Upload
queensu
View
0
Download
0
Embed Size (px)
Citation preview
Brain and Cognition 55 (2004) 209–219
www.elsevier.com/locate/b&c
Understanding orbitofrontal contributionsto theory-of-mind reasoning: Implications for autism
Mark A. Sabbagh
Department of Psychology, Queen�s University, Kingston, Canada K7L 3N6
Accepted 17 April 2003
Available online 21 February 2004
Abstract
Autism is a lifelong developmental disorder that is associated with severe difficulties with ‘‘theory-of-mind’’—the understanding
that others� behaviors are motivated by internal mental states. Here, we raise the possibility that research examining the neural bases
of theory-of-mind reasoning has the potential to inform researchers about the elusive functional neural impairments associated with
autism. Evidence from our lab and others� suggests that theory-of-mind reasoning may be fractionated into at least two functionally
and anatomically distinct neural circuits. Specifically, the ability to decode others� mental states from observable cues (such as facial
expressions) may rely on contributions from the orbitofrontal/medial temporal circuit within the right hemisphere. In contrast, the
ability to reason about others� mental states may rely left medial frontal regions. We conclude by reviewing evidence suggesting that
the developmental roots of autism might lie in abnormal functioning of the orbitofrontal/medial temporal circuit which may, in
turn, underlie the abnormal development of social-cognitive skills among individuals with autism.
� 2004 Elsevier Inc. All rights reserved.
1. Introduction
Researchers generally agree that human social inter-
action gains much of its richness and complexity becausewe possess a ‘‘theory-of-mind’’—an understanding of
how other�s behaviors are motivated by their internal
mental states, such as beliefs, desires, intentions, and
emotions (Wellman, 1990). The conceptual framework
underlying research on theory of mind has its roots in
developmental psychology. At present, issues sur-
rounding aspects of children�s development of a theory-
of-mind have come to guide research on children�ssocial-cognitive development (Taylor, 1996). Through
this work, researchers have established links between
theory-of-mind and executive functions (Carlson &
Moses, 2001; Frye, Zelazo, & Burack, 1998; Perner &
Lang, 1999), language development (Baldwin & Toma-
sello, 1998; Bloom, 2000; Sabbagh & Baldwin, 2001),
prosocial behavior, and emotional functioning (Dunn,
1999; Happ�e & Frith, 1996).
E-mail address: [email protected].
0278-2626/$ - see front matter � 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.bandc.2003.04.002
Perhaps because of its wide ranging applications,
theory-of-mind research has extended well beyond the
domain of developmental psychology. Indeed, the topic
has become an interdisciplinary hub for clinical, social,cognitive, and comparative psychologists and philoso-
phers interested in understanding the foundations of
everyday social cognition (Malle, Moses, & Baldwin,
2001). Still more recently, cognitive neurologists and
neuroscientists have begun to explore the neural bases of
theory-of-mind skills (see Siegal & Varley, 2002, for a
recent review). This latter branch of research is contrib-
uting to our developing understanding of how the cog-nitive skills that enable high-level social understanding
are organized in the human brain (Cacioppo et al., 2002).
Perhaps most importantly, this research has opened new
windows for understanding the neuropathological bases
of clinical disorders in which social-cognitive and theory-
of-mind skills may be specifically impaired (e.g., Doody,
Goetz, Johnstone, Frith, & Cunningham-Owens, 1998).
One disorder in which a theory-of-mind approach hasbeen especially fruitful is autism (Baron-Cohen, 1995).
Autism is a lifelong and debilitating neuro-cognitive
developmental disorder that affects between 1/250
210 M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219
(Gillberg & Wing, 1999) and 1/500 (Fombonne, 1999)individuals. Autism is characterized by a triad of fea-
tures, including impaired social development, delayed
and deviant language, and an insistence on sameness
(American Psychiatric Association, 1994; Rutter, 1978).
Baron-Cohen (Baron-Cohen, 1995; Baron-Cohen, Les-
lie, & Frith, 1985) was one of the first to suggest that this
apparently disparate triad of symptoms may stem from
a fundamental deficit in theory-of-mind reasoning.Through his pioneering and continuing studies, it is now
well-established that many individuals with autism, in-
cluding high-functioning autism and Asperger syn-
drome, experience difficulties engaging theory-of-mind
reasoning (see Baron-Cohen, 2001, for a review).
What is perhaps most compelling about many of
these cases is that, especially among high-functioning
individuals with autism, theory-of-mind deficits stand incontrast to surprising strengths in other areas of cogni-
tive functioning. For instance, individuals with autism
have difficulty recognizing emotional mental states in
pictures of eyes, but no trouble discerning other un-
derlying traits (such as gender or age) (Baron-Cohen,
Wheelwright, & Joliffe, 1999). Also, individuals with
autism have difficulty reasoning about others� false be-
liefs (i.e., beliefs that do not match the true state of theworld), but are adept at reasoning about non-mental
representations that also can be false such as photo-
graphs (Leslie & Thaiss, 1992). This pattern of findings
has been taken by some to suggest that individuals with
autism� theory-of-mind deficit represents a rather spe-
cific neuro-cognitive impairment (Baron-Cohen, 1994).
While there is general agreement now that social-
cognition is particularly impaired in autism, there isconsiderable debate regarding the best way to charac-
terize the neuro-cognitive nature of this impairment (see
commentary on Baron-Cohen, 1994). In our lab, we
have been attempting to gain insight into this question
by characterizing the neural correlates of theory-of-
mind reasoning in normally developing adults using the
same tasks that have been used to index a theory of
mind impairment in autism. This work has two mainresearch aims. First, by gaining an appreciation of the
functional cortical regions that are important for theo-
ry-of-mind reasoning, we can begin to develop hypoth-
eses regarding the neuro-cognitive processes that
distinguish theory-of-mind reasoning from reasoning in
other domains. Second, insights from this approach can,
in turn, provide a way of understanding how known
neuro-cognitive characteristics associated with autismmay map onto the primary social-cognitive deficits that
characterize this developmental disorder.
1.1. Separating component processes in theory-of-mind
Before discussing our work, it is important to func-
tionally define what is meant by Theory-of-Mind in a
way that makes it amenable to cognitive neuroscienceinvestigations. Researchers typically functionally define
theory-of-mind as a unitary skill that is engaged anytime
one needs to represent the mental states of others.
Sometimes, the skill is referred to by the term mind
reading (Whiten, 1991). Nonetheless, it is important to
note that theory-of-mind reasoning can be separated
into at least two component processes (see also Tager-
Flusberg, 2001): (1) detecting or decoding others� mentalstates based on immediately available observable infor-
mation and (2) reasoning about those mental states in
the service of explaining or predicting others� actions.Examples of detecting or decoding mental states might
include some of the relatively rudimentary skills, such as
identifying others� focus of attention based on eye gaze.
In contrast, examples of reasoning about mental states
within the context of explaining or predicting actionmay include some more complex aspects of theory of
mind reasoning, such as distinguishing jokes from lies,
or predicting story characters� behavior on the basis of
false beliefs.
Ordinarily, these two aspects of theory-of-mind work
in concert to produce reliable judgments about others�mental states. However, the distinction between these
two processes is important because they each rely pri-marily on different kinds of social-information process-
ing skills. Mental state decoding relies principally on
social information that is gleaned in the immediate and
observable environment (e.g., the person�s actions, toneof voice, facial expression, etc.). By contrast, reasoning
about others� mental states to explain or predict action
requires one to access knowledge and facts about either
the person in question, or their contextual circum-stances. For example, in order to correctly infer that
someone is sad because she got a poor mark on an exam
one needs to detect sadness from the observable infor-
mation, know that she received a poor mark, and per-
haps know that she had wanted to do well.
As was noted above, there is some evidence to suggest
that high-functioning individuals with autism have dif-
ficulties with both mental state decoding (e.g., poor atdecoding mental states relative to gender) and mental
state reasoning (e.g., poor at reasoning about beliefs
relative to photographs). The fact that the two compo-
nent processes involved in theory-of-mind reasoning
may be affected in autism has interesting implications
for understanding the neuro-cognitive bases of theory-
of-mind reasoning. One possibility is that detecting and
reasoning about mental states are subserved by partiallyoverlapping neural circuitry. The overlapping regions
may represent a circumscribed sub-circuit that is spe-
cialized for representing information relevant to ‘‘theo-
ry-of-mind’’ no matter what the specific task (see Frith
& Frith, 1999). If this were the case, this ‘‘theory-of-
mind’’ site may be a key to understanding the neuro-
pathology of autism. An alternative possibility is that, as
Fig. 1. Approximate topographic layout of the 128-channel Geodesic
Sensor Net used in our electrophysiological studies. All recordings
were digitized at 500Hz (bandpass filter: 0.1–200Hz) and referenced to
vertex. Scalp impedances were kept below 50kX throughout the entire
recording session. After recording, data were submitted to algorithmic
artifact rejection routines that identified and excluded trials that were
contaminated with artifact (e.g., eye blink, eye movement). Data from
artifact free trials were averaged to create the ERP, re-referenced using
the average reference derivation, corrected to 100-ms baseline, and
then filtered (low-pass: 15Hz) for analysis.
M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219 211
is the case with language (Neville, Mills, & Lawson,1992; Pulvermueller, 2000), the different components of
theory-of-mind reasoning may be subserved by distinct
and non-overlapping circuitry. Individuals with autism
may be impaired on both components of theory of mind
for one of two reasons: (1) a fundamental deficit in both
systems or (2) a core deficit in one component—such as,
mental state detection—that affects appropriate devel-
opment of the other.Below, we will review evidence from our laboratory in
support of the hypothesis that different components of
theory-of-mind reasoning are associated with distinct,
non-overlapping neural systems. In particular, we will
review evidence suggesting that decoding others� mental
states is associated with anterior frontal systems, most
likely related to orbitofrontal/medial temporal circuit.
Moreover, these systems may be lateralized to the righthemisphere. In contrast, we will review evidence sug-
gesting that reasoning about mental states is related to a
relatively circumscribed area of medial prefrontal cor-
tex, that may be lateralized to the left hemisphere. After
reviewing this evidence, we will discuss the extent to
which our findings regarding the neural circuitry of
theory-of-mind relates to current hypotheses regarding
the developmental neuropsychological underpinnings ofautism (Dawson et al., 2002).
1.2. Neural correlates of decoding mental states
To date, very few studies have been designed specif-
ically with the goal of investigating the neural bases of
simply decoding mental states (see Haxby, Hoffman, &
Gobbini, 2002, for a recent review). However, someevidence bearing on this issue might come from studies
investigating the neural bases of facial emotion recog-
nition. Emotion is a mental state, and judgments about
others emotions can be made solely on the basis of im-
mediately available information. Neuropsychological
evidence suggests that bilateral amygdala damage im-
pairs recognition of emotional states, especially fear
(e.g., Adolphs, Tranel, Damasio, & Damasio, 1994;Calder et al., 1996). A role for medial temporal struc-
tures such as the amygdala in emotion identification has
also found some support in a small-scale MEG study
(Streit et al., 1999). In addition to the amygdala, there is
some evidence that individuals with ventral-medial
prefrontal lesions (including orbitofrontal cortex) are
impaired in their abilities to decode emotions from facial
expressions (Hornak, Rolls, & Wade, 1996).These studies have led to the intriguing hypothesis
that a circuit including the orbitofrontal cortex and
medial temporal regions (including the amygdala) may
be crucial for decoding others� emotional mental states
(Baron-Cohen, 1995; Rolls, 1996). However, interpre-
tation of these studies is limited somewhat by their small
scale, and by the fact that the stimuli used for emotional
judgments were different from ones used for controljudgments (e.g., gender, identity). A notable exception
comes from Baron-Cohen et al. (1999), who attempted
an fMRI investigation of emotional mental state de-
coding in which gender judgments of the same stimuli
were used as a control condition. Unfortunately, these
authors noted that there were a number of difficulties
with the precise design of their study that limited in-
terpretations of their results.
1.2.1. ERP study: Decoding mental states
We recently completed a study aimed at delineating
the spatial and temporal aspects of the neural systems
that are important for decoding others� emotional
mental states (Sabbagh, Moulson, & Harkness, in
press). To do this, we adapted the ‘‘Reading the Mind in
the Eyes’’ task (Baron-Cohen, Wheelwright, Hill, Raste,& Plumb, 2001) for use in a dense-array ERP paradigm
(see Fig. 1). In this modified version, 18 normal partic-
ipants first saw a word labeling either a mental/emo-
tional state (e.g., embarrassed) or a gender (e.g., female).
The word was then replaced with a fixation point, and
then a picture of eyes. Participants� task was to judge
(yes/no) whether the eyes depicted the emotion or gen-
der labeled by the preceding word. Event-related po-tentials were recorded time-locked to the onset of the
212 M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219
picture of the eyes, as participants made their decisions(see Fig. 2). Because emotional mental state and gender
judgments were made with respect to the same photo-
graphs, differences in ERP activity could not be attrib-
uted to general differences between stimuli.
Using this procedure, we found two aspects of the
ERP that distinguished mental state decoding from
gender decoding. The first component to show the
condition effect was the N270-400 that was present overfrontal and anterior temporal areas. We found that at
right hemisphere sites, the N270-400 component elicited
in the mental state decoding condition was more nega-
tive than in the gender decoding condition (see Fig. 3).
The N270-400 difference was not detected at left hemi-
sphere sites. The second ERP component to show a
condition effect was the P300. Specifically, the P300
elicited in the mental/emotional state condition wasmore positive than in the gender condition. The frontal
N270-400 differences and the P300 differences were the
only ones to meet significance criteria.
We conducted a source estimation analysis using the
low-resolution electromagnetic tomography (LORETA)
technique to identify candidate neural generators con-
tributing to the differences in scalp potentials at 320ms
post-stimulus.While this technique is limited by relatively
Fig. 2. Schematic of the structure of a single tria
Fig. 3. Three-dimensional spherical spline scalp interpolation of the voltage
stimulus with representative waveforms from regions that showed differences i
time period of N270-400 where condition differences were significant.
low spatial resolution, previous research has demon-strated that it can be effective in providing highly accurate
description of the cortical regions responsible for a given
effect (Pascual-Marqui, 1999). This analysis identified two
cortical regions that made strong contributions to the
scalp electrical differences: orbitofrontal cortex and me-
dial temporal cortex. While each of these areas showed
some bilateral activation within each region, higher levels
of activation were found in the right hemisphere.Of course, there are a number of problems with
source localization using ERP, and as such, these anal-
yses must be considered cautiously. It is noteworthy,
however, that these ERP findings dovetail nicely with
the neuropsychological data suggesting that orbito-
frontal and medial temporal lesions disrupt emotional
recognition. Thus, these findings do provide initial evi-
dence that the neural circuitry underlying one aspect oftheory-of-mind reasoning—decoding others� mental
states—may rely on neural circuits within orbitofrontal
and medial temporal regions.
1.3. Neural correlates of reasoning about mental states
Although the study described above represents one of
the first attempts to understand the neural correlates of
l in the mental state decoding ERP study.
of the difference wave (mental state—gender) sampled at 300ms post-
n the N270-400 component. Shaded regions of the waveforms represent
Table 1
Example of story from mental state reasoning task
Belief Photo
Fred put a flower and a bottle
of wine on the table
Fred put a flower and a bottle
of wine on the table
Fred showed them to his
friend Marianne
Fred showed them to his
friend Marianne
Marianne thought they
looked pretty
Marianne took a picture of
them with her camera
Then, Marianne fell asleep in
her room
Then, Marianne put her cam-
era away
While Marianne was asleep,
Fred moved the flower
Later on that day, Fred
moved the flower
Fred put the flower on the
piano
Fred put the flower on the
piano
He left the bottle of wine on
the table
He left the bottle of wine on
the table
M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219 213
decoding others� mental states, there is a fairly largebody of neuroimaging studies aimed at delineating the
neural correlates of reasoning about others� mental
states (see Frith & Frith, 2001, for a review). Typically,
these studies have relied on tasks in which the partici-
pants� neural activation is assessed as they perform tasks
that require reasoning about mental states as compared
with tasks that are closely matched, but do not require
reasoning about mental states. For instance, Fletcherand colleagues (Fletcher et al., 1995) developed a PET
paradigm in which participants were asked to predict or
explain behavior on the basis of either physical causality
or story characters� mental states. Similarly, Goel and
colleagues (Goel, Grafman, Sadato, & Hallett, 1995)
asked participants to make judgments about whether
objects would be considered familiar to either oneself or
an historical figure (e.g., Christopher Columbus). Eachof these studies, and a number of others that came later
(see Siegal & Varley, 2002), have converged on the result
that medial frontal areas, particularly in the left hemi-
sphere, make critical contributions to reasoning about
others� mental states.
When taken together with our findings on the neural
systems important for decoding others� mental states,
these findings seem to provide some evidence in supportof our hypothesis that decoding and reasoning about
mental states are subserved by non-overlapping neural
systems. Specifically, orbitofrontal and medial temporal
areas may be important for decoding mental states,
while medial frontal areas are important for reasoning
about mental states. Before drawing this conclusion,
however, it is important to consider some other inter-
pretations of the mental state reasoning studies. In eachof these studies, brain activation was demonstrated as
blood flow changes induced as participants reasoned
about mental state representations (like beliefs or per-
son-specific knowledge) relative to a control condition in
which participants reasoned about physical causation or
a real-world state of affairs. Arguably, then, the ‘‘mental
state’’ condition places more emphasis on the ability to
reason about abstract representations (e.g., beliefs) rel-ative to the other conditions. Thus, it remained possible
that left medial frontal cortex may be particularly im-
portant for making judgments about abstract represen-
tations, relevant to theory-of-mind or otherwise.
This is a particularly important concern because, as
discussed above, individuals with autism have special
difficulties in reasoning about mental representations
(e.g., beliefs) while they are unimpaired in their ability tomake judgments about non-mental representations (e.g.,
photographs).
1.3.1. Reasoning about mental states: ERP study
With this in mind, we undertook a study that was
designed to investigate whether left prefrontal regions
make a greater contribution to reasoning about mental
representations relative to non-mental representations(see Sabbagh & Taylor, 2000, for fuller description of
this work). Our study took as its starting place the false
belief/false photograph dissociation apparent in autism.
In our study, 23 typically developing adults had to
reason about outdated (or false) beliefs and photo-
graphs. These beliefs and photographs were outdated in
the sense that the scene changed subsequent to the cre-
ation of the representation, and as such, the represen-tation no longer matched the scene. Participants read a
belief and a photo version of 40 stories for a total of 80
stories (see Table 1 for an example). Following each
story, participants were asked a series of control ques-
tions to ensure that they understood the story, and a test
question asking them to judge the contents of the rep-
resentation. The test question was constructed such that
participants could not answer until the final word ap-peared. ERPs were recorded to the onset of the final
word in the question. The events associated with the test
question and recording of ERPs are illustrated in Fig. 4.
The key results of this study are illustrated in Fig. 5.
Even in this tightly controlled comparison, reasoning
about mental states was uniquely associated with left
frontal activity beginning 300ms following the onset of
the stimulus. Specifically, the ERP was characterized bya slow-wave effect, which was more positive in the belief
condition relative to the photograph condition. This
slow-wave condition effect was remarkably focal—it was
present only on a cluster of four sites over lateral frontal
areas. This effect was mirrored at left parietal sites where
a slow-wave effect also existed, but in the opposite di-
rection: the ERP elicited in the belief condition was
more negative relative to the photograph condition. Thispattern of differences was maintained throughout the
recording epoch, and was most distinct at around
820ms.
Unfortunately, we were unable to conduct source
localization analyses on these data, because precise
electrode locations were not measured. Nonetheless, the
Fig. 4. Schematic of the structure of a single trial in the mental state reasoning ERP study. (A) Question presented, word-by-word, 200ms. ISI. (B)
ERP epoch events timeline (all stimuli on screen for 250ms).
Fig. 5. Three-dimensional spherical spline scalp interpolation of the voltage of the difference wave (mental state—gender) sampled at 820ms post-
stimulus with representative waveforms from regions that showed differences in the time period between 300 and 850ms post-stimulus. Shaded
regions of the waveforms represent time periods within which condition differences were significant.
214 M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219
pattern of differences is consistent with a left anterior
medial frontal generator that is suggested by the fMRIand PET studies reviewed briefly above. These findings
find converging evidence for the importance of medial
prefrontal regions of the left hemisphere in reasoning
about others� mental states.
2. Summary and discussion of ERP findings
One of the most striking aspects of our ERP studies is
that, when viewed together, there was essentially no
reliable overlap between the neural systems that were
associated with decoding versus reasoning about mental
states. In particular, the ERP differences that indexed
mental state decoding were localized to inferior frontaland anterior temporal regions of the right hemisphere.
Source localization analyses estimated that these ERP
differences may have been generated within orbitofron-
tal prefrontal and medial temporal regions of the right
hemisphere. In contrast, reasoning about others� mental
states was associated with ERP activity that was focal
over prefrontal areas of the left hemisphere. Although
we did not conduct source localization analyses on thesedata, the pattern of activity was consistent with a medial
frontal generator within the left hemisphere that has
been described in fMRI and PET studies of mental state
reasoning (see Frith & Frith, 2001).
M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219 215
These findings support the notion that no single brainarea is active when humans engage theory-of-mind rel-
evant informational content. Instead, the neural systems
that support successful theory-of-mind judgments are
distributed across different cortical regions. These find-
ings raise important questions regarding how to char-
acterize the contributions of these different neural
circuits. One possibility is that instead of there being one
specialized circuit for reasoning about mental states,there may be several—each one dedicated to carrying out
a specific step in the process of making appropriate
theory-of-mind judgments. For instance, there may be a
specialized pathway within the orbitofrontal/medial
temporal circuit that is specifically dedicated to decod-
ing mental states. It seems possible that one aspect of
this specialization may be its laterality. There are a
number of findings now that suggest that decodingothers� mental states from observable information may
be a skill that is lateralized to the right hemisphere
(Sabbagh, 1999; Sabbagh et al., in press). Comple-
menting this right lateralized orbitofrontal/medial tem-
poral circuit, there may be a specialized region of medial
frontal regions of the left hemisphere may be especially
important for reasoning about mental states. This hy-
pothesis gains support from the fair amount of consis-tency across studies that have used tasks that require
theory-of-mind reasoning and revealed activation of left
medial frontal areas (Frith & Frith, 1999; Siegal &
Varley, 2002).
A second possibility is that instead of being special-
ized, each of these circuits performs a domain-general
computation that is required when engaging different
theory-of-mind related processes, but not specific to thissocial-cognitive domain. For instance, orbitofrontal ar-
eas may be important not only for decoding mental
states but also for decoding the meaning of stimuli in
other non-social domains. Similarly, the left medial
frontal contributions could reflect a domain-general
cognitive contribution that happens to be heavily taxed
when participants are engaged in theory-of-mind rea-
soning.Our studies do not provide us with the opportunity to
determine whether we have indexed domain-specific or
domain-general neural systems. However, some insight
on this issue may be taken from examining the tasks that
we used—and that have been used with individuals with
autism—to reveal electrophysiological dissociations as-
sociated with theory-of-mind. In both the case of mental
state decoding and mental state reasoning, the taskswere chosen because the mental state and non-mental
state conditions are, at least on their surfaces, very well
matched in terms of peripheral domain-general cogni-
tive demands. For example, the cognitive demands in
the false belief task seem very similar to those required
in the false photograph task (e.g., working memory,
logical reasoning, etc.). Similarly, there is no a priori
reason to suspect that the information-processing de-mands associated with decoding mental states from
pictures of eyes would be quantitatively and qualita-
tively different from those required to decode gender.
From this basis, then, it seems reasonable to suggest that
our electrophysiological differences reveal the operation
of neural circuits functionally dedicated to decoding and
reasoning about mental states (e.g., Frith & Frith, 2001).
It is important to note, however, that the assumptionsregarding how well these experimental tasks are mat-
ched have not been assessed directly. Thus, it remains
possible that the neural systems we suggest are crucial
for decoding and reasoning about mental states may be
related to domain-general cognitive skills that are more
heavily tapped in course of these processes. Although no
research has investigated this possibility directly, it is
possible that the findings from our ERP studies mayprovide some hypotheses about candidate cognitive
skills that may be especially important for theory-of-
mind reasoning. For instance, reasoning about mental
states may be rely on contributions from left prefrontal
areas because this skill may be related to executive
functioning and ‘‘inhibitory control.’’ A number of re-
searchers have argued that prefrontal regions are asso-
ciated with executive functions including workingmemory (Smith & Jonides, 1998), error monitoring
(Gehring & Knight, 2000), and inhibitory control (Di-
amond, 1998). Moreover, recent research strongly sug-
gests that these executive skills are crucial for reasoning
about others� mental states (Carlson & Moses, 2001;
Russell, 1996). Together, these findings strengthen the
possibility that left frontal contributions to theory-of-
mind may be related to executive functions in children.One caveat is that the bulk of research from both animal
and human work suggests that the aspects of executive
function that are typically related to theory-of-mind
performance in children are located with in dorsal–lat-
eral prefrontal cortex as opposed to medial frontal
cortex. Clearly, more research is needed to better un-
derstand this discrepancy.
In a similar vein, decoding mental states may recruitneural circuitry within orbitofrontal contributions be-
cause these areas seem to play a special role in decoding
the content of emotionally reinforcing stimuli generally
(Rolls, 1996, this issue). A considerable amount of re-
search now suggests that the orbitofrontal cortex plays a
crucial role in decoding the content of gustatory and
olfactory stimuli (see Rolls, 1996, for a review). Simi-
larly, Bechara and colleagues have demonstrated thatindividuals with damage to orbitofrontal cortex have
difficulty with so-called gambling tasks—card games in
which they are required to distinguish ‘‘rewarding’’
decks from ‘‘punishing’’ decks (e.g., Bechara, Tranel, &
Damasio, 2000). Arguably, the emotional value of social
interaction crucially relies upon the ability to decode
others� mental states. Thus, when mental states are
216 M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219
decoded, they may be represented primarily as socialemotional reinforcers. It may be, then, that orbitofron-
tal contributions to mental state decoding may be re-
lated to the emotionally reinforcing character of mental
states themselves.
As was noted above, we cannot marshal strong evi-
dence in favor of either a domain-specific or domain-
general interpretation of our results regarding the neural
bases of decoding or reasoning about mental states. Onething that is clear, however, is that neural systems as-
sociated with either component of theory-of-mind are
distinguishable from those that are associated with
similar judgments in other domains.
3. Neurobiology of autism: Is there a core deficit?
From the outset, a primary motivation for conduct-
ing our studies was to better understand the neuro-
pathological factors that contribute to autism. This is an
important aim because although it is generally recog-
nized that the cognitive and social deficits seen in autism
have their roots in neurodevelopmental abnormalities,
researchers have had considerable difficulty in accu-
rately characterizing specific sites of pathology (Min-shew & Goldstein, 1998; Piven & O�Leary, 1997). More
specifically, neuroimaging studies have failed to con-
verge on a single focal defect that characterizes the au-
tistic brain (see Rumsey & Ernst, 2000, for a recent
review). There are likely a number of reasons for this
lack of convergence. For one, autism is a heterogeneous
disorder with respect to presentation characteristics.
Affected individuals can range from having profound IQdeficits to relatively normal intelligence. Or, affected
individuals may present either with or without epilepti-
form symptoms. These relatively large variations may
obscure more subtle common neural characteristics
among individuals.
A second, and perhaps more important, reason for
the lack of convergence is that autism is a developmental
disorder. Building on recent work in the domain of de-velopmental cognitive neuroscience, Karmiloff-Smith
(1997) argues that development itself plays an important
role in shaping the functional organization of neural
systems. As such, it does not necessarily follow that
individuals affected by a given developmental disorder
would have a characteristic brain lesion. Instead, the
whole pattern of functional brain organization must be
considered in light of abnormal initial states (e.g., Deb &Thompson, 1998; Elman et al., 1996). This shift in per-
spective militates for understanding the neuro-cognitive
bases of autism from a fundamentally developmental
perspective.
The theory-of-mind framework may provide an ex-
cellent route to adopting just this perspective. Delin-
eating the developmental trajectory of theory-of-mind
skills from infancy through the preschool period hasbeen a central concern for many researchers in the area
of cognitive development. Although even a partial re-
view of this literature is beyond the scope of this paper,
one consistent finding is that children are able to detect
or decode others� mental states before they are able to
reason about those mental states in a sophisticated
manner. One example of this progression is the case of
eye-gaze monitoring. Infants as young as 6-month-oldshow evidence of attending to others� direction of eye-
gaze (Hood, Willen, & Driver, 1998), although it is not
until the age of 18-months (Baldwin, 1991) or later (Lee,
Eskritt, Symons, & Muir, 1998) that they are able to
reliably use this information to reason about others�intentions in a sophisticated manner. Research suggests
that this progression may also hold for later developing
aspects of theory-of-mind reasoning, such as under-standing of knowledge—even young children seem to be
able to reliably indicate whether someone is knowl-
edgeable (Bartsch & Wellman, 1994; Pratt & Bryant,
1990), but they have difficulty reasoning on the basis of
that information (Wimmer, Hogrefe, & Perner, 1988).
That decoding emerges prior to reasoning in the
theory-of-mind domain has important implications for
autism. Specifically, given that autism is a develop-mental disorder, it may be the case that at the core of
this disorder is the ability to engage the right orbito-
frontal/medial temporal circuit that we hypothesize
makes an important contribution to mental state de-
coding. This core deficit may, in turn, lead to additional
abnormalities in functional brain organization, and
theory of mind reasoning. A key piece of evidence in
evaluating this question comes from determining whe-ther young children with autism are specifically impaired
on tasks designed to tap orbitofrontal/medial temporal
functioning relative to other anterior systems.
Unfortunately, the evidence in this regard is sparse.
Dawson and colleagues (Dawson et al., 2002) recently
reported that autistic and non-autistic children who are
matched for mental age do not differ from one another
on tasks that are typically thought to tap dorsolateralversus orbitofrontal functioning (Dawson et al., 2002).
However, this same group of researchers has also re-
ported two very intriguing findings that indirectly link
variability in orbitofrontal functioning to both autism
and to theory-of-mind. First, Dawson and colleagues
(Dawson, Meltzoff, Osterling, & Rinaldi, 1998; Dawson,
Osterling, Rinaldi, Carver, & McPartland, 2001) have
shown that individuals with autism are relatively moreimpaired than mental age matched controls on the de-
layed non-matching to sample task, which taps the
ability to form rules regarding the links between stimuli
and rewards. This is important because research has
shown that this skill is mediated through ventromedial
prefrontal cortex, including orbitofrontal cortex (Meu-
nier, Bachevalier, & Mishkin, 1997). Second, across all
M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219 217
participants, tasks associated with orbitofrontal func-tioning was more likely to be impaired in the ability to
establish ‘‘joint attention,’’ which requires the ability to
detect others� attentional and intentional states based on
the direction of their eye gaze. Developmentally, chil-
dren�s abilities to establish joint attention is widely re-
garded as one of the first milestones on the road to
acquiring a theory-of-mind (Moore, 1999). Moreover, it
is gaining validity as one of the earliest warning signs ofautism (Baird et al., 2000). These findings linking joint
attention abilities with ventral-medial prefrontal func-
tion provide converging evidence for both (1) the fun-
damental role of ventral-medial regions in decoding
mental states (including joint attention), and (2) the
possibility that aspects of orbitofrontal dysfunction may
play a crucial role in understanding the developmental
neuropsychological factors that lead to the social deficitsevidenced by individuals with autism.
Finally, it should be noted that this hypothesis re-
garding the central importance of the orbitofrontal/
medial temporal circuit for both social cognition and
autism extends a long standing hypothesis regarding the
pathological role of the hippocampus and amygdala in
autism. Bachevalier (1994) and others have argued from
the basis of neuropsychological, animal lesion, andneuroimaging studies that the medial temporal lobe is an
important source of dysfunction in autism (see also
Howard et al., 2000). By including a role for orbito-
frontal regions, we may gain important insight into the
neuro-cognitive bases of the theory-of-mind deficits as-
sociated with autism.
4. Conclusions and potential clinical implications
In summary, our findings suggest first that there is no
one highly circumscribed neural circuit that is active any
time one engages in theory-of-mind relevant reasoning.
Instead, the neural circuitry that underlies the ability to
decode mental states from facial displays differs from
that underlying the ability to reason about mental states.Specifically, orbitofrontal cortex may be especially im-
portant for decoding mental states, while left medial
frontal regions may be important for reasoning about
mental states. It is presently unclear as to whether a
domain-specific or domain-general interpretation of
these neuro-cognitive contributions is most warranted.
Nonetheless, these findings may provide a springboard
for exploring some hypotheses about the neuropatho-logical bases of autism. Specifically, it seems possible
that autism may be characterized by a core deficit in the
functioning of the orbitofrontal/medial temporal circuit
that may underlie their difficulties in decoding mental
states. Given that decoding mental states is widely
considered to be the earliest emerging building block of
theory-of-mind development, it could be that a funda-
mental abnormality in mental state decoding is the basisof social cognitive deficits in autism.
We would like to conclude with a speculation about
the possible clinical implications of our findings. A
principal aim of therapists working with individuals
with autism is to improve their social functioning skills.
We, along with others, have hypothesized that the ‘‘core
deficits’’ associated with autism lie in the orbitofrontal/
medial temporal circuit that is important for decodingmental states. In contrast, the left prefrontal regions that
are important for reasoning about mental states may be
relatively less affected. If this were the case, it is possible
that individuals with autism may reveal strengths in
their ability to reason about mental states if the links
between mental states and their observable correlates
(e.g., facial expression) were made more transparent.
Put in another way, perhaps an effective therapeuticstrategy would be to help individuals with autism to
recognize mental states by presenting their contents
more transparently. For instance, individuals with au-
tism might have difficulty recognizing that someone is
sad, but if told explicitly, may be able to make appro-
priate judgments regarding that person�s attendant
mental states.
There is some evidence to suggest that this might bethe case. For instance, Wellman et al. (2002) have shown
that when others� mental states are made explicit
through the use of cartoon-like ‘‘thought bubbles’’ in-
dividuals with autism can solve basic theory-of-mind
tasks, including false belief. Similarly, Tager-Flusberg
(2001) reviews evidence suggesting that individuals with
autism can use language-based strategies to ‘‘hammer
out’’ solutions to problems that require reasoning aboutothers� mental states. More research is required to de-
termine whether these strategies lead to lasting im-
provements in social cognitive skills in individuals with
autism. Nonetheless, this approach may ultimately be an
effective way of capitalizing on the neuro-cognitive
strengths of individuals with autism to better their ev-
eryday social cognitive skills.
Acknowledgments
This article was prepared with support of an NSERC
Discovery Grant. The author thanks Sid Segalowitz and
PhilZelazo for very helpful comments on aprevious draft.
References
Adolphs, R., Tranel, D., Damasio, H., & Damasio, A. (1994).
Impaired recognition of emotion in facial expressions following
bilateral damage to the human amygdala. Nature, 372, 669–672.
American Psychiatric Association 1994. Diagnostic and statistical
manual of mental disorders (4th ed.). Washington, DC: Author.
218 M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219
Bachevalier, J. (1994). Medial temporal lobe structures and autism: A
review of clinical and experimental findings. Neuropsychologia, 32,
627–648.
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 and Adolescent Psychiatry, 39, 694–
702.
Baldwin, D. A. (1991). Infants� contribution to the achievement of
joint reference. Child Development, 62, 875–890.
Baldwin, D. A., & Tomasello, M. (1998). Word learning: A window on
early pragmatic understanding. In E. Clark (Ed.), Proceedings of
the twenty-ninth annual child language research forum (pp. 3–23).
Chicago, IL: CSLI.
Baron-Cohen, S. (1994). How to build a baby that can read minds.
Current Psychology of Cognition, 13, 513–552, and commentary.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory
of mind. Cambridge, MA: The MIT Press.
Baron-Cohen, S. (2001). Theory of mind and autism: A review. In L.
M. Glidden (Ed.), International review of research in mental
retardation: Autism (pp. 169–184). San Diego, CA: Academic Press.
Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic
child have a theory of mind? Cognition, 21, 37–46.
Baron-Cohen, S., Wheelwright, S., & Joliffe, T. (1999). Is there a
‘‘language of the eyes’’? Evidence from normal adults and adults
with autism or Asperger syndrome. Visual Cognition, 4, 311–331.
Baron-Cohen, S., Wheelwright, S., Hill, J., Raste, Y., & Plumb, I.
(2001). The ‘‘Reading the mind in the eyes’’ Test revised version: A
study with normal adults, and adults with Asperger syndrome or
high-functioning autism. Journal of Child Psychology and Psychi-
atry and Allied Disciplines, 42, 241–251.
Bartsch, K., & Wellman, H. M. (1994). Children talk about the mind.
Cambridge: Cambridge University Press.
Bechara, A., Tranel, D., & Damasio, H. (2000). Characterization of
the decision-making deficit of patients with ventromedial prefrontal
cortex lesions. Brain, 123, 2189–2202.
Bloom, P. (2000). How children learn the meanings of words.
Cambridge, MA: MIT Press.
Cacioppo, J. T., Berntson, G. G., Adolphs, R., Carter, C. S.,
Davidson, R. J., McClintock, M. K., McEwen, B. S., Meaney,
M. J., Schacter, D. L., Sternberg, E. M., Suomi, S. S., & Taylor, S.
E. (Eds.). (2002). Foundations in Social Neuroscience. Cambridge,
MA: MIT Press.
Calder, A. J., Young, A. W., Rowland, D., Perrett, D. I., Hodges, J.
R., & Etcoff, N. L. (1996). Facial emotion recognition after
bilateral amygdala damage: Differentially severe impairment of
fear. Cognitive Neuropsychology, 13, 699–745.
Carlson, S. M., & Moses, L. J. (2001). Individual differences in
inhibitory control and children�s theory of mind. Child Develop-
ment, 72, 1032–1053.
Dawson, G., Meltzoff, A., Osterling, J., & Rinaldi, J. (1998).
Neuropsychological correlates of early symptoms of autism. Child
Development, 69, 1277–1285.
Dawson, G., Munson, J., Estes, A., Osterling, J., McPartland, J., Toth,
K., Carver, L., & Abbott, R. (2002). Neurocognitive function and
joint attention ability in young children with autism spectrum
disorder versus developmental delay. Child Development, 73, 345–
358.
Dawson, G., Osterling, J., Rinaldi, J., Carver, L., & McPartland, J.
(2001). Recognition memory and stimulus-reward associations:
Indirect support for the role of ventromedial prefrontal dysfunc-
tion in autism. Journal of Autism and Developmental Disorders, 31,
337–341.
Deb, S., & Thompson, B. (1998). Neuroimaging in autism. British
Journal of Psychiatry, 173, 299–302.
Diamond, A. (1998). Evidence for the importance of dopamine for
prefrontal cortex functions early in life. In A. C. Roberts & T. W.
Robbins (Eds.), The prefrontal cortex: Executive and cognitive
functions (pp. 144–164). London: Oxford.
Doody, G. A., Goetz, M., Johnstone, E. C., Frith, C. D., &
Cunningham-Owens, D. G. (1998). Theory of mind and psychoses.
Psychological Medicine, 28, 397–405.
Dunn, J. (1999). Making sense of the social world: Mindreading,
emotion, and relationships. In P. D. Zelazo, J. W. Astington, & D.
R. Olson (Eds.), Developing theories of intention: Social
understanding and self control (pp. 229–242). Mahwah, NJ:
Erlbaum.
Elman, J., Bates, E., Karmiloff-Smith, A., Johnson, M., Parisi, D., &
Plunkett, K. (1996). Rethinking innateness. Cambridge MA: MIT
Press.
Fletcher, P. C., Happe, F., Frith, U., Baker, S. C., Dolan, R. J.,
Frackowiak, R. S. J., et al. (1995). Other minds in the brain: A
functional imaging study of �theory of mind� in story comprehen-
sion. Cognition, 57, 109–128.
Fombonne, E. (1999). The epidemiology of autism: A review.
Psychological Medicine, 29, 769–786.
Frith, U., & Frith, C. (1999). Interacting minds—a biological basis.
Science, 286, 1692–1695.
Frith, U., & Frith, C. (2001). Biological basis of social interaction.
Current Directions in Psychological Science, 10, 151–155.
Frye, D., Zelazo, P. D., & Burack, J. A. (1998). Cognitive complexity
and control: I. Theory of mind in typical and atypical development.
Current Directions in Psychological Science, 7, 116–121.
Gehring, W. J., & Knight, R. T. (2000). Prefrontal-cingulate interac-
tions in action monitoring. Nature Neuroscience, 3, 516–520.
Gillberg, C., & Wing, L. (1999). Autism: Not an extremely rare
disorder. Acta Psychiatrica Scandinavia, 99, 399–406.
Goel, V., Grafman, J., Sadato, N., & Hallett, M. (1995). Modeling
other minds. Neuroreport, 6, 1741–1746.
Happ�e, F. G. E., & Frith, U. (1996). Theory of mind and social
impairment in children with conduct disorder. British Journal of
Developmental Psychology, 14, 385–398.
Haxby, J. V., Hoffman, E. A., & Gobbini, M. I. (2002). Human neural
systems for face recognition and social communication. Biological
Psychiatry, 51, 59–67.
Hood, B. M., Willen, J. D., & Driver, J. (1998). Adults� eyes trigger
shifts of visual attention in human infants. Psychological Science, 9,
131–134.
Hornak, J., Rolls, E. T., & Wade, D. (1996). Face and voice expression
identification in patients with emotional and behavioral changes
following ventral frontal lobe damage. Neuropsychologia, 34, 247–
261.
Howard, M. A., Cowell, P. E., Boucher, J., Broks, P., Mayes, A.,
Farrant, A., & Roberts, N. (2000). Convergent neuroanatomical
and behavioural evidence of an amygdala hypothesis of autism.
NeuroReport, 11, 2931–2935.
Karmiloff-Smith, A. (1997). Crucial differences between developmental
cognitive neuroscience and adult neuropsychology. Developmental
Neuropsychology, 13, 513–524.
Lee, K., Eskritt, M., Symons, L. A., & Muir, D. (1998). Children�s useof triadic eye gaze information for ‘‘mind reading’’. Developmental
Psychology, 34, 525–539.
Leslie, A. M., & Thaiss, L. (1992). Domain specificity in conceptual
development: Neuropsychological evidence from autism. Cogni-
tion, 43, 225–251.
Malle, B. F., Moses, L. J., & Baldwin, D. A. (Eds.). (2001). Intentions
and intentionality: Foundations of social cognition. Cambridge, MA:
The MIT Press.
Meunier, M., Bachevalier, J., & Mishkin, M. (1997). Effects of orbital
frontal and anterior cingulate lesions on object and spatial memory
in rhesus monkeys. Neuropsychologia, 35, 999–1015.
Minshew, N. J., & Goldstein, G. (1998). Autism as a disorder of
complex information processing. Mental Retardation and Develop-
mental Disabilities Research Reviews, 4, 129–136.
M.A. Sabbagh / Brain and Cognition 55 (2004) 209–219 219
Moore, C. (1999). Gaze following and the control of attention. In P.
Rochat (Ed.), Early social cognition: Understanding the first months
of life (pp. 241–256). Mahwah, NJ: Erlbaum.
Neville, H. J., Mills, D. L., & Lawson, D. S. (1992). Fractionating
language: Different neural systems with different sensitive periods.
Cerebral Cortex, 2, 244–258.
Pascual-Marqui, R. D. (1999). Review of methods for solving the EEG
inverse problem. International Journal of Bioelectromagnetism, 1,
75–86.
Perner, J., & Lang, B. (1999). Development of theory of mind
and executive control. Trends in Cognitive Science, 3,
337–344.
Piven, J., & O�Leary, D. (1997). Neuroimaging in autism. Child and
Adolescent Clinics of North America, 6, 305–323.
Pratt, C., & Bryant, P. (1990). Young children understand that looking
leads to knowing (so long as they are looking into a single barrel).
Child Development, 61, 973–982.
Pulvermueller, F. (2000). The neuroscience of language: On brain
circuits of words and serial order. New York: Cambridge University
Press.
Rolls, R. (1996). The orbitofrontal cortex. Philosophical Transactions
of the Royal Society of London, Series B, Biological Sciences, 351,
1433–1444.
Rumsey, J. M., & Ernst, M. (2000). Functional neuroimaging of
autistic disorders. Mental Retardation and Developmental Disabil-
ities Research Reviews, 6, 171–179.
Russell, J. (1996). Agency. Cambridge: Cambridge University Press.
Rutter, M. (1978). Diagnosis and definitions of childhood autism.
Journal of Autism and Childhood Schizophrenia, 8, 139–161.
Sabbagh, M. A. (1999). Communicative intentions and language:
Evidence from right-hemisphere damage and autism. Brain and
Language, 70, 29–69.
Sabbagh, M. A., & Baldwin, D. A. (2001). Learning words from
knowledgeable versus ignorant speakers: Links between preschool-
ers� theory of mind and semantic development. Child Development,
72, 1054–1070.
Sabbagh, M. A., Moulson, M., & Harkness, K. L. (in press). Neural
correlates of mental state recognition in human adults: An ERP
study. Journal of Cognitive Neuroscience, 16.
Sabbagh, M. A., & Taylor, M. (2000). Neural correlates of the theory-
of-mind reasoning: An event-related potential study. Psychological
Science, 11, 46–50.
Siegal, M., & Varley, R. (2002). Neural systems involved in �theory of
mind�. Nature Neuroscience, 3, 463–471.
Smith, E. E., & Jonides, J. (1998). Neuroimaging analyses of human
working memory. Proceedings of the National Academy of Sciences,
95, 12061–12068.
Streit, M., Ioannides, A. A., Liu, L., Woelwer, W., Dammers, J.,
Gross, J., et al. (1999). Neurophysiological correlates of the
recognition of facial expressions of emotion as revealed by
magnetoencephalography. Cognitive Brain Research, 7, 481–491.
Tager-Flusberg, H. (2001). A reexamination of the theory of mind
hypothesis of autism. In J. A. Burack, T. Charman, N. Yimiya, &
P. R. Zelazo (Eds.), The development of autism: Perspectives from
development and theory (pp. 173–193). Mahwah, NJ: Erlbaum.
Taylor, M. (1996). A theory of mind perspective on social cognitive
development. In R. Gelman & T. Au (Eds.), Perceptual and
cognitive development. Handbook of perception and cognition (2nd
ed., pp. 283–329). San Diego, CA: Academic Press.
Wellman, H. M. (1990). The child�s theory of mind. Cambridge, MA:
MIT Press.
Wellman, H. M., Baron-Cohen, S., Caswell, R., Gomez, J. C.,
Swettenham, J., Toye, E., & Lagattuta, K. (2002). Thought bubbles
help children with autism to acquire an alternative to a theory of
mind. Autism, 6, 343–363.
Wimmer, H., Hogrefe, G., & Perner, J. (1988). Children�s understand-ing of informational access as source of knowledge. Child Devel-
opment, 59, 386–396.