Deciding with feeling

The storyteller in the head

Ask people why they chose something and they will tell you. One of the odder findings of the last half century is how readily the telling comes, whether or not it is true.

The clearest early evidence came from patients whose brains had been surgically divided. Surgeons had cut the corpus callosum, the broad band of nerve fibers that connects the brain’s two halves, to stop severe epilepsy from spreading since the 1940s, and a Los Angeles team revived the operation in the 1960s. Roger Sperry of Caltech, who shared the 1981 Nobel Prize in Physiology or Medicine for this line of work, and his graduate student Michael Gazzaniga tested what the operation had changed. Because of the way the eyes connect to the brain, whatever is shown on the left side of the visual field goes only to the right hemisphere, and in most people language lives in the left. A patient called W.J., a former paratrooper, was shown a shape on the left. Asked what he had seen, he said, “Nothing.” Asked to point with his left hand, which the right hemisphere controls, he pointed to the right shape.

In a later test, which Gazzaniga and Joseph LeDoux published in a 1978 book and Gazzaniga has retold many times, a patient known as P.S. was shown a chicken claw to the left hemisphere (the one that speaks) and a snow scene to the right. Each hand chose from an array of cards. The right hand chose a chicken, matching the claw. The left hand chose a shovel, matching the snow. When P.S. was asked why he had chosen those two, his speaking half, which had seen only the claw, did not say “I don’t know.” It said, in effect, that the claw goes with the chicken, and you need a shovel to clean out the chicken shed. Gazzaniga named the part of the left hemisphere that does this the interpreter: a system that produces a plausible story for whatever the person finds themselves doing, without access to the real reason.

You do not need a split brain to see the same thing. In 2005 a team at Lund University in Sweden, led by Petter Johansson, showed 120 volunteers pairs of photographs of women’s faces and asked them to point to the more attractive one in each pair. After each choice the experimenter handed over the chosen photograph and asked why it was chosen, but on some trials, with a card trick, he had quietly swapped in the face the person had not chosen. Only 13 percent of the swaps were noticed on the spot, and at most 26 percent by any measure. The rest explained, fluently and with feeling, why they preferred the face they had rejected. One man said he would have rather approached her in a bar and that he liked her earrings. A woman said the face looked like her aunt. The authors called it choice blindness. A 2013 version, during the 2010 Swedish election, asked 162 people about their political views and slipped in altered answers: 92 percent accepted the altered summary of their views as their own, and 48 percent said they would consider switching to the other political bloc. The much earlier classic is a 1977 paper by Richard Nisbett and Timothy Wilson. Shoppers asked to pick the best of four identical pairs of stockings chose the right-most pair nearly four times as often as the left-most one. None mentioned position. When asked directly if it could have mattered, virtually all denied it.

What does this have to do with feeling? A storyteller that reports reasons it did not use will also tell stories about feelings. People who are anxious tend to treat the anxiety as evidence of danger: in a 1995 study of 186 people, anxious patients inferred danger from their own anxiety, and healthy people did not. A feeling is information about something. It is not always information about what you think it is about, and the storyteller will gladly supply an answer.

What feeling adds, and what it costs

Course 10, Lesson 1, told the story of Elliot, Antonio Damasio’s patient who could reason perfectly well and could not decide, and the card game that suggested why. This lesson takes up what came after. If feeling is what lets us choose, what does it add, and what does it cost?

People often judge how risky or beneficial something is partly by how it feels. Paul Slovic and his colleagues called this the affect heuristic. Things we like are judged to be low in risk and high in benefit, and things we dislike the reverse, even where facts should make the two independent. In a survey Slovic reported, members of the British Toxicological Society rated 30 chemicals as good or bad in a moment, and then judged how risky a very small exposure would be, which should have been close to zero for all of them. The two judgments were correlated, and 95 of the 97 toxicologists who answered showed the pattern. Under time pressure, Melissa Finucane and colleagues found in 2000, the inverse link between judged risk and judged benefit got stronger. A direct replication in 2022 found the strong negative link again, but telling people that a technology had few benefits did not make them judge it riskier, which suggests the effect is lopsided.

The patients tell more. Michael Koenigs and colleagues in 2007 gave moral dilemmas to six people with damage to the ventromedial prefrontal cortex, the region that was damaged in Elliot, and to two groups of controls. On ordinary dilemmas they answered like everyone else. On the hardest personal ones, such as whether to smother your own crying baby to save a larger group, the patients endorsed the harmful act far more often: the odds were about 4.7 to 5.4 times higher than in either control group. Their other judgments were normal. What was missing, the authors argued, was the emotional response that stops most people.

The missing feeling can also help. In 2005 Baba Shiv, George Loewenstein, Antoine Bechara, and the Damasios gave 41 people a simple game. Everyone started with $20 and played 20 rounds. In each, you could invest $1 on a coin toss: heads, you lost the dollar; tails, you got $2.50. Investing was a good bet, worth $1.25 on average for each dollar risked, but a loss stings. The 19 normal volunteers invested in 57.6 percent of the rounds and finished with $22.80 on average. A group of 15 patients with damage to emotion-related regions of the brain, in the orbitofrontal cortex, the right insula and somatosensory cortex, or the amygdalae, invested in 83.7 percent of rounds and finished with $25.70. The difference was in what happened after a loss: normal volunteers invested only 40.5 percent of the time after losing, while the patients invested 85.2 percent of the time, as often as after a win. A group of seven brain-damaged controls behaved like the healthy volunteers. Try it.

Another patient takes us back to the first lesson. In 2010 Benedetto De Martino, Colin Camerer, and Ralph Adolphs offered two women with Urbach-Wiethe disease, whose amygdalae are damaged, and twelve healthy controls 256 coin-flip gambles, with possible gains of $20 to $50 against possible losses of $20 to $50. Most people need the potential gain to be about one and a half to two times the potential loss before they take a gamble: the loss-aversion measure in the study was 1.52 and 1.76 for the two groups of controls. For the patient S.M., almost certainly the same woman as in the first lesson (same disease, same lesions, 43 in this study and 44 in the fear studies), it was 0.76: she weighed losses slightly less than gains. For the other patient, A.P., it was 1.06. Their caution about risk in general was normal. Only the sting of losing was missing.

So emotion is both an alarm that guards us and a bias that can make us pass up a good bet. Patients with damage to the front of the brain can be ruined by everyday choices and yet endorse harsh moral acts that most people refuse. Patients with damage to the amygdala take good gambles that others turn down, and S.M. walked into danger without fear. There is no free lunch in either direction.

Feelings from one thing also leak into others. Jennifer Lerner and Dacher Keltner found in 2001 that fearful people gave gloomy estimates of risks and chose cautiously, while angry people gave sunny estimates and took risks, in both naturally occurring and induced moods. In 2004 Lerner, Deborah Small, and Loewenstein showed 199 people, mostly students, a short film clip, either a sad scene from The Champ, a disgusting scene from Trainspotting, or a neutral nature film, and then asked them to name a price for a set of highlighters. Sadness lowered the price at which people would sell the set and raised the price at which they would buy it, which reversed the usual gap. A preregistered replication in 2023 found the sadness result again, but not the disgust one. Finance studies claiming that a national team’s loss in the World Cup knockout rounds lowers its stock market the next day by about 0.5 percent, and that morning sunshine lifts markets, are small and disputed, and the sunshine authors themselves note that trading costs would eat the profit.

Feelings are also poor forecasters. In a 1998 study at the University of Texas at Austin, Daniel Gilbert and colleagues asked assistant professors to predict how happy they would be in the five years after being denied tenure. They predicted 3.42 on a scale of 1 to 7, against 5.90 if they got it. Former professors who had in fact been denied, on average two and a half years before (there were only seven of them), reported 4.71. A later study of five experiments by Linda Levine and colleagues argued that people predict how an event will feel quite well, and that the large error appears mainly when they are asked about their general happiness. How big the forecasting error is depends heavily on how you ask. The pattern that survives is more modest: strong feeling is a loud signal and a poor clock.

Accepting a feeling

If feelings are signals, the question is how to treat them. Two familiar answers pull in opposite directions: push them down, or let them rule. A third is to accept them, to notice a feeling and let it be there while you decide what to do.

The case against pushing down begins with a polar bear that was white. In 1987 Daniel Wegner and colleagues asked people to say whatever came to mind aloud, with one rule: they were not to think of a white bear. They rang a bell each time one came to mind. Afterward, those who had tried to suppress it thought of the bear more than people who had been free to think of it from the start. Wegner called it ironic: the part of the mind that monitors for the forbidden thought keeps bringing it up. Later work shows the rebound is smaller and less consistent than the story suggests: a 2001 meta-analysis found a small-to-moderate rebound, and a recent critique argues that in healthy people suppression can work. It is not a law.

Studies of acceptance compare it directly with suppression. In 2006 Laura Campbell-Sills and colleagues showed an upsetting film to 60 people with anxiety or mood disorders, who were told either to suppress their feelings or to accept them. Distress during the film was similar in both groups, but the people who had been told to accept recovered with less negative feeling afterward, and their heart rates fell rather than rose. In 2018 Brett Ford and colleagues, across three studies, found that people who habitually accepted their negative emotions and thoughts were better off. In a sample of 1,003 students, acceptance went with greater well-being and fewer anxiety symptoms, correlations of .26 to .49 and −.41 to −.57. In a lab task, 156 women who accepted their reactions to a three-minute recorded speech felt less negative. A group of 222 people kept a diary for 14 nights: those who accepted had less daily negative emotion, which in turn accounted for better psychological health six months later. These are correlational, and mostly self-reported.

“Acceptance” here does not mean resignation or approval. It means allowing an unwanted thought or feeling to be present without struggling against it, while continuing to act on what matters to you. It is one of six processes in Acceptance and Commitment Therapy, developed by Steven Hayes and colleagues. In a 2015 meta-analysis of 39 trials with 1,821 patients, that therapy beat waiting lists and treatment as usual, with an overall effect of 0.57, and was not significantly different from cognitive behavioral therapy. A 2014 review of 60 trials by Lars-Göran Öst, with a stricter eye on quality, put the overall effect at 0.42 and called it “probably efficacious” only for chronic pain and tinnitus. Mindfulness, a practice that overlaps with acceptance, has been tested more widely: a 2013 meta-analysis of 209 studies with 12,145 people found a moderate effect (0.55) over time, 0.53 against a waiting list, 0.33 against other active treatments, and no edge over cognitive behavioral therapy. In the largest school trial, MYRIAD, 8,376 teenagers in 84 UK schools were randomly assigned to mindfulness training or ordinary lessons, and the training made no difference to their risk of depression, with an effect of 0.005. The honest summary is that acceptance-based methods help, about as much as other good therapies and no more, and they are oversold when sold as a cure.

Reason and feeling: a false split

The old picture sets reason and feeling against each other, with reason as the driver and feeling as the passenger who grabs the wheel. The research of this course points elsewhere. In 1739 the Scottish philosopher David Hume wrote that reason “is, and ought only to be the slave of the passions.” He did not mean that people should be irrational. He meant that reason alone cannot move us: it finds facts and means, while some desire supplies the end. Damasio’s Descartes’ Error (1994) argues something close for the brain, that Descartes was wrong to split mind from body and reason from emotion, and that rational thought depends on signals from the body.

The research gives some practical rules. A feeling is evidence, not a verdict: it tells you that something matters, not always what, and the storyteller will offer an answer whether or not it is right. Naming the feeling and asking what it is about is a cheap first step, and, as the first lesson showed, the alarm may have fired at nothing. When the situation has stable cues and you have practiced it with quick feedback, as a firefighter or a chess player has, a gut feeling deserves trust (Daniel Kahneman and Gary Klein, 2009). When it does not, it deserves less, however certain it feels. The folk advice to “sleep on it” works less well than claimed: a famous 2006 study said that complex choices came out best after a period of distraction, but a 2015 analysis of the studies, with a new test of 399 people, found no reliable support. And where the odds are clear, as in the investing game, a feeling about losses may simply mislead.

Regret research adds one more. Over a lifetime people tend to regret what they did not do more than what they did. In 1994 Thomas Gilovich and Victoria Medvec found that actions bring the sharper regret at first and inactions the longer-lasting one. In 2005 Neal Roese and Amy Summerville pooled 11 studies and found that people most regret education, career, and romance, the areas where they still see room to act, and a 2003 study in China, Japan, and Russia found the same long-run pattern. The popular list of the “top five regrets of the dying” is anecdote, not research. What follows is this course’s own synthesis, not an experimental result: for choices that are big and hard to undo, ask which version you could live with. The answer comes partly from a feeling and partly from reasons you can check.