Afraid Because You Run

William James, in 1884, proposed something that still sounds wrong on first hearing.

The commonsense view is that you see a bear, feel afraid, and the fear makes your heart race and your legs move. James argued the sequence runs the other way: you see the bear, your body reacts (heart, breath, muscles, gut) and the feeling of fear is your perception of those changes. Remove the bodily response, he claimed, and nothing is left but a cold intellectual judgement that a bear is present.

Carl Lange proposed something similar independently, and the position is known as the James-Lange theory.

Walter Cannon attacked it on two grounds that seemed decisive at the time. Bodily responses are too slow; emotions arrive faster than visceral changes. And bodily responses are too undifferentiated: the racing heart of fear looks much like the racing heart of rage or excitement, so how could the perception of it tell you which emotion you are having?

The second objection turned out to be the productive one, because it invites an obvious question. If the arousal is ambiguous, what resolves it?

Schachter and Singer's answer, in 1962, was the situation. They injected participants with adrenaline, told some what to expect and misinformed others, then left them in a room with a confederate behaving either euphorically or angrily. Participants who had an explanation for their arousal were relatively unaffected. Those who did not reported feeling whatever the confederate was performing.

Arousal plus an attribution. The same physiological state becomes different emotions depending on what is available to explain it.

The study has a mixed replication record and a small effect, and the idea it introduced has been enormously productive regardless, partly because it predicts something odd and checkable, which is that arousal from one source will be misattributed to another.

Dutton and Aron tested this on two bridges in Vancouver. Men crossing a narrow, swaying suspension bridge high above a canyon were approached by a female interviewer who gave them her number; so were men on a low, solid bridge. The men from the high bridge were substantially more likely to call.

The arousal from the height was available for reinterpretation, and an attractive person was the salient thing in the environment. The effect has been replicated and complicated (it depends on the interviewer and on timing) and the general principle is well supported. Arousal from exercise, from crowds, from an argument, routinely gets attributed to whatever is nearby.

Now the claim most people think is settled, and which has taken heavy damage.

Paul Ekman's research programme from the 1960s onward argued for a small number of basic emotions with universal facial expressions: anger, disgust, fear, happiness, sadness, surprise. The evidence came from showing photographs to people in different cultures, including people in Papua New Guinea with limited Western contact, and finding high agreement about which emotion each face showed.

The method contained a problem. Participants were given a short list of labels and asked to pick one (forced choice) which does a great deal of work. Studies using free labelling, where people describe the face in their own words, find dramatically lower agreement. Work by Gendron, Barrett and colleagues with the Himba of Namibia and the Trobriand Islanders found that the supposedly universal wide-eyed fear face was not reliably read as fear, and the scowl not reliably as anger.

A large review of the evidence concluded that the mapping between facial configurations and emotion categories is far looser than the basic-emotion account requires: that people scowl when angry a minority of the time, and scowl for many reasons other than anger.

Lisa Feldman Barrett's alternative is the theory of constructed emotion, and it is worth understanding because it reframes the whole question.

On her account the brain does not have dedicated circuits that fire for anger or fear. It has core affect (a continuously updated state of valence and arousal, pleasant to unpleasant, activated to quiet) and it categorises that state using learned emotion concepts, in context, much as it categorises a continuous spectrum of light into colours.

The prediction is that there is no fingerprint for any emotion in the face, the body or the brain, and meta-analyses of imaging and physiology broadly bear this out. A further prediction is that emotion concepts shape emotional experience (that having a word for a state makes the state available) which is testable and being tested, and connects to the finding that greater emotional granularity is associated with better regulation.

The practical payoff of all of this is one of the most useful findings in applied psychology, and it follows directly.

If an emotion is arousal plus an interpretation, then changing the interpretation is a short move and changing the arousal is a long one. Alison Wood Brooks tested this with people about to sing, speak publicly or take a maths test. One group was told to calm down. Another was told to say "I am excited."

The excited group performed better, and reported feeling better, across all three tasks. Anxiety and excitement are both high-arousal states; relabelling requires no change in the body at all. Trying to calm down, under pressure, requires shifting the arousal itself, which is difficult and tends to fail, and the failure then becomes further evidence that you are anxious.