Mental state attribution and the temporoparietal junction: An fMRI study comparing belief, emotion, and perception
Introduction
The development during childhood of a theory of mind – the ability to attribute mental states to others – has been characterized as involving both an early referential theory and a later representational one (Flavell, 1999, Perner, 1993, Wellman et al., 2001) The early referential theory supports the ability to attribute desires, goals, perceptions, and emotions, an ability evident even in toddlers, while the later representational theory underlies the ability to attribute true and false beliefs, an ability not acquired until the preschool years.
By the time children are 2 years old, they spontaneously refer to desires (Bartsch & Wellman, 1995) and indicate their understanding that another person's desires may differ from their own (Repacholi & Gopnik, 1997). They speak about perceptions and emotions as well, referring to what people see and what people feel. Moreover, they speak about the causes of their emotions (Bretherton et al., 1986, Wellman et al., 1995), especially noting that frustrated desires lead to anger or sadness (Dunn and Brown, 2001, Lagattuta and Wellman, 2001). It has been claimed that such a coherent causal framework, one that allows prediction and explanation of human action, constitutes a genuine theory of mind (Wellman & Woolley, 1990). Still, it has been argued, this early theory does not require any understanding of mental representation. A goal or perception or emotion may be understood by the very young child as a connection of a particular sort toward an object or event (Flavell, 1988). For example, a desire for an object may be understood as an attraction, a pull, that refers to the object. It would not be understood by the child as a representation. Indeed, a representational understanding of the mind is not evident until about age 4, when the child first succeeds in attributing a false belief to a deceived actor. The knowledge that beliefs can be false (that is, they can misrepresent) necessarily indicates the understanding that beliefs are representations (Wimmer and Perner, 1983, Wellman and Cross, 2001, Zaitchik, 1990). Different models of theory of mind have agreed that the ability to attribute a false belief represents a developmental watershed, and a good deal of current research is dedicated to specifying the precise cognitive mechanisms underlying this development (Leslie, 2000, Baron-Cohen, 1995). If new cognitive mechanisms come on-line to support the later theory of mind, an interesting question arises: is the neural circuitry subserving the attribution of the early referential mental states (desires, perceptions, emotions) different from the neural circuitry subserving the attribution of beliefs? Is there a neural substrate specialized for belief attribution? Rebecca Saxe and her colleagues have laid out two criteria that they consider necessary to support a claim that a particular brain region is specialized for the attribution of belief (Saxe, Xiao, Kovacs, Perrett, & Kanwisher, 2004). First, the area must show an increased response to stimuli inviting true belief attributions as well as false belief attributions. Second, it must respond specifically to belief attributions. The present study focuses on the issue of specificity, directly comparing brain activity during attributions of beliefs vs. attributions of emotions and perceptions.
Areas within the broad anatomic area termed TPJ have consistently been found to be active during tasks involving the attribution of mental states. A recent extensive meta-analysis (Van Overwalle, 2009) reports that 13 of the 15 studies involving attribution of beliefs showed activation of left TPJ (Ferstl and von Cramon, 2002, Wang et al., 2006), right TPJ (Grézes et al., 2006, Grèzes et al., 2004, Sommer et al., 2007, Wang et al., 2006), or bilateral TPJ (Gallagher et al., 2000, German et al., 2004, Perner et al., 2006, Saxe and Kanwisher, 2003, Saxe and Powell, 2006, Saxe et al., 2006b). Importantly, these same regions have been shown to be specific to mental states: for example, they do not respond merely to the presence of a person or to non-mental representations such as an outdated photograph (Saxe & Kanwisher, 2003) and they do not respond to subjective sensations such as hunger or cold (Saxe & Powell, 2006). It seems likely then that any neural area responding preferentially to belief attribution will be located in the TPJ. Still, the TPJ is recruited during the attribution of other mental states as well – and these include referential states such as desires, intentions, and goals (Van Overwalle, 2009). Indeed, Van Overwalle (2009) has suggested that the TPJ is recruited for attribution of all temporary mental states. These would include perceptions and emotions as well as beliefs (it would exclude enduring properties, such as personality traits or long-standing dispositions). Thus, it is not clear whether there are regions within the TPJ that respond selectively to representational mental states (beliefs), as opposed to referential mental states (emotions and perceptions), or whether these regions are relatively unselective behaviorally.
This lack of specification is in part related to the design of tasks used in previous neuroimaging studies of theory of mind. To test the claim that different circuitry subserves early and late theory of mind, a neuroimaging study must directly compare the attribution of referential states (e.g., emotions, perceptions) with the attribution of representational states (e.g., beliefs). To date, however, there have been no published neuroimaging studies that directly compare beliefs with perceptions and emotions. Moreover, stimuli used in these paradigms vary widely, such that neuroimaging studies of beliefs have primarily used story stimuli, requiring the participant to infer the story character's belief from the causal events specified in the narrative (Ferstl and von Cramon, 2002, Gallagher et al., 2000, Perner et al., 2006, Saxe and Kanwisher, 2003, Saxe and Powell, 2006, Saxe et al., 2006a, Saxe et al., 2006b, Vogeley et al., 2001). In contrast, studies of emotions and perceptions have generally used visual displays such as pictures of emotionally expressive faces (Narumoto, Okada, Sadato, Fukui, & Yonekura, 2001) or videos of moving eyes (Pelphrey, Singerman, Allison, & McCarthy, 2003). These studies require visual detection of a mental state, a psychological process quite different from inference. Some studies mix the two. For example, one recent paper (Gobbini, Koralek, Bryan, Montgomery, & Haxby, 2007) compares the areas recruited during two tasks – one involving stories about false beliefs, the other involving animated movements of geometric blobs. These tasks vary in many ways in addition to the specific types of mental states involved. Restricting the stimuli to stories would not completely solve the problem either: story tasks involve complex narratives whose length, causal structure, and variable content allow for differences among conditions that extend beyond the type of mental state. While previous studies have contributed in a variety of important ways to our understanding of the neural substrates of social cognition, it seems important to approach an investigation of perceptions, emotions, and beliefs using a tightly controlled behavioral paradigm. Such a study would use a within-subject design, a single type of stimulus (visual or verbal), a single type of psychological process (for example, target detection, causal inference, or sentence comprehension), and identical stimuli (save for the mental state to be attributed). This is the goal of the present study.
In addition to the issue of behavioral specificity, interpretation of the prior literature presents a challenge related to anatomic specificity as well. The TPJ is relatively large, encompassing inferior parietal lobule and superior temporal brain regions, and exhibits substantial anatomic variability between individuals. While previous analytic models often treated these regions as one, contemporary models suggest that they may have different functions. Specifically, the superior temporal cortex may be specialized for reading mental states from body cues; for example, regions of STS show increased activation in response to visual displays of eye, mouth, and hand movements (Allison et al., 2000, David et al., 2008, Pelphrey et al., 2005), as well as whole body movements (Saxe, Xiao, et al., 2004). The inferior parietal lobule (IPL), in contrast, may be responsible for mediating one's sense of agency – the feeling that the self is causing or generating an experience, not just passively undergoing it (Decety & Chaminade, 2003).
The issues outlined above have led some investigators to call for the use of developmentally motivated hypotheses, carefully defined regions of interest, and verbal stimuli in the design of future theory of mind neuroimaging studies (Saxe, 2006a, Saxe, 2006b). The present study adopts this approach. As noted above, the developmentally motivated hypothesis is that one or more regions within the TPJ will respond selectively to representational mental states as opposed to referential states. With respect to anatomic specificity, our analytic approach employs the a priori use of anatomically defined regions of interest including the inferior parietal lobule and caudal superior temporal sulcus, defined at the individual subject level. The task uses verbal stimuli in the form of single sentences. Finally, to address the issue of behavioral specificity of the paradigm, the task conditions vary on only a single parameter: the type of mental state specified in the stimulus statement.
Section snippets
Participants
Participants were 15 right-handed, native English speakers (8 males, 7 females; mean age = 22.4 years, range of 20–24; mean education = 15.7 years, range of 11–18 years) who were recruited via local advertisements and received $100 each for participation. A standard health screen was administered over the phone to ensure no history of medical, psychiatric, or neurological illness prior to enrollment. Upon arrival to the testing center, all participants were administered a handedness screen (the
Behavioral data1
One-way ANOVA revealed significant effects of condition on both accuracy [F, df(3, 2863) = .003] and reaction time [F, df(3, 2863) = .001]. Post hoc tests show that these effects were due solely to the syntax CONTROL condition, which led to less accurate responses than the EMOTION and PERCEPTION conditions (ps < 0.02) and slower responses than the BELIEF and EMOTION conditions (ps < 0.01). There were no significant differences in either accuracy or reaction time among the three mental state conditions.
Functional imaging
Discussion
This study was motivated by a very puzzling developmental phenomenon – a 2-year lag between the toddler's ability to attribute referential states (e.g., emotions and perceptions) and the preschool child's ability to attribute representational mental states (e.g., beliefs). As noted above, Saxe and Wexler (2005) posed the question of whether representational mental states differentially activate selective regions of the neural substrate. If so, perhaps developmental constraints on brain function
Acknowledgements
This study was supported by the NIA (K23-AG22509, R01-AG29411), the NCRR (P41-RR14075, U24-RR021382), and the Mental Illness and Neuroscience Discovery (MIND) Institute. The authors thank Mary Foley, Larry White, and Jill Clark for technical assistance.
References (52)
- et al.
Social perception from visual cues: Role of the STS region
Trends in Cognitive Science
(2000) - et al.
When the self represents the other: A new cognitive neuroscience view on psychological identification
Consciousness and Cognition
(2003) - et al.
An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest
NeuroImage
(2006) - et al.
What does the frontomedian cortex contribute to language processing: Coherence or theory of mind?
NeuroImage
(2002) - et al.
The neural basis of mentalizing
Neuron
(2006) - 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.
Amygdala activation when one is the target of deceit: Did he lie to you or to someone else?
NeuroImage
(2006) - et al.
Inferring false beliefs from the actions of oneself and others: An fMRI study
NeuroImage
(2004) - et al.
Amygdala activation when one is the target of deceit: Did he lie to you or to someone else?
Neuroimage
(2006) - et al.
Attention to emotion modulates fMRI activity in human right superior temporal sulcus
Cognitive Brain Research
(2001)