Social Cognition and Theory of Mind: False Beliefs and Mentalizing

Knowing That Others Do Not Know What You Know

A three-year-old child, watching a puppet hide a marble in a basket while another puppet is absent, will predict that the returning puppet will look for the marble where the first puppet left it — not where the child knows the marble actually is. That failure reveals one of the most important milestones in human cognitive development: the acquisition of a genuine theory of mind — the ability to attribute mental states (beliefs, desires, intentions, knowledge) to others and use those attributed states to predict behavior. The shift from failing to passing the false belief task between ages 3 and 5 marks the point where children begin reasoning about minds rather than just behavior.

The Wimmer-Perner 1983 False Belief Task

Heinz Wimmer and Josef Perner, developmental psychologists at the University of Salzburg, published the foundational false belief task study in 1983 in Cognition. Their original version used a character named Maxi who places chocolate in location X, then leaves the room while his mother moves the chocolate to location Y. Children are asked where Maxi will look for his chocolate when he returns. The correct answer — location X, where Maxi believes it to be — requires suppressing what the child knows (the chocolate is in Y) and representing Maxi's false belief.

Wimmer and Perner found that children younger than approximately 4–5 years of age consistently predicted Maxi would look in location Y (the actual location), indicating they could not yet distinguish their own knowledge from another person's belief state. This inability to pass false belief tasks — treating others as having access to the same knowledge the self has — was termed failure to attribute false beliefs, or, in later terminology, limited theory of mind. The study generated over 400 subsequent replications and variations across cultures, languages, and populations, establishing false belief performance as the operationalized benchmark for theory of mind assessment.

The Sally-Anne Test and Autism

Simon Baron-Cohen, Alan Leslie, and Uta Frith adapted the false belief paradigm into the Sally-Anne test in a landmark 1985 paper in Cognition, explicitly designed to test theory of mind in children with autism. The task used two dolls: Sally places a marble in her basket, then leaves; Anne moves the marble to her box. Children are asked: "Where will Sally look for her marble?"

The results were striking. Eighty percent of typically developing 4-year-olds answered correctly (Sally's basket). Eighty-six percent of children with Down syndrome answered correctly. But only 20% of autistic children answered correctly — despite the autistic participants having higher mental ages than the Down syndrome group on standardized tests. Baron-Cohen, Leslie, and Frith concluded that autism involves a specific deficit in theory of mind — a failure of the cognitive mechanism for representing mental states — rather than a general intellectual deficit.

This "mindblindness" hypothesis (later elaborated by Baron-Cohen in his 1995 book of the same name) was enormously influential. It reframed autism as a social-cognitive impairment rather than an emotional or motivational disorder, directing research toward theory of mind mechanisms. Subsequent work has refined this picture: many autistic adults do pass explicit false belief tasks, though they show differences in spontaneous mentalizing (using mental state reasoning automatically in real-time social contexts) and in more demanding second-order false belief tasks ("What does A think B believes?"). The "double empathy problem" — proposed by Damian Milton in 2012 — offers an alternative framing: autistic and non-autistic individuals have mutual difficulty understanding each other's mental states, making mentalizing difficulty bidirectional rather than a unidirectional deficit in autism alone.

Dennett's Intentional Stance

Philosopher Daniel Dennett distinguished three stances one can adopt toward a system: the physical stance (predicting behavior from physical laws), the design stance (predicting from function and purpose), and the intentional stance (predicting behavior by attributing beliefs, desires, and rationality to the system). The intentional stance is predictively powerful whenever a system is sufficiently complex: it works for humans, chess computers, thermostats, and corporations. Dennett's crucial point is that adopting the intentional stance does not commit one to claiming that the target system has genuine mental states — it is an instrumental strategy that works when it works. The question of whether humans have "real" beliefs that justify the intentional stance is, for Dennett, a separate (and less important) question from the pragmatic utility of the stance.

• Dennett's framework is deflationary: it does not require solving the hard problem of consciousness to explain why theory of mind is useful and how it evolved
• The intentional stance applies to entities without consciousness — explaining why humans attribute goals to robots, weather systems, and sports teams
• Theory of mind, on Dennett's view, is the cognitive implementation of the intentional stance — a specialized prediction system that evolved because accurately modeling others' mental states confers major fitness advantages in social species
• Dennett's The Intentional Stance (1987) and Brainstorms (1978) remain primary references in philosophy of mind

Mirror Neurons: The Overclaim

Mirror neurons — neurons in macaque premotor cortex (area F5) that fire both when the monkey performs an action and when it observes another performing the same action — were discovered by Giacomo Rizzolatti's group at the University of Parma in the early 1990s. The original 1996 Brain paper documenting them triggered enormous extrapolation: mirror neurons were described as the basis of empathy, language evolution, autism (the "broken mirror" hypothesis), imitation, and social cognition generally.

Cognitive neuroscientist Gregory Hickok systematically critiqued these claims in his 2009 Nature Reviews Neuroscience paper and his 2014 book The Myth of Mirror Neurons. Hickok's challenges are empirically grounded. First, mirror neurons in macaques do not respond to most human communicative gestures — they respond to transitive, object-directed actions (grasping), not pantomime, speech-related movements, or social interactions generally. Second, the mirror neuron system is not necessary for imitation: patients with lesions to mirror-neuron regions (premotor cortex) can still imitate; patients without such lesions can show severe imitation deficits. Third, the broken mirror hypothesis for autism has not been confirmed: fMRI and EEG studies of mirror system activity in autism have produced inconsistent results, and behavioral evidence does not support a global imitation deficit tied to mirror dysfunction.

The Mentalizing Network

Neuroimaging studies of theory of mind tasks consistently implicate a distributed brain network distinct from the mirror neuron system. This mentalizing network includes:

The TPJ is the most reliably activated region in false belief tasks across neuroimaging studies. Rebecca Saxe and Nancy Kanwisher's 2003 NeuroImage paper showed that TPJ activation is specific to false belief understanding — not to general reasoning about story characters or physical causality. TMS disruption of right TPJ impairs false belief task performance without affecting matched control tasks in healthy adults, establishing the TPJ as causally necessary for representing others' beliefs. The mentalizing network overlaps substantially with the default mode network, consistent with findings that social cognition and spontaneous self-generated thought share common neural substrates.