The Trouble With Rewards

In the early 1970s, 55 kids at Stanford University’s Bing Nursery School found themselves playing an odd little game with some visiting adults.



In the early 1970s, 55 kids at Stanford University’s Bing Nursery School found themselves playing an odd little game with some visiting adults.

One by one, they were escorted to a “surprise room” where they found Magic Markers and sheets of paper. There, they were asked if they’d like to draw for a visitor. In addition, some were told that if they did a nice job they’d get a “Good Player Award” (with their name, the name of the school, a ribbon and a gold seal) which they’d get to put on a bulletin board. A third cohort wasn’t told about the reward in advance, but got it as a surprise after they made their pictures.

The children, all of whom had been picked because they looked particularly interested in drawing with Magic Markers, got to work immediately. Then they were escorted back to their regular classroom and their regular 3-, 4- and 5-year-old lives.

It was that aftermath that really interested the researchers — psychologists whose work at Bing helped overturn our understanding of motivation. In their regular classroom, the children were free to choose among many options (like going outside, playing with blocks, using Play-Doh, as well as drawing with Magic Markers). Watching through a one-way mirror in the days after the experiment, the team saw that children who had not expected a reward spent about a sixth of their playtime with Magic Markers. And the kids who had been told they were drawing in order to earn that shiny incentive? They spent only half as much of their time drawing.

Classical economic theory, of course, predicts just the reverse, not just for children but for adults. Concrete rewards are supposed to have a simple proportional relationship to effort: More incentive, more output. People work because work leads to other things that they desire. Their motives are extrinsic to them. These rewards, of course, needn’t be pay raises or annual bonuses. In ancient times, they might be food, sex or status. Often, these motivators are the sort of shiny “toys” that, Napoleon observed, have always been important in getting the best out of people. (Having created the Legion of Honor to recognize achievement in the new French republic, he said of its ranks, with their various medals and sashes: “We call these children’s toys, I know, it’s been said already. Well, I replied that it’s with such toys that one leads men.”)

Whatever the external reward, there’s no denying that mapping it to motivation makes it easy for a leader to shape incentives: It’s just a matter of figuring out how much (or how little) to give in pay, benefits, bonuses, shiny trophies and back-pats in order to get the needed amount of cooperation and work out of your people.

Of course, any good model is supposed to be simpler than its original. No one ever thought extrinsic motives explained all human actions. It just seemed that the theory captured what incentive-makers needed to be effective. We’ve always known that motives varied among individuals, and that people did remarkable things for love, honor, revenge, pride or other “externalities” that can’t be quantified. (It was to express the fundamental mystery of human motives that Woody Allen once asked, “Why does man kill? He kills for food. And not only food: frequently there must be a beverage.”) People value things other than bonuses and pay packages and employee-of-the-month citations. (Self-employed people, the economist Bruno Frey has noted, tend to report more satisfaction at work than their higher-paid peers at corporations). And many professionals feel, as does Seattle doctor Carol Wiley Cassella, that “more money isn’t the answer.”

“Salaried physicians tend to be more satisfied with their work than more highly paid, fee-for-service physicians, even when their work hours are comparable,” she wrote recently, in an essay on how incentives turned her from a happy and idealistic general practitioner into a more burned-out specialist. Wikipedia, Firefox and Apache are also proof that compensation and “gold stars” don’t make the world go round: Those enterprises were all created, and are maintained, by volunteers who are paid exactly zero and remain little known.

Nonetheless, for decades, the model-world workers chasing pay and perks stayed separate from the real world of individual psychology. Today, though, this is no longer true.

Since the Bing Nursery study and similar experiments in the 1970s, the two worlds have met. Today, it’s clear that leaders, either in business or politics, can’t afford to ignore what goes on inside people’s heads — the drive to do something for its own sake, not because of what it will bring in in pay, fame or perks. These intrinsic motives mix with extrinsic motives to make people perform. It’s a dynamic mix — a mix that can be managed. In fact, it has to be, if managers want to avoid suboptimal results and unpleasant surprises.

In many situations, for example, offering only money or other external incentives can reduce motivation. One case in point: In the spring of 1993 Frey and another economist, Felix Oberholzer-Gee, interviewed 305 people in two Swiss towns that were potential sites for a nuclear-waste repository. Asked for their opinion, nearly 51 percent of the interviewees said they would consent to having the nuclear-waste facility in their community. The researchers also asked how people would feel if the Swiss Parliament agreed to pay each villager a hefty sum in compensation for putting the facility in their midst. Result: The percentage in favor was cut in half, to 24.6 percent.

Frey calls this a “crowding-out effect”: Money, the extrinsic motive, crowded out the intrinsic desire to “do our part” as responsible citizens, just as the “Good Player Award” crowded out the kids’ desire to draw because drawing is fun. This is almost always unfortunate, because intrinsic motives (such as genuine interest in the task, a sense of community and “doing your share,” control over the outcome of the work, and even, Frey says, simple acknowledgement that one has intrinsic motives) are stronger and longer-lasting than extrinsic ones.

“Crowding out,” Frey argues, explains why monitoring workers often doesn’t yield higher productivity: Where there is a personal relationship and sense of community between employee and employer, monitoring is felt as a statement of distrust, a rupture of warm human bonds. Pay and benefits aren’t sufficient to replace the lost desire to be a good citizen.

Being a leader, Frey argues, is the art of managing the environment so that intrinsic motives aren’t crowded out by extrinsic ones in this way. The true artist, he suggests, can go one better than that by managing matters to encourage “crowding-in” effects: circumstances where extrinsic and intrinsic motivators reinforce each other.

To understand the difference between intrinsic motives and extrinsic ones, consider how you feel when doing something for its own sake. It could be working on your backhand, or your fantasy baseball team, or your memoir, or your kid’s dollhouse. Whatever the activity, your priorities, as you lose yourself in this pleasant experience, are about its inherent interest, or perhaps the satisfactions of being a good mother, sister, neighbor or the like. One thing you’re almost certain not to be thinking about is a bottom-line result — not because that goal doesn’t matter, but because the measurable product (how much in how much time) is not central to the experience. Your game will get better as you work at it. The dollhouse might be done Sunday, or maybe it will be next Saturday. You’re absorbed in the process, but the time and size of the payoff is uncertain.

It could be that this uncertainty is a byproduct — that you don’t measure your progress because you’re “in the zone,” thinking of other things. But recent evidence from psychology and brain science suggests that this is not so. Instead, researchers have found, the mind is drawn to uncertainty. In other words, the fact that you can’t be sure of your results might not be a side-effect when you’re doing what you love. It may be a crucial part of the fun. Uncertainty, then, may be part of what makes an intrinsic incentive feel intrinsic.
Of course, this implies something about motivation that is very much at odds with rational-economic-man expectations (even more than the fact that paying people to do something can cause them to do less of it). It implies that there are times when uncertainty about a payoff motivates people more than a guaranteed reward.

After all, educational theory emphasizes “reward consistency,” and students, when they have a choice on test questions, tend to pick the ones they’re sure to solve, notes Paul A. Howard-Jones of the University of Bristol’s School of Education. Howard-Jones thinks that caution stems from fear of looking bad. It’s actually not that difficult to get students to reach higher, he has found. But the trick is to introduce an element of randomness into the scoring.

In one recent experiment, Howard-Jones had 11-year-old pupils at a school in Cyprus play a computer game in which they had to answer 30 math questions like “Is it true that 13 times 42 is 564?” Before each question, though, the child had to decide whether to receive it from “Mr. Certain” or “Mr. Uncertain.” Mr. Certain gave a point for each correct answer. Mr. Uncertain flipped a virtual coin and awarded either two points or zero for the same right response.

If you expected that the students would prefer Mr. Certain, you’d be wrong. Mr. Uncertain was chosen, on average, more than 61 percent of the time. To explore why, Howard-Jones and his collaborators interviewed 10 of their 50 participants. Asked why they had picked Mr. Uncertain, they explained that, much as they understood that he could cost them points, there was something just more interesting and exciting about him.

“You are stressed,” one child said, “but if you win then you forget about that!” In the language of academia, Howard-Jones and his colleague Skevi Demetriou wrote, “a learner can find a task more emotionally appealing when an element of uncertainty is introduced that is not wholly defined by the learner’s own ability.” In another of his experiments, teams of teen-aged students answering biology questions would sometimes say they hated to lose points by bad luck (“73 points and I got wiped out!”), at the same time as they were saying they wanted to play another round.

In other words, by turning the learning and testing process into a game, Howard-Jones had motivated students to reach far beyond their usual comfort zone. It’s a result that wouldn’t surprise any parent who has grumbled about the lousy test scores of a teen who has memorized every anatomical and historical detail about every character in World of Warcraft. Framing a problem as a game — an effort whose outcome is by definition unpredictable — motivates extra attention and effort.

That’s true in the world of work as well as school. In September 2011, a paper in the journal Nature Structural and Molecular Biology announced that an international team had solved a longstanding problem in medical research on HIV and similar viruses: What is the precise three-dimensional structure of an enzyme that the AIDS virus needs to reproduce itself? Medical researchers presented their problem to the online game Foldit, whose players compete in teams to find the best solutions to protein-folding problems. Those players came up with the key insight that allowed the medical researchers to solve the mystery. Conventional lab work hadn’t cracked the problem in 15 years of trying. The gamesters did it in three weeks.

The appeal of games over plain old work, Howard-Jones believes, is that “element of uncertainty” that piques our interest and seems to goose our motivation. And part of uncertainty’s appeal might just be that it makes extrinsic rewards feel like the ones we get when we are pursuing something for its own sake — when precise outcome is uncertain, hoped for, always worth chasing but not entirely under our control. In other words, an element of uncertainty can make an extrinsic reward feel like an intrinsic one.

Even as psychologists have found that randomness can help to motivate people, neuroscientists have begun to explain how and why the brain is wired in this seemingly irrational way.

Whenever we enjoy or desire something, the brain’s “reward chemical,” dopamine, is secreted by specialized cells and absorbed by other neurons. The uptake of dopamine generally makes us feel interested, confident and happy. Having something we like triggers the “reward circuit” in which dopamine-making cells send the neurotransmitter to other parts of the brain. But, interestingly, this also happens when we want something we like.

About ten years ago, Christopher D. Fiorillo, a neuroscientist now at the Korea Advanced Institute for Science and Technology, discovered the dopamine-driven effects of anticipation while working with a pair of monkeys. After the primates had done an experimental task and were waiting hopefully, licking their lips, for their juice reward, the computer would give them a crucial piece of information: It told them how likely they were to get their reward a couple of seconds later. Some of the animals’ dopamine-making cells were responding in a way Fiorillo and his team hadn’t expected: When the probability of getting the reward was zero, the cells were quiet. The same was true when the animals were 100 percent certain of getting their juice. But when the outcome was not so certain, with reward probability at, say, 25 percent or 75 percent, the cells were much more active. In other words, this was a reward-circuit that responded not to the reward itself, but rather to the odds of winning it. And this circuit doesn’t get excited by certainty, but rather by its opposite. The cells measured by Fiorillo were most active when the chance of reward was exactly 50-50.

A couple of years later, Jean-Claude Dreher of the National Institutes of Mental Health and his colleagues found that this tuning-to-uncertainty works the same way in human brains. The researchers’ MRI scans of 31 volunteers found that the dopamine-producing parts of their brains were most active when their chances of winning a reward (money, this time, instead of juice) were, again, 50-50. “During anticipation, uncertainty increases arousal,” says Brian Knutson, a professor of psychology and neuroscience at Stanford University.

There are, it turns out, sound evolutionary reasons that we should be drawn to the uncertain payoff. Uncertain situations yield information that could be important to future success. (This has been represented formally in Claude Shannon’s information theory, which defines 50-50 odds, maximum uncertainty, as the condition that contains the most new information. Absolute certainty about anything, of course, brings you no new information at all.) As one of the students interviewed by Howard-Jones put it, in a regular soccer game, “whenever I kick I’m never sure that I will score.” But if it’s just a practice without a goalkeeper “everything is easier and I don’t feel so excited when I score.” A similar feeling is common in law firms, writes Jennifer Alvey, a blogger and “recovering lawyer”: “Once you’ve gotten a bonus or two, you start expecting it, and stop appreciating it.”

Of course, it does not take an MRI scanner to know that people are tantalized by a payoff that may or may not happen, and that this excitement can feed on itself. That mode of reward-chasing is called gambling, and it has been popular for millennia. What’s new in the insights from behavioral and brain science is the evidence that the power of the gamble isn’t confined to casinos. In a 2007 experiment by the economist Ernst Fehr of the University of Zurich, for instance, he and his colleagues gave teams of Munich university students a set of tasks along with a choice about how they would be compensated for their work. One was a hard-and-fast guarantee of a payoff for work above a baseline (the old-fashioned incentive approach). Another was a generous payment in advance (call it the “employer-trusts-you” approach). And the third was a bonus to be paid at the “employer’s” discretion (you could call this one “you-trust-the employer”). The option the students preferred, which also happened to produce the best work? No. 3, the bonus that might or might not be paid.

This isn’t to suggest that any company should replace its payroll organization with a craps table. Any employee’s motivation is a mix of intrinsic and extrinsic drivers, and it’s a dynamic mix, changing from day to day. (That’s not peculiar to work; it’s a feature of motivation anywhere. Consider a parent putting together a kid’s sandwich for the school day. Monday’s motive might be mostly love for the child; Tuesday’s a mix of love and obligation; Wednesday’s a grim feeling that someone has to do it, since the other parent forgot, and so on.) Uncertainty is not a magic bullet. But it isn’t your enemy, either, and it can be a part of your motivational toolkit.

Download the PDF