Emergent Natural Selection for Engineering
Living Systems
Bradly Alicea, Michigan State University
http://www.msu.edu/~aliceabr/
INTRODUCTION
Motivating Question:
Can we focus on the role of natural
selective forces in the construction of
complex organisms and living systems?
* perhaps, if we treat a complex organism as an
emergent, hierarchical directed graph with
control properties.
* fundamental units can exist above cellular
level, but subject to aggregate or selective
genetic control.
* useful heuristic for approximating emergent
properties and aggregate effects. Units may or
may not have a genomic representation.
B) Free energy (e.g. Gibbs, Landau de
Gennes [1,2]) = information, natural
selection = constraint .
* framework for self-assembly at single spatial
scale [3].
Must rely on the following:
A) fundamental unit: a volume that serves as a
minimal description of biological function at the
cellular level.
* based on a prediction of hierarchy theory [4]
for ecological systems.
* at cellular population level, this can be a
single cell.
* makes approach amenable to fault-tolerance,
robustness.
* also a tenet of complex adaptive systems
(CAS) approaches [5] to emergence
Abstraction:
1) organism is a hierarchy of “fundamental
units” [6].
* fundamental units in hierarchy can be
connected in various modes. Provide modular
and structural organization.
* connections are pruned when selection
exceeds free energy input.
Focus is on organizational complexity,
trophic complexity of cellular populations.
* fundamental units are replicators that exhibit
diversity
embodied
in
a
genotypic
representation.
Selection,
magnitude
Scope
(S)
Transformity
(T)
Connectivity
(C)
Low, large
1
10-1
small
Low, small
1
10-1
large
High, small
1
10-max
large
High, large
1
10-max
small
* replicator fitness is maximized when the
energy (E) value is high and large aggregate
effect is approaches 1.0.
3) basic set of relations.
2) Metrics for characterizing hierarchical
natural selection. For more details, see end
of poster and [6].
* describe both selection and information.
* scope (S), transformity (T) [see 1], and
connectivity (C) variables contribute to the
magnitude of selection in a relative manner.
* pervasiveness and relative embodied energy
contribute to fitness in an additive fashion.
Figure 1: Example of relations between
fundamental units, different hierarchical levels.
Figure
3:
example of two
laminar
flow
fields on an
unconnected set
of fundamental
units.
Composite flow
field
example:
laminar
and
coriolis.
Figure 2: an example of recursive sampling [7] and
representation from skeletal muscle.
3) geometric assembly conditions.
To achieve the types of tissues seen in
Figure 2, a geometric component must be
added to this model.
If third dimension added to graph
(unconnected array structure of units), shape
properties can be approximated.
* structures result from flows, which are
constrained by minimization of energy.
Consistent with constructal theory [8].
* flows added to model, act externally to
network (have an aggregate effect on entire
structure).
* complex flow fields (composite flows) might
be able to structure units at a single scale to
mimic modularity [see 9] in multicellular
organisms.
* laminar field (Figure 3) can act upon
connectivity, replication ability of units.
* tissues and structures at single scales can
also form structural networks [10] in response
to local mechanical conditions.
In Figure A, fundamental units can be
treated as either “black boxes” (Organ and
Tissue A in Figure 1) or groups of units in
an array (cell populations, Figure 1)).
System Behavior:
The goal of these networks is not to imitate
a neural network or to simulate the
evolution of organisms, but to approximate
the dynamical behavior of self-assembled
systems.
Connections between fundamental units are
critical to determining the hierarchical
nature of these evolutionary systems:
1) Information: acts as
mechanism for parental units.
a
support
Biological analogue: trophic and tropic
signaling.
Examples:
* diffusion (stochastic and directional).
* one-to-one chemical signaling.
2) Selection: acts as a constraint on child
units.
Determined by rules such as logical
functions, simple relational equations [11].
Examples:
*
helps to maintain function modules
composed of child units.
* feedback mechanism for determining which
cell populations, biological components are
needed to form a functional organ.
3) Aggregate selection: acts on all units in
a single module or at a specific scale.
Governed by force fields and flow fields.
Approximates
physical
aspects
of
development and tissue self-assembly [12].
Examples:
*
cell sorting in development based on
chemotactic signals [13].
* local cyclic stresses determine cell fate in
stem cells [14].
Examples from neural evolution/development:
Parcellation theory
(left)
[15]
and
population matching
(right) [16].
Displacement
hypothesis and
developmental
plasticity (right)
[16, 17].
Figure 4: Four versions of
connectivity.
Natural
selection acts on populations
recursively.
A: selection imposed at
lower level of hierarchy,
small magnitude of effect.
B: selection imposed at
higher level of hierarchy,
little to no effect.
C: selection imposed at
lower level of hierarchy,
small magnitude of effect.
D: selection imposed at
higher level of hierarchy,
large magnitude of effect.
Potential application domains.
There are many potential applications of this
theoretical framework, from tissue engineering
to biological analysis. Below are two key areas
for further study.
Nanoassembly.
At the micro- and nano-scale, self- and human-assisted
assembly of synthetic elements is impacted greatly forces and
stochastic processes. One way to do this is to directly
implement a genetic algorithm [18]. However, the method
proposed here might be a more open-ended way to approach
the problem.
Biomimetic Control.
Knowing which units and modules lead to higher-level
structures requires the implementation of a biological control
policy. This control policy could be loosely based on
hierarchical natural selection.
Variables and Definitions:
Scope.
Connectivity.
Uhs = # of units at 10-n under selection.
Uhns = # of units at 10-n not under
selection.
Un = single unit at 10-n.
Cu = # connections to
units at other levels.
Transformity.
Scaling
hierarchical levels [1].
Energy.
factor
between
Gn = free energy, Bn =
bond energy. max, min =
maximum,
minimum
energy across hierarchy.
Cmax = maximum
connectivity across
hierarchy.
Pervasiveness.
References:
[1]
Odom, H.T. and Odum, E.C.
Modeling at All Scales: an introduction to
system simulation. Acadmeic Press, New
York (2000).
[2] DeGennes, P.G. Theory of Liquid
Crystals. Clarendon Press, Oxford, UK
(1974).
[3] Whitesides, G.M. and B. Grzybowski
Self-assembly at all scales. Science 295,
2418–2421 (2002).
[4] O'Neill, R.V., D. DeAngelis, J.B.
Waide, and T.F.H. Allen. A Hierarchical
Concept of Ecosystems. Princeton
University Press, Princeton (1986).
[5] Holland, J.H. Adaptation in natural
and artificial systems. MIT Press,
Cambridge, MA (1992).
[6]
Alicea,
B.
Hierarchies
of
Biocomplexity: modeling life’s energetic
complexity. arXiv:0810.4547 [q-bio.PE]
(2008).
[7] Weibel, E.R. Symmorphosis: on form
and function in shaping life. Harvard
Press, Cambridge, MA (2000).
[8] Bejan, A. Shape and Structure, from
Engineering to Nature. Cambridge
University Press, 2000.
[9] Schlosser, G. and Wagner, G.P.
Modularity in development and evolution.
University of Chicago Press, Chicago
(2004).
10] Mizuno, D., Tardin, C., Schmidt, C.F.,
MacKintosh,
F.C.
Nonequilibrium
mechanics of active cytoskeletal
Networks. Science, 315, 370 (2007).
[11] Wolfram, S. A new kind of science.
Wolfram Press, Champaign, IL (2002).
[12] Kafer et.al Moving Forward
Moving Backward. PLoS Computational
Biology, 2(6), e56 (2006).
[13] Steinberg, M.S. Differential
adhesion in morphogenesis: a modern
view. Current Opinion in Genetics &
Development, 17, 281–286 (2007).
[14] Chowdhury et.al Material properties
of the cell dictate stress-induced
spreading
and
differentiation
in
embryonic stem cells. Nature Materials,
9, 82-88 (2010).
[15] Ebbesson, S.O.E. Evolution and
Ontogeny of neural circuits. Behavioral
and Brain Research, 7, 321-366 (1984).
[16] Striedter, G.F. Principles of Brain
Evolution. Sinauer, Sunderland, MA
(2006).
[17]
Deacon, T.W. Rethinking
mammalian brain evolution. American
Zoologist, 30, 629-705 (1990).
[18]
Alicea,
B.
Hierarchical
Evolutionary
Dynamics
for
Understanding Self-Assembly in NanoMechanical
Systems.
IEEE
Fall
Confernece
on
Nanotechnology,
Nanorobots,
Nanobusiness,
and
Nanoeducation. Ypsilanti, MI (2009).