The False Belief Task: When Children Learn That Others Can Be Wrong

Ask a typical three-year-old where Sally thinks her marble is, after you have shown the child that Anne moved it, and the three-year-old will tell you where the marble actually is. They cannot yet hold in mind that Sally, who did not see the move, has a different — and false — belief about the marble's location. Ask a typical five-year-old the same question and they will answer correctly. That shift, between the two ages, marks one of the most consequential cognitive transitions in human development. The false belief task, and the theory of mind it measures, explains how we become social beings capable of genuinely understanding other people.

What Theory of Mind Is — and Why It Matters

Before the experiment, there is a concept to define.

Theory of mind — the cognitive capacity to attribute mental states — beliefs, desires, intentions, knowledge, ignorance — to other people, and to understand that those mental states can differ from one's own and from reality.

This is not the same as empathy, though the two are related. Empathy is feeling something in response to another person's emotional state. Theory of mind is understanding, cognitively, that another person has a mental state at all — that their beliefs, knowledge, and perspective are distinct from yours and from objective reality.

The name comes from a 1978 paper by primatologists David Premack and Guy Woodruff, who asked whether chimpanzees have a theory of mind: that is, whether they attribute mental states to the humans around them. Their paper framed the question formally for the first time and triggered the research program that produced the false belief task.

False belief — a belief that an individual holds about the world that is, in fact, incorrect. The ability to understand that someone else holds a false belief is the key test of theory of mind, because it requires holding two models simultaneously: what you know to be true, and what the other person (incorrectly) believes to be true.

The analogy that clarifies why this is hard: imagine you are watching a stage play and you know the villain is about to step into a trap. The audience shouts warnings. They know something the character does not, and they can model the character's ignorance while simultaneously holding their own correct knowledge. A three-year-old cannot do this reliably. The character's ignorance doesn't register as a real mental state separate from what the three-year-old knows. Theory of mind is the capacity that makes the stage play work on us emotionally — we can inhabit other minds.

Source: Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of mind? Behavioral and Brain Sciences, 1(4), 515–526.

Wimmer and Perner: The 1983 Experiment

Austrian psychologists Heinz Wimmer and Josef Perner designed the first systematic false belief task in 1983. Their version used a narrative about a boy named Maxi and a piece of chocolate.

In the story, Maxi puts his chocolate in a green cupboard and leaves the room. While he is away, his mother moves the chocolate to a blue cupboard. The child being tested watches the whole sequence. Then comes the critical question: "Where will Maxi look for his chocolate when he comes back?"

The correct answer is the green cupboard — where Maxi left it and where he believes it still is. Maxi has a false belief that his chocolate is in the green cupboard. A child who understands false belief will answer "green cupboard." A child who does not will typically answer "blue cupboard" — the place the chocolate actually is — because they cannot separate their own knowledge from Maxi's ignorance.

Wimmer and Perner tested children aged three to nine and found a clear developmental pattern:

Age Group	Approximate % Passing False Belief Task
3–4 years	~15–20%
4–5 years	~50–60%
5–6 years	~75–85%
6–9 years	~85–90%

The shift between three and five years was the finding. Something happened cognitively in that window that allowed children to hold Maxi's false belief in mind as a genuine representation — a real mental state belonging to a real person who doesn't know what the child knows.

Source: Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition, 13(1), 103–128.

The Sally-Anne Version: Baron-Cohen, Leslie, and Frith, 1985

Two years after Wimmer and Perner, Simon Baron-Cohen, Alan Leslie, and Uta Frith adapted the false belief paradigm into what became its most recognizable form. Their goal was not primarily to study typical development — they wanted to test whether children with autism lacked theory of mind.

Their version used two dolls named Sally and Anne, acted out physically rather than narrated:

1. Sally has a basket. Anne has a box. Both are on a table.
2. Sally places a marble in her basket, then leaves the scene.
3. While Sally is gone, Anne takes the marble from Sally's basket and puts it in her box.
4. Sally returns.
5. The child is asked: "Where will Sally look for her marble?"

The correct answer is the basket — where Sally left it, where Sally believes it still is. The incorrect answer is the box — where the marble actually is.

Baron-Cohen and colleagues tested three groups:

Group	n	Mean Mental Age	% Passing Sally-Anne
Typically developing children	27	~4.5 years	85%
Children with Down syndrome	14	~5.5 years	86%
Children with autism	20	~9.3 years	20%

The autism finding was the headline result. Children with autism had a mean mental age of over nine years — well above the age at which typically developing children pass the task — yet 80% of them failed it. Their intellectual capabilities were not the limiting factor. Something specific to social cognition was impaired.

Baron-Cohen, Leslie, and Frith proposed that autism involves a specific deficit in theory of mind: the capacity to attribute mental states to others. This "mindblindness" hypothesis became one of the most influential frameworks in autism research of the subsequent four decades.

Source: Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a "theory of mind"? Cognition, 21(1), 37–46.

What Failing the Task Looks Like — and Why

When a three-year-old answers "the box" to the Sally-Anne question, they are not confused about the narrative or failing to pay attention. They understand the sequence perfectly. They know Sally left, Anne moved the marble, and the marble is now in the box. The problem is that they cannot hold Sally's outdated mental model of the marble's location as a real thing — a genuine state of Sally's mind — distinct from their own knowledge that the marble has moved.

This failure is called reality bias — the tendency to answer questions about belief by substituting what is actually true rather than what the target person believes to be true. It is not a reasoning failure in any general sense; it is a specific inability to represent mental representations.

Reality bias — in theory of mind research, the tendency of younger children to answer belief questions by reporting the actual state of the world rather than the content of the target person's belief.

Leslie's theoretical account proposed that representing false beliefs requires decoupling — the ability to hold a representation as a representation, bracketed from reality, rather than as a direct reflection of the world. A child who can decouple can maintain "Sally thinks the marble is in the basket" as a distinct propositional attitude — a belief attributed to Sally — without confusing it with the true proposition "the marble is in the box."

This is why the task is specifically hard for young children and not merely about knowledge. The child knows the correct answer about the marble's location. The task requires them to simultaneously represent someone else's wrong answer as a genuine mental state. That is a different cognitive operation.

Source: Leslie, A. M. (1987). Pretense and representation: The origins of "theory of mind." Psychological Review, 94(4), 412–426.

The Developmental Timeline in Detail

The four-year shift is robust across cultures and replication studies, but the picture is more nuanced than a simple on/off switch at age four.

Research conducted in the 1990s and 2000s using modified procedures — simpler language, more familiar contexts, reduced working memory demands — found evidence of implicit theory-of-mind understanding in children as young as 15 to 18 months. Infants show surprise (measured by looking time) when an agent acts inconsistently with what they should believe, suggesting some precursor of belief attribution exists well before children can verbally pass the standard false belief task.

This led to a distinction between implicit and explicit theory of mind:

• Implicit theory of mind: Early, automatic, non-verbal sensitivity to others' mental states, visible in looking-time paradigms from infancy
• Explicit theory of mind: Later, deliberate, verbally expressible capacity to reason about and report on others' beliefs, developing around age four

The standard Sally-Anne task measures explicit theory of mind — it requires a child to verbally report what Sally believes. The infant looking-time studies suggest that some precursor sensitivity exists much earlier. Whether those two systems are continuous developmental stages of the same capacity or genuinely different cognitive systems is still debated.

Source: Onishi, K. H., & Baillargeon, R. (2005). Do 15-month-old infants understand false beliefs? Science, 308(5719), 255–258.

Theory of Mind Across Cultures

A potential concern with the false belief paradigm is cultural specificity. The task was developed in Western academic settings. Could the age-of-passing reflect Western children's particular experience with narrative, testing situations, or conversation about mental states?

Cross-cultural research has addressed this directly. Studies in China, Japan, Canada, Peru, India, and West Africa have all found the same developmental transition occurring around age four to five. The specific age varies somewhat — children in some societies pass slightly earlier, in others slightly later — but the existence of the transition and its rough timing appear universal.

Studies of children in cultures with notably different conversational norms around mental state language — cultures where adults less frequently discuss others' beliefs and intentions in child-directed speech — do find somewhat later passing ages. This suggests that social exposure to mental-state talk influences the pace of development, even if it does not determine whether the capacity emerges at all.

Source: Callaghan, T., Rochat, P., Lillard, A., Claux, M. L., Odden, H., Itakura, S., Tapanya, S., & Singh, S. (2005). Synchrony in the onset of mental-state reasoning. Psychological Science, 16(5), 378–384.

The Autism Connection: Mindblindness and Its Limits

Baron-Cohen's mindblindness hypothesis — that autism involves a specific deficit in theory of mind — generated enormous research activity. It has also been refined and challenged substantially since 1985.

The original 1985 finding — that 80% of autistic children failed the standard Sally-Anne task — has been replicated many times. It is real. But subsequent decades of research identified important complications.

First, most autistic adults do pass the standard first-order false belief task. The deficit in explicit theory of mind in autism is developmental, not permanent. The question shifted to whether autistic individuals pass because they have acquired an explicit, laboriously learned workaround for a capacity that others acquire automatically.

Second, more demanding theory-of-mind measures — second-order false belief tasks, the "Reading the Mind in the Eyes" task, and naturalistic social situations — continue to differentiate autistic from non-autistic individuals even when both groups pass the standard Sally-Anne task. This suggests the underlying difference may be one of degree and automaticity rather than categorical absence.

Third, the spectrum is wide. Many autistic individuals demonstrate rich social understanding in some contexts. The blanket characterization of autism as "theory of mind deficit" overgeneralizes from a specific task failure to a broad social capacity.

Baron-Cohen himself has refined his position, emphasizing that autistic individuals may process social information differently — using more deliberate, explicit reasoning where others use automatic, intuitive processing — rather than simply lacking the capacity.

Source: Baron-Cohen, S. (1995). Mindblindness: An Essay on Autism and Theory of Mind. MIT Press.

Source: Senju, A. (2012). Spontaneous theory of mind and its absence in autism spectrum disorders. The Neuroscientist, 18(2), 108–113.

The Neural Basis of Theory of Mind

Neuroscience has identified a network of brain regions that activate consistently during theory-of-mind tasks, including the false belief task.

The core network includes:

Brain Region	Role in Theory of Mind
Temporoparietal junction (TPJ)	Distinguishing self from other; attributing mental states
Medial prefrontal cortex (mPFC)	Reasoning about others' thoughts and beliefs
Posterior superior temporal sulcus (pSTS)	Processing intentional agents and social cues
Precuneus	Perspective-taking and self-referential processing

The temporoparietal junction is particularly central. Research by Rebecca Saxe at MIT found that the TPJ responds more strongly to mental state content than to physical descriptions of the same scenarios. It appears to be specifically recruited for the task of representing what others know, believe, or intend — not just for processing social information generally.

The development of this network over childhood corresponds broadly to the behavioral transition measured by the false belief task, though the neural maturation extends well beyond age five and continues into adolescence.

Source: Saxe, R., & Kanwisher, N. (2003). People thinking about thinking people: The role of the TPJ in "theory of mind." NeuroImage, 19(4), 1835–1842.

Beyond Sally-Anne: What Second-Order False Belief Tells Us

The standard Sally-Anne task tests first-order false belief: what does Sally think about where the marble is? Around age six to seven, children master second-order false belief: what does one person think that another person thinks?

An example: "Mary thinks that John thinks the ice cream van is in the park." To pass a second-order task, the child must represent a belief about a belief — a recursive mental state attribution.

Second-order false belief tasks predict social understanding at a more sophisticated level and are more sensitive to differences in social cognition between autistic and non-autistic individuals who have both mastered the first-order task. The ability to follow complex social narratives, understand gossip, appreciate dramatic irony in literature, and navigate workplace politics all depend on second-order (and higher) mental state reasoning.

This is why theory of mind researchers consider the false belief task a threshold measure — passing it shows that the capacity has come online, but it says nothing about the depth or speed of the social reasoning that builds on it.

Source: Perner, J., & Wimmer, H. (1985). "John thinks that Mary thinks that...": Attribution of second-order beliefs by 5- to 10-year-old children. Journal of Experimental Child Psychology, 39(3), 437–471.

Frequently Asked Questions

Q: What is the difference between the Wimmer-Perner task and the Sally-Anne task?
A: They test the same capacity — first-order false belief — through different presentation methods. Wimmer and Perner used a narrated story about a character named Maxi. Baron-Cohen, Leslie, and Frith used physically enacted dolls named Sally and Anne. The doll version was designed to reduce language and memory demands and to be usable with autistic children who may have different responses to narrative versus enacted scenarios.

Q: Do all children with autism fail the false belief task?
A: No. In Baron-Cohen et al.'s 1985 study, 20% of autistic children passed. Subsequent research has found higher pass rates, particularly in older autistic children and those with higher verbal ability. The finding is a group-level tendency, not a universal deficit. Many autistic individuals eventually pass standard false belief tasks, though differences appear on more demanding measures.

Q: Does passing the Sally-Anne task mean a child is socially competent?
A: Not necessarily. The task is a threshold measure. Passing it indicates that the explicit capacity for false belief reasoning has come online. But social competence in the real world draws on a much richer set of abilities, including processing speed, emotional recognition, pragmatic language use, and the automatic application of mental state reasoning in fast-moving social situations.

Q: Can non-human animals pass false belief tasks?
A: This remains actively debated. Some studies using looking-time methods have found evidence that great apes, particularly chimpanzees, show sensitivity to what other agents can see and know, suggesting a precursor to theory of mind. But no non-human animal has consistently passed the explicit verbal false belief task in the form used with children. The capacity appears to reach its fullest human form only in our species.

Q: How does theory of mind relate to lying and deception?
A: Deliberate deception requires theory of mind. To successfully deceive someone, you must represent their current (mistaken) belief, predict that your action will change their belief in a specific direction, and plan accordingly. Children begin to succeed at deception tasks around the same age they pass false belief tasks — the two abilities appear to share cognitive foundations.

Q: Are there cultural differences in when children pass the false belief task?
A: Yes, but small ones. The transition occurs universally around age four to five across cultures studied to date. Cultures where adults frequently discuss mental states in conversation with children show slightly earlier passing ages, suggesting environmental input modulates the pace without determining whether the capacity emerges.

Q: What is the "Reading the Mind in the Eyes" task and how does it relate?
A: Developed by Baron-Cohen as a more demanding measure for adults, it presents photographs of the eye region of faces and asks participants to select which of four mental state words best describes what the person is feeling or thinking. Adults with autism typically score lower than non-autistic adults on this task even when both groups pass standard false belief tasks, suggesting that the underlying difference persists but becomes more subtle at higher levels of social-cognitive demand.

The Takeaway

The false belief task is a small, elegant procedure that probes one of the most consequential capacities in human cognition: the ability to hold a model of another mind as genuinely separate from both your own perspective and from objective reality. That capacity, which comes online around age four in most children, is the cognitive foundation for everything from everyday conversation to empathy, from literature to law. The finding that autistic children show a specific, selective delay in this capacity — in the absence of general intellectual impairment — reoriented a generation of autism research. The finding that the capacity has deep evolutionary roots, measurable in infants through looking-time paradigms, suggests it is not a cultural acquisition but a biological inheritance that human development brings gradually online.

---

Read next:

• The Little Albert Experiment
• The Visual Cliff Experiment
• The Iowa Gambling Task