Lecture 20: Theory of Mind and Mentalizing

What this lecture covers

This lecture asks how humans infer what other people are thinking, a uniquely human ability the lecture calls mentalizing or theory of mind. It builds the case, using behavioral tasks and fMRI, that reasoning about another person's beliefs is a distinct cognitive function rather than a byproduct of general reasoning about people, and that it depends heavily on one brain region, the right temporal parietal junction (RTPJ). It follows the course's earlier lectures on uniquely human abilities such as language and music.

After watching, you should be able to explain the false belief task and why it isolates belief attribution from other confounds, describe the evidence that the RTPJ selectively represents another person's thoughts (not their bodily sensations or appearance), and explain how moral reasoning experiments and autism studies have been used to test the causal role of this region.

Key ideas

• Mentalizing: inferring another agent's hidden mental states, specifically what they can perceive, what they believe, and what they desire, none of which are directly observable.
• False belief paradigm: a task that asks what an agent will do based on a belief the experimenter knows to be false, isolating belief attribution from simply predicting what is true in the world; the classic version is the Sally-Anne task.
• False photo task: a control task designed by Debbie Zaitchik that is logically identical to the false belief task but asks about an outdated physical photograph instead of a person's belief; children with autism spectrum disorder (ASD) fail the belief version much more than the photo version.
• Right temporal parietal junction (RTPJ): a brain region that activates more for false belief than false photo stories, and, in further experiments, responds specifically to reasoning about another person's thoughts rather than their visceral states or external appearance.
• Medial prefrontal cortex (MPFC): a region that also activates more for belief than photo tasks but, unlike the RTPJ, responds to thoughts, visceral states, and external appearance alike, suggesting it is engaged more generally by thinking about another person.
• Intentional versus accidental harm: a moral reasoning paradigm where the same outcome is judged more or less permissible depending on what the agent believed at the time, used as a test case for theory of mind.
• TMS causal test: disrupting the RTPJ with transcranial magnetic stimulation reduces the moral distinction people draw between intentional and accidental harm, supporting a causal role for the region.

Walkthrough

Why social cognition and theory of mind matter (0:10)

The lecture opens by arguing that social interaction dominates human experience and may have shaped the evolution of the human brain. A video of 18-month-old infants spontaneously helping an unfamiliar adult is used to show that even very young children infer another person's intentions and goals from nonverbal behavior alone.

What mentalizing requires: percepts, beliefs, and desires (8:23)

Using the example of a character named Romeo reaching for a bottle, the lecture distinguishes an observable action (reaching) from the layers of hidden mental state needed to explain it: an intention to drink, a belief that the liquid is poison, and a further belief and desire behind that. It stresses that beliefs and desires cannot be seen directly and must be inferred, and that this ability is essentially absent in current computer systems and severely limited in animals.

The false belief task and its developmental time course (14:25)

The Sally-Anne task is introduced as the classic method for testing belief attribution: because it uses a false belief rather than a true one, it separates predictions based on another agent's mind from predictions based on the true state of the world. A video of Rebecca Saxe running the task with children shows 3-year-olds failing it while 5-year-olds pass, and shows a further shift around age 7 in how children make moral judgments about the same scenario.

Isolating belief attribution with the false photo task (23:30)

To rule out alternative explanations for why children or people with autism fail the false belief task, such as general difficulty holding two conflicting representations in mind, the lecture presents Zaitchik's false photo task, a logically identical scenario built around an outdated photograph rather than a belief. Children with ASD fail the belief version much more than the photo version, supporting the idea that belief attribution specifically, not the general format of the task, is what is difficult.

Locating theory of mind in the brain (28:31)

An fMRI experiment scanning people reading false belief and false photo stories identifies two responsive regions: the RTPJ and the MPFC. A follow-up experiment with three story types, external appearance, visceral sensation, and thoughts, shows that the RTPJ responds selectively to stories about another person's thoughts, while the MPFC responds broadly across all three conditions. The lecture also describes a generalization experiment using a wordless Pixar short film, showing the same RTPJ activation pattern without any language, which supports the claim that the region reflects a deep cognitive process rather than verbal reasoning.

Moral reasoning as a test case, and the RTPJ in autism (47:03)

Using a scenario where a character named Grace poisons a friend's coffee either by accident or knowingly, the lecture shows that typical adults judge accidental harm as far more permissible than intentional harm, a judgment that depends on tracking what the agent believed. People with autism show a smaller gap between the two judgments, and disrupting the RTPJ with TMS in neurotypical subjects reduces the same gap, pointing to a causal role for the region. A large study then found that the RTPJ's size, location, and overall activation look the same in high-functioning autistic adults and neurotypical controls, but a pattern-based (MVPA) analysis showed that only in neurotypical subjects does the RTPJ's activation pattern distinguish intentional from accidental harm.

Before you watch

• Review earlier lectures on multivariate pattern analysis (MVPA) and functional localizers, since both are used to interpret the RTPJ findings here.
• Some familiarity with the fusiform face area discussion from the vision lectures is useful, since the lecture compares the RTPJ's selectivity to it.

Check your understanding

1. Why does the false belief paradigm use a false rather than a true belief to test whether someone can represent another agent's mind?
2. How does the false photo task rule out the explanation that children with autism simply struggle with holding two conflicting representations in mind?
3. What is the key difference in how the RTPJ and the MPFC responded across the external appearance, visceral state, and thought conditions?
4. Why did generalizing the theory-of-mind experiment to a wordless Pixar film strengthen the interpretation of the RTPJ's role?
5. What did the pattern-based (MVPA) analysis reveal about the RTPJ in autism that a standard activation-magnitude analysis did not?