History

According to research on operant conditioning and behaviorism in the 1950s, extrinsic rewards should increase the chances of the rewarded behavior occurring, with the greatest effect on behavior if the reward is given immediately after the behavior.^[5] In these studies, often removing the reward quickly led to a return to the pre-reward baseline frequency of the behavior. These findings led to popular calls for the adoption of incentives as motivational tools in a variety of professional and educational contexts. Moreover, according to standard economics, providing extrinsic incentives for a behavior has an immediate relative-price effect which should produce more of that behavior by making that behavior more attractive. Literature in economics has myriad examples of this.^[6]

However, Titmuss argued that sometimes adding incentives can actually diminish the rewarded behavior.^[1] Exploring this idea, Edward Deci noticed in the early 1970s that some actions appear to provide their own reward. These behaviors are described as being intrinsically motivated, and their enjoyment or rewards come from the act of engaging in the task itself.^[7] In this case, behavior does not require any extrinsic reward.^[7]

These observations led researchers to ask how providing extrinsic rewards for a given activity would influence intrinsic motivation toward that activity. While the relative-price effect would predict that rewards should only enhance the attractiveness of the behavior, there appeared to be indirect psychological effects of offering extrinsic incentives that, in some cases, have the opposite effect of making the behavior seem less attractive.^[8]

Experimental manipulations

Early evidence

Early research in this area in the 1970s found that providing an extrinsic incentive for completing a task could undermine intrinsic motivation and subsequent effort devoted to that task across a broad range of contexts. This research considered the effect of monetary,^[11] tangible (e.g. gifts), and symbolic rewards^[12] among young children,^[13] college students, and adults^[14] doing a wide variety of tasks.^[15] In a classic study, Deci paid all subjects for participation in a psychological experiment that involved solving multiple puzzles or IQ test questions.^[16] Half of the subjects were paid a flat fee just for showing up to the experiment, but the other half of subjects were informed that they would be paid per their completion of the study's tasks. After the presumptive experiment was over, subjects were left with free time during which they could either sit idly or complete more tasks. Deci measured the number of additional IQ questions or puzzles completed during this non-compensated time as well as self-reported measurements of interest in the task. Deci found that, compared to those who were paid simply for showing up, subjects who were paid specifically to complete the tasks were significantly less likely to complete additional non-compensated tasks and gave lower ratings of interest levels in the tasks themselves. Deci interpreted these findings to suggest that motivation for and interest in the tasks had been displaced by the provision of extrinsic incentives.^[16]

These studies typically find that if incentives are large then, once removed, they can have long-run crowding out effects.^[17] However, more recent research has found that even if workers find incentives to be insufficient then there can be short-run crowding out of the rewarded behavior too.^[10][9] Examples of early crowding out studies include:

• In a pioneering study, Deci had college students attempt to solve a puzzle game called a Soma cube.^[13] During Phase I of the experiment, all subjects had the opportunity to play with the cube and attempt to solve several puzzles. During Phase II, half of the subjects (control) repeated Phase I whereas the other half of subjects were paid $1 for each puzzle they could solve. During Phase III, no one was paid but experimenters interrupted participants in the middle of the session, telling them a cover story about why the experimenter needed to leave the room for a few minutes. Secretly, experimenters were able to observe how participants spent their free time. Deci found that those who had been paid during Phase II were significantly less likely to play with the cube during the free, uncompensated time during Phase III.
• In a follow-up study that replicated the basic pattern of results from 1971, Deci later found that offering verbal praise as a reward for task completion did not have a similar backfiring effect as offering a monetary reward had.
• Kruglanski et al. found that if high school students were promised an extrinsic incentive before engaging in a variety of tasks, the students showed less creativity and subsequently reported enjoying the task less compared to those who were not promised payment at the outset.
• Lepper et al. found that children who were told that they would receive a reward in exchange for drawing—something they had previously shown to be intrinsically interested in—subsequently became less interested in drawing after the reward was given, compared to those who received a reward unexpectedly or who received no reward at all.^[12]

Other research has shown that a similar effect of crowding out can occur from negative disincentives for behavior, too. For instance, economic studies have shown that increasing penalties can actually lower obedience with the law and decrease worker performance.^[18][19] While all of these early investigations demonstrated that providing extrinsic incentives could undermine motivation for the rewarded behavior, researchers had not yet established the psychological process involved that could explain this consistent pattern of results.

Motivational theories

On this view—sometimes referred to as cognitive evaluation theory—the post-behavioral significance people assign to the reward determines subsequent motivation. Deci and Ryan argue that rewards can be seen to have two components: one that controls people's behavior and thus infringes on their autonomy, and a different, status-signaling component that enhances people's sense of competence. For instance, an employee recognition award could be seen as either the reason why an employee worked so hard in a given month (i.e. to win the award) or could be seen simply as a recognition of the employee's performance in general. If an extrinsic reward for some behavior appears to be controlling (i.e. the reason a person plausibly performed that behavior), this is argued to supplant intrinsic motivation for engaging in the behavior. Insofar, however, as the extrinsic incentive is seen not as an inducement but rather as a signal of high status or high achievement in general (e.g. a merit-based award), the incentive will engender more effort without crowding out motivation. On this account, then, the extent to which a given extrinsic incentive crowds out motivation is determined by the balance of the controlling versus status-signaling nature of the awards as perceived by the actor.^[20][8]

Attributional theories

The application of self-perception theory to motivation suggests that people sometimes form post-behavior judgments about the causes of their actions by considering the external circumstances of their decision. While intrinsic motivation for doing the activity might be a cause, the presence of an extrinsic reward could also be sufficient for explaining a behavior. The overjustification account of motivational crowding, most prominently advanced by Lepper et al., argues that people recognize the presence of a significant extrinsic incentive, attribute their motivation for doing the rewarded activity to the reward itself, and consequently lower their feelings of intrinsic motivation toward the activity. Thus, they infer, if effort for engaging in a task becomes too onerous or if an extrinsic reward is removed, people feel less internally motivated to engage in the task compared to those who were never offered a reward for doing so.^[12]

Behavioral theories

Various attempts have been made by behaviorists to explain the apparent phenomenon of crowding out in terms of reward conditioning.^[21] Behaviors that are typically thought to be intrinsically motivated, these theories argue, are actually motivated by the social praise they tend to engender. Dickinson argues that part of the reason why these behaviors are socially praised is precisely because they are not connected with any particular reinforcers. When a person helps someone else, he argues, he receives praise in part because there does not seem to be any specific private incentive for doing helping. Thus, the introduction of a specific reinforcer such as an extrinsic reward lowers the public praise, Dickinson argues. If the loss of praise is larger than the size of the specific reinforcer, she argues, then free-choice selection of that behavior will decrease. Hence, what appears as crowding out of intrinsic motivation can instead be explained, according to these theories, by shifting perceptions and incentives.^[10]

Economic utility theories

Some have argued that certain utility functions can be modeled to explain crowding out. Bénabou and Tirole, for instance, have theoretically established that crowding out can reliably occur if an agent's utility function for some behavior is composed of three things: intrinsic motivation, extrinsic motivation, and image-signaling concerns.^[22]