How to Understand Emotions | Dr. Lisa Feldman Barrett

The conversation begins by challenging the assumption that science has a settled definition of emotion. Lisa Feldman Barrett explains that researchers have not agreed on one for roughly 150 years, partly because the proposed building blocks of emotion—changes in the body, brain activity, facial movement, and action—also occur during ordinary moments. A coordinated response therefore does not by itself distinguish fear, anger, sadness, or any other named state.

The older model searched for diagnostic patterns: fear was expected to produce a characteristic heart-rate change, action tendency, and facial configuration. Barrett argues that the evidence does not support a one-to-one mapping. A widened-eyed, open-mouthed face can have different meanings across cultures, and the same person can make similar facial movements in different situations. The problem is not that emotional experiences are unreal; it is that no single physical marker reliably separates one emotion category from another.

Barrett calls the question of whether a facial-expression system feeds an emotion system ill-posed because it assumes two discrete systems that have not been established. There is circuitry for moving the facial muscles, but a movement is not the same thing as an expression. An expression is an interpretation of what a movement means, and that meaning depends on the surrounding sensory situation.

Seeing someone cry does not give an observer direct access to that person’s emotional state. The brain combines facial movement and tears with sounds, smells, context, and signals from the observer’s own body. A cry is not acoustically specific to hunger or fatigue, and a context-free image can invite a stereotyped judgment that changes when the wider scene is revealed. Barrett therefore describes perception as inference: brains use ensembles of signals to predict what a movement means rather than reading an emotion directly from a face.

There is feedback from the face to the brain, but it is not a special route reserved for emotion. The brain sends motor commands to the face, the body sends sensory signals back, and the brain uses learned patterns to predict what will happen next. Smiling can therefore participate in a larger pattern that is associated with a pleasant state, but simply posing a smile is not a mechanical switch that activates an emotion system. This account also rejects a simplistic interpretation of posture studies in which taking up space or shrinking the body automatically produces empowerment or depression.

Barrett places the classic universal-expression account in a broader network view of the nervous system. Cortisol is not inherently bad, and body posture cannot be treated as an isolated switch that makes someone powerful or depressed. Claims about facial expressions likewise need to be separated from the interpretations researchers attach to observed movements.

Her own change in view came from trying to use stereotyped faces in experiments. Participants often could not identify a widened-eyed, gasping face as fear, and a five-person consensus group later reviewed more than 1,000 papers over two and a half years. The group concluded that the evidence did not support universal facial expressions of emotion. It did support substantial variability within and across situations.

The disagreement becomes concrete in the scowl-and-anger example. Even in Western samples, meta-analyses found scowling in about 35 percent of angry episodes—above chance, but absent in the other 65 percent. A scowl is also produced for many non-anger reasons, so it has limited specificity. Barrett’s conclusion is that anger is not one fixed thing but a variable category of instances whose bodily movements depend on context and goals. Calling anger a process rather than a noun captures that claim without making the instances random.