What’s domain-specific about theory of mind?
Valerie E. Stone
University of Queensland, St Lucia, Australia
Philip Gerrans
University of Adelaide, Adelaide, Australia
Twenty years ago, Baron-Cohen and colleagues argued that autistic performance on false belief tests was
explained by a deficit in metarepresentation. Subsequent research moved from the view that the mind
has a domain-general capacity for metarepresentation to the view that the mind has a domain-specific
mechanism for metarepresentation of mental states per se, i.e., the theory of mind mechanism (ToMM).
We argue that 20 years of data collection in lesion patients and children with autism supports a more
parsimonious view closer to that of the 1985 paper. Lower-level domain-specific mechanisms*e.g.,
tracking gaze, joint attention*interacting with higher-level domain-general mechanisms for metarepresentation,
recursion, and executive function can account for observed patterns of deficits in both
autism and neurological patients. The performance of children with autism or orbitofrontal patients on
ToM tests can be explained more parsimoniously by their deficits in lower-level domain-specific
mechanisms for processing social information. Without proper inputs, the intact capacity for
metarepresentation by itself cannot make correct ToM inferences. Children with autism have no
impairment in false photograph tests because their metarepresentational capacity is intact and they have
no impairment in inputs required for such tests. TPJ patients have equivalent deficits on ToM and nonToM
metarepresentational tasks, consistent with a failure in domain-general processing. If deficits on
ToM tasks can result from deficits in low-level input systems or in higher-level domain-general capacities,
postulating a separate ToM mechanism may have been an unnecessary theoretical move.
In 1985, Simon Baron-Cohen, Alan Leslie, and
Uta Frith published an influential paper entitled
‘‘Does the autistic child have a theory of mind?’’
They argued that a child who lacked a capacity
for metarepresentation would be unable to apply
it to the developmentally vital task of understanding
the representational nature of other’s
thoughts, and that the abnormal developmental
trajectory of autism would result. They presented
evidence that children with autism were more
likely to fail ‘‘false belief’’ tests, which require the
metarepresentation of mental states, than were
typically developing children. These findings have
been replicated and extended over the past two
decades (Wellman, Cross, & Watson, 2001). Work
on autism in general has expanded greatly over
the same time, with many accounts of the core
deficit in autism emerging as alternatives to the
theory of mind account: deficits in executive
function, emotion and empathy, central coherence,
attentional effects following from cerebellar
abnormalities (e.g., Baron-Cohen, 2004; BaronCohen
& Wheelwright, 2004; Courchesne et al.,
1994; Happe´, 1999; Ozonoff, Pennington, &
Rogers, 1991; Schultz, 2005). As a result, the
empirical database on early deficits in autism is
much richer than it was two decades ago, and we
believe it is time to revisit the question of
domain-specificity in theory of mind in light of
new evidence.
Correspondence should be addressed to: Valerie E. Stone, School of Psychology, McElwain Bldg., University of Queensland,
St Lucia, QLD 4072, Australia. E-mail: stone@psy.uq.edu.au
# 2006 Psychology Press, an imprint of the Taylor & Francis Group, an informa business
SOCIAL NEUROSCIENCE, 2006, 1 (3Á4), 309Á319
www.psypress.com/socialneuroscience DOI:10.1080/17470910601029221
The original paper by Baron-Cohen et al. says
specifically that theory of mind (ToM) is ‘‘one of
the manifestations of a basic metarepresentational
capacity’’ (1985, p. 37, emphasis added).
‘‘Metarepresentation’’ is used in the literature to
mean representing the relation between representation
and referent. Following the spirit of that
argument, several theorists have put forth a case
for metarepresentation being a domain-general
capacity that includes but is not limited to
metarepresentation of mental states (Corballis,
2003; Perner, 1991; Stone & Gerrans, 2006;
Suddendorf, 1999; Suddendorf & Whiten, 2001).
In contrast, other developmental theorists have
pursued the idea that autism is a failure of a
domain-specific capacity for the metarepresentation
of mental states underwritten by a specific
cognitive mechanism: the Theory of Mind Mechanism
(ToMM; Leslie, 1987, 1994). As children
with autism often have difficulty with metarepresentation
of mental states, but not other metarepresentational
tasks (Leslie & Thaiss, 1992), it
seemed necessary to propose such a specific
module. Evolutionary psychologists embraced
the idea of such a module with enthusiasm, as it
represented a high-level domain-specific mechanism
with a specific adaptive function, the metarepresentation
of mental states (Cosmides &
Tooby, 1995). Given that neuroscience is one
important source of evidence for modularity,
neuroscientists began the search for evidence of
specialized brain regions that instantiate this
module, and for evidence of the dissociation
from other cognitive functions that would be the
hallmark of modularity (e.g., Saxe, Carey, &
Kanwisher, 2004; Stone, Baron-Cohen, & Knight,
1998). Those researchers who dissented from the
view that children with autism have a metarepresentation
deficit that is specific to mental states
are the very same theorists who did not espouse
the modular view (Perner, 1991; Russell, Saltmarsh,
& Hill, 1999).
The concept of ‘‘modularity’’ has been interpreted
in several different ways. On some views
modularity does not imply anatomical localization.
A distributed circuit can be specialized for
performance of a particular cognitive function
(Atkinson & Wheeler, 2004; Coltheart, 1999;
Stone et al., 1998). On any view of domain
specificity, however, functional specialization
must depend on specialized neural circuitry,
distributed or localized. Both fMRI and lesion
studies can show that particular sets of neurons
are necessary for performance of a certain
cognitive function. However, from the fact that
a set of neurons is necessary for performance of a
particular cognitive function it does not follow
that the neural circuit is specific to that function.
It may be necessary for other cognitive functions
as well. Working memory, which has a particular
neural instantiation, is necessary for understanding
complex grammatical sentences; however,
working memory and its neural substrate are
not specialized for that function. The boundaries
of domain-specific mechanisms (functional and
anatomical) are drawn on the basis of both
necessity and specialization for particular tasks.
This minimal characterization of domain-specificity
makes no claims (as modular theorists do)
about speed, automaticity, encapsulation, localization,
and representational format, which do not
concern us here. When we discuss a domainspecific
mechanism, we mean a neural circuit that
is not necessarily localized but is both necessary
for, and specific to, a particular cognitive task.
We argue that the postulation of a domainspecific
ToM mechanism represents an unnecessary
move in theoretical understanding of both
autism and normal development. An alternative
model of ToM can account for the data from both
autism and neurological patients more parsimoniously.
This alternative view, however, will need
to account for the data which led to the postulation
of a ToM mechanism in the first place: the
apparent dissociation between metarepresentation
per se and metarepresentation of mental
states.
Let us begin by stating what is in common
between all theorists of ToM development. First,
development of ToM depends on the prior
development of a suite of specialized lower-level
cognitive mechanisms. These precursor mechanisms
represent vital information about the social
world of the infant and toddler and mediate her/
his earliest interactions with others. These mechanisms
enable face processing, emotion processing,
representations of gaze direction, gaze
monitoring, detection of animacy, tracking of
intentions and goals, and joint attention (BaronCohen,
1995, 2004; Charman, Baron-Cohen,
Swettenham, Baird, Cox, & Drew, 2000; Crichton
& Lange-Ku¨ttner, 1999; Csibra, Biro, Koos, &
Gergely, 2003; Dawson, Meltzoff, Osterling, Rinaldi,
& Brown, 1998; Saxe et al., 2004; Schultz,
2005; Stone, 2005, 2006; Wellman, Phillips, Dunphy-Lelii,
& LaLonde, 2004; Woodward, 1999).
These capacities appear to be domain specific.
They seem to be specific to social stimuli, to be
310 STONE AND GERRANS
shared with other primates, and to depend on
neural circuitry that responds preferentially to
social stimuli (e.g., Blakemore, Boyer, PachotClouchard,
Meltzoff, Segebarth, & Decety, 2003;
Campbell, Heywood, Cowey, Regard, & Landis,
1990; Hare, Call, Agnetta, & Tomasello, 2000;
Kumashiro, Ishibashi, Itakura, & Iriki, 2002;
Perrett et al., 1990). 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
non-social stimuli, even other stimuli that are
physically similar to images of eyes (Campbell et
al., 1990; Hoffman & Haxby, 2000). Similar
preferential neural responses are involved in
assessing others’ goals and intentions, which
seems to depend on specific representations of
certain movement patterns: limb movement combined
with gaze, head, or body orientation
(Blakemore et al., 2003; Jellema, Baker, Wicker,
& Perrett, 2000).
The development of these capacities ensures
that the normal toddler is equipped with a
sophisticated battery of domain-specific mechanisms
that enable her/him to negotiate the social
world on the basis of perceptually available
information. In addition, the toddler appears to
have a domain-general capacity for secondary
representation (the capacity to hold in mind and
compare two different representations), but cannot
metarepresent beliefs (Suddendorf, 1999;
Suddendorf & Whiten, 2001). In this respect
she/he has similar social capacities to the great
apes, using essentially the same cognitive equipment
(Suddendorf & Whiten, 2001, 2003).
Leslie (1987, 1992) has maintained that pretense
in toddlers is evidence of early mental state
metarepresentational abilities, and one recent
study credits 15-month-olds with this ability
(Onishi & Baillargeon, 2005). Other developmental
research questions these claims, arguing
that they can be explained by secondary representations
or understanding pretense as a special
category of action (cf. Sobel & Lillard, 2002;
Suddendorf & Whiten, 2001; Wellman & Lagatutta,
2000). Developmental research does not
address the question of domain-specificity directly,
unless it can be shown that: (1) a ToM
task definitely depends on metarepresentation,
which is debated in the case of pretense; and (2)
infants can solve a ToM metarepresentational
task before they can solve any other well-matched
metarepresentational task, which has not been
established in the case of 15-month-olds. Thus,
just as debates about where ToM is in the brain do
not resolve the question of domain specificity,
neither do current debates about when ToM
metarepresentation emerges.
Another point of agreement among ToM
theorists is that by age four, the typically developing
child is equipped with a battery of higher
level domain-general cognitive mechanisms that
serve metacognitive computational functions. Advanced
executive function (meaning working
memory, inhibition and flexible control of attention),
secondary representations, recursion (the
ability to compute embedded representations),
and metarepresentation (understanding the representational
nature of the relationship between
representation and referent) all come on line by
age four, though they may develop further
beyond that age (De Villiers & Pyers, 2002;
Perner, 1991; Smith, Apperly, & White, 2003;
Suddendorf, 1999; Suddendorf & Whiten, 2001).
A further point agreed on by many theorists is
that by age four a child has some level of semantic
knowledge about what mental states are like
(Gopnik & Wellman, 1992; Leslie, 1987; Wellman,
1988). This semantic knowledge includes an
understanding that beliefs and desires are private
and changeable independent of the external state
of reality changing (Wellman & Lagattuta, 2000).
Positing semantic knowledge of a particular content
domain, however, is quite different from
positing that there are specialized computational
processes for a certain domain. Semantic knowledge
of a particular content domain is not a
‘‘domain-specific mechanism.’’
The original paper by Baron-Cohen et al.
(1985) suggested that passing the false belief
test engaged some or all of these metacognitive
mechanisms. The ToMM theory suggests that as
well as these domain-general mechanisms the child
develops a domain-specific mechanism for the
metarepresentation of mental states. See Figure 1
for a graphic representation. A child equipped
with both a ToMM and domain-general mechanisms
essentially has two metarepresentational
devices: one for social metarepresentations and
one for metarepresentations in other domains.
We think that it is unlikely that two separate
metarepresentational mechanisms (a) evolved
twice and (b) develop twice.
One of the main challenges to the domainspecific
view of ToM deficits in autism has been
the view that deficits in a domain-general ability,
executive function (EF), can account for autistic
children’s failures on ToM tasks (e.g., Ozonoff
DOMAIN SPECIFICITY OF TOM 311
et al., 1991). We wish to claim something quite
different, namely that ToM abilities depend, not
only on EF, language, recursion, or metarepresentation
per se, but on their developmental and
on-line interaction with the low level precursor
mechanisms previously described, e.g., gaze processing,
emotion recognition. Failure of these
low-level abilities can also cause failure on ToM
tasks, even in subjects with intact EF. This view
might explain why recent empirical results show
that early deficits in EF in autism cannot always
account for the difficulties children with autism
have with ToM tasks. Toddlers with autism show
evidence of joint attention deficits, but not always
early EF deficits (Griffith, Pennington, Wehner,
& Rogers, 1999; Rutherford & Rogers, 2003).
Furthermore, deficits in EF are not always
apparent when children with autism are tested
by a computer rather than a person (Ozonoff,
1995). In our view, this pattern is to be expected if
the cause of the failure is a deficit of low-level
input systems.1
How, then, can one account for
the fact that children with autism show deficits in
the metarepresentation of mental states, but not
in other metarepresentational tasks?
1
Indeed, much recent work in explaining autism has
looked at how early social deficits contribute to later theory
of mind difficulties (e.g., Baird et al., 2000; Baron-Cohen,
1995; Boucher & Lewis, 1992; Dawson et al., 2004a, 2004b;
Rutherford & Rogers, 2003; Schultz, 2005).
Figure 1. Mental state inferences with a ToM module. Architecture postulated by the ToMM theory. A child equipped with both a
ToMM and domain-general mechanisms essentially has two metarepresentational devices: one for social metarepresentations and
one for metarepresentations in other domains.
312 STONE AND GERRANS
The interaction between low-level precursor
domain-specific mechanisms and high-level domain-general
mechanisms can account for normal
performance on ToM tasks (Gerrans, 2003; Stone
& Gerrans, 2006). The result of this interaction is
not the creation of an extra domain-specific
mechanism, but the wiring up of distributed
metarepresentational circuitry that can take social
information as input and deliver ToM inferences
as output. See Figure 2 for a graphic
representation of this possible architecture. Metarepresentation
operating on information about
eye gaze2
and attention (who saw or was attending
to what) allows us to represent others’ knowledge
states (who knew what) (Baron-Cohen,
1995). Without the necessary inputs, no proper
inferences are made. Recursion operating on
metarepresentations of mental states can allow
us to reason about not just others’ thoughts, but
Figure 2. Mental state inferences without a ToM module. ToM is simply an interaction between low-level precursor domainspecific
mechanisms and high-level domain-general mechanisms. The result of this interaction is not the creation of an extra domainspecific
mechanism, as in Figure 1, but the wiring up of distributed metarepresentational circuitry that can take social information as
input and deliver ToM inferences as output. The performance of children with autism on ToM tests can be explained more
parsimoniously by the view that they have deficits, not in metarepresentation, but in lower-level domain-specific mechanisms for
processing social information. Without proper inputs, the intact capacity for metarepresentation by itself cannot make correct ToM
inferences.
2
To say that eye gaze is an input to ToM inferences is not
to say that it is the only necessary input. Obviously, other
senses can compensate, as they do in enabling blind children to
make ToM inferences, even if their development is slightly
delayed (Baron-Cohen, 1995).
DOMAIN SPECIFICITY OF TOM 313
also others’ thoughts about thoughts (Corballis,
2003). Indeed, children’s ability to use embedded
syntactical structures comes on line shortly before
the ability to solve false belief tasks, supporting
the view that recursion may be one general ability
serving both tasks (De Villiers & Pyers, 2002;
Smith et al., 2003). Executive function allows one
to keep the elements of a social interaction in
mind, and inhibit one’s 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 (Carlson &
Moses, 2001; Keenan, 1998; Stone et al., 1998).
Thus, deficits on ToM tasks can 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 module.
It is worth pointing out that recursive embedding,
e.g., of a clause within a sentence, is not the
same thing as metarepresentation, which is the
representation of a representational relationship.
Metarepresentation requires the recursive embedding
of representational relationships. Recursion
provides a schema but does not explain how
elements embedded within that schema are
generated. That metarepresentation and recursion
are separate domain-general abilities explains
why chimps who understand recursive
dominance relationships still cannot metarepresent
mental states: they do not understand
representational relationships (Suddendorf &
Whiten, 2001, 2003). Syntactic recursion seems
closely tied to ToM in development (De Villiers
& Pyers, 2002; Smith et al., 2003) but this does not
mean either that the two abilities are identical or
that ToM is domain specific.
The performance of children with autism on
ToM tests can be explained more parsimoniously
by the view that they have deficits, not in
metarepresentation, but in lower-level domainspecific
mechanisms for processing social information.
Without proper inputs, the intact capacity
for metarepresentation by itself cannot make
correct ToM inferences. Autistic children have
no impairment on false photograph tests because
they have no impairment in inputs required for
those tests. See Figure 3 for a graphic representation.
Thus, we argue that autistic deficits on ToM
tasks are evidence not of impaired metarepresentation,
but are instead evidence of impaired
inputs to ToM inferences. Research on social
deficits in autism has shown clearly that children
with autism have deficits in many early domainspecific
social competences: face recognition,
facial expression recognition, processing of gaze
direction, and joint attention (Baird, Charman,
Cox, Swettenham, Wheelwright, & Drew, 2000;
Baron-Cohen, 1995; Boucher & Lewis, 1992;
Dawson et al., 2004a; Dawson, Webb, Carver,
Panagiotides, & McPartland, 2004b; Rutherford
& Rogers, 2003; Schultz, 2005). These deficits in
lower-level domain-specific mechanisms have
been proposed as causes of the ToM impairment
seen in autism, because the causal antecedents of
ToM have not developed properly. Strong evidence
against our view would be a case of an
individual with deficits on metarepresentational
ToM tasks, but no lower-level social deficits. We
note that no such case has yet been demonstrated
in the autism literature.
We suggest that metarepresentation in autism
is intact and that evidence supports this view. In
the absence of comorbid intellectual disability,
individuals with autism seem to have intact
capacities for both metarepresentation and recursion.
They perform well on a non-mental metarepresentational
task, the false photograph test,
and can sometimes pass false belief tests (BaronCohen,
1989; Baron-Cohen, Wheelwright, Stone,
Jones, & Plaisted, 1999a; Leslie & Thaiss, 1992).
Baron-Cohen, Wheelwright, Stone, and Rutherford
(1999b) report three cases of high-functioning
individuals with ASD, a mathematician, a
physicist and an engineering student at a prestigious
university. All three fields require expertise
in mathematics, that most recursive of cognitive
abilities. Mathematics requires people to represent
the relations between symbols and their
objects (magnitudes, 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 capacities for metarepresentation,
recursion, and EF were intact. All did well
on the Tower of Hanoi, an EF test. 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.,
1999b), indicating a problem with lower-level
domain-specific mechanisms for face and gaze
processing rather than with metarepresentation.
There is currently no evidence for a pure metarepresentational
ToM deficit in autism.
In our view, the two routes to deficits on ToM
tasks are deficits in low-level input systems (e.g.,
representations of gaze, joint attention) or in
higher-level domain-general capacities (executive
314 STONE AND GERRANS
function, metarepresentation, and recursion). In
the spirit of empiricism, we offer to be proven
wrong. An individual with a deficit in ToM
metarepresentation without any deficit in lowerlevel
social competences and without any more
general deficit in other types of metarepresentation,
executive function, or recursion would
provide conclusive evidence against our view.
No such case from autism has yet been demonstrated
in the literature.
If children with autism do not provide such
evidence, then perhaps data from neurological
patients with lesions and selective deficits would
provide stronger evidence. Existing evidence
from neuropsychology, however, 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
other domain-general abilities. Patients with
orbitofrontal cortex (OFC) damage, for example,
have been found to be impaired on ToM tasks
(Gregory et al., 2002; Stone et al., 1998). However,
they are also impaired at recognizing facial
expressions and at judging mental states from eye
gaze or expression in the eye region of the face
(Gregory et al., 2002; Hornak, Rolls, & Wade,
1996; Snowden et al., 2003). Furthermore, even
their problems with some ToM tasks may result
from a difficulty in tracking intentions rather than
beliefs, as these are ToM tasks that, unlike false
belief tasks, do not specifically require metarepresentation
of belief (Stone, 2005). Thus, their
mentalizing deficits can be explained as a result of
deficits in lower-level domain-specific mechanisms,
not in higher-order domain-general mechanisms.
Indeed, some can perform at ceiling on false
Figure 3. Architecture supporting false photograph inferences. Children with autism have no impairment on false photograph tests
because their metarepresentational capacity is intact and they have no impairment in inputs required for such tests.
DOMAIN SPECIFICITY OF TOM 315
belief tasks, even those requiring 2nd- and 3rdorder
mental state inferences (Stone, 2005; Stone
et al., 1998). Patients with medial frontal damage
who have ToM deficits all have accompanying
executive function deficits, and extensive medial
frontal damage does not necessarily cause impairment
in ToM (Apperly, Samson, Chiavarino, &
Humphreys, 2004; Bird, Castelli, Malik, Frith, &
Husain, 2004; Gregory et al., 2002; Happe´, Mahli,
& Checkley, 2001; Stone, 2005).
There has been much recent excitement over
the possibility that patients with temporoparietal
junction (TPJ) lesions might represent a group
with specific ToM deficits (Apperly et al., 2004;
Samson, Apperly, Chiavarino, & Humphreys,
2004). However, the most recent evidence from
such patients indicates that these deficits are not
ToM specific either (Apperly, Samson, Chiavarino,
Bickerton, & Humphreys, in press). Language
and EF deficits can account for deficits on
language-based false belief tasks that require
inhibiting knowledge of the correct state of
reality, although some patients with lesions in
the same areas did well on these same tasks
(Apperly et al., 2004; Samson et al., 2004). More
interesting is the finding that other TPJ patients
who failed to perform above chance on nonverbal
false belief tasks also failed to perform above
chance on a very closely matched nonverbal test
of non-mental metarepresentation, the false
photograph test (Apperly, Samson, & Humphreys,
2005; Apperly et al., in press). Thus, there
is as yet no evidence from neurological patients
that supports the claim that a capacity for ToM
inferences can be impaired independently of
damage to lower-level input mechanisms or
higher-level domain-general mechanisms.
However, we do not believe that excitement
over the role of TPJ was unwarranted. Rather,
the TPJ might be interesting for a different
reason: Apperly et al.’s (in press) data suggests
that metarepresentation might be a crucial ability
subserved by TPJ. These patients also have
deficits in language, another ability that uses
metarepresentation (Stone, 2006). Although TPJ
does not appear to be the locus of the fabled ToM
module, it may be the locus of a much more
interesting ability. Metarepresentation, some researchers
have argued, is one of the key abilities
that separates humans from the great apes, one of
the key cognitive capacities that emerged over the
course of hominid evolution (Corballis, 2003;
Suddendorf, 1999). Thus, excitement over the
importance of this region should be excitement
over looking at a region subserving a domaingeneral
ability rather than a domain-specific
ability.
Some researchers, however, have argued that
neuroimaging data gives support to the domainspecific
view. Recent results indicate that the
TPJ is differentially active for false belief tasks
and false photograph tasks. This apparent discrepancy
would be resolvable with some additional
analyses of current imaging data. Saxe and
Kanwisher (2003) present contrasts between false
belief and false photograph trials, and between
false belief task and non-metarepresentational
task trials. They do not, however, present any
contrasts for the whole brain between false
photograph task and non-metarepresentational
task trials.3
If such a contrast did show differential
activation in TPJ, then this would support
the view that TPJ might underwrite metarepresentation
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 metare
presentation. In contrast, if TPJ was not differentially
active when non-metarepresentational
tasks are subtracted from false photograph tasks,
there would be stronger evidence for TPJ’s
subserving belief state metarepresentation speci-
fically.
A variety of tasks have been used to measure
ToM in neuroscience (Stone, 2005). False belief
tasks are considered the strongest tests because
they require metarepresentation. There are some
differences, however, in the false belief tasks used
by different experimenters. Saxe and Kanwisher
(2003) used verbal false belief tasks and verbal
false photograph tasks, tasks that differed in
difficulty for their participants. Apperly et al.
(2005, in press) have used nonverbal false belief
and false photograph tasks closely equated for
3
Figure 4 of Saxe and Kanwisher (2003) compares how
certain voxels respond to particular types of stimulus stories:
false belief, false photograph, physical descriptions of people,
desire, and nonhuman. However, these voxels were
preselected to be more active in false belief than in false
photograph tasks in ‘‘more than half’’ of the 14 participants.
Thus, not all voxels were compared on all tasks. Furthermore,
for some subsets of these 14 participants, there was no
significant difference in the activation between false belief
and false photograph tasks, a situation that is, on some
interpretations, at odds with ToM being a universal domainspecific
mechanism. The comparison we are suggesting (false
photograph vs. non-metarepresentational tasks) needs to be
done for all participants for all voxels.
316 STONE AND GERRANS
difficulty in neurotypical participants. Convergence
between neuroimaging and lesion studies
depends on researchers using the same paradigms.
These results will be clearer when a
uniform paradigm, verbal or nonverbal, is used
in both types of research.
In the absence of strong evidence for dedicated
neural circuitry or the dissociation of ToM
from metarepresentational or lower-level deficits,
it seems that the interaction of several
domain-general mechanisms and lower-level domain-specific
mechanisms can account for the
flexibility and sophistication of behavior, which
has been taken to be evidence for a domainspecific
ToM mechanism. One can understand,
as an episode in the history of science, why it
seemed necessary to posit a specific ToM mechanism.
Early evidence from autism seemed to
show a pattern of impairment in metarepresentation
of belief, without a corresponding impairment
in metarepresentation in other domains
(Baron-Cohen et al., 1985; Frith, Morton &
Leslie, 1991; Leslie & Thaiss, 1992). Only later
did it become clear that the primary deficits in
autism were at a lower level. Furthermore, early
explorations of the idea of metarepresentation
primarily centered around its role in the embedding
of beliefs (Dennett, 1987; Leslie, 1987). The
relevance of metarepresentation and recursion
for other cognitive functions*language, mathematics,
episodic memory, and future planning
(Corballis, 2003; Perner, 1991; Suddendorf,
1999)*was obscured by the significance of the
apparent dissociation between ToM and domaingeneral
metarepresentation.
The history of science is replete with examples
of theorists postulating unnecessary entities. In
insisting that the earth was at the center of
the universe, Ptolemy proposed the notion of
epicycles to explain the movements of the planets
across the heavens: separate circular orbits in
which the planets moved whose center was
attached to the celestial spheres, which surrounded
the earth. Copernicus and Kepler, moving
the sun to the center, could dispense with this
unnecessary theoretical addition. The positing of
a separate ToM module appears to have been an
‘‘epicycle,’’ an unnecessary addition to psychological
theorizing. We argue that metarepresentation,
as a domain-general ability, should be moved
back to its rightful place at the center of our
species’ uniquely human cognitive abilities.
REFERENCES
Apperly, I. A., Samson, D., Chiavarino, C., Bickerton,
W., & Humphreys, G. W. (in press). 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.
Apperly, I. A., Samson, D., Chiavarino, C., & 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(10), 1773Á1784.
Apperly, I. A., Samson, D., & Humphreys, G. W.
(2005). Domain-specificity and theory of mind:
Evaluating neuropsychological evidence. Trends in
Cognitive Sciences, 9(12), 572Á577.
Atkinson, A. P., & Wheeler, M. (2004). The grain of
domains: The evolutionaryÁpsychological case
against domain-general cognition. Mind & Language,
19(2), 147Á176.
Baird, G., Charman, T., Baron-Cohen, S., Cox, A.,
Swettenham, J., Wheelwright, S., & 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 & Adolescent Psychiatry,
39(6), 694Á702.
Baron-Cohen, S. (1989). The autistic child’s theory of
mind: A case of specific developmental delay.
Journal of Child Psychology & Psychiatry, 30(2),
285Á297.
Baron-Cohen, S. (1995). Mindblindness: An essay on
autism and theory of mind. Cambridge, MA: MIT
Press.
Baron-Cohen, S. (2004). The cognitive neuroscience of
autism. Journal of Neurology, Neurosurgery &
Psychiatry, 75(7), 945Á948.
Baron-Cohen, S., Leslie, A., & Frith, U. (1985). Does
the autistic child have a theory of mind. Cognition,
21, 37Á46.
Baron-Cohen, S., & Wheelwright, S. (2004). The empathy
quotient: An investigation of adults with
Asperger syndrome or high functioning autism,
and normal sex differences. Journal of Autism and
Developmental Disorders, 34(2), 163Á175.
Baron-Cohen, S., Wheelwright, S., Stone, V. E., Jones,
R., & Plaisted, K. (1999a). Recognition of faux pas
by normally developing children and children with
Asperger syndrome or high-functioning autism.
Journal of Autism and Developmental Disorders,
29(5), 407Á418.
Baron-Cohen, S., Wheelwright, S., Stone, V. E., &
Rutherford, M. (1999b). A mathematician, a physicist,
and a computer scientist with Asperger syndrome:
Performance on folk psychology and folk
physics tests. Neurocase, 5(6), 475Á483.
Bird, C. M., Castelli, F., Malik, O., Frith, U., & Husain,
M. (2004). The impact of extensive medial frontal
lobe damage on ‘‘theory of mind’’ and cognition.
Brain, 127(4), 914Á928.
Blakemore, S. J., Boyer, P., Pachot-Clouchard, M.,
Meltzoff, A., Segebarth, C., & Decety, J. (2003).
The detection of contingency and animacy from
DOMAIN SPECIFICITY OF TOM 317
simple animations in the human brain. Cerebral
Cortex, 13(8), 837Á844.
Boucher, J., & Lewis, V. (1992). Unfamiliar face
recognition in relatively able autistic children.
Journal of Child Psychology & Psychiatry, 33(5),
843Á859.
Campbell, R., Heywood, C., Cowey, A., Regard, M., &
Landis, T. (1990). Sensitivity to eye gaze in prosopagnosic
patients and monkeys with superior temporal
sulcus ablation. Neuropsychologia, 28(11),
1123Á1142.
Carlson, S., & Moses, L. (2001). Individual differences
in inhibitory control and children’s theory of mind.
Child Development, 72(4), 1032Á1053.
Charman, T., Baron-Cohen, S., Swettenham, J., Baird,
G., Cox, A., & Drew, A. (2000). Testing joint
attention, imitation and play as infancy precursors
to language and theory of mind. Cognitive Development,
15, 481Á498.
Coltheart, M. (1999). Modularity and cognition. Trends
in Cognitive Sciences, 3(3), 115Á120.
Corballis, M. (2003). Recursion as the key to the human
mind. In K. Sterelny & J. Fitness (Eds.), From
mating to mentality: Evaluating evolutionary psychology
(pp. 155Á171). New York: Psychology Press.
Cosmides, L., & Tooby, J. (1995). Foreword. In S.
Baron-Cohen (Ed.), Mindblindness: An essay on
autism and theory of mind. Cambridge, MA: MIT
Press.
Courchesne, E., Townsend, J., Akshoomoff, N. A.,
Saitoh, O., Yeung-Courchesne, R., Lincoln, A. J.,
et al. (1994). Impairment in shifting attention in
autistic and cerebellar patients. Behavioral Neuroscience,
108(5), 848Á865.
Crichton, M., & Lange-Ku¨ttner, C. (1999). Animacy
and propulsion in infancy: Tracking, waving, and
reaching to self-propelled and induced moving
objects. Developmental Science, 2(3), 318Á324.
Csibra, G., Biro, S., Koos, O., & Gergely, G. (2003).
One-year-old infants use teleological representations
of actions productively. Cognitive Science,
27(1), 111Á133.
Dawson, G., Meltzoff, A. N., Osterling, J., Rinaldi, J., &
Brown, E. (1998). Children with autism fail to orient
to naturally occurring social stimuli. Journal of
Autism and Developmental Disorders, 28, 479Á485.
Dawson, G., Toth, K., Abbott, R., Osterling, J.,
Munson, J., Estes, A., et al. (2004a). Early social
attention impairments in autism: Social orienting,
joint attention and attention to distress. Developmental
Psychology, 40(2), 271Á283.
Dawson, G., Webb, S. J., Carver, L., Panagiotides, H., &
McPartland, J. (2004b). Young children with autism
show atypical brain responses to fearful versus
neutral facial expressions of emotion. Developmental
Science, 7(3), 340Á359.
De Villiers, J., & Pyers, J. (2002). Complements to
cognition: A longitudinal study of the relationship
between complex syntax and false-belief understanding.
Cognitive Development, 17, 1037Á1060.
Dennett, D. (1987). The intentional stance. Cambridge,
MA: MIT Press.
Frith, U., Morton, J., & Leslie, A. M. (1991). The
cognitive basis of a biological disorder: Autism.
Trends in Neurosciences, 14(10), 433Á438.
Gerrans, P. (2003). The theory of mind module in
evolutionary psychology. Biology & Philosophy, 17,
305Á321.
Gopnik, A., & Wellman, H. M. (1992). Why the child’s
theory of mind really is a theory. Mind & Language,
7(1Á2), 145Á171.
Gregory, C., Lough, S., Stone, V. E., Erzinclioglu, S.,
Martin, L., Baron-Cohen, S., & Hodges, J. (2002).
Theory of mind in frontotemporal dementia and
Alzheimer’s disease: Theoretical and practical implications.
Brain, 125, 752Á764.
Griffith, E. M., Pennington, B. F., Wehner, E. A., &
Rogers, S. (1999). Executive functions in young
children with autism. Child Development, 70, 817Á
832.
Happe´, F. (1999). Autism: Cognitive deficit or cognitive
style? Trends in Cognitive Sciences, 3(6), 216Á222.
Happe´, F., Mahli, G. S., & 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, 83Á90.
Hare, B., Call, J., Agnetta, B., & Tomasello, M. (2000).
Chimpanzees know what conspecifics do and do not
see. Animal Behaviour, 59, 771Á785.
Hoffman, E. A., & Haxby, J. V. (2000). Distinct
representations of eye gaze and identity in the
distributed human neural system for face perception.
Nature Neuroscience, 3(1), 80Á84.
Hornak, J., Rolls, E. T., & Wade, D. (1996). Face and
voice expression identification in patients with
emotional and behavioural changes following ventral
frontal lobe damage. Neuropsychologia, 34(4),
247Á261.
Jellema, T., Baker, C. I., Wicker, B., & Perrett, D. I.
(2000). Neural representation for the perception of
the intentionality of actions. Brain & Cognition,
44(2), 280Á302.
Keenan, T. (1998). Memory span as a predictor of false
belief understanding. New Zealand Journal of Psychology,
27(2), 36Á43.
Kumashiro, M., Ishibashi, H., Itakura, S., & Iriki, A.
(2002). Bidirectional communication between a
Japanese monkey and a human through eye gaze
and pointing. Current Psychology of Cognition,
21(1), 3Á32.
Leslie, A. M. (1987). Pretence and representation: The
origins of ‘‘theory of mind. Psychological Review,
94(4), 412Á426.
Leslie, A. M. (1992). Pretense, autism, and the theoryof-mind
module. Current Directions in Psychological
Science, 1(1), 18Á21.
Leslie, A. M. (1994). Pretending and believing: Issues in
the theory of ToMM. Cognition, 50(1Á3), 211Á238.
Leslie, A. M., & Thaiss, L. (1992). Domain specificity in
conceptual development: Neuropsychological evidence
from autism. Cognition, 43, 225Á251.
Onishi, K. H., & Baillargeon, R. (2005). Do 15-monthold
infants understand false beliefs? Science,
308(5719), 255Á258.
318 STONE AND GERRANS
Ozonoff, S. (1995). Reliability and validity of the
Wisconsin Card Sorting Test in studies of autism.
Neuropsychology, 9(4), 491Á500.
Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991).
Executive function deficits in high-functioning autistic
individuals: Relationship to theory of mind.
Journal of Child Psychology & Psychiatry, 32, 1081Á
1105.
Perner, J. (1991). The representational theory of mind.
Cambridge, MA: MIT Press.
Perrett, D., Harries, M., Mistlin, A., Hietanen, J.,
Benson, P., Bevan, R., et al. (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(1), 25Á55.
Russell, J., Saltmarsh, R., & Hill, E. (1999). What do
executive factors contribute to the failure on false
belief tasks by children with autism? Journal of
Child Psychology & Psychiatry, 40(6) 859Á868.
Rutherford, M., & Rogers, S. J. (2003). Cognitive
underpinnings of pretend play in autism. Journal
of Autism & Developmental Disorders, 33(3), 289Á
302.
Samson, D., Apperly, I. A., Chiavarino, C., & Humphreys,
G. W. (2004). The left temporoparietal
junction is necessary for representing someone
else’s belief. Nature Neuroscience, 7(5), 499Á500.
Saxe, R., & Kanwisher, N. (2003). People thinking
about thinking people: The role of the temporoparietal
junction in ‘‘theory of mind. NeuroImage,
19, 1835Á1842.
Saxe, R., Carey, S., & Kanwisher, N. (2004). Understanding
other minds: Linking developmental psychology
and functional neuroimaging. Annual
Review of Psychology, 55, 87Á124.
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(2Á3), 125Á141.
Smith, M., Apperly, I., & White, V. (2003). False belief
reasoning and the acquisition of relative clause
sentences. Child Development, 74(6), 1709Á1719.
Snowden, J. S., Gibbons, Z., Blackshaw, A., Doubleday,
E., Thompson, J., Craufurd, D., et al. (2003). Social
cognition in frontotemporal dementia and Huntington’s
disease. Neuropsychologia, 41, 688Á701.
Sobel, D. M., & Lillard, A. S. (2002). Children’s
understanding of the mind’s involvement in pretense:
Do words bend the truth? Developmental
Science, 5(1), 87Á97.
Stone, V. E. (2005). Theory of mind and the evolution
of social intelligence. In J. Cacciopo (Ed.), Social
neuroscience: People thinking about people. Cambridge,
MA: MIT Press.
Stone, V. E. (2006). The moral dimensions of human
social intelligence. Philosophical Explorations, 9(1),
55Á68.
Stone, V. E., Baron-Cohen, S., & Knight, R. T. (1998).
Frontal lobe contributions to theory of mind.
Journal of Cognitive Neuroscience, 10, 640Á656.
Stone, V. E., & Gerrans, P. (2006). Does the normal
brain have a theory of mind? Thoughts on modularity
in response to Apperly et al. Trends in
Cognitive Sciences, 10(1), 3Á4.
Suddendorf, T. (1999). The rise of the metamind. In M.
C. Corballis & S. Lea (Eds.), The descent of mind:
Psychological perspectives on hominid evolution (pp.
218Á260). London: Oxford University Press.
Suddendorf, T., & Whiten, A. (2001). Mental evolution
and development: Evidence for secondary representation
in children, great apes and other animals.
Psychological Bulletin, 127(5), 629Á650.
Suddendorf, T., & Whiten, A. (2003). Reinterpreting
the mentality of apes. In J. Fitness & K. Sterelny
(Eds.), From mating to mentality: Evaluating evolutionary
psychology (pp. 173Á196). New York: Psychology
Press.
Wellman, H., Cross, D., & Watson, J. (2001). Metaanalysis
of theory-of-mind development: The truth
about false belief. Child Development, 72(3), 655Á
684.
Wellman, H., & Lagattuta, K. H. (2000). Developing
understandings of mind. In S. Baron-Cohen, H.
Tager-Flusberg, & D. Cohen (Eds.), Understanding
other minds: Perspectives from developmental cognitive
neuroscience (2nd ed., pp. 21Á49). Oxford,
UK: Oxford University Press.
Wellman, H., Phillips, A., Dunphy-Lelii, S., & LaLonde,
N. (2004). Infant social attention predicts
preschool social cognition. Developmental Science,
7(3), 283Á288.
Wellman, H. M. (1988). First steps in the child’s
theorizing about the mind. In J. W. Astington, P. L.
Harris, & D. R. Olson (Eds.), Developing theories of
mind (pp. 64Á92). New York: Cambridge University
Press.
Woodward, A. (1999). Infants’ ability to distinguish
between purposeful and non-purposeful behaviors.
Infant Behavior & Development, 22(2), 145Á160.
DOMAIN SPECIFICITY OF TOM 319