Theory of mind is the ability to recognize that other people have their own thoughts, beliefs, desires, and feelings, and to use that understanding to predict how they’ll behave. If a coworker storms out of a meeting, you automatically infer that something upset them, even though you can’t directly observe their emotions. That mental leap, attributing an invisible internal state to another person, is theory of mind in action. The concept was first named by researchers Premack and Woodruff in 1978, originally in the context of studying chimpanzees, and has since become one of the most studied topics in cognitive science.
How Theory of Mind Works
Reading other people’s minds (in the everyday, non-mystical sense) draws on three core components. First, you rely on shared context: knowing what both you and the other person have experienced in a given situation. Second, you pick up on social cues like gaze direction, facial expressions, and tone of voice. Third, you interpret actions, working backward from what someone does to figure out what they were thinking or wanting.
Humans express their mental states through a surprisingly rich set of behaviors. A raised eyebrow, a shift in posture, a pause before answering a question: these are all data points your brain processes rapidly and mostly unconsciously. Theory of mind is what allows you to go beyond noticing these signals and actually construct a model of what’s happening inside someone else’s head.
Two Competing Explanations
Scientists have debated how we actually pull this off, and two major frameworks have emerged. The first, called theory-theory, proposes that your brain maintains a kind of informal rulebook about how minds work. You hold concepts like “belief” and “desire” and apply general principles to them: people act to satisfy their desires based on what they believe to be true. Under this view, predicting someone’s behavior is like running a logical equation.
The second framework, simulation theory, takes a different approach. It suggests that when you try to understand what someone else is thinking, you essentially run their situation through your own mind. You imagine yourself in their position and use whatever your brain produces as a rough prediction of their mental state. Rather than consulting an abstract rulebook, you’re using yourself as a model. Most researchers now think both processes play a role, depending on the situation.
When Children Develop It
Theory of mind isn’t something you’re born with. It develops gradually during early childhood, and the classic way researchers measure it is with the Sally-Anne test, first described by Simon Baron-Cohen and colleagues in 1985. In this task, a child watches a simple scenario: Sally places a marble in a basket and leaves the room. While she’s gone, Anne moves the marble to a box. When Sally comes back, the child is asked, “Where will Sally look for her marble?”
To answer correctly, the child has to understand that Sally holds a false belief. Sally doesn’t know the marble was moved, so she’ll look in the basket where she left it, not in the box where it actually is. Children typically pass this test around age 4, which marks a significant milestone in social-cognitive development. Before that age, most children answer based on reality (“the box”) rather than on Sally’s perspective. The test also includes control questions confirming the child remembers where the marble started and where it actually is, to make sure a correct answer truly reflects understanding of false belief rather than guessing.
This is called first-order false belief: understanding that someone else can believe something that isn’t true. Second-order false belief, grasping what one person thinks another person believes, develops later, typically around age 6 or 7. These nested layers of mental reasoning continue to grow more sophisticated through adolescence and into adulthood.
What Happens in the Brain
Brain imaging studies have consistently identified two regions that activate during theory of mind tasks. The medial prefrontal cortex, a region behind your forehead involved in thinking about yourself and others, shows up in roughly 93% of neuroimaging studies on theory of mind. Some researchers consider it critical to the process. The temporoparietal junction, located where the temporal and parietal lobes meet near the back and side of the brain, appears in about 58% of studies. This area seems especially important for distinguishing your own perspective from someone else’s.
Other areas, including regions involved in emotional processing and conflict monitoring, also contribute. The picture that emerges is not a single “mind-reading center” but a network of brain regions working together, with the medial prefrontal cortex and temporoparietal junction as consistent anchors.
Theory of Mind and Autism
Difficulties with theory of mind have long been recognized as one of the most significant challenges for people with autism spectrum disorder. Individuals with autism often struggle with tasks that require inferring the intentions, beliefs, or emotions of others, particularly in unfamiliar social situations. This shows up in practical ways: difficulty reading facial expressions (especially from eyes alone), trouble interpreting sarcasm or irony, and challenges with the spontaneous social back-and-forth that many people take for granted.
These difficulties persist even in individuals with high-functioning autism, who may learn explicit rules for social interaction but still find it hard to respond spontaneously to social cues. The core issue isn’t a lack of empathy or caring. It’s that the automatic process of constructing a model of another person’s mental state doesn’t work the same way. This leads to real consequences in communication, relationships, and daily social functioning.
How It’s Measured in Adults
Theory of mind doesn’t stop developing at age 4. Adults vary considerably in how well they read others, and researchers have developed several tools to measure this. The Reading the Mind in the Eyes Test presents photographs showing only the eye region of a face. You choose from four descriptors (such as “suspicious,” “pensive,” or “hostile”) to identify what the person is thinking or feeling. It’s a surprisingly difficult task that reveals real differences in social perception.
Other adult assessments include higher-order false belief tasks, which test increasingly complex layers of mental reasoning (“A thought that B wanted C to know that…”), sometimes up to seven levels deep. Strange Stories tests present brief social scenarios involving sarcasm, white lies, or deception, and ask you to explain the speaker’s intended meaning. Even simple animations of geometric shapes moving in ways that suggest goals or emotions can reveal whether someone naturally attributes mental states to what they see. These tools are used in both research and clinical settings to understand social cognition across a range of conditions, not just autism.
Do Other Animals Have It?
The question that launched the entire field, whether chimpanzees have a theory of mind, remains surprisingly unresolved. While researchers have pointed to behaviors like deception, perspective-taking, and social maneuvering as evidence that great apes understand mental states, a thorough review of the evidence found that every one of these behaviors could be explained without assuming the animals understand minds at all. Associative learning, pattern recognition, and inferences based on observable behavior (rather than invisible mental states) can account for what’s been observed.
This doesn’t mean chimpanzees definitely lack theory of mind. It means the experimental tools available haven’t been able to clearly distinguish between “this animal understands what another animal is thinking” and “this animal has learned that certain behaviors lead to certain outcomes.” The bar for demonstrating genuine mental state attribution in non-human animals remains high, and no study has definitively cleared it.
AI and Theory of Mind
Recent research has tested whether large language models can perform theory of mind tasks, and the results are striking. Some models now achieve adult-level performance on higher-order theory of mind tasks, successfully tracking nested beliefs across multiple characters in complex scenarios. This has raised the possibility of using AI to help people who struggle with social cognition, for instance by providing personalized social skills training in simulated environments for individuals with autism or intellectual disabilities.
At the same time, researchers have flagged a genuine concern: if people begin relying on AI systems for social interpretation, it could reduce opportunities for developing organic social skills and authentic human interaction. Whether AI systems truly “understand” mental states or are simply producing convincing text patterns remains an open and actively debated question.