Constructed Emotion

What the science says

The classical view of emotion, dominant in Western psychology for decades, holds that emotions are discrete, universal, and hardwired. Each emotion (fear, anger, sadness, disgust, happiness, surprise) has a dedicated neural circuit, a characteristic physiological signature, and a universal facial expression readable across cultures. This view was systematized by Sylvan Tomkins and operationalized by Paul Ekman's cross-cultural expression research.

Lisa Feldman Barrett's meta-analyses challenge this picture. Looking at the full body of emotion research:

• The same internal state can produce different facial expressions depending on context
• Facial expressions don't reliably indicate internal emotional states across individuals or contexts
• There are no neurological "fingerprints" unique to specific emotions — the same brain regions are implicated across multiple emotions
• Emotional experience varies substantially across individuals and cultures in ways the classical view dismisses as noise

Barrett's alternative is the theory of constructed emotion. The brain builds emotional experience from three components:

Core affect: The brain continuously receives interoceptive signals — information about the body's internal state. These signals are undifferentiated: they carry valence (pleasant/unpleasant) and arousal (activation level), but no label, no category, no narrative. This is the raw material — what you might notice as "I feel off" or "something feels good" before you know what it is. Core affect is universal; it is the constructed layer that varies.

Emotion concepts: Through social learning, the brain acquires conceptual categories — "anger," "anxiety," "contempt," "awe," "schadenfreude" — that it uses to make sense of core affect in context. These are not passive labels applied after the fact; they are active categorization tools the brain uses during construction. Without the concept, the discrete emotion doesn't arise as a distinct experience. This is why cultures genuinely differ in what emotions they have — the Portuguese saudade, the Japanese amae, the German Schadenfreude are not just different words for universal experiences; they are different constructions.

Predictive simulation: Consistent with the predictive processing framework, the brain constructs emotion by running simulations — drawing on past emotional experiences to predict what the current interoceptive state means and what action it calls for. The emotion is the brain's best guess about what is causing the body's current state and what to do about it. This is why context matters so much: the same physiological arousal state can be constructed as excitement, anxiety, or anger depending on what the brain predicts is the cause.

This account integrates with both predictive processing (Barrett works explicitly in Friston's framework) and allostatic regulation: emotions serve body budget function — they are the brain's primary mechanism for managing physiological resources in anticipation of demands, not an epiphenomenon riding on top of a separate regulatory process.

Holism, not reductionism

Barrett et al. (2025) make explicit what the books implied: the TCE is a holistic theory, not a reductionist one. Basic Emotion Theory is reductionist — it understands an emotion by reducing it to its simpler causes (a gene, a circuit, a physiological type). The TCE understands the causes in terms of the whole: an instance of emotion is a brain-wide causal ensemble whose simultaneous features include allostatic regulation, skeletomotor action, and lived experience. These features cannot be meaningfully separated into distinct phenomena caused by separate mechanisms.

The specific neuroanatomical basis: allostatic control neurons (including those in the anterior cingulate cortex) simultaneously send motor control signals to the spinal cord and prediction signals that construct features of lived experience. These efferent copies of the same allostatic signals mean that the feeling is generated by the same neural event that generates the bodily state. Allostasis and feeling are the same process, not cause and effect.

Population thinking, not typological thinking

Barrett et al. (2025) introduce the distinction from evolutionary biologist Ernst Mayr: Basic Emotion Theory uses typological thinking — each emotion category has an immutable prototype that defines all instances. The TCE uses population thinking — the prototype is a statistical simplification that doesn't exist in any actual instance; variation is real and structured by context.

The basic unit is the ad hoc category: something you do, not something you have. Each instance of ANGER begins as a population of possible prediction signals — possible futures for allostasis and action in this specific situation — and collapses into a single contextualized instance. The emotion word is an abstract feature that trains the brain to generalize across physically dissimilar instances that serve the same allostatic function, not a label stuck on a pre-formed state.

What the evidence does NOT support

• That emotions are purely arbitrary cultural constructions. Core affect (valence + arousal) is universal and biological. The constructive layer operates within real biological constraints.
• That the classical view is entirely wrong. Ekman's cross-cultural data is real; Barrett disputes the interpretation (universality of expression), not the existence of the data. Some degree of biological preparedness for threat-relevant processing is consistent with her account.
• That constructed emotion theory is settled science. The debate between classical and constructionist views is active in affective neuroscience. Paul Ekman, Jaak Panksepp, and others have offered substantive critiques.
• That because emotions are constructed, they are consciously controllable. Construction happens largely below conscious access. You can influence the inputs (body state, available concepts, context) but not the construction process itself in the moment.
• That emotions are illusions. Emotion categories are real — but real the way money is real: they require a perceiver for existence and have no physical essence shared across all instances. Currencies throughout human history share no physical similarity; money is still real. The constructed view denies neural essences, not reality. (Barrett, 2017 SCAN)
• That brain networks are Lego blocks with fixed functions. Networks have functional motifs — repertoires of possible configurations constrained by structural connectivity — not a single essential function they always perform. The default mode network is not the simulation machine; the salience network is not the attention selector. Each has a repertoire; context determines which configuration is active.
• That automaticity means a "System 1" and conscious control means a "System 2." Automaticity and conscious control are different brain modes, each achievable via multiple network configurations. This is not the same as two anatomically distinct systems in competition. The dual-process framing locates function where it doesn't reside.

Buddhist parallel

The constructed emotion account has several credible Buddhist parallels.

Vedanā — one of the five aggregates — refers to the hedonic tone of experience: pleasant, unpleasant, or neutral, arising with every moment of sensory contact. It is pre-conceptual and pre-intentional: the bare quality of pleasantness or unpleasantness before the mind elaborates. This maps closely onto core affect. Vedanā is the raw material; what the mind does with it is a subsequent process.

Papañca (conceptual proliferation) refers to the mind's tendency to elaborate raw experience into conceptual chains — narrative, judgment, self-reference. This is recognizable as the construction layer: the transformation of undifferentiated core affect into full emotional experience through conceptual categorization. In Buddhist framing, papañca is a process to be seen clearly and potentially loosened, not just a cognitive fact about how the mind works.

Anattā (non-self) is relevant in a specific way: constructed emotion theory denies that there is a fixed emotional self with stable, hardwired responses waiting to be triggered. Emotional character is constructed from experience, concept, and context — it is not a pre-given nature. The Buddhist claim is stronger (no self at all) than Barrett's (no hardwired emotion circuits), but they point in the same direction.

The strain in the mapping: Buddhist practice, particularly Vipassana, aims at loosening the grip of papañca — seeing through conceptual elaboration to what's actually present. Barrett's account implies intervention through concept expansion (developing higher granularity, acquiring new emotion categories). These are different therapeutic logics operating on the same mechanism: one expands the conceptual repertoire, the other investigates the constructed nature of any concept. They are not contradictory, but they are not the same.

Key sources

• Barrett, 2017 — primary source for this theory; accessible treatment of core affect, emotion concepts, and body budget
• Barrett, 2017 (SCAN) — formal peer-reviewed statement; active inference account of interoception and categorization; source for anti-misconception list and mental inference fallacy
• Barrett et al., 2025 — updated theoretical statement; holism vs. reductionism in BET; population thinking; efferent copies mechanism
• Russell, James A. "A circumplex model of affect." Journal of Personality and Social Psychology 39, no. 6 (1980): 1161–1178. — foundational for core affect
• Barrett, Lisa Feldman. "Are emotions natural kinds?" Perspectives on Psychological Science 1, no. 1 (2006): 28–58. — earlier theoretical statement
• Lindquist, Kristen A., and Lisa Feldman Barrett. "A functional architecture of the human brain: emerging insights from the science of emotion." Trends in Cognitive Sciences 16, no. 11 (2012): 533–540.