Brit. J. Phil. Sci. 59 (2008), 121–141
Generous or Parsimonious
Cognitive Architecture? Cognitive
Neuroscience and Theory of Mind
Philip Gerrans and Valerie E. Stone
ABSTRACT
Recent work in cognitive neuroscience on the child’s Theory of Mind (ToM) has pursued
the idea that the ability to metarepresent mental states depends on a domain-specific
cognitive subystem implemented in specific neural circuitry: a Theory of Mind Module.
We argue that the interaction of several domain-general mechanisms and lower-level
domain-specific mechanisms accounts for the flexibility and sophistication of behavior,
which has been taken to be evidence for a domain-specific ToM module. This finding
is of more general interest since it suggests a parsimonious cognitive architecture can
account for apparent domain specificity. We argue for such an architecture in two
stages. First, on conceptual grounds, contrasting the case of language with ToM, and
second, by showing that recent evidence in the form of fMRI and lesion studies supports
the more parsimonious hypothesis.
1
2
3
4
5
6
Theory of Mind, Metarepresentation, and Modularity
Developmental Components of ToM
The Analogy with Modularity of Language
Dissociations without Modules
The Evidence from Neuroscience
Conclusion
1 Theory of Mind, Metarepresentation, and Modularity
In 1985, Simon Baron-Cohen, Alan Leslie, and Uta Frith published an
influential paper entitled ‘Does the autistic child have a theory of mind
(ToM)?’ They presented evidence that children with autism were impaired
relative to typically developing and Downs syndrome children on nonverbal
‘false-belief’ tests of ToM, the ability to form beliefs about the mental
states of others. Recently, some researchers in cognitive neuroscience have
argued, on the basis of neuroimaging results, that the temporoparietal
 The Author (2008). Published by Oxford University Press on behalf of British Society for the Philosophy of Science. All rights reserved.
doi:10.1093/bjps/axm038
For Permissions, please email: [email protected]
122
Philip Gerrans and Valerie E. Stone
junction (TPJ) contains a domain-specific cognitive system specialized for
ToM. This suggests an elegant theoretical and empirical marriage between
developmental psychology and neuroscience. The basis of this marriage
is the idea that cognitive neuroscientists and developmental psychologists
are studying the same domain-specific ‘module’ at different levels; with
neuroscience revealing the neural substrate of a discrete cognitive capacity.1
We argue, however, that both fields are studying the emergent outcome of
interaction between numerous domain-specific low-level cognitive systems
and domain-general cognitive systems. The low-level domain-specific systems
such as joint attention, gaze tracking, animacy detection, and recognition of
emotional expression develop early (Baron-Cohen [1995]; Rutherford et al.
[2006]; Stone [2005]). The domain-general systems such as metarepresentation
(MR), secondary representation (SR), executive function (EF), recursive
embedding, and natural language develop later (Perner [1991]; Suddendorf
[1999]; Suddendorf and Whiten [2001]; De Villiers and Pyers [2002]; Smith
et al. [2003]). It is their interaction in development, rather than the maturation
of a mindreading module, which explains the emergence of ToM. Thus the
notion of ‘modularity’ of ToM is misleading if taken to mean a discrete
cognitive entity implemented in specialized neural circuitry.
The argument against the presence of a ToM module (or a ToM domainspecific cognitive mechanism) provides a general lesson for the explanation of
cognition and a cautionary tale for evolutionary psychology: modulae non sunt
multiplicanda praeter necessitatem. We develop the argument in five stages:
1. Passing the false-belief test does not necessarily depend on a
domain-specific cognitive mechanism. We suggest an alternative
cognitive architecture that can explain the phenomena that led to
the postulation of the ToM module.
2. Recent evidence from neuroscience does not support the inference to
the presence of a ToM module.
1
If so, we would have identified a cognitive device evolved to solve an adaptive problem (Sterelny
and Griffiths [1999]). Evolutionary psychologists suggest we identity such devices in order to
assist in determining which features of the cognitive phenotype are adaptations (Tooby &
Cosmides [1992]). Atkinson and Wheeler ([2004]), however, point out that the relationship of
explanatory priority between cognitive device and adaptive function is theoretically problematic.
Although many of Atkinson and Wheeler’s arguments are situated within the framework of
evolutionary psychology, rather than developmental psychology and neuroscience, we are
sympathetic to a general line of argument used by critics of naı̈ve adaptationism. The inference
from apparent cognitive structure in the phenotype (in our case a dissociation in performance
on false-belief and false-photo tasks) to the presence of domain-specific cognitive mechanisms
is not straightforward. Just as Samuels ([1998]) proposes an alternative general processing
architecture (the library conception) to argue against the inference from apparent domain
specificity to modularity, we provide a specific alternative architecture to show that the selective
activation in the TPJ in ToM tasks is not evidence for the presence of a domain-specific cognitive
mechanism.
Generous or Parsimonious Cognitive Architecture?
123
3. Our suggested architecture is more parsimonious.
4. We support our case in part by comparison to the case of language,
which many theorists argue is a high-level module, which takes
outputs from domain-specific cognitive subsystems as inputs. The
interest of the case of language is that if those arguments are sound
then we have a case of a module constructed in development by
the type of interaction we think explains ToM. We argue, however,
that the arguments in support of modularity of language do not
apply to the case of ToM. This is particularly important because the
case of language acquisition has served as a general model for the
postulation of domain-specific cognitive architecture (Pinker [1989];
Cosmides and Tooby [1995]; Cosmides and Tooby [1994]; Gelman
and Hirschfeld [1994]).
5. Thus, activation in TPJ revealed in fMRI studies does not reveal the
neural substrate of a ToM module. In fact, TPJ may help implement
a domain-general cognitive capacity, MR. This result would be
consistent with the explanation of ToM proposed by Baron–Cohen,
Leslie and Frith 20 years ago.
Before developing our argument we need to discuss a central distinction
between domain-specific and domain-general cognitive abilities. Domaingeneral abilities include working memory, EF, recursion, attention, and MR.
These capacities process information without being restricted to a particular
representational format. While the visual system, for example, can process only
optical information, attention can be focused on visual, auditory, sensory,
spatial, or higher-order representations produced by any more-specialized
system. The same is true of EF (inhibition, sequencing, flexible control
of attention): any of these processes must operate over a wide variety of
representational formats.
Domain-specific subsystems, such as the visual system, process only their
proprietary inputs using representations specialized for their task and produce outputs in specialized representational formats. It is this specialization
that provides the contrast between domain-general and domain-specific cognitive systems. Indeed, the lack of specialization of domain-general systems is
required in order for them to be able to process the variety of representations
produced by the mind’s domain-specific systems. The existence of domain
specificity necessitates domain generality for any creature that needs to integrate different types of information. The terms domain-specific and modular
are sometimes used interchangeably; however, the notion of a ‘module’ is a
contested one. For example, it has been variously characterized as a restricted
database, a specialized neural mechanism, distributed or localized, an informationally encapsulated cognitive process, a cognitive process vulnerable to
124
Philip Gerrans and Valerie E. Stone
selective deficit, a functionally discrete component of a computational system,
or a content-specific cognitive capacity.2 The advent of fMRI has led most
researchers in the field to adopt one of these conceptions of modularity to
explain the selective activation of different neural circuits observed under
different task conditions: that is, to treat neural activation as implementation
of one of these definitions. It is worth noting that all these characterizations
of ‘module’ attempt to preserve the basic distinction between specialization
and generality while doing justice to the complexity of cognition at different
levels of analysis. We do not wish to adjudicate debates between proponents of
different conceptions because the decision to adopt a definition depends very
much on which aspects of cognition one wants to emphasize.3 For example,
encapsulation theorists need to deal with the large amount of top-down processing, recurrence, and cross-talk in the brain, which means that few systems
are entirely informationally discrete. Localization theorists must address the
distributed nature of cognition, and so on. Each view seems to encounter
some difficulties and counter examples. Nonetheless, despite the problem of
providing a watertight definition, brain research continues (i) to demonstrate
a very high degree of functional specialization down to the molecular level; (ii)
to show that some cognitive functions, such as attention or working memory,
are not restricted in the representations they can take as inputs.
Consequently, we will use the terms domain-specific cognitive mechanism
and domain-general cognitive mechanism to capture this distinction in our
2
3
(Coltheart [1999]) contains a helpful survey of costs and benefits associated with adopting
different conceptions. As he points out, following Shallice ([1988]) neuropsychologists use
acquired lesion to determine functional architecture without being committed to a stronger
notion of anatomical modularity. He also points out that Fodor’s influential discussion did
not propose necessary and sufficient conditions for modularity but a cluster of related features
modules might possess in degrees. Like Pylyshyn ([1999]), however, Fodor does think that
modules are informationally encapsulated from central processes of inference. Samuels ([1998])
helpfully makes a distinction between the algorithmic conception and the hardware conception
of modularity. A laptop is a domain-general piece of hardware, which runs many domainspecific programs (e.g. a word processor and an internet browser). If hardware is domain-specific
then the algorithms running on that hardware must also be domain-specific. This is the intuitive
idea behind the idea that a neurally specialized system (such as vision) is encapsulated from
domain-specific processes advanced in the work of Fodor ([1983]). But it does not follow
from the fact that algorithms are domain-specific (e.g. a graphics program), that hardware
is domain-specific. So the inference from the presence of domain specificity in the cognitive
phenotype to domain specificity in the neural phenotype is invalid.
Atkinson and Wheeler ([2004]) point out that the possibility of different foci of explanation
generates a problem of indeterminacy for the boundaries of modules. Like Samuels, their target
is the use of adaptationist methodology to determine the boundary or domain of a module.
Which of an organism’s features (cognitive or physical) is an adaptation requires us to solve a
simultaneous equation with variables being the adaptive problem and the feature (Sterenly and
Griffiths [2003]). Since both features and adaptive problems decompose into microconstituents
the problem of indeterminacy rapidly ramifies. As they point out, this does not invalidate the
program of linking neural and cognitive specialization, but it requires the theorist to move
‘within and between levels of organization and levels of analysis’ (p. 170). Our tour of this
territory did not reveal any evidence of either hardware or algorithm specialized for ToM.
Generous or Parsimonious Cognitive Architecture?
125
discussion. This is adequate for our purpose, which is to recapture the insight
of the original proposal by Baron-Cohen et al. ([1985]) that ToM is the
application of a domain-general capacity to a specific task.
In their 1985 paper, Baron–Cohen, Leslie & Frith proposed a cognitive
account of ToM which treats it as ‘one of the manifestations of a basic
metarepresentational capacity’ ([1985], p. 37, emphasis added). MR is the
ability to represent the relationship between a representation and its object. As
the quote implies it is not necessarily limited to MR of mental states (Perner
[1991]; Suddendorf [1999]; Suddendorf and Whiten [2001]; Corballis [2003];
Povinelli [1996]) and is thus a domain-general ability. Subsequently, other
developmental theorists pursued the idea that ToM depends on a domainspecific capacity for MR of mental states: the ToM mechanism (e.g. Leslie
[1987], [1994]). Theorists differ over the nature of domain specificity in ToM.
All agree ToM is content specific (its domain is the MR of mental states) but the
nature of its representational format, relationship with other domain-specific
capacities, and neural implementation are controversial.
Evidence for domain specificity in ToM is provided by an apparent
dissociation between domain-general MR and MR of mental states. In
particular, many children with autism who fail the false-belief test pass the
false-photograph test, which requires MR of representational relationships
with nonmental state content, while normal children exhibit a converse pattern
of performance (Zaitchik [1990]).
For domain-specific theorists, ToM is a domain-specific capacity, and
autism results from impairment to the development of a ToM mechanism:
a neurocomputational system specialized for the MR of mental states. The
fact that autism is a developmental rather than an acquired disorder also
suggests to some theorists that the domain-specific mechanism is innate, in the
same way that disorders such as specific language impairment support nativist
inferences about the nature of human language ability. Although the coherence
of nativism and the inference from developmental deficit to innateness have
been contested (Elman et al. [1996]; Karmiloff-Smith [1998]; Karmiloff-Smith
et al. [2003]), those supporting the domain-specific conception of ToM regard
it as a domain-specific cognitive adaptation that depends on the genetically
guided maturation of specialized neural circuitry. It is for this reason that the
near-universal presence of ToM in the human phenotype has been recruited by
Evolutionary Psychology as evidence for domain-specific nativism (Cosmides
& Tooby [1995]).
More recently, cognitive neuroscience, using data from fMRI and lesion
studies, has investigated ToM by locating neural circuitry which subserves it
and attempting to distinguish ToM from other capacities, such as working
memory or more general inferential abilities (Stone et al. [1998]; Saxe &
Kanwisher [2003]; Apperly et al. [2004]; Samson et al. [2004]; Saxe et al. [2004]).
126
Philip Gerrans and Valerie E. Stone
Initial studies focused on the frontal lobes, either medial prefrontal cortex
(MPFC) (Fletcher et al. [1995]; Gallagher and Frith [2003]) or orbitofrontal
cortex (OFC) (Stone et al. [1998]; Gregory et al. [2002]). More recent studies
have implicated the TPJ in ToM performance. The TPJ is of particular interest
because fMRI studies showed it to be active during false-belief tasks but
not false-photograph tasks, suggesting a neural substrate for the behavioral
dissociation that prompted the initial postulation of a ToM module (Saxe and
Kanwisher [2003]; Saxe [2004]).
These convergent lines of inquiry from developmental psychology and neuroscience have persuaded many theorists that ToM is a domain-specific cognitive ability. However, we argue that an alternative model of ToM development
can account more parsimoniously for the data from autism and neuroscience.
Our main concern is to defuse the arguments for domain specificity in
ToM rather than nativism, since the two are logically independent. However,
the inference to nativism from empirical evidence about ToM is less persuasive once the relationships between developmental components of ToM are
clarified.
2 Developmental Components of ToM
Before we consider the evidence from neuroscience let us recall what is in
common between all theorists of ToM development. First, development of
ToM depends on the prior development of a suite of specialized lowerlevel cognitive mechanisms. These precursor mechanisms represent vital
information about the social world of the infant and toddler, and mediate her
earliest interactions with others. These mechanisms enable face processing,
representations of gaze direction, gaze monitoring, detection of animacy,
tracking of intentions and goals, and joint attention (Baron-Cohen [1995];
Dawson et al. [1998]; Crichton and Lange-Kuttner [1999]; Woodward [1999];
Charman et al. [2000]; Hare et al. [2000]; Carpenter et al. [2001]; Csibra et al.
[2003]; Saxe et al. [2004]; Schultz [2005]; Stone [2005]; Rutherford et al. [2006]).
These capacities seem to be specific to social stimuli, are shared with other
primates, and appear to depend on neural circuitry that responds specifically to
social stimuli (e.g. Campbell et al. [1990]; Kumashiro et al. [2002]; Blakemore
et al. [2003]; Csibra et al. [2003]; Povinelli et al. [2003]). Gaze monitoring, for
example, seems to involve specific regions of the superior temporal sulcus that
respond to the stimulus of eye gaze direction but not to other stimuli, even other
stimuli that are physically similar to images of eyes (Campbell et al. [1990];
Perrett et al. [1990]; Hoffman and Haxby [2000]). Assessing others’ goals and
intentions seems to depend on specific representations of certain movement
patterns, limb movement, combined with gaze, head or body orientation, and
Generous or Parsimonious Cognitive Architecture?
127
involves both superior temporal sulcus and superior parietal and lateral frontal
areas (Jellema et al. [2000]; Blakemore et al. [2003]).
The development of these domain-specific mechanisms ensures that the
normal toddler is equipped with a sophisticated battery of mechanisms, which
enable her to negotiate the social world on the basis of perceptually available
information. In this respect, she is like the great apes and uses essentially
the same cognitive equipment. Like the great apes she cannot compute
over abstract representations of unobservable states to predict and explain
conspecific behavior. She cannot metarepresent beliefs (Suddendorf [1999];
Suddendorf and Whiten [2001]).
Thus, one way to describe the issue between domain-specific and domaingeneral accounts of ToM abilities is as a debate about the nature of an
important cognitive difference between the great apes and ourselves. We
argue that homo sapiens’ superior ToM capacity does not depend on the
evolution in the hominid line of a domain-specific ToM mechanism. Our
domain-general capacities in combination with domain-specific precursors,
such as gaze processing and emotion recognition, explain ToM performance.
The presence and nature of these domain-general capacities is another
point of agreement between all theorists. The normal 4-year-old is
equipped with a battery of cognitive devices which subserve domain-general
metacognitive computational functions. Advanced EF (meaning working
memory, inhibition, and flexible control of attention), SRs (the capacity to
hold in mind and compare two different representations), recursion (the
ability to compute embedded representations), and MR (understanding
the representational nature of the relationship between representation and
referent) have all come online by ages 2–4, though they may continue to
develop further after age 4 (Perner [1991]; Suddendorf [1999]; Suddendorf and
Whiten [2001]; De Villiers and Pyers [2002]; Smith et al. [2003]). It is advanced
EF, MR and recursion–all domain-general cognitive abilities–that distance
us from our primate ancestors (Suddendorf [1999]; Suddendorf and Whiten
[2001]), rather than the fact that we possess a domain-specific ToM mechanism
and they do not.
However, our claim is not that presence of ToM in normal children, or
its absence in autism, is entirely accounted for by these metarepresentational
abilities. Toddlers, chimps, and children with Downs syndrome, who lack
metarepresentational abilities in different ways and to different degrees, do
not resemble individuals with autism. Further evidence is the fact that although
solving false-belief tasks clearly places demands on EF, there are no clear early
deficits in EF in autism (Griffith et al. [1999]). Furthermore, apparent deficits
in EF disappear when children with autism are tested by a computer rather
than a person (Ozonoff et al. [1991]; Ozonoff [1995]).
128
Philip Gerrans and Valerie E. Stone
In our view, these results are to be expected since ToM abilities depend, not
on MR, EF, language, or recursion per se, but on their developmental
interaction with the low-level precursors previously described, e.g. gaze
processing, emotion recognition, and joint attention (see Figures 1 and 2 for a
graphic representation of this architecture in comparison to that postulated by
modular explanations of ToM). The result of this interaction is not the creation
of an extra module but the wiring up of distributed metarepresentational
circuitry which can take social information as input (Gerrans [2003]; Stone
[2005]) and, ultimately, compute over abstract representations of mental
states not tied to a perceptual modality. The manipulation of those abstract
representations is not, however, essentially different from the manipulation of
other abstract representations involving recursion, natural language, and MR.
Nor does it need to involve a distinct neurocomputational mechanism.
MR operating on information about eye gaze and attention (who saw or
was attending to what) allows us to represent others’ knowledge states (who
knew what) (Baron-Cohen [1995]). Recursion operating on MRs of mental
states allows us to reason about not just others’ thoughts, but others’ thoughts
about thoughts (Corballis [2003]). Indeed, children’s abilities to use embedded
Figure 1. Mental state inferences with a TOM module.
Generous or Parsimonious Cognitive Architecture?
129
Figure 2. Mental state inferences without a TOM module.
syntactical structures comes online shortly before the ability to solve falsebelief tasks, supporting the view that recursion may be a general ability serving
both tasks (Apperly et al. [2004], [2005]). EF allows us to keep the elements
of a social interaction in mind, and inhibit our own knowledge of the state of
reality when asked what someone else’s mental state is (Stone [2005]). Both
working memory and inhibition have been shown to be related to performance
on ToM tasks (Keenan [1998]; Stone et al. [1998]; Carlson and Moses [2001]).
We note in this regard, that Sabbagh et al. ([2006]) found that 3–5-year-old
performance on an executive task requiring inhibition predicted performance
on false-belief and false-sign tasks but not false-photo tasks. In summary,
deficits on ToM tasks could potentially result from deficits in either low-level
input systems (e.g. joint attention) or higher-level domain-general capacities,
rather than in a separate ToM mechanism.
3 The Analogy with Modularity of Language
Our view seems to face two related problems. The first is that a modular
theorist could point out that maturation of any cognitively sophisticated
module would require developmental interaction between domain-general
130
Philip Gerrans and Valerie E. Stone
cognition and domain-specific precursors. In the case of language, for example,
domain-specific mechanisms such as phonological processing interact with
domain-general ones such as EF, SR, and MR and the consequence of this
interaction is the construction of a language module (Smith [1999]; Saxe
et al. [2004]), which is vulnerable to both developmental and acquired double
dissociations. In fact, the case of language is instructive because when we
compare language and ToM we find that the main arguments for domain
specificity between the two cases are not analogous. Our discussion is brief
because we are conceding the controversial points that language is modular
and that Specific Language Impairment is a developmental dissociation of that
module. We merely point out that arguments for those conclusions are much
less persuasive in the case of ToM.
First, Poverty of Stimulus arguments for modularity of language rely on
the fact that the evidence available to children in their linguistic environment
(primary linguistic data) underdetermines possible grammars. In order to
explain the near-universal convergence of developmental trajectories across
the human cognitive phenotype we need to postulate specialized cognitive
architecture (Chomsky [1986]). The inference to ToM from the child’s
experience, however, is not similarly unconstrained. The child’s domainspecific precursors to ToM are a rich body of evidence for the hypothesis that
other people act on the basis of intentional states. The child does not need a
special neurocomputational device to generate and test that hypothesis, she
merely needs to apply her domain-general capacities for hypothesis formation
to the ‘primary psychological data’ in which she is immersed.
Second, the theoretical argument for language modularity is empirically
reinforced by the presence of acquired dissociations, such as aphasia, in which
syntactic processing is impaired while both domain-specific precursors to
language and domain-general mechanisms remain intact. However, there are
no equivalent cases of acquired deficits to ToM that spare other cognitive
abilities (see Section 5 for a full discussion of recent neuroscientific work
on this issue). If ToM is a genuine domain-specific mechanism, this lack of
empirical evidence is surprising.
Third, the modular theorist of language can point to cases of developmental
double dissociation of the capacity for syntactic processing. Smith and
Ianthi–Simpli ([1995]) report the case of a quite severely cognitively impaired
man who, nonetheless, has an amazing facility with the syntactic aspects of over
twenty languages. This man’s facility with syntax could be explained by the
sparing of his language module. His case, however, is only a single dissociation,
of spared syntactical abilities from impaired domain-general cognition. Double
dissociations are more convincing arguments for modularity, and the reverse
dissociation is in fact known. There are cases of selective impairment to
syntactic processing with sparing of other cognitive capacities: the disorder
Generous or Parsimonious Cognitive Architecture?
131
known as Grammatical Specific Language Impairment is interpreted in that
way. (Gopnik and Goad [1997]; Van der Lely and Stollwerk [1997]).
Whether autism is evidence for the presence of a selective developmental
single dissociation in a ToM module is still at issue, as discussed below.
However, there is no evidence for a double dissociation, the selective sparing in
development of ToM. Researchers were initially misled by the hypersociability
and expressiveness of children with Williams syndrome into thinking that
they had a spared ToM (Karmiloff-Smith et al. [1995]) However, subsequent
studies have shown that their ToM also develops abnormally, as do their
language abilities (Tager-Flusberg [2000]; Gerrans [2002]). In our view, this
connection results from the fact that language and ToM both depend on
some of the same domain-general capacities, MR and recursion, which may
be impaired in Williams syndrome.
Thus, the theoretical and empirical considerations which support modularity
theses about language do not apply straightforwardly to ToM. The empirical
evidence in it is simply not there, as discussed further below, and the theoretical
arguments (poverty of the stimulus) are difficult to uphold in light of the social
information children are exposed to.
4 Dissociations without Modules
However, even if proponents of domain specificity in ToM accepted these
points, they would press a second difficulty for our account. We still
need to explain dissociations between performance on false-belief and falsephotograph tasks. This difficulty seems especially pressing given that attempts
to explain ToM deficits in terms of deficits in EF or MR have not been entirely
successful. Attempts to explain autism in terms of EF have foundered on a
lack of early EF deficits in autism, and attempts to explain it as a general
metarepresentational deficit have foundered on autistic children’s excellent
performance on false-photograph tasks.
Our reply to domain-specific theorists of ToM is that if children with
autism have intact MR, but impaired inputs to MR from impaired lowlevel domain-specific mechanisms (e.g. joint attention), then one would
predict exactly this dissociation between false-photograph and false-belief
performance. Individuals with autism perform well on the false-photograph
task because the inputs to the metarepresentational system, in that task, are
not impaired in autism. See Figure 3 for a graphic representation.
We believe that deficits on ToM tasks in autism arise essentially from deficits
in lower-level input systems. Research on social deficits in autism has shown
clearly that children with autism have deficits in many early domain-specific
social competences: face recognition, facial expression recognition, processing
of gaze direction, and joint attention; (Osterling and Dawson [1994]; Boucher
132
Philip Gerrans and Valerie E. Stone
Figure 3. False photo inferences.
and Lewis [1992a]; Dawson et al. [1998], [2004a], [2004b]; Klin et al. [1999];
Baird et al. [2000]; Grice et al. [2005]; Schultz [2005]). The performance of
children with autism on ToM tests can be explained more parsimoniously by
the view that they have deficits, not in MR, but in lower-level domain-specific
mechanisms for processing social information. Without proper inputs, the
intact capacity for MR by itself cannot make correct ToM inferences with the
same fluency and automaticity as the mind of the normal 4-year-old. Thus, we
argue that autistic deficits on ToM tasks are evidence, not for impaired MR,
but are instead evidence of impaired inputs to metarepresentational systems
recruited by ToM tasks. Strong evidence against this view would be a case of
an individual with specific deficits on metarepresentational ToM tasks, such
as false belief, but no lower-level social deficits, but as we noted, no such case
has been reported.
In our view, then, there should be cases of high-functioning autism spectrum
disorder (ASD) individuals in whom MR is intact. Evidence supports this view.
In the absence of comorbid intellectual disability, individuals with autism seem
to have intact capacities for MR, recursion and EF. Children with autism
perform well on a nonmental metarepresentational task, the false-photograph
test (Leslie and Thaiss [1992]). Baron–Cohen et al. ([1999]) report three cases
Generous or Parsimonious Cognitive Architecture?
133
of high-functioning individuals with ASD, a mathematician, a physicist, and an
engineering student. All three fields require people to represent the relations
between symbols and their objects (magnitudes, abstract spatial relations)
and to perform recursive computations over these symbols. These three men
excelled in their fields (in fact, the mathematician was a Fields medalist),
showing that their capacity for MR and recursion was intact. Furthermore,
all did well on the Tower of Hanoi, an EF test that requires not only
EF, but also recursion. However, all three had difficulty in inferring what
someone was feeling, or paying attention to, from pictures of the eye region
of the face (Baron–Cohen et al. [1999]), indicating a problem with lower-level
domain-specific capacities for face and gaze processing rather than MR.
The weight of evidence indicates that the ToM impairments in autism and
related disorders are a result of abnormal developmental interaction between
low-level input systems and higher-level domain-general capacities, caused
essentially by deficits in ToM precursors. An individual with a deficit in ToM
MR, without any deficit in lower-level social competences and without any
more general deficit in other relevant domain-general competences would
provide conclusive evidence against our view. No such case has yet been
invoked in support of the domain-specific ToM hypothesis. Yet, as we
remarked earlier, such cases of selective deficits are the core evidence for
modularity hypotheses in other areas, such as language. We note in passing
that our view of the evidence is more parsimonious. The domain-specific ToM
theorist requires that MR emerges twice, once for mental states and once for
domain-general computation, in both evolution and development.
5 The Evidence from Neuroscience
If children with autism do not provide developmental evidence for domain
specificity in ToM, then perhaps data from neurological patients with lesions
would provide stronger evidence. These might provide evidence for selective
acquired deficits to an existing domain-specific mechanism. In contrast, on
our view, if deficits on ToM tasks result from deficits in higher-level domaingeneral capacities, rather than in a separate domain-specific mechanism, we
would expect ToM deficits to be accompanied by deficits in other tasks
requiring domain-general metacognition. Our prediction is vindicated by
existing evidence from neuropsychology. Current research seems only to
support the view that it is not possible to be impaired on ToM tasks without
a concurrent deficit in either lower-level social abilities or deficits in other
domain-general abilities.
Patients with OFC damage, for example, have been found to be impaired
on ToM tasks (Stone et al. [1998]; Gregory et al. [2002]). However, they are
also impaired at recognizing facial expressions, judging mental states from eye
134
Philip Gerrans and Valerie E. Stone
gaze or expression in the eye region of the face (Hornak et al. [1996]; Gregory
et al. [2002]; Snowden et al. [2003]). Furthermore, even their problems with
some ToM tasks may result from a difficulty in tracking intentions rather than
beliefs (Stone [2000], [2005]). Thus, their mentalizing deficits can be explained
as a result of deficits in lower-level domain-specific mechanisms, not in higherorder domain-general mechanisms. Indeed, they can perform at ceiling-levels
on false-belief tasks, even those requiring 2nd- and 3rd-order mental state
inferences (Stone et al. [1998]; Stone [2005]).
Early neuroimaging studies of ToM implicated the MPFC as an area needed
for the MR of mental states. If MPFC is necessary for ToM, then patients
with damage in that area should show specific deficits. However, patients with
medial frontal damage who have ToM deficits all have accompanying EF
deficits (Happé et al. [2001]; Gregory et al. [2002]; Stone [2005]). Furthermore,
extensive medial frontal damage does not necessarily cause impairment in
ToM (Bird et al. [2004]). Hence, while MPFC is recruited during ToM tasks
its contribution does not seem to be specific to ToM inferences.
Perhaps, then, more recent fMRI studies implicating the TPJ in the
attribution of mental states might provide the basis for the ToM module?
Saxe and Kanwisher ([2003]) found that the TPJ is more active in false-belief
tasks than in tasks requiring the attribution of intention or social information.
They also found that the TPJ was less strongly activated in false-photo tasks
than in false-belief tasks, which they say suggests a crucial role for the TPJ
in ToM inferences (Saxe and Kanwisher [2003]). However, in discussing their
results, they also noted that while the false-photo tasks were ‘logically similar’
to the ToM tasks they were also ‘more difficult’ (p. 1840). This raises the
possibility that the differential activation in TPJ is a resource artefact within
MR. That is to say, perhaps the difference in difficulty of the tasks accounts
for the difference in activation.
A possible explanation for the difference in difficulty is that the
TPJ, although a domain-general metarepresentational system, is constantly
receiving social information from the early domain-specific social input systems
we identified. Thus, the baseline level of activation for metarepresenting social
information is already high. Any new, nonsocial, task needs either to bring TPJ
activation up to the same baseline or to inhibit the social input systems to allow
the nonsocial information to command TPJ processing resources. This could
explain differences in activation levels in TPJ produced by false-belief and
false-photograph tasks, and explain why Saxe and Kanwisher’s participants
took longer to do false-photograph than false-belief tasks.
Thus, more decisive evidence for TPJ specialization for ToM would
be provided by false-belief and false-photo studies, which posed identical
executive and cognitive resource demands. By using better-matched tasks in a
neuroimaging paradigm, this issue could be resolved. The issue could also be
Generous or Parsimonious Cognitive Architecture?
135
approached by further analyses of current imaging data. Saxe and Kanwisher
([2003]) present contrasts between false-belief and false-photograph trials, and
between false-belief task and nonmetarepresentational-task trials. They do not,
however, present any whole-brain contrasts for all participants between falsephotograph task and nonmetarepresentational-task trials. If such a contrast
did show differential activation in TPJ, then this would support the view
that TPJ might underwrite MR more generally. In that case, the differential
activation in TPJ between false-belief and false-photograph trials might simply
reflect how much each task demands of a metarepresentational capacity,
though both require MR. In contrast, if TPJ was not differentially active when
nonmetarepresentational tasks are subtracted from false-photograph tasks,
there would be stronger evidence for TPJ’s specialization for belief state MR.
In the absence of imaging studies of false-belief and false-photograph tasks
that control for executive demands, we can turn to recent work with TPJ lesion
patients (Apperly et al. [2004]; Samson et al. [2004]). The hypothesis that TPJ is
the site of the ToM mechanism suggests that TPJ patients should have specific
ToM deficits. Indeed, initial research with TPJ patients showed deficits on
false-belief tasks and not other EF tasks even when the general demands of EF
were controlled for (Apperly et al. [2004]; Samson et al. [2004]). However, the
most recent research shows that all TPJ lesion patients who performed below
chance on false-belief tasks also performed poorly on a false-photograph test
(Apperly et al., [2007]). Thus, there is as yet no conclusive evidence from
neurological patients that supports the claim that the TPJ is the site of a
domain-specific mechanism dedicated to ToM inferences.
In fact, Apperly et al.’s data suggests that, although TPJ does not appear
to be specialized for ToM, it may be specialized for MR and recruited by
metarepresentational aspects of ToM tasks.
6 Conclusion
As we noted earlier, MR is one of the abilities that emerged over the course
of hominid evolution that separates humans from the great apes, (Suddendorf
[1999]; Corballis [2003]). In our view, it is not surprising that our ToM abilities
emerge in concert with the suite of other abilities which distance us from
the apes, all of which are essentially dependent on the architecture of the
prefrontal cortex and temporal lobes. The fact that domain-general cognition,
which allows us to detach from perceptual stimuli and represents unobservable
facts, is largely occupied in the first few years with social interaction is a natural
consequence of our altricial developmental trajectory.
In the absence of strong evidence for dedicated neural circuitry or the dissociation of ToM from metarepresentational or lower-level deficits, it seems that
136
Philip Gerrans and Valerie E. Stone
the interaction of several domain-general mechanisms and lower-level domainspecific mechanisms can account for the flexibility and sophistication of behavior which has been taken to be evidence for a domain-specific ToM module.
One can understand, as an episode in the history of science, why it seemed
necessary to posit a specific ToM module. Early evidence from autism seemed
to show a pattern of impairment in MR of belief, without a corresponding
impairment in MR in other domains (Baron-Cohen et al. [1985], Leslie and
Thaiss [1992]). Only later did it become clear that the deficits in autism were at
a lower level. Furthermore, early explorations of the role of MR understood as
a general prerequisite for higher cognition (Perner [1991]; Suddendorf [1999];
Corballis [2003]) were overlooked once an apparent dissociation between ToM
and domain-general MR was demonstrated (Leslie and Thaiss [1992]).
However, postulating a domain-specific ToM mechanism to explain autistic
deficits appears to have been an unnecessary elaboration on an already
adequate basic theoretical structure: i.e one containing domain-specific input
systems like shared attention and domain-general systems such as EF, MR,
SR, and language, for higher-order cognition. We believe that this episode in
the history of cognitive science can serve as a cautionary tale to evolutionary
psychologists and modular theorists: it is important, in proposing a certain
cognitive architecture, to make sure that one has not postulated more modules
than are warranted by the data: modulae non sunt multiplicanda praeter
necessitatem. The data for language justifies the postulation of a separate
module–not so for ToM. Just as the heliocentric theory of celestial motion
removed the need for epicycles in explanations of planetary orbits, a proper
understanding of the role of MR allows us to see past the limitations of
modulocentric theorising about ToM.
Philip Gerrans
Department of Philosophy
University of Adelaide
[email protected]
Valerie E. Stone
Department of Psychology
University of Queensland
[email protected]
References
Apperly, I. A., Samson, D., Chiavarino, C. and Humphreys, G. W. [2004]: ‘Frontal
and Temporo-Parietal Lobe Contributions to Theory of Mind: Neuropsychological
Evidence from a False-belief Task with Reduced Language and Executive Demands’,
Journal of Cognitive Neuroscience, 16, pp. 1773–84.
Generous or Parsimonious Cognitive Architecture?
137
Apperly, I. A., Samson, D., Chiavarino, C., Bickerton, W. and Humphreys, G. W.
[2005]: ‘Testing the Domain-specificity of a Theory of Mind Deficit in Brain-injured
Patients: Evidence from an Improved False Photograph Task with Reduced Language
and Executive Demands’, Cognitive Neuroscience Society Abstracts, B189, p. 71.
Apperly, I. A., Samson, D., Chiavarino, C., Bickerton, W-L. and Humphreys, G. W.
[2007]: ‘Testing the domain-specificity of a theory of mind deficit in brain-injured
patients: Evidence for consistent performance on non-verbal, ‘‘reality-unknown’’
false belief and false photograph tasks’, Cognition, 103(2), pp. 300–21.
Atkinson, A. P. and Wheeler, M. [2004]: ‘The Grain of Domains: The Evolutionarypsychological Case Against Domain-general Cognition’, Mind and Language, 19,
pp. 147–76.
Baird, G., Charman, T., Baron-Cohen, S., Cox, A., Swettenham, J., Wheelwright, S.
and Drew, A. [2000]: ‘A Screening Instrument for Autism at 18 Months of Age: A
6-year Follow-up Study’, Journal of the American Academy of Child and Adolescent
Psychiatry, 9, pp. 694–702.
Baron-Cohen, S. [1995]: Mindblindness: An Essay on Autism and Theory of Mind,
Cambridge, MA: MIT Press.
Baron-Cohen, S., Leslie, A. and Frith, U. [1985]: ‘Does the Autistic Child have a
‘‘Theory of Mind’’’? Cognition, 21, pp. 37–46.
Baron-Cohen, S., Wheelwright, S., Stone, V. E. and Rutherford, M. [1999]: ‘A
Mathematician, a Physicist, and a Computer Scientist with Asperger Syndrome:
Performance on Folk Psychology and Folk Physics Tests’, Neurocase, 5, pp. 475–83.
Bird, C. M., Castelli, F., Malik, O., Frith, U. and Husain, M. [2004]: ‘The Impact
of Extensive Medial Frontal Lobe Damage on ‘‘Theory of Mind’’ and Cognition’,
Brain, 127, pp. 914–28.
Blakemore, S. J., Boyer, P., Pachot-Clouchard, M., Meltzoff, A., Segebarth, C.
and Decety, J. [2003]: ‘The Detection of Contingency and Animacy from Simple
Animations in the Human Brain’, Cerebral Cortex, 13, pp. 837–44.
Boucher, J. and Lewis, V. [1992a]: ‘Unfamiliar Face Recognition in Relatively Able
Autistic Children’, Journal of Cognitive Neuroscience, 8, pp. 551–65.
Boucher, J. and Lewis, V. [1992b]: ‘Unfamiliar Face Recognition in Relatively Able
Autistic Children’, Journal of Child Psychology and Psychiatry, 33, pp. 843–59.
Campbell, R., Heywood, C., Cowey, A., Regard, M. and Landis, T. [1990]: ‘Sensitivity
to Eye Gaze in Prosopagnosic Patients and Monkeys with Superior Temporal Sulcus
Ablation’, Neuropsychologia, 28, pp. 1123–42.
Carlson, S. and Moses, L. [2001]: ‘Individual Differences in Inhibitory Control and
Children’s Theory of Mind’, Child Development, 72, pp. 1032–53.
Carpenter, M., Pennington, B. F. and Rogers, S. J. [2001]: ‘Understanding of Others’
Intentions in Children with Autism’, Journal of Autism and Developmental Disorders,
31, pp. 589–99.
Charman, T., Baron-Cohen, S., Swettenham, J., Baird, G., Cox, A. and Drew, A. [2000]:
‘Testing Joint Attention, Imitation and Play as Infancy Precursors to Language and
Theory of Mind’, Cognitive Development, 15, pp. 481–98.
Chomsky, N. [1986]: Knowledge of Language: Its Nature, Origin and Use, New York:
Praeger.
138
Philip Gerrans and Valerie E. Stone
Coltheart, M. [1999]: ‘Modularity and Cognition’, Trends in Cognitive Sciences, 3,
pp. 115–20.
Corballis, M. [2003]: ‘Recursion as the key to the human mind’, in K. Sterelny and J.
Fitness (eds), 2003, From Mating to Mentality: Evaluating Evolutionary Psychology,
New York: Psychology Press pp. 155–71.
Cosmides, L. and Tooby, J. [1992]: ‘Cognitive adaptations for social exchange’, in
J. Barkow, L. Cosmides, and J. Tooby (eds) The adapted mind, New York: Oxford
University Press.
Cosmides, L. and Tooby, J. [1994]: ‘Beyond Intuition and Instinct Blindness: Towards
an Evolutionarily Rigorous Cognitive Science’, Cognition, 58, pp. 1–73.
Cosmides, L. and Tooby, J. [1995]: ‘Foreword’, in S. Baron-Cohen (ed.), 1995,
Mindblindness: An Essay on Autism and Theory of Mind, Cambridge, Mass: MIT
Press.
Crichton, M. and Lange-Kuttner, C. [1999]: ‘Animacy and Propulsion in Infancy:
Tracking, Waving, and Reaching to Self-propelled and Induced Moving Objects’,
Developmental Science, 2, pp. 318–24.
Csibra, G., Biro, S., Koos, O. and Gergely, G. [2003]: ‘One-year-old Infants
Use Teleological Representations of Actions Productively’, Cognitive Science, 27,
pp. 111–33.
Dawson, G., Meltzoff, A., Osterling, J. and Brown, E. [1998]: ‘Children with Autism
Fail to Orient to Social Stimuli’, Journal of Autism and Developmental Disorders, 28,
pp. 479–85.
Dawson, G., Toth, K., Abbott, R., Osterling, J., Munson, J. and Estes, A. [2004a]:
‘Defining the Early Social Attention Impairments in Autism: Social Orienting, Joint
Attention and Responses to Emotions’, Developmental Psychology, 40, pp. 271–83.
Dawson, G., Webb, S. J., Carver, L., Panagiotides, H. and McPartland, J. [2004b]:
‘Young Children with Autism Show Atypical Brain Responses to Fearful Versus
Neutral Facial Expressions of Emotion’, Developmental Science, 7, pp. 340–59.
De Villiers, J. and Pyers, J. [2002]: ‘Complements to Cognition: A Longitudinal Study
of the Relationship between Complex Syntax and False Belief Understanding’,
Cognitive Development, 17, pp. 1037–60.
Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D. and Plunkett,
K. [1996]: Rethinking Innateness: A Connectionist Perspective on Development,
Cambridge, MA: MIT Press.
Fletcher, P. C., Happé, F., Frith, U., Baker, S. C., Dolan, R. J., Frackowiak, R. S.,
et al. [1995]: ‘Other Minds in the Brain: A Functional Imaging Study of ‘‘Theory of
Mind’’ in Story Comprehension’, Cognition, 57, pp. 109–28.
Fodor, J. [1983]: The Modularity of Mind, Cambridge, Mass: MIT Press.
Gallagher, H. L. and Frith, C. D. [2003]: ‘Functional Imaging of ‘‘Theory of Mind’’’,
Trends in Cognitive Science, 7, pp. 77–83.
Gelman, S. A. and Hirschfeld, L. A. (eds). [1994]: Mapping the Mind: Domain Specificity
in Cognition and Culture, New York: Cambridge University Press.
Gerrans, P. [2002]: ‘Nativism and Neuroconstructivism in the Explanation of Williams
Syndrome’, Biology and Philosophy, 17, pp. 1–11.
Generous or Parsimonious Cognitive Architecture?
139
Gerrans, P. [2003]: ‘The Theory of Mind Module in Evolutionary Psychology’, Biology
and Philosophy, 17, pp. 305–21.
Gopnik, M. and Goad, H. [1997]: ‘What Underlies Inflectional Errors in SLI’, Journal
of Neurolinguistics, 10, pp. 109–237.
Gregory, C., Lough, S., Stone, V. E., Erzinclioglu, S., Martin, L., and Baron-Cohen,
S. [2002]: ‘Theory of Mind in Frontotemporal Dementia and Alzheimer’s Disease:
Theoretical and Practical Implications’, Brain, 125, pp. 752–64.
Grice, S., Halit, H., Farroni, T., Baron-Cohen, S., Bolton, P. and Johnson, M. [2005]:
‘Neural Correlates of Eye-gaze Detection in Young Children with Autism’, Cortex,
41, pp. 342–53.
Griffith, E. M., Pennington, B. F., Wehner, E. A. and Rogers, S. [1999]: ‘Executive
Functions in Young Children with Autism’, Child Development, 70, pp. 817–32.
Happé, F., Mahli, G. S. and Checkley, S. [2001]: ‘Acquired Mind-blindness Following
Frontal Lobe Surgery. A Single Case Study of Impaired ‘‘Theory of Mind’’ in
a Patient Treated with Stereotactic Anterior Capsulotomy’, Neuropsychologia, 39,
pp. 83–90.
Hare, B., Call, J., Agnetta, B. and Tomasello, M. [2000]: ‘Chimpanzees Know What
Conspecifics Do and Do not See’, Animal Behaviour, 59, pp. 771–85.
Hoffman, E. A. and Haxby, J. V. [2000]: ‘Distinct Representations of Eye Gaze and
Identity in the Distributed Human Neural System for Face Perception’, Nature
Neuroscience, 3, pp. 80–4.
Hornak, J., Rolls, E. and Wade, D. [1996]: ‘A Differential Neural Response in the
Human Amygdala to Fearful and Happy Faces’, Nature, 383, pp. 812–5.
Jellema, T., Baker, C. I., Wicker, B. and Perrett, D. I. [2000]: ‘Neural Representation
for the Perception of the Intentionality of Actions’, Brain & Cognition, Special Issue:
Cognitive neuroscience of actions, 44, pp. 280–302.
Karmiloff-Smith, A. [1998]: ‘Development Itself is the Key to Understanding
Developmental Disorders’, Trends in Cognitive Science, 2, pp. 389–983.
Karmiloff-Smith, A., Scerif, G. and Ansari, D. [2003]: ‘Double Dissociations
in Developmental Disorders. Theoretically Misconceived, Empirically Dubious’,
Cortex, 39, pp. 161–3.
Karmiloff-Smith, A., Klima, E., Bellugi, U., Grant, J. and Baron-Cohen, S. [1995]:
‘Is There a Social Module? Language, Face Processing, and Theory of Mind
in Individuals with Williams Syndrome’, Journal of Cognitive Neuroscience, 7,
pp. 196–208.
Keenan, T. [1998]: ‘Memory Span as a Predictor of False Belief Understanding’, New
Zealand Journal of Psychology, 27, pp. 36–43.
Klin, A., Sparrow, S., de Bildt, A., Cicchetti, D., Cohen, D. and Volkmar, F. [1999]:
‘A Normed Study of Face Recognition in Autism and Related Disorders’, Journal of
Autism and Developmental Disorders, 29, pp. 499–508.
Kumashiro, M., Ishibashi, H., Itakura, S. and Iriki, A. [2002]: ‘Bidirectional
Communication Between a Japanese Monkey and a Human Through Eye Gaze
and Pointing’, Current Psychology of Cognition, 21, pp. 3–32.
Leslie, A. M. [1987]: ‘Pretence and Representation: The Origins of ‘‘Theory of Mind’’’,
Psychological Review, 94, pp. 412–26.
140
Philip Gerrans and Valerie E. Stone
Leslie, A. M. and Thaiss, L. [1992]: ‘Domain Specificity in Conceptual Development:
Neuropsychological Evidence from Autism’, Cognition, 43, pp. 225–51.
Leslie, A. M. [1994]: ‘Pretending and believing: issues in the theory of ToM’, Cognition,
50, pp. 211–38.
Osterling, J. and Dawson, G. [1994]: ‘Early recognition of children with autism: A study
of first birthday home videotapes’, Journal of Autism and Developmental Disorders,
24, pp. 247–57.
Ozonoff, S. [1995]: ‘Reliability and Validity of the Wisconsin Card Sorting Test in
Studies of Autism’, Neuropsychology, 9, pp. 491–500.
Ozonoff, S., Pennington, B. F. and Rogers, S. J. [1991]: ‘Executive Function Deficits in
High-functioning Autistic Individuals: Relationship to Theory of Mind’, Journal of
Child Psychology and Psychiatry, 32, pp. 1081–105.
Perner, J. [1991]: The Representational Theory of Mind, Cambridge, Massachusetts:
MIT Press.
Perrett, D., Harries, M., Mistlin, A., Hietanen, J., Benson, P., Bevan, R., Thomas,
S., Oram, M., Ortega, J. and Brierley, K. [1990]: ‘Social Signals Analyzed at the
Single Cell Level: Someone is Looking At Me, Something Touched Me, Something
Moved!’, International Journal of Comparative Psychology, 4, pp. 25–55.
Pinker, S. [1989]: Learnability and Cognition: The acquisition of Argument Structure,
Cambridge, Mass: MIT Press.
Povinelli, D. [1996]: ‘Chimpanzee theory of mind? The long road to strong inference’,
in P. Carruthers and P. Smith (eds), 1996, Theories of Theories of Mind, Cambridge:
Cambridge University Press.
Povinelli, D. J. and Vonk, J. [2003]: ‘Chimpanzee minds: suspiciously human?’, Trends
in Cognitive Sciences, 7(4), pp. 157–60.
Pylyshyn, Z. [1999]: ‘Is Vision Continuous with Cognition?’, Behavioral and Brain
Sciences, 22, pp. 341–65.
Rutherford, M. D., Pennington, B. F., and Rogers, S. J. [2006]: ‘The Perception of
Animacy in Young Children with Autism’, Journal of Autism and Developmental
Disorders, 36, pp. 983–92.
Sabbagh, M., Shiverick, S. and Mosels, L. [2006]: ‘Executive Functioning and
Preschoolers’ Understanding of False Beliefs, False Photographs and False Signs’,
Child Development, 77, pp. 1034–49.
Samson, D., Apperly, I. A., Chiavarino, C. and Humphreys, G. W. [2004]: ‘The Left
Temporoparietal Junction is Necessary for Representing Someone Else’s Belief’,
Nature Neuroscience, 7, pp. 499–500.
Samuels, R. [1998]: ‘Evolutionary Psychology and The Massive Modularity
Hypothesis’, British Journal for the Philosophy of Science, 49, pp. 575–602.
Saxe, R. [2004]: ‘On Human Nature: Reading Your Mind’, Boston Review, Feb/March
2004, available online at <bostonreview.net/BR29.1/saxe.html>.
Saxe, R. and Kanwisher, N. [2003]: ‘People Thinking About Thinking People: The
Role of the Temporo-parietal Junction in ‘‘Theory of Mind’’’, Neuroimage, 19,
pp. 1835–42.
Generous or Parsimonious Cognitive Architecture?
141
Saxe, R., Carey, S. and Kanwisher, N. [2004]: ‘Understanding Other Minds: Linking
Developmental Psychology and Functional Neuroimaging’, Annual Review of
Psychology, 55, pp. 87–124.
Schultz, R. T. [2005]: ‘Developmental Deficits in Social Perception in Autism: The Role
of the Amygdala and Fusiform Face Area’, International Journal of Developmental
Neuroscience, 23, pp. 125–41.
Shallice, T. [1988]: From Neuropsychology to Mental Structure, New York: Cambridge
University Press.
Smith, N. and Ianthi-Simpli, M. [1995]: The Mind of a Savant: Language learning and
Modularity, Oxford: Blackwell.
Smith, N. [1999]: Chomsky, Ideas and Ideals, Cambridge: Cambridge University Press.
Smith, M., Apperly, I. and White, V. [2003]: ‘False Belief Reasoning and the Acquisition
of Relative Clause Sentences’, Child Development, 74, pp. 1709–19.
Snowden, J. S., Gibbons, Z., Blackshaw, A., Doubleday, E., Thompson, J., Craufurd,
D., Foster, J., Happé, F. and Neary, D. [2003]: ‘Social Cognition in Frontotemporal
Dementia and Huntington’s Disease’, Neuropsychologia, 41, pp. 688–701.
Sterelny, K. [2003]: Thought in a Hostile World, Oxford: Blackwell.
Sterelny, K. and Griffiths, P. E. [1999]: Sex and Death: An Introduction to the Philosophy
of Biology, Chicago: University of Chicago Press.
Stone, V. E. [2000]: ‘The role of the frontal lobes and the amygdala in theory of mind’,
in S. Baron-Cohen, D. Cohen and H. Tager-Flusberg (eds), Understanding Other
Minds: Perspectives from Autism and Developmental Cognitive Neuroscience, Oxford:
Oxford University Press.
Stone, V. E. [2005]: ‘Theory of mind and the evolution of social intelligence’, in J.
Cacciopo (ed.), 2005, Social Neuroscience: People Thinking About People, Cambridge,
MA: MIT Press.
Stone, V. E., Baron-Cohen, S. and Knight, R. T. [1998]: ‘Frontal Lobe Contributions
to Theory of Mind’, Journal of Cognitive Neuroscience, 10, pp. 640–56.
Suddendorf, T. [1999]: ‘The rise of the metamind’, in M. C. Corballis and S. Lea
(eds), 1999, The Descent of Mind: Psychological Perspectives on Hominid Evolution,
London: Oxford University Press, pp. 218–60.
Suddendorf, T. and Whiten, A. [2001]: ‘Mental Evolution and Development: Evidence
for Secondary Representation in Children, Great Apes and Other Animals’,
Psychological Bulletin, 127, pp. 629–50.
Tager-Flusberg, H. [2000]: ‘A Componential View of the Mind: Evidence from Williams
Syndrome’, Cognition, 76, pp. 59–89.
Van der Lely, H. and Stollwerk, L. [1997]: ‘Binding Theory and Grammatical Specific
Language Impairment in Children’, Cognition, 62, pp. 245–90.
Wellman, H., Cross, D. and Watson, J. [2001]: ‘Meta-analysis of Theory-of-mind
Development: The Truth About False Belief’, Child Development, 72, pp. 655–84.
Woodward, A. [1999]: ‘Infants’ Ability to Distinguish between Purposeful and Nonpurposeful Behaviours’, Infant Behavior and Development, 22, pp. 145–60.
Zaitchik, D. [1990]: ‘When Representations Conflict with Reality: The Preschooler’s
Problem with False Beliefs and ‘‘False’’ Photographs’, Cognition, 35, pp. 41–68.