to have more ways of generating mental entities than the
number of mental entities that have explicitly been stored in it.
Under what conditions does R increase faster than s? Abstraction plays the crucial role. To
determine how abstraction affects R, let us assume for the moment that memory is fully
distributed. Clearly memory is not fully distributed—as we saw, distributions are graded or
constrained—but this simplification illustrates trends which also apply to a more realistic model
of memory. We will be conservative and limit the sort of retrieval event under consideration to
remindings amongst episodic mental entities, abstraction, and the redescription of an idea as an
instance of an abstraction. Abstractions have n properties, where n ranges from a minimum of m
to a maximum of M. RA, the number of ways a retrieval can be evoked through abstraction, equals
the number of retrieval paths enabled by an n-dimensional abstraction, multiplied by the number
of n-dimensional abstractions, summed over all values of n from m to M-1. The number of
retrieval paths equals the number of mental entities that are instances of an n-property abstraction
= 2M-n. The number of n-property abstractions is equal to the binomial coefficient of M and n. The
result is multiplied by two since an abstraction can evoke an instance, and likewise, an instance
can evoke an abstraction.

M
M
M
√
√
√√
R A = 2 2 M −m
+ 2 M −( m +1)
+ ... 2
√
√
√
m↵
m + 1↵
M − 1√
↵↵
=2
M −1
n=m
130
2 M −n
M
√
n√
↵
(8.3)
Chapter 8: Autocatalyic Closure in a Cognitive System
The key idea is that abstraction increases s by creating a new mental entity, but it increases R
more, because the more abstract the concept, the greater the number of mental entities a short
Hamming distance away (since |xi - ki| = 0 for the irrelevant properties, so they make no
contribution to Hamming distance). In other words, mental entities with fewer properties enable
exponentially more retrieval paths.
A second thing to note is that as n starts to decrease, the number of possible abstractions for
each value of n increases (up to M/2, after which it starts to decrease). Thus the more likely any
given mental entity is to get activated and participate in a given retrieval. Whereas R increases as
abstraction makes relationships amongst mental entities increasingly explicit, s levels off as new
experiences have to be increasingly unusual in order to count as new, and get stored in a new
constellation of locations. Furthermore, when the carrying capacity of the memory is reached, s
plateaus, but R does not. Taken together, these points mean: the more deeply a mind delves into
fewer-property abstractions, the more the distribution in Figure 5.1 rises and becomes skewed to
the left25. The effect is magnified by the fact that the more active a region of conceptual space, the
more likely an abstraction is to be positioned there, and thus abstractions beget abstractions
recursively through positive feedback loops.
Thus, as long as the activation threshold is large enough to permit abstraction and small
enough to permit temporal continuity, the average value of n decreases, and sooner or later, the
system is expected to reach a critical percolation threshold such that R increases exponentially
faster than s as in Figure 8.3. The memory becomes so densely packed that any mental entity that
comes to occupy the focus is bound to be close enough in Hamming distance to some previouslystored mental entity to evoke it. The memory (or some portion of it) is holograph-like in the sense
that there is a pathway of associations from any one mental entity to any other; together they form
an autocatalytic set. What was once just a collection of isolated memories is now a structured
network of concepts, instances, and relationships—a worldview.
It seems reasonable to suggest that animals are not prohibited from evolving complex
cognition a priori, but that without the physical capacity to generate and manipulate complex
artifacts and vocalizations there is insufficient evolutionary pressure 26 to tinker with the activation
threshold until it achieves the requisite delicate balance to sustain a stream of thought, or to
25
These distances are no longer strictly Hamming distances, because when we allow for abstractions, it is
no longer a reasonable assumption that mental entities are of the same length i.e. have the same number of
properties. However, the idea holds; for any given stimulus there are more mental entities in memory that
share properties with that stimulus.
26
The phrase ‘evolutionary pressure’ can misleading be construed as ‘active force’, as opposed to
‘vulnerability to change’. Unfortunately, I don’t know of another way of referring to it.
131
Liane Gabora
establish and refine the necessary feedback mechanisms to dynamically tune it to match to the
degree of conceptual fluidity needed at any given instant. It may be that humans are the only
species for which the benefits of this tinkering process have outweighed the risks.
8.6
IMPLICATIONS AND EVOLUTIONARY CONSIDERATIONS
The conceptual closure proposal outlined here has implications for not just cultural but also
biological evolution. Here we address some of them.
8.6.1
Slowly Obviating the Need for Hardwiring
We mentioned that animals are hardwired to respond appropriately to certain stimuli, as are
humans. However, the ability of humans to develop worldviews with which they can make
decisions about what action to take may obviate the need for some of this hardwiring. Genetic
mutations that interfere with this hardwiring may not be selected against, and may actually be
selected for, because in the long run they promote the formation of concepts that generate the
same responses but can be used in a more context-sensitive manner. However this increases the
amount of processing necessary to achieve a workable worldview.
8.6.2
The Proposed Transition Might Not Leave Archeological Footprints
Note that the cognitive reorganization proposed here would not necessarily have an effect on the
size or shape of the cranium, and therefore may not be detectable in the archeological record. This
is consistent with the following statement by Leakey [1984]:
The fact that the anatomically modern human populations in the Middle East appear to have
manufactured Mousterianlike technology rather than the innovation-rich tool assemblages so
characteristic of the Upper Paleolithic means that they were modern in form only, and not in
their behavior. The link between anatomy and behavior therefore seems to break.
8.6.3
Worldviews are Primitive Replicators
Recall from chapters two and three that mental entities do not constitute replicators because they
consist of what von Neumann called uninterpreted information—a self description—but not
interpreted information—symbolically coded instructions for how to self replicate; it is we who
do the replication for them. The exciting implication that falls out of the theory of conceptual
closure in the present chapter is that although mental entities do not constitute replicators,
interconnected networks of mental entities—worldviews—do constitute replicators. However
these replicators do not work with the efficiency of a symbolic code. Their self replication is
132
Chapter 8: Autocatalyic Closure in a Cognitive System
primitive and clumsy, like that of autocatalytic sets of molecules prior to the genetic code. During
childhood, many different kinds of situations are bound to arise sooner or later that provide the
appropriate context for a child to be exposed to various mental entities from elders and artifacts.
As these stories, actions, and bits of knowledge get assimilated, elements of the worldviews of its
parents and other influential members of its culture get fitted together in a somewhat, but not
altogether, new way. It is through this process that worldviews replicate, not all at once, but piece
by piece.
8.6.4
Autocatalysis as an Explanation of Evolutionary Origins
Returning briefly to the origin of life puzzle, recall that traditional attempts to explain how
something as complex as a self-replicating entity could arise spontaneously entail the
synchronization of a large number of vastly-improbable events. Proponents of such explanations
argue that the improbability of the mechanisms they propose does not invalidate them, because
they only had to happen once; as soon as there was one self-replicating molecule, the rest could
be copied from this template. However, Kauffman’s theory that life arose through the selforganization of a set of autocatalytic polymers suggests that life might not be a fortunate chain of
accidents but rather an expected event. Although there is much evidence for this hypothesis,
definitive proof that it is the correct explanation of how life began will not be easy to come by.
However, we now have another data point, another evolutionary process to figure into the picture,
that of culture. And as part of this evolutionary process, a relationally-structured web has come
into existence billions of times; in fact, in the mind of every young child. So in this case,
explanations that involve the synchronization of a large number of vastly-improbable events are
unconvincing to say the least! The emergence of conceptual closure in the mind of a child is the
topic of the next chapter.
8.7
SUMMARY
Cultural evolution presents a puzzle analogous to the origin of life: the origin of an internal model
of the world, or worldview, that both generates and is generated by self-sustained streams of
thought. It is the capacity for abstract thought that enables us to plan and predict, to generate
novelty, and to tailor behavior according to context. The question is: until discrete memories have
been woven into a relationally-structured web, or worldview, how can they generate a stream of
thought? And conversely, until a mind can generate a stream of thought, how does it weave its
memories into a worldview? This chapter presented a plausible scenario for how cultural
evolution, like biological evolution, could have originated in a phase transition to a self-organized
133
Liane Gabora
web of catalytic relations between patterns. In this case, the patterns are memories, and the
catalytic events are reminding events.
Kauffman’s proposal that life originated with the self-organization of a set of autocatalytic
polymers suggests a mechanism for how this comes about. When polymers interact, the number
of different polymers increases exponentially. However, the number of reactions by which they
can interconvert increases faster than their total number. Thus, as their diversity increases, so does
the probability that some subset of the total reaches a critical point where there is a catalytic
pathway to every member. Such a set is autocatalytically closed because each molecule can
catalyze the replication of some other molecule in the set, and likewise, its own replication is
catalyzed by some other member of the set.
Much as catalysis increases the number of different polymers, which in turn increases the
frequency of catalysis, reminding events increase the density of the conceptual space by
triggering the emergence of abstractions, which in turn increases the frequency of remindings.
And just as catalytic polymers undergo a phase transition to a state where there is a catalytic
pathway to each polymer present and together they constitute a self-replicating set, mental entities
undergo a phase transition to a state where each memory and abstraction is retrievable through a
pathway of remindings/associations. Together the mental entities now constitute a transmittable
worldview, an internalized tapestry of reality, that both weaves, and is woven by, threads of
experience. Social interaction and artifacts are vital to the process, and ensure that the continued
evolution of mental entities does not hinge on the survival of any particular individual. The
fascinating implication that falls out of this theory is that although mental entities do not
constitute replicators, worldviews—interconnected networks of mental entities—do constitute
replicators, but of a primitive, clumsy sort, like the autocatalytic sets of molecules prior to the
genetic code. They replicate, not all at once, but piece by piece when we pass on knowledge,
actions, and artifacts to children who have not yet some idea of how the various aspects of their
world fit together and relate to one another.
The scenario outlined here is nascent. Putting the pieces together would require the
cooperation of neuroscientists, developmental psychologists, cognitive scientists, sociologists,
anthropologists, archeologists, and perhaps others. Whether or not this theory of the origin of
culture turns out to be correct, it illustrates how the analogy to biology can focus our study of
culture by providing a scaffold around which explanatory theories can be built. Nevertheless, I
know of no other serious attempt to provide a functional account of how cultural evolution got
started. Whether or not the scenario outlined here turns out to be precisely correct, my hope is that
it draws attention to the problem of cultural origins, suggests what a solution might look like, and
134
Chapter 8: Autocatalyic Closure in a Cognitive System
provides a concrete example of how we gain a new perspective on cognition by viewing it as an
architecture that has been sculpted to support a second evolutionary process, that of culture. The
proposal has implications for biological evolution as well. Consistent with Kauffman’s assertion
that the bootstrapping of an evolutionary process is not an inherently improbable event, the ‘it
only had to happen once’ argument does not hold water here because the cultural analog to the
origin of life takes place in the brain of every young child. Autocatalysis may well be the key to
the origin of not only biological evolution, but any information-evolving process.
135
Liane Gabora
the
the
136
Chapter Nine
9
Embryology of One and Many Worldviews
A human individual starts out with instinctive stimulus-response associations. These instincts
reflect the world its ancestors lived in, and thus may be considered tiny fragments of a worldview,
but this worldview does not encode how these fragments relate to one another. Eventually,
however, these fragments of a worldview become woven into a relationally structured whole,
which is able to generate streams of thought that further explore these inter-relations. This chapter
investigates some implications of the hypothesis that this happens through the process of
conceptual closure presented in the previous chapter. The idea, in other words, is that not only is
conceptual closure responsible for the origin of culture, but it must happen anew in the mind of
every young child who is to meaningfully participate in culture. The resulting conceptual network
enables the child to understand not just in a passive sense what the world is like, but in an active
sense how to make its way in this world.
9.1
CREATIVITY INITIALLY HINDERS CONCEPTUAL CLOSURE
Abstraction creates mental entities with fewer properties, which makes the conceptual space
denser, and increases susceptibility to the autocatalytically closed state. On the other hand,
creative blending to form more complex mental entities could interfere with this by opening up
new regions of conceptual space, and thereby decreasing density. If cross-category blending
indeed disrupts conceptual closure, one might expect it to be less evident in younger children than
in older ones, and this expectation is born out experimentally [Karmiloff-Smith 1990]. There is in
fact evidence of a similar shift in human history from an emphasis on ritual and memorization
toward an emphasis on innovation [Donald 1990].
137
Liane Gabora
9.2
WEAVING AND REWEAVING A WORLDVIEW
We now investigate processes through which deeper levels of conceptual closure are achieved,
and through which inconsistencies can induce a ‘ripping apart’ and re-closure of the worldview.
9.2.1
Pros and Cons of Thorough Processing and Variable Fluidity
Recall from chapter six that a low activation threshold would increase the tendency toward long
streams of thought that more deeply process provocative stimuli, while with a high activation
threshold this tendency would not be so great, leaving the individual more alert to new stimuli.
Neither thorough processing of provocative stimuli, nor remaining alert to new stimuli, is
inherently better. However the thorough processing option affords greater possibility of making a
creative contribution to the world. Once the overall framework of a unique idea has been painted
in the broad strokes characteristic of the high-fluidity individual, failure to be understood may
motivate a decrease in fluidity. This affords more delicate control over what gets evoked and
refined. The details can then be fleshed out, grounding the idea more firmly in consensus reality,
so that when it is born it is less vulnerable and more understandable. In sparse regions of
conceptual space one might expect memories to be more widely distributed (to encourage the
formation of abstractions there), and in dense regions, narrower (to permit finer distinctions) as
illustrated schematically in Figure 9.1. By adapting the degree of fluidity to the density of
concepts in the region of the worldview being explored, one can remain at the edge of chaos at all
times. In other words, this enables the degree of connectivity amongst concepts from one instant
to the next to remain relatively constant.
Figure 9.1 A schematic diagram of a sparsely populated region of conceptual space on the left, with abstractions more
widely distributed, and a more densely populated region on the right, with abstractions less widely distributed. The ease
with which a stimulus can make contact with a stored entity is approximately the same in the two regions.
The more variable the degree of fluidity, the more hierarchical levels at which inconsistency can
manifest. Thus the more detailed (both fine/coarse-grained) the memory can potentially become.
138
Chapter 9: Embryology of One and Many Worldviews
However, in general one can expect that the more hierarchical levels that must be climbed to
achieve closure over the assimilated entities, the longer it takes for closure to be achieved. By
analogy, a spider web can be both bigger and more tightly woven, but that takes longer. So it may
take longer for the creative individual’s worldview to become integrated such that there exists an
association pathway from any one item in memory to any other. Nevertheless, when their
memories finally do get woven into a unified whole, the resulting worldview is not only more
unique and self-made, but it also captures more of richness and subtlety of the world we live in.
Support for this comes from the finding that the less rule-driven, and thus the more contextual the
domain, the older the age at which performance peaked, which suggests that it took longer for a
unique closure to be achieved [Simonton 1984, 1999].
9.2.2
Subcritical versus Supracritical Conceptual Closure
Recall that Kauffman differentiates between autocatalytic sets of molecules that are
supracriticalable to incorporate new moleculesand subcritical phaseunable to incorporate
new molecules. He argues that a slow but steady influx of novelty is most adaptive, and thus
complex systems would spontaneously evolve to lie right at the boundary between these two
regimes. Think of how parents interact with a child, talking baby talk when the child is very
young, and then gradually speaking more quickly and using more complex words, ‘feeding’ the
child cultural entities it doesn’t quite but almost understands. This kind of parental behavior may
be analogous to handcrafting new polymers that are particularly suited to be readily integrated
into a particular autocatalytic set. It may in effect keeps the child’s mind perpetually poised at the
threshold between subcritical and supracritical, such that there is a steady influx of novelty, but
not so much that the new cultural entities disrupt the conceptual structure built up thus far.
Varying the degree of conceptual fluidity by altering the activation threshold is one way the
cognitive system could oscillate between subcritical and supracritical, depending on whether the
demands of the situation it faces. The distinction between subcritical and supracritical thus sheds
a new light on the creative process discussed in chapter six. During the inspirational phase, the
conceptual system may become supracritical, a state more conducive to the incorporation and
spontaneous generation of novelty. Then during the refining of the idea, by lowering the
threshold, the system becomes increasingly subcritical, a state more conducive to the stabilization
of existing structures.
139
Liane Gabora
9.2.3
Annealing on a New Worldview
Piaget [1952] distinguished between the assimilation of new experiences into the
worldview—integration of the new mental entity into the existing structure—and the
accomodation of the worldview to be able to incorporate a mental entity that is inconsistent with
the existing worldview. The latter is clearly more extreme. When inconsistencies in one’s
worldview abound, large-scale worldview renovation is in order. It seems reasonable to expect
that at these times there is a tendency to temporarily ‘loosen’ one’s internal model of reality,
weaken inter-concept relationships, so as to allow new insights to more readily percolate through
and exert the needed revolutionary impact. The notion of decreasing the degrees of freedom as
the creative or problem solving process proceeds, presented in chapter six, is seen in simulated
annealing, a mathematical technique for solving problems with multiple, frequently conflicting
criteria. This kind of problem is particularly difficult to solve because the trade-offs between the
multiple objectives or constraints are often unknown. The best that can be done is to find a Pareto
‘optimal’ solution—one that cannot be improved with respect to any one objective without
worsening some other objective [Steuer, 1986]. A compromise is reached. Simulated annealing
was inspired by the physical process of annealing, during which a metal is heated and then
gradually cooled. The shapes and charges of the atoms confer varying degrees of attraction and
repulsion toward one another. Configurations that maximize attractive forces and minimize
repulsive ones are most stable; thus, the shapes and charges of the atoms define the constraints of
the system. Heating decreases the stability of the forces that bind the atoms together—it loosens
global structure. The subsequent cooling has the opposite effect. As the temperature is lowered,
the correlation length—that is, how far apart the components of a system must be before their
mutual information falls to zero—increases. The result is that a perturbation to any one
component can percolate through the system and affect even distant components. The slower the
cooling process, the more opportunity the atoms have to settle into a stable, low-energy
arrangement.
Simulated annealing computationally mimics this physical annealing process by decreasing
and then gradually increasing the stability of the connections amongst the parts of a system in a
series of either random or deterministic updates. If there are too few update steps, the system
settles on a state wherein few of the constraints imposed by the structure and dynamics of its
components are met. It may, for example, result in islands of mutually compatible components
which are themselves incompatible, in which case the system has difficulty functioning as a
whole. The greater the number of updates, the more harmonious the state the system eventually
settles into, i.e., the more likely it is to find a Pareto ‘optimum’.
140
Chapter 9: Embryology of One and Many Worldviews
Simulated annealing has been used to improve the performance of a neural network [Cohen
1994; Cozzio 1995]. The evidence of variable fluidity suggests that something analogous happens
in the human mind. The individual asks questions and becomes receptive to new ways of
perceiving the world. Then one slowly ‘anneals’ as the details of how to best structure this new
and improved worldview fall into place. This could be achieved by decreasing the activation
threshold; that is, increasing the width of the distribution function, so that there is a higher degree
of conceptual fluidity. Thus there is an increased potential for any thought to trigger a chain
reaction of novel associations. The activation threshold then slowly increases, thereby stabilizing
associations that are consistent and fruitful.
9.2.4
Self-organized Criticality
Note that most experiences are either so consistent, or so inconsistent, with the worldview that
they have little impact on it. Others percolate deep, renewing our understanding of myriad other
concepts or events. The conscious realization of the logical operators ‘and’, ‘or’ and ‘not’, are
expected to exert a significant effect by enabling the willful manipulation of symbols. Other
particularly useful abstractions such as ‘mine’, depth’, or ‘time’, as well as scripts [Schank &
Abelson 1977], and schemas [Minsky 1985], may generate large-scale reorganization of the
conceptual system. Just as in a sand pile perched at the proverbial ‘edge of chaos’ a collision
between two grains occasionally triggers a chain reaction that generates a large avalanche, one
thought occasionally triggers a chain reaction of others that dramatically reconfigures the
conceptual network. This may result in a new and finer grained conceptual closure. Rosch’s
[1978] work on basic level categories suggests that the way we organize information is not
arbitrary but emerges in a way that maximizes explanatory power. It would not be surprising to
find that the number of categories and their degree of abstraction exhibit the same kind of power
law relationship as one finds in other emergent systems [Bak, Tang, & Weisenfeld 1988].
9.3
FRAGMENTING THE WORLDVIEW
Here we look at some phenomena that disrupt integration of the worldview.
9.3.1
Censorship and Repression
Initially a child is expected to be unselective about what it assimilates, since (1) it does not know
much about the world yet, so it has no basis for choosing, and (2) its parents have lived long
enough to reproduce, so they must be doing something right. But excessive receptivity makes the
worldview vulnerable, and over time receptivity is expected to decrease. Just as importing foreign
141
Liane Gabora
plants can bring ecological disaster, assimilation of an alien idea can upset the established state of
harmony in a cognitive network, with unforeseen and potentially harmful consequences (such as
confusion or depression).
Thus eventually a child is expected to develop mental censors that ward off the
internalization of potentially threatening mental entities [Minsky 1985]. Much as biological
closure creates a system that internalizes the food necessary for maintenance and growth yet
shields off toxic substances, conceptual closure creates a system that internalizes seemingly
useful ideas, yet censors potentially harmful mental entities. This includes any realization, idea, or
attitude that has the potential to disrupt the belief structure (such as the idea of natural selection to
a creationist) as well as those that could damage the ego (such as the realization that you are
ugly). It includes any thought that could in some way threaten survival (such as the realization
that you don’t believe in the product your company produces, or a host of others that I can’t tell
you about because my censors prohibit them). Censorship could easily be accomplished by
temporarily increasing the activation threshold, and thereby prematurely terminating assimilation
of the current contents of awareness into the conceptual network.
When a dangerous mental entity gets assimilated despite our censoring mechanism, and we
avoid dwelling on it, this is referred to as repression. The repressed thought is isolated from the
memory at large, much as are the episodic memories in a primate’s brain. Repressing a thought
alters both the probability that it gets evoked (activated into awareness) by other mental entities,
and the probabilities involved in determining which other mental entities are evoked by it once it
has become active. It is like a portion of fabric that is fenced in on all sides by knots. Pulling on a
fiber of the fabric outside the fenced region does not exert much of a pull on that fiber inside the
fenced region, and vice versa; the knots dampen the force of the pull by diffusing it across the
tangled mass of other fibers. Much as erecting a real fence increases the probability that people
will stay on either one side or the other, the repressed thought is either avoided, or dwelt on
excessively. This is consistent with our bipolar attitude toward highly repressed subjects such as
aggression and sexuality, and seems to correspond closely to what psychiatrists refer to as altered
schema valence, wherein specific topics elicit in the patient either latent valency (excessive
avoidance), or hypervalency (excessive preoccupation) [Beck & Freeman 1990].
Recall the discussion of the previous section concerning simulated annealing and the concept
of correlation length. If the above line of reasoning is correct, repression lowers the mutual
information and the average correlation length between mental entities. The individual is less able
to respond spontaneously because contextual information is blocked from spreading through the
network. Much as it would be hard to stay physically balanced if the nerve endings from one of
142
Chapter 9: Embryology of One and Many Worldviews
your legs were blocked, it is hard to stay psychologically balanced if significant portions of your
conceptual network are fenced off. It also provides us with a relatively tangible interpretation of
the common phrases ‘psychologically unstable’, ‘fragmented reality’ and ‘split personality’.
Eventually, however, annealing repressed entities may produce a feeling of unity. Although one
way of annealing the repressed material may be harmful, another way may be found that
enhances the worldview.
These processes may take place not only at the level of an individual’s worldview, but also
at the level of the collective worldview of a society. Once again we have a trade-off. This time it
is between social mores that encourage free thinking, and thereby risk the proliferation of
potentially dangerous thought trajectories, and social mores that discourage free thinking, at the
risk of increased censorship and repression.
It is sometimes suggested repressed material is a source of creativity [e.g. Minsky 1985]. It
makes sense to expect that an idea that has been repressed—for example because it would evoke
unbearable sadness or anger—would be vulnerable to being targeted as an area where worldview
cohesion could be increased. Since at the time the repressed material was experienced it was
prohibited from forming associations to obviously-related ideas, it in turn can not be retrieved
through these expected or straightforward associations. It can only be retrieved via backdoor
entrances, associations that reflect structural congruity at an abstract level. Thus a musician may
come to habitually funnel patterns encountered in a variety of domains—and particularly
repressed material—through modules that filter out hitchhiker and enabler features, and adapt the
core features (or feature schemata) to the constraints of music. It is in this repackaged format that
mental entities originally deemed dangerous can be integrated into the memory at large without
harm, and it is through this process that the creator establishes a sense of control over mental
entities that were previously off limits.
9.3.2
How Deception Invites Worldview Distortion
We have all felt at one time or another the strain of telling a lie, or of living a lie. A reasonable
mechanism underlying dishonesty is that one increases the activation threshold in certain regions
of conceptual space, thereby impeding the natural flow of associations, such that the stream of
expressed thought is fenced off and cannot emanate from these regions. In a sense one
relinquishes ‘integrity’ by splitting into a self that is aware of the real situation, and a self that
pretends that what lies beyond the fenced off regions does not exist. Once you lie, you never
know what other lies will have to be told to maintain consistency. For example, the teenager says
he was at the library doing homework. His mother asks why then does he smell like marijuana.
143
Liane Gabora
He lies again and says that it’s his after-shave that has that smell. And so forth. Much as a fold in
fabric induces folding to the left and to the right of it, deception perpetuates more deception in
nearby regions of conceptual space. The more habitual the deception, the more one then comes to
exist as a collection of disintegrated selves living in different worldview fragments.
9.4
CO-EVOLUTION OF MENTAL ENTITIES AND MULTIPLE WORLDVIEWS
The process of worldview development happens in the context of a social milieu of other such coevolving worldviews, mutually decoding fragments of one another. Recall that worldviews are
primitive replicators. A question acts like an external catalyst that enters the primitive replicator
and ‘catalyzes’ a reaction, a response. Every glimmer of an idea in every mind alters, however
minutely, the structure of a worldview, which, through a smile or spoken word interacts with
other worldviews, and is part of a vast chain of worldviews evolving through space and time.
Mental entities could fool us into thinking we need or identify with them by associating
themselves with other mental entities that represent things we do need or identify with (as
advertisers are well aware). Thus the greater the extent to which we identify with or value
ourselves in terms of the mental entities (including those that pertain to the self) and implemented
artifacts we possess or lack, the more vulnerable we are to ever-more-seductive forms of
persuasion and advertising which tie up time, energy, and resources that could be applied toward
other goals [Brodie 1996]. One way to defend oneself against painful or manipulative mental
entities is to construct what Dennett [1995] refers to as a ‘meme-immunological system’ that is,
formulate new mental entities specifically to deflect these ‘cultural antigens’. However
constructing ‘cultural antibodies’ of this sort is time-consuming, and like any immunological
response it has to be repeated every time the outside agent evolves a counter-response.
9.5
ONE AND MANY WORLDVIEWS
Although we have a common origin in the obvious sense that we are all descended from the same
original first living entity, and in the possible sense that our worldviews are all descended from
Oga’s worldview, we are all separated through the fact that we look out on the world with
different eyes, from different points of view. In this sense we are many separate entities. The
craving we have for intimacy and a sense of one-ness is reminiscent of how physical systems
spontaneously move toward a state of decreased entropy. This craving for intimacy inspires us to
chat, play music together, and engage in other forms of shared activity and communication, and
thereby seam together our worldviews into one.
144
Chapter 9: Embryology of One and Many Worldviews
9.6
A POSSIBLE EXPLANATION FOR CONSCIOUSNESS
Since the dawn of reason (perhaps as long ago as Og and Oga’s time) human have wondered:
what makes one entity conscious and another entity not? In a bold move, but bolstered with
persuasive and influential arguments, Chalmers [1995] proposed that all information has a
conscious, or phenomena aspect. This is referred to as the double aspect theory of information.
Thus he provides a tentative solution to the problem of how complex an entity must be to be
conscious, and what sort of complication grants it consciousness, by positing that
consciousness—or at least, a primal form of it—is in the very building blocks, starting from the
lowest level. One obvious question this speculative theory leaves us with is: how do you get from
phenomenal information to the real McCoy, human consciousness?
Following up on Chalmers’ proposal, in [Gabora 2000, in press] I put forth a tentative
hypothesis that the degree to which an entity is conscious is a function of the degree to which it
amplifies and integrates information. Two transitions in the degree to which information is
amplified and integrated were discussed in the preceding chapter. There have in fact been several
transitions toward increased complexity [Maynard Smith and Szathmary 1994; de Duve 1995 for
reviews] which would also have this effect. If the double aspect theory of information is correct,
with each transition, the degrees of freedom of what could be experienced would correspondingly
increase. Moreover, when sub-entities become autocatalytically closed, as described in the
previous chapter for biological and again for cognitive systems, the amount of information they
are capable of storing and communicating increases. Since such entities are entropy-defying
[Prigogine & Stengers 1984], the direction of information processing in such a system is
nonsymetrically inward-biased; thus information is locally amplified and integrated. Moreover,
they generate not just novel information-processing components, but exactly those whose
information-providing potential can be exploited by what is already in place, thereby enabling the
system to function as a unified whole. Thus if Chalmers’ theory is true, there is a corresponding
phase transition in degree of consciousness; the emergence of autocatalytic structure at the
organic level as Kauffman proposes causes individual bits of phenomenally-endowed information
to integrate their individual subjectivities into a single, concentrated subjectivity. And if the line
of reasoning developed in chapter eight is also correct, in cognitive systems such as ourselves the
effect is magnified through a second, cognitive level of closure, wherein memories and
abstractions are woven into an interconnected conceptual web. Note that since closure fences an
entity off, to some extent, from its environment, it would not only amplify the entity’s own
consciousness, but tend to make the entity underestimate the degree to which other entities are
conscious.
145
Liane Gabora
9.7
SUMMARY
The previous chapter presented a model of the origin of culture through the formation of an
autocatalytically closed conceptual web, or worldview, in one or more interacting individuals.
Here it is suggested that this transformation must repeat itself in the mind of every young child in
order for that child to meaningfully participate in culture. This chapter explored some of the
implications for cognitive development.
Some experiences are either so consistent, or so inconsistent with one’s worldview, that they
have little impact on it. Others assimilate readily with existing mental entities, renewing our
understanding of a myriad other concepts or events. We looked at processes that interfere with the
cohesiveness or integration of the conceptual network, such as censorship, repression, and
deception, and also at processes that cause it to be richer—that is, more deeply penetrated with
abstractions that enable finer levels of distinction—such as cognitive analogs to simulated
annealing and self-organized criticality. Processes such as repression or deception can make one
mind become many, whereas processes such as exchanging ideas or playing music together can
make many become one.
Although conceptual closure may initially be disrupted by creativity, the long term effect of
creativity is the potential for a more finely structured worldview. We saw that autocatalytic sets
can be either subcritical—unable to incorporate new elements—or supracritical—able to
incorporate new elements, and that the most adaptive state is at border between these two
regimes. Parents may interact with children in a way that keeps them poised at the subcriticalsupracritical transition zone. During the creative process, one might slide from supracritical to
subcritical by decreasing the activation threshold and slowly annealing on a worldview that can
consistently incorporate new speculations.
It is noted that autocatalytic closure, first at the organismic level, and then at increasingly
complex cognitive levels, causes local integration and amplification of information. Following
Chalmers’ suggestion that information has a phenomenal or conscious aspect, it is tentatively
hypothesized that each new closure generates a corresponding transition in degree of
consciousness.
146
Chapter Ten
10 What is Missing in Current Evolutionary Theory
We have seen various ways in which the origin and evolution of culture is and is not like
biological evolution. We now start to look at what is necessary for a theory of evolution general
enough to encompass both these processes. In so doing, not only does the rationale for viewing
culture as an evolutionary process become more scientifically grounded, but we also gain a
clearer picture of what is going on in biological evolution, and also processes of physical
evolution.
10.1 REVIEW OF NEO-DARWINIAN THEORY OF EVOLUTION
The term evolution is usually construed as a shorthand for neo-Darwinian evolution—a process of
descent with modification, wherein a species adapts to an environment through repeated iterations
of replication with random variation followed by natural selection. However, as discussed in
chapter two, it is becoming increasingly evident that neo-Darwinian evolution, powerful though it
is, does not provide a comprehensive account of biological change [e.g. Boone 1998; Crutchfield
& Schuster 2001; Kauffman 1993; Newman and Muller 1999; Schwartz 1999]. Moreover, as we
have also seen, adaptive change in response to an environment is not limited to the biological
case; something similar takes place in culture. Physicists too are interested in how micro-physical
entities change, and describe it as a process of either collapse or dynamical evolution. Is there
anything all these supposed evolutionary processes have in common? Perhaps not. On the other
hand, perhaps the time is ripe for a more general framework for evolution, of which biological,
cultural, and physical evolution are but different manifestations.
We will start by briefly reviewing our most abstract formulation to date of what evolution
entails:
147
Liane Gabora
•
A pattern of information (an entity that occupies a state within a space of possible states).
•
A means of randomly varying the pattern (exploring or transforming the space).
•
A rationale for selecting variations that are adaptive, i.e. that tend to perform better than their
predecessors in a given environment (a fitness landscape applied to the space).
Thus, according to the neo-Darwinian picture, the generation of biological variation through
random mutation and recombination brings about a divergence of forms, and natural selection
subsequently induces convergence by weeding out those that are least fit. This picture has served
as an adequate description of evolution for some time. But even a non-biologist, with a minute’s
reflection, can easily see that random variation and natural selection are not the whole story.
Natural selection takes into account the forces that caused your parents to live long enough to
bear children (rather than, say, starve or freeze to death in their youth). But it does not go far
toward accounting for the forces that caused your parents to have you. It does not incorporate the
factors that caused them to fall for each other rather than for someone else, or to make love on
one particular day rather than another, et cetera.
One could argue that for humans things are more complicated because of culture, but that for
other species the assumption of statistical randomness is valid. But this is not the case. Contrary
to Mendel’s law of independent assortment, we find that the generation of variation is not
completely random; convergent pressures are already at work prior to the physical realization of
organisms. For example, since Cairns [1988] initial report, there is increasing evidence and
acceptance of directed mutation, where the frequency of beneficial mutations is much higher than
chance. Furthermore, mating is often assortative—mates are chosen on the basis of traits they
possess or lack, rather than at random—and relatives are avoided as mates. Moreover, processes
other than variation and selection—such as self-organized emergence—cause change of
biological form. Thus the neo-Darwinian model does not even capture what is going on in
biological evolution, let alone cultural evolution, where convergent and divergent processes more
clearly work hand in hand. In order to really understand how culture evolves we are going to need
a more resilient theory of evolution.
10.2 THE NEED TO INCORPORATE POTENTIALITY AND ACTUALIZATION
So let us see if we can do better. The state of an entity at a given instant endows that entity with a
certain capacity to evolve, and makes it more prone to evolve in some directions than others.
Recall from chapter one that by ‘potential’ we do not mean potential in the sense of something
waiting to be actualized in a prefabricated, deterministic sort of way, as biologists sometimes use
148
Chapter 10: What is Missing in Current Evolutionary Theory?
the word. We use the term potentiality to refer to any and all possible ways which, given the
current state of the entity it could evolve. Potentiality encompasses both direction and degree of
possible evolutionary change. It can be viewed as the gulf that lies between what an entity is, and
what, given any environment, context, or situation it might encounter, it could become or give
rise to. The concept of potentiality is related to the concept of genotype in the sense that knowing
the genotype of a biological entity we know something about its potentiality. However,
potentiality is a much more general concept. While an organism is said to possess a single
genotype throughout its lifetime, its potentiality changes with each change of state of that
genotype. For instance, if a certain segment of the genotype either starts or ceases to be
transcribed in response to changes in the internal and external milieu of the organism, its
potentiality changes. Potentiality is also related to the concept of fitness, but again it is more
general in that it is not limited to number of offspring. Moreover, one can speak of the potentiality
of a single gene, an organism, or an entire species. One can also speak of the potentiality of a
concept, or a mind, or a micro-physical particle.
Actualization is the articulation or realization of a state that was previously potential for an
entity. It is related to the concept of phenotype in the sense that both deal with how potentiality
manifests concretely in the world. However, once again, actualization is more general, since it
applies to any sort of entity. Sometimes an actualized result is easily detected, as when an
organism reproduces. Other times the result may not be so readily detected, as when an allergen
changes the state of the genome. However, a relatively invisible newly actualized state may
trigger other changes that do manifest outwardly. For example, the change of state of the genome
in response to the allergen may cause hives. As a cognitive example, the actualization of a state of
mind may manifest as a facial expression, a vocalization, an action, or possibly an invention.
When hereditary information undergoes descent with modification, it repeatedly actualizes
one or another of the forms that became potential for it through its history of previous
actualizations. For example, when your parents conceived you, they actualized a being whose
potential had come about because of their specific genotypes, and all the various circumstances
that led them to (hopefully) fall in love. With each expression of their love, the potential for your
existence came a tiny step closer to being realized. Your conception27 was an event wherein each
parent constituted the most significant element of the other’s environment. Sexual reproduction is
an event with exceptionally high potentiality because of the vast number of ways the genomic
27
Some, such as Philip Polk, appropriately suggest a note of caution here, because the being that came into
existence at this time was not what we could call ‘you’. I refer to it in this way only because it is somewhat
cumbersome to say ‘the being that was conceived at this time that eventually became you’.
149
Liane Gabora
material of two parents can give rise to new form. Moreover, by actualizing offspring, one comes
closer to actualizing the potential for one’s offspring’s offspring, their offspring, and so forth.
We can view natural selection as one factor that influences which of the potential forms a
species is capable of in one generation are indeed realized in the next, and we can view
reproduction as one means of realizing these new forms. However, natural selection is certainly
not the only factor involved in determining which potential forms become actual; phenomena
such as assortative mating mentioned earlier being others. Nor is reproduction the only means of
realizing new forms, self-organized emergence being another. In this case, what is actualized is,
not offspring, but a level of structure for a whole that was collectively potential given the
structure of its parts. Each time one part changed state in a way that reflected the constraints and
affordances of the other parts, this collective structure became a little more clearly articulated.
Since the system becomes increasingly committed to a particular actualization, its potential
decreases.28
The same general phenomenon of reiterated actualization of potential is even more evident
in minds, and the social interactions that occur between them. The generation of an instant of
experience arises through the actualization of something that was potential given the previous
state of the mind, body, and social or asocial environment. The current instant of experience
changes the state of one’s mind, however slightly, in such a way as to affect what could
potentially be experienced in the next instant. Thus, neo-Darwinian evolution through natural
selection seems to be a special case of this more general phenomenon wherein an entity
repeatedly changes by actualizing a new state that was potential for it given its previous state.
10.2.1 Evolution as a Means of Preserving and Enhancing Potentiality
Potentiality can be lost in the process of actualization. This is easily seen in the example from
chapter three of black and white paint. When white paint is blended with black paint to actualize
gray paint, it loses the potential to ever again be as purely white. However, an evolutionary
system not only preserves its potentiality in the course of its actualizations, but actually enhances
it. How does it do this?
Biological systems preserve potentiality through particulate inheritance and parallelism.
Recall from chapter three how a cross of AA x aa will yield all Aa individuals, each of whom can
have offspring that are either AA, aa, or Aa. The fact that AA and aa individuals reappear in the
third generation shows how the particulate nature of hereditary information guards against the
28
See [Shaw & Turvey 1999] or a thoughtful discussion of the relationship between potential, constraint,
and degrees of freedom.
150
Chapter 10: What is Missing in Current Evolutionary Theory?
dilution of potentiality from one generation to the next. The fact that this same process is
happening in parallel in other members of the species ensures that even if a beneficial allele is lost
in one lineage it will probably not be lost in another.
The sparse, distributed, content addressable, hierarchical structure of the mind enables it to
preserve and enhance potentiality to an even greater degree. Preservation is made possible
through the storage and retrieval capacity of memory, and the capacity for learning enables one to
bias the actualization of potentiality in ways that stand a greater than average chance of being
beneficial. (In chapter four this was simplistically implemented in the MAV computer model
using strategic operators.) Potentiality is enhanced, and its actualization further biased, through
abstract thought, strategic planning, imagining, and perhaps intuition. We have seen that in a
stream of thought, thinking along the lines ‘what would happen if....’ can influence the action or
artifact that eventually results. (In MAV this was implemented using mental simulation.) Patterns
at many levels of abstraction, both those which have been directly experienced and those that
could be potentially imagined, can be honed to meet the constraints of a given circumstance, or to
please one’s artistic tastes or fantasies. Potentiality is further preserved by the explicit parallel
actualization of thoughts and ideas in culture, as for example when a politician gives a speech and
millions watch, or even when graffiti is painted on a wall and people walk by and see it and it
affects the streams of thought taking shape in their minds.
10.3 THE NEED TO INCORPORATE CONTEXT AND CONTEXTUALITY
Another factor that affects the evolution of an entity is its environment. The environment and the
selective pressures it exerts continuously generates contexts that cull adaptive potential in an
entity. Natural selection acting on previously-actualized entities is but one type of contextual
influence; other contexts operate on potential form rather than form that has already been
actualized. In the example above, the potential of the Aa individuals gets actualized differently, in
parallel, depending on the genotype of the individual’s mate. In the context of an AA mate, an Aa
individual’s potential becomes constrained to include just AA or Aa offspring. In the context of an
aa mate, the Aa individual has the potential for Aa or aa offspring, and once again some of this
potential might get actualized. And so forth. But while the mate constrains the individual’s
potential, the mate is necessary to actualize some of this potential in the form of real physical
offspring. Thus the genome of the mate simultaneously makes some aspects of the Aa organism’s
potentiality realizable, and others unrealizable. Contextual events occur not only at the interface
between one generation and the next, but also in between. The allergic response considered
earlier is an example. The change of state is brought about through interaction between the
151
Liane Gabora
genome and its environment, which includes signals from within the organism and mediated
signals from outside.
The contextuality of a biological entity derives in part from it being an autocatalytically
closed system, or hypercycle of such closure systems, as discussed in chapter eight. Because such
a structure is richly interconnected (e.g., by way of sensorimotor and endocrine systems), a
perturbation can percolate through the interconnected system and elicit a response specifically
tailored to it. Conceptual closure, also discussed in chapter eight, is a closure in conceptual space
rather than physical space. That is, the mind is conceptually closed in the sense that every
concept, belief, et cetera, impacts and is impacted by, a graded hypersphere of related concepts
and beliefs, and there exists a ‘conceptual pathway’ through streams of associative recall from
any one concept to any other. This second form of closure further enhances contextuality by
enabling the individual to engage in relational streams of associative thought that refine potential
behaviors in light of goals or imagined outcomes. Thus in cognition, contexts can be external, as
in the case of a sensed stimulus, or generated internally, as in the case of mental simulation,
where one considers what would happen if something were to be the case.
10.3.1 Deterministic versus Nondeterministic Contextuality
For the present purposes, the most important parameter concerning context is whether the
interaction between an entity and a context is deterministic or not. In the sort of contextuality we
are interested in, the effect of context is not deterministic; we do not have complete knowledge
about the effect of the context on the state of the entity. An example we will look at shortly is the
effect of a measurement on a quantum entity. Not all change of state due to context is of this sort.
For example, a book dropped from a third floor window is very much affected by the context of
gravity, but the effect of this context on this entity is deterministic.
As is the case with physical systems, in biological, cognitive, and cultural systems not all
change of state is nondeterministic. This is why classical mechanical modes of description have
worked fairly in these fields. We can distinguish between situations that are an unfolding of a
process that is underway, and hence classified as influenced by context in a deterministic way,
and situations that are influenced by context in a nondeterministic way. An example of a
deterministically driven change would be a the change of state from not eating to eating that a
healthy, hungry dog undergoes when food is put before it. This change of state is
deterministically mediated, like the changes of state of a dynamically evolving micro particle
between measurement situations.
152
Chapter 10: What is Missing in Current Evolutionary Theory?
However, consider the situation where something is put before a dog that might just barely
be considered by a dog to be food, such as a shoe. We lack knowledge about whether a hungry
dog will consider the shoe to be food and undergo a change of state from not eating to eating.
This is an example of a nondeterministically mediated change. The sexual reproduction of
offspring discussed earlier is another example. The context consists directly of the interaction
with the mate, but indirectly of the history contextual interactions that affect which of the
offspring an organism could potentially have are actualized, survive, and have offspring of their
own. We noted earlier that natural selection is but one factor influencing which of the forms one
generation could potentially give birth to are indeed realized in the next generation. For example,
natural selection takes into account the forces that caused your parents to live long enough to bear
children, but it does not go far toward accounting for all the forces that caused your parents to
have you. However, these forces are accounted for if we consider how the potential of each
parent was affected repeatedly by contextual interactions with an environment.
The cognitive scientist can similarly distinguish between situations of context driven, but
deterministic evolution, and situations of context-driven, but nondeterministic evolution. If you
are in love with someone and you hear that person’s name, then this is a context-driven but fairly
deterministic change of state. However, if you see an abstract painting and for some reason it
makes you think of the person you are in love with, then this is a context-driven nondeterministic
change of state.
10.4 THE PROBLEM OF INCOMPLETE KNOWLEDGE
Potentiality and contextuality can thus be seen to be two facets of the same underlying
phenomenon: incomplete knowledge of the universe in which an entity is operating. More
specifically, one is forced to think in terms of potentiality when the entity of interest behaves
differently in different contexts, and one does not have complete information about all possible
contexts and their possible effects on the entity. It is possible to ignore the problem of incomplete
knowledge if context has a limited effect on the heritability of the entity, or if all contexts are
equally likely. In biology, neither of these assumptions is too far from the truth. Since acquired
traits are not heritable, the only contextual interactions that exert much of an effect on evolution
are those that affect survival or the procurement of a mate. And although physical traits such as
odors and displays bias decisions about mating to some extent, one doesn’t go too far astray
assuming that any male mates with any female. Thus, in biology, the problem of incomplete
knowledge is not acute. In culture, however, neither does context have a limited effect on the
heritability of cultural entities, nor are all contexts equally likely. Our minds are engaged in an
153
Liane Gabora
ongoing contextual interaction with each other, and these interactions not only affect the
evolution of ideas and artifacts, but they immediately feed back on and affect the potentiality of
our minds.
10.5 SUMMARY
The neo-Darwinian picture of evolution is not general enough to provide a full description of how
culture evolves. It ignores potential; it focuses on natural selection, a process that works on
variation that has already been explicitly actualized. But even in biological evolution, this is just
part of the story. When hereditary information undergoes descent with modification, it repeatedly
actualizes one or another of the forms that has become potential for it because of its history of
previous actualizations, and this happens not just at the interface between one generation and the
next, but between generations as well. Each actualization, in turn, provides constraints and
opportunities for further evolution. This is also what is happening when one refines an idea in the
course of a stream of thought, or when an invention is refined by a society of individuals over
many generations. In all these cases, context plays a role in determining how potential is
actualized. While replication with variation and selection of particulate traits has served as an
adequate theory of evolution for some time, for a complete and accurate theory we require a
formalism that can describe nondeterministic potentiality and contextuality. The need to
incorporate potentiality and the need to incorporate contextuality both arise through the problem
of incomplete knowledge.
154
Chapter Eleven
11 Potentiality, Context, and Change of State
In order to develop a general theory of evolution that takes potentiality and contextuality into
account, it is necessary to introduce these concepts in a more precise way. To do so, we will
borrow from physics, where methods for handling them have already been developed. The ideas
in the next few chapters were developed in collaboration with Diederik Aerts.29
11.1 GENERAL SCHEME FOR CHANGE OF AN ENTITY UNDER THE INFLUENCE OF A CONTEXT
Consider an entity—whether it be a physical biological, mental, or some other sort of entity—at a
certain instant of time t0. At that instant, the entity is in a certain state p(t0). This state incorporates
the various ways this entity would be influenced by the different contexts that could be present at
the instant t0. Each context could change the entity in a particular way, and this is described by a
change of the state p(t0) of the entity to another state p(t1) at time t1. Once the state of the entity
has changed to p(t1), the process repeats itself. The states p(t0), p(t1), p(t2), …, p(ti), … constitute
the trajectory of the entity through the state space, and describe its evolution in time.
11.1.1 Representation of the Trajectory of an Entity
We now represent this type of change in a more explicit manner. The entity starts at time t0 in
state p(t0). It is under the influence of a context e(t0), and as a consequence its state changes to
one of the states p1(t1), p2(t1), …, pn(t1), … . Thus all the states of the set {p1(t1), p 2(t1), …, pn(t1),
… } are potentially the state that the entity will actualize at time t1. Let us suppose that the change
happens and hence that one of the potential states gets actualized, for example state p4(t1). This
means that at time t1, the entity is in state p4(t1), but it could have been in one of the other states of
the set {p1(t1), p2(t1), …, pn(t1), … }. Starting from state p4(t1), the entity is influenced by a
155
Liane Gabora
context e(t1), and it evolves to the time t2. Again there are different potential states that it could
change to. Let us denote the set of all these states by {p1(t2), p2(t2), …, pn(t2), … }. As time
evolves from t1 to t2, one of these states, for example the state p3(t2), becomes actualized. The
process of change then starts all over again.
We say that this change of state is nondeterministic because the actual trajectory of the entity
is one choice amongst many potential trajectories. Concretely what this means is that a given
entity in a certain state p(ti) under a certain context can change to any of the states in the set
{p1(ti+1), p2(ti+1), …, pn(ti+1), … }. This is depicted graphically in Figure 11.1.
Figure 11.1 A graphical representation of the general evolution process. The contexts e(t0), e(t1), e(t2) and e(t3) at different
times t0, t1, t2, and t3, are represented by vertical lines. The different states of the entity are represented by little circles on
the vertical lines. At time t0 the entity is in state p(t0) and under the influence of the context e(t0). Its state can potentially
change to one of the states of the set {p1(t1), p2(t1), p3(t1), p4(t1), … , pn(t1), … } under the influence of the context e(t0).
These potential changes are represented by thin lines connecting the different states. Only one of these potential changes
actually happens. This one is represented by a thick line, i.e. state p(t0) changes to state p4(t1) under the influence of
context e(t0). At time t1 the entity will in general be under the influence of another context e(t1). It is now in state p4(t1) and
will potentially change to one of the states of the set {p1(t2), p2(t2), p3(t2), p4(t2), … , pn(t2), … } under the influence of
context e(t1). Again only one of these changes happens, the one indicated by a thick like, i.e. state p4(t1) changes to state
p3(t2). The process then starts all over again. Under the influence of a new context e(t2), the entity, which is now in state
29
The strict mathematical definitions of terms in this chapter were provided by Diederik Aerts.
156
Chapter 11: Potentiality, Context, and Change of State
p3(t2) can change its state to one of the states of the set {p1(t3), p2(t3), p3(t3), p4(t3), … , pn(t3), … }. Again only one of these
potential changes happens, namely state p3(t2) changes to state p5(t3). The dashed lines from the states that have not
been actualized at a certain instant of time are to indicate that much more potentiality is present at time t0 than we are
able to present in the picture. Indeed, suppose for example that state p(t0) would have changed to state p2(t1) instead of to
state p4(t1), which was possible at time t0. Then the context e(t1) would have a different effect on the entity at time t1
(because the entity is in another actualized state) such that a new vertical line at time t1 would have to be drawn for this
situation, representing another possible pattern of change.
11.1.2 The Concept of a Potentiality State
Consider an entity in a given state p under the influence of a context e. There are two possibilities
for how p will change. The first is that the context has a deterministic influence on the state of the
entity, and thus changes it deterministically to another state. In this kind of situation we will refer
to the state as a deterministic state with respect to the considered context.
The second possibility is that the context has a nondeterministic influence on the state of the
entity, such that this state can change to different possible states. Thus, the indeterminism is due
to lack of knowledge of the context, not lack of knowledge of the state of the entity. In such a
situation, we will refer to the state of the entity as a potentiality state30 with respect to the
considered context. It is specifically this potentiality to change under the influence of context to
many different states that we are getting at when we speak of the state of an entity as a
potentiality state.
As we will see in the next section, the structure of change of state of entities as it appears in
quantum mechanics is exactly of this nature. The situation in biology and culture is not so
different. Much as a quantum entity has the potential to collapse to different other states, a gene
has the potential to mutate into different alleles, a species has the potential to branch off into other
species, a gesture, phrase, or idea has the potential to be adapted to different situations. Note that
a state is always a potentiality state in relation to a certain context. It is possible for a state to be a
potentiality state with respect to one context, and a deterministic state with respect to another
context. A deterministic state with respect to a context can also be considered as a limiting case of
a potentiality state, with zero potentiality.
30
The concept of potentiality state was introduced in [Aerts, D’Hooghe & Gabora, 2000; Gabora & Aerts
2000], and is formalized in a detailed and rigorous way in [Aerts & Gabora 2001]. Neither in standard
quantum mechanics, nor in the generalized quantum theories, is the term potentiality state used. What we
call a potentiality state here is just called a state there. However, we think it is clarifying to refer to the state
explicitly as a potentiality state. This is particularly true in the present approach, because the presence or
absence of potentiality plays an important role in the general framework for evolution we are developing.
157
Liane Gabora
11.2 HOW STANDARD QUANTUM MECHANICS DESCRIBES CHANGE
Now that we have seen the kind of change we need for a theory of evolution that takes seriously
the kind of contextuality present in culture, let us look at formal systems for describing change of
state to see if they can help us describe this kind of change more precisely. It would seem that the
obvious next step would be to see if classical theories, such as perhaps some branch of chaos
theory, or Markov processes, would be useful here. However, before looking at classical theories,
we will look at how change of state is described in standard quantum mechanics. This may seem
like a strange leap, but the reason for it, and the reason for introducing it now rather than after
classical theories, will become clear shortly.
Standard quantum mechanics can in fact describe the type of change presented in the
previous section, although not in a sufficiently general manner. Before we see why, let us look at
how this type of change is described by standard quantum mechanics. The central mathematical
object in quantum mechanics is a complex Hilbert space, which is a vector space over the field of
complex numbers. The unit vectors of this space represent the states of a quantum entity. To show
that these states behave as we have pictured the general behavior of potentiality states, we must
first explain how context appears in standard quantum mechanics.
11.2.1 Dynamical Evolution
In standard quantum mechanics there are two different modes of change. First there is the
dynamical change of the state that takes place when no measurement is executed. This is the
effect of fields present in the rest of the universe and that steer the change of state of the quantum
entity. This dynamical change is described by the Schrödinger equation. It is important to note
that this dynamical change is deterministic. This means that if the quantum entity at a certain
moment of time t0 is in state p(t0), and the only change that takes place is this dynamical change
governed by the Schrödinger equation, then the state p(ti) at any instant of time ti later than t0 is
determined. This also means that any state is a deterministic state with respect to the context that
is present in this dynamical change.
11.2.2 Change of State Due to a Measurement
The second type of evolution in quantum mechanics is the change of state under the influence of a
measurement. Let us now look at what this involves.
158
Chapter 11: Potentiality, Context, and Change of State
11.2.2.1
Eigenstates and Superposition States
In quantum mechanics, a measurement is described by a self-adjoint operator, which is a linear
function on the Hilbert space. A self-adjoint operator always has a set of special states connected
to it, that are called the eigenstates of the operator31 [for technical details, see Aerts & Gabora
2001]. An eigenstate does not change under the influence of the measurement context described
by the considered self-adjoint operator. This means that an eigenstate is a deterministic state with
respect to the measurement context. However, if the state of the quantum entity is a genuine
superposition state 32 of the eigenstates, then the change of state provoked by the measurement is
such that this superposition state changes to one of the eigenstates of the self-adjoint operator
representing the measurement. A superposition state can also be referred to as a pure quantum
potentiality state. That two states can always give rise to a superposition state is called the
superposition principle.
11.2.2.2
Quantum Collapse
The change of state itself is referred to as collapse. If the state of a quantum entity with respect to
a certain context is a superposition state, then its various possible future states are irretrievable as
separate states until the measurement evokes a collapse onto one or another of them.
The probability that a superposition state collapses to a specific eigenstate is related to the
weight of the vector representing the genuine superposition state in its linear sum over the vectors
representing the eigenstates. In general, depending on how many of the weights are different from
zero, many of the eigenstates are possible states to collapse to under the influence of this
measurement; in other words, the collapse is not deterministic. This means that such a genuine
superposition state is a potentiality state with respect to the context that corresponds to the
measurement situation. It is with this type of change that the quantum entity reveals its
potentiality state nature. This state of affairs makes it tempting to conclude that what we refer to
as a context is the same thing as what in the standard quantum case is referred to as the
measurement, i.e. it is a measurement context33.
31
For sake of mathematical completeness, we mention that this is only the case when the operator has a
point spectrum. But it is sufficient for our purpose to consider this case. Measurements that are described
by operators that have not a point spectrum have to be treated in a more sophisticated way, but this is of no
relevance for the points we want to make here.
32
A genuine superposition state to a self-adjoint operator is a state that is a renormalized sum of the
eigenstates corresponding to this self-adjoint operator but not one of the eigenstates itself.
33
Previous work on this can be found in [Aerts, D’Hooghe & Gabora 2000; Aerts & Gabora 2001; Gabora
& Aerts 2000].
159
Liane Gabora
11.2.2.3
The Role of Lack of Knowledge
At this point let us return briefly to the subject of lack of knowledge. Consider the situation where
we have an entity in a potentiality state, which theoretically under different contexts would
collapse to different future states, but we know with complete certainty the context it is actually
exposed to in complete detail, and also the way this context interacts with the entity. For practical
purposes, the entity is not in a potentiality state, and its evolution is determined. In reality the
universe is so complex that this is never completely the case; there is always some possibility of
even very unlikely outcomes. However, there are situations in which we do know the relevant
variables to the extent that it is not necessary to consider the entity as existing in a potentiality
state, and other situations in which it is necessary. Thus a formalism for describing these entities
must take into account the degree of knowledge we as observers have about the context.
11.2.3 The Two Evolutions of Quantum Mechanics United Under the Context View
While we were working on these ideas out, Diederik Aerts had an insight which is presented here
in detail because it has far-reaching consequences, not just for the present application of the
quantum formalism, but also for the quantum structures themselves.34
For historical reasons, physicists do not think of the measurement influencing the state of the
entity under study as a context. They think of the measurement as an influence, but also most of
all as a process that gives rise to outcomes, to be read off on the measurement apparatus. In this
scheme of thought, the simplest measurements are assumed to be those with two possible
outcomes. If however, one is interested in how context influences and changes the state of the
entity under consideration as the basic process, then the simplest of these changes is the
‘deterministic’ change. Such a change does not correspond to a measurement with two outcomes,
but to a measurement with one outcome: namely, always the same. Indeed, the influence of a
measurement with two outcomes, so thought to be the most simple situation by the physicists
working in the foundations of quantum mechanics, has to be already nondeterministic, because
both outcomes have to be possible. And a measurement with one outcome would rightly not
being thought of as a measurement, because if the same outcome always occurs, nothing has been
compared and/or measured. However, by shifting our attention to how states are changed under
the influence of the measurement context, it has become clear that the simplest of these changes is
exactly the one under the influence of this measurement with one determined outcome. A state
that is under the influence of such a context with a determined change has potentiality zero; it can
160
Chapter 11: Potentiality, Context, and Change of State
only change to one state. A measurement with two outcomes is already only the second most
simple change possible, where the potentiality of the state is not zero because it involves a choice
between two possible changes. This means that the effect of context on change in quantum
mechanics is as follows:
•
When the context is the rest of the universe, the influence of this context on the state of the
quantum entity is deterministic, and the change of state is described by the Schrödinger
equation.
•
When the context is the measurement context, the influence of this context on the genuine
superposition state is nondeterministic, and the change of state is described according to the
change of state by a measurement as outlined above.
•
When the context is the measurement context, the influence of this context on an eigenstate is
deterministic. The eigenstate does not change at all.
11.2.4 Entanglement and the Generation of New States with New Properties
As remarked already, explaining how novelty arises is one of the unsolved problems for a theory
of cultural evolution. The quantum formalism contains within it a peculiar feature, which could
provide a way to describe the birth of novelty and the process of creativity. . Let us explain this
peculiar feature of the quantum formalism in some detail. If for two classical entities described by
classical mechanics, the composite entity of these entities is considered, then the state space of
this composite entity is constructed by making the Cartesian product of the two state spaces of the
subentities. Each state of this Cartesian product state space is of the form (p1, p2) where p1 is a
state of the first subentity, and p2 a state of the second subentity. Although these product states
(p1, p 2) are new states, they do not describe anything new; they just describe the two subentities
together, as one new entity. The procedure that is used in quantum mechanics to describe joint
entities leads however in an unexpected way to new states, that do not just describe the two
entities together. Indeed, if in quantum mechanics two subentities combine to form a composite
entity, the state space of the composite entity is formed by considering the Hilbert space which is
the tensor product of the two Hilbert spaces that represent the state spaces of the subentities. This
tensor product Hilbert space contains the product states of the form (p1, p2) as in the classical
34
Note that this interesting feedback effect back on the original source points to the value of
transdisciplinary research.
161
Liane Gabora
case, but it also contains all possible superpositions of such product states. In the quantum jargon,
these superpositions of product states are referred to as entangled states. These entangled states
are new states that describe genuine new properties of the joint entity that cannot be reduced to
properties of the subentities. The word entanglement has been introduced to express this rise of
novelty that appears within the formalism of quantum mechanics. Entangled states are genuine
superposition states with respect to measurement contexts that subdivide the joint entity back into
subentities.
11.3 CLASSICAL MECHANICS CANNOT DESCRIBE INDETERMINISM DUE TO CONTEXT
The insight that the deterministic change described by the Schrödinger equation in quantum
mechanics can be interpreted as change under the influence of the context of the fields that are
present in the rest of the universe (or in that part of the universe that we want to take into account
as relevant), also holds for the case of classical physics and how it describes the behavior of
classical physical entities. (This is why quantum mechanics was discussed before classical
mechanics.) In classical physics, this is in fact the only type of influence possible to describe;
measurements giving rise to nondeterministic changes of state cannot be described within the
formalism. Moreover, within one classical model of description, only one deterministic context
can be introduced. It may be composed of many seemingly different contexts, but because the
entirety is determined with respect to when and how it exerts an effect, it must be regarded as one
big deterministic context. Thus we classify classical mechanics within our general framework for
describing change under the influence of context as the situation where only deterministic change
is possible. This is why the classical physics framework is definitely not general enough to satisfy
the need to incorporate potentiality and contextuality as discussed in the previous chapter. Thus
we must therefore take recourse to (at least) a quantum mechanical framework. It also shows that
the concept of potentiality state has a very limited meaning in the classical framework. If the only
kind of influence a context can have on a state is deterministic, then only deterministic states (or
states of zero potentiality) are present.35
35
Note that classical physics was incorrectly classified in two previous papers on potentiality state and
context [Aerts, D’Hooghe & Gabora 2000; Gabora & Aerts 2000]. There we classified classical physics as
a theory where context cannot be taken into account at all, because then we were still thinking about
context strictly as measurement context, and not as dynamical context.
162
Chapter 11: Potentiality, Context, and Change of State
11.3.1 Measurements in Classical Physics
One could ask the question: “If in classical physics only deterministic contexts can be described,
how could a measurement with more than one possible outcome possibly happen? Isn’t it the case
that if two or more outcomes are possible, something nondeterministic has to happen?”
The situation of a measurement in classical physics is considered as follows. The state of the
entity has been prepared, and the measurement procedure is put into action. At this moment, one
does not yet know the outcome; hence the state of the entity could still change in different ways,
leading to different outcomes. But the classical physicist has the view that the outcome is
determined in principle; it is just knowledge about this state that is lacking. For a given state the
outcome of each experiment is determined and experiments in classical physics are only made to
gather knowledge about the state. The measurement context may eventually influence the state,
such that the outcome is biased by the measurement, but this influence will also be deterministic.
Hence even then the outcome was predictable, but just unknown. We will come back to this
situation in a more general way when we consider stochastic processes in section 11.7.1.
11.3.2 The Unstable Equilibrium as a First Order Quantum Situation
There is one type of situation where the classical physicist comes into problems with this limited
view, not in principle, but in practice. This is the situation of an unstable equilibrium. However,
since it is not very common, it has not been identified as problematic. As an example of an
unstable equilibrium, consider a pencil standing straight on its point on a desk. It is evident that
the equilibrium state of the pencil is unstable, because the tiniest perturbation will make it fall to
the desk. If it is moved slightly to the north, it will fall in this direction and end up lying on the
desk to the north. If it is moved slightly to the west, it will fall to the west. And so forth.
However, despite the fact that the pencil will obviously not stay in this state for long, as far as its
classical description is concerned, it is in equilibrium, meaning that if it really stands straight on
its point, classical mechanics predicts that it will remain like this forever. This means that ‘in
principle’ also the unstable equilibrium state of a classical entity can be treated consistently with
how we classified the classical theories within our general framework of change; only
deterministic contextual influence is possible. Of course in practice it is obvious that the unstable
equilibrium state will not remain, because there are always perturbations that will make it start to
change. In this sense often unstable equilibrium states are treated within classical physics by
introducing models of perturbations. Eventually even models that do not deterministically disturb
the unstable equilibrium state could be introduced, and doing this would bring a real quantum-like
163
Liane Gabora
element into the theory. But still it would remain an ad hoc introduction, case by case, and not
following from a general framework as is the case in quantum mechanics.
The situation is much the same as it is in biology where, in the absence of a complete model
of biological change—which would incorporate all the factors that lead any one organism to meet
and mate with any other—phenomena such as assortative mating and inbreeding avoidance are
tacked on in an ad hoc manner to the neo-Darwinian model.
11.3.3 Chaos and Complexity Theories Cannot Describe Indeterminism Due to Context
The above limitation of classical theories is present even in modern classical theories such as
chaos and complexity theory; context still only influences the state of the entity deterministically.
What is specific to chaos and complexity theories is that from similar initial conditions, identical
entities can end up in very different final conditions. However, the chaotic behavior that results
through lack of complete knowledge of initial conditions is not related to the indeterminism that
results when contexts can exert nondeterministic influences on the state of an entity at any point
along its evolution.
11.4 WHY WE NEED THE GENERALIZED QUANTUM FORMALISMS
The phenomena we need to describe are the potentiality and contextuality of the change of state
of an entity and the appearance of states with new properties. We saw that potentiality is only
present when context influences the change of the state nondeterministically. For help, we looked
to two formal systems for describing entities and how they change: classical mechanics and
standard quantum mechanics. Classical mechanics copes with potentiality and contextuality in a
very limited way, because it can only describe situations where the context influences the state
deterministically. Standard quantum mechanics provides a much more general means of
incorporating potentiality and contextuality. However, the standard quantum mechanical
framework is not general enough for the present purposes. This is due to several limitations of
standard quantum mechanics, all of which find their origin in the particular mathematical
structure in which the theory of standard quantum mechanics is formulated. Let us first point out
the most important of these limitations. Then we explain how the generalized theories of quantum
mechanics overcome these limitations while maintaining all of the richness of how potentiality
and context are presented in standard quantum mechanics.
First a comment on the application of these formalisms beyond their original domain of
quantum mechanics. This is not as strange as it may seem. It can even be viewed as an
unavoidable sort of evolution, analogous to what has been observed for chaos and complexity
164
Chapter 11: Potentiality, Context, and Change of State
theory. Although chaos and complexity theory were first applied to inorganic physical systems,
they quickly found applications to biological and cognitive systems. When one thinks of
complexity and chaos theory, one thinks of a general formalism that can be applied to many
different domains. The same is true of the mathematics that underlies the generalized quantum
formalisms. Although they were originally developed to describe the behavior of microscopic
physical entities in the micro-world, there is no theoretical reason why their application should be
limited to the microphysical realm. In fact, given the presence of potentiality and contextuality in
biological and cultural evolution, it seems natural to look to these formalisms for guidance in the
development of a formal description of evolution. Using the concept of a potentiality state and its
relationship to a context, these formalisms enable us to incorporate potentiality and contextuality
directly into the description of the evolutionary system. (Theoretically—although perhaps not
historically—it would have been possible that the need for this type of formal structure could
have first arisen in the study of the mind, and been given the name ‘mind mechanics’. In this case,
it would perhaps have taken time to abstract it from its original context as pertaining to the mind,
and show that it is also relevant to the micro-world.)
11.4.1 Overcoming the Limitation of the Linear State Space
As mentioned previously, the states of a standard quantum mechanical entity are represented by
unit vectors of a Hilbert space, which is a linear vector space. This means that the states are
strictly defined within this linear mathematical framework. This is too limited for the formulation
of a general evolutionary framework incorporating also biological and cultural evolution36 [Aerts
& Gabora 2001]. The generalized quantum formalisms structure the state space in such a way that
the limitation due to the linearity is overcome.
11.4.2 Describing Situations Between Quantum and Classical
There is growing evidence of macro-level structure that shares some of the key properties of
quantum structure [Aerts 1982b, 1985a, b, 1991; Aerts & Durt 1994a, b; Aerts et al. 2000a; Durt
36
This note is only for the technically interested reader. It is not completely correct to state that it is the
linearity of the state space that is at stake. The situation is more complex since states correspond in fact to
the rays of the vector space (or unit vectors), and hence are incorporated into the projective structure of the
vector space rather than in the linear structure. This means that correctly speaking the limitations brought
about from the mathematical structure of the standard quantum mechanical state space can be best
expressed in the language of projective geometry. For example, if states are considered to be points of the
corresponding projective geometry, and planes are considered to be lines, then for any two points, there is
always a third point that is on the line that connect these two points, which is an axiom of projective
geometry. It is exactly in a category on the projective geometry level, more specifically speaking, the
165
Liane Gabora
1996, D’Hooghe 2000]. The inverse, that classical structure exists at the micro-level, is
undisputed; for example, charge is a classical property37. Thus, although it is commonly assumed
that the divide between classical and quantum corresponds exactly to the divide between macro
and micro, this assumption is unfounded. It has also been made clear that the quantum structure
that has been identified in the macro-world incorporates situations that have to be classified
structurally as ‘between quantum and classical’. The aspects of quantum structure that we identify
for our general evolutionary scheme also incorporates these ‘between quantum and classical’
aspects. It has been proven that the ‘between quantum and classical’ structure cannot be described
by standard quantum mechanics, whereas the generalized quantum formalisms describe ‘between
quantum and classical’ structure without problem. In fact, they contain standard quantum
mechanics and classical mechanics as special cases. For biological, cognitive, and cultural
systems, the effect of contextual influence is usually not negligible, nor maximal, but in between.
Therefore it is the generalized quantum formalisms, which can describe both quantum and
classical situations, as well as situations between quantum and classical, that are useful to us.
11.4.3 Classical and Pure Quantum as Special Cases in the Generalized Formalisms
The development of the generalized quantum formalism has been a work of many and a variety of
such generalized quantum formalisms have been developed38. [Aerts 1982a, 1983, 1992; Aerts &
Durt 1994a, 1994b, Durt 1996; D’Hooghe 2000; Emch 1984; Foulis & Randall 1981; Foulis et al.
1983; Gudder 1988; Jauch 1968; Mackey 1963; Piron 1976, 1989, 1990; Pitowsky 1989; Randall
& Foulis 1976, 1978; Segal 1947]. The finding that classical mechanics and pure quantum
mechanics appear as special cases arises in all of these generalized quantum formalisms.
Let us explain how it is that classical and quantum are two extremes of a more general
situation, and why this is relevant to the present purpose. Recall that for a quantum entity, one can
give only a probabilistic account of its final state. The origin of this quantum uncertainty is
generally considered to be unknown. In the FUND group in Brussels, however, a possible
explanation for this uncertainty is put forward. It is introduced as being due to a lack of
knowledge about the fluctuations present in the interaction between the measurement apparatus
category of the complete orthocomplemented atomistic lattices and their morphisms, that the generalized
quantum formalisms are formulated.
37
In physics this is expressed by saying that superposition states of different charges do not exist, or that
charge is a superselection rule.
38
Note that the formalism used in [Aerts & Gabora 2001] is more general than these generalized quantum
formalisms, because we consider the influence of the change of state on the change of context. This is not
the case in the generalized quantum formalisms, because there the measurement apparatus—representing
the context—is not influenced by the change of the state.
166
Chapter 11: Potentiality, Context, and Change of State
and the entity, and it is shown that this can completely explain the quantum uncertainty [Aerts
1986, 1993, 1998; Aerts & Durt 1994a, 1994b, Durt 1996; D’Hooghe 2000, Aerts et al. 2000a].
Translating this idea to our view amounts to stating that the uncertainty involved when the state
of an entity changes under the influence of a context is due to our lack of knowledge concerning
exactly how this context interacts with this state of the entity. It is indeed very plausible that the
exact knowledge about how a state of an entity interacts with a context is neither incorporated
into the description of the state of the entity (because this description only has to incorporate the
elements that are proper to the entity, and it is still open to interactions with all possible contexts),
nor into the description of the context (because this description has to incorporate the elements
that are proper to the context, and is still open to interactions of this context with different
possible entities). This means that even if we had a complete description of a specific state of an
entity and a complete description of a specific context, the interaction between this entity in this
state and this context, would in general contain elements of reality that are not incorporated into
the two complete descriptions of state and context. It is the lack of knowledge about these new
elements that gives rise to the indeterminism that is present, and that creates a situation where the
state can still collapse to different potential states under the influence of this context. A way of
mathematically describing this situation is by introducing fluctuations in this interaction. This is
what has been done in the FUND research group, and it has been shown that models isomorphic
with any quantum entity can be constructed. In section 11.5.2 one of these models is exposed in
detail. The following picture emerges. If the fluctuations present in the interaction between entity
and context are negligible, and can be ignored, the structure becomes classical, and we are in the
situation of classical physics. If the fluctuations are maximal, the structure become the one of
standard quantum mechanics. In general, however, the fluctuations are neither negligible nor
maximal, and the structure is ‘in between quantum and classical’, a situation that can only be
described by the generalized quantum formalisms [Aerts 1986, 1993, 1998; Aerts & Durt 1994a,
1994b, Durt 1996; D’Hooghe 2000, Aerts et al. 2000a].
11.5 EVOLUTION OF POTENTIALITY STATE UNDER THE INFLUENCE OF CONTEXT
This section introduces macroscopic examples of quantum structure that will hopefully provide a
clearer picture of how a potentiality state changes under the influence of a context.
11.5.1 The Water Example
Consider the following situation: a vase stands in the middle of a table filled with 20 liters of
water. The entity that we consider is the 20 liters of water. We now introduce the following
167
Liane Gabora
measurement context. The measurement apparatus consists of two cups that can be connected to
the vase by tubes of some length and width. The lengths and the widths of the tubes are not
specified because they represent the interaction between the measurement apparatus and the
entity, and we lack knowledge of this interaction. The influence of the measurement context on
the water in the vase is such that it flows from the vase through the tubes to the two cups, and gets
distributed across them. The final state of the water, after the measurement context has influenced
it, is given by the particular distribution of water across the two cups. The state the water is in
when it is contained in the vase we call p, and the different states that the water is in when it is
distributed amongst the cups we call p(a, b), where a is the volume of water in the first cup and b
the volume of water in the second cup. This means that a + b = 20 liters. We see that there are
many potential states that the state p can be changed to by this one measurement context, namely
all the states contained in the set {p(a, b): 0 ≤ a ≤ 20, 0 ≤ b ≤ 20, a + b = 20}.
In this example, we can easily identify the deterministic states and the potentiality states.
The state of 20 liters of water inside the vase, denoted p, is a potentiality state with respect to the
measurement context of the cups and tubes. As mentioned previously, the tubes of various lengths
and widths that go with the same measurement context stand for the different ways that
measurement context can interact with the water entity in state p. It is this lack of knowledge that
gives rise to the nondeterminism in the interaction between the context and the state of the entity.
Those unfamiliar with the rather recent findings of this type of quantum structure in the
macroscopic world might object that water moving between vessels must have to be a pure
classical physics situation. The reason quantum structure is manifested here is specifically due to
the way that the measurement context is defined, leaving open the possibility of different
interactions between context and entity, of which we lack knowledge in the description. In [Aerts
1982b , 1985a, b] this situation is examined in great detail from a structural point of view.
11.5.2 The Epsilon Model Example
The next example, even more than the previous one, illustrates how the aspects of quantum
structure that we need for our description arise. The entity we consider is a point P located on the
surface of a sphere (Figure 11.2). The different points of the surface of the sphere where the point
P could be located represent the different possible states of the entity. Thus the state space of the
entity P is given by the set of all points on the surface of the sphere. These points could be,
literally, points on a sphere in physical space, or they could represent possible states of an abstract
model of some other sort of structure. We want to try to locate P on the sphere, but we cannot just
look at the sphere and see where P is. To locate it, we perform a specific kind of experiment, and
168
Chapter 11: Potentiality, Context, and Change of State
it is the restriction on the type of experiments that introduces quantum structure for our entity.
(Note that in the water example it was also the type of experiments considered that made it
possible to detect the presence of quantum structure.) The experiments require an elastic or rubber
band the length of the diameter of the sphere. They proceed as follows:
Figure 11.2 The state space of the Particle P, the entity that we consider, is the set of points of the surface of the sphere.
We consider the particle P to be in one of these points (see a). The measurements that we consider are the following. An
elastic is put between two diametrically opposite points u and - u of the surface of the sphere (see b). The particle P falls
orthogonally to the elastic and sticks to it (see c). Then the elastic breaks uniformly in one of its points, and the particle P
is pulled to the point u or -u, depending on where the elastic breaks.
§
Place the elastic or rubber band between two diametrically opposite points of the surface of
the sphere, u and -u (Figure 11.2b). This is the first step of the experiment.
§
The experiment continues in the following way. Once the elastic is placed, the point P falls
from its initial place orthogonally onto the elastic, and sticks to it (Figure 11.2c). This is the
second step of the experiment.
169
Liane Gabora
§
The third step of the experiment consists of the elastic breaking at some arbitrary point. The
chance of breakage is evenly spread over the entire length of the elastic, and is independent of
where P stuck to it.
§
In the fourth step, the point P, which is attached to one of the two pieces of the elastic, is
pulled to one of the two endpoints, u or -u (Figure 11.2d). Depending on where point P ends
up, we say the outcome of the experiment is u or -u. This completes the experiment.
We denote as e[u, -u] each experiment where the diametrically opposite points that are chosen are
u and -u. Let us see now what is the net effect of the experiment e[u, -u]. Suppose that entity P
was originally on a point v of the surface of the sphere (as indicated), then the experiment e[u, -u]
has changed the location of P such that it is now on point u or on point -u. It is important to note
that the change of state to u or - u is not deterministic. Whether the point P, our entity, will after
the experiment finally end up at u or end up at -u depends on where the elastic breaks. And where
the elastic breaks is not known, because it depends on the specific interaction that takes place
between the measuring context (the elastic placed in u and - u ) and the point P. Also, the point
where the elastic breaks may depend on all sorts of other effects present in the environment when
the measurement takes place. Thus the indeterminism that is modeled by this example is of
exactly the same nature as the indeterminism that is present in the sorts of cognitive and cultural
situations that we want to describe.
Let us now describe the evolution of entity P over a longer lapse of time, such that different
contexts play a role. Suppose that the sphere consisting of points that constitute possible states for
P is an actual, physical sphere, and thus the states are literally possible locations on this sphere.
Suppose that this sphere is located in a gravitational field, and that this gravitational field acts as a
deterministic context for entity P. (Thus we suppose the entity P to have a certain weight that is
influenced by the gravitational field). Since the sphere is in the gravitational field, we call the
highest point of the sphere its North Pole and the lowest point its South Pole. Suppose that at time
t0 our entity, the point P, is in state v(t0) as shown in Figure 11.3. In the time interval [t0, t1], only
the deterministic context of the gravitational field is present. Under the influence of the context of
this gravitational field, point P changes from its original state v(t0) to a new state v(t1). The big
circle that entity P moves on during the time interval [t0, t 1] cuts through the North Pole and the
South Pole, and through the original point v(t0). Thus, the entity moves to the south on this circle
under the influence of the gravitational field until it reaches the point v(t1).
170
Chapter 11: Potentiality, Context, and Change of State
Figure 11.3 During the time interval [t0, t 1], particle P moves on a big circle that cuts through the North Pole, the South
Pole, and the original point v(t0). It moves southward under the influence of a gravitational field until it reaches the point
v(t1). Then, during the time interval [t1, t2], a nondeterministic context comes into play, namely the elastic that is put
between the two diametrical points u and - u. P moves under the influence of this context to one of the points u or - u, let
us say to - u. This means that v(t2) = -u. In the time interval [t2, t 3] the gravitational field again works as a deterministic
context that makes the particle P move to v(t3). Then, during the time interval [t3, t4] a new nondeterministic context arises;
an elastic put between the points w and -w,. The particle P then movse to one of the points w or -w, lets say w. This
means that v(t4) = w. Again the deterministic context of the gravitational field takes over then. And so on.
Meanwhile, let us say during a time interval that lies within the time interval [t0, t1] a new context
has come in. It is the measurement context e(t1) = e[u, -u]. This means that an elastic has been put
between the diametrically opposite points u and -u, and in the time interval [t1, t 2] the four steps
of the measurement indicated in Figure 11.2 take place. It also means that at time t2, the state of
entity P, will have changed to one of the points u or - u. Suppose that the actual change that takes
place is the one of the state v(t1) moving to -u, hence we have then v(t2) = -u. We have drawn
only the actual path that the particle P follows, but with each nondeterministic context, there are
other potential paths that the particle P could also have followed. Once again at time t2 the
deterministic context of the gravitational field takes over. This means that in the time interval [t2,
171
Liane Gabora
t3] the entity P again moves on a big circle to the south and ands up in point v(t3). At time t3 a new
measurement context e(t3) is ready to take over the gravitational field context, and it is given by
the measurement context e[w, -w]. During the time interval [t3, t4], the four steps of the
measurement indicated in Figure 11.2 take place. Hence at time t4, the state of the entity P will
have changed to one of the points w or -w. Suppose that the actual change that takes place is the
one of the state v(t3) moving to w, hence we have v(t4) = w. Then again the gravitational field
takes over, and the entity P moves on a big circle to the south. In subsequent time intervals, new
measurement contexts come in, and the evolution process continues as in the four time intervals
[t0, t1], [t1, t2], [t2, t3] and [t3, t4] described here, with the elastic put in different directions.
Again we can easily identify the deterministic states and the potentiality states. For the
epsilon model, if we consider a measurement context corresponding to the measurement e[u, -u],
then every state v of the entity P that is different from u and different from -u is a potentiality
state with respect to this context. The states v = u and v = -u are deterministic states with respect
to this context. It is easy to see that for each state there exists two contexts such that this state is a
deterministic state with respect to these contexts. Indeed, if we consider a state v, then v is a
deterministic state with respect to the measurement contexts e[v, -v] and e[-v, v]. The state v is a
potentiality state with respect to all contexts different from these two.
This model, the epsilon model, was introduced in [Aerts 1986], and has been studied in great
detail in the FUND research group [Aerts 1986, 1987 1988, 1991; Aerts, Durt & Van Bogaert,
1993; Aerts. & Durt 1994a, b; Aerts, Coecke, Durt, & Valckenborgh, 1997a,b]. It can be shown
that the transition probabilities that appear when the elastic breaks uniformly are exactly the
quantum mechanical transition probabilities of the spin of a spin _ particle. This means that the
evolution process considered here can be described using the standard quantum mechanical
description of the spin of a spin _ particle measured upon by a Stern Gerlach apparatus.
11.6 IDENTIFYING QUANTUM STRUCTURE THROUGH VIOLATION OF BELL INEQUALITIES
The violation of Bell inequalities is an experimental indication of the presence of quantum
structure. (This is why in chapter eleven we investigate situations in cognition where Bell
inequalities are violated). To make clear how the violation of Bell inequalities indicates the
presence of genuine potentiality states, we introduce the inequalities here, and show how they are
violated in the above examples.
Let us begin by presenting a brief account of the most relevant historical results related to
Bell inequalities, and show why the violation of these inequalities is an experimental indication
for the presence of quantum structure. In the 1970s, a series of experiments was carried out to test
172
Chapter 11: Potentiality, Context, and Change of State
for the presence of nonlocality in the microworld described by quantum mechanics [Clauser,
1976; Faraci et al., 1974; Freedman & Clauser, 1972; Holt & Pipkin, 1973; Kasday et al. 1975]
culminating in decisive experiments by Aspect and his team in Paris [Aspect et al., 1981, 1982].
They were inspired by three important theoretical results: the EPR Paradox [Einstein et al., 1935],
Bohm’s thought experiment [Bohm, 1951], and Bell’s theorem [Bell, 1964]. Einstein, Podolsky
and Rosen believed to have shown that quantum mechanics is incomplete, in that there exist
elements of reality that cannot be described by it [Einstein et al. 1935; Aerts 1984, 2000]. Bohm
took their insight further with a simple example: the ‘coupled spin-_ entity’ consisting of two
particles with spin _, of which the spins are coupled such that the quantum spin vector is a
nonproduct vector representing a singlet spin state. It was Bohm’s example that inspired Bell to
formulate a condition that would test experimentally for nonlocality. The result of his efforts are
the infamous Bell inequalities. Bell’s theorem states that the statistical results of experiments
performed on a physical entity satisfy the inequalities if and only if the reality in which this
physical entity is embedded is local. Experiments performed to test for the presence of
nonlocality confirmed the predicted results, such that it is now commonly accepted that the
micro-physical world is incompatible with local realism.
Bell inequalities are defined with the following experimental situation in mind. We
consider a physical entity S, and four experiments e1, e2, e3, and e4 that can be performed on the
physical entity S. Each of the experiments ei, i equals 1, 2, 3, or 4, has two possible outcomes,
respectively denoted oi(up) and oi(down). Some of the experiments can be performed together,
which in principle leads to coincidence experiments eij, where i and j equal 1, 2, 3, or 4. For
example, ei and ej together will be denoted eij. Such a coincidence experiment eij has four possible
outcomes, namely (oi(up),oj(up)), (oi(up),oj(down)), (oi(down),oj(up)) and (oi(down),oj(down)).
Following Bell, we will introduce the expectation values Eij, where i and j equal 1, 2, 3, or
4, for our coincidence experiments, as
Eij = P(oi(up),oj(up)) + P(oi(down),oj(down)) - P(oi(up),oj(down)) - P(oi(down),oj(up))
(11.1)
From the assumption that the outcomes are either +1 or -1, and that the correlation Eij can be
written as an integral over some hidden variable of a product of the two local outcome
assignments, one derives Bell inequalities:
|E13 - E14|+|E23 + E24| ≤ 2
(11.2)
173
Liane Gabora
To reach the point where we can use the violation of Bell inequalities as an indication of the
presence of quantum structure, it is necessary to mention one more thing. Pitowsky [1989] proved
that if Bell inequalities are satisfied for a set of probabilities concerning the outcomes of the
considered experiments, there exists a classical Kolmogorovian probability model. The
probability can then be explained as being due to a lack of knowledge about the precise state of
the system. If, however, Bell inequalities are violated, Pitowsky proved that no such classical
Kolmogorovian probability model exists. Hence, the violation of Bell inequalities shows that the
probabilities involved are nonclassical. The only type of nonclassical probabilities that are well
known in nature are the quantum probabilities. The probability structure that is present in our
examples is nonclassical and nonquantum.
11.6.1 The Original Violation of Bell Inequalities in the Microworld
When Bell introduced the inequalities, he had in mind the quantum mechanical situation
originally introduced by Bohm [Bohm, 1951] of correlated spin-_ particles in the singlet spin
state. Here e1 and e2 refer to measurements of spin at the left location in spin directions a1 and a2,
and e3 and e4 refer to measurements of spin at the right location in spin directions a3 and a4. Let us
introduce it in some detail and see how, for well chosen directions of spin, the quantum
theoretical calculation in this situation gives the value 2 √2 for the left side of the equation, and
hence violates the inequalities.
Our physical entity S is now a pair of quantum particles of spin-_ that ‘fly to the left and the
right’ respectively, along a certain direction v of space, and are prepared in a singlet state Ψ (S)
for the spin (see Figure 11.4). We consider four experiments e1, e2, e3, and e4, that are
measurements of the spin of the particles in directions a1, a2, a3, and a4, that are four directions of
space orthogonal to the direction v of flight of the particles. We choose the experiments such that
e1 and e2 are measurements of the spin of the particle flying to the left and e3 and e4 of the particle
flying to the right. For the Bohm example, the experiment e1 can be performed together with the
experiments e3 and e4, which leads to experiments e13 and e14, and the experiment e2 can also be
performed together with the experiments e3 and e4, which leads to experiments e23 and e24.
174
Chapter 11: Potentiality, Context, and Change of State
a3
45ϒ
a4
a2
a1
Figure 11.4 : The violation of Bell inequalities by the singlet spin state.
Quantum mechanically, this corresponds to the expectation value <
1(a),
2(b)
> = -a b which
gives the well known predictions:
E13 = -cos (a1, a3)
E14 = -cos (a1, a4)
E23 = -cos (a2, a3)
E24 = -cos (a2, a4)
(11.3)
Let us specify the situation that gives rise to a maximal violation of Bell inequalities. Let a1, a2,
a3, and a4 be coplanar directions such that the angle between a1 and a3, the angle between a3 and
a2, and the angle between a2 and a4 equals 45 degrees, and the angle between a1 and a4 equals 135
degrees (see Figure 11.4). Then we have:
E13 = E23 = E24 = -√2/2
(11.4)
and
E14 = √2/2
(11.5)
This gives:
|E13 - E14| + |E23 + E24| = |-√2/2 - √2/2| + |-√2/2 - √2/2| = 2√2
(11.6)
which is bigger than 2, which shows that Bell inequalities are violated.
175
Liane Gabora
11.6.2 The Violation of Bell Inequalities by the Potentiality State of Water
To show how Bell Inequalities are violated by the water example, we will adapt the water
example of section 11.5.1 slightly. Instead of considering the 20 liters of water to be present in
one vase, we now consider two vases, one at the left and one at the right, that are connected by a
tube. Thus, the 20 liters of water is inside the two vases and also inside the tube that connects
them, as in Figure 11.5. Since the vases are connected by a tube, they behave like one vase for the
experiments. The experiments consist of taking out the water at two sides and collecting it in two
cups, one to the left, below the first vase, and one to the right, below the second vase. This means
that if, for example a liters of water is taken out of the left vase, then b liters of water will be
taken out at the right side, such that a + b = 20 liters. Thus, this is essentially the same situation as
the one considered earlier, but we will see that the adaptation will make it easier to see why Bell
inequalities can be violated.
Figure 11.5 The vessels of water example violating Bell inequalities. The entity S consists of two vessels containing 20
liters of water that are connected by a tube. Experiments are performed on both sides of the entity S by introducing
siphons K1 and K2 in the respective vessels and pouring out the water and collecting it in reference vessels R1 and R2.
Carefully chosen experiments reveal that Bell inequalities are violated by this entity S.
To be able to check for the violation of Bell inequalities, we must introduce four experiments, of
which some can be performed together. Let us introduce the experiment e1 that consists of putting
a siphon K1 in the vase of water at the left, taking out water using the siphon, and collecting this
water in a reference vessel R1 placed to the left of the vessel. If we collect more than 10 liters of
water, we call the outcome o1(up), and if we collect less or equal to 10 liters, we call the outcome
o1(down). We introduce another experiment e2 that consists of taking with a little spoon, from the
left, a bit of the water, and determining whether it is transparent. We call the outcome o2(up)
when the water is transparent and the outcome o2(down) when it is not. We introduce the
experiment e3 that consists of putting a siphon K3 in the vessel of water at the right, taking out
water using the siphon, and collecting this water in a reference vessel R3 to the right of the vessel.
176
Chapter 11: Potentiality, Context, and Change of State
If we collect 10 or more liters of water, we call the outcome o3(up), and if we collect less than 10
liters, we call the outcome o3(down). We also introduce the experiment e4 which is analogous to
experiment e2, except that we perform it to the right of the vessel.
The experiment e1 can be performed together with experiments e3 and e4, and we denote the
coincidence experiments e13 and e14. Also experiment e2 can be performed together with
experiments e3 and e4, and we denote the coincidence experiments e23 and e24. For the 20 liters of
water distributed over the two vases, the coincidence experiment e13 always gives one of the
outcomes (o1(up), o 3(down)) or (o1(down), o 3(up)), since more than 10 liters of water can never
come out of the vessel at both sides. This shows that E13 = -1. The coincidence experiment e14
always gives the outcome (o1(up), o4(up)) which shows that E14 = +1, and the coincidence
experiment e23 always gives the outcome (o2(up), o3(up)) which shows that E23 = +1. Clearly
experiment e24 always gives the outcome (o2(up), o4(up)) which shows that E24 = +1. Let us now
calculate the terms of Bell inequalities,
|E13 - E14| + |E23 + E24| = |-1 - 1| + |1 + 1| = 2 + 2 = +4
(11.7)
This shows that Bell inequalities are violated. The potentiality state related to the measurements
that divide up the 20 liters of water in the two reference vessels, is at the origin of this violation.
We will see that the violation of Bell inequalities makes it possible to us to identify exactly the
type of quantum indeterminism and distinguish it from classical indeterminism.
11.7 TWO TYPES OF INDETERMINISM
One might wonder why the general evolution process could not just be described using the theory
of stochastic processes. After all, it is claimed that the theory of stochastic processes can describe
evolution where one lacks knowledge concerning the transition from one state to another state;
that is, this transition is not deterministic. This is an important issue, so it is addressed here in
some detail.
11.7.1 Pure States versus Mixed States in Stochastic Processes
It is true that stochastic processes describe nondeterministic change, but the source of the type of
indeterminism they are able to describe is different from that we have been considering. To see
this more clearly, let us differentiate between two uses of the concept of state. The first way it is
used is to refer to the set of actual properties of an entity. This is sometimes called the pure state.
This is how we have used the concept of state until now, whether it was deterministic with respect
177
Liane Gabora
to a context, or nondeterministic in the sense of being a potentiality state with respect to a context,
and this is how it will continue to be used. Often, however, the situation is that we lack
knowledge of the pure state of the entity. To describe this situation, one can use another concept
of state referred to as a mixed state. A mixed state does not describe the set of actual properties of
an entity; it describes the knowledge (and lack of knowledge) we have about this set of actual
properties. If a description of change works on mixed states, it is also nondeterministic, not
because we don’t know how it relates to the context, but because we don’t precisely know the
state itself, irregardless of context. (In the epsilon model, this type of indeterminism could be
shown by representing the entity P as a fuzzy spot on the surface of the sphere that encompasses
not just one point but many, and there would be no elastics involved at all.) This is the type of
indeterminism described by stochastic processes, which is why stochastic processes are classified
as statistical physical theories.39
11.7.2 Can Stochastic Processes Describe Context-driven Indeterminism?
One can ask whether the stochastic approach can nevertheless describe the nondeterministic
evolution outlined above. This has been the subject of quite some investigation, and it turns out
that the answer is negative. The reason why is that the mathematical structure of the state space
used in these approaches is too restricted. In a stochastic process, the state space—the set of
mixed states—is taken to be a set of probability measures on a σ-algebra of events in an outcome
space. It was already known from the beginning years of quantum mechanics that the stochastic
approach is not able to describe the nondeterministic evolution of quantum entities, because the
set of events of a quantum entity do not form a σ-algebra in a set of outcomes [Foulis & Randall
1972, 1973]. (That is the rationale behind the field of quantum stochastic processes.) This is why
when the indeterminism present in an evolution process such as the change of state in the epsilon
model is due to a lack of knowledge of the interaction between the context and the state of the
entity, this evolution process cannot be described by a classical (as opposed to quantum)
stochastic process. Indeed, the evolution process in the epsilon model is structurally completely
identical to the evolution process of the spin of a quantum particle of spin _ under the
measurement contexts of spin measurements (for example by a Stern Gerlach apparatus). In this
sense, the sole existence of the epsilon model is theoretically sufficient to prove that there is at
least one non-classical stochastic process possible in macroscopic reality. Meanwhile, it was
39
Note that we did not mean to give the impression that we do not want to consider situations involving
mixed states. It is more just a matter of treating the less complicated situation first. Moreover, in [Aerts &
178
Chapter 11: Potentiality, Context, and Change of State
proven that the probability model generated by the epsilon model cannot have a set of events of
some arbitrary outcome space that forms a σ-algebra, as would have to be the case if the
stochastic processes approach applied here [Aerts 1986]. Moreover, Pitowski showed that there is
an interesting link with Bell inequalities. He proved that Bell inequalities test exactly whether the
set of experiments that are considered to formulate the inequalities emerge from the presence of a
σ-algebra structure on the outcome set for the set of events corresponding to these experiments
[Pitowski 1989]40 This means that we can use the violation of Bell inequalities to indicate the
presence of the type of indeterminism due to lack of knowledge about the interaction between the
context and the state of the entity, and to distinguish it from indeterminism due to considering
mixed states instead of pure states. It also means that Bell inequalities will never be violated in an
evolution described by a stochastic process. It furthermore shows that changes of state of mind
should be classified as a nonclassical stochastic processes, because we have come up with an
example where Bell inequalities are violated in cognition (described in chapter thirteen). For how
classical probabilistic structures are related to lack of knowledge of the state of the entity, and
nonclassical probabilistic (e.g. quantum probabilistic) structures are related to a lack of
knowledge of the interaction between the context and the state of the entity, we refer the reader to
[Aerts 1999]. For a detailed analysis of the nature of this nonclassical probability model, and how
it is related to the quantum probability model, see [Aerts 1985a, b, 1991a, b].
11.7.3 The Water Example in a Classical Statistical Description
In this section we present a classical statistical version of the water example, where we can see
that indeed Bell inequalities are no longer violated. Suppose we have to produce (at first sight) the
same type of indeterminism as that identified in the water example of sections 11.5.1 and 11.6.2,
but now arising from a lack of knowledge concerning the state of the entity instead of a lack of
knowledge of the interaction between the measurement context and the state of the entity. We
proceed as follows. Instead of considering the potentiality state p where the two vases containing
a total of 20 liters of water are connected by a tube, and hence form in fact one vase, we consider
now a huge collection of different pairs of two vases, wherein one contains a liters of water and
the other b liters of water, such that a + b = 20 liters. These couples of vases are collected inside a
big room. The preparation of the mixed state that we want to consider proceeds now as follows.
We elect at random one couple of vases, but we do not know which one. Hence before us in the
Gabora 2001] show how a statistical version of the evolution process must in fact be of a quantum
stochastic nature.
179
Liane Gabora
laboratory where we will execute our experiment we have one such couple of vases, and the only
thing we know is that it is elected from this room, hence the sum of the amounts of water
contained in each vase is 20 liters. If we now perform the experiments considered earlier to
violate Bell Inequalities, we see that the first coincidence experiment e13 always gives one of the
outcomes (o1(up), o3(down)) or (o1(down), o 3(up)), which shows that E13 = -1. More importantly,
this is exactly the same type of indeterminism as when we performed this experiment on the
potentiality state p. Indeed, we find in our two reference vessels to the left and the right two
amounts of water, such that their sum equals 20 liters. Thus, considering just this one experiment,
we would not be able to distuinguish between the two situations: the first, where the
indeterminism arises because of the potentiality state p of the connected vessels of water, and
hence from the lack of knowledge of the interaction of the context and the state of the entity, and
the second, where the indeterminism arises because of the mixed state, the random choice
between all the couples of separated vases in the room, and hence from a lack of knowledge of
the state of the entity.
Let us see, however, whether Bell inequalities are still violated for the mixed state situation.
The coincidence experiment e14 now gives the outcome (o1(up), o4(up)) or (o 1(down), o4(up)) with
equal probability, and this is definitely different from what the coincidence experiment gave in
the case of the measurement being executed on the potentiality state. Indeed there it always gave
the outcome (o1(up), o4(up)). For the coincidence experiment e23 we have a similar situation; it
will give the outcome (o2(up), o3(up)) or (o2(up), o3(down)) with equal probability, while in the
the situation of the potentiality state p, it always gave outcome (o2(up), o3(up)). What is important
is that e14 gives outcome (o1(up), o 4(up)) if and only if e23 gives outcome (o2(up), o 3(down)) and
e14 gives outcome (o1(up), o4(down)) if and only if e23 gives outcome (o2(up), o3(up)). This means
that two possibilities arise with equal probability: (1) E14 = +1 and E 23 =-1, or, (2) E14 = -1 and
E23 = +1. For the coincidence experiment e24 nothing has changed, it always gives the outcome
(o2(up), o 4(up)) which shows that E24 = +1. Putting these values inside Bell inequalities we find
in case (1):
|E13 - E14| + |E23 + E24| = |-1 - 1| + |-1 + 1| = 2
(11.8)
And in case (2):
40
To come to this result Pitowsky introduces some more general type of inequalities, now referred to as
180
Chapter 11: Potentiality, Context, and Change of State
|E13 - E14| + |E23 + E24| = |-1 + 1| + |-1 + 1| = 2
(11.9)
This shows that Bell inequalities will not be violated.
11.8 SUMMARY
A theory of evolution requires a description of how an entity undergoes a change of state under
the influence of a context. However, in general, we do not have complete knowledge of the state
of the entity, the context, and how they interact. Thus when we believe one situation to be
identical to another—i.e. same entity in the same state under the same context—this is often not
really the case. The states and contexts may be different, though we have identified them as being
the same because of our lack of knowledge of the situation. This gives rise in a natural way to
nondeterministic change.
There are different possible reasons for a lack of knowledge concerning a change of state,
each of which gives rise to a different kind of nondeterminism. The theory of classical stochastic
processes (Markov processes) claims to provide a formalism for the description of
nondeterministic change of states. However, it can only describe nondeterministic change that
arises through a lack of knowledge concerning the state of the entity, and not a lack of knowledge
of the interaction between this state and the context. The reason is that lack of knowledge of the
interaction between state and context introduces a non-Kolmogorovian probability model41 on the
state space, while classical stochastic processes work within a framework where the state space
(in this case, the set of mixed states, which are probability measures on the set of pure states) is
described within a Kolmogorovian probability structure. Moreover, because classical mechanics
describes the formation of joint entities only using product states, it cannot describe the
spontaneous appearance of new states with new properties. Chaos and complexity theory provide
a means of describing emergent new states, and entities that started out in similar initial states can
end up in widely different final states, but the effect of context on change of state is always
deterministic.
Quantum mechanics does provide a means of describing states of potentiality, referred to as
superposition states. It can also describe changes of state that are nondeterministic, referred to as
collapse events, where this nondeterminism originates through a lack of knowledge about how a
context—specifically, a measurement—interacts with the state of the entity. Quantum mechanics
can also describe the appearance of new states with new properties. We saw from two examples,
Pitowsky’s inequalities, of which Bell inequalities are a special case.
181
Liane Gabora
one involving water and one involving the epsilon model, that quantum structure can be present
in the macroworld, and that one can test for the presence of quantum structure by seeing if the
entity under consideration violates Bell inequalities. However, if one were to use the pure
quantum formalism for describing how a quantum entity undergoes a change of state in response
to a measurement to develop a general theory of evolution, one would be confronted with several
limitations; including the linearity of the Hilbert space, and the fact that one can only describe the
extreme case where potentiality is always present and change of state is maximally contextual.
The generalized quantum formalisms overcome these limitations by using a state-property
system instead of a Hilbert space. The original motivation for the development of generalizations
of the formalisms of quantum mechanics was in fact theoretical (as opposed to the need to
describe the reality revealed by experiments). However, it turn out to be able to describe
situations of nonlinearity, involving entities that vary in degree of potentiality and contextuality,
as well as the appearance of states with new properties. This is why they are outlined here in the
context of beginning to construct a generalized theory of evolution. The state of potentiality with
respect to a context is referred to as a potentiality state, and the change of state of the entity
whose state is a potentiality state with respect to a context is described as a process of collapse
under the influence of a context such that a state that was previously potential is now actual.
41
This means that Baye’s axiom of conditional probability is not satisfied.
182
Chapter Twelve
12 Toward a General Theory of Evolution
The previous chapter investigated what is the best way to describe change of state when
potentiality and contextuality—that is, indeterminism due to context—may be present. A good
theory of change of state takes us part of the way toward a theory of evolution. But only part of
the way. We also want in our theory of evolution to know how different factors involved in how
an entity changes state under the influence of a context lead to global patterns over time such as
adaptation to an environment. The goal of this chapter is to, drawing on concepts from the
generalized quantum mechanical formalisms, develop a more precise, less domain-specific
formulation of evolution.
12.1 ON THE RE-APPLICATION OF CONCEPTS FROM PHYSICS
The generalization of the process of evolution will involve some re-application of concepts from
physics presented in the previous chapter. Although it may eventually be appropriate to come up
with a new terminology, for now, for simplicity, we will borrow from the original terminology of
physics. However, it is necessary to specify exactly what is meant by certain concepts as they are
reapplied in this new context pertaining to evolution, culture, and cognition, since their use in this
domain is not exactly the same as their use in their original domain.
12.1.1 Collapse
We will refer to the change of a potentiality state under the influence of a context as a ‘collapse’.
We will also refer to a deterministic change of state as a collapse, but specify that it is
deterministic. Thus we use the term collapse in a more general way than it is used in quantum
mechanics, where it is restricted to the nondeterministic case.
183
Liane Gabora
12.1.2 Superposition
Let us return to the general scheme for evolution presented in the previous chapter. We have a
potentiality state with respect to a context, and this potentiality state can change to different
collapsed states. If we take the special case of a quantum entity, then the potentiality state is a
genuine superposition state, and the collapsed states are eigenstates of the measurement context.
In quantum mechanics, there exists an inverse procedure, wherein the superposition state is
retrieved as a possible ‘superposition’ of the eigenstates. For the general situation of a potentiality
state with respect to a context, we also want to be able to express this inverse relation, and as in
quantum mechanics we will use the same word ‘superposition’ to refer to it. This means that for a
potentiality state that has a set of collapsed states with respect to a certain context, we also say
that this potentiality state is a possible superposition of some of these collapsed states with respect
to this context.
12.2 GENERALIZING THE PROCESS OF EVOLUTION
We now outline a general scheme that can be viewed as the backbone of any evolution process.
Afterward we will look more specifically at how this general scheme manifests differently in
different kinds of evolution.
12.2.1 The Basic Idea
Let us now see how treating evolving entities as potentiality states provides the basis for a more
accurate theory of evolution. Evolution is still viewed as a process wherein a stream of
information incrementally adapts to environmental constraints. However, now it requires:
•
A potentiality state which incorporates probabilistically the various states the entity might
actualize given the contexts it might encounter.
•
A means of actualizing, or collapsing on, states that tend to be adaptive given its particular
context.
•
A means of diverging from actualized states to merely potential ones; in other words, of
determining how the actualization feeds back on the potentiality state.
Thus, an evolving entity is in a potentiality state with respect to its given context, and this
context actualizes some aspect of this potentiality by causing it to collapse to another state, one
184
Chapter 12: Toward a General Theory of Evolution
of the possible states it could collapse to. The specifics of the context elicit in the entity one of
the states that were previously potential for it These changes of state feed back on the potentiality
state of entity, thus affecting the way it is then influenced by context, and so forth recursively. So
when a lineage—whether it be biological, cultural, or of some other form—undergoes descent
with modification, it repeatedly collapses onto one or another of the forms that became potential
for it through its history of previous collapses. As the cycle repeats itself, novel adaptations arise.
Thus, evolutionary systems are described as processes of recursively iterated, context-driven
actualization of potential (abbreviated as CAP), and the degree and direction of evolution depends
on the state of the entity, its context, and the nature of their interaction. This formulation gives us
the capacity to address (1) the potentiality of the evolving system, (2) all the contextual factors
(not just natural selection) involved in the actualization of certain aspects of this potentiality, and
(3) the effect of a given actualization on the potentiality. The evolution of the entity cannot be
examined without introducing contexts, and the contexts themselves unavoidably affect its
evolution. Its state at any time reflects both its history of previous states, and the contexts to
which it has been exposed.
12.2.2 Quantum Structure in Evolutionary Systems
In a biological application of this formalism, for reasoning at the organism level the state of the
genome can be treated as a superposition state of the genes, the potential of which becomes
actualized in the context of a certain ecological milieu to generate an organism. For reasoning at
the population level, one considers the potentiality of a species in a given generation as a
superposition state of the genotypes of all reproducing individuals, and one thinks in terms of
distributions of potential genotypes in the next generation given the potentiality of the current
generation. Whereas in population genetics one is forced to ignore potentiality until the point
where it has had an impact on the distribution of genotypes, the current approach enables one to
incorporate it directly into the formalism.
In the cognitive application of this formalism, abstractions are treated as potentiality states.
More specifically, we focus on concepts (since they are the most straightforward type of
abstraction), and view them as superposition states of their instances or exemplars. Note that this
theory has something in common with both prototype and exemplar theories (see chapter five).
Like the exemplar theory, concepts consist of instances (that is, exemplars, or interpretations), but
these instances are ‘woven together’ like a prototype. Moreover, the state space where concepts
‘live’ is not limited a priori to only those properties which appear to be most relevant; thus
185
Liane Gabora
concepts retain in their representation the contexts in which they have, or even could potentially
be, evoked or collapsed onto. This is what enables them to be contextual.
Borrowing from the generalized quantum mechanical formalism, it may be possible to
describe situations where exemplars are superposed to varying degrees in a concept, and concepts
are superposed to varying degrees in more abstract concepts, such that hierarchical structure with
varying degrees of superposition can be described. Thus the entire mind would be viewed as a
superposition state of memories, concepts, and other sorts of abstractions, with respect to the
given situation the individual is in. The network of memories and concepts, or worldview,
undergoes collapse to generate an instant of experience—an impression, observation,
interpretation, judgement, emotional reaction, or assessment of some kind. The specific quality of
this collapse emerges through the interaction between the worldview, and the world itself as
perceived at that instant. The potentiality of the mind at time t0 is the thoughts or experiences it
could have at time t1. This potentiality collapses to generate the thought or experience it does in
fact have at time t1.
We can refer to an environmental perturbation or situation that evokes a conceptual collapse
as a stimulus. The stimulus plays a role analogous to that of the measurement in quantum
mechanics; it evokes an experience that was latent, one the individual was potentially capable of,
given the appropriate context. Although the concept B that follows a collapse will sometimes be
referred to as an exemplar or instance of the concept A that preceded the collapse, it is good to
keep in mind that many sorts of relationships between A and B are possible. Not only could B be
an instance of A, but A could be an instance of B, or A could simply remind one of B. In abstract
thought, the previous stimulus is re-experienced in the context of the changes to the conceptual
network that it just evoked. Stimuli are categorized as instances of a concept according not to how
well they match a static prototype or set of typical exemplars, but to the extent to which they
activate and thereby evoke a collapse to one or another of the potential instances of the concept.
(As a metaphorical explanatory aid, if concepts were apples, and the stimulus a knife, then the
qualities of the knife determine not just which apple to slice, but which direction to slice through
it. The formulation is similar to a holograph in the sense that, much like changing the angle of a
reference beam can elicit a different stored image, changing the context in which a stimulus
situation is embedded can change can cause a different version of the concept to be elicited.)
The contextual interaction between the stimulus and the concept(s) it activates in turn feed
back on the potentiality of the mind; what thoughts or experiences become accessible to this mind
in the future, and thus what idea, actions, or artifacts it might eventually express or manifest in
the cultural dynamic. At the cultural level, it may be useful to view a society as a superposition
186
Chapter 12: Toward a General Theory of Evolution
state of minds with respect to the global context of the state of our shared environment. All this at
the moment is still at the level of speculative though intriguing possibilities. However, in the next
chapter we will see that we do have preliminary evidence that this approach indeed reflects the
underlying architecture of cognition.
12.3 DISTINGUISHING DIFFERENT EVOLUTIONARY SYSTEMS
We now distinguish several ways in which evolutionary systems differ with respect to how they
actualize potential.
12.3.1 Degree to which Potential is Retained
Having taken a careful look at deterministic and nondeterministic change of state under the effect
of a context and how they are dealt with in the formalisms of physics, we are better able to see the
significance of the fact that all biological entities eventually die. A change of state of a biological
entity only affects its lineage if it has an effect on the generation of progeny, either through the
interpretation of self-replication instructions as discussed in chapter three, or through parental
care. Thus, for example, if a rabbit, say, hops into a rabbit hole, the resulting change of state from
above ground to below ground will not affect the evolution of its rabbit lineage (unless it causes
death or influences the generation of offspring). In contrast, a change of state to a cultural entity
can constitute its next generation. For example, if you hear of an idea, and you think about that
idea a new way, the change of state it undergoes in your mind can affect its ‘cultural lineage’.
We can now see why the neo-Darwinian view of evolution has been satisfactory for so long,
and why it wasn’t until after cultural, as well as cosmological [Smolin 1997], change started to be
viewed in evolutionary terms that the time became ripe for potentiality and contextuality to be
taken seriously. Biological evolution is genotypically mediated; that is, it is the mutation and
recombination of genetic material that gives rise to biological novelty, and from which the
phenotype develops. Thus, phenotypic modifications acquired during an organism’s lifetime
cannot be passed on to offspring. Most of the effects of contextual interaction in a biological
system is in a certain sense wasted; it has relatively little effect on the set of genetic instructions
passed on to the next generation. Such interactions affect biological evolution only indirectly
through their effect on fitness, by influencing the number and nature of offspring. In culture, on
the other hand, modifications to an idea since its conception can be passed on to others, as we
saw in chapter three. Contextual interaction affects both the quantity of new ideas, and the form
these new ideas take. Thus in culture there is much more opportunity than in biology for context
to exert a lasting impact on the evolutionary process.
187
Liane Gabora
12.3.2 Degree of Determinism
Another dimension along which evolutionary systems differ, the one we have concentrated on
here, is the degree of determinism involved in the changes of state the entity undergoes. However
what remains to be explained is the relationship of indeterminism to the degree to which context
acts upon (1) explicitly actualized versus (2) potential variations of the entity. It is when context
is acting upon potential variation that indeterminism comes in. More concretely, Darwinian
evolution is an example of (1); the entity splits off into physically realized and separated variants
of itself, and context chooses or selects amongst them. An extreme example of (2) is a quantum
entity, the various possible future states of the entity are superposed, irretrievable as separate
states until the context (in this case a measurement) evokes a collapse upon one or another of
them. The evolution of a stream of thought in a single mind is closer to (2), but since a mind is
embedded in a web of other minds and artifacts, culture may come closer to (1).
12.3.3 Degree of Sensitivity to Context
Another parameter is the degree of sensitivity to context. This can be restated as follows: the
degree to which a change of state of the context causes a change of state of the entity. Two related
dimensions are the degree to which a change of state of the entity causes a change of state of the
context, and the degree to which the process of undergoing a change of state feeds back
recursively on the state of the entity (regardless of context). As an illustrative example, let us
return to the question from chapter three: ‘are you angry’. A meek person would be likely to
answer ‘no’, no matter how the question was asked. However, if you were to ask a moody,
volatile person this question, it is quite possible that the response would be highly sensitive to
your facial expression and the tone of voice in which the question was asked.
The degree of sensitivity varies from one entity to another and from one context to another.
In one environment a given entity will be completely adapted, and this environment may only
rarely offer contexts that cause a change of state. In another it will barely survive, and this
environment may continuously provide contexts that cause change.
12.3.4 Context Dependence and Context Independence
The degree to which the structure of an evolving entity reflects particular aspects of its
environment will be referred to as context dependence. One can conceive of a sort of
contextuality that would not, however, qualify as context dependence. The degree to which an
entity is able to withstand, not a particular environment, but any environment, will be referred to
as context independence. Thus, sensitivity to context can lead, in the long term, to either context
188
Chapter 12: Toward a General Theory of Evolution
dependence or context independence. It seems reasonable to expect that this depends on the
variability of the contexts to which an entity is exposed. A static environment is likely to provide
contexts that foster specializations tailored to that particular environment, whereas a dynamic
environment is likely to foster more general coping mechanisms.
An example of context dependence is a species that develops an intestine specialized for the
absorption of nutrients from a certain vegetation that is abundant in its environment. An example
of context independence is a species that becomes more able to consume any sort of vegetation. A
cultural example of the first is a person (or society) who mirrors the behavior and attitudes of
those around them. A cultural example of the second is a person (or society) that finds general
principles that prescribe ways of being that work in many situations.
Whether an evolutionary system exhibits context dependence or context independence may
simply reflect what one chooses to define as the entity of interest. If the entity splits into alternate
parallel versions of itself (as through reproduction) each of which adapts to a different context
and thus becomes more context dependent, then when all parallel versions are considered
different lineages of one joint entity, that joint entity is becoming more context-independent.
12.4 DIFFERENT WAYS REALITY HAS FOUND OF ACTUALIZING POTENTIAL
We now turn to how many different kinds of evolution are all means of actualizing potential that
was present due to the state of the entity, the context, and the nature of their interaction. For
simplicity, the abbreviation CAP will be used to refer to context-driven actualization of potential.
12.4.1 Nondeterministic Collapse of a Quantum Entity as a Kind of CAP
A quantum entity exists in a genuine superposition state and a measurement will cause it to
collapse to an eigenstate of that measurement in a nondeterministic way. The specifics of the
measurement provide the context that elicit in the entity one of the states that were previously
potential for it. The evolution of such an entity cannot be examined without performing
measurements—that is, introducing contexts—but the contexts themselves unavoidably affect its
evolution. Thus the evolution of a quantum particle is a paradigmatic example of CAP because its
state at any point in time reflects both its history of previous states, and the contexts to which it
has been exposed.
12.4.2 Deterministic Evolution of Quantum or Classical Entities as a Second Kind of CAP
We saw that the deterministic change of state of an entity that is not in a superposition state can
be viewed as an experiment for which there is only one potential outcome; there is only one way
189
Liane Gabora
the context can cause the entity to collapse. So the dynamical evolution of a quantum entity as per
the Schrödinger equation reduces to a collapse for which there was only one way to collapse. It
too is an example of the actualization of potential because its state at any point in time reflects its
history of previous states. Deterministic change under the influence of a context is really just a
reflection of the fact that we have sufficient knowledge about the context to make acceptably
accurate predictions. This also holds for the deterministic evolution of classical entities. Also, if a
quantum entity is in an eigenstate with respect to a context, the change produced by this context is
deterministic. In this case, the context has no influence on the state of the entity.
12.4.3 Biological Evolution as a Third Kind of CAP
Biological evolution can be viewed as the situation where an entity with the capacity to selfreplicate—the first living organism—splits into different lineages, each of which is exposed to a
different series of contexts. Most of the changes of state these various lineages of the biological
entity undergoes are relatively deterministic. However, when the interaction between its internal
state and its context is such that the outcome could be either death, or the generation of offspring,
it enters a situation more like the context-driven collapse of a quantum entity. In the life-or-death
situation, the potential for life and the potential for death constitute a potentiality state, and
through interaction with the context the entity collapses to either one or the other of them. In the
mating situation, the different offspring (different variants of itself) it could have with this mate
constitute a potentiality state, and it collapses to one or more of them. Thus the mate constitutes
the context for which the state of the biological entity is a potentiality state. Natural selection is
now viewed as the effect of the physical and social environment on the collapses involved in lifeor-death and mating events.
Speciation can be viewed as the situation wherein one kind of variant no longer has the
potential to create a context for the other for which its state is a potentiality state with respect to
offspring. A species or lineage is adapted to its environment to the extent that its previous states
have been potentiality states that could collapse to different possible outcomes, and thus its
history of collapses has reflected the contexts to which it has been exposed. Note that although
each species becomes over time increasingly context dependent, as a collective entity, living
organisms are becoming more context independent. (That is, for virtually any environment one
can think of there exists at least some branch of organic life that can cope with it.)
190
Chapter 12: Toward a General Theory of Evolution
12.4.4 Stream of Thought as a Fourth Kind of CAP
A stream of thought is similarly composed of both relatively deterministic segments and collapse
events. During the relatively deterministic segments, the state of the mind is not in a potentiality
state with respect to the current context. Thus the state of mind changes state in a way that reflects
the context to which it is exposed, but only deterministically. A typical example might be the
change of state of your mind as you are driving a car along a familiar route, not thinking about
anything in particular, just following the car behind you, obeying the road signs, and so forth. In
[Aerts and Gabora 2001] we discuss how one can approach the formal description of this sort of
change of state of mind in an analogous way to the description of change provided by the
Schrödinger equation.
During collapse events, the mind is in a potentiality state with respect to the current context,
where that context could be external, or generated internally, such as through imagining or
fantasizing or mentally simulating a possible scenario or event. A typical example is a state of
indecision, such as the example of the state of indecision as to which candidate to vote for,
presented in chapter three. There are many ways the state of mind could potentially change,
depending on the specifics of the interaction between this state of mind and this internally or
externally generated context. Under the influence of this context, the mind undergoes a collapse
event that actualizes one of the states that was previously potential. Thus the state of mind
nondeterministically reflects the stimulus or situation to which it was exposed.
12.4.5 Cultural Evolution as a Fifth Kind of CAP
In culture, interacting individuals often provide the context that evokes such collapse events in
one another. This kind of nondeterministic change of state may, as in biology, be more likely to
occur when the individuals involved have the potential to either end each other’s stream of
thought (a life-or-death situation for the idea under reflection), or develop it further (generate
progeny for the idea under reflection). Thus cultural entities evolve through this combination of
deterministic and collapse-driven changes of state in one or more minds interacting either
directly, or indirectly via artifacts.
Cultural entities, like biological entities, also have the potential to split off into multiple
variants of themselves. These variants can range from virtually identical to virtually impossible to
trace back to the same ‘parent’ idea. They affect, and are affected by, the minds that encounter
them. For example, books can affect all the individuals who read them, CDs can affect all the
individuals who play them, movies can affect all the individuals who watch them, and so forth,
191
Liane Gabora
and these individuals subsequently provide new contexts for the possible further evolution of the
ideas they described and stories they told.
12.5 SUMMARY
Borrowing from how potentiality and contextuality are handled in the generalized quantum
mechanical formalism, we start to construct a theory where evolution is more broadly construed
as the incremental change that results from the recursive, context-driven actualization of
potential. The description of the state of an evolving entity is treated as a potentiality state, and
the contextual interaction with the environment plays the role of the measurement by causing the
entity to collapse to an eigenstate of this interaction.
When one takes this more general perspective, the scientific basis for viewing culture as an
evolutionary process becomes clearer. Moreover, we reach a more general understanding of how
it is that something could evolve. Quantum, classical, biological, conceptual, and cultural
evolution become united under one umbrella. In this general approach they appear as different
ways in which potential that is present due to the state of an entity, its context, and the nature of
their interaction, is recursively actualized. These different kinds of evolution differ with respect to
amount of indeterminism, sensitivity to context, whether the evolution is bringing about context
dependence or context independence, and the degree to which the effects of changes of state are
retained in the further evolution of the entity and/or its lineages. In general, the states of quantum
entities collapse to different states depending on context. The deterministic evolution of classical
entities and the dynamical evolution of quantum entities can be viewed as collapse events for
which there is only one possible way to collapse; thus their evolution is steered deterministically
by context. In biological evolution, the entity splits into different variants of itself, each of which
is exposed to a different series of contexts. Most of its changes of state are relatively
deterministic. However, when the interaction between its internal state and its context is such that
the outcome could be either death, or the generation of offspring, it enters a situation that is akin
to the context-driven collapse of a quantum entity. Cultural entities can similarly split into
different variants, each of which is exposed to a different series of contexts. Since cultural entities
do not necessarily die, nor rely on the interpretation of self-replication instructions, their changes
of state can constitute their next generation, and directly affect the lineage of which they are a
part. As a consequence, cultural evolution is less deterministic than biological evolution.
192
Chapter Thirteen
13 Contextualizing Theories of Concepts and Culture
This chapter explores in more detail how the theoretical approach to evolution introduced in
chapter ten applies to cultural entities and the minds that evolve them. We begin by proving that
the mind has the sort of structure that is necessary for the approach to work. We then show how
the approach can overcome problems that have arisen with other approaches to cognition and
culture. The chapter ends with some speculative ideas for future research.
13.1 QUANTUM STRUCTURE IN THE MIND
In chapter ten we showed that the type of structure first revealed in the microworld, involving
superposition states and indeterminism due to the context, can be tested for using Bell
inequalities. Thus, to show the approach outlined there is indeed valid, it is suitable to begin by
determining whether there is a violation of Bell inequalities in the entities of interest. In this
section we show how Bell inequalities are violated in the mind in virtue of the relationship
between abstract concepts and specific instances of them. This was first presented in [Aerts &
Gabora 1999, Aerts et al. 2000]. An analogous line of reasoning could be used to demonstrate the
presence of potentiality states and indeterminism due to the context in the genome.
13.1.1 The Violation of Bell Inequalities in Cognition
To make clear how Bell’s insight applies to concepts, the explanation is framed in a concrete
example. Keynote players in this example are the two cats, Glimmer and Inkling, that live at our
research center (Figure 13.1). The experimental situation has been set up by one of the authors
(Diederik) to show that the mind of another of the authors (Liane) violates Bell inequalities. The
situation is as follows. Liane is preparing food for her cats, thinking abstractly about cats, but not
about any cat in particular. On the table where she prepares the food she finds a little note that
193
Liane Gabora
says: ‘Think of one of your cats now’. To show that Bell inequalities are violated we must
introduce four experiments e1, e 2, e 3 and e 4. Experiment e1 consists of Glimmer showing up at the
instant Liane reads the note. If, as a result of the appearance of Glimmer and Liane reading the
note, the state of her mind is changed from the more general concept ‘cat’ to the instance
‘Glimmer’, we call the outcome o1(up), and if it is changed to the instance ‘Inkling’, we call the
outcome o1(down). Experiment e3 consists of Inkling showing up at the instant that Liane reads
the note. We call the outcome o3(up) if the state of her mind is changed to the instance ‘Inkling’,
and o3(down) if it is changed to the instance ‘Glimmer’, as a result of the appearance of Inkling
and Liane reading the note. The coincidence experiment e13 consists of Glimmer and Inkling both
showing up when Liane reads the note. The outcome is (o1(up), o3(down)) if the state of her mind
is changed to the instance ‘Glimmer’, and (o1(down), o3(up)) if it changes to the instance ‘Inkling’
as a consequence of their appearance and the reading of the note.
Figure 13.1 Inkling on the left and Glimmer on the right. (This picture was taken after Glimmer decided that the quantum
cat superstar life was not for him, and removed his bell.)
It is necessary to know that occasionally the secretary puts bells on the cats’ necks, and
occasionally she takes the bells off. Thus, when Liane comes to work, she does not know whether
or not the cats will be wearing bells, and she is always curious to know. Whenever she sees one of
the cats, she eagerly both looks and listens for the bell. Experiment e2 consists of Liane seeing
Inkling and noticing that she hears a bell ring or doesn’t. We give the outcome o2(up) to the
experiment e2 when Liane hears the bell, and o2(down) when she does not. Experiment e4 is
identical to experiment e2 except that Inkling is interchanged with Glimmer. The coincidence
experiment e14 consists of Liane reading the note, and Glimmer showing up, and her listening to
whether a bell is ringing or not. It has four possible outcomes: (o1 (up), o 4(up)) when the state of
Liane’s mind is changed to the instance ‘Glimmer’ and she hears a bell; (o1(up), o 4(down)) when
194
Chapter 13: Contextualizing Theories of Concepts and Culture
the state of her mind is changed to the instance ‘Glimmer’ and she does not hear a bell;
(o1(down), o 4(up)) when the state of her mind is changed to the instance ‘Inkling’ and she hears a
bell and (o1 (down), o 4(down)) when the state of her mind is changed to the instance ‘Inkling’ and
she does not hear a bell. The coincidence experiment e23 is defined analogously. It consists of
Liane reading the note and Inkling showing up and her listening to whether a bell is ringing or
not. It too has four possible outcomes: (o2(up), o 3(up)) when she hears a bell and the state of her
mind is changed to the instance ‘Inkling’; (o2(up), o 3(down)) when she hears a bell and the state
of her mind is changed to the instance ‘Glimmer’; (o1(down), o3(up)) when she does not hear a
bell and the state of her mind is changed to the instance ‘Inkling’ and (o1(down), o 3(down)) when
she does not hear a bell and the state of her mind is changed to the instance ‘Glimmer’. The
coincidence experiment e24 is the experiment where Glimmer and Inkling show up and Liane
listens to see whether she hears the ringing of bells. It has outcome (o2(up), o4(up)) when both
cats wear bells, (o2(up), o 4(down)) when only Inkling wears a bell, (o2(down), o 4(up)) when only
Glimmer wears a bell and (o2 (down), o4(down)) when neither cat wears a bell.
We now formulate the necessary conditions such that Bell inequalities are violated in this
experiment:
•
The concept ‘cat’ is activated in Liane’s mind.
•
She does what is written on the note, namely think of one of her two cats.
•
The state of her mind changes from the concept ‘cat’ to the instance ‘Glimmer’ when she sees
Glimmer, and to the instance ‘Inkling’ when she sees Inkling.
•
Both cats are wearing bells around their necks.
The coincidence experiment e13 gives outcome (o1(up), o 3(down)) or (o1(down), o 3(up)) because
from (2) it follows that Liane will think of Glimmer or Inkling. This means that E13 = -1. The
coincidence experiment e14 gives outcome (o1(up), o4(up)), because from (3) and (4) it follows
that she thinks of Glimmer and hears the bell. Hence E14 = +1. The coincidence experiment e23
also gives outcome (o2(up), o 3(up)), because from (3) and (4) it follows that she thinks of Inkling
and hears the bell. Hence E23 = +1. The coincidence experiment e24 gives (o2(up), o4(up)),
because from (4) it follows that she hears two bells. Hence E24 = +1. As a consequence we have:
| E13 - E14 | + | E23 + E24 | = +4
(13.1)
195
Liane Gabora
The reason that Bell inequalities are violated is that Liane’s state of mind changes from activation
of the concept ‘cat’, to activation of either ‘Glimmer’ or ‘Inkling’, and that this change is
provoked by the context. It also shows that the indeterminism that is revealed when the two cats
appear and Liane thinks of one of them, is of the quantum nature, due to a lack of knowledge of
how the context (Glimmer and Inkling both appearing) interacts with the state of the entity (the
state of Liane’s mind with the concept ‘cat’ actualized). This means that the concept ‘cat’ cannot
be regarded as being a statistical mixture of the instances Glimmer and Inkling, as if in a hidden
way, Glimmer or Inkling was already actualized, but this was unknown and got revealed when
they both appeared. It also shows that this process of change cannot be described by means of a
stochastic process.
We can thus view the state ‘cat’ as a superposition state of these two instances of it. This
means that concepts can, like micro-particles, be in superposition states, and that the architecture
of the mind is in this way similar to the architecture of the micro-world.
I end this section by apologizing for the pun on Bell’s name, but it seemed like a good way
to ring in these new ideas.
13.1.2 Implications
The previous section showed that the mind does indeed have the same sort of structure as the
microworld, with potentiality states and nondeterministic due to context. This indicates the
validity of treating concepts as existing in potentiality states that can collapse to one state or
another depending on context. One might argue that the architecture of cognition doesn’t seem
superposed. When you look at a cat you see a discrete entity that occupies a specific location.
Even when you think of the abstract concept ‘cat’ or ‘chair’, you think of a certain kind of thing
in a certain kind of place. But when you consciously think of specific memories or concepts, you
usually re-experience them in their collapsed form. They are only superposed in the form they are
in when they are not in a collapsed state, when they are not being consciously experienced, or
when they are still vague or undecided upon. However, the fact that they were in an superposed
state just previous to their being evoked does trickle into conscious experience, perhaps more than
we are generally aware. If a devastatingly beautiful and fascinating person sat down in your chair
today, your experience of it tomorrow might be vividly different, even cause your heart to beat
faster, and so forth. In other words, an experience of a chair isn’t as disconnected from everything
else as one might be tempted to think.
The theory that both a stream of thought and the evolution of ideas and artifacts through
cultural exchange can be broadly viewed as context driven actualization of potential, and that this
196
Chapter 13: Contextualizing Theories of Concepts and Culture
potential exists largely in virtue of a quantum type of structure in the mind, has potentially farreaching implications. The next few section highlight some of the preliminary explorations that
are underway. But it is perhaps a good idea to first say a few words about how this approach
compares to other seemingly similar approaches with which the reader might be familiar. Over
the past several decades, numerous attempts have been made to forge a connection between
quantum mechanics and the mind [e.g. Penrose & Hameroff 1998]. Despite a superficial
similarity between these approaches and our own, the methods and underlying motivations are
different. In these other approaches, it is generally assumed that the only way the two could be
connected is through micro-level quantum events in the brain exerting macro-level effects on the
cognitive functions of the mind. In contrast, our approach deals not with quantum particles or
events, but rather with the formal structure they brought to light.42
13.2 DESCRIBING STATES OF MIND THAT ARE UNFOCUSED OR UNDECIDED
Chapter three showed that not all forms of thought can be viewed as a Darwinian process of
variation and selective retention, because in order to select one thought from a set of possible
variant thoughts it is necessary to first consider and evaluate each as a distinct, separate entity.
This cannot be done without actualizing it, not necessarily in physical space, but at least in
conceptual space. But in so doing, one changes both the initial state—such that it cannot be
returned to in order to consider the next variant—and the selection criterion—such that the next
variant is subjected to a different selective pressure. The problem boils down to the fact that
selection theory works with distinct, actualized states, but that it was not developed to handle, nor
can it cope with potentiality states, such as states of indecision, particularly when one hasn’t yet
reached the point of yet being able to clearly articulate the alternatives, or where different
outcomes are possible depending on the context to which the entity is exposed. Moreover we saw
that to represent a state of potentiality as a set of possible subsequent, states and say that this
42
Nevertheless, our approach is not incompatible with theirs. Consider the water example of Chapter 10. It
is because the water has the potential to collapse into particular distribution across the cups given a certain
context—a set of tubes of particular (but undetermined) lengths and widths—that it has quantum structure.
This quantum structure rests on the fact that fluids by nature are able to flow together to become one, and
separate again to become many as well as the fact that the situation that forces the water to separate is not
completely deterministic. In turn, the macroscopic behavior of water rests on the fact that the molecules
that constitute a fluid behave the way they behave, and this behavior is due to their quantum nature. So in
the end it is more accurate to say is that the quantum structure described here is compatible with two
possibilities, and not committed to either of them: (1) a micro-level entity in the brain that can harness
quantum mechanical effects resulting in a unique sort of macro structure in the mind as Hameroff and
Penrose suggest, or (2) a commonplace macro structure, like water, situated such that it has ample
opportunity to exhibit its potential for context-dependency, where that context-dependency can be traced to
its micro-level structure.
197
Liane Gabora
subsequent state is one or another of them does not give us an ontologically accurate description
of its state if there is any degree of contextuality present.
How then does one describe a state of mind that could change in different ways depending
on the stimulus or biological drive or overarching goal? Aerts and Aerts [1996] proved that in
situations where one moves from a state of indecision to a decided state, the probability
distribution is non-Kolmogorovian, and therefore a classical probability model cannot be used.
Moreover, they proved that such situations can be accurately described by treating the indecision
state as a potentiality state that undergoes collapse, and viewing the evolution as context-driven
actualization of potential, as we have been discussing.
13.3 HOW THE PET FISH PROBLEM IS SOLVED BY CONTEXTUALIZING CONCEPTS
In chapter five, we discussed two current representational theories of concepts: prototype and
exemplar theories. These theories are adequate for predicting experimental results for many
dependent variables including typicality ratings, latency of category decision, exemplar
generation frequencies, and category naming frequencies. However, they run into problems trying
to account for the creative generation of, and membership assessment for, conjunctions of
concepts. They cannot account for phenomena such as the so-called guppy effect, where ‘guppy’
is not a good example of ‘pet’, nor is it a good example of ‘fish’ but it is indeed a good example
of ‘pet fish’ [Osherson & Smith 1981]. This is problematic because if (1) activation of ‘pet’ does
not cause activation of ‘guppy’, and (2) activation of ‘fish’ does not cause activation of ‘guppy’,
how is it that (3) ‘pet fish’, which activates both ‘pet’ AND ‘fish’, causes activation of ‘guppy’?
(In fact, it has been demonstrated experimentally that other conjunctions are better examples of
the guppy effect than ‘pet fish’ [Storms et al. 1998], but since it is well-known we will continue
to use it here as an example.)
Zadeh [1965, 1982] tried, unsuccessfully, to solve the conjunction problem using a minimum
rule model, where the typicality of an item as a conjunction of two concepts (conjunction
typicality) equals the minimum of the typicalities of the two constituents. Storms et al. [2000]
showed that a weighted and calibrated version of the minimum rule model can account for a
substantial proportion of the variance in typicality ratings for conjunctions exhibiting the guppy
effect, suggesting that the effect could be due to the existence of contrast categories. However,
another study provided negative evidence for contrast categories [Verbeemen et al., in press].
The problems that arise with conjunctions reflects a more general problem with
representational theories [see Riegler, Peschl & von Stein 1999, for overview]. As Rosch [1999]
puts it, they do not account for the fact that concepts “have a participatory, not an identifying
198
Chapter 13: Contextualizing Theories of Concepts and Culture
function in situations”; that is, they cannot explain the contextual way in which concepts are
evoked and used [see also Gerrig & Murphy 1992; Hampton, 1987; Komatsu, 1992; Medin &
Shoben, 1988; Murphy & Medin, 1985]. Not only does a concept give meaning to a stimulus or
situation, but the situation evokes meaning in the concept, and when more than one is active they
evoke meaning in each other. It is this contextuality that makes them difficult to model when two
or more arise together, or follow one another, as in a creative construction such as a conjunction
or a sentence.
This story has a precedent. The same problem arose in physics in the last century. Classical
mechanics could describe situations where the effect of the measurement was negligible, but not
situations where the measurement intrinsically influenced the evolution of the entity. This
shortcoming of classical mechanics was also revealed when it came to explaining what happens
when many entities become entangled. In quantum mechanics, if some arbitrary context causes
two quantum entities to interact, then according to the dynamical evolution described by the
Schrödinger equation, the two quantum entities spontaneously enter an entangled state which
contains new properties that the original entities did not have. This means that the mathematical
structure incorporates the possibility to describe the birth of new states and new properties. As we
know, for these situations it was necessary to develop the formalism of quantum mechanics,
which takes contextuality into account.
Similarly, the Pet Fish Problem can be overcome by taking a contextual approach to
concepts. Conjunctions such as this are easily handled by incorporating contextual effects: (1)
activation of ‘pet’ still rarely causes activation of ‘guppy’, and likewise (2) activation of ‘fish’
still rarely causes activation of ‘guppy’. But now (3) ‘pet fish’ causes activation of the potentiality
states ‘pet’ in the context of ‘pet fish’ AND ‘fish’ in the context of ‘pet fish’. Since for both, the
probability of collapsing on to the state ‘guppy’ is high, it is very likely to be activated. Thus we
have a formalism for describing concepts that is not stumped by a situation wherein an entity that
is neither a good instantiation of A nor of B is nevertheless a good instantiation of A AND B.
Examples such as this are evidence that the mind handles nondisjunction (as well as negation) in
a nonclassical manner [Aerts et al. 2000b].
To gain deeper insight into what is happening here, let us look at how the quantum
formalism can be used to model conceptual space. For a given stimulus, the probability that a
potentiality state representing a certain concept will, in a given context, collapse onto another
state representing another concept is related to the algebraic structure of the total state space, and
to how the context is represented in this state space. In the case of a mathematical model that used
pure quantum mechanics, this algebraic structure would be given by the vector space structure of
199
Liane Gabora
the complex Hilbert space: states are represented by unit vectors of the Hilbert space, and a
context by a self-adjoint operator on the Hilbert space. In the case of a mathematical model that
uses the generalized quantum formalisms, the algebraic structure is then, for example, a state
property system. So now let us consider how this way of describing conceptual space enables
contextuality to be modeled in our example. Whereas in representational approaches such as
prototype and exemplar theories, relations between concepts arise through overlapping contextindependent distributions, in the present approach, the closeness of one concept to another
(expressed as the probability that its potentiality state will collapse to an actualized state of the
other) is context-dependent. Thus it is possible for two states to be far apart from each other with
respect to a one context (for example ‘fish’ and ‘guppy’ in the context of just being asked to name
a fish), and close to one another with respect to another context (for example ‘fish’ and ‘guppy’ in
the context of both ‘pet’ and being asked to name a ‘fish’).
Future plans include to compare the performance of the contextualized theory of concepts
with prototype and examplar theories using previous data sets for typicality ratings, latency of
category decision, exemplar generation frequencies, category naming frequencies on everyday
natural language concepts, such as ‘trees’, ‘furniture’, or ‘games’. The purpose of these initial
investigations will be to make sure that the proposed formalism is at least as successful as
representational approaches for the simple case of single concepts. Assuming this to be the case,
we will then concentrate our efforts on conjunctions of concepts, since this is where the CAP
approach is expected to supercede representational theories. We will re-analyze previously
collected data for noun-noun conjunctions such as ‘pet fish’, and relative clause conjunctions
such as ‘pets that are also fish’ [Storms et al. 1996]. A new study will be performed which will
compare the proposed approach with representational approaches at predicting the results of
studies using situations that are highly contextual. Typicality ratings for conjunctions will be
compared with, not just their components, but with other conjunctions that share these
components. (Thus, for example, does ‘brainchild’ share features with ‘childbirth’ or
‘brainstorm’? Does ‘brainstorm’ share features with ‘birdbrain’ or ‘sandstorm’?)
13.4 USING THE CAP APPROACH TO MODEL CREATIVITY
Conjunction is a useful place to begin applying the CAP approach, not just because it is the
creative process—where an entity with new properties comes into existence—that most highlights
its advantages, but also because conjunction is a relatively simple form of creativity. However,
most creativity is of course more complex than the mere conjunction of concepts. Here I will try
to convey how the CAP approach can be used to describe more complex instances of creativity.
200
Chapter 13: Contextualizing Theories of Concepts and Culture
13.4.1 An Example: Oga Invents the Torch
For concreteness, let us take as an example Og and Oga. They now sit before a blazing bonfire,
listening to the chanting and drumming of their fellow clanspeople. Og is actually one of the
drummers, the best drummer of the clan in fact, and he is completely immersed in the situation.
His state of mind is very similar to that of most of the other clanspeople. However, Oga’s state of
mind is not quite harmonized with that of the others in this way because she is focused on the fact
that there is a wind coming from behind her, and her back is therefore cold. She yearns for a
warm water buffalo fur to drape over her shoulders to keep her back warm. She knows there is
one in the cave, but doesn’t think she would be able to find it on this dark, moonless night. The
entity S that we consider is Oga’s mind, and the trajectory of her stream of thought is summarized
in Figure 13.2.
Figure 13.2. The trajectory of the changes of state that take place in entity S, Oga’s mind, culminating in the invention of
the torch. (The effects of the contexts are not explicitly shown.)
The set of states of mind {p1, p2., p3... pn} relevant to what happens to S during the considered
period of time is denoted Σ. Note that n is not the number of states that get actualized during one
of the possible set of events that could happen in this time interval, but the number of relevant
states (including those that could take place in different possible realizations). The set of contexts
{e1, e2., e3... em} relevant to what happens to S during the considered period of time is denoted M.
Again, note that m is not the number of actualized contexts during one of the possible set of
events, but the number of relevant contexts (including those that might remain potential in one of
the possible sets of events, but could get actualized in another one).
201
Liane Gabora
If an entity S is in a state pi _ Σ and a context ej _ M is present, we denote (pi, ej, pk, el) the
probability that the context ej will make the state pi collapse to the state pk, and the state pi will
make the context ek change to the context el. The probability
is a function defined as follows:
: Σ x M x Σ x M → [0, 1]
(13.2)
(pi, ej, pk, el) |→ (pi, ej, pk, el)
such that for pi _ Σ and ej _ M we have:
∑k, l
(pi, ej, pk, el) = 1
(13.3)
Equation 13.3 follows from the fact that the interaction between state pi and context ej provokes a
collapse to one of the states pk _ Σ and contexts el _ M with certainty. We remark that for a given
state pi and context ej we have 0 ≤ (pi, ej, pk, el) ≤ 1 for all pk _ Σ and el _ M. For a state pi and
context ej it is interesting to consider the set of states and contexts for which there is a non-zero
probability of collapsing to. We call this the range of potentiality state pi under context ej.
During the interval of time [t0, t1], the entity S which is Oga’s mind has undergone a change
of state that brought her to the initial state of mind that we consider. This state is realized at time
t1, and we denote it p1. The state p1 is experienced as the thought ‘I must see better in cave’. The
context in which this state p1 occurs, which involves sitting before a bonfire with a cold back, is
denoted e1. The state p1, constitutes a potentiality state with respect to the current context, e1.
There are different possible changes of state from p1 given context e1. These possible changes of
state derive in part through the relational structure of her worldview. For instance, she knows
something about the relationship of fire to light, and of light to the act of seeing; she knows that
light enables one to see better, and that the farther away one goes from the source of light, the less
it helps one to see. Let us say that the range of S at the next instant of time t2 consists of three
possible states. The first is the state of mind p2 experienced as the idea ‘move bonfire to cave’.
The second possibility exists only because she is an exceptionally brilliant cave girl, as renowned
for her ingenuity as Og is for his drumming. It is the possibility that she will have the thought that
a burning ember could be used as a torch, and she could carry it with her to light up the cave. The
state of mind corresponding to this thought is denoted p3. Even for Oga, the probability of have
this thought is, at this point anyway, still very small: 0.01. The third possibility is the state of
mind where she does not have this idea, and so at t2 she is still in state p1. Let us say that the
202
Chapter 13: Contextualizing Theories of Concepts and Culture
probability of having this idea is 0.29. Hence the probability of not having this idea is 0.7. Thus
for the interval of time [t1, t2] we have:
(p1, e1, p1, e1) = 0.7 and
(p1, e1, p2, e1) = 0.29 and
(p1, e1, p3, e1) = 0.01
(13.4)
Furthermore, we have that:
(p1, e1, pi, ej) = 0 for i ≠ 1, 2, 3 or j ≠ 1
(13.5)
Note that (p1, e1, pi, ej) = 0 for j different from 1 expresses the fact that the state p1 of Oga’s
mind does not cause a change of context e1 to one of the other contexts ej, for any j different from
1. In other words, state p1 does not exert a significant feedback effect on Oga context. We could
have imagined a more complicated situation. For example, an annoyed expression on Oga’s face
due to the cold could have caused Og to stop playing the drums, and this in turn could perhaps
have caused the others to stop, which would really change Oga’s context. Or, for example, Og
might have put his arm around Oga, thus making her back warmer, which would have taken away
the incentive to find the fur.
Let us say that at the instant of time t2, Oga has the idea ‘move bonfire to cave’. Thus the
state of mind p1, experienced as the thought ‘I must see better in cave’ changes to the state of
mind p2 experienced as the idea ‘move bonfire to cave’, under the influence of the context e1 of
sitting before a bonfire with a cold back.
Her next thought, given both the thought experienced in the state of mind p2, and the
relational structure of her model of the world, is an inevitable thought to have. It is the realization
that moving the bonfire to the cave is impossible. We denote this state of mind p4. Since in this
context the state of the mind does not constitute a potentiality state, it can be modeled as a
deterministic context. Hence for [t2, t3]:
(p2, e1, p4, e1) = 1 and
(p2, e1, pi, ej) = 0 for i ≠ 4 or j ≠ 1
(13.6)
During the interval of time [t3, t4], the state of entity S, Oga’s mind changes to a potentiality state
p5 that is similar to p1 experienced as the thought ‘must see better in cave’. Like the change of
state from p2 to p4, this is a deterministic change of state. Since the new idea wasn’t workable
there is no point to continue further with it; there is nothing to do but go back to the problem of
how to see better in the cave and look for another solution. Note that the new state p5 is not
203
Liane Gabora
identical to state p1, even though we characterize them both as being experienced as the thought
‘must see better in cave’. The reason p5 is not identical to state p1 is that it takes place in the
context of having considered, as a possible solution, the state p2 experienced as ‘move bonfire to
cave’. The realization that this wouldn’t work modified ever so slightly the relational structure of
her worldview. Since the worldview that generated the state p5 is not exactly the worldview that
generated the state p1, the state p5, which is the fruit of this slightly modified worldview, takes into
account something that was not taken into account by state p1. Even though the external
environment has not changed, Oga’s context has undergone a change. It changed to a new
context, e2, which incorporates not only the external contextual elements of e1 (the wind at her
back, and so forth), but also the internal restructuring her worldview underwent in the process of
considering the possibility of moving the bonfire to the cave. It undergoes this change with
certainty, hence for the interval of time [t3, t4], we have:
(p4, e1, p5, e2) = 1 and (p4, e1, pi, ej) = 0 for i ≠ 5 or j ≠ 2
(13.7)
The next moment, a burning ember rolls out of the fire several centimeters toward Oga. The
presence of the ember constitutes a new context, e3. Her perception of Which is the presence of
the ember this ‘moving fire stick’ stimulus at t5 generates a new state of mind p6. The change of
context, and change of state of mind happen with certainty, hence for the time interval [t4, t5]:
(p5, e2, p6, e3) = 1 and (p5, e2, pi, ej) = 0 for i ≠ 5 or j ≠ 3
(13.8)
With respect to this new context e3, the new state p6 is a potentiality state. For simplicity, let us
say that there are only two possibilities for what happens next. One is that her mind collapses to
the state p3, experienced as the idea ‘move fire stick to cave’; the idea to invent the torch, which
had probability 0.01 of being realized during the interval of time [t1, t2]. The other is that she does
not have this idea, but becomes mesmerized by the burning ember, so S goes to the state p7.
Having considered the very similar idea of moving the entire bonfire, and also because as we
know she is very brilliant, the probability that she now has this idea under the influence of the
rolling ember context e3 is 0.8. Hence the probability of her not having this idea and becoming
mesmerized by the ember is 0.2. Hence for the interval of time [t5, t6], we have:
(p6, e3, p7, e3) = 0.2 and (p6, e3, p3, e3) = 0.8
204
(13.9)
Chapter 13: Contextualizing Theories of Concepts and Culture
Furthermore:
(p6, e3, pi, ej) = 0 for i ≠ 3, 7 or j ≠ 3
(13.10)
Now, given that the probability was so high, perhaps it is not so surprising that we find that at the
instant of time t5, Oga does indeed have the idea ‘move fire stick to cave’ which is the experience
of the state p3. This state was potential all along but it required the appropriate thought trajectory
to make it more realizable, and the appropriate stimulus context to really elicit it. Oga picks up
the ember, makes her way to the cave, and comes back with the fur. During this period of going
to the cave and coming back with the fur, S undergoes various different changes of state and
changes of context, which we will not elaborate since the creative event we were interested in has
already taken place. Let us skip forward to when she sits before the bonfire with the fur draped
over her back. The state of her mind is as a state of contentment. The context involves sitting
before a flickering fire with a warm back and enjoying Og’s drumming and the chanting of the
clanspeople. Now there is no reason for Oga’s state of mind to undergo collapse events because it
is no longer a potentiality state with respect to her present context.
13.4.2 Why Oga’s Stream of Thought Could Not be Described Classically
The change of state that took place in Oga’s mind in the above example could not be described
using a classical mathematical formalism. The reason is that, throughout her stream of thought,
context exerts a nondeterministic effect on change of state, and this introduces a non
Kolmogorivian probability on the state space as discussed extensively in chapter eleven.
Another reason a classical formalism would not work is that the invention of the torch
involved the spontaneous appearance of a new state, the state of mind that conceives of the torch,
with a new property, the property of being able to move fire. Recall from chapter three that in
classical physics, a composite or joint entity can only be described as a product state of the two
subentities. Thus if X1 is the state space of the first subentity, and X2 the state space of the second
subentity, the state space of the joint entity is the Cartesian product space X1 _ X2. So if the first
subentity is ‘door’ and the second is ‘bell’, one can give a description of the two at once, but they
are still two, with separate properties. Classical physics cannot even describe the situation
wherein the two entities give a new entity that has all the properties of its subentities, let alone a
new entity that involves only certain properties of one subentity and certain of the properties of
the other. We saw that the problem can be solved ad hoc by starting all over again with a new
state space every time an episode of impossibilist creativity occurs; that is, every time there
205
Liane Gabora
appears a state that was not possible given the previous state space. However, in so doing we fail
to include exactly those changes of state that are most relevant to the generation of cultural
novelty (such as the change of state of the mind that occurred when the torch was invented).
Another solution could be to make the state space infinitely large to begin with. However, as we
saw in chapter five, we can only hold approximately seven properties in our mind at any one time.
So to model the change of state that takes place during a stream of thought, this is not a viable
solution.
13.4.3 Why a Pure Quantum Formalism Would Not Work Either
In chapter eleven we saw that in quantum mechanics, if H1 is the Hilbert space representing the
state space of the first subentity, and H2 the Hilbert space representing the state space of the
second subentity, the state space of the composite is not the Cartesian product, as it in classical
physic, but the tensor product, i.e., H1 ⊗ H2. The tensor product, due to the mathematical
procedure of how it is formed, always gives rise to new states with new properties. The new states
are specifically the entangled states that are potentiality states with respect to the two subentities.
This means that with this formalism it is possible to describe the spontaneous generation of new
states with new properties.
However, the invention of the torch also could not have been described with the pure
quantum formalism. One reason is that in the pure quantum formalism, a state can only collapse
to itself with a probability equal to one. This was not the case in the time intervals [t1, t2] and [t5,
t6]. In other words, it cannot describe situations of intermediate contextuality. Another reason is
that the pure quantum formalism cannot model the situation where the context is changed by the
state, as happened in the interval [t3, t4].
13.4.4 Why Generalized Quantum Formalisms can Describe Impossibilist Creativity
The generalized formalisms similarly enable one to describe the spontaneous generation of new
states with new properties. However, here the new states are not just the completely contextual
potentiality states, but also the potentiality states with varying degrees of contextuality. Thus it
can describe situations of intermediate contextuality as took place in the time intervals [t1, t2] and
[t5, t 6], where there was some probability of being affected by context, and some probability of
not being affected by context. Recall from chapter six that a creative insight can happen through
the simultaneous retrieval of memories or concepts that, although they always had the potential to
be evoked simultaneously, had to wait until the right context came along to be actualized. The
same basic explanation for the invention of ‘TropicAle’ applies to the invention of the torch.
206
Chapter 13: Contextualizing Theories of Concepts and Culture
However now we can see that the simultaneous retrieval can be viewed as a collapse, and we see
the role of context in evoking the relevant mental entities. This kind of approach is useful, and in
fact necessary, if one is to incorporate the possibility of unpredictable contexts forging
unexpected—although often useful or beautiful—connections between concepts, ideas, or actions.
Since the generation of novelty is the crux of cultural evolution, a move in this general direction
is also unavoidable in furthering our understanding of culture.
As another example of the need for being able to describe intermediate contextuality, given a
mind that is in the state of thinking about a door, and the context of hearing the song ‘Thriller’,
one isn’t limited to describing changes of state of mind that stick just to obvious thoughts about
doors, nor to changes of state of mind that switch completely to Michael Jackson. One can
describe changes of state of mind that involve contextually modifying the thought of the door to
incorporate Michael Jackson, such as, for example, transitioning to the state of mind of imagining
the door being opened by Michael Jackson. Or, given the context of needing to hear that someone
is at the door, it could describe the appearance in the mind of the concept ‘doorbell’. At first it
would be a potentiality state that could collapse to many different possible manifestations of the
idea, such as perhaps (1) just a loud bell at the door for the person to ring, or (2) something that
caused a bell to ring when the person stepped on something that detects weight, or (3) something
that, when pressed by the finger caused a bell to ring (the doorbell we actually use now). Which
direction the doorbell invention took would depend on how the mind used internally and
externally generated contexts to refine it. If the first route didn’t work, it might go back to this
potentiality state and try another route, but otherwise it would just continue refining until the idea
could be feasibly implemented. At this point, the state of the mind is no longer a potentiality state
with respect to the evolution of this idea in this mind, so there are no more collapses pertaining to
this context.
13.5 CONTEXTUALIZING A THEORY OF CULTURAL INTERACTION
The context that drives a collapse is not necessarily a direct interpersonal interaction, but an
indirect interpersonal interaction mediated through an artifact, or even a natural object such as the
ember of the Og and Oga example. Contextual effects can happen in subtle ways; for example,
when you look at the cup on your desk you are affected by collapses that went on in the mind of
whoever invented cups, and collapses that went on in the mind of whoever designed that
particular cup. Artifacts carry much of the potentiality of the mind—conceptual space—into the
world—physical space. In other words, even as manifest objects, they retain the potential to be
207
Liane Gabora
transformed in many ways depending on the context. For example, if a war breaks out, food
storage cellars might be used as bomb shelters to protect people from the violence.
However, let us now focus not on artifacts but on more direct interactions amongst minds,
since these after all constitute the original basis of culture. Chapter ten showed that potentiality
and contextuality are important aspects of evolution, but that it has been possible to get away
without paying serious attention to them in biology because their impact is largely not heritable.
With culture, things are not so simple. Contextual effects can restructure the worldview(s) of an
individual or entire society, and this restructuring can have an immediate impact on the evolution
of behaviors, artifacts, and so forth.
13.5.1 The Final Appearance of Og and Oga
In the bonfire scenario described above, the invention of the torch took place in one single mind.
But it could just as easily have happened in the course of social interaction, one person ‘carrying
the torch’, so to speak, for the next. Let us sketch out another scenario. This time we will not
outline the mathematical treatment in detail, but it would be carried out exactly as above. As
before, Oga’s state of mind becomes ‘must see better in cave’, and the context of desiring the fur
in the cave causes collapse to actualize the state ‘move bonfire to cave’. But as before, this cannot
be actualized. However, now let us say that as Oga collapses on ‘move bonfire to cave’, she
watches her mate Og roast a hunk of yak meat on the end of a stick. He moves the stick back and
forth over the fire, rotating it in hopes that the meat will cook evenly on all sides and not catch
fire. Oga’s state of mind becomes ‘cook meat on stick’. As it happens, the piece of meat contains
a lot of fat, and when the stick gets too close to the fire, the fat starts to burn, creating a flame at
the end of the stick. Og throws the stick with the burnt meat to the ground and growls in disgust.
However, in Oga’s mind, the context of desiring to retrieve fur from a cave by the light of a fire,
the potentiality state ‘stick burning’ collapses onto states which have to do with the properties of
carrying light, and being moveable, rather than, say, cooking meat, or being made from a branch.
So in this context it collapses on: ‘burning stick moves light’. She picks up Og’s stick and runs to
her cave. By the light of the stick she grabs the fur, and returns excitedly to the bonfire. Others of
the clan watch her, and the new potentiality state ‘burning stick moves light’ is thereby culturally
transmitted. Some of them imitate what she did (sometimes modifying the basic idea to suit their
own needs; see Figure 13.3). Thrilled with the new invention, they run around waving burning
sticks, lighting parts of their world that were in the dark.
208
Chapter 13: Contextualizing Theories of Concepts and Culture
Figure 13.3 Og was so excited about the idea of a torch that sometimes he didn’t even take the time to light it.
13.5.2 Another Example: Mike and Abby and the Four Mad Scientists
As another example of what a powerful impact contextuality has on the cultural dynamic,
consider even the tiny segment of cultural exchange contained in the following scenario:
Two teenagers, Mike and Abby, sit next to one another on a train and get into a conversation. Let us make
the situation slightly more complex. Four scientists, who independently developed models of cultural
evolution, decide to use this train conversation as a miniature testbed against which to evaluate their
theories. Before the train starts, they plant a tape recorder under the seat that the teenagers eventually
occupy. When the train stops, they retrieve the tape recorder, listen to the conversation, and debate which
of their models best describes how ideas unfolded and evolved as they got mulled over and expressed in the
minds of the two people who sat there.
The first scientist, a mathematical biologist named Richardson, presented a population geneticsinspired model of information transmission. It demonstrated some interesting phenomena, such as that
reducing the size of the society increased that chance of fixation on a smaller number of cultural traits.
However the emphasis on transmission evoked skepticism on the part of the other scientists, who argued
that the two teenagers were not just so creative but also so responsive to one another’s cues and signals that
nothing in their conversation seemed likely to have been transmitted in a form that much resembled the
form it was likely to have been in when it was originally learned or assimilated. Everything they said, and
the voice in which they said it, was tailored to match this particular person and this particular situation.
“Cultural change is represented by a mutation operator,” Richardson defended himself.
209
Liane Gabora
“That’s one heck of a mutation operator,” said the second scientist, a computer scientist named
Shankson, “if it can come up with something as self-referentially weird as ‘Freudian flip’, or a theory that
nylon sleeping bags create a ‘bad’ electromagnetic field around you that counteracts the positive effects of
sleeping close to the ground within reach of the supposed ‘good’ electromagnetic field of Mother Earth.
And what in your theory could begin to address, for example, why Mike toned down his theory,
presumably in response to an expression on Abby’s face?”
Shankson had written a computer program that used scripts and grammatical rules to generate a
human-like conversation. The other scientists agreed that this program did capture somewhat the contextual
manner in which the conversation unfolded; the subject matter moved fluidly from a watch someone had
received for their birthday, to a friend’s upcoming birthday party, to the subject of age, to the subject of
parents. However, when they scrutinized the computer program, they found that it was put together in an ad
hoc way; it did not describe, neither at an abstract nor a concrete level, the formal structure of the
conversation that took place between the two teenagers.
The third scientist, a many worlds physicist named Deutschson, had written a program similar to
Shankson’s, but it used a random number generator to generate a large number of widely-diverging
conversations.
“See,” said Deutschson, “one of the conversations my program generated even touches on one of the
topics that actually came up in their discussion: Freud.”
“But spitting out lots of different conversations is cheating,” Schankson scoffed. Only one
conversation actually took place.”
“I was hoping you’d say that,” said Deutschson. “Now we get to the exciting part of my theory. The
other conversations were recorded by tape recorders left by alternate versions of ourselves in other
universes!”
The other scientists didn’t buy it. They turned to the fourth scientist, a quantum physicist who we will
call Dr. Q. What Dr. Q unveiled was a formal description of culture that drew upon, of all things,
mathematical formalisms for dealing with potentiality and contextuality that were originally developed for
quantum mechanics.
“Let it run,” said Shankson.
“Well, I must admit that for it to be completely accurate, we would have to know the entire history of
conceptual collapses in the minds of Mike and Abby.”
“Huh?” said Deutschson.
“Let me explain. Like Richardson’s, it is a mathematical model, but I think you will find that mine
really does capture the mathematical structure of what took place. Like Shankson’s, and to an even greater
extent Deutschson’s, it incorporates the contextual manner in which the conversation flowed. But unlike
Deutschson’s, it represents the two cognitive culture-generating structures using hierarchically structured
potentiality states. The various context-specific possibilities are encompassed by incorporating contextual
210
Chapter 13: Contextualizing Theories of Concepts and Culture
effects right into the mathematical description of an entity, and the output of one such entity is the context
that evokes a quantum-like collapse of the other.”
The other scientists were speechless, but under the given context, it’s not precisely certain what that
indicates.
If the approach proposed in this thesis proves correct, real-life cultural exchange such as this
conversation—with all its subtly, suggestiveness, spontaneity, and unpredictability—evolves
through the recursive assimilation and expression of collapsed actualizations of conceptual
networks. Each thought expressed in a conversation, each move in a game of chess or soccer,
each riff in a musical jam, is a collapse of cultural information in one mind that provides the
context for the collapse and further evolution of that information in one or more other minds. In
multiple minds, the cultural entity can evolve in ways it never could have evolved in a single
mind. In the mind of a single reflective individual, an idea can repeatedly collapse anew through a
series of internally generated contexts, a process referred to earlier as mental simulation.
However, all of these contexts come from the same worldview, the same interwoven conceptual
structure, with its particular memories, abilities, goals, and convictions. Sooner or later the idea
reaches the point where it has evolved as far as it can; in this mind it is no longer a potentiality
state, but an ‘almost’ deterministically steered state under this context. But in another mind, with
a different conceptual structure, prone to simulate different contexts, it might still be a potentiality
state. In a conversation between two or more individuals, each refinement of the idea in one mind
provides the other(s) with a context it probably would not have generated on its own. This is
clearly illustrated in the following fragment of the conversation that took place between the two
teenagers, Mike and Abby:
“Dude, did you notice how that creepy old guy stared at us as he walked by again?”
“Yeah. Hey, wouldn’t it be wicked if he was a mad scientist and we were the subjects of some kind of
weird experiment he was conducting?”
“Like I’m sure. Like he’s going to walk by and inject steroids into us. No way.”
“Way. He could be like, you know, Freud, analyzing what we talk about and stuff.”
“What I think would be really wicked is if us and the creepy guy were all just the inventions of some
scientists to show that our minds are quantum mechanical!”
13.6 SUMMARY
This chapter provided evidence that the CAP approach to cultural evolution developed in the
previous chapter is valid. First we saw that Bell inequalities are violated in the relationship
211
Liane Gabora
between an abstract concept (‘cat’), and specific instances of this concept (the two cats that live at
our research center and who sometimes wear bells). This proves the presence of superposition
states in the architecture of the mind. Then we reviewed why cognitive states cannot always be
viewed as situations of variation and selection, particularly when the state of mind is contextual
or undecided with respect to some issue or situation. The present approach overcomes this
shortcoming, and in fact is necessary for the description of states of indecision. Then we moved
to the cognitive situation where the contextual approach truly outshines other approaches: the
emergence of new states with new properties in the creative process. A simple sort of creativity is
conjunction. Even this simple sort of creativity has proved difficult for other approaches to
cognition. However the problem of conjunction is overcome by the present approach, as
illustrated using the Pet Fish Problem. An attempt is made to convey how this approach can be
used for more complex instances of creativity using as an example the invention of the torch.
Finally, we saw how the present approach can be used to understand the contextual manner in
which ideas and actions play off one another in real life cultural exchange.
212
Chapter Fourteen
14 Summary and Synthesis of Main Points
The goals of this thesis have been (1) to compare the mechanisms by which cultural change takes
place with those of biology, (2) where the mechanisms of culture differ, to explicate the
underlying mechanisms, (3) to propose a plausible explanation of how cultural evolution may
have begun, and (4) to attempt to arrive at a more general theory of evolution, of which culture
and biology are two types. In this final chapter, I summarize the contents of this thesis and
explain how these contents contribute to an understanding or clarification of these goals.
14.1 COMPARISON OF CULTURAL AND BIOLOGICAL EVOLUTION
It is clear that cultural entities—such as ideas, artifacts, mannerisms, and attitudes—evolve in the
general sense of adaptation to environmental constraints and affordances through descent with
modification. Agricultural techniques become more efficient, computers get faster, scientific
theories predict and account for increasingly more observed phenomena, artistic forms
simultaneously build upon and creatively modify previous ones. This isn’t to say that such
changes are necessarily improvements. It is rather to say that, as in biology, novelty generates
new situations that provoke or inspire more novelty.
This has led many to suggest that cultural evolution, like biological evolution, is Darwinian,
a process of natural selection causing the differential replication of randomly generated variants.
However, the application of Darwinian (or neo-Darwinian) theory to culture has proven far from
straightforward. We have looked at a number of approaches taken, including epidemiological and
population genetics inspired approaches, universal Darwinism and memetics, ideas from
complexity theory, selectionism and evolutionary epistomology. Each of these has helped clarify
the issues and led us closer to an understanding of how culture evolves. However, the application
213
Liane Gabora
of Darwinian theory to culture has never been fleshed out, nor gained universal support, in my
opinion because it provides only a partial explanation.
The primary difficulty applying evolutionary theory to culture arises largely because of the
complexity of the mind—the hub of cultural change—and the strategic, contextual, creative
manner in which it conceives, refines, expresses, assimilates, and evaluates or selects cultural
novelty. Therefore I have argued that a theory of cultural evolution must incorporate cognitive
processes acting on mental entities (such as ideas and attitudes) that are only potential, not
actualized, and not yet directly perceivable in the everyday world. It is through these processes
that cultural entities (such as actions and artifacts) take form. Thus the overall aim of developing
a coherent theory of the evolution of culture gave rise to a secondary goal of elucidating the
relevant underlying cognitive mechanisms.
14.1.1 Relation Between Biological and Cultural Evolution
It is useful to begin by clarifying the similarities and differences between biological and cultural
evolution. The following are the primary points established regarding the relationship between
these two forms of evolution.
14.1.1.1
Culture is not Simply an Extension of Biology
If you were to go back to some time during the first billion years of Earth’s history, you would
not find plants or animals, but you would find pattern in, for example, the weather, and the
trajectories of lava and water. The only causal principle you would need to invoke to account for
this pattern would be the physical constraints and self-organizing properties of matter. If you
were to go back to some time after the origin of life, approximately three billion years ago, this
would no longer be the case because it would be virtually impossible for, say, a giraffe to appear
in an information space not acted upon by natural selection. Another causal principle—biological
evolution—would have to be invoked from this point on. Today the Earth is embedded with
things like paintings, circuses, and computer networks that cannot be accounted for by appeal to
either the properties of matter or biological evolution. These entities are manifestations of yet
another causal principle: the evolution of culture. Since biology does not provide adequate
explanatory power to account for the existence of cultural entities any more than the properties of
matter can explain the existence of giraffes, culture cannot be dismissed as a natural extension of
biology.
214
Chapter 14: Summary and Synthesis of Main Points
14.1.1.2
Biological Needs Constrain but don’t Control the Evolution of Culture
Since many of our needs have a biological basis—e.g. the need for food, shelter, and so
forth—the generation of cultural novelty is largely constrained by our heritage as products of
biological evolution. However, as many have noted, the existence of self-destructive behaviors
indicates that survival needs do not exert complete control over the evolution of cultural entities.
My own contribution to this discussion is merely to try to clarify the relationship between
biological and cultural needs, and the resulting evolution of biological and cultural entities. Both
sorts of entities become increasingly complex over time; much as the evolution of rabbits created
new ecological niches for species that eat them and species that parasitize them, the invention of
cars created new cultural niches for gas stations, seat belts, and garage door openers. However,
whereas biologically-based needs remain relatively constant within a given generation, because of
the speed with which cultural entities evolve, culturally-derived needs change within a single
human lifetime. Thus the topology of the cultural fitness landscape does loosely follows that of
the biological fitness landscape, but there are places where it deviates, and this becomes
increasingly pronounced over the life span of a civilization, or even the lifetime of an individual.
14.1.2 Similarities Between Biological and Cultural Evolution
This section summarizes some of the most important similarities found between these two forms
of evolution. We first consider those derived through theoretical argumentation. We then consider
those revealed through the Meme and Variations (MAV) computer program.
14.1.2.1
Particulate versus Blending Inheritance
Genetic inheritance is said to be particulate because alternative alleles of a gene, rather than
blending together to form a new kind of allele, retain their original structure through successive
generations. For example, a cross of AA x aa will yield all Aa individuals, each of whom can
potentially have offspring that are either AA, aa, or Aa, depending on who they mate. In other
words, unlike how white and black paint mixed together can never become as purely white or as
purely black as the constituents of which it is composed, the pure AA and aa genotypes can
reappear full strength. (Also, new phenotypes can arise even in the absence of new alleles when
alleles at one or more loci get combined in new ways.)
Similarly, items in a memory get stored in different locations, and therefore retain their
original structure, even when they are retrieved from that memory and combined in new ways.
This is even true of a distributed memory such as a human memory, where items are stored with
varying strengths in multiple overlapping locations (though to a lesser extent than for a computer
215
Liane Gabora
memory or expert system; however the distributed memory is more able to be creative). Thus,
once a mind has conceived of the idea of a ‘houseboat’ it is still able to retrieve the concepts of
which it was composed: ‘house’ and ‘boat’. It is because a memory is an excellent potentialpreserving device that culture, like biology, can be characterized as a form of inheritance that is
particulate rather than blending. Just as biological evolution preserves the potential to re-manifest
AA or aa, the mind preserves the potential to re-manifest ‘house’ or ‘boat’.
14.1.2.2
Conceptual Linkage Disequilibrium
Recall that linkage disequilibrium refers to a state of nonrandom association amongst different
alleles of genes that are linked (on the same chromosome). Over generations the probability of
approaching a state of linkage equilibrium increases, because mutation and recombination disrupt
arbitrary associations amongst co-occurring alleles. The phenomenon wherein historical
association inhibits the re-association of parts to generate a new kind of whole appears also in
culture. It is seen for example in mental set, where an individual has difficulty applying an idea or
problem solving technique to situations other than the one where it was originally encountered, or
conversely, knowledge of one problem-solving technique interferes with the ability to solve a
problem using another technique. And as in the biological situation, this historical bias gradually
disappears with time. For example, although the first motorized vacuum cleaners and washing
machines looked like their pre-motorized predecessors, gradually they lost the aspects that had
been imposed by constraints that no longer existed, and took on their modern forms.
14.1.2.3
A Cultural Analog of Genetic Hitchhiking
Alleles that are important to survival often go to fixation in a population. The fixated allele is
surrounded by a region of reduced polymorphism, the size of which depends on the local
recombination rate. The alleles in this region are said to exemplify genetic hitchhiking. They
confer no fitness advantage, but endure because they are linked to the fixated allele.
This concept translates readily to culture. If an idea goes to fixation in a society due to
some advantage it confers, the region of conceptual space (and the domains of physical space)
containing related ideas will exhibit a ‘window’ of reduced polymorphism, the size of which may
vary according to how closely related or dependent they are upon the fixated idea. For example,
once midi became standard for electronic music keyboards, other electronic music gadgetry
became midi compatible.
216
Chapter 14: Summary and Synthesis of Main Points
14.1.2.4
Individuation in Families and Social Groups
Different species result when populations of organisms become so different from one another that
they cannot interbreed, which leads them to become even more different from one another. This
process is referred to as speciation. The concept of speciation can be applied to the individuation
and division of labor in a family or society. In both the biological case and its cultural analog,
small differences are amplified through positive feedback leading to the formation and
transformation of viable niches. This could provide answers to questions such as why siblings are
often so different from one another. As one sibling becomes increasingly identified as the athlete
of the family, other siblings, unable to compete in this particular domain, veer toward other
domains such as academics or music.
Other similarities between biological and cultural evolution were pointed out. For example, (1)
the digital revolution can be viewed as a phase transition analogous to the explosion of new
species in the Cambrian era, (2) the distinctive character of a certain peer group or society can
often be traced to a few distinguishing ideas invented by founding or influential members, a
cultural equivalent of the Founder Effect in biology, (3) the phenomenon of runaway selection in
culture, and (4) the insight that culture as well as biology suggests a straightforward explanation
for altruism.
Still other similarities were revealed in Meme and Variations (MAV), a computer model of
the process by which culture evolves. It was inspired by the genetic algorithm (GA), a minimal
computational abstraction of biological evolution. If culture, like biology, is a form of evolution,
it should be possible to analogously develop a minimal computational model of it. The model
consists of an artificial society of neural network based agents that don’t have genomes, and
neither die nor have offspring, but that invent, imitate, and implement ideas. Each iteration, every
agent has the opportunity to acquire a new idea, either through 1) innovation, by creatively
modifying a previously learned idea, or 2) imitation, by copying an action performed by a
neighbor. MAV explored the impact of three cultural phenomena that have no obvious biological
equivalents. The program exhibited numerous features observed in biology, some of which are
briefly summarized below.
14.1.2.5
Evolution Does Take Place in a Model of Culture
The average fitness of ideas gradually increased as agents modified ideas through innovation,
implemented them as actions, and imitated neighbors whose actions were fitter than their own.
Eventually one or more agents began to implement maximally fit actions, which spread through
217
Liane Gabora
imitation, and the society converged on these maximally fit actions. Evolution then stopped
because the society was completely adapted to its simple environment.
14.1.2.6
Cultural Drift
The model exhibited a cultural analog of genetic drift—changes in the relative frequencies of
different alleles due to random sampling processes in a finite population. One a particular
maximally fit idea happened to be found, it was there for good, and was more likely than other
possible maximally fit ideas to become widespread.
14.1.2.7
Frequency of Change Must be Intermediate
As in a GA, evolution did not occur in the absence of the creative or variation-generating
operations responsible for innovation. Also as in a GA, increasing the frequency of variation
increased diversity, both as the society was evolving, and in the final set of actions it stabilized
upon. However, increasing the frequency of variation much beyond the minimum necessary for
evolution caused average fitness to decrease and increased the number of generations required to
stabilize. This is also seen in a GA when the mutation rate is much above the minimum necessary
for evolution. In both MAV and the GA, the frequency of change must be intermediate between
the two extremes.
14.1.2.8
Epistasis Increases Time to Fixation
MAV exhibited a cultural analog of epistasis, where the fitness at one locus depends on which
allele is present at another locus. As in a GA, epistatic loci took longer to reach fixation than
nonepistatic loci. This provides at least a partial explanation for the previous finding that
increasing the frequency of change does not necessarily increase the rate of evolution, because it
increases the probability of disrupting schemata, or coadapted genes with epistatically linked
alleles.
14.1.3 Differences Between Biological and Cultural Evolution
This section summarizes the most important ways in which I have showed that these two forms of
evolution differ. Again, some are the result of theoretical argumentation, and others are results
observed in the MAV computer model. (Note that some of these differences will be elaborated
upon in forthcoming sections.)
218
Chapter 14: Summary and Synthesis of Main Points
14.1.3.1
Cultural Entities are not Replicators
A replicator, by definition, is an entity that makes copies of itself. A cultural entity is not a
replicator because it does not interpret instructions that constitute part of itself to create new
copies of itself. That is, although it may constitute a self-description—what von Neumann
referred to as uninterpreted code—it lacks instructions for how to self-replicate—what von
Neumann referred to as interpreted code. A cultural entity (even a symbolically coded one, like a
book) just impinges on minds like a radio signal received by multiple radios, and it is we who do
the interpreting, not the entities themselves. In so doing, we change them however we wish to suit
our current needs and tastes. Thus acquired characteristics can be inherited, and cultural evolution
is therefore sometimes said to be Lamarckian. This is an important reason for the aforementioned
speed with which culture evolves. It is because individual organisms die, and therefore biological
lineages thrive only through reproduction via an interpreted code, that in biology acquired
characteristics are not inherited, and thus not retained.
The view that a cultural entity constitutes a ‘second replicator’ has had the unfortunate
consequence of leading some to assume that only ideas transmitted through imitation participate
in culture. The argument is that imitation is the only form of social exchange that is high fidelity
enough to guard against change or degradation. However, firstly, imitated ideas are as subject to
modification as ideas acquired through any other means. Secondly, by excluding emotions,
individual learning, and information acquired through other forms of social exchange, one is
forced to some implausible conclusions, such as that a haiku inspired by nature cannot enter into
culture, or that the mood of a teacher has no cultural impact on students. Rather than viewing
cultural entities as replicators, one can say that both in a stream of thought and in interactions
amongst individuals, some aspects are retained or preserved, and others lost or varied. It is thus
retention, not self-replication, that lends continuity to the cultural process.
14.1.3.2
Cultural Novelty is Generated not Randomly but Strategically and Contextually
The production of novelty in culture is less random than it is in biology. Cultural novelty is
generated and assimilated strategically and contextually. The form it takes reflects the
accumulated knowledge of individuals, the circumstances they found themselves in, and the
social structure in which they are embedded. Thus new entities have a much greater than chance
probability of being fitter than their predecessors. This was seen in the MAV computer model
where it was possible for variation to be generated either randomly or strategically (using
knowledge-based operators). It was found that the strategic generation of variation both increased
the rate at which fitter ideas evolved, and decreased the time to stabilization.
219
Liane Gabora
The difference between random and strategic generation of novelty can be characterized in
terms of parallel versus heuristic search. A biological organism produces as many as several
billion of gametes. Most don’t survive, but a few do. The occasional one is fitter than average,
and over generations it tends to proliferate. This kind of approach—search the entire space of
possibilities without devoting much effort to any one possibility, and probably at least one of
them will be better than what exists now—is referred to as parallel search.
Human creativity, on the other hand, is generated using knowledge of relationships amongst
the various elements of the problem domain, and taking into account specifics of how the present
situation differs from previously encountered ones. This kind of approach—explore few
possibilities, but choose them wisely and explore them well—is referred to as heuristic search.
Cultural novelty can be generated in a heuristic manner because it is the product of a relationally
structured model of the world, or worldview, that is used to figure out what to do whenever a
situation is too complicated for hardwired instincts.
14.1.3.3
Mental Simulation
Humans use their models of the world not just to strategically solve problems, but also to
imagine, fantasize, or mentally simulate possible scenarios and actions prior to (or sometimes
without ever) carrying them out. The agents in MAV could engage in a simple form of mental
simulation; that is, they could assess the relative fitness of an action before implementing it. This
provided them with a rudimentary type of selection before the phenotypic expression of an idea.
Turning off this capacity reduced the speed at which fitter ideas evolved. This is an indication that
the capacity to mentally simulate events before actualizing them is evolutionarily beneficial.
14.1.3.4
Culture Works with not just Actual but Potential Entities
Mental simulation actualizes a mental entity, not in physical space, but in conceptual space. If
multiple scenarios have been conceptually actualized in this way, and are then considered
simultaneously such that one gets chosen to be thought about further or actualized in the external
world, this kind of cognitive process is relatively amenable to Darwinian description. But a
Darwinian approach is only possible when the relevant entities are actualized (if not in physical
space then at least in conceptual space) and uniquely distinguishable one from another; otherwise
there is no basis for them to be selected amongst. A full description of culture requires us to be
able to describe changes of state undergone by mental entities that are not even actualized in
conceptual space, and that can only be said to be potential (such as the state this thesis was in a
year ago). A Darwinian approach is not appropriate when the state of the mind is unfocused,
220
Chapter 14: Summary and Synthesis of Main Points
contextual, undecided in a nondeterministic way with respect to some issue or situation, or when
we are refining or honing in an idea in a heuristic manner.
14.1.3.5
Tradeoff Between Innovation and Imitation
Humans encounter a tradeoff in that doing one thing interferes with doing another. This was
implemented in MAV as a trade-off for agents between innovating versus imitating a neighbor. It
was found that the higher the ratio of innovation to imitation, the greater the diversity, and the
higher the fitness of the fittest action being implemented by any single agent. Interestingly
however, for the artificial society as a whole, the optimal innovation-to-imitation ratio was
approximately 2:1 (but diversity is then lower), which indicates that a combination of creativity
and imitation is evolutionarily beneficial.
Another result observed in the computer model was that, for the agent with the fittest idea,
the less it imitated (i.e. the more computational effort reserved for innovation) the better its
performance. This suggests that although a combination of creativity and imitation is beneficial
for the society as a whole, the optimal ratio of the two depends upon the relative fitness of the
mental entities of the individual; specifically, the fitter the mental entities, the fewer the benefits
of imitation, and vice versa.
14.1.4 Conclusion Regarding Similarities and Differences
The invention and development of cultural entities constitutes a form of evolution that is
intertwined with, and in numerous ways similar to, biological evolution. Because of these
similarities, the explananda of Darwinian evolution can, to some extent, be put to use to describe
cultural phenomena. However, in other respects cultural evolution is very different from
biological evolution. An observation that can be made here is that the similarities have primarily
to do with inter-individual processes, and the differences with intra-individual processes. It is
these intra-individual processes that have proven most problematic for (and thus avoided by)
Darwinian and neo-Darwinian approaches to culture. So we must ask: what is the source of these
intra-individual processes? This leads us to ask what is the origin of the unique aspects of human
cognition that make it capable of evolving culture. Note that the question of how an evolutionary
process begins—whether biological or cultural—is not answered by Darwinian theory, which
focuses on what happens once the process of evolving is underway.
221
Liane Gabora
14.2 HOW DID CULTURE BEGIN TO EVOLVE?
If culture is an evolutionary process, an important question that must be addressed is what is it
about humans that enabled us to evolve culture in the first place. This section summarizes the key
elements of the tentative answer proposed in this thesis.
14.2.1 Arguments that the Bottleneck was Creativity rather than Imitation
It is widely believed that culture began with the onset of the capacity to imitate. The intuition that
imitation was the bottleneck to culture may stem from a biological mindset; in biology, the advent
of self-replication through reproduction was the critical milestone. However, it does not follow
that cultural evolution began with the onset of imitation, even if imitation plays a somewhat
analogous role, because as we have seen, cultural entities are not replicators. They contain a selfdescription, but not instructions for their self-replication. It is we who replicate them. Biology, a
breath-first strategy, could not get going until it developed an elaborate system for replicating.
Culture, a heuristic strategy, had to develop an elaborate system for generating novelty.
There are several sources of evidence that the source of our uniqueness stems more from our
creativity than from our capacity for imitation. The first comes from findings that animals also
imitate. The second comes from archeological evidence that the origin of culture is associated
with increased variety, which results not from imitation but innovation. (For example, if someone
invents a new method of baking bread, the variety of methods increases. However, if your
method of baking bread works better than mine and I stop using my method and start imitating
you, the variety of methods decreases.) Thirdly, we saw in the MAV computer program that a
society that could imitate but not invent could not evolve culture. However, the converse was not
true; a society that could invent but not imitate could still evolve ideas, albeit with much
reinventing of the wheel. Of course, in the absence of social exchange, we might not want to use
the word ‘culture’ here. The point, though, is that since innovation is sufficient to evolve ideas
without imitation, it follows that a lack of culture (as in animals) is due to the incapacity for
creativity, or at least of the strategic, contextual sort exhibited by humans.
14.2.2 Creativity Arises through Variability of the Activation Threshold
What then is it about the human mind that enabled it to become strategically and contextually
creative? To explain the theory put forth here it must first be said that mental entities consist of
properties, commonly referred to as features or dimensions. From a dynamical perspective, they
can be viewed as hierarchically nested phase relations. Their storage in memory is distributed (but
not fully distributed), meaning they are stored in multiple locations with graded strengths. Their
222
Chapter 14: Summary and Synthesis of Main Points
storage is also content-addressable, meaning that entities stored in overlapping regions are
correlated or share properties. It is hypothesized that the creative generation of cultural novelty
originates in the simultaneous retrieval of items whose distributed storage regions encompass
overlapping memory locations. The lower the activation threshold, the wider the memory region
stored to and searched from at any given instant. Thus, the greater the conceptual fluidity, and the
more likely this kind of simultaneous retrieval is to happen.
However, it is not just a lower activation threshold that facilitates greater creativity, but a
variable one; in other words, the possibility for the threshold to increase or decrease depending on
the nature of the situation. Although some ideas elicit spontaneous creative expression, many
must first be recursively redescribed or refined in a stream of thought. By gradually decreasing
the activation threshold during the creative process one goes from a state of mind that is more
likely to simultaneously evoke items that are correlated (in some sense, but perhaps a sense that
we would never think of) and therefore stored in overlapping memory locations, to a state of mind
that is more conducive to establishing relationships of causation. This could involve going from
what Kauffman [1993] refers to (in the origin of life context) as a supracritical state—which is
able to incorporate new items—to a subcritical state—which does not incorporate new items, but
could be conducive to the stabilizing of the structure already present. Thus one slowly anneals on
a worldview that incorporates aspects of the idea that are useful or relevant while weeding out
irrelevant or misleading aspects. This hypothesis is consistent with research into the attributes of
creative individuals: defocused attention, heightened sensitivity, and flat association hierarchies.
14.2.3 Creativity Requires a Relationally Structured Worldview
Donald [1991] and others give evidence that the human mind underwent at least one major
cognitive transition. Whereas the animal mind stores memories and stimulus-response
associations independently, humans organize them in a way that captures abstract similarities and
relationships. We saw that strategic creativity is made possible through the associative structure
of mental entities in a model of the world, or worldview. This associative structure enables
novelty to be generated using knowledge of relationships amongst the various elements of the
problem domain, and taking into account how the present situation differs from previous ones.
The less stable the environment, the more our ability to make predictions, invent solutions to
problems, and evaluate possible plans of action, hangs on the accuracy of this worldview.
223
Liane Gabora
14.2.4 The Existence of a Worldview Generates a Chicken and Egg Paradox
But now we come to a paradox. How does this reorganization get underway? That is, until
discrete memories and associations have been woven into a conceptual web, how can they
generate a stream of thought? And conversely, until a mind can generate a stream of thought, how
does it weave its memories and associations into a connected worldview? That is, how does a
mind come to assume a stream of thought that progressively shapes, and is shaped by, a highly
structured model of the world?
A stream of thought requires each mental entity that enters awareness to activate one or
more associations to evoke a retrieval or re-interpretation. However, thinking is the process that
puts related mental entities within reach of one another in the first place; it is what recognizes
abstract similarities and restructures the memory to take them into account. How could an entity
composed of complex, mutually dependent parts have come into existence?
14.2.5 Origin of Culture through Conceptual Closure to Yield a Worldview
The approach I have taken to this question is inspired by an approach to the analogous chickenand-egg paradox concerning how biological evolution began. The origin of life presents the
following paradox not unlike the one we just looked at: if living things come into existence when
other living things give birth to them, how did the first living thing arise? How could something
able to reproduce itself come to be? Over the last century, we learned that self-replication is
orchestrated by an intricate network of interactions between proteins and DNA; proteins are made
by decoding DNA, and DNA requires the catalytic action of proteins to get decoded. So the
question became: how could a system composed of complex, mutually-dependent parts come into
existence? Kauffman proposed a solution to this chicken-and-egg problem: life began with the
emergence of a set of autocatalytic polymers. None of the polymers can catalyze its own
replication, but there exists a catalytic pathway to the formation of each polymer in the set.
My research has involved adapting Kauffman’s origin-of-life solution to the origin of the
kind of cognitive architecture capable of sustaining cultural evolution. Initially a genetic mutation
causes the activation threshold to decrease, leading to the more widely distributed storage and
associative retrieval of memories. Given this wider storage, and because the memory is content
addressable, similar memories are stored in overlapping regions of conceptual space, and
sometimes get retrieved simultaneously. This causes the emergence of abstractions—mental
entities derived from multiple experiences or abstract thinking, such as concepts, stories, plans
and attitudes. Abstractions make the memory denser, which thus increases the frequency of
reminding events. Reminding events themselves begin to evoke reminding events recursively,
224
Chapter 14: Summary and Synthesis of Main Points
thus generating streams of associative thought, which increase in both duration and frequency. In
the course of these streams of thought yet more abstractions emerge, which themselves become
connected in conceptual space through higher-level abstractions. Eventually the memories, as
well as instinctive stimulus-response associations, become connected through hierarchical layers
of abstractions into an autocatalytically closed relationally structured conceptual network, or
worldview. In sum then, the basic idea is that culture began with the formation of an
autocatalytically closed conceptual web, or worldview, in one or more interacting individuals.
14.2.6 Conceptual Closure Recurs in the Mind of Every Encultured Child
I propose that this transformation recurs in the mind of every young child in order for the child to
meaningfully participate in culture. Some speculative factors that may have an impact on this
process are discussed. Although conceptual closure may initially be disrupted by creativity, the
long-term effect of a tendency toward creativity is the potential for a more finely structured
worldview. Some experiences are so consistent, or so inconsistent, with a worldview, that they
have little impact on it. Others percolate deep into the worldview, renewing our understanding of
a myriad other concepts or events. Some processes interfere with the cohesiveness or integration
of the worldview. Censorship arrests the process by which dangerous thoughts infiltrate the
conceptual network, repression blocks dangerous thoughts that have already been assimilated, and
deception separates you from others in conceptual space. These processes may evoke feelings of
fragmentation at the level of the individual or the society. Other processes, which include
cognitive analogs to self-organized criticality and simulated annealing, cause the worldview to be
more deeply penetrated with abstractions that enable finer levels of distinction.
14.2.7 Worldviews are Primitive Replicators
Recall that mental entities do not constitute replicators because they consist of uninterpreted
information—a self description—but not interpreted information—instructions for how to self
replicate. The fascinating implication that falls out of the theory that worldviews emerge through
an autocatalytic process of conceptual closure is that although mental entities do not constitute
replicators, interconnected networks of them—worldviews—do constitute replicators, but of a
primitive, clumsy sort, like the autocatalytic sets of molecules prior to the genetic code.
Worldviews replicate, not all at once, but piece by piece, particularly when we pass on
knowledge, actions, and artifacts to children who have not yet some idea of how the various
aspects of their world fit together and relate to one another.
225
Liane Gabora
14.2.8 Conceptual Networks Contain Quantum-like Structure
One of the motivations for approaching the evolution of worldviews through culture as described
below, using generalizations of the mathematics originally developed for quantum mechanics, is
that we were able to show that Bell inequalities are violated in the relationship between an
abstract concept (‘cat’), and specific instances of this concept (the two cats that live at our
research center). Thus, in at least this cognitive situation, a concept can correctly be viewed as a
superposition state of its instances, and therefore quantum structure exists in the mind. This
tentatively suggests that the process through which a child develops an autocatalytically closed
conceptual network is, in fact, a process of modifying the architecture of memory such that it
offers the potential for superposition states to be entered in the course of a stream of thought. The
exact nature of the relationship between closure and quantum structure is still unknown, but it is
under investigation at CLEA and FUND. Already, however, something can be said. A big deal
has been made of the fact that distributions are graded, that the degree of gradation depends upon
the activation threshold, that this in turn affects the degree of conceptual fluidity, and that by
varying the activation threshold depending on the density of concepts in the currently active
region of conceptual space, one can remain at the edge of chaos. It seems clear that such an
arrangement is what gives the cognitive architecture an intermediate degree of contextuality, and
thereby a structure that is midway between classical and quantum.
14.3 TOWARD A GENERAL, TRANS-DISCIPLINARY THEORY OF EVOLUTION
In considering the process through which potential mental entities become actualized, and the role
of context in this process, it became clear that not all processes acting on mental entities are
Darwinian. It is also clear that Darwinism could not explain the sort of change of state observed
in the micro-world. Thus the need for a general, trans-disciplinary theory of evolution. This
section outlines some steps toward the development of such a theory.
14.3.1 The Inadequacy of Neo-Darwinian Theory
Having looked in depth at the origin of life and the origin of an interconnected worldview, we
come face to face with a severe limitation of neo-Darwinism: it does not provide an explanation
for this most fundamental evolutionary issue, that of origins. Moreover, it cannot cope with the
issue of how an evolutionary system originates because it ignores potential; it focuses on natural
selection, a process that works on variation that has already been actualized. But even in
biological evolution, actualized variation is not the only phenomenon that begs for explanation.
When hereditary information undergoes descent with modification, the most formative states it
226
Chapter 14: Summary and Synthesis of Main Points
enters (and the ones we perhaps feel intuitively to be the most interesting) are those that can only
be described as potential with respect to a certain context: those where it has the possibility of
sudden death, or where it is about to mutate, or recombine through sexual reproduction. As
scientists, our tendency is to ignore these states of potentiality, and begin the descriptive process
when things become concrete; when a state that was potential becomes actualized. When we
really need to describe a state that could actualize in different ways to different new states, we use
a stochastic model, and assign a probability to the different possible ways of actualizing. We then
imagine that the real state is one of these new states; we just don’t know which one. But this is
not correct if the probabilities arise from the interaction with the context. In biology the
limitations of this approach are not overwhelmingly evident. Potentiality and contextuality are
important aspects of biological evolution, but it has been possible to get away without paying
serious attention to them because their impact is largely not heritable. In culture, however, they
are so omnipresent and exert such a large impact that they are in fact crucial to arrive at a
scientific basis for viewing culture as an evolutionary process. Contextual effects can restructure
the worldview(s) of an individual or entire society, and this restructuring can have an immediate
impact on the evolution of ideas, behaviors, and artifacts.
Thus, to explain how culture evolves, Darwinian theory is insufficient. How then does
culture evolve? To answer this question we need a stronger theory of evolution.
14.3.2 Context, Potential, Actualization, and Lack of Knowledge
Let us look at exactly what is going on when an entity—whether it be biological, cultural, or of
another sort—evolves. This entity undergoes many changes of state. When it changes, it
actualizes one or another of the forms that has become potential for it because of its history of
previous actualizations. Each actualization, in turn, provides constraints and opportunities for its
further evolution. This is also what happens whether the entity is a species, or an idea being
refined in the course of a stream of thought, or an invention being refined by an entire society of
individuals interacting over the course of many generations. In all these cases, context plays a role
in how potential is actualized. Self-organizing, emergent processes (such as the proposals for the
origin of life and origin of culture described above) which the neo-Darwinian picture also does
not account for, are also ways of actualizing potential. Thus, while replication with variation and
selection of particulate traits has served as an adequate theory of evolution for some time, for a
complete and accurate theory we require a formalism that can describe this kind of
nondeterministic potentiality and contextuality.
227
Liane Gabora
In general, we do not have complete knowledge of the state of the entity, the context, and
how they interact. Thus when we believe one situation to be identical to another—i.e. same entity
in the same state under the same context—this is often not really the case. The states and contexts
may be different, though we have identified them as being the same because of our lack of
knowledge of the situation. This gives rise in a natural way to nondeterministic change.
There are different possible reasons for a lack of knowledge concerning a change of state,
each of which gives rise to a different kind of nondeterminism. If the lack of knowledge concerns
the state of the entity the probability model that describes this lack of knowledge is
Kolmogorovian. But if the lack of knowledge concerns the context, the probability model that
describes this lack of knowledge is non-Kolmogorovian. Since this is a crucial point, is is briefly
elaborates upon.
14.3.3 Describing Change of State when One Lacks Knowledge about the State
The general theory of classical stochastic processes (e.g. Markov processes) claims to provide a
formalism for the description of nondeterministic change of states. It can be shown, however, that
classical stochastic processes can only describe nondeterministic change that arises through a lack
of knowledge of the state of the entity, not through lack of knowledge of the contextual
interaction between this state and the context (or a lack of knowledge concerning the context
itself). The reason is that lack of knowledge of the interaction between state and context (or the
context itself) introduces a non-Kolmogorovian probability model on the state space, while
classical stochastic processes work within a framework where the state space (in this case, the set
of mixed states, which are probability measures on some outcome set) is described within a
Kolmogorovian probability structure.
Moreover, because classical mechanics describes the formation of joint entities only using
product states, it cannot describe the spontaneous appearance of new states with new properties.
Chaos and complexity theory provide a means of describing emergent new states, and entities that
started out in similar initial states can end up in widely different final states, but the effect of
context on change of state is always deterministic.
14.3.4 Describing Contextuality (Lack of Knowledge about State-Context Interaction)
Quantum mechanics does provide a means of describing states of potentiality, and changes of
state that are nondeterministic, where this indeterminism finds its origin in a lack of knowledge
about how the context interacts with the state of the entity. It can also describe the appearance of
new states with new properties. Thus if one wanted to use the stochastic processes approach to
228
Chapter 14: Summary and Synthesis of Main Points
develop a theory of evolution that is general enough to cover contextual change of state, one
might be tempted to use quantum stochastic processes. If, however, for these quantum stochastic
processes, one takes recourse to the formalisms of pure quantum mechanics (measures on a
Hilbert space to generate the state space of mixed states) rather than the generalized quantum
formalisms suggested here, one is confronted with several limitations; including the linearity of
the Hilbert space, and the fact that one can only describe the extreme case where potentiality is
always present and change of state is maximally contextual.
14.3.5 Intermediate Contextuality and the Generalized Quantum Formalisms
The generalized quantum formalisms overcome these limitations by using a state-property system
instead of a Hilbert space. The original motivation for the development of generalizations of the
formalisms of quantum mechanics was in fact theoretical (as opposed to the need to describe the
reality revealed by experiments). However, it turns out to be able to describe situations of
nonlinearity, involving entities that vary in degree of potentiality and contextuality, as well as the
appearance of states with new properties. This is why they are re-applied here to begin to
construct a generalized theory of evolution.
14.3.6 Evolution as Context-driven Actualization of Potential (CAP)
The general theory of evolution proposed here grew out of viewing evolution as incremental
adaptation in response to environmental constraints and affordances. The concept of
‘environment’ was further generalized to context, and the concept ‘adaptation’ to context-driven
actualization of potential. The process of actualizing potential feeds back on the potentiality state
of an entity, thus affecting the way it is next influenced by context, and so forth recursively.
Novelty generates new situations that provoke or inspire more novelty. Thus, evolution is more
broadly construed as the incremental change that results from the recursive, context-driven
actualization of potential. The description of the state of an evolving entity is treated as a
potentiality state, and the contextual interaction with the environment plays the role of the
measurement by causing the entity to collapse to an actualization state of this interaction.
With this general approach to evolution we are able to unite quantum, classical, biological,
conceptual, and cultural evolution under one roof. These various kinds of evolution are different
ways in which potential that is present due to the state of an entity, its context, and the nature of
their interaction, is recursively actualized. They differ with respect to amount of indeterminism,
sensitivity to context, whether the evolution is bringing about context dependence or context
229
Liane Gabora
independence, and the degree to which the effects of changes of state are retained in the further
evolution of the entity and/or its lineages.
14.3.6.1
Nondeterministic Collapse of a Quantum Entity
A quantum entity exists in a superposition state, or entangled state, and a measurement causes it
to collapse to an eigenstate of that measurement. The specifics of the measurement provide the
context that elicit in the entity one of the states that were previously potential for it. The evolution
of such an entity cannot be examined without performing measurements—that is, introducing
contexts—but due to extreme sensitivity to context, the contexts themselves unavoidably affect its
evolution. Thus the evolution of a quantum particle is a paradigmatic example of CAP because its
state at any point in time reflects both its history of previous states, and the contexts to which it
has been exposed.
14.3.6.2
Deterministic Evolution of Quantum or Classical Entity
We saw that the deterministic change of state of an entity can likewise be viewed as an
‘experiment’ of sorts for which there is only one potential outcome; there is only one way the
context can cause the entity to collapse. Thus the evolution of the entity appears to be
deterministic. So the dynamical evolution of such an entity as per the Schrödinger equation
reduces to a a deterministic collapse. It too is an example of the actualization of potential because
its state at any point in time reflects its history of previous states. The same line of reasoning also
holds for classical entities. The apparent deterministic influence of the context is really just a
reflection of the fact that we have sufficient knowledge about the context to make acceptably
accurate predictions.
14.3.6.3
Biological Evolution
Biological evolution can be viewed as the situation where an entity (the original living organism)
splits into different variants of itself, giving rise to different lineages, each of which is exposed to
a different series of contexts. Most of the changes of state a biological entity undergoes are
relatively deterministic. However, when the interaction between its internal state and its context is
such that the outcome could be either death, or the generation of offspring, it enters a situation
more like the context-driven collapse of a quantum entity. In the life-or-death situation, the
potential for life and the potential for death constitute a potentiality state, and through interaction
with the context, the entity collapses to either one or the other of them. In the mating situation, the
different offspring (different variants of itself) it could have with this mate constitute a
potentiality state, and it collapses to one or more of them. Thus the mate constitutes the context
230
Chapter 14: Summary and Synthesis of Main Points
for which the state of the biological entity is a potentiality state. Speciation can be viewed as the
situation wherein one kind of variant no longer has the potential to create a context for the other
for which its state is a potentiality state with respect to offspring. Natural selection is now viewed
as the effect of the physical and social environment on the collapses involved in life-or-death and
mating events. A species or lineage is adapted to its environment to the extent that its previous
states have been potentiality states that could collapse to different possible outcomes, and thus its
history of collapses has reflected the specific contexts to which it has been exposed.
14.3.6.4
A Stream of Thought
A stream of thought is similarly composed of both (1) relatively deterministic segments, where
the mental entity under consideration changes state in a way that only deterministically reflect the
context to which it is exposed. and (2) collapse events, where the change of state of the mental
entity nondeterministically reflects the stimulus or situation to which it was exposed. Interacting
individuals provide the context that evokes collapse events in one another. As in biology, the
opportunity for this may be higher when the individuals involved have the potential to either end
each other’s stream of thought (a life-or-death situation for the idea under reflection), or develop
it further (generate progeny for the idea under reflection).
A reasonable first step toward an understanding of how culture evolves is to understand how
novel ideas arise in a stream of thought. To this end, it is vital to be able to describe those
transitions in the state of the mind wherein features or dimensions that were not under
consideration just previously suddenly come under consideration; their relevance to the situation
at hand becomes apparent. Unlike other approaches to cognition, which are classical theories in
the sense that they fall out of classical mechanics, the CAP approach can be used to model such
changes of state as occur in the creative process because the mathematical formalism can account
for the appearance of new states with new properties. Even simple sorts of creativity, and not just
its generation but even its interpretation, cause difficulties for other approaches. For example,
they do not explain how people interpret creative conjunctions. They cannot account for the socalled guppy effect where ‘guppy’ is not viewed as a good example of ‘pet’, nor a good example
of ‘fish’ but it is viewed as a good example of ‘pet fish’ [Osherson & Smith 1981]. The CAP
approach handles this by incorporating contextual effects, as follows. Activation of ‘pet’ rarely
causes activation of ‘guppy’. Likewise, activation of ‘fish’ rarely causes activation of ‘guppy’.
But ‘pet fish’ causes activation of the potentiality states ‘pet’ in the context of ‘pet fish’ AND
‘fish’ in the context of ‘pet fish’. Since for both, the probability of collapsing on to the state
‘guppy’ is high, it is very likely to be activated. Thus we have a formalism that is not stumped by
231
Liane Gabora
a situation wherein an entity that is neither a good instance of A, nor B, is nevertheless a good
instance of A AND B. We also saw more concretely how the CAP approach can be used to
describe instances of what has been referred to as ‘impossibilist’ creativity, through an example
involving the cave people Og and Oga, and the invention of the torch.
14.3.6.5
Cultural Evolution
One could refer to a stream of thought as another form of evolution, which we might call
cognitive evolution, except that were it not for culture the process would terminate; that is, the
evolving entities would cease to exist when the individual thinking them died. However, these
entities also evolve when minds interact, either directly, or indirectly via artifacts. An idea may
not be able to evolve further in the mind of a particular individual—that is, it no longer
constitutes a potentiality state with respect to the contexts that individual encounters or is able to
mentally simulate. However, if it is worked out at least to the point where it can be expressed, it
may constitute a potentiality state in the mind of someone else, who can evolve it further. And so
forth. If the theory developed here is correct, it is through this process that ideas and actions play
off one another in real life cultural exchange.
Like biological entities, cultural entities are also able to split into multiple variants of
themselves which evolve as separate lineages. These variants can range from virtually identical to
one another to virtually impossible to trace back to the same ‘parent’ idea. They affect, and are
affected by, the minds that encounter them. For example, books can affect all the individuals who
read them, CDs can affect all the individuals who play them, movies can affect all the individuals
who watch them, and so forth, and these individuals subsequently provide new contexts for the
possible further evolution of the ideas they described and stories they told.
14.4 CONCLUSION
If we are to take seriously the idea that culture is an evolutionary process, we can look to
evolution to provide the kind of overarching framework for the humanities that it provides for the
biological sciences. Many aspects of culture are amenable to Darwinian description, particularly
those that involve artifacts, or interactions amongst individuals (such as the competition of
artifacts in the marketplace). Others, such as the origin of culture (and even the origin of
biological life) are more aptly described using concepts from complexity such as self organization
and emergence. Still other aspects of culture, particularly those that involve the generation and
refinement of novel ideas within individuals, require for their full description a formalism for
dealing with potentiality and nondeterminism. This is because ideas can be modified at any point
232
Chapter 14: Summary and Synthesis of Main Points
along their evolution according to our needs, tastes, and desires, and importantly, these
modifications are inherited. The reason for this stems from the fact that mental and cultural
entities sometimes constitute a self-description, but never instructions for their self-replication. It
is we who replicate them, strategically and contextually, using knowledge and intuition.
Nevertheless, although mental entities themselves are not replicators, interconnected
networks of them, such as in a mental model of the world, or worldview, do constitute a
replicator. I propose that this is because they undergo a process of conceptual closure analogous to
the self organization of catalytic molecules into an autocatalytically closed set. Such a structure
can replicate, but only clumsily; its self replication is the emergent result of local interactions
rather than through a symbolic code. It is in this primitive sense that worldviews replicate, little by
little, like organisms did prior to the genome, as elders pass on procedural and semantic
knowledge, directly or mediated through artifacts.
Because of the omnipresence of nondeterminism, potentiality, and contextuality in culture,
neither the Darwinian theory of evolution through replication, variation, and natural selection, nor
complexity theory, is sufficient to fully describe culture. We require a more general theory of
evolution, wherein Darwinian evolution is one manifestation of this broader framework. In the
theory proposed here—which draws upon the approach to nondeterminism, potentiality, and
contextuality offered by the generalized quantum formalism—biological, cultural, quantum, and
classical evolution are all described as different instantiations of the recursive, context-driven
actualization of potential. Thus we reach a more general understanding of how it is that
something could evolve. In sum, I view cultural evolution as consisting of two interwoven parts,
one more like biological evolution, dealing with entities in states that are external and actualized,
and the other more similar to the evolution of quantum entities in the micro-world, dealing with
entities in states that are internal and potential.
233
Liane Gabora
the
234
REFERENCES
Abrahamson, E. (1991) Managerial Fads and Fashions: The Diffusion and Rejection of Innovations.
Academy of Management Review, 16 (3), July, p. 586-612.
Abrahamson, E. & Rosenkopf, L. (1993) Institutional and competitive bandwagons: Using mathematical
modeling as a tool to explore innovation diffusion. Academy of Management Review, 18 (3), July, p. 487517.
Abrahamson, E. & Rosenkopf, L. (1997) Social Network Effects on the Extent of Innovation Diffusion: A
Computer Simulation. Organization Science, 8 (3), May, p. 289-309.
Aerts, D. (1981) The One and the Many: Towards a Unification of the Quantum and the Classical
Description of One and Many Physical Entities. Doctoral Thesis, Brussels Free University.
Aerts, D.(1982a) Description of many physical entities without the paradoxes encountered in quantum
mechanics. Foundations of Physics, 12, p. 1131-1170.
Aerts, D. (1982b) Example of a macroscopical situation that violates Bell inequalities. Lettere al Nuovo
Cimento, 34 (107).
Aerts, D. (1983) Classical theories and non classical theories as a special case of a more general theory.
Journal of Mathematical Physics, 24, p. 2441-2454.
Aerts, D. (1985a) The physical origin of the EPR paradox and how to violate Bell inequalities by
macroscopical systems. In (Lathi, P. & Mittelstaedt, P., eds.) On the Foundations of Modern Physics, 305,
World Scientific, Singapore.
Aerts, D. (1985b) A possible explanation for the probabilities of quantum mechanics and a macroscopical
situation that violates Bell inequalities. In (P. Mittelstaedt & E. W. Stachow, eds.) Recent Developments in
Quantum Logic, Grundlagen der Exacten Naturwissenschaften, 6, Wissenschaftverlag, Bibliographisches
Institut 235, Mannheim, Germany.
235
Liane Gabora
Aerts, D. (1986) A possible explanation for the probabilities of quantum mechanics. Journal of
Mathematical Physics, 27, p. 202-210.
Aerts, D. (1991) A mechanistic classical laboratory situation violating the Bell inequalities with √2, exactly
‘in the same way’ as its violations by the EPR experiments. Helvetica Physica Acta, 64, p. 1-24.
Aerts, D. (1993) Quantum structures due to fluctuations of the measurement situations. International
Journal of Theoretical Physics, 32, p. 2207-2220.
Aerts, D., (1998) The hidden measurement formalism: what can be explained and where paradoxes remain?
International Journal of Theoretical Physics, 37 (291).
Aerts, D. (1999) The Stuff the World is Made Of: Quantum Mechanics and the Nature of Reality. In (Aerts,
D., Broekaert, J., & Mathijs, E., eds.) Einstein Meets Magritte: An Interdisciplinary Reflection, Kluwer
Academic, Dordrecht, The Netherlands.
Aerts, D., Aerts, S., Broekaert, J., & Gabora, L. (2000a) The violation of Bell inequalities in the
macroworld. Foundations of Physics, 30 (9), 1387-1414. [quant-ph/0007044]
Aerts, D., Broekaert, J. & Gabora, L. (2000c) Intrinsic contextuality as the crux of consciousness. In (K.
Yasue, ed.) Proceedings of Fundamental Approaches to Consciousness, Tokyo '99, John Benjamins
Publishing Company, Amsterdam, The Netherlands.
Aerts, D., D’Hooghe, B., & Gabora, L., (2000) Emergence and states of potentiality. In (F. Gorzelak, ed.)
Proceedings of the Fourth International Conference on Emergence, ECHO IV, July 31-August 3, 2000,
University of Southern Denmark, Odense, Denmark.
Aerts, D., D’Hondt, E. & Gabora, L., (2000b) Why the disjunction in quantum logic is not classical.
Foundations of Physics, 30 (9), p. 1473-1480. [quant-ph/ 0007041].
Aerts, D. & Durt, T. (1994a) Quantum, classical and intermediate: A measurement model. Proceedings of
the International Symposium on the Foundations of Modern Physics 1994, Helsinki, Finland, eds. C.
Montonen et al., Editions Frontieres, Gives Sur Yvettes, France.
Aerts, D. & Durt, T. (1994b) Quantum, classical, and intermediate, an illustrative example. Foundations of
Physics 24, p. 1353-1368.
236
References
Aerts, D. & Gabora, L. (1999) Quantum Mind: Quantum approaches to understanding the mind. Lecture
Ten, http://www.consciousness.arizona.edu/Millennium/
Aerts, D. & Gabora, L. (2001) Towards a general theory of evolution. To appear in Foundations of Science.
Akins, C. K. & Zentall, T. R. (1998) Imitation in the Japanese quail: The role of reinforcement of
demonstrator responding. Psychonomic Bulletin and Review, 5 (694).
Alexander, R. D. (1980) Darwinism and Human Affairs. Pitman, London.
Alexander, R. D. (1975) The search for a general theory of behavior. Behavioral Sciences, 20, p. 77-100.
Alexander, R. D. (1979) Darwinism and Human Affairs. University of Washington Press, Seattle WA.
Aspect, A., Grangier, P. and Roger, G. (1981) Experimental tests of realistic local theories via Bell’s
theorem. Physical Review Letters, 47, p. 460.
Aspect, A., Grangier, P. and Roger, G. (1982) Experimental realization of Einstein-Podolsky-Rosen-Bohm
gedankenexperiment: a new violation of Bell’s inequalities. Physical Review Letters, 48, p. 91.
Axelrod, R. (1985) Modeling the Evolution of Norms. Paper presented at the American Political Science
Association Meeting, New Orleans, LA.
Axelrod, R. (1997) The Complexity of Cooperation: Agent-based Models of Competition and
Cooperation,.Princeton University Press, Princeton.
Baars, B. (1988) A Cognitive theory of Consciousness. Cambridge University Press, Cambridge.
Bak, P., Tang, C., & Weisenfeld, K. (1988) Self-organized criticality. Physical Review A, 38 (364).
Bandura, A. & Walters, R. (1963) Social Learning and Personality Development. Holt, Rinehart &
Winston, New York.
Barkow, J. H., Cosmides, L. & Tooby, J. (1992) The Adapted Mind: Evolutionary Psychology and the
Generation of Culture. Oxford University Press, Cambridge.
Barron, F. (1963) Creativity and Psychological Health. Van Nostrand, New York.
237
Liane Gabora
Barsalou, L. W. (1999) Perceptual symbol systems. Behavioral and Brain Sciences, 22 (4).
Bartlett, F. C. (1932) Remembering: A Study in Experimental and Social Psychology. Columbia University
Press, New York.
Bartlett, F. C. (1958) Thinking: An Experiment and Social Study. Basic Books, New York.
Bateson, G. (1972) Steps to an Ecology of Mind. Ballantine Books, New York.
Beck, A. T., Freeman, A. & Associates. (1990) Cognitive Therapy of Personality Disorders. Guilford, New
York.
Beck, B. B. (1976) Tool use by captive pigtailed monkeys. Primates, 17, p.301-310.
Begun, D. J. & Aquadro, C. F. (1992) Levels of naturally occurring DNA polymorphism correlate with
recombination rates in D. melanogaster. Nature, 356, p. 519-520.
Bell, J. S. (1964) On the Einstein Podolsky Rosen paradox. Physics, 1, p. 195.
Best, M. (1997) Models for interacting populations of memes. Journal of Memetics, 1 (2), December issue.
Best, M. (1998) Memes on memes—A critique of memetic models. Journal of Memetics, 2 (1).
Best, M. (1999) How culture can guide evolution: An inquiry into gene/meme enhancement and opposition.
Journal of Adaptive Behavior, 7 (3/4), p. 289-306.
Bhargava, S. C., Kumar, A. & Mukherjee, A. (1993) A stochastic cellular automata model of innovation
diffusion. Technological Forecasting and Social Change, 44, p. 87-97.
Bickerton, D. (1990) Language and Species. University of Chicago Press, Chicago.
Bickhard, M. H. & Campbell, D. T. (1996) Variations in variation and selection: The ubiquity of the
variation-and-selective-retention ratchet. Submitted.
Blackmore, S. (1998). Imitation and the definition of a meme. Journal of Memetics: Evolutionary Models
of Information Transmission, 2.
238
References
http://www.cpm.mmu.ac.uk/jom-emit/1998/vol2/blackmore_s.html
Blackmore, S. (1999) The Meme Machine. Oxford University Press, Oxford.
Boden, M. A. (1991) The Creative Mind: Myths and Mechanisms. Cambridge University Press, Cambridge
MA.
Boesch, C. and Tomasello, M. Chimpanzee and Human Cultures. Current Anthropology, 39 (5), December
issue.
Bohm, D. (1951) Quantum Theory. Prentice-Hall, Englewood Cliffs, NY.
Bonner, J. T. (1980) The Evolution of Culture in Animals. Princeton University Press, Princeton NJ.
Bowers, K. S. & Keeling, K. R. (1971) Heart-rate variability in creative functioning. Psychological
Reports, 29, p. 160-162.
Bowers, K. S., Regehr, G., Balthazard, C. & Parker, K. (1990) Intuition in the context of discovery.
Cognitive Psychology, 22, p. 72-110.
Boyd, R. & Richerson, P. J. (1985) Culture and the Evolutionary Process. The University of Chicago
Press, Chicago IL.
Braitenberg, V. (1984) Vehicles. MIT Press, Cambridge MA.
Brodie, R. (1996) Virus of the Mind. Integral Press, Seattle WA.
Brooks, R. A. (1991) Intelligence without representation. Artificial Intelligence, 47, p. 139-159.
Brooks, V. (1986) The Neural Basis of Motor Control. Oxford University Press, Oxford.
Brown, L. A. (1981) Innovation Diffusion: A New Perspective. Methuen Press, London.
Burnet, F. M. (1959) The Clonal Selection Theory of Acquired Immunity. Cambridge University Press,
London.
239
Liane Gabora
Burt, R. S. (1987) Social Contagion and Innovation: Cohesion versus Structural Equivalence. American
Journal of Sociology, 92, p. 1287-1335.
Byrne, R. W. & Russon, A. (1998) Learning by imitation: A hierarchical approach. Behavioral and Brain
Sciences, 21, 667-721. http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.byrne.html
Calvin, W. (1996) The Cerebral Code. MIT Press, Cambridge MA.
Campbell D. T. (1965) Variation and selective retention in socio_cultural evolution. In Social Change in
Developing Areas: A Reinterpretation of Evolutionary Theory, eds. H. R. Barringer, G. I. Blanksten, & R.
W. Mack, Schenkman, Cambridge, MA.
Campbell, D. T. (1974) Evolutionary Epistemology. In (P.A. Schlipp, ed.) The Library of Living
Philosophers, Vol. XIV: The Philosophy of Karl Popper. Open Court, LaSalle IL.
Campbell, D. (1987) Evolutionary epistomology. In (Radnitzky, G. &. Bartley III, W. W., eds.)
Evolutionary Epistomology, Rationality, and the Sociology of Knowledge. Open Court, LaSalle, IL.
Cariani, P. (1995) As if time really mattered: temporal strategies for neural coding of sensory information.
In (K. Pribram, ed.) Origins: Brain and Self-Organization, Erlbaum, Hillsdale, NJ, p. 161-229.
Cariani, P. (1997) Temporal coding of sensory information. In (J. M. Bower, ed.) Computational
Neuroscience: Trends in Research 1997, Plenum, New York, p. 591-598.
Cavalli-Sforza, L. L. & Feldman, M. W. (1981) Cultural Transmission and Evolution: A Quantitative
Approach. Princeton University Press, Princeton NJ.
Cemin, S. C. & Smolin, L. (1997) Coevolution of membranes and channels: A possible step in the origin of
life. Journal of Theoretical Biology (October issue) [adap-org/9709004].
Chalmers, D. (1996) The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press,
Oxford UK.
Cheney, D. L. & Seyfarth, R. M. (1990) How Monkeys See the World: Inside the Mind of Another Species.
University of Chicago Press, Chicago.
Chomsky, N. (1957) Syntactic Structures. Mouton, The Hague, The Netherlands.
240
References
Clark, A. & Thornton, C. (1996) Trading spaces: Computation, representation, and the limits of uninformed
learning. Brain and Behavioral Sciences, 20, p. 57-90.
Clauser, J. F. (1976) Experimental investigation of a polarization correlation anomaly. Physical Review
Letters, 36, p. 1223-1236.
Cohen, A. B. (1994) Training synaptic delays in a recurrent neural network. M.S. Thesis, Tel-Aviv
University, Tel-Aviv, Israel. ftp://archive.cis.ohio-state.edu/pub/neuroprose/Thesis/cohen.thesis.ps.Z
Combs, A. (1996) The Radiance of Being: Complexity, Chaos and the Evolution of Consciousness. Paragon
House, St. Paul, MN.
Combs, A., Germine, M, & Goertzel, B., Eds. (in press) Mind in Time: The Dynamics of Thought, Reality,
and Consciousness. Hampton Press, Cresskill NJ.
Conte, R. & Paolucci, M. (in press) Cognitive properties and memetic processes: The case of reputation.
Corballis, M. C. (1991) The Lopsided Ape: Evolution of the Generative Mind. Cambridge University Press,
Cambridge MA.
Cozzio-Bu, R. A. (1995) The design of neural networks using a priori knowledge. Ph.D. Thesis, Swiss
Federal Institute of Technology, Zurich, Switzerland.
ftp://archive.cis.ohio-state.edu/pub/neuroprose/Thesis/cozzio.thesis.ps.Z
Crutchfield, J. P. (2001) When evolution is revolution: Origins of innovation. In (Crutchfield, J. P. &
Schuster, P., eds.) Evolutionary Dynamics: Exploring the Interplay of Accident, Selection, Neutrality, and
Function. Oxford University Press, Oxford.
Crutchfield, J. P. & Schuster, P., eds. (2001) Evolutionary Dynamics: Exploring the Interplay of Accident,
Selection, Neutrality, and Function. Oxford University Press, Oxford.
Czikszentmihalyi, M. (1990) Flow: The Psychology of Optimal Experience. Harper & Row, New York.
Csikzentmihalyi, M. (1993) The Evolving Self: A Psychology for the Third Millenium. Harper Collins, New
York NY.
241
Liane Gabora
Csikzentmihalyi, M. (1999) Implications of a systems perspective for the study of creativity. In (Sternberg,
R. J., ed.) Handbook of Creativity, Cambridge University Press, Cambridge, p. 313-335.
Cziko, G. (1997) Without Miracles: Universal Selection Theory and the Second Darwinian Revolution.
MIT Press, Cambridge MA.
Darwin, C. (1871) The Descent of Man. John Murray Publications, London.
Dawkins, R. (1976) The Selfish Gene. Oxford University Press, Oxford UK.
Dawkins, R. (1982) The Extended Pphenotype: The Gene as the Unit of Selection. W. H. Freeman and
Company.
De Duve, C. (1995) Vital Dust: Life as a Cosmic Imperative. Basic Books, New York NY.
Dennett, D. C. (1995) Darwin's Dangerous Idea. Simon & Schuster, New York.
Dewing, K. & Battye, G. (1971) Attentional deployment and non-verbal fluency. Journal of Personality
and Social Psychology, 17, p. 214-218.
D’Hooghe, B. (2000) A study of the effect of varying fluctuations in the measurement context
and state transitions due to experiments, Doctoral Dissertation Brussels Free University.
Donald, M. (1991) Origins of the Modern Mind. Cambridge: Harvard University Press, Cambridge UK.
Donald, M. (1993a) Precis of Origins of the modern mind: Three stages in the evolution of culture and
cognition. Behavioral and Brain Sciences 16, p. 737-791.
Donald, M. (1993b) Human cognitive evolution: What we were, what we are becoming. Social Research,
60 (1), p. 143-170.
Dugatkin, L. A. (2001) Imitation Factor: Imitation in Animals and the Origin of Human Culture. Free
Press, New York.
Dunbar, K. & Schunn, C. D. (1990) The temporal nature of scientific discovery: The roles of priming and
analogy. In: Proceedings of the Twelfth Annual Meeting of the Cognitive Science Society, p. 93-100.
242
References
Durham, W. H. (1991) Coevolution: Genes, Culture, and Human Diversity. Stanford University Press,
Stanford CA.
Durt, T. (1996) From Quantum to Classical: A Toy Model, Doctoral Dissertation, Free University of
Brussels.
Dykes, M. & McGhie, A. (1976) A comparative study of attentional strategies in schizophrenics and highly
creative normal subjects. British Journal of Psychiatry, 128, p. 50-56.
Edelman, G. M. (1987) Neural Darwinism: The Theory of Neuronal Group Selection. Basic Books, New
York.
Edelman, G. M. (1992) Bright Air, Brilliant Fire. Basic Books, New York.
Edmonds, B. (1999) Joining the Dots: Extending the Autocatalytic Picture to a Credible Evolutionary
Process, Commentary on Gabora on Origin-Culture. Psycoloquy, 10 (7).
http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?10.007
Eigen. M. & Schuster, P. (1979) The Hypercycle: A Principle of Natural Self-organization. Springer.
Einstein, A., Podolsky, B. & Rosen, N. (1935) Can quantum mechanical description of physical reality be
considered complete? Physical Review, 47, p. 777.
Elman, J. (1996) Rethinking Innateness: A Connectionist Perspective on Development. MIT Press,
Cambridge MA.
Emch, G. G. (1984) Mathematical and Conceptual Foundations of 20th Century Physics, North-Holland,
Amsterdam.
Evans, S. H. (1967). A brief statement of schema theory. Psychonomic Science, 8, p. 87-88.
Evans, S. H., Hoffman, A. A., Arnoult, M. D., & Zinser, O. (1968). Pattern enhancement with schematic
operators. Behavioral Science, 13, p. 402-404.
Eysenck, H. J. (1995) Genius: The Natural History of Creativity. Cambridge University Press, Cambridge
MA.
243
Liane Gabora
Faraci, G. Gutkowski, S., Notarrigo, S., and Pennisi, A. R. (1974) An experimental test of the EPR
paradox. Lettere al Nuovo Cimento, 9, p. 607-611.
Farmer, J. D., Kauffman, S. A., & Packard, N. H. (1987) Autocatalytic replication of polymers. Physica D,
22 (50).
Fehrer, E. & Raab D. (1962) Reaction time to stimuli masked by metacontrast. Journal of Experimental
Psychology, 63, p. 143-147.
Feist, G. J. (1999) The influence of personality on artistic and scientific creativity. In (Sternberg, R. J., Ed.)
Handbook of Creativity, Cambridge University Press, Cambridge, p. 273-296.
Fodor, E. M. (1995) Subclinical manifestations of psychosis-proneness, ego-strength, and creativity.
Personality and Individual Differences, 18, p. 635-642.
Fodor, J. (1983) The Modularity of Mind. MIT Press, Cambridge MA.
Foulis, D. & Randall, C. (1972) Operational Statistics, I. Basic Concepts. Journal of Mathematical Physics,
13, p. 1667 - 1675.
Foulis, D. & Randall, C. (1973) Operational Statistics, II. Manuals of Operations and Their Logics. Journal
of Mathematical Physics, 14, p. 1472-1481.
Foulis, D., Piron C. & Randall, C. (1983) Realism, operationalism, and quantum mechanics. Foundations
of Physics, 13 (813).
Foulis, D. & Randall, C. (1981) What are quantum logics and what ought they to be? In (E. Beltrametti &
B. van Fraassen, eds.) Current Issues in Quantum Logic, 35, Plenum Press, New York.
Freedman, S. J. & Clauser, J. F. (1972) Experimental test of local hidden-variable theories. Physical Review
Letters, 28, p. 938-941.
Freeman, W. & Skarda, C. (1990) Representations: Who needs them? In (J. L. McGaugh, N. M.
Weinberger, & G. Lynch, eds.) Brain Organization and Memory Cells, Systems, and Circuits, Oxford
University Press, New York NY.
Gabora, L. (1995) Meme and variations: A computer model of cultural evolution. In (L. Nadel &
D. Stein, eds.) 1993 Lectures in Complex Systems. Addison-Wesley, NewYork.
http://www.lycaeum.org/~sputnik/Memetics/variations.txt
244
References
Gabora, L. (1996a) Culture, evolution and computation. In Proceedings of the Second Online Workshop on
Evolutionary Computation. Society of Fuzzy Theory and Systems.
http://www.bioele.nuee.nagoya-u.ac.jp/wec2/papers/p023.html
Gabora, L. (1996b). A day in the life of a meme. Philosophica, 57, p. 901-938.
http://www.lycaeum.org/~sputnik/Memetics/day.life.txt
Gabora, L. Memetics. (1997) In (Paul Bouissac, Ed.) The Encyclopedia of Semiotics and Cultural Studies,
Oxford University Press, Oxford.
Gabora, L. (1997) Taking memes seriously. Semiotic Review of Books, 8 (2), May issue.
http://www.univie.ac.at/Wissenschaftstheorie/srb/srb/memes.html
Gabora, L. (1997) The origin and evolution of culture and creativity. Journal of Memetics: Evolutionary
Models of Information Transmission, 1 (1).
http://www.cpm.mmu.ac.uk/jom-emit/1997/vol1/gabora_l.html
Gabora, L. (1998) Autocatalytic closure in a cognitive system: A tentative scenario for the origin of culture.
Psycholoquy, 9 (67) psyc.98.9.67.origin-culture.l.gabora
http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?9.67 [adap-org/9901002]
Gabora, L. (1999) Weaving, bending, patching, mending the fabric of reality: A cognitive science
perspective on worldview inconsistency. Foundations of Science, 3 (2), p. 395-428.
Gabora, L. (1999) A review of ‘The Meme Machine’ by Susan Blackmore. The Journal of Artificial
Societies and Social Systems, 2 (2).
http://www.soc.surrey.ac.uk/JASSS/2/2/review2.html Reprinted with permission in Dynamical Psychology
and Journal of Consciousness Studies.
Gabora, L. (1999) The Interwoven Conceptual Matrix of the Cultural Replicator. Commentary on ‘Niche
construction, biological evolution, and cultural change’ by K. N. Laland, J. Odling-Smee, & M. W.
Feldman, Behavioral and Brain Sciences, 23 (1). http://www.vub.ac.be/CLEA/liane/Reviews/Laland.htm
Gabora, L. (1999) Grounded in perception yet transformed into amodal symbols. Commentary on
‘Perceptual symbol systems’ by L. W. Barsalou. Behavioral and Brain Sciences, 22, p. 577-660.
http://www.vub.ac.be/CLEA/liane/Reviews/Barsalou.htm
245
Liane Gabora
Gabora, L. (2000) Conceptual closure: Weaving memories into an interconnected worldview. In (G. Van de
Vijver & J. Chandler, eds.) Closure: Emergent Organizations and their Dynamics. Vol. 901 of the Annals
of the New York Academy of Sciences, New York.
Gabora, L. (2000) The beer can theory of creativity. In (P. Bentley & D. Corne, eds.) Creative Evolutionary
Systems, Morgan Kauffman, London. http://www.vub.ac.be/CLEA/liane/PAPERS00/ces/beercan.htm
Gabora, L. (2000) Toward a theory of creative inklings. In (R. Ascott, Ed.) Art, Technology, and
Consciousness, Intellect Press, Bristol, UK. http://www.vub.ac.be/CLEA/liane/PAPERS00/atc/inklings.htm
Gabora, L. (2000) From a double aspect theory of information to human consciousness. Consciousness and
Cognition, 9 (2).
Gabora, L. & Aerts, D. (2000) Distilling the essence of an evolutionary process, and implications for a
formal description of culture. In (W. Kistler, Ed.) Proceedings of Center for Human Evolution Workshop
#5: Cultural Evolution, May 2000, Foundation for the Future, Seattle, WA.
Gabora, L. (in press) Amplifying phenomenal information: From double aspect theory to human
consciousness. [adap-org/9911003]
Gabora, L. M. & Colgan, P. W. (1990) A model of the mechanisms underlying exploratory behaviour. In
(S. Wilson and J. A. Mayer, eds.) The Simulation of Adaptive Behavior, MIT Press, Cambridge MA.
Gardner, M. H. (1983) Frames of Mind: The Theory of Multiple Intelligences. Basic Books, New York.
Gardner, M. H. (1993) Multiple Intelligences: The Theory in Practice. Basic Books, New York.
Gentner, D. (1983) Structure-mapping: A theoretical framework for analogy. Cognitive Science, 7 (2), p.
155-170.
Gick, M. L. & Holyoak, K. J. (1983) Schema induction and analogical transfer. Cognitive Psychology, 15,
p. 1-38.
Gould, S. J. (1991) Bully for Brontosaurus: Reflections in Natural History. W. W. Norton & Company,
NewYork, p. 63-66.
246
References
Gould, S. J. & Vrba, E. S. (1982) Exaptation—a missing term in the science of form. Paleobiology, 8 (1),
p. 4-15.
Greene, P.J. (1978) From genes to memes? Contemporary Sociology 7, p. 706-709.
Guilford, J.P. (1979) Some incubated thoughts on incubation. Journal of Creative Behavior, 13, p. 1-8.
Galef, B. G., Manzig, L. A., & Field, R. M. (1986) Imitation learning in budgerigars: Dawson and Foss
(1965) revisited. Behavioural Processes, 13, p. 191-202.
Gerrig, R. J., & Murphy, G. L. (1992). Contextual influences on the comprehension of complex concepts.
Language and Cognitive Processes, 7, p. 205-230.
Goldberg, D. E. (1989) Genetic Algorithms in Search, Optimization and Machine Learning. AddisonWesley, New York.
Goodall, J. (1986) The Chimpanzees of Gombe: Patterns of Behavior. The Belknap Press, Cambridge MA.
Gudder, S. P. (1988) Quantum Probability. Academic Press, Inc. Harcourt Brave Jovanovich, New York.
Hales, D. (in press) Understanding our creations: Using machine learning to understand machine models.
Hallpike, C. R. (1986) The Principles of Social Evolution. Clarendon Press, Oxford.
Hampton, J. A. (1987). Inheritance of attributes in natural concept conjunctions. Memory & Cognition, 15,
p. 55-71.
Hancock, P. J. B., Smith, L. S. & Phillips, W. A. (1991) A biologically supported error-correcting learning
rule. Neural Computation, 3 (2), p. 201-212.
Hauser, M. D. (1988) Invention and social transmission: New data from wild vervet monkeys. In (eds. R.
Byrne & A. Whiten) Machiavellian Intelligence. Clarendon Press, Oxford.
Hebb, D. (1949) The Organization of Behavior. Wiley and Sons, New York.
Heit, E., & Barsalou, L. W. (1996) The instantiation principle in natural language categories. Memory, 4, p.
413-451.
247
Liane Gabora
Heyes, C. M. (1995) Rats imitate and apes are flattered. Animal Behaviour, p. 1421-1424.
Heyes, C. M. (1998) Theory of mind in nonhuman primates. Behavioral and Brain Sciences, 21 (1), p. 104134.
Heyes, C. M., & Dawson, G. R. (1990) A demonstration of observational learning in rats using a
bidirectional control. Quarterly Journal of Experimental Psychology, 42B, p. 59-71.
Heyes, C. M., Dawson, G. R. & Nokes, T. (1992). Imitation in rats: initial responding and transfer evidence
from a bidirectional control procedure. Quarterly Journal of Experimental Psychology. Section B:
Comparative and Physiological Psychology, 45B, p. 229-240.
Heyes, C. M., Jaldow, E. & Dawson, G. R. (1994a) Imitation in rats: conditions of occurrence in a
bidirectional control paradigm. Learning and Motivation, 25, p. 276-287.
Heyes, C. M., Jaldow, E., Ray, E. & Dawson, G. R. (1994b) Imitation in rats: The role of demonstrator
action. Behavioural Processes, 32, p. 173-182.
Heyes, C. M. & Plotkin, H. C. (1989) Replicators and interactors in cultural evolution. In (M. Ruse, ed.)
What the Philosophy of Biology Is. Kluwer Academic Publishers, Dordrecht, The Netherlands, p. 139-162.
Heylighen, F. (1991) Modeling emergence. World Futures, 31, p. 89-104.
Heylighen, F. (1992) Evolution, Selfishness and Cooperation (Part 1). Selfish Memes and the Evolution of
Cooperation (Part 2). Journal of Ideas, 2 (4), p. 77-84.
ftp://ftp.vub.ac.be/pub/projects/Principia_Cybernetica/Papers_Heylighen/Memes&Cooperation.txt
Hinton, G. E. (1981) Implementing Semantic Networks in Parallel Hardware. In (G. E. Hinton & J. A.
Anderson, eds.) Parallel Models of Associative Memory. . Lawrence Erlbaum Associates, Hillsdale NJ.
Hinton, G. E. & J. A. Anderson. (1981) Parallel Models of Associative Memory. Lawrence Erlbaum
Associates, Hillsdale NJ.
Hochberg, J. (1969) In the mind’s eye. In (R. N. Haber, ed.) Contemporary Theory and Research
in Visual Perception. Holt Rinehart & Winston, New York, p. 309-331.
248
References
Hofbauer, J. & Sigmund, K. (1988) The Theory of Evolution and Dynamical Systems. Cambridge
University Press, Cambridge.
Holden, S.B. & Niranjan, M. (1997) Average-case learning curves for radial basis function networks.
Neural Computation, 9 (2), p. 441-460.
Holgate, P. (1966) A mathematical study of the founder principle of evolutionary genetic. Journal of
Applied Probability., 3, p. 115-128.
Holland, J. K. (1975) Adaptation in Natural and Artificial Systems. University of Michigan Press, Ann
Arbor MI.
Holland, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1986) Induction. MIT Press, Cambridge
MA.
Holt, R. A. & Pipkin, F. M. (1973) Preprint, Harvard University, Cambridge MA.
Hopfield, J. J. (1982) Neural networks and physical systems with emergent collective computational
abilities. Proceedings of the National Academy of Sciences (Biophysics), 79 (8), p. 2554-2558.
Hoyle, F. & Wickramasinghe, N. C. (1981) Evolution from Space. Dent, London.
Hubel, D.H. & Wiesel, T.N. (1979) Brain mechanisms and vision. Scientific American, 241 (3), p. 150-62.
Hughes, D.M. & Nei, M. (1988) Patterns of nucleotide substitution at major histocompatibility complex
class I loci reveals overdominant selection. Nature, 335, p. 167-170.
Hull, C. L. (1943) Principles of Behavior. Appleton Century Crofts, New York.
Hull. D. L. (1988a) Science as a process: An Evolutionary Account of the Social and Conceptual
Development of Science. The University of Chicago Press, Chicago.
Hull, D. L. (1988b) Interactors versus vehicles. In (H. C. Plotkin, ed.) The Role of Behavior in Evolution,
MIT Press, Cambridge, MA.
Hutchins, E. and B. Hazelhurst. (1992) Learning in the cultural process. In In (C. G. Langton, C. Taylor, J.
D. Farmer & S. Rasmussen, eds.) Artificial Life II. Addison-Wesley, Redwood City CA.
249
Liane Gabora
Hutter, M. (2001) Fitness Uniform Selection to Preserve Genetic Diversity.
http://arxiv.org/abs/cs.AI/0103015
Jacobs, R.A., Jordan, M.I., & Barto, A.G. (1991). Task decomposition through competition in a modular
connectionist architecture: The what and where vision tasks. Cognitive Science, 15, 219-250.
Jauch, J. (1968) Foundations of Quantum Mechanics. Addison-Wesley, Reading, MA.
Jeffreys, M. (2000) The meme metaphor. Perspectives in Biology and Medicine, 43 (2), p. 227-242.
Jerne, N. (1967) Antibodies and learning: selection versus instruction. In (G. Quarton, T. Melnechuk, & F.
Schmitt, eds.) The Neurosciences: A Study Program, Rockefeller University Press, New York, p. 200-205.
Jordan, M.I. and Jacobs, R.A. (1995) Modular and Hierarchical Learning Systems. In (M. A. Arbib, ed.)
The Handbook of Brain Theory and Neural Networks, p. 579-581.
Joyce, G. F. (1987) Non-enzymatic, template-directed synthesis of informational macromolecules. In: Cold
Spring Harbor Symposia on Quantitative Biology 52. Cold Spring Harbor Laboratory, New York.
Kanerva, P. (1988) Sparse Distributed Memory. MIT Press, Cambridge MA.
Kaplan, N. L., Hudson, R. R., & Langley, C. H. (1989) The Hitchhiking effect revisited. Genetics, 123, p.
887-899.
Karmiloff-Smith, A. (1986) From meta-processes to conscious access: Evidence from children’s
metalinguistic and repair data. Cognition 23, p. 95-147.
Karmiloff-Smith, A. (1990) Constraints on representational change: Evidence from children’s drawing.
Cognition 34, p. 57-83.
Karmiloff-Smith, A. (1992) Beyond Modularity: A Developmental Perspective on Cognitive Science. MIT
Press, Cambridge MA.
Kasday, L. R., Ullmann, J. D. & Wu, C. S. (1975) Angular correlation of compton-scattered annihilation
photons and hidden variables. Nuovo Cimento, 25, p. 633-660.
250
References
Kauffman, S. A. (1993) Origins of Order. Oxford University Press, New York.
Kauffman, S. A. (1999) Darwinism, neo-Darwinism, and an autocatalytic model of cultural evolution,
Commentary on Gabora on Origin-Culture. Psycoloquy, 10 (22). http://www.cogsci.soton.ac.uk/psycbin/newpsy?10.022
Keeler, J. D. (1988) Comparison between Kanerva’s SDM and Hopfield-type neural networks. Cognitive
Science, 12, p. 299-329.
Kendal, J. R. & Laland, K. N. (2000) Mathematical models for memetics. Journal of Memetics:
Evolutionary Models of Information Transmission, 4.
http://www.cpm.mmu.ac.uk/jom-emit/2000/vol4/kendal_jr&laland_kn.html
King, A. P. & West, M. J. (1989) Presence of female cowbirds (Molothrus aterater) affects vocal imitation
and improvisation in males. Journal of Comparative Psychology, 103 (39).
Klimesch, W. (1995) Memory processes described as brain oscillations. Psycoloquy, 6 (6),
http://cogsci.soton.ac.uk/~lac/psy/articles/6.6.html
Koestler, A. (1964) The Act of Creation. Picador, Nottingham UK.
Kojima, K. & Schaeffer, H. E. (1967) Survival processes of linked mutant genes. Evolution, 21, p. 518-531.
Krebs, J. R. & Dawkins, R. (1984) Animal signals: mind reading and manipulation. In (J. R. Krebs & N. B.
Davies, eds.) Behavioral Ecology. Sinauer Associates, Sunderland MA.
Kubose, S. K. & Umenoto, T. (1980) Creativity and the Zen koan. Psychologia, 23 (1), p. 1-9.
Lake, M. (1998) Digging for memes: the role of material objects in cultural evolution. In (C. Renfrew & C.
Scarre, eds.) Cognition and Material Culture: The Archeology of Symbolic Storage. McDonald Institute
Monographs, p. 77-88.
Langton, C. G. (1992) Life at the edge of chaos. In (C. G. Langton, J. D. Farmer & S. Rasmussen, eds.)
Artificial Life II, Addison-Wesley, New York NY.
Latimer, C. R. & Stevens, C. J. (1997) Some remarks on wholes, parts and their perception. Psycoloquy, 8
(13). http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?08.13
251
Liane Gabora
Leakey, R. (1984) The Origins of Humankind. Science Masters Basic Books, New York.
Lee, D. H., Granja, J. R., Martinez J. A., Severin, K. & Ghadiri, M. R. (1996) A Self-Replicating Peptide.
Nature. 382 (8) August, p. 525-528.
Lee, D., Severin, K., Yokobayashi & Ghadiri, M. (1997) Emergence of symbiosis in peptide selfreplication through a hypercyclic network. Nature, 390 (11) December, p. 591-594.
Lenat, D. B. (1974) The ubiquity of discovery. Artificial Intelligence Journal, 9, p. 257-86.
Lieberman, P. (1991) Uniquely human: The Evolution of Speech, Thought, and Selfless Behavior. Harvard
University Press, Cambridge MA.
Lorenz, K. (1971) Studies in Animal and Human Behavior. Harvard University Press, Cambridge MA.
Luchins, A.S. (1942) Mechanization in problem solving. Psychological Monographs, 54 (248).
Lumsden, C. and E. O. Wilson. (1981) Genes, Mind, and Culture. Harvard University Press, Cambridge
MA.
Lu, Y.W., Sundararajan, N. & Saratchandran, P. (1997) A sequential learning scheme for function
approximation using minimal radial basis function neural networks. Neural Computation, 9 (2), p. 461-478.
Lynch, A. (1996) Thought Contagion: How Belief Spreads Through Society. Basic Books, New York.
Mackey, G. W. (1963) Mathematical Foundations of Quantum Mechanics. Benjamin, Reading MA.
Mackintosh, D. R. & Stewart, G. T. (1979) A mathematical model of a heroin epidemic: Implications for
control policies. Journal of Epidemiology and Community Health, 33, p. 299-301.
Mandelbrot, B. B. (1982) The Fractal Nature of Geometry. W.H. Freedman and Company, San Francisco.
Marr, D. (1982) Vision. Freeman, Berkeley Heights, NJ.
Martindale, C. (1977) Creativity, consciousness, and cortical arousal. Journal of Altered States of
Consciousness, 3, p. 69-87.
252
References
Martindale, C. (1999) Biological bases of creativity. In (R. J. Sternberg, ed.) Handbook of Creativity,
Cambridge University Press, Cambridge, p. 137-152.
Martindale, C. & Armstrong, J. (1974) The relationship of creativity to cortical activation and its operant
control. Journal of Genetic Psychology, 124, p. 311-320.
Martindale, C. & Hasenfus, N. (1978) EEG differences as a function of creativity, stage of the creative
process, and effort to be original. Biological Psychology, 6, p. 157-167.
Maturana, H. R. & Varela, F. J. (1980) Autopoiesis and Cognition: The Realization of the Living.
Reidel, Boston. (Original work published 1973.)
Maynard Smith, J. & Haigh, J. (1974) The hitchhiking effect of a favourable gene. Genetic Research, 23, p.
23-35.
Maynard-Smith, J. & Szathmary, E. (1994) The Major Transitions in Evolution. Oxford University Press,
Oxford.
McCulloch, W.S. & Pitts, W. (1943) A logical calculus of the ideas immanent in nervous activity. Bulletin
of Mathematical Biophysics, 5, p. 115-133.
McFarland, D. J. & Sibly, R. M. (1975) The behavioural final common path. Philosophical Transactions of
the London Royal Society, 270B, p. 265-93.
Medin, D. L., & Shoben, E. J. (1988) Context and structure in conceptual combinations. Cognitive
Psychology, 20, p. 158-190.
Mednick, S. A. (1962) The associative basis of the creative process. Psychological Review, 69, p. 220-232.
Mellars, P. (1973) The character of the Middle-Upper transition in south-west France. In (C. Renfrew, ed.)
The Explanation of Culture Change, Duckworth, London UK, p. 255-276.
Mellars, P. (1989a) Major issues in the emergence of modern humans. Current Anthropology, 30, p. 349385.
253
Liane Gabora
Mellars, P. (1989b) Technological changes at the Middle-Upper Paleolithic transition:Economic, social,
and cognitive perspectives. In (Eds. P. Mellars and C. Stringer) The Human Revolution. Edinburgh
University Press, Edinburgh, p. 338-365.
Mendelsohn, G. A. (1976) Associative and attentional processes in creative performance. Journal of
Personality, 44, p. 341-369.
Mignault, C. (1965) Transition between sensorimotor and symbolic activities in nursery-reared
chimpanzees (Pan troglodytes). Journal of Human Evolution, 14, p. 747-758.
Miller, G. A. (1956) The magic number seven, plus or minus two: Some limits on our capacity for
processing information. Psychological Review, 63, p. 81-97.
Miller, S. L. (1955) Production of some organic compounds under possible primitive earth conditions.
Journal of the American Chemistry Society, 77, p. 23-51.
Minsky, M. (1975) A framework for representing knowledge. In (P. H. Winston, ed.) The Psychology of
Computer Vision. McGraw-Hill, New York, NY, p. 211-280.
Minsky, M. (1985) The Society of Mind. Simon and Schuster, New York NY.
Mitchell, M. (1993) Analogy-making as Perception. MIT Press, Cambridge MA.
Mitchell, R. W. & Thompson, N. S., eds. (1986) Deception: Perspectives on Human and Nonhuman
Deceit. SUNY Press, Binghamton NY.
Morowitz, H. J. (1992) The Beginnings of Cellular Life. Yale University Press, New Haven CT.
Murphy, G. L. & Medin, D. L. (1985) The role of theories in conceptual coherence. Psychological Review,
92, p. 289-316.
Nersessian, N. (1993) In the theoretician’s laboratory: Thought experimenting as mental modeling. In PSA
1992, 2, eds. D. Hull, M. Forbes, & K Okrulik. PSA.
Newman S. A. & Müller, G. B. (1999) Morphological evolution: Epigenetic mechanisms. In: Embryonic
Encyclopedia of Life Sciences, Nature Publishing Group,London.
254
References
Nishida, T. (1986) Local traditions and cultural transmission. In (B. B. Smuts, D. L. Cheney, R. M.
Seyfarth, R. W. Wrangham, & T. T. Struhsaker, eds.) Primate Societies, University of Chicago Press,
Chicago IL.
Olton, D. S. (1977) Spatial memory. Scientific American, 236, p. 82-98.
Orgel, L. E. (1987) Evolution of the genetic apparatus: A review. In Cold Spring Harbor Symposia on
Quantitative Biology Vol. 52. Cold Spring Harbor Laboratory, New York.
Orr, H. A. (1996) Dennett’s strange idea. Boston Review, 21 (3).
http://bostonreview.mit.edu/BR21.3/Orr.html See also his reply to Dennett’s response in 21 (5).
http://bostonreview.mit.edu/BR21.5/Orr.html
Orsucci, F. (1998) The Complex Matters of the Mind. World Scientific Publishing Co, Danvers MA.
Oparin, A. I. (1971) Routes for the origin of the first forms of life. Subcellular Biochemistry 1 (75).
Palmer, M. W. (1992) The coexistence of species in fractal landscapes. American Naturalist, 139, p. 375397.
Penrose, R., & Hameroff, S. (1998) Quantum computation in brain microtubules? The Penrose-Hameroff
“Orch OR” model of consciousness, Philosophical Transactions Royal Society London (A) 356, p. 18691896.
Piaget, J.(1926) The Language and Thought of the Child. Harcourt Brace, Kent UK.
Piaget, J. (1936/52) The Origin of Intelligence in Children. International Universities Press, Madison CT.
Pinker, S. (1995) The Language Instinct. Harper Perrenial, New York.
Piron, C. (1976) Foundations of Quantum Physics. Benjamin, Reading, Mass.
Piron, C. (1989) Recent developments in quantum mechanics. Helvetica Physica Acta, 62 (82).
Piron, C. (1990) Mécanique Quantique: Bases et Applications, Press Polytechnique de Lausanne,
Lausanne, Suisse.
255
Liane Gabora
Pitowsky, I. (1989) Quantum Probability—Quantum Logic, Lecture Notes in Physics 321. Springer, Berlin,
New York.
Plotkin, H. C. (1988) The Role of Behavior in Evolution. MIT Press, Cambridge MA.
Plotkin, H. (1998) Evolution in Mind: An Introduction to Evolutionary Psychology. Harvard Press,
Cambridge MA.
Popper, K. (1963) Conjectures and Refutations. Routledge, New York.
Port, R., Anderson, S. & McAuley, D. (1996) Toward simulated audition in open environments. In (Covey,
E., Hawkins, H., and Port, R., eds.) Neural Representation of Temporal Patterns, Plenum, New York NY.
Premak, D. & Woodruff, G. (1978) Does the chimpanzee have a theory of mind? Behavioral and Brain
Sciences, 1, p. 515-526.
Pribram, K.H. (1974) How is it that sensing so much we can do so little? In (F. O. Schmidt & F. G.
Worden, eds.) The Neurosciences Third Study Program, MIT Press, Cambridge MA.
Pribram, K. H., Ed. (1994) Origins: Brain and Self Organization. Erlbaum, Hillsdale NJ.
Prigogine, I. & Stengers, I. (1984) Order out of Chaos. Bantam Books, New York NY.
Quillian, M. R. (1968) Semantic memory. In (M. Minsky, ed.) Semantic Information Processing, MIT
Press, Cambridge MA.
Ram, A. (1993) Creative conceptual change. Proceedings of the Fifteenth Annual Conference of the
Cognitive Science Society, p. 17-26.
Randall, C. & Foulis, D. (1976) A mathematical setting for inductive reasoning. In (C. A. Hooker, ed.)
Foundations of Probability Theory, Statistical Inference, and Statistical Theories of Science III, 169, Reidel
Publishing, Dordrecht, Holland.
Randall, C. & Foulis, D. (1978) The Operational Approach to Quantum Mechanics. In (C. A. Hooker, ed.),
Physical Theories as Logico-Operational Structures, 167, Reidel Publishing, Dordrecht, Holland.
256
References
Ray, T. (1991) An approach to the synthesis of life. In (C. G. Langton, C. Taylor, J. D. Farmer & S.
Rasmussen, eds.) Artificial Life II, Addison-Wesley, New York.
Reed, S. K. (1972). Pattern recognition and categorization. Cognitive Psychology, 3, p. 382-407.
Rescher, N., & Oppenheim, P. (1955) Logical analysis of gestalt concepts. British Journal for the
Philosophy of Science, 6, p. 89-106.
Richards, R. L., Kinney, D. K., Lunde, I., Benet, M., & Merzel, A. (1988) Creativity in manic depressives,
cyclothymes, their normal relatives, and control subjects. Journal of Abnormal Psychology, 97, p. 281-289.
Richerson, P. (1998) The pleistocene and the origins of human culture: built for speed. Paper presented at
The 5th Biannual Symposium on the Science of Behavior: Behavior, Evolution, and Culture, University of
Guadalajara, Mexico.
Riegler, A. Peschl, M. & von Stein, A. (1999) Understanding Representation in the Cognitive Sciences.
Kluwer, Dordrecht, The Netherlands.
Robert, M. (1990) Observational learning in fish, birds, and mammals: A classified bibliography spanning
over 100 years of research. Psychological Record 40, p. 289-311.
Rodgers, J.L. & Rowe, D.C. (1993) Social contagion and adolescent sexual behavior: A developmental
EMOSA model. Psychological Review, 100 (93), p. 479-510.
Rogers, E. M. (1995) Diffusion of Innovations. Free Press, New York.
Rosch, E. (1975) Cognitive reference points. Cognitive Psychology, 7, p. 532-547.
Rosch, E. (1978) Principles of categorization. In (E. Rosch & B. B. Lloyd, eds.), Cognition and
Categorization, Erlbaum, Hillsdale NJ, p. 27-48.
Rosch, E. (1983) Prototype classification and logical classification: The two systems. In (E. K. Scholnick,
ed.), New Trends in Conceptual Representation: Challenges to Piaget’s Theory?, Erlbaum, Hillsdale NJ, p.
73-86.
Rosch, E. (1987) Principles of categorization. In (E. Rosch & B. B. Lloyd, eds.) Cognition and
Categorization. Erlbaum, Hillsdale, NJ.
257
Liane Gabora
Rosch, E. (1999) Reclaiming concepts. Journal of Consciousness Studies, 6 (11), p. 61-78.
Rosch, E., & Mervis, C. B. (1975). Family resemblances: Studies in the internal structure of categories.
Cognitive Psychology, 7, p. 573-605.
Rosen, R. (1985) Anticipatory Systems: Philosophical, Mathematical & Methodological Foundations,
Pergamon Press, New York.
Rosen, R. (1991) Life Itself: A Comprehensive Inquiry into the Nature, Origin, and Fabrication of Life.
Columbia University Press, New York.
Rosenfield, I. (1993) The Strange, Familiar, and Forgotten: An Anatomy of Consciousness. Vintage,
Newbury Park CA.
Roszak, T., Gomes, M. E., & Kanner, A. D. (1995) Ecopsychology: Restoring the Earth, Healing the Mind,
Sierra Club Eco Books, Oakland CA.
Rozin, P. (1976) The evolution of intelligence and access to the cognitive unconscious. In Progress in
Psychobiology and Physiological Psychology. Eds. J. M. Sprague and A. N. Epstein. P. 245-277. Academic
Press, New York.
Rumelhart, D. E. & McClelland, J. L. eds. (1986) Parallel Distributed Processing. Bradford/MIT Press,
Cambridge MA.
Russ, S. W. (1993) Affect and Creativity. Erlbaum, Hillsdale NJ.
Russon, A. E., Galdikas, B. M. F. (1993) Imitation in free-ranging rehabilitant orangutans. Journal of
Comparative Psychology, 107, p. 147-161.
Schank, R. C. (1983) Dynamic Memory. Cambridge University Press, Cambridge MA.
Schank, R., & Abelson, R. P. (1977) Scripts, Plans, Goals, and Understanding: An Inquiry into Human
Knowledge Structures. Lawrence Erlbaum Associates, Mahwah NJ.
Schank, R. C. & Leake, D. B. (1989) Creativity and learning in a case-based explainer. Artificial
Intelligence, 40 (13), p. 353-385.
258
References
Scheuring, I. & Riedi, R. H. (1994) Application of multifractals to the analysis of vegetation pattern.
Journal of Vegetation Science, 5, p. 489-496.
Schuster, P. & Sigmund, K. (1983) Replicator dynamics. Journal of Theoretical Biology, 100, p. 533-38.
Schwartz, J. H. (1999) Sudden Origins. Wiley, New York.
Segal, I. E. (1947) Postulates for general quantum mechanics. Annals of Mathematics, 48 (930).
Severin, K., Lee, D. H., Kennan, A. J. & Ghadiri, M. R. (1997) A synthetic peptide ligase. Nature, 389 (16)
October, p. 706-709.
Shannon, C.E. & Weaver, W. (1963) The Mathematical Theory of Communication. University of Illinois
Press, Chicago.
Shear, J., ed. (1997) Explaining Consciousness: The Hard Problem. MIT Press, Cambridge MA.
Sheets-Johnstone, M. (1999) The Primacy of Movement. John Benjamins, Amsterdam.
Simonton, D. K. (1984) Genius, Creativity and Leadership: Historiometric Inquiries. Harvard University
Press, Cambridge MA.
Simonton, D. K. (1999) Creativity from a historiometric perspective. . In (R. J. Sternberg, ed.) Handbook of
Creativity, Cambridge University Press, Cambridge, p. 116-136.
Sims, K. (1991) Artificial evolution for computer graphics. Computer Graphics, 25 (4), p. 319-328.
Skarda, C. A. (1999) The perceptual form of life. In Journal of Consciousness Studies, 6 (11), p. 79-94.
Smith, E. E., & Medin, D. M. (1981). Categories and Concepts. Harvard University Press, Cambridge,
MA.
Smith, G. J. W. & Van de Meer, G. (1994) Creativity through psychosomatics. Creativity Research
Journal, 7, p. 159-170.
Smith, W. J. (1977) The Behavior of Communicating. Harvard University Press, Cambridge MA.
259
Liane Gabora
Smolin, L. & Cemin, S. C. (1997) Coevolution of membranes and channels: A possible step in the origin of
life. Journal of Theoretical Biology (October issue), adap-org/9709004.
Spector, L. & Luke, S. (1996) Culture enhances the evolvability of cognition. In Proceedings of the 1996
Cognitive Science Society Meeting.
Speel, H. C. (1997) A Memetic Analysis of Policy Making. Journal of Memetics: Evolutionary Models of
Information Transmission, 1. http://www.cpm.mmu.ac.uk/jomemit/1997/vol1/speel_hc.html
Sperber, D. (1994) The modularity of thought and the epidemiology of representations. In (L. Hirschfeld &
S. A. Gelman, eds.) Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge
University Press, Cambridge, p. 39-67.
Squire, L.R. (1992) Memory and the hippocampus: A synthesis from findings with rats, monkeys, and
humans. Psychological Review, 99, p. 195-231.
Steels, L. (1995) Intelligence—Dynamics and Representations. In: (L. Steels, ed.) The Biology and
Technology of Intelligent Autonomous Agents. Springer-Verlag, Berlin, Germany.
Steels, L. (1996) The origins of intelligence. In Proceedings of the Carlo Erba Foundation Conference on
Artificial Life. Fondazione Carlo Erba, Milano, Italy.
Steels, L. (1996) Perceptually grounded meaning creation. In (M. Tokoro, ed.) Proceedings of ICMAS-96,
Kyoto 1996. AAAI Press, San Francisco, USA.
Steels, L. (1997) Synthesising the origins of language and meaning using co-evolution, self-organisation
and level formation. In (Hurford, J., C. Knight and M. Studdert-Kennedy, eds.) Evolution of Human
Language. Edinburgh Univ. Press, Edinburgh, Scotland.
Steuer, R. E. (1986) Multiple Criteria Optimization: Theory, Computation, and Applications. John Wiley
and Sons, New York.
Storms, G., De Boeck, P., Hampton, J. A., & Van Mechelen, I. (1999). Predicting conjunction typicalities
by component typicalities. Psychonomic Bulletin and Review, 4, p. 677-684.
260
References
Storms, G., De Boeck, P., & Ruts, W. (2000). Prototype and exemplar based information in natural
language categories. Journal of Memory and Language, 42, p. 51-73.
Storms, G., De Boeck, P., Van Mechelen, I. & Ruts, W. (1996). The dominance effect in concept
conjunctions: Generality and interaction aspects. Journal of Experimental Psychology: Learning, Memory
& Cognition, 22, p. 1-15.
Storms, G., De Boeck, P., Van Mechelen, I. & Ruts, W. (1998). Not guppies, nor goldfish, but tumble
dryers, Noriega, Jesse Jackson, panties, car crashes, bird books, and Stevie Wonder. Memory and
Cognition, 26, p. 143-145.
Sumita, K., Kitahara-Frisch, J. & Norikoshi, K. (1985). The acquisition of stone-tool use in captive
chimpanzees. Primates, 26, p. 168-181.
Sutcliffe, J. P. (1993). Concepts, class, and category in the tradition of Aristotle. In (I. Van Mechelen, J. A.
Hampton, R. S. Michalski, & P. Theuns, eds.), Categories and concepts: Theoretical views and inductive
data analysis (p. 35-65). Academic Press, London.
Terrace, H. S., Petitto, L. A., Sanders, R. J. & Bever, T. G. (1979) Can an ape create a sentence? Science,
206, p. 891-902.
Thagard, P. (1980) Against evolutionary epistomology. In (P. D. Asquith & R. N. Giere, eds.) PSA 1980, p.
187-96.
Thelen, E. & Smith, L. (1994) A Dynamical Systems Approach to the Development of Cognition and
Action. MIT Press, Cambridge MA.
Todd, S. & Latham, W. (1992) Evolutionary Art and Computers. Academic Press, London.
Tomasello, M. Kruger, A. C., & Ratner, H. H. (1993) Cultural learning. Behavioral and Brain Sciences, 16,
p. 495-552.
Tooby, J. & Cosmides, L. (1989) Evolutionary psychology and the generation of culture, Part I. Ethology
and Sociobiology, 10, p. 29-49.
Trivers, R. L. (1976) Forward in R. Dawkins’ The Selfish Gene. Oxford University Press, Oxford.
261
Liane Gabora
Trivers, R. L. (1985) Social Evolution. Cummings, Menlo Park CA.
van Gelder, T. (1995) What might cognition be, if not computation? Journal of Philosophy, xi, p. 345-381.
Van Gelder, T. (1998) The dynamical systems hypothesis. Behavioral and Brain Sciences, 21, p. 1-14.
van Gelder, T. & Port, R. (1995) It’s About Time: An Overview of the Dynamical Approach to Cognition.
MIT Press, Cambridge MA.
Verbeemen, T., Vanoverberghe, V., Storms, G., & Ruts, W. (in press). Contrast categories in natural
language concepts. Journal of Memory and Language.
Wisniewski, E. J. (1997). When concepts combine. Psychonomic Bulletin & Review, 4, p. 167-183.
Wisniewski, E. J., & Gentner, D. (1991). On the combinatorial semantics of noun pairs: Minor and major
adjustments. In (G. B. Simpson, ed.) Understanding Word and Sentence, Elsevier, Amsterdam, p. 241-284.
Wheeler, M. (1994) From activation to activity: Representation., computation, and the dynamics of neural
network control systems. Artificial Intelligence and the Simulation of Behavior Quarterly, 87, p. 36-42.
White, R. (1982) Re-thinking the Middle/Upper Paleolithic transition. Current Anthropology, 23, p. 169192.
White, R. (1993) Technological and social dimensions of ‘Aurignacian-Age’ body ornaments across
Europe. In (Eds. H. Knecht, A. Pike-Tay, & R. White) Before Lascaux: The Complex Record of the Early
Upper Paleolithic. CRC Press, Boca Raton, p. 247-299.
Varela, F. J. (1979) Principles of biological autonomy. North-Holland, New York.
Varela, F., Thompson, E., & Rosch, E. (1991) The Embodied Mind. MIT Press, Cambridge MA.
Von Neumann, J. (1966) Theory of Self-reproducing Automata. University of Illinois Press, Chicago.
Vygotsky, L.S. (1978) Mind in Society: The Development of Higher Psychological Processes. In eds. M.
Cole, V. John-Steiner, S. Scribner & E. Souberman, Harvard University Press, Cambridge MA.
262
References
Waal, F. (1991) Complementary methods and convergent evidence in the study of primate social cognition.
Behaviour, 118, p. 297-320.
Wallace, B. (1973) Misinformation, fitness, and selection. American Naturalist, 107, p. 1-7.
Wallas, G. (1926) The Art of Thought. Harcourt, Brace & World, New York.
Walsh, R. & Vaughan, F. (1993) Paths Beyond Ego, Jeremy P. Tarcher/Perigee, Los Angeles CA.
Weisberg, R. W. (1986) Creativity: Genius and other Myths. Freeman, New York.
Wesson, R. (1994) Beyond Natural Selection. The MIT Press, Cambridge MA.
Westergaard, G. C. (1988) Lion-tailed macaques (Macaca silenus) manufacture and use tools. Journal of
Comparative Psychology, 102, p. 152-159.
Whiten, A. (1998) Imitation of the social structure of actions by chimpanzees (Pan troglodytes). Journal of
Comparative Psychology, 112 (270).
Willshaw, D. & Dayan, P. (1990) Optimal plasticity from matrix memories: What goes up must come
down. Neural Computation, 2 (1), p. 85-93.
Wright, S. (1969) Evolution and the Genetics of Populations. University of Chicago Press, Chicago.
Zadeh, L (1965) Fuzzy sets. Information & Control, 8, p. 338-353.
Zadeh, L. (1982) A note on prototype theory and fuzzy sets. Cognition, 12, p. 291-297.
263