The neuroscience of emotions and how to regulate them

Emotions are not just mental noise. According to neuroscientist Ralph Adolfs, professor of psychology, neuroscience and biology at Caltech, emotions are functional states that evolved to solve very specific survival problems. In a conversation with Andrew Huberman on the Huberman Lab podcast, Adolfs explains that understanding what emotions are, and how to regulate them, starts by no longer treating them as vague feelings and instead seeing them as systems with a precise function, similar to a reflex, but far more flexible.

Four features that define an emotion

Adolfs, together with neuroscientist David Anderson, proposed a framework to identify what emotions as different as fear, anger and joy have in common. Under this framework, an emotion meets four criteria:

• Priority: it can take over behavior, much like a reflex, though in a more flexible way that adapts to context.
• Valence: it sits on an approach or avoidance axis, from pleasant to unpleasant, the most basic dimension for classifying emotions.
• Scalability: it grows in intensity gradually, for example from anxiety to fear and then to panic as a threat gets closer.
• Temporal persistence: it outlasts the stimulus that triggered it, unlike a reflex, which ends the instant you pull your hand off a hot surface.

This temporal persistence turned out to be especially revealing. Adolfs cites a study on patients with severe amnesia, conducted by former colleagues of his: after watching a sad movie, these patients felt sad, just like a control group with no brain damage. Minutes later, with no memory of having watched the movie, they still felt sad and had no idea why. The emotion persisted independently of declarative memory, showing that its function does not depend on remembering the cause.

Do animals feel emotions, and can AI have them too?

Applying these four functional criteria, many animals, including domestic mammals, clearly qualify. Adolfs goes further, arguing that in a purely functional sense, certain artificial intelligence systems could also qualify as having emotions. Here he draws an important distinction: having a functional emotion that guides behavior is one thing, having the conscious experience of that emotion is something else entirely. Separating the two lets scientists study emotions in animals and machines without first having to solve the much harder problem of consciousness.

The myth of reading emotions from faces

For decades it was assumed that facial expressions like fear or surprise are recognized universally and reliably across cultures. Adolfs, co-author of a recent review on the topic, complicates that picture with data: the classic studies used expressions posed by actors, far more exaggerated than what shows up in real life, and asked participants to pick from a closed list of words. When researchers use spontaneous expressions instead and ask people to freely describe the emotion they observe, far more variability shows up than expected. In practice, people trust their ability to read faces, both human and their pets', much more than that ability actually deserves, and that confidence clearly outstrips real accuracy in controlled studies.

Cold exposure as emotion-regulation training

One of the most practical examples in the conversation is Adolfs's use of ice baths. At first, the experience is simply painful: the body responds with a spike in heart rate and breathing. With weekly repetition, that same body learns to regulate the stress response faster and with less conscious effort. What stands out is that this training appears to generalize to other situations. Adolfs describes how, after months of regular ice baths, everyday scenarios that used to trigger his anger, like someone honking while he was trying to park, stopped affecting him the same way. He interprets this as autonomic training of the capacity to regulate emotions, one that becomes progressively more automatic and effortless with consistent practice.

Awe and long-term regulation

The conversation also covers emotion regulation on a much longer timescale, through the experience of running ultramarathons, a challenge Adolfs takes on himself as a runner. He explains that, mile after mile, the body cycles through feeling terrible and then recovering, and that learning to trust that this cycle repeats is itself a form of emotion regulation: refusing to give up despite an immediate signal of distress. These kinds of extreme experiences, he argues, are related to awe, an emotion that appears when we widen our perspective, whether in space or in time, and one that seems to be uniquely human because it depends on our ability to detach conscious experience from the here and now.

The takeaway

Understanding emotions as functional states, defined by priority, valence, scalability and persistence, offers a practical way to work with them day to day. It helps identify them more precisely, accept that some can be trained through repeated exposure, as with cold, and that others are cultivated by seeking perspective, as with awe. The goal is not to eliminate emotions or suppress them, but to learn to regulate them without losing the function they evolved to serve.