Learn it → Apply it → Save as Skill
Paying someone to do what they already loved doing can make them love it less. Adding a strong, expected, task-tied reward on top of an activity someone already found interesting can shift the reason they do it from "I want to" toward "I get something for it," and interest can drop once the reward stops.
The Dr. Wallace’s Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
When someone already enjoys a task, adding a strong, expected, task-tied reward can crowd out their own reason for doing it. The activity gets reframed as "paid work" rather than something chosen freely, and once the reward disappears, some of the original enthusiasm can disappear with it.
Important: this chart is illustrative. The classic studies reported a statistically significant drop, not a specific percentage you should expect to reproduce.
The model is most useful before you design pay, prizes, grades, points or praise for an activity that already has a willing, interested performer.
The goal is not to avoid all rewards. It is to notice when a reward is landing on top of motivation that was already present.
Was this person already doing the activity, or enjoying it, before any reward was on the table?
Is the reward tangible, expected in advance, and tied tightly to completing the task? That combination carries the most risk.
Where possible, favor unexpected recognition, informational feedback about competence, or rewards for trying something new rather than for continuing something already loved.
Watch voluntary engagement after the reward shrinks or ends. A drop below the original baseline is the signal this model is describing.
Open a use case to see what the model changes.
A standing cash bonus tied to a specific recurring task can turn "I do this because I'm good at it and like it" into "I do this because I'm paid to." Consider whether recognition, autonomy, or a one-time unexpected thank-you preserves more of the original drive than a routine, expected bonus.
If a child already reads for pleasure, a per-book cash reward can shift their reason for reading. If the child does not yet read voluntarily, a reward may be a reasonable way to build the habit; the risk is highest when the interest already exists.
Introducing even a small stipend can reframe "I volunteer because it matters to me" as "I show up because I'm paid." If funding for the stipend later disappears, participation can fall below the original volunteer rate, not just back to it.
Turning a hobby into paid work, with deadlines, client briefs, and per-piece rates, can change how the work feels even when nothing else about the craft has changed. Separating a protected, unpaid creative practice from paid client work is one common workaround.
Gamified rewards can boost short-term adherence, especially for people who were not already exercising. For someone who already enjoyed the activity, heavy point systems can sometimes make the workout feel like a chore to be completed for score.
Bounty programs can attract new contributors, but can also change how existing volunteer maintainers feel about the same work once a price tag is attached to it. Consider whether the bounty targets new contributions rather than replacing goodwill-based ones.
This is an illustrative example designed to teach the method.
Expand each item for the research behind the model and what it does not claim.
Edward Deci had participants work on Soma spatial puzzles across three sessions. In the middle session, one group was paid a dollar for each puzzle solved while a control group was not. In a "free-choice" period, where participants were told they could keep solving puzzles, read magazines, or do nothing while the researcher was supposedly busy, the previously-paid group spent much less time voluntarily working on the puzzles in the final free-choice period than they had in the first one, while the unpaid control group's interest declined far less. Deci used this free-choice persistence as his behavioral measure of intrinsic motivation.
Mark Lepper, David Greene, and Richard Nisbett extended the question to children in "Undermining Children's Intrinsic Interest with Extrinsic Reward: A Test of the 'Overjustification' Hypothesis," published in the Journal of Personality and Social Psychology in 1973 (volume 28, pages 129 to 137). Preschoolers who already liked drawing with felt-tip markers were assigned to one of three conditions: told in advance they would receive a "Good Player Award" for drawing, given the same award afterward with no advance promise, or given no award at all. In follow-up free-play observation roughly one to two weeks later, the children who had been promised the award in advance spent significantly less time drawing with the markers than children in either the unexpected-award or no-award groups, whose interest held up.
Two theories offer complementary mechanisms. Daryl Bem's self-perception theory suggests people partly infer their own motives by observing their own behavior and its surrounding context; when a salient, expected reward is present, a person may reinterpret "I do this because I enjoy it" as "I do this for the reward." Edward Deci and Richard Ryan's cognitive evaluation theory, part of their broader self-determination theory, instead emphasizes a perceived locus of causality: a controlling reward can make the activity feel externally imposed rather than freely chosen, which undermines the sense of autonomy that intrinsic motivation depends on. The two accounts are not mutually exclusive and are often discussed together.
It does not claim that all rewards reduce motivation, that praise is dangerous, or that paying people for their work is a mistake. The effect is most reliably observed for tangible, expected, task-contingent rewards layered onto a task someone already found intrinsically interesting, measured by later voluntary engagement. Unexpected rewards, rewards framed as informational feedback about competence rather than control, and rewards for tasks with little pre-existing interest behave differently. A 1996 meta-analysis by Robert Eisenberger and Judy Cameron also questioned how broadly the undermining effect generalizes outside tightly controlled lab conditions, which is why this model is best used as a design caution rather than an absolute rule.
Most workplace tasks were never purely intrinsically motivating to begin with. There is little to "overjustify" in a task nobody wanted to do for its own sake, so ordinary pay and incentives for that work are not at risk the same way.
A reward framed as controlling ("do this or you don't get it") carries more risk than one framed as informational ("you did this well"). The framing often matters more than the dollar amount.
Overcorrecting into "never reward anything someone enjoys" can remove useful encouragement and competence feedback that people genuinely want.
Rewards frequently raise output while they are active. The model is about what happens to voluntary engagement after the reward shrinks or stops, not about whether the reward "worked" in the moment.
Without knowing how someone behaved before any reward existed, a later drop in interest can be misread. The clean comparison is always against a credible baseline or an unrewarded control, not against intuition.
If the activity was never intrinsically motivating for this particular person, a reward is simply doing its ordinary job of motivating behavior. The effect is specific to rewards layered on top of existing interest.
the task has no meaningful pre-existing intrinsic interest and simply needs to get done, such as routine compliance work or unpleasant chores.
fairness, legal compliance, or consistency requires an explicit, predictable compensation structure, such as overtime pay or a sales commission that is a core part of the job.
someone's drop in motivation looks like burnout, disengagement, or a wellbeing issue rather than a response to a specific reward. That calls for a conversation and, where appropriate, professional support, not a reward redesign.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the Overjustification Effect, what questions to ask, what evidence to seek, what mistakes to avoid, and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
No. It means an expected, tangible, task-tied reward is the riskiest combination when real intrinsic interest already exists. Unexpected recognition, informational praise about competence, and rewards for brand-new effort generally do not carry the same risk, and in some cases support motivation further.
If a reward was not promised or anticipated, it could not have been the reason someone engaged in the activity in the first place. Self-perception theory suggests people only reattribute their motive to a reward when that reward was a known, salient part of their decision to act before they acted.
Goodhart's Law is about a measure stopping being a good measure once it becomes a target; people optimize the number instead of the underlying goal. The Overjustification Effect is about motivation itself changing character, from internally driven to externally driven, once a reward is attached. A single badly designed bonus program can trigger both at once.