Short answer: The Overjustification Effect is the drop in how much people freely do an activity they already enjoyed, after it starts getting externally rewarded. Mark Lepper, David Greene, and Richard Nisbett demonstrated it in 1973, building on Edward Deci’s 1971 puzzle experiment.
The cause is attribution. People see the reward and conclude they were doing the activity for the reward, not for itself.
Not all rewards backfire. Deci, Koestner and Ryan’s 1999 meta-analysis of 128 experiments found expected tangible rewards cut intrinsic motivation by about d = -0.40, while verbal praise raised it by about d = +0.31.
Walk into any modern office, classroom, or app and you will find a manager, a teacher, or a product owner trying to motivate someone with a reward. The bonus, the gold star, the badge, the points, the variable cash incentive sitting at the bottom of the next push notification. The assumption underneath all of it is that motivation is additive — that giving a person a reward for doing something is, in the worst case, neutral, and in the best case a multiplier on whatever drive they already have.
The Overjustification Effect says the assumption is wrong, and that the worst case is not neutral. It says that for behaviors a person already wants to do, paying them to do it can systematically poison the want. Mark Lepper, David Greene, and Richard Nisbett demonstrated it in 1973 with a roomful of preschoolers, magic markers, and a Good Player Award; they watched a perfectly enjoyable activity collapse into a chore the moment the children’s drawing was rebranded as work-for-pay. Half a century later the finding has been replicated, meta-analyzed, contested, defended, qualified, and absorbed into the load-bearing assumptions of behavioral economics, gamification, education policy, and product design — and yet the assumption that “motivation is additive” still ships in nearly every onboarding flow, performance management system, and points economy in production today.

This is the S-Tier Behavioral Designer’s Guide to the Overjustification Effect: where the original 1973 study held up and where it has been narrowed by replication, what the cognitive evaluation theory machinery underneath it actually predicts, why “all rewards are bad” is the wrong takeaway and why the meta-analytic record is more interesting than either the folk version or the backlash version, and how to map the whole thing onto the Octalysis Framework so that the difference between a Black-Hat reward that erodes Core Drive 4 (Ownership & Possession) and a White-Hat reward that amplifies it stops being a cocktail-party distinction and becomes a screen-by-screen design decision.
Poisoned rewards are a recurring villain across my whole library, not just this guide. Hyperbolic Discounting shows the instant payout crowding out the long game, and the Endowment Effect shows why pulling a reward back stings far more than never offering it. I keep all of these frameworks, cross-linked and mapped to Octalysis, in my Behavioral Framework Library. When you want the interlocking map instead of one guide at a time, start there. For now, back to the preschoolers.
Speed Run Notes
- The Overjustification Effect is the documented decrease in intrinsic motivation that happens when an already-enjoyed activity is given an external reward, especially when that reward is expected, salient, and contingent on engagement.
- Lepper, Greene & Nisbett’s 1973 nursery-school study is the canonical proof: drawing time in free play dropped roughly in half for the Expected Award group versus the Unexpected and No-Award controls.
- Deci, Koestner & Ryan’s 1999 meta-analysis of 128 experiments confirmed that tangible expected rewards reliably reduce free-choice intrinsic motivation; verbal rewards generally enhance it.
- The mechanism is Cognitive Evaluation Theory inside Self-Determination Theory: rewards have a controlling face that erodes autonomy and an informational face that affirms competence — the controlling face is the toxic one.
- The effect is real but tightly conditional: requires baseline interest, expected and salient reward, engagement-contingency, and a target activity that the user already considers their own.
- Design implication: treat every reward layered onto an activity users already enjoy as a Cognitive Evaluation Theory audit before shipping, because overjustification is a slow-burn churn driver that A/B tests rarely catch.
In This Article
- What Is the Overjustification Effect
- The Core Findings
- What Lepper, Greene & Nisbett Got Right
- Where the Overjustification Effect Falls Apart
- The Brain on Overjustification
- Overjustification vs Other Theories
- Overjustification in the Real World
- The Elephant in the Room
- How to Apply Overjustification with the Octalysis Framework
- Practical Steps to Apply Overjustification
- Frequently Asked Questions
Author Credibility: Yu-kai Chou

Yu-kai Chou created the Octalysis Framework after studying gamification since 2003 — years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.
Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.
His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.
The reason the Overjustification Effect deserves a full S-Tier treatment, and not just a paragraph in a list of “things designers should know,” is that it sits on one of the two axes of the Octalysis Framework. The Left Brain Core Drives (2, 4 and 6) tend to rely on extrinsic motivation; the Right Brain Core Drives (3, 5 and 7) are the ones where the activity is rewarding in itself. The Overjustification Effect is what happens when a Left Brain reward gets bolted onto a Right Brain activity and the reward starts carrying the load the activity used to carry. Most onboarding flows, points economies and completion bonuses have at least one screen where a designer pays the user to do something the user was already doing for free. This guide is the long version of the audit for those screens.
What Is the Overjustification Effect
The Overjustification Effect is the phenomenon in which providing an external reward for an activity a person already finds intrinsically interesting reduces the person’s subsequent free-choice engagement with that activity. The reward, in effect, is “too much” justification for the behavior — beyond what the activity itself already supplied — and the cognitive system resolves the surplus by attributing the behavior to the reward rather than to the underlying interest. Once the reward is removed, the originally intrinsic motivation no longer carries the load it did before, and the activity is performed less than it was when no reward was in the picture at all.
The phrase itself comes from attribution theory. If two sufficient causes are present for a behavior — the activity is fun and there is a reward — the discounting principle Harold Kelley described in his 1972 work on causal schemata predicts that the perceiver downweights either cause when the other is sufficient on its own. Children watching themselves draw with markers and receive a Good Player Award conclude, in effect, that they are drawing because of the award, since the award is “enough” to explain the drawing all by itself. The intrinsic explanation gets discounted, the activity becomes mentally categorized as work-for-pay, and once the pay disappears, so does much of the work.
Definition Drilldown
For working purposes, three operating definitions clarify the construct without losing the technical content:
- The broad definition covers any case where an external reward reduces subsequent intrinsic motivation for an already-interesting activity. This is the version that shows up in popular writing and is often overgeneralized into “rewards always backfire.”
- The narrow operational definition from Lepper, Greene & Nisbett 1973 specifies four moderating conditions: the reward must be expected before the activity begins, salient enough that the person notices it, contingent on engagement (rather than on quality of performance), and offered for an activity that the person already finds intrinsically interesting at baseline. When all four conditions hold, the effect reliably appears.
- The Cognitive Evaluation Theory definition embedded inside Self-Determination Theory (Deci 1975; Ryan & Deci 2000) specifies the mechanism: rewards have a controlling functional significance and an informational functional significance, and the controlling face of a reward — the part that signals “I am doing this because someone is paying me” — undermines the perceived autonomy of the action and therefore the intrinsic motivation to perform it.
The three definitions are not in conflict. The broad definition names the phenomenon. The narrow operational definition tells you when to expect it. The Cognitive Evaluation Theory definition tells you why it happens. A serious behavioral designer keeps all three on hand: the broad version for client conversations, the narrow version for predicting which features will trigger it, and the mechanistic version for designing around it.
What Counts as a Reward Here
One reason the construct gets confused is that “reward” is treated as a single category by folk psychology and as a multi-dimensional space by the actual research. The dimensions that matter for whether a reward will trigger the Overjustification Effect are: tangibility (cash and physical goods are riskier than verbal praise), expectation (announced-up-front rewards are riskier than surprise rewards), contingency (rewards tied to mere participation are riskier than rewards tied to meeting an explicit performance standard), salience (rewards the user is constantly reminded of are riskier than rewards mentioned once and then receded), and competence-information (rewards that convey mastery information protect against the effect; rewards that feel purely controlling do not). Deci, Koestner & Ryan’s 1999 meta-analysis of 128 experiments organized the entire literature along these dimensions, and the result is more nuanced than the bumper-sticker version of overjustification: tangible expected engagement-contingent rewards reliably hurt; verbal informational rewards reliably help; performance-contingent rewards with informational feedback can go either way depending on whether they read as primarily controlling or primarily informational.
The Origin Story
Lepper, Greene & Nisbett ran the original 1973 study at the Bing Nursery School at Stanford. Fifty-one children between roughly three and five years old were preselected on the basis of showing high baseline interest in drawing with magic markers during free-play observations. Each child was then randomly assigned to one of three conditions. In the Expected Award condition, the child was told before drawing that they would receive a “Good Player Award” — a printed certificate with a gold star and a red ribbon — for their effort. In the Unexpected Award condition, the child drew with the markers without any mention of an award, and only at the end of the session was given the same award as a surprise. In the No-Award control condition, the child drew with no award mentioned and none given. One to two weeks later, the same markers were placed in the classroom during free play and an observer recorded how much time each child spent drawing with them. The result is the textbook chart: the Expected Award children drew at about half the rate of the other two groups. The other two groups, importantly, did not differ from each other or from baseline. The reward was not the problem; the expected, salient, engagement-contingent reward was the problem.
The choice of design details matters because it pinned down the four moderators that the meta-analytic literature later confirmed. The kids selected into the study were already intrinsically interested. The reward was announced before the activity. The reward was contingent on participation, not on quality. The reward was salient. Subtract any of those and the effect attenuates. Hold all of them and the effect ships predictably.
The Core Findings
The fifty-three years between the 1973 paper and the present have produced a cluster of findings about the Overjustification Effect that any designer who wants to use the construct in production should be able to recite. The headline holds up at the meta-analytic level. The boundary conditions are real and important. The mechanism is well-specified inside Cognitive Evaluation Theory. And the controversy of the 1990s has clarified the boundary conditions further rather than disconfirmed the effect.
Finding 1 — The Headline Effect Replicates
Deci, Koestner & Ryan’s 1999 meta-analysis pooled 128 controlled experiments through 1995 and found that tangible expected rewards contingent on doing, completing, or excelling at a task reliably reduced intrinsic motivation as measured by both free-choice behavior and self-reported task interest. The aggregated effect size for tangible expected engagement-contingent rewards on free-choice behavior was approximately d = -0.40, a moderate-sized negative effect. The aggregated effect size for the same reward family on self-reported interest was around d = -0.15, smaller but still negative and statistically reliable. Free-choice behavior — measured by how much time a participant spends on the activity when no one is offering them a reward to do so — is the cleanest field-relevant outcome, because it operationalizes exactly what designers care about: does the user keep coming back when the carrot is removed.
Finding 2 — Verbal Rewards Move the Other Way
Verbal rewards — praise, positive feedback, recognition — were associated with a positive effect on intrinsic motivation in the same Deci, Koestner & Ryan 1999 meta-analysis, with an aggregated d of approximately +0.33 for free-choice behavior and +0.31 for self-reported interest, as long as the praise was not delivered in a controlling manner. The free-choice benefit was carried by college students. For children it was small and not statistically significant (d = 0.11), so praise is not a free pass with kids. The standard interpretation inside Cognitive Evaluation Theory is that informational verbal rewards primarily affirm competence (CD2-flavored Development & Accomplishment) without thwarting autonomy (CD4-flavored Ownership & Possession), and the net effect on intrinsic motivation is positive. The implication for designers is direct: verbal recognition that affirms competence on a user’s own terms is structurally different from a points-and-badges reward economy, and the two should not be conflated under a single “extrinsic motivation” label.
Finding 3 — Unexpected Rewards Are Mostly Safe
The original 1973 finding that the Unexpected Award condition did not differ from the No-Award control has held up across the meta-analytic record. Unexpected, surprise rewards, even tangible ones, do not reliably reduce intrinsic motivation, because the agent did not have the chance to attribute their engagement to the reward in advance — the activity is already underway when the reward arrives. The design implication is that unexpected appreciation, surprise upgrades, and post-hoc bonuses sit in a different risk class than expected, announced rewards, and can be deployed without triggering the overjustification mechanism. This is one of the cleanest distinctions in the literature and one of the most consistently overlooked in product design.
Finding 4 — Engagement and Completion Rewards Do the Most Damage
Within the family of expected tangible rewards, the meta-analysis found the largest free-choice drops for engagement-contingent rewards (tied to merely doing the task, d = -0.40) and completion-contingent rewards (tied to finishing it, d = -0.36). The two are close enough that a designer should treat them as one risk class rather than rank them. Performance-contingent rewards (meet a standard and get the reward) also undermined free-choice behavior on average (d = -0.28), and they sit in a more ambiguous space: they can preserve or even enhance intrinsic motivation when the standard reads as informational competence feedback, and they undermine it when the standard reads as a controlling target. The pattern is consistent with the Cognitive Evaluation Theory distinction between informational and controlling functional significance: the more a reward feels like “you must do this,” the worse it performs on intrinsic motivation; the more it feels like “here is a marker of how well you did,” the better.
Finding 5 — Initial Interest Is a Necessary Condition
The Overjustification Effect requires baseline intrinsic interest. If the activity is boring, dull, or aversive at baseline, adding an extrinsic reward does not undermine intrinsic motivation, because there is no intrinsic motivation to undermine. In fact, in low-interest activities, extrinsic rewards reliably increase engagement during the reward period, exactly as folk theory predicts. The implication is that the Overjustification Effect is a tax that designers pay only on the most valuable parts of their product — the screens, surfaces, and behaviors that users actually enjoy. The dull mandatory-compliance flows are immune. This is the deeply uncomfortable design truth: rewarding the boring features helps; rewarding the loved features can hurt.
What Lepper, Greene & Nisbett Got Right
It is easy, half a century later, to forget that the 1973 paper was contrarian when it shipped. Behaviorism was the dominant paradigm in motivation research, and the prediction from a strict reinforcement-learning view was the opposite of what Lepper, Greene, and Nisbett found. A reward should reinforce a behavior. Reinforced behaviors should occur more often, not less. The fact that the children in the Expected Award condition drew less in free play after being rewarded was a frontal challenge to the prevailing theory, and the design choices in the original paper made the challenge difficult to dismiss.
They Built In the Right Controls
The three-condition design — Expected, Unexpected, No Award — is a small thing on paper but a load-bearing thing in interpretation. Including the Unexpected Award condition was the move that ruled out the obvious behaviorist alternative explanations. If rewards in general damaged intrinsic motivation, the Unexpected condition should have shown the same drop. It did not. This single design decision narrowed the explanatory space to “expectation and contingency are the toxic variables,” which is exactly where the subsequent fifty years of research converged. The discipline of putting the rival explanation in the design rather than dismissing it in the discussion is what gave the original paper its staying power.
They Hooked Into a Real Mechanism
The mechanism Lepper et al. proposed — an attribution-based discounting of intrinsic interest in the presence of a sufficient extrinsic cause — was not just a description of the result, it was a claim about cognition that made testable predictions. The prediction that unexpected rewards should not produce the effect, that verbal rewards should not produce the effect to the same extent, and that the effect should be larger when the reward is more salient, all flowed naturally from the attribution mechanism. The fact that subsequent research confirmed those predictions across multiple labs and populations is the strongest evidence that the original framing was identifying something structural rather than picking up an artifact. Self-Perception Theory (Bem 1967) and Cognitive Evaluation Theory (Deci 1975) layered additional mechanism-level detail on top of the original attribution account, and the layered mechanism remains the standard explanation today.
They Picked the Right Outcome Measure
The choice to measure free-choice behavior — how much time the child spent with the markers when the experimenter was no longer in the room and no reward was on offer — instead of in-task performance or self-report alone is the methodological move that converted the result into a useful design construct. Self-reported interest is downstream of social desirability and post-hoc rationalization. Performance during the reward period is downstream of the reward itself and tells you nothing about what happens after the carrot is removed. Free-choice behavior in a follow-up session asks the only question that matters from a retention design perspective: when no one is paying you, does the user still come back. The answer in the Expected Award condition was: half as often. That is the operational definition of churn caused by overjustification.
Where the Overjustification Effect Falls Apart
The contested-construct discipline asks the honest question: where does this finding stop being load-bearing, and what should a designer worry about when they read a confident claim about it. The Overjustification Effect is not an exception to that discipline. The construct is real but tightly bounded, and three critiques in the literature deserve to be carried into every design conversation, not buried in a discussion section.
Critique 1 — The Cameron & Pierce 1994 Challenge and Its Resolution
The single most consequential challenge to the Overjustification Effect was the 1994 meta-analysis by Judy Cameron and W. David Pierce, published in Review of Educational Research. Their analysis of 96 experiments concluded that the negative effect of rewards on intrinsic motivation was largely an artifact of methodological idiosyncrasies, that the average effect was near zero across the broad reward category, and that the practical implication was that educators and managers should feel comfortable using rewards without fear of damaging intrinsic motivation. The paper provoked one of the more consequential methodological fights in motivation psychology, and it was eventually resolved in favor of the original finding — but with the boundary conditions sharpened in the process. Deci, Koestner & Ryan’s 1999 meta-analysis re-coded the same studies (and added the studies Cameron and Pierce had excluded) and showed that when the studies were partitioned by reward type and contingency the negative effect on free-choice intrinsic motivation reliably appeared in the expected-tangible-engagement-contingent cells. The Cameron and Pierce conclusion, in other words, was correct about the average across all reward types, including the verbal and unexpected ones that move intrinsic motivation in the opposite direction; it was wrong as a claim about the toxic variant the original literature had identified. The lesson for production design is that the construct does not shrink to zero — but it also does not generalize to “all rewards are bad.” The careful inference is the only one that survives both meta-analyses, and it is the inference Deci and colleagues had been making since 1971.
Critique 2 — The Effect Is Highly Conditional and Hard to Detect Without Long-Term Measurement
Even taking the 1999 meta-analysis at face value, the Overjustification Effect is structurally hard to detect in production telemetry. The toxic version of the effect shows up in free-choice behavior after the reward is removed, sometimes weeks later in classroom and field studies. Standard A/B testing infrastructure rarely runs experiments long enough to capture this dynamic, and standard product analytics rarely separate “engagement during the reward period” from “engagement after the reward period.” A points-and-badges economy that shows a reliable lift in week-one engagement may be hiding a slow-burn intrinsic-motivation tax that only manifests as a churn signal three or six months later, by which point the connection back to the design decision is buried under a dozen confounders. Cerasoli, Nicklin & Ford’s 2014 meta-analysis of 183 independent samples adds a related warning from the performance side. Intrinsic motivation predicted performance whether or not incentives were present, but it mattered less when incentives were tied directly to performance and more when they were tied only indirectly; intrinsic motivation was the better predictor of quality, incentives the better predictor of quantity. That is not a direct measure of crowding out, but it points the same way: the more salient the incentive, the more the incentive, rather than the person’s own interest, ends up steering the work. The implication for designers is sobering. The behaviors most worth measuring with respect to the Overjustification Effect are precisely the behaviors product analytics infrastructure measures worst: long-tail free-choice engagement, multi-month retention conditional on incentive removal, and the differential between users who internalized the activity and users who learned to expect the reward.
Critique 3 — Cultural and Individual Moderation Limits Generalization
The original studies were conducted with predominantly Western, middle-class, white populations, and the cross-cultural generalizability of the effect has been more mixed than the popular literature acknowledges. Iyengar & Lepper (1999) found that Asian-American children were most intrinsically motivated when a trusted member of their in-group chose the task for them (their mother in one study, their classmates in another), while Anglo-American children did best when they chose for themselves. That study manipulated choice, not rewards, so it speaks to the autonomy mechanism rather than to the reward effect directly. The Cognitive Evaluation Theory account is salvageable here — autonomy is being supported, just relationally rather than individually — but the operational prediction that “external rewards always damage intrinsic motivation” does not generalize uniformly across cultural contexts, and a designer working on a global product should not treat the original 1973 American-sample finding as a universal law. Individual-difference moderators are similarly real. Hagger & Chatzisarantis (2011) found the reverse of what intuition suggests: in a paid puzzle-task experiment, people high in control orientation (measured with Deci & Ryan’s General Causality Orientations Scale) showed the classic undermining effect, while autonomy-oriented people were protected from it. The same reward can be a CD4 (Ownership) violation for one user and a CD2 (Accomplishment) signal for another. Population-level effects in the literature average across this individual variance and obscure design-relevant heterogeneity.
The Brain on Overjustification
The neuroscience of the Overjustification Effect did not exist when Lepper, Greene, and Nisbett ran the 1973 study. A small number of imaging studies have since looked at what happens in the brain when an extrinsic reward attaches itself to an intrinsically motivated activity. The evidence is much thinner than the behavioral record and should be read that way, but what exists points in the same direction as the Cognitive Evaluation Theory account.
The Striatal Reward System
The ventral and dorsal striatum, particularly the nucleus accumbens, encode prediction errors for both primary rewards (food, water) and secondary rewards (money, social approval). Murayama, Matsumoto, Izuma & Matsumoto (2010) ran the best-known fMRI test: participants performed a stopwatch task either for a monetary reward or for the inherent enjoyment of accurate timing. When everyone performed the task again after the payment had stopped, the previously paid group showed reduced activity in the anterior striatum and lateral prefrontal areas compared with the group that had never been paid. The paid group also spent less of an unrewarded break voluntarily playing with the stopwatch, which is the behavioral undermining effect, measured outside the scanner. It is one small study, so treat it as a strong pointer rather than settled neuroscience. What it does show is that the effect is not just a verbal-report artifact. Something measurable changes in the brain’s reward processing of the activity itself when the activity has been externally rewarded.
Intrinsic and Extrinsic Reasons Look Different to the Brain
Lee, Reeve, Xue & Xiong (2012) used fMRI to compare how the brain handles intrinsic reasons for acting against extrinsic ones. Participants considered reasons for doing things rather than being paid for a task and then having the payment withdrawn, so the study does not test the undermining effect itself. It supports a narrower point: “because I want to” and “because I am being paid” are represented as different kinds of reason, which is the precondition for a reward being able to re-label an activity.
Autonomy-Supportive Versus Controlling Rewards
Behaviorally, the controlling-versus-informational distinction is well supported: Deci, Koestner & Ryan’s meta-analysis and the causality-orientation work above both show that how a reward lands matters as much as whether it is given. Neurally, it is not yet pinned down. Murayama and colleagues’ follow-up (2015, Cerebral Cortex) found that simply letting people choose their task improved performance, with the ventromedial prefrontal cortex playing a key role, which fits the autonomy side of the story. No imaging study cited here holds reward value constant and varies only controlling versus informational framing. The practical conclusion survives anyway: spend design effort on how a reward is framed, not only on how big it is.
The Practical Take
The neural evidence does not change the design implications, but it adds weight to them. The Overjustification Effect is not just a story participants tell on a questionnaire: in the one direct imaging test, paying for an enjoyable task changed how the brain responded to that task after the payment stopped. That is consistent with the risk this guide keeps returning to. A loved feature that has been paid for may not be the same feature to the user once the payment stops.
Overjustification vs Other Theories
The Overjustification Effect does not live alone in the motivation literature. It sits at the intersection of several adjacent constructs that designers regularly conflate, and the boundaries between them matter for choosing the right design intervention. Five comparisons are worth working through explicitly.
Versus Self-Determination Theory
The Overjustification Effect is the empirical phenomenon that Cognitive Evaluation Theory — the most developed sub-theory inside Self-Determination Theory — was designed to explain. Self-Determination Theory specifies that behavior is regulated along a continuum from amotivation through external regulation, introjected regulation, identified regulation, integrated regulation, to intrinsic motivation, and that movement toward the intrinsic end is supported by the satisfaction of three basic psychological needs (autonomy, competence, relatedness). The Overjustification Effect is the diagnostic case in which a controlling reward thwarts autonomy and shifts a previously intrinsic regulation toward external regulation, with the predicted decline in free-choice engagement. If you want one motivation framework on the wall and you can only have one, Self-Determination Theory is the one — the Overjustification Effect is the clearest example of why.
Versus Operant Conditioning
The Skinner-tradition behaviorist position predicts that any reinforcer will increase the rate of the behavior it follows, full stop. The Overjustification Effect is the canonical counterexample to that prediction in humans, and the resolution is that human behavior is mediated by attribution and self-perception in ways that animal operant chambers are not. Under operant conditioning, the reward changes the behavior directly. Under the cognitive account that the Overjustification Effect supports, the reward changes the agent’s representation of why they are doing the behavior, and the behavior changes downstream of the representation. The two views can be reconciled — operant principles still apply for behaviors with no baseline intrinsic interest — but the Overjustification Effect demarcates the boundary at which the operant model breaks down for human cognitive agents.
Versus Self-Perception Theory
Bem’s 1967 Self-Perception Theory provides the underlying mechanism for the Overjustification Effect. People infer their own attitudes and motivations by observing their own behavior, the same way they would infer another person’s attitudes from observed behavior. If I observe myself drawing for a Good Player Award, I infer that I draw for awards. If I observe myself drawing with no award offered, I infer that I draw for the joy of drawing. The Overjustification Effect is what Self-Perception Theory predicts will happen when extrinsic rewards become salient enough to shift the self-attribution. Self-Perception Theory is the broader construct; the Overjustification Effect is the specific design-relevant case.
Versus Goal-Setting Theory
Locke and Latham’s Goal-Setting Theory argues that specific, challenging goals reliably produce better performance than vague or easy goals, and the Goal-Setting tradition often pairs goals with rewards as part of the standard performance-management toolkit. The Overjustification Effect is the warning label on that pairing. Specific challenging goals are likely to enhance performance during the goal period; performance-contingent rewards layered on top of those goals can preserve or enhance intrinsic motivation when they read as informational competence feedback, and undermine it when they read as controlling. The Goal-Setting tradition and the Overjustification tradition are not opposed but they are not the same thing, and the design move that respects both is to set goals while keeping the controlling framing of rewards low — the goal supplies the structure, the reward affirms competence, neither is presented as a controlling demand.
Versus Expectancy Theory
Vroom’s Expectancy Theory models motivation as the product of expectancy (effort yields performance), instrumentality (performance yields reward), and valence (the reward is desired), and is the standard managerial framework for thinking about pay-for-performance. The Overjustification Effect sits underneath Expectancy Theory as a quiet correction: the valence term is not a constant. The valence of the reward changes the valence of the underlying activity, and a designer who optimizes only the Expectancy formula without auditing the impact of instrumentality on intrinsic motivation will reliably produce systems that look optimal in the short run and erode in the long run. The two theories are not in conflict so much as Expectancy is a horsepower-and-acceleration model and Overjustification is the brake-pad-wear curve underneath it. You need both to operate the vehicle.
Overjustification in the Real World
The construct earns its keep when it predicts what will happen in the field, not just in the lab. Four domains demonstrate the operational footprint of the Overjustification Effect in current production design, and in each one the same lesson recurs: the behaviors most worth protecting are the behaviors most easily damaged by a sloppy reward.
Education and Reading Programs
Pizza Hut’s Book It! program, which has paid children in personal pan pizzas for reading since 1984, is the case critics reach for first. Alfie Kohn (1993) argued that it teaches children to read for pizza rather than for the story, and that is a fair application of the theory. It is a prediction, though, not a measured finding: nobody has tracked Book It! readers’ later reading-for-pleasure against a control group. The best randomized evidence is more interesting. Roland Fryer (2011) ran incentive trials in Chicago (paying for grades), New York (paying for test performance) and Dallas (paying second-graders $2 per book read). The overall effects were statistically zero in each city, the one significant gain was among English-speaking students in Dallas, and Fryer reported little evidence that the payments decreased intrinsic motivation. Reading for most second-graders is still effortful, not yet a loved activity, which is the low-baseline-interest case where the theory says rewards can help. The design move that protects intrinsic interest is to reward inputs and skill development informationally rather than outputs controllingly, to watch most closely the children who already read for fun, and to phase rewards out once the activity has been internalized.
The Workplace and Pay-for-Performance
The corporate compensation literature has rediscovered the Overjustification Effect every decade since the 1970s under different labels: “crowding out” in Bruno Frey’s economic-psychology tradition, “the hidden costs of reward” in Lepper and Greene’s edited volume, Daniel Pink’s critique of “if-then” rewards in Drive. The empirical pattern in workplace data is that pay-for-performance produces measurable lift on simple, well-defined, low-creativity tasks, and produces null or negative effects on complex, ill-defined, creativity-dependent work. Glucksberg’s 1962 candle-problem experiments are the canonical demonstration: monetary incentives speeded performance on the procedural version of the task and slowed it on the insight version. Ariely, Gneezy, Loewenstein & Mazar (2009) showed that very large monetary stakes worsened performance on cognitively demanding tasks across a range of populations. The implication for managers and product designers is the same: the more the work in question depends on intrinsically motivated discretionary effort, the riskier the layered extrinsic incentive, and the better-served the system is by ensuring that compensation feels fair and adequate while leaving the actual day-to-day motivation to come from the work itself. This is the empirical foundation of what Pink popularized as Autonomy-Mastery-Purpose, and it is one of the few cases where the popular distillation tracks the research closely.
Gamification and Points Economies
The case of greatest interest to designers reading this guide is the gamification-of-already-loved-features case. A user is enthusiastically using a feature, and the product team decides to add a points-and-badges economy on top of that feature to “boost engagement.” The Overjustification Effect predicts that the addition will produce a short-run lift on engagement metrics, a slow erosion of free-choice engagement once the points become routine, and a measurable drop on retention conditional on incentive removal. Some of that now has field evidence. When GitHub quietly removed the public activity-streak counters from developer profiles, Moldon, Strohmaier & Wachs (2021) found that long streaks were abandoned, weekend activity fell, and some developers chasing a 100-day goal gave it up, which shows how much of the streak behavior the counter itself had been carrying. In a semester-long classroom study, Hanus & Fox (2015) found that students in a course gamified with badges and a leaderboard ended up with lower intrinsic motivation, lower satisfaction and lower final-exam scores than students in the same course without them. The record is not one-sided: Mekler and colleagues (2017) found that points, levels and leaderboards raised performance on an image-tagging task without lowering intrinsic motivation. Social-media engagement design runs the same risk at another scale, since every “like” notification is structurally an engagement-contingent reward whose cost per event is small but repeats millions of times. The right design response is not to never use points-and-badges but to be specific about which Core Drive a given mechanic is targeting and which it is putting at risk.
Healthcare and Behavior Change
Cash incentives for health behaviors (quit-smoking bonuses, exercise rebates, weight-loss prizes) are the most rigorously tested reward programs outside the lab, and they mostly show the opposite of overjustification. In the best-known trial, Volpp and colleagues at the University of Pennsylvania (2009) offered 878 General Electric employees up to $750 for quitting smoking. Quit rates were roughly three times higher than in the information-only group at nine to twelve months (14.7% vs 5.0%), and the gap held at fifteen to eighteen months, after the payments had ended (9.4% vs 3.6%). That is what the theory predicts for a behavior with little baseline intrinsic appeal: there is no prior love of quitting for the reward to crowd out, so the incentive works as scaffolding. The lesson for designers is the boundary condition again. Health incentives are a strong tool for behaviors people do not already want to do, and they say little about what happens when you pay people for something they already enjoy.
The Elephant in the Room
The elephant in the room of the Overjustification Effect is that the construct is often invoked as an excuse for designers to ship low-effort points-and-badges economies and then call them “gamification” — and almost as often invoked as a reason for product leaders to refuse to use any extrinsic motivation at all on the grounds that it will damage intrinsic motivation. Both moves are wrong, and the both-wrong observation is the one a serious behavioral designer needs to be able to articulate.
The first elephant is that the loud version of the construct — “rewards always backfire, never use them” — is a popular distillation that the underlying research does not support. Verbal informational rewards reliably enhance intrinsic motivation. Unexpected rewards do not reliably damage it. Performance-contingent rewards delivered with informational framing can preserve it. Cash incentives on behaviors with no baseline intrinsic interest reliably increase those behaviors, and the increase often persists past the incentive period if the behavior was internalized along the way. A behavioral designer who refuses to use any extrinsic motivation under the banner of “the Overjustification Effect” is leaving entire classes of legitimate design intervention on the table, and is doing so on the basis of a popular reading the literature does not endorse.
The second elephant is that the cynical version of the construct — “Cameron and Pierce 1994 disproved the effect, so points and badges are fine” — is also wrong. The 1999 re-analysis by Deci, Koestner, and Ryan resolved the dispute in favor of the original finding for the toxic variant the literature had identified. Tangible expected engagement-contingent rewards on intrinsically interesting activities reliably reduce free-choice intrinsic motivation. The fact that the average across all reward types across all studies is small does not mean the toxic variant is small. It means the verbal and unexpected rewards that move the metric in the opposite direction are washing out the average. A designer who reads only Cameron and Pierce 1994 and concludes that the Overjustification Effect is a methodological artifact is missing the design implications the meta-analytic literature has actually produced.
The honest synthesis is that the Overjustification Effect is a real, mechanistically grounded, neurally plausible, meta-analytically supported phenomenon with tightly specified boundary conditions, and the design discipline it asks of you is to know which boundary condition you are in before you ship. The boundary conditions are: is there baseline intrinsic interest in the target activity, is the reward expected, is the reward salient, is the reward engagement-contingent, is the reward presented controllingly, and is the activity culturally framed as autonomous or relationally embedded. Run those questions across every reward in your design surface and you will find a small set of mechanics that are obviously high-risk, a larger set that are clearly low-risk, and a middle set that requires either an A/B test long enough to capture post-reward free-choice behavior or a careful judgment call from someone who has actually read the literature. That middle set is where most production design decisions live, and it is the set where the Overjustification Effect earns its place on the wall.
How to Apply the Overjustification Effect with the Octalysis Framework
The Octalysis Framework treats motivation as the interaction of eight Core Drives arranged around an octagon, with the top three (Core Drive 1: Epic Meaning & Calling, Core Drive 2: Development & Accomplishment, Core Drive 3: Empowerment of Creativity & Feedback) classed as White Hat, the bottom three (Core Drive 6: Scarcity & Impatience, Core Drive 7: Unpredictability & Curiosity, Core Drive 8: Loss & Avoidance) as Black Hat, Core Drives 2, 4 and 6 on the left as Left Brain (extrinsic tendency), and Core Drives 3, 5 and 7 on the right as Right Brain (intrinsic tendency), with Core Drive 1 at the top center and Core Drive 8 at the bottom center. The two axes measure different things. White Hat versus Black Hat is about how the motivation feels: powerful and in control at the top, urgent and out of control at the bottom. Left Brain versus Right Brain is about where the motivation comes from, and that is the axis the Overjustification Effect lives on. Without the effect, the Left Brain and Right Brain split would be a tidy way to sort mechanics. With it, the split becomes a prediction about what happens to free-choice engagement after a reward is removed.
Primary Core Drive: CD4 Ownership & Possession (the Drive Being Threatened)
The Overjustification Effect is, at its core, an attack on Core Drive 4 (Ownership & Possession). The activity that the user previously felt they owned — they were drawing because they wanted to draw, exercising because they liked exercising, contributing to the open-source project because they cared about it — is reattributed by the user themselves to an external party who is paying them. Ownership of the activity transfers from the user to the system, and the moment the system stops paying, the user no longer feels the ownership that was carrying the engagement. The design move that defends Core Drive 4 against the Overjustification Effect is to make the user’s ownership of the activity highly salient at every reward delivery — the language of “your streak,” “your collection,” “your progress,” and the practice of letting the user choose which subset of the activity to engage with — so that the controlling functional significance of the reward is diluted by an autonomy-supportive frame. This is the root reason that the Octalysis Framework places “Game Techniques” like Avatar (#13), Collection Set (#16), and Alfred Effect (#83) inside CD4: each one is a structural way of routing reward delivery through a sense of personal ownership, which lowers the controlling reading of the reward and protects the underlying intrinsic motivation.
Supporting Core Drive: CD3 Empowerment of Creativity & Feedback (the Intrinsic Motivation Being Crowded)
Core Drive 3 (Empowerment of Creativity & Feedback) is where intrinsic motivation lives in the Octalysis Framework. The activities most vulnerable to the Overjustification Effect are activities the user is doing because they enjoy the creative or feedback-rich loop the activity offers — drawing, writing, coding, designing, playing a musical instrument, working on a craft. The design implication is that any reward layered onto a CD3 surface should be presented as informational competence feedback rather than as engagement-contingent payment. A “you mastered the chord progression” feedback signal is structurally different from a “you earned 50 points for practicing for 15 minutes” signal, and the difference shows up in post-reward free-choice engagement. The cleanest CD3-friendly reward design is one where the reward is the natural feedback the activity itself produces — better output, faster mastery, visible progress — and any meta-layer the system adds is curated to amplify the natural feedback rather than substitute a controlling alternative for it.
Supporting Core Drive: CD2 Development & Accomplishment (the Reward That Crowds)
Core Drive 2 (Development & Accomplishment) is the design surface where extrinsic rewards usually live: points, badges, levels, leaderboards, achievements. The Overjustification Effect is the structural risk on every CD2 mechanic that touches a previously CD3-driven activity. The design audit on every CD2 mechanic is whether the mechanic reads to the user as informational competence feedback (safe) or as controlling engagement-contingent payment (risky), and whether it is being layered onto a baseline-low-interest activity (where the operant lift is real and the overjustification cost is minimal) or a baseline-high-interest activity (where the lift is small and the cost compounds). The Octalysis Game Techniques inside CD2 — Step-By-Step Overlay Tutorials (#6), Progress Bar (#4), Crowning (#18) — are explicitly designed to lean toward the informational reading, but the framing of the surrounding language is what determines which reading the user actually performs.
The Black Hat Cluster: When Urgency Turns a Reward Controlling
The Black Hat Core Drives (CD6 Scarcity & Impatience, CD7 Unpredictability & Curiosity, CD8 Loss & Avoidance) are where Overjustification mechanics most often migrate when a designer optimizes a points-and-badges economy for short-run engagement. A streak that started as a CD2 accomplishment becomes a CD8 loss-avoidance trap when the user begins to engage primarily to avoid losing the streak. A loot-box reward that started as a CD7 unpredictability hit becomes a controlling engagement-contingent payment when the user begins to engage primarily for the variable reinforcement schedule. A countdown-timer scarcity that started as a CD6 urgency signal becomes overjustification when the user is no longer doing the activity for itself but to avoid letting the timer run out. Black Hat does not mean bad: people often choose Black Hat motivation on purpose, to get themselves to the gym or over the finish line of a project. The overlap with overjustification is narrower. Black Hat mechanics feel urgent by design, so when they are bolted onto an activity the user already enjoyed, they push the reward toward the controlling side of the Cognitive Evaluation Theory line, which is the side that does the damage.
The Endgame Phase Test
The cleanest production audit of an Octalysis design for Overjustification risk is the Endgame Phase test. The Octalysis 4 Experience Phases — Discovery, Onboarding, Scaffolding, Endgame — separate the lifetime of a user’s engagement into stages where different motivation profiles dominate. In Discovery and Onboarding, extrinsic Core Drives can do real work because the user has not yet developed an intrinsic relationship with the activity. By Scaffolding and Endgame, the user has either internalized the activity (in which case the design should pivot to White Hat, Right Brain Core Drives that support the intrinsic regulation) or has not (in which case the system is at risk of the multi-month free-choice drop the Overjustification Effect predicts). The audit question for every screen in Scaffolding and Endgame is: am I still rewarding this user the way I rewarded them in Onboarding, and if so, am I sustaining their intrinsic motivation or training them to wait for the reward.
Practical Steps to Apply the Overjustification Effect
Eight design moves convert the Overjustification Effect from a literature finding into a screen-by-screen audit a serious behavioral designer can run on production work.
Step 1 — Audit Every Reward for Its Functional Significance
For every reward in the design surface, ask whether it is functioning primarily as informational competence feedback or as controlling engagement-contingent payment. The audit unit is the screen, not the system. The same point reward can read informationally on one screen (“You leveled up — you have now mastered intermediate listening”) and controllingly on another (“Earn 50 points by completing today’s lesson”). Replace controlling framings with informational framings wherever the meaning is preserved, and treat any reward that cannot be re-framed as a candidate for removal.
Step 2 — Distinguish High-Interest from Low-Interest Activities
Map every behavior the system is trying to reinforce on a baseline-interest scale. Behaviors that users would do without any extrinsic motivation belong in the high-interest cell. Behaviors that users would not perform without compensation belong in the low-interest cell. The reward design should differ by cell: high-interest behaviors get verbal informational rewards or no reward at all, low-interest behaviors get the extrinsic incentive structure that the operant literature actually supports. Mismatching the reward type to the interest cell is the most common Overjustification failure in production design.
Step 3 — Make Unexpected Rewards a Standard Tool
The Unexpected Award condition in the original 1973 study did not produce the effect, and the meta-analytic literature confirms that surprise rewards are structurally safer than expected rewards. Build a deliberate cadence of unexpected appreciation into the design — an end-of-month gift to engaged users with no announcement, a surprise upgrade for a long-tenured customer, a thank-you note that arrives without having been promised. Unexpected rewards land as gifts, expected rewards land as payment, and the difference matters.
Step 4 — Use Verbal Informational Feedback Liberally
Verbal rewards reliably enhance intrinsic motivation when they affirm competence on the user’s terms. Build the system’s natural-language layer to deliver competence-affirming feedback on every meaningful interaction — but watch for the slip where verbal praise becomes controlling (“Good job following the instructions!” reads more controllingly than “You picked a clever path through that puzzle”). The design move is to make the praise specific, contingent on the user’s own choices, and free of language that frames the activity as something the user is being asked to do.
Step 5 — Phase Rewards Out as Internalization Stabilizes
The healthcare incentive literature supports a “scaffolding” use of incentives: pay enough during the early period to get the user past the initial behavioral barrier, and phase the reward out as the user stabilizes the new behavior. The phase-out is the critical design decision. Reward removal that is sudden and unexplained reads as withdrawal; reward removal that is paced, narrated, and tied to the user’s own demonstrated mastery reads as graduation. The graduation framing is what protects intrinsic motivation through the transition.
Step 6 — Set Performance-Contingent Rewards Carefully
Rewards contingent on meeting an explicit performance standard sit in the more ambiguous middle of the Cognitive Evaluation Theory landscape. They preserve intrinsic motivation when the standard reads as informational competence feedback (“you reached the proficient tier on the assessment”) and undermine it when the standard reads as a controlling target (“complete five lessons by Friday or lose your discount”). The design move that lands the reward on the informational side is to let the user choose the standard, frame achievement as a marker of mastery rather than as a hurdle to clear, and never present the reward as a condition of access.
Step 7 — Measure Free-Choice Behavior, Not Just Reward-Period Engagement
The Overjustification Effect is invisible to A/B tests that measure only the reward period. Build the analytics infrastructure to separate engagement during the reward period from engagement after the reward is removed, and run experiments long enough to observe the post-reward free-choice window. The honest dashboard for any reward mechanic shows three numbers: lift during the reward period (operant signal), free-choice engagement after the reward is removed (intrinsic motivation signal), and the difference between the two (overjustification cost). Most production analytics show only the first.
Step 8 — Treat Every Reward as a Cognitive Evaluation Theory Audit
The discipline that converts the Overjustification Effect from a piece of academic content into a design tool is to run the Cognitive Evaluation Theory audit on every reward in the system. The audit is four questions: does this reward primarily affirm autonomy or threaten it; does this reward primarily affirm competence or substitute for it; does this reward primarily affirm relatedness or position the user as a transactional agent; and does the cumulative effect of this reward and its peers across the surface push the user’s regulation toward the intrinsic end of the SDT continuum or the external end. Run those questions across every reward delivery in the design surface, and the points-and-badges economies that started looking inevitable will start looking optional, and the optional ones that were doing real work will start looking earned.
Closing Thoughts
The Overjustification Effect is the clearest research case for the Left Brain / Right Brain distinction in the Octalysis Framework, and it is the single piece of motivation research most worth knowing if your job involves designing systems that depend on long-run discretionary engagement. The headline holds: tangible expected engagement-contingent rewards on intrinsically interesting activities reduce free-choice engagement after the reward is removed. The mechanism is well-specified inside Cognitive Evaluation Theory: the controlling functional significance of the reward thwarts autonomy and shifts regulation from intrinsic toward external. The boundary conditions are tight: the effect requires baseline interest, expected and salient rewards, engagement-contingency, and a controlling presentation. Verbal informational rewards and unexpected rewards are not in the same risk class. Cultural and individual moderators are real and a global product design cannot rely on the original 1973 American sample as a universal law.
The right design move is calibrated extrinsic motivation. Use rewards where the operant literature actually supports them — on baseline-low-interest behaviors, on early-phase scaffolding, on unexpected appreciation — and protect the high-interest behaviors that the rewards would damage. Frame every reward informationally rather than controllingly. Phase rewards out as users internalize the activity. Measure free-choice behavior, not just engagement during the reward period. Treat the reward audit as a screen-by-screen Cognitive Evaluation Theory question rather than a one-time architecture decision. And remember that the popular reading of the construct — “rewards always backfire” — is a distortion of the literature in the same way the cynical reading — “Cameron and Pierce disproved it” — is. The honest synthesis is the only one that survives both meta-analyses, and it is the one a serious behavioral designer carries forward into production work. If you want the load-bearing application of this in product work, start with the Octalysis Framework and the Core Drive 4: Ownership & Possession pillar — the two pages that turn the Overjustification Effect from a research finding into a screen-by-screen design audit.
The Overjustification Effect is fifty-three years old and it has not aged. The design discipline it demands is the same one it demanded in 1973: know which reward you are giving, know which behavior you are rewarding, know whether the user already wanted to do that behavior, and pay attention to what happens to free-choice engagement once the reward is gone.
Frequently Asked Questions
What is the Overjustification Effect in simple terms?
The Overjustification Effect is the documented decrease in a person’s free-choice engagement with an activity they previously enjoyed, after the activity has been externally rewarded. Adding a reward to an already-enjoyed task can reduce how much the person does that task once the reward is removed, because the person re-attributes their previous engagement to the reward rather than to the activity itself. The effect is reliable when the reward is expected, salient, contingent on engagement, and presented in a controlling rather than informational frame.
What is an example of the Overjustification Effect?
The original 1973 study is the cleanest example. Preschoolers who already loved drawing with markers were promised a Good Player Award for drawing. One to two weeks later, during free play, that group spent about half as much time drawing as children who got no reward or an unexpected one.
Who first identified the Overjustification Effect?
Mark Lepper, David Greene, and Richard Nisbett demonstrated the effect in their 1973 paper “Undermining children’s intrinsic interest with extrinsic reward: A test of the ‘overjustification’ hypothesis,” published in the Journal of Personality and Social Psychology. Edward Deci’s 1971 puzzle-solving experiment with college students preceded the Lepper, Greene & Nisbett study and identified the same phenomenon under a different framing. Both are foundational citations.
What is the difference between intrinsic and extrinsic motivation in the Overjustification Effect?
Intrinsic motivation is the drive to perform an activity for the inherent satisfaction of the activity itself; extrinsic motivation is the drive to perform an activity for a separable outcome such as a reward, recognition, or punishment avoidance. The Overjustification Effect describes what happens when the introduction of extrinsic motivation crowds out pre-existing intrinsic motivation: the activity gets recategorized as something done for the external outcome, and the originally intrinsic engagement diminishes once the external outcome is removed.
Does the Overjustification Effect mean all rewards are bad?
No. The meta-analytic literature is specific about which kinds of rewards are risky: tangible, expected, engagement-contingent rewards on already-interesting activities, presented in a controlling frame. Verbal informational rewards reliably enhance intrinsic motivation; unexpected rewards do not reliably damage it; performance-contingent rewards delivered with informational framing can preserve it; and rewards on baseline-low-interest activities reliably increase engagement without crowding out an intrinsic motivation that was not there to begin with. The popular reading “rewards always backfire” is an oversimplification that the literature does not support.
How does the Overjustification Effect relate to Self-Determination Theory?
The Overjustification Effect is the empirical phenomenon that Cognitive Evaluation Theory — the most developed sub-theory inside Self-Determination Theory — was designed to explain. Cognitive Evaluation Theory specifies that rewards have a controlling functional significance and an informational functional significance, and that controlling rewards thwart autonomy and undermine intrinsic motivation while informational rewards affirm competence and preserve it. The Overjustification Effect is the diagnostic case for the controlling-side prediction; the broader Self-Determination Theory framework provides the mechanism that explains it.
Why does the Overjustification Effect happen? What is the mechanism?
The mechanism is attribution-based: in the presence of two sufficient causes for a behavior (the activity is interesting and there is a reward), the perceiver discounts one cause when the other is sufficient on its own. Self-Perception Theory (Bem 1967) extends the same logic to self-attribution: people infer their own motivations from observing their own behavior, and the presence of a salient extrinsic reward leads them to infer that they are doing the activity for the reward rather than for itself. Cognitive Evaluation Theory adds the autonomy-thwarting layer: controlling rewards undermine the perceived autonomy of the action, and the loss of perceived autonomy is what produces the downstream decrease in intrinsic motivation.
Is the Overjustification Effect replicated? What does the meta-analytic evidence say?
The Deci, Koestner & Ryan 1999 meta-analysis of 128 controlled experiments found a reliable negative effect of tangible expected engagement-contingent rewards on free-choice intrinsic motivation, with an aggregated effect size of approximately d = -0.40. Verbal rewards moved intrinsic motivation in the positive direction (d = +0.33 for free-choice behavior, carried mainly by college students). The earlier Cameron & Pierce 1994 meta-analysis had concluded the average effect across all reward types was near zero, but this conclusion was washing the toxic and beneficial reward types together; partitioning by type produces the differentiated pattern in the 1999 re-analysis. Cerasoli, Nicklin & Ford (2014), across 183 samples, found that intrinsic motivation still predicts performance when incentives are present, but matters less when incentives are tied directly to performance. The construct is replicated; the boundary conditions are tight.
Can the Overjustification Effect be reversed once it has happened?
There is little direct research on reversing it, which is a good reason to treat prevention as the main strategy. The interventions that work involve narrating the reward removal as a transition to autonomy (“you have mastered this and don’t need the points anymore”), restoring autonomy support around the activity, and rebuilding the user’s sense of personal ownership of the behavior through CD4 design moves. The general lesson is that prevention is much cheaper than recovery, which is the design implication that makes the Cognitive Evaluation Theory audit on every reward worth running before the reward ships.
How does the Overjustification Effect apply to gamification design?
The Overjustification Effect is the structural risk on every points-and-badges economy layered onto a previously intrinsically interesting activity. The design implication is that gamification mechanics should be matched carefully to baseline interest, presented in informational rather than controlling framing, phased out as users internalize the underlying behavior, and audited screen-by-screen with the Cognitive Evaluation Theory questions. In Octalysis terms, the risk sits on the Left Brain / Right Brain axis: Left Brain rewards (Core Drives 2, 4 and 6) layered onto Right Brain activities (Core Drives 3, 5 and 7) are where crowding out happens, and Black Hat urgency makes it worse by making the reward feel controlling.
What is the difference between expected and unexpected rewards in the Overjustification Effect?
Expected rewards are announced before the activity and create an explicit instrumental link in the user’s mind: “I am doing this in order to get that.” Unexpected rewards arrive after the activity is already underway and do not create the same instrumental link. The original Lepper, Greene & Nisbett 1973 study showed that the Unexpected Award condition did not differ from the No-Award control in subsequent free-play behavior, while the Expected Award condition showed roughly half the engagement of either control. The meta-analytic record has consistently confirmed this distinction. Surprise appreciation is structurally safer than promised payment.
References
- Lepper, M. R., Greene, D., & Nisbett, R. E. (1973). Undermining children’s intrinsic interest with extrinsic reward: A test of the “overjustification” hypothesis. Journal of Personality and Social Psychology, 28(1), 129–137.
- Deci, E. L. (1971). Effects of externally mediated rewards on intrinsic motivation. Journal of Personality and Social Psychology, 18(1), 105–115.
- Deci, E. L. (1975). Intrinsic Motivation. Plenum Press.
- Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74(3), 183–200.
- Cameron, J., & Pierce, W. D. (1994). Reinforcement, reward, and intrinsic motivation: A meta-analysis. Review of Educational Research, 64(3), 363–423.
- Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 125(6), 627–668.
- Ryan, R. M., & Deci, E. L. (2000). Intrinsic and extrinsic motivations: Classic definitions and new directions. Contemporary Educational Psychology, 25(1), 54–67.
- Cerasoli, C. P., Nicklin, J. M., & Ford, M. T. (2014). Intrinsic motivation and extrinsic incentives jointly predict performance: A 40-year meta-analysis. Psychological Bulletin, 140(4), 980–1008.
- Iyengar, S. S., & Lepper, M. R. (1999). Rethinking the value of choice: A cultural perspective on intrinsic motivation. Journal of Personality and Social Psychology, 76(3), 349–366.
- Murayama, K., Matsumoto, M., Izuma, K., & Matsumoto, K. (2010). Neural basis of the undermining effect of monetary reward on intrinsic motivation. Proceedings of the National Academy of Sciences, 107(49), 20911–20916.
- Lee, W., Reeve, J., Xue, Y., & Xiong, J. (2012). Neural differences between intrinsic reasons for doing versus extrinsic reasons for doing: An fMRI study. Neuroscience Research, 73(1), 68–72.
- Frey, B. S., & Jegen, R. (2001). Motivation crowding theory. Journal of Economic Surveys, 15(5), 589–611.
- Ariely, D., Gneezy, U., Loewenstein, G., & Mazar, N. (2009). Large stakes and big mistakes. Review of Economic Studies, 76(2), 451–469.
- Volpp, K. G., Troxel, A. B., Pauly, M. V., Glick, H. A., Puig, A., Asch, D. A., et al. (2009). A randomized, controlled trial of financial incentives for smoking cessation. New England Journal of Medicine, 360(7), 699–709.
- Kohn, A. (1993). Punished by Rewards: The Trouble with Gold Stars, Incentive Plans, A’s, Praise, and Other Bribes. Houghton Mifflin.
- Hagger, M. S., & Chatzisarantis, N. L. D. (2011). Causality orientations moderate the undermining effect of rewards on intrinsic motivation. Journal of Experimental Social Psychology, 47(2), 485–489.
- Murayama, K., Matsumoto, M., Izuma, K., Sugiura, A., Ryan, R. M., Deci, E. L., & Matsumoto, K. (2015). How self-determined choice facilitates performance: A key role of the ventromedial prefrontal cortex. Cerebral Cortex, 25(5), 1241–1251.
- Fryer, R. G. (2011). Financial incentives and student achievement: Evidence from randomized trials. Quarterly Journal of Economics, 126(4), 1755–1798.
- Moldon, L., Strohmaier, M., & Wachs, J. (2021). How gamification affects software developers: Cautionary evidence from a natural experiment on GitHub. Proceedings of the 43rd International Conference on Software Engineering (ICSE).
- Hanus, M. D., & Fox, J. (2015). Assessing the effects of gamification in the classroom: A longitudinal study on intrinsic motivation, social comparison, satisfaction, effort, and academic performance. Computers & Education, 80, 152–161.
- Mekler, E. D., Brühlmann, F., Tuch, A. N., & Opwis, K. (2017). Towards understanding the effects of individual gamification elements on intrinsic motivation and performance. Computers in Human Behavior, 71, 525–534.
Related Reading
- The Octalysis Complete Gamification Framework — the eight-Core-Drive system the Overjustification Effect is most directly applied through.
- Self-Determination Theory — the broader motivation framework that contains Cognitive Evaluation Theory and the autonomy-competence-relatedness machinery the Overjustification Effect rides on.
- Cognitive Dissonance — the adjacent attribution-and-self-perception literature that interlocks with Bem’s Self-Perception Theory and the Overjustification mechanism.
- Locus of Control — the trait-level perceived-contingency construct that moderates the size of the Overjustification Effect across users.
- Reactance Theory — the parallel motivational-defense construct that fires when freedom is threatened, which is the mechanism by which controlling rewards do their damage.
- Expectancy Theory — the standard pay-for-performance framework the Overjustification Effect is the structural correction on.
- Actionable Gamification — the book-length presentation of the Octalysis Framework, with extended treatment of the White-Hat / Black-Hat and Left-Brain / Right-Brain distinctions the Overjustification Effect supplies the empirical backing for.



