How We Decide

Jonah Lehrer's How We Decide (Houghton Mifflin Harcourt, 2009) is a smart and entertaining look at recent research into how the brain functions when we're faced with a choice. One of the main findings is that both the rational centers (the prefrontal cortex) and emotional centers (the amygdala) are involved in decisionmaking, but that each tends to be best at certain kinds of problems. Counter-intuitively, Lehrer claims that the rational brain is best at simple problems, but can be overwhelmed when there are many complex and competing variables--then it's best to rely on gut instinct (the emotional brain) and not to overthink a problem.

He offers selectively chosen experimental evidence, but even his carefully chosen examples are sometimes contradictory. Buying a house is an example: there are so many variables to consider that the rational brain can't sort them out; better decisions are made (Lehrer claims) when people trust their impressions and instincts, rather than try to weigh every variable. But then Lehrer suggests that when people use their emotional brain to choose a mortgage--another complex choice--they often go for the option that is cheaper in the short term but vastly more expensive over the life of the loan: in other words, a really bad deal. So trusting your gut in complex decisions is best, except when it isn't--typically, when numbers are involved.

Do you want cake with that telephone number? I'm also concerned about the experimental design of some of the studies that Lehrer recounts. In one experiment, subjects were asked to memorize either a two-digit or a seven-digit number, and then walk down a hall to another room where they would be asked to repeat it. Halfway down the hall they were offered a snack (and this was the real experiment): they could choose either a serving of fruit salad, or a slice of German chocolate cake.

You might think that personal preference would determine which snack was chosen, but instead it was found that the determining factor was the size of the number the subject had to memorize. The ones who'd memorized the seven-digit number went for the chocolate cake much more often than the ones who'd memorized the two-digit number. The explanation offered is that the rational centers of the brain were overwhelmed even by the apparently simple task of memorizing a seven-digit number, and so those subjects were obeying an emotional impulse towards the chocolate dessert (instead of exercising rational self-control and going for the healthful fruit salad). From the point of view of the scientists who designed the experiment, people are being led into the less optimal choice by their emotional centers. But aren't they also expressing their true preference? And isn't a true preference a more optimal choice, even if it's a bit decadent?

Van Gogh vs. the Kittens: In another study of preferences, college students were asked to select a poster to take home, and given a choice between a Monet reproduction, a Van Gogh reproduction, and three different "humorous cat posters." One group of subjects was simply asked to rate the posters on a scale of 1 to 9 before selecting their favorite. A second group was asked to explain why they liked or disliked each poster before rating them. Only five percent of the first group chose one of the kitten posters, while half of the second group did. When the students were contacted several weeks later, 75 percent of the ones with the kitten posters were dissatisfied with their choices, while no one with an art poster regretted their choice. The scientists' explanation was that in the second group, the engagement of the rational centers (having to explain their choice) had blocked the input from the emotional centers, leading to a less satisfying choice.

But a key aspect of this experiment is that there wasn't one art poster among the choices, but two. A subject who chose either the Van Gogh or the Monet as her favorite may have felt that in order to adequately explain her choice to the scientists running the experiment (not just authorities, but representatives of the college faculty who would be judging her academic performance for the rest of her college career), she would have to discuss matters of art history or aesthetics, areas in which she may have not felt comfortable. Since a preference for a kitten poster was easier to explain, subjects may have rated them highest. So the ratings in the second experiment may not have reflected a subject's actual preferences, just her wish to avoid looking ignorant in front of a professor. It's little wonder most of those subjects felt dissatisfied later on.

Drinking champagne through a straw: Another preference experiment with design problems is the famous wine-tasting experiment, where subjects expressed a preference for wine that they were told was more expensive. The twist was that the wine the subjects were told cost $45 a bottle was actually the same wine that they'd been told cost $5 a bottle (its true retail price) when they'd sampled it earlier in the experiment. But brain imaging showed that their prefrontal cortexes were more active when they drank the "$45" wine. That is, they weren't just claiming to like the more expensive wine in order to appear sophisticated. When they thought a wine cost more, they actually enjoyed it more. When the subjects tasted the same wines without any price information, they reversed their preferences, and gave the highest rating to the cheapest wine.

The scientists were able to take brain scans of people who were drinking wine by having them sip it through straws. But the problem with this procedure is that smell is intimately bound up with our perception of taste--that's why food can seem tasteless when you have a stuffy nose. Sipping wine through a straw removes many of its flavor components. So wines drunk through a straw will taste more alike than when the same wines are drunk from wine glasses; the experiment doesn't allow the subjects to fully distinguish among the wines by flavor, leaving price as the only other information the subjects have.

Of course, the main point of this experiment that our expectations can affect our perceptions, which seems indisputable. But my own experience shows that things in the real world are less clear-cut than in this experiment. A friend of mine used to give regular wine-tasting parties, where three or four different wines would be sampled blind (only one person knew which wine was which, and he would indicate his preference last). At every single one of the more than half-dozen parties we attended, the group preferred the wine that turned out to be the most expensive. Occasionally a cheap wine would wind up in second place, but it was never close--the most expensive wine would typically be considered superior by a large margin. The difference between us and the subjects of the experiment? We were drinking the wine out of glasses, and so the differences in flavor and aroma were more apparent.

A bird in the hand, or...? But for me the most fascinating chapters of the book concern behavioral economics--the choices we make when money is at stake. Lehrer describes a game show that I'm glad I've never had a chance to see, because I'd instantly become an addict. It's called "Deal or No Deal," and it features 26 briefcases containing amounts of money ranging from 1 cent to a million dollars (or euros). At the beginning of the game you choose one briefcase, which is placed in a lockbox. Then you start opening the other briefcases one by one. After every couple of rounds, a man called The Banker makes you a cash offer for the briefcase you've set aside, and the amount of the offer fluctuates with what's revealed in each opened briefcase. As long as the big-money briefcases remain unopened, your offer rises steadily; as soon as you eliminate them, it plummets. That means your offer can change by a huge amount of money very quickly.

Lehrer describes the case of Frank, a contestant on the Dutch version of the show. With five briefcases left, the only high-value case that remained contained 500,000 euros. The Banker offered Frank 102,006 euros for his locked briefcase. Should he have taken the offer?

The answer is yes. There was only a one in five chance that Frank's briefcase contained the 500,000 euros, and he's being offered more than 100,000 euros for it.

But Frank said no. If he could eliminate another low-value briefcase, the Banker's offer would rise. But Frank was unlucky--the next briefcase he opened contained the 500,000 euros. The most the remaining briefcases could contain was 10,000 euros. The Banker then offered 2508 euros. In the space of a few seconds, Frank's offer had declined by 100,000 euros. Should Frank have taken this offer, painful as it was?

The answer is again yes. But Frank refused again, because of a mechanism that is termed "loss aversion." Losses are more painful to us than gains are pleasurable, by a substantial margin. Frank decided that after a loss of 100,000 euros that a gain of 2500 wasn't worth it, even though it was a fair deal. He still had the possibility of winning 10,000 euros, after all.

Finally, two briefcases were left: one containing 10,000 euros, and one containing 10 euros. The Banker made a very generous offer: 6500 euros. Frank should definitely have taken the offer: he only had a 50 percent chance of winning 10,000 euros, and he was being offered nearly two-thirds of that amount, guaranteed. But he decided to open his locked briefcase instead. It contained 10 euros.

Frank's experience makes you think that perhaps economic models in which all actors are assumed to be rationally maximizing their own utility might be a bit of an oversimplification...

How We Decide is filled with thought-provoking studies and anecdotes like these. Interestingly, the book avoids entirely the subject of romantic choice. That fascinating area is one that I hope Lehrer will explore in his next book on how the brain works.

Update 6 February 2010: WNYC's excellent science program RadioLab has a podcast episode on choice that I highly recommend; Jonah Lehrer is one of the interviewees.