Rapid publicationMaking sense of another mind: The role of the right temporo-parietal junction
Introduction
By a ‘Theory of Mind’, we mean the process(es) by which most healthy human adults (1) attribute unobservable mental states to others (and under certain circumstances, to the self, [cf. Bem, 1967, Happe, 2003]), and (2) integrate these attributed states into a single coherent model (Gopnik & Meltzoff, 1997) that can be used to explain and predict the target's behaviour and experiences. In this paper, we show that the hemodynamic response of one brain region – the right temporo-parietal junction (RTPJ) – reflects both of these characteristics of a Theory of Mind. First, we find that enhanced BOLD response in this region is selective to the attribution of mental states, and is not recruited by processing other socially relevant facts about a person. Second, activity in the RTPJ is modulated by the congruence or incongruence of multiple relevant facts about the target's mind. RTPJ activity was enhanced when the protagonist of a story professed a belief or desire that was inconsistent with the subject's expectations, based on the protagonist's background. Finally, none of the other brain regions commonly implicated in Theory of Mind reasoning – the left temporo-parietal junction (LTPJ), posterior cingulate (PC) and medial prefrontal cortex (MPFC) – showed an equally selective profile of response.
Recent neuroimaging work has suggested that multiple regions of cortex in the human brain are dedicated to components of the process of perceiving and reasoning about other people, including recognising and identifying human faces (e.g. Kanwisher, McDermott, & Chun, 1997; Hoffman & Haxby, 2000; Grill-Spector, Knouf, & Kanwisher, 2004), perceiving other human bodies (e.g. Downing, Jiang, & Kanwisher, 2001; Saxe, Jamal, & Powell, 2005), identifying human-like biological motion (e.g. Vaina et al., 2001; Grossman & Blake, 2002; Beauchamp, Lee, Haxby, & Martin, 2003; Pelphrey, Singerman, Allison, & McCarthy, 2003), perceiving intentional actions (e.g. Castelli, Happe, Frith, & Frith, 2000; Schultz et al., 2003; Saxe, Xiao, Kovacs, & Perret, 2004), and orienting towards and recognising basic emotional expressions (e.g. Whalen et al., 2001; LaBar, Crupain, Voyvodic, & McCarthy, 2003; Winston, O’Doherty, & Dolan, 2003; Wicker et al., 2003; Pessoa & Ungerleider, 2004). Beyond perceiving the physical appearance and behaviours of others, though, we intuitively conceive of each person as a being with a mind, and attribute to one another specific, content-ful mental states like perceptions, desires and beliefs. That is, we understand other people using a ‘Theory of Mind’ (Premack and Woodruff, 1978).
Could the human brain contain one or more specialised neural substrate(s) for Theory of Mind (e.g. a ‘Theory of Mind Module’ (Leslie & Thaiss, 1992)? At least four cortical regions are consistently identified as possible candidates (Saxe & Kanwisher, 2003; Gallagher et al., 2000, Fletcher et al., 1995, see reviews by Frith & Frith, 2003; Saxe, Carey, & Kanwisher, 2004): the right and left temporo-parietal junctions (RTPJ and LTPJ), posterior cingulate (PC) and medial prefrontal cortex (MPFC). The cognitive neuroscience of Theory of Mind has mostly depended on adaptations of the False Belief paradigm from developmental psychology. In this task, subjects must predict a character's action based on the character's false belief (Wimmer & Perner, 1983). False beliefs provide a useful behavioural test of Theory of Mind, because when the character's belief is false, the action predicted by the belief is different from the action that would be predicted by the true state of affairs (Dennett, 1978).
Increased activation has been reported in the same four brain regions when subjects reason about the false belief either of a character in a story (e.g. Fletcher et al., 1995, Gallagher et al., 2000, Vogeley et al., 2001) or in a cartoon (Gallagher et al., 2000, Brunet et al., 2000), or an imaginary ill-informed protagonist (Goel, Grafman, Sadato, & Hallett, 1995, a medical lay person, Ruby & Decety, 2003), relative to when belief attribution is not required. The control conditions have included scrambled texts, scrambled pictures (Fletcher et al., 1995, Gallagher et al., 2000), texts describing logical relations between events (Vogeley et al., 2001), judgements about the true function of an object (Goel et al., 1995), and stories about a physical representation of the world (e.g. a photograph) that becomes false (Saxe & Kanwisher, 2003, Experiment 2).
However, False Belief tasks alone are not enough to establish that a brain region is selectively recruited for Theory of Mind (e.g. Scholl & Leslie, 2001; Saxe, Carey et al., 2004), let alone to determine which component of Theory of Mind is reflected in its response. The current study was designed to help characterise the contributions of the RTPJ, LTPJ, PC, and MPFC to perceiving and understanding other people.
First, we asked whether the RTPJ, LTPJ, PC and/or MPFC was recruited selectively for the attribution of mental states, or more broadly when subjects reasoned about any socially relevant information about a person. A previous study (Saxe & Kanwisher, 2003) found that these regions were not involved in representing the mere physical appearance of another person. In the current study, we extended these earlier results by comparing the neural response during the attribution of mental states (e.g. ‘he wants to be a neurosurgeon’, ‘she likes to watch a little TV’, ‘he thinks it is a good idea to have sex before marriage’), with the response to information about a protagonist's social, geographical or cultural background (e.g. ‘Kevin is from Ireland and was raised strictly Catholic’, ‘Olivia comes from a middle class family’, ‘Carla has a top position at a large company’). The background information allowed subjects to begin to form an impression of the protagonist, without containing an explicit description of her mental states. Brain regions recruited selectively for mental state attribution should therefore respond little to the social background of the protagonist, while brain regions involved more broadly in social cognition and person perception might be recruited equally for both background and mental state information.
In addition, we asked whether the neural response would be affected by a manipulation of the protagonist's background: half of the protagonists had the same kinds of backgrounds as our subjects (the ‘Familiar’ backgrounds, e.g. middle class, American, urban), while the other half had ‘Foreign’ backgrounds (e.g. aristocratic, orthodox, isolated; see Appendix A for examples). We reasoned that the ‘Foreign’ background would constitute relevant distinctive social information about a person, in the absence of mental state attribution, and therefore would produce an enhanced response in brain regions involved in person perception but not restricted to Theory of Mind. For any brain region that was truly selectively involved in reasoning about mental states, we predicted (1) a low overall response to background information, and (2) no difference in the response to ‘Familiar’ versus ‘Foreign’ backgrounds.
Second, we asked whether the recruitment of each brain region would be influenced by the effort required to create an integrated coherent model of the protagonist's mind. Although tasks tapping Theory of Mind often measure the attribution of specific individual beliefs (or belief–desire pairs, e.g. Wimmer & Perner, 1983; Repacholi & Gopnik, 1997), mental state attribution is fundamentally holistic: a belief or desire can only be used to explain an action against the background of many (probably infinitely many) other beliefs and desires. Even 2- and 3-year-olds, in their spontaneous speech about the mind, obey the rule that mentioned mental states must be consistent with one another, and relevant to the action or situation (Bartsch & Wellman, 1995). Furthermore, there is extensive evidence that perceivers expect other people to be coherent, unified entities, and strive to resolve inconsistencies with that expectation (see review by Hamilton & Sherman, 1996). Consequently, when a target's behaviour violates the perceiver's previous impression of that person, the perceiver spends more time processing the behaviour (Bargh & Thein, 1985) and searching for the behaviour's causes (Hamilton, 1988), and later shows enhanced memory for the incongruent information (e.g. Wyer & Gordon, 1982, see review by Higgins & Bargh, 1987). We hypothesised that a similar process of integration and inconsistency-resolution could be elicited in the context of mental state attribution.
Expectations about the background mental states of a target individual may be influenced by schematic knowledge about the group membership of the target, and about the typical beliefs and desires of members of that group. We sought to set up such expectations about the protagonists of our stories in the background information, described above. Following the background information, we gave subjects a description of the protagonist's beliefs and/or desires. This mental state could be ‘Normal’ (similar to those of our subjects) or ‘Norm-Violating’ (unusual, and even inappropriate, in our subjects’ social environment). The experiment followed a 2 × 2 design: the protagonist's mental state could either be congruent with her background (e.g. a ‘Normal’ mental state in a protagonist from a ‘Familiar’ background), or incongruent with her background (e.g. a ‘Norm-violating’ mental state in a protagonist from a ‘Familiar’ background, Terwogt & Rieffe, 2003).
After the mental state was described, the stories concluded with the outcome for the protagonist – whether her preference was fulfilled – and then subjects were asked to predict whether the protagonist would feel positive or negative about this outcome. Successful performance of the task depended on integration of the stated mental state of the protagonist with the subsequent outcome, but did not involve the protagonist's background. Subjects could therefore have adopted a policy of ignoring the background information altogether. By contrast, we anticipated that subjects would attempt to integrate all of the information about the protagonist's mind as it became available. Consequently, we predicted that the incongruent stories would elicit an effort to resolve the inconsistency, at the time when the mental state information was presented, and that this inconsistency resolution would be reflected in the BOLD response of brain regions involved in Theory of Mind.
These same stimuli also allowed us to test a third hypothesis about the way perceivers reason about other minds: namely, that the mind of the target is represented fundamentally in terms of the similarity between the target's mind, and the perceiver's own. This alternative hypothesis can be derived from one currently popular class of theories of understanding other minds, collectively called the Simulation Theory (ST, Stich & Nichols, 1992; Gordon, 1998, Goldman, 1992, Harris, 1992, Heal, 1998, Nichols et al., 1995). ST proposes that an observer reasons about other minds by ‘putting herself in the other person's shoes’ and then passively reading off the mental states that arise in her own mind, within the pretend context.
There are many different specific versions of ST, and so there can be no monolithic prediction for neural activity from an ST perspective. We reasoned that one way to cash out the central notion of simulation would be to predict a linear relationship between the similarity of the modelled mind to the modellers mind, and the response of brain regions involved in the simulation. In our paradigm, the protagonists from a ‘Familiar’ background who professed a ‘Normal’ belief or desire were the most similar to the observers (our subjects). The protagonists from a ‘Foreign’ background who professed ‘Norm-violating’ beliefs were the least similar, since they differed from our subjects both in background and in mental state. The other two conditions were intermediate: each contained one similar and one dissimilar element.
These two hypotheses – one inspired by ToM and the other by ST – thus make different predictions about the response of the RTPJ, LTPJ, PC, and MPFC (or any neural substrate of reasoning about other minds). The ToM perspective predicts a higher response for the incongruent conditions, relative to the congruent conditions. On our reading, ST might predict that the response of these regions would increase linearly with the similarity between the subject and the protagonists, reflecting the ease or success of the simulation. Alternatively, ST might predict that the neural response would increase linearly with the dissimilarity between the subject and the protagonists, reflecting the effort of the simulation or the number of changes to the default assumption of similarity (Harris, 1992, Nichols et al., 1995). However, our simple first-order version of ST does not predict an interaction between the background and mental state of the character. In the current study, we tested these competing predictions.
Section snippets
Methods
Twelve naïve right-handed subjects (6 female; 1 Asian, 2 African-American, 1 Hispanic) gave written informed consent in accordance with the requirements of Internal Review Boards at Massachussetts General Hospital and MIT. All subjects were native speakers of English, and had normal or corrected-to-normal vision. Furthermore, all subjects were raised in middle class families in the United States (for more details, see Section 3.1).
Subjects were scanned at 3T (at the MGH scanning facility in
Behavioural results
The average familiarity score for familiar backgrounds was 4.32 out of a maximum of 5 (S.D. 0.72). For foreign backgrounds, the average score was 1.42 (S.D. 0.31, t(1, 10) = 12.14, p < 0.001, paired-samples t-test). The average agreement subjects reported with normal desires was 4.16 (S.D. 0.66) and for norm-violating desires it was 1.55 (S.D. 0.35, t(1, 10) = 11.30, p < 0.001, paired-samples t-test). No subject reported strong (>3) identification with any ‘foreign’ background or ‘norm-violating’
Discussion
Our subjects appeared to be very adept – even formulaic – at applying the ToM maxim that ‘people's feelings have to be predicted from their own subjective desires’ (Terwogt & Rieffe, 2003). A previous behavioural study found that when asked to ‘really consider’ the protagonist's feelings, subjects tended to overrule the protagonist's stated desire under very specific circumstances: when a protagonist from a ‘Familiar’ background professed a ‘Norm-violating’ desire. Thus, the subjects were
Acknowledgements
This work was funded by grant NIMH 66696. Thanks to Andrew Baron, Daniel Gilbert, Susan Carey, Lindsey Powell, Yuhong Jiang, Laura Schulz, Tania Tzelnic, Jason Mitchell, and especially to Nancy Kanwisher.
References (60)
- et al.
A PET investigation of the attribution of intentions with a nonverbal task
Neuroimage
(2000) - et al.
Movement and mind: a functional imaging study of perception and interpretation of complex intentional movement patterns
Neuroimage
(2000) - et al.
Other minds in the brain: a functional imaging study of ‘theory of mind’ in story comprehension
Cognition
(1995) - et al.
Functional imaging of ‘theory of mind’
Trends Cognitive Science
(2003) - et al.
Reading the mind in cartoons and stories: an fMRI study of ‘theory of mind’ in verbal and nonverbal tasks
Neuropsychologia
(2000) - et al.
Brain areas active during visual perception of biological motion
Neuron
(2002) - et al.
Domain specificity in conceptual development
Cognition
(1992) - et al.
Brain activation evoked by perception of gaze shifts: the influence of context
Neuropsychologia
(2003) - et al.
People thinking about thinking people: fMRI investigations of theory of mind
Neuroimage
(2003) - et al.
A region of right posterior superior temporal sulcus responds to observed intentional actions
Neuropsychologia
(2004)
Stereotyped beliefs about desirability: implications for characterizing the child's theory of mind
New Ideas in Psychology
Mind reading: neural mechanisms of theory of mind and self-perspective
Neuroimage
Both of us disgusted in My insula: the common neural basis of seeing and feeling disgust
Neuron
Beliefs about beliefs: representation and constraining function of wrong beliefs in young children's understanding of deception
Cognition
Common and distinct neural responses during direct and incidental processing of multiple facial emotions
Neuroimage
Forming Impressions of Personality
Journal of Abnormal and Social Psychology
Theory of mind: independence of executive function and the role of the frontal cortex in acquired brain injury
Cognitive Neurospsychiatry
Individual construct accessibility, person memory, and the recall-judgment link: the case of information overload
Journal of Personality and Social Psychology
Children talk about the mind
FMRI responses to video and point-light displays of moving humans and manipulable objects
Journal of Cognitive Neuroscience
Self-perception: an alternative interpretation of cognitive dissonance phenomena
Psychological Review
The impact of extensive medial frontal lobe damage on ‘Theory of Mind’ and cognition
Brain
Beliefs about beliefs
Behavioural and Brain Sciences
A cortical area selective for visual processing of the human body
Science
Development and neurophysiology of mentalizing
Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences
Ordinary personology
Modeling other minds
Neuroreport
In defense of the simulation theory
Mind and Language
Words, thoughts, and theories
Radical simulation
In theories of theories of mind
Cited by (749)
Anodal tDCS over TPJ reduces bidding in Tullock contest: Implications for social decision-making
2023, Neuroscience Letters