Soon more boys arrive, all of whom tumble into Tom’s trap and end up whitewashing the fence—several times over—on his behalf. From this episode, Twain extracts a key motivational principle, namely “that Work consists of whatever a body is OBLIGED to do, and that Play consists of whatever a body is not obliged to do.” He goes on to write:There are wealthy gentlemen in England who drive four-horse passenger-coaches twenty or thirty miles on a daily line, in the summer, because the privilege costs them considerable money; but if they were offered wages for the service, that would turn it into work and then they would resign.1
In other words, rewards can perform a weird sort of behavioral alchemy: They can transform an interesting task into a drudge. They can turn play into work. And by diminishing intrinsic motivation, they can send performance, creativity, and even upstanding behavior toppling like dominoes. Let’s call this the Sawyer Effect.a A sampling of intriguing experiments around the world reveals the four realms where this effect kicks in—and shows yet again the mismatch between what science knows and what business does.
Intrinsic Motivation
Behavioral scientists like Deci began discovering the Sawyer Effect nearly forty years ago, although they didn’t use that term. Instead, they referred to the counterintuitive consequences of extrinsic incentives as “the hidden costs of rewards.” That, in fact, was the title of the first book on the subject—a 1978 research volume that was edited by psychologists Mark Lepper and David Greene.
One of Lepper and Greene’s early studies (which they carried out with a third colleague, Robert Nisbett) has become a classic in the field and among the most cited articles in the motivation literature. The three researchers watched a classroom of preschoolers for several days and identified the children who chose to spend their “free play” time drawing. Then they fashioned an experiment to test the effect of rewarding an activity these children clearly enjoyed.
The researchers divided the children into three groups. The first was the “expected-award” group. They showed each of these children a “Good Player” certificate—adorned with a blue ribbon and featuring the child’s name—and asked if the child wanted to draw in order to receive the award. The second group was the “unexpected-award” group. Researchers asked these children simply if they wanted to draw. If they decided to, when the session ended, the researchers handed each child one of the “Good Player” certificates. The third group was the “no-award” group. Researchers asked these children if they wanted to draw, but neither promised them a certificate at the beginning nor gave them one at the end.
Two weeks later, back in the classroom, teachers set out paper and markers during the preschool’s free play period while the researchers secretly observed the students. Children previously in the “unexpected-award” and “no-award” groups drew just as much, and with the same relish, as they had before the experiment. But children in the first group—the ones who’d expected and then received an award—showed much less interest and spent much less time drawing. 2 The Sawyer Effect had taken hold. Even two weeks later, those alluring prizes—so common in classrooms and cubicles—had turned play into work.
To be clear, it wasn’t necessarily the rewards themselves that dampened the children’s interest. Remember: When children didn’t expect a reward, receiving one had little impact on their intrinsic motivation. Only contingent rewards—if you do this, then you’ll get that—had the negative effect. Why? “If-then” rewards require people to forfeit some of their autonomy. Like the gentlemen driving carriages for money instead of fun, they’re no longer fully controlling their lives. And that can spring a hole in the bottom of their motivational bucket, draining an activity of its enjoyment.
Lepper and Greene replicated these results in several subsequent experiments with children. As time went on, other researchers found similar results with adults. Over and over again, they discovered that extrinsic rewards—in particular, contingent, expected, “if-then” rewards—snuffed out the third drive.
These insights proved so controversial—after all, they called into question a standard practice of most companies and schools—that in 1999 Deci and two colleagues reanalyzed nearly three decades of studies on the subject to confirm the findings. “Careful consideration of reward effects reported in 128 experiments lead to the conclusion that tangible rewards tend to have a substantially negative effect on intrinsic motivation,” they determined. “When institutions—families, schools, businesses, and athletic teams, for example—focus on the short-term and opt for controlling people’s behavior,” they do considerable long-term damage.3
Try to encourage a kid to learn math by paying her for each work-book page she completes—and she’ll almost certainly become more diligent in the short term and lose interest in math in the long term. Take an industrial designer who loves his work and try to get him to do better by making his pay contingent on a hit product—and he’ll almost certainly work like a maniac in the short term, but become less interested in his job in the long term. As one leading behavioral science textbook puts it, “People use rewards expecting to gain the benefit of increasing another person’s motivation and behavior, but in so doing, they often incur the unintentional and hidden cost of undermining that person’s intrinsic motivation toward the activity.”4
This is one of the most robust findings in social science—and also one of the most ignored. Despite the work of a few skilled and passionate popularizers—in particular, Alfie Kohn, whose prescient 1993 book, Punished by Rewards, lays out a devastating indictment of extrinsic incentives—we persist in trying to motivate people this way. Perhaps we’re scared to let go of Motivation 2.0, despite its obvious downsides. Perhaps we can’t get our minds around the peculiar quantum mechanics of intrinsic motivation.
Or perhaps there’s a better reason. Even if those controlling “if-then” rewards activate the Sawyer Effect and suffocate the third drive, maybe they actually get people to perform better. If that’s the case, perhaps they’re not so bad. So let’s ask: Do extrinsic rewards boost performance? Four economists went to India to find out.
High Performance
One of the difficulties of laboratory experiments that test the impact of extrinsic motivators like cash is the cost. If you’re going to pay people to perform, you have to pay them a meaningful amount. And in the United States or Europe, where standards of living are high, an individually meaningful amount multiplied by dozens of participants can rack up unsustainably large bills for behavioral scientists.
In part to circumvent this problem, a quartet of economists—including Dan Ariely, whom I mentioned in the last chapter—set up shop in Madurai, India, to gauge the effects of extrinsic incentives on performance. Because the cost of living in rural India is much lower than in North America, the researchers could offer large rewards without breaking their own banks.
They recruited eighty-seven participants and asked them to play several games—for example, tossing tennis balls at a target, unscrambling anagrams, recalling a string of digits—that required motor skills, creativity, or concentration. To test the power of incentives, the experimenters offered three types of rewards for reaching certain performance levels.
One-third of the participants could earn a small reward—4 rupees (at the time worth around 50 U.S. cents and equal to about a day’s pay in Madurai) for reaching their performance targets. One-third could earn a medium reward—40 rupees (about $5, or two weeks’ pay). And one-third could earn a very large reward—400 rupees (about $50, or nearly five months’ pay).
What happened? Did the size of the reward predict the quality of the performance?
Yes. But not in the way you might expect. As it turned out, the people offered the medium-sized bonus didn’t perform any better than those offered the small one. And those in the 400-rupee super-incentivized group? They fared worst of all. By nearly every measure, they lagged behind both the low-reward and medium-reward participants. Reporting the results for the Federal Reserve B
ank of Boston, the researchers wrote, “In eight of the nine tasks we examined across the three experiments, higher incentives led to worse performance.”5
Let’s circle back to this conclusion for a moment. Four economists—two from MIT, one from Carnegie Mellon, and one from the University of Chicago—undertake research for the Federal Reserve System, one of the most powerful economic actors in the world. But instead of affirming a simple business principle—higher rewards lead to higher performance—they seem to refute it. And it’s not just American researchers reaching these counterintuitive conclusions. In 2009, scholars at the London School of Economics—alma mater of eleven Nobel laureates in economics—analyzed fifty-one studies of corporate pay-for-performance plans. These economists’ conclusion: “We find that financial incentives . . . can result in a negative impact on overall performance.”6 On both sides of the Atlantic, the gap between what science is learning and what business is doing is wide.
“Many existing institutions provide very large incentives for exactly the type of tasks we used here,” Ariely and his colleagues wrote. “Our results challenge [that] assumption. Our experiment suggests . . . that one cannot assume that introducing or raising incentives always improves performance.” Indeed, in many instances, contingent incentives—that cornerstone of how businesses attempt to motivate employees—may be “a losing proposition.”
Of course, procrastinating writers notwithstanding, few of us spend our working hours flinging tennis balls or doing anagrams. How about the more creative tasks that are more akin to what we actually do on the job?
Creativity
For a quick test of problem-solving prowess, few exercises are more useful than the “candle problem.” Devised by psychologist Karl Duncker in the 1930s, the candle problem is used in a wide variety of experiments in behavioral science. Follow along and see how you do.
You sit at a table next to a wooden wall and the experimenter gives you the materials shown below: a candle, some tacks, and a book of matches.
The candle problem presented.
Your job is to attach the candle to the wall so that the wax doesn’t drip on the table. Think for a moment about how you’d solve the problem. Many people begin by trying to tack the candle to the wall. But that doesn’t work. Some light a match, melt the side of the candle, and try to adhere it to the wall. That doesn’t work either. But after five or ten minutes, most people stumble onto the solution, which you can see below.
The candle problem solved.
The key is to overcome what’s called “functional fixedness.” You look at the box and see only one function—as a container for the tacks. But by thinking afresh, you eventually see that the box can have another function—as a platform for the candle. To reprise language from the previous chapter, the solution isn’t algorithmic (following a set path) but heuristic (breaking from the path to discover a novel strategy).
What happens when you give people a conceptual challenge like this and offer them rewards for speedy solutions? Sam Glucksberg, a psychologist now at Princeton University, tested this a few decades ago by timing how quickly two groups of participants solved the candle problem. He told the first group that he was timing their work merely to establish norms for how long it typically took someone to complete this sort of puzzle. To the second group he offered incentives. If a participant’s time was among the fastest 25 percent of all the people being tested, that participant would receive $5. If the participant’s time was the fastest of all, the reward would be $20. Adjusted for inflation, those are decent sums of money for a few minutes of effort—a nice motivator.
How much faster did the incentivized group come up with a solution? On average, it took them nearly three and a half minutes longer.7 Yes, three and a half minutes longer. (Whenever I’ve relayed these results to a group of businesspeople, the reaction is almost always a loud, pained, involuntary gasp.) In direct contravention to the core tenets of Motivation 2.0, an incentive designed to clarify thinking and sharpen creativity ended up clouding thinking and dulling creativity. Why? Rewards, by their very nature, narrow our focus. That’s helpful when there’s a clear path to a solution. They help us stare ahead and race faster. But “if-then” motivators are terrible for challenges like the candle problem. As this experiment shows, the rewards narrowed people’s focus and blinkered the wide view that might have allowed them to see new uses for old objects.
Something similar seems to occur for challenges that aren’t so much about cracking an existing problem but about iterating something new. Teresa Amabile, the Harvard Business School professor and one of the world’s leading researchers on creativity, has frequently tested the effects of contingent rewards on the creative process. In one study, she and two colleagues recruited twenty-three professional artists from the United States who had produced both commissioned and noncommissioned artwork. They asked the artists to randomly select ten commissioned works and ten noncommissioned works. Then Amabile and her team gave the works to a panel of accomplished artists and curators, who knew nothing about the study, and asked the experts to rate the pieces on creativity and technical skill.
“Our results were quite startling,” the researchers wrote. “The commissioned works were rated as significantly less creative than the non-commissioned works, yet they were not rated as different in technical quality. Moreover, the artists reported feeling significantly more constrained when doing commissioned works than when doing non-commissioned works.” One artist whom they interviewed describes the Sawyer Effect in action:Not always, but a lot of the time, when you are doing a piece for someone else it becomes more “work” than joy. When I work for myself there is the pure joy of creating and I can work through the night and not even know it. On a commissioned piece you have to check yourself—be careful to do what the client wants.8
Another study of artists over a longer period shows that a concern for outside rewards might actually hinder eventual success. In the early 1960s, researchers surveyed sophomores and juniors at the School of the Art Institute of Chicago about their attitudes toward work and whether they were more intrinsically or extrinsically motivated. Using these data as a benchmark, another researcher followed up with these students in the early 1980s to see how their careers were progressing. Among the starkest findings, especially for men: “The less evidence of extrinsic motivation during art school, the more success in professional art both several years after graduation and nearly twenty years later.” Painters and sculptors who were intrinsically motivated, those for whom the joy of discovery and the challenge of creation were their own rewards, were able to weather the tough times—and the lack of remuneration and recognition—that inevitably accompany artistic careers. And that led to yet another paradox in the Alice in Wonderland world of the third drive. “Those artists who pursued their painting and sculpture more for the pleasure of the activity itself than for extrinsic rewards have produced art that has been socially recognized as superior,” the study said. “It is those who are least motivated to pursue extrinsic rewards who eventually receive them.” 9
This result is not true across all tasks, of course. Amabile and others have found that extrinsic rewards can be effective for algorithmic tasks—those that depend on following an existing formula to its logical conclusion. But for more right-brain undertakings—those that demand flexible problem-solving, inventiveness, or conceptual understanding—contingent rewards can be dangerous. Rewarded subjects often have a harder time seeing the periphery and crafting original solutions. This, too, is one of the sturdiest findings in social science—especially as Amabile and others have refined it over the years.10 For artists, scientists, inventors, schoolchildren, and the rest of us, intrinsic motivation—the drive do something because it is interesting, challenging, and absorbing—is essential for high levels of creativity. But the “if-then” motivators that are the staple of most businesses often stifle, rather than stir, creative thinking. As the economy moves toward more right-brain, conceptual work—as
more of us deal with our own versions of the candle problem—this might be the most alarming gap between what science knows and what business does.
Good Behavior
Philosophers and medical professionals have long debated whether blood donors should be paid. Some claim that blood, like human tissue or organs, is special—that we shouldn’t be able to buy and sell it like a barrel of crude oil or a crate of ball bearings. Others argue that we should shelve our squeamishness, because paying for this substance will ensure an ample supply.
But in 1970, British sociologist Richard Titmuss, who had studied blood donation in the United Kingdom, offered a bolder speculation. Paying for blood wasn’t just immoral, he said. It was also inefficient. If Britain decided to pay citizens to donate, that would actually reduce the country’s blood supply. It was an oddball notion, to be sure. Economists snickered. And Titmuss never tested the idea; it was merely a philosophical hunch.11
But a quarter-century later, two Swedish economists decided to see if Titmuss was right. In an intriguing field experiment, they visited a regional blood center in Gothenburg and found 153 women who were interested in giving blood. Then—as seems to be the custom among motivation researchers—they divided the women into three groups.12 Experimenters told those in the first group that blood donation was voluntary. These participants could give blood, but they wouldn’t receive a payment. The experimenters offered the second group a different arrangement. If these participants gave blood, they’d each receive 50 Swedish kronor (about $7). The third group received a variation on that second offer: a 50-kronor payment with an immediate option to donate the amount to a children’s cancer charity.