A two-year-old watches you place a cookie in box A. You leave the room. While you're gone, the researcher moves the cookie to box B. You return. The child watches your face as you scan the boxes. You obviously look for the cookie in box A — where you placed it. The child knows with certainty it's in box B.
But ask the child, "When he comes back, where will he look for the cookie?" The two-year-old points to box B. She knows it's there, therefore everyone knows it's there. Her mind cannot yet contain the possibility that someone's beliefs can be different from reality.
Around age three or four, something shifts. The same setup. The same question. Now the child points to box A. "He'll look there because he doesn't know it moved." The child can reason about something she knows to be false. She can model someone else's incorrect belief. She has acquired Theory of Mind — the ability to hold in her awareness that other people have separate mental states, different knowledge, different beliefs, different desires than she does.1
This is not a trivial achievement. It means you can now predict what someone will do based not on what's true, but on what they believe to be true. You can manipulate them by controlling what they believe. You can empathize with them by understanding their perspective is genuinely different from yours. You can lie — create a belief in someone else's mind that contradicts reality.
Multiple brain regions collectively support Theory of Mind. The ventromedial prefrontal cortex (vmPFC) is involved — the same emotional integration region that processes your own social decisions. But the critical regions are newer to the neuroscience understanding: the temporoparietal junction (TPJ) and nearby regions in the temporal cortex, the precuneus, and the superior temporal sulcus. These regions form a circuit dedicated to modeling other people's mental states.2
The temporoparietal junction is particularly fascinating. It sits at the intersection of multiple sensory and cognitive systems — a crossroads for integrating information about what others are doing, where they're looking, what they're paying attention to. When you need to figure out what someone's thinking, your TPJ activates. Temporarily inactivate the TPJ, and people stop incorporating others' intentions into their moral judgments. They'll blame someone for an accident as harshly as an intentional harm — because they can't model the person's mental state (intention vs. accident) anymore.3
The superior temporal sulcus (STS), running along the side of the brain, tracks biological motion and social cues: the direction someone is looking, the direction they're moving, the gestures they're making. This region feeds information to the TPJ and vmPFC, creating an ongoing model of "what is this person doing, and what are they aware of?"4
These regions don't work in isolation. They work as a system: the temporal-parietal-prefrontal circuit for understanding other minds. Damage to any of these regions degrades Theory of Mind. Autistic individuals, who have limited Theory of Mind abilities, show reduced gray matter and activity in these regions, particularly the superior temporal sulcus.5
Theory of Mind unfolds in stages, shaped as much by social motivation as by neural maturation.
By nine months, babies understand gaze following — if someone looks somewhere, there's probably something interesting there. This is not yet ToM; it's just attending to where others attend.
By three to four years, children pass the false belief test. They can model a discrepancy between what they know and what someone else believes.
By five to seven years, children grasp irony — understanding that someone said the opposite of what they mean, requiring you to model both their literal statement and their intended meaning.
Secondary Theory of Mind arrives later: understanding person A's theory about person B's mind. "He thinks she doesn't like him, but she actually does." This requires nested mental modeling — not just understanding one person's beliefs, but understanding what one person thinks about another person's beliefs.6
What's remarkable is that development isn't purely cognitive. Social factors reshape ToM capacity in real time. Children demonstrate more sophisticated ToM when interacting with peers than when interacting with researchers. They show better ToM when something motivating — food, for instance — is at stake. They develop ToM faster if they have older siblings (practice modeling complex minds). Children in cultures emphasizing individualism develop explicit ToM later than children in collectivist cultures, but they show equivalent sophistication — just activated through different neural circuits, because they practice modeling minds in different social contexts.7
This reveals something crucial: ToM isn't purely about the neural hardware. It's about practice. Children in collectivist cultures spend more time modeling others' perspectives (a core value of the culture), so their minds become exquisitely sensitive to what others think. Children in individualist cultures practice this less in everyday social interaction, so they develop explicit ToM more slowly. But when tested, they're equally capable — they just use different neural strategies.8
Here's the disturbing part: Theory of Mind doesn't require empathy. A sociopath can have perfect Theory of Mind — understanding exactly what someone believes, feels, wants, fears — and use that understanding to manipulate, coerce, and harm.
The sociopath uses ToM to navigate social situations with frightening precision. They model your mind to exploit you. They understand what will make you trust them, what will trigger your guilt, what you value most. They can describe your emotions accurately while feeling nothing themselves.
Conversely, ToM is not strictly necessary for empathy. A toddler too young for formal Theory of Mind will still try to comfort someone who's crying, offering their pacifier (though the comfort might be more about stopping the distressing sound than understanding the other person's pain). Very young children show rudiments of empathy — a form of feeling-with that precedes the cognitive sophistication of understanding what the other person thinks.
But the mature form of empathy — truly understanding and resonating with someone's experience because you comprehend their perspective, their constraints, their impossible situation — requires ToM. It requires being able to hold in awareness: "This person has a different mind than me, with different knowledge and fears and hopes, and their actions make sense from inside their perspective even if I wouldn't choose them."9
As children develop, their empathy shifts from concrete to abstract, paralleling changes in their ToM regions. When a seven-year-old watches someone experiencing pain (a finger being poked), the empathic response is still concrete: their own sensory cortex activates (imagining their own finger), their motor cortex activates (tensing their own finger), and their periaqueductal gray (the pain processing center) fires.
But in older children and adolescents, the pattern changes. The vmPFC becomes more engaged, coupled to limbic structures processing emotion. By adolescence, when watching someone in pain, activation is strongest in the vmPFC and regions supporting Theory of Mind — not in the sensory cortex. The adolescent is understanding why the pain matters to that person, what it means to them, how their life is changed by it — not just simulating the sensation in their own body.10
This shift from concrete resonance to abstract understanding allows for a deeper form of empathy but also creates a vulnerability: you can understand someone's pain intellectually while remaining emotionally detached. The vmPFC-ToM circuit allows you to comprehend suffering without being overwhelmed by it — but it also allows you to comprehend suffering and decide it's not your problem.
An adult's prefrontal cortex gives them the capacity for compassionate action (you can feel empathy and think strategically about helping). The cost is that the same detachment makes it easier to turn away.
ToM as Gateway vs. ToM as Instrument: Theory of Mind enables both empathy (understanding others' perspectives) and exploitation (manipulating others' beliefs). The same cognitive capacity serves opposite ends. The distinction lies not in ToM itself but in what you do with it — whether you use the understanding to increase others' wellbeing or your own advantage.
Explicit vs. Implicit ToM: Children in individualist vs. collectivist cultures develop explicit ToM at different rates, yet both reach equivalent sophistication. This suggests that "ToM capacity" is not a single skill but a collection of social-cognitive practices that different cultures emphasize at different ages. The neural circuits are flexible.
Development as Maturation vs. Development as Cultural Learning: Piaget framed ToM development as stages of cognitive maturation, but social context reshapes development dramatically. The tension reveals that "development" is not a fixed biological program but an interaction between neural capacity and cultural practice.
The behavioral-mechanics counterpart to understanding Theory of Mind neurobiologically is recognizing it as the capacity for tactical understanding of others. If ToM allows you to model what someone believes, wants, fears, and values, then understanding ToM reveals how manipulation works: by crafting beliefs in someone else's mind.
A skilled manipulator is someone with exceptionally developed ToM who uses that understanding not to increase another person's wellbeing but to serve their own interests. They understand which buttons to push (what the target fears, what they value, what they're ashamed of). They create narratives that fit the target's existing beliefs while shifting their conclusions. They know what the target doesn't know and exploit that ignorance.
The neurobiological understanding of ToM reveals why it's so hard to resist manipulation: your TPJ is constantly modeling what others think, making you vulnerable to someone who understands your modeling process better than you do. Conversely, developing explicit awareness of your own ToM — noticing when you're modeling someone else's mind — creates space for skepticism. "Am I adopting this belief because it's true, or because this person has skillfully crafted it to fit what I already think?"
Buddhist and contemplative traditions emphasize "perspective-taking" as central to spiritual development — the capacity to step outside your own mind and see from another's vantage. This is neurologically identical to Theory of Mind. The practice of imagining "what is this person experiencing?" activates the TPJ and vmPFC.
But contemplative practice adds a layer: it uses perspective-taking not to manipulate or even primarily to empathize, but to dismantle the illusion of a fixed self. By repeatedly stepping into others' perspectives, you begin to see your own perspective as just one viewpoint among infinite viewpoints. You're practicing the dissolution of boundaries between self and other.
The Buddhist concept of anatta (non-self) is theoretically compatible with the neurobiology of ToM: your sense of being a unified, separate self is a construction your brain maintains. By practicing perspective-taking (activating vmPFC/TPJ), you're revealing that other minds are equally constructed, equally real, equally valid. The spiritual path and the neurobiology converge: both dissolve the illusion of an isolated, separate self by forcing the mind to hold multiple perspectives simultaneously.