A review of Lisa Feldman Barrett's theory of constructed emotion — what the old model got wrong, what the new science shows, and why it changes everything about how we understand our inner lives.
For most of the twentieth century, the dominant model of emotion went something like this: emotions are universal, biological categories. Fear, anger, sadness, joy, disgust, surprise — six basic emotions, wired into the brain, expressed the same way across all cultures and all people. Each one has a dedicated neural circuit, a characteristic physiological signature, a recognizable facial expression. The brain reads the body, detects an emotion, and reports it. Emotions happen to you. Your job is to receive them accurately and manage them well.
This model is intuitive. It maps onto how emotions feel — sudden, involuntary, arriving from somewhere outside conscious control. It also has decades of research behind it, from Paul Ekman's facial expression studies to lesion research identifying brain regions associated with fear. It is what most of us were taught. It is what most clinical training still assumes.
It is also, according to Lisa Feldman Barrett, largely wrong.
Barrett spent years trying to find the neural fingerprints of basic emotions. The assumption was that fear would have a consistent, identifiable pattern of brain activity — that you could look at a scan and say: this person is afraid. The research kept failing to produce that fingerprint. Fear activated different regions in different people, in different contexts, with different intensities. The same region that lit up for fear also lit up for excitement, for pain, for effort, for anticipation.
At some point Barrett realized the problem wasn't the data. The problem was the question. Researchers had been looking for dedicated emotion circuits because they assumed emotions were discrete biological categories waiting to be found. They were searching for concepts — and then finding what they were looking for, because that is what concept-driven research does. When you design a study around the category "fear," you find activation associated with fear. That doesn't mean fear is a natural category in the brain. It might just mean your experimental design was organized around a concept.
The brain does not come pre-loaded with emotion categories. It builds them — from the raw material of body states, past experience, and the concepts a culture has provided for making sense of inner life.
What Barrett and her colleagues found when they looked without the assumption of discrete categories was messier and more interesting. There is no single "fear circuit," no dedicated "sadness center." What the brain does have is an interoceptive network — a system for receiving and processing signals from the body — and a prediction system that is constantly trying to make sense of those signals by comparing them to what it has experienced before. Emotions emerge from that process. They are not detected. They are constructed.
The body is generating signals constantly. Heart rate, respiration, muscle tension, gut activity, hormonal state — an unrelenting stream of information traveling up into the brain through the interoceptive network. This is the bottom-up signal: raw, undifferentiated, arriving before any interpretation has occurred.
Crucially, this signal does not arrive labeled. The body does not send up a message saying "this is fear" or "this is excitement." It sends up something more like: intensity, arousal, quality of tension, rate of change. The brain receives this and has to figure out what it means.
Before it figures out what it means, what you experience is affect — the background valence and arousal that color everything. Affect is not an emotion. It is more like the weather of the body: pleasant or unpleasant, activated or calm. You are always in some affective state. It is always influencing how you perceive and interpret what is happening. But it is not yet an emotion. An emotion is what happens next.
The brain is not a passive receiver of information. It is, as Barrett describes it, a prediction machine — constantly generating models of what is happening and updating them when reality doesn't match. When interoceptive signals arrive, the brain does not simply read them. It asks: given everything I know — this context, this history, this body state — what is the most likely explanation for these signals?
This is where the default mode network comes in. The DMN is the brain's concept library — the accumulated store of past experiences, social knowledge, linguistic categories, and cultural frameworks that the brain uses to make predictions. When a strong interoceptive signal arrives, the DMN reaches for the concept that best fits: given this context, given this body state, given what I have experienced before in situations like this — this is probably fear. Or grief. Or excitement. Or love.
That concept-capture is not a neutral description. It is a construction. The concept shapes the experience — it determines what the signal means, how intense it feels, what memories get activated, what actions become available. The same raw interoceptive signal can be constructed into completely different emotions depending on the concept that captures it. High heart rate, shallow breathing, muscle tension — that is fear at the dentist, excitement before a performance, anger in a conflict, love at the beginning of a relationship. The body state is similar. The constructed experience is entirely different.
This is not a philosophical abstraction. It has direct, measurable consequences. People with more precise emotional concept vocabularies — what Barrett calls higher emotional granularity — regulate their emotions more effectively, seek medical care less often, use medications less, and recover from setbacks more quickly. Not because they understand themselves better in some general sense. Because the concept that captures the signal determines what response options are available.
A vague concept collapses the signal into a broad category with limited options: I feel bad. A precise concept opens it: I feel the specific quality of anticipatory dread that comes from a threat to a bond I care about. The second one points somewhere. The first one just sits there.
Underneath all of this is something even more fundamental. Barrett, drawing on the work of Peter Sterling, argues that the brain's primary job is not emotion regulation. It is not even perception. It is allostasis — the continuous management of the body's energy resources. The brain is constantly predicting what the body will need, allocating resources accordingly, and adjusting those allocations as conditions change.
Affect — that background weather of pleasant/unpleasant, calm/activated — is largely a readout of the body budget. When the budget is running low, everything feels harder, more threatening, more urgent. A deficit in the body budget is experienced as negative affect. A surplus is experienced as positive affect. This is not metaphorical. The pleasant feeling of being well-rested and well-fed and safe is the felt experience of a body budget in good standing. The grinding dysphoria of chronic depletion is the felt experience of a body budget under sustained pressure.
This matters enormously for how we understand emotional experience. Much of what we interpret as emotional reactivity — increased sensitivity, lower tolerance, difficulty with regulation — is not primarily an emotion problem. It is a budget problem. The same signal, arriving at a depleted system, will be constructed into a more intense, more threatening emotional experience than the same signal arriving at a well-resourced one. The emotion is downstream of the budget. Treating the emotion without addressing the budget is working at the wrong level.
One of Barrett's most practically significant findings is that the emotion concepts you have available — and use habitually — are not just ways of describing experience after the fact. They are part of how the experience is constructed in the first place.
The brain uses past emotional experiences to predict current ones. If your concept of "anger" has been built from thousands of instances of a specific kind of experience, that concept will be applied readily and will shape what gets constructed. If you only have broad, coarse categories — sad, bad, angry, fine — your interoceptive signals will be captured by those broad categories, and the resulting experience will be less differentiated, less informative, and harder to work with.
This means that expanding your emotional vocabulary is not a sentimental exercise. It is a genuine intervention at the level of construction. Learning more precise concepts — not just "anxious" but "the specific quality of threat to belonging that arrives when someone I care about goes quiet" — creates new prediction possibilities. The brain has new options for what to build with the signal it receives. Different concept, different experience, different response options.
You are not discovering your emotions. You are building them — with the concepts you have available, in the context you're in, from the body state you bring to the moment. Change any of those three things and you change what gets built.
Barrett's theory of constructed emotion is not a minor update to the old model. It is a paradigm shift that has not yet fully reached clinical practice, education, or popular understanding of mental health.
If emotions are constructions, then the goal of emotional intelligence is not to receive them more accurately — it is to build them more wisely. That requires a different set of skills: interoceptive awareness (actually noticing what the body is sending up before the concept captures it), emotional granularity (having precise enough concepts to build with), and understanding of how context, body budget, and prior experience are shaping what gets built.
It also means that working "on emotions" — trying to feel differently, reframe experiences, regulate responses — is working downstream. The construction has already happened. What can be worked with more directly is the interoceptive signal before capture, the concepts available for capture, and the body state that is shaping the whole process.
The emotion is not the problem. The emotion is already a solution — the brain's best current prediction of what is happening and what to do about it. The question is whether it is the most accurate, most useful solution available given what the body is actually sending up. And that question can only be answered upstream — before the concept arrives and the experience solidifies around it.
Barrett, L.F. (2017). How Emotions Are Made: The Secret Life of the Brain. Houghton Mifflin Harcourt. — The full argument, accessible and rigorous.
Barrett, L.F. (2006). Are emotions natural kinds? Perspectives on Psychological Science, 1(1), 28–58. — The foundational challenge to the basic emotion model.
Kashdan, T.B., Barrett, L.F., & McKnight, P.E. (2015). Unpacking emotion differentiation. Current Directions in Psychological Science, 24(1), 10–16. — The emotional granularity findings.
Sterling, P. (2012). Allostasis: A model of predictive regulation. Physiology & Behavior, 106(1), 5–15. — The body budget model Barrett builds on.