Rewards, and when they undermine the thing

The idea that paying someone for something they enjoyed makes them enjoy it less is widely repeated. The research is more specific and more interesting than the summary.

The original finding

Experiments in the 1970s found that people given an external reward for an activity they already found interesting subsequently spent less time on it when the reward was removed, compared with people who were never rewarded.

The effect was named the overjustification effect, and the proposed mechanism is that the reward provides an alternative explanation for the behaviour, displacing the internal one.

The conditions that turned out to matter

Subsequent meta-analytic work established that the effect is real and conditional, and the conditions do most of the work.

The activity must already be intrinsically interesting. For dull tasks nobody would do voluntarily, rewards do not undermine motivation that was not there.

Which covers a great deal of paid work, and is why the finding is much less applicable to employment than it is usually made to sound.

The reward must be expected and contingent on the activity itself rather than on performance quality.

Unexpected rewards given afterward show little or no undermining effect. Rewards contingent on doing well convey information about competence, which can increase motivation.

The reward must be perceived as controlling. This is the crux.

A reward experienced as an attempt to control behaviour undermines autonomy, which self-determination theory identifies as a basic need. A reward experienced as recognition of competence does not.

The same payment can be either, depending on how it is framed and how it is delivered.

Verbal praise generally does not undermine and frequently enhances, provided it is specific and not experienced as controlling.

What this means practically

For something you already enjoy, be cautious about monetising or gamifying it. Turning a hobby into a source of income changes the relationship, and people report this consistently.

Not that it is always wrong — plenty of people build good work out of an interest — but that the change is real and should be expected rather than discovered.

For something you do not enjoy, external structure is a reasonable tool and the undermining concern largely does not apply.

A commitment device, a deposit at risk, an accountability partner: these are external and there was no intrinsic motivation to protect.

For rewarding other people, the practical guidance is to reward outcomes rather than presence, to make rewards informational rather than controlling, and to be aware that a reward introduced becomes very difficult to remove.

That last point is underrated. A bonus given twice becomes an expectation, and its withdrawal is experienced as a penalty.

The measurement problem

The larger risk with incentives, and it is not about motivation at all.

Any measure attached to a reward becomes a target, and behaviour reorganises around it — including in ways nobody intended.

Sales targets producing pressure selling. Response-time targets producing calls terminated early. Publication counts producing thin papers.

The general principle is that people optimise what is measured and rewarded, which is a feature until the measure diverges from the thing you wanted.

Which means the design question is less about whether to reward and more about whether the measure captures the actual objective, and it usually does not capture all of it.

What self-determination theory suggests instead

Where sustained motivation is the goal, the theory points to three conditions.

Autonomy — a sense of choosing rather than being directed.

Competence — a sense of being effective and improving.

Relatedness — connection to others through the activity.

These are supported by a substantial body of work, and in workplace terms they translate into control over how work is done, visible progress, and colleagues who matter.

All three are harder to arrange than a bonus, and all three predict sustained engagement better.

The honest summary

Rewards work well for behaviours nobody wants to do, badly for behaviours people already value, and unpredictably wherever the measure is a poor proxy for the goal.

Which is a more useful conclusion than either that incentives are corrosive or that they solve motivation.