Intrinsic vs extrinsic motivation: what Deci and Ryan actually showed

Intrinsic vs extrinsic motivation comes down to why a person does an activity: because it is inherently interesting or enjoyable, or because it leads to a separable outcome such as a paycheque, a grade or praise. Those are the definitions Richard Ryan and Edward Deci drew in their 2000 review, and most revision notes stop there. The part they drop is what comes after it. Extrinsic motivation is not one thing but four, arranged from bare compliance to a reason a person has genuinely made their own, and a reward only undermines interest under conditions that are now fairly well mapped.

• Intrinsic motivation: the point of the activity is the activity. Extrinsic motivation: the point is something the activity produces.
• The distinction belongs to self-determination theory, the framework Ryan and Deci have built since the 1970s. It sorts reasons for acting, which makes it a different kind of model from Maslow's hierarchy of needs and its levels of need.
• Rewards can reduce interest, but the effect is specific: it is strongest for tangible rewards that are expected in advance and tied to doing the task.
• Extrinsic motivation comes in four forms, external, introjected, identified and integrated regulation, ordered by how far the person endorses the reason for acting.
• Intrinsic and extrinsic motivation are not opposites. In student data the two correlate positively, and a single task can carry both at once.

The difference in one table

Both kinds of motivation can produce identical behaviour, so the useful question is not what someone is doing but why. Each row below is a question that separates the two.

Intrinsic motivation: the activity is the point

Intrinsic motivation is the pull toward exploring and learning that needs no external push. Ryan and Deci call it the prototype of self-determined activity: the person acts because the acting is its own reward. The student who reads past the syllabus, and the cook who tries a new recipe on a free evening, are both working from it.

Self-determination theory holds that this kind of interest depends on three psychological needs. Autonomy is acting by choice rather than under pressure. Competence is getting better at something and having that registered. Relatedness is doing it among people who matter. When a setting supports all three, interest tends to hold up; when it frustrates them, interest tends to fall.

Extrinsic motivation: the outcome is the point

Extrinsic motivation is doing something for what it leads to. A bonus for hitting a sales target, a grade for finishing a module and buckling a seatbelt to silence a chime are all extrinsically motivated: something separable from the act is doing the work. The early laboratory studies contrasted this with intrinsic interest, and the form they studied, working for a reward or to avoid a punishment, is what the theory calls external regulation.

Treating all extrinsic motivation as identical is where the simple version of the story goes wrong. The theory has always treated it as a family, and the members differ in how much they feel like the person's own choice.

The overjustification effect: when a reward backfires

The overjustification effect is the finding that rewarding an activity someone already enjoys can reduce their interest in it once the reward stops. It comes from a group of studies run in the early 1970s, and the two most quoted used very different tasks.

In Deci's 1971 puzzle experiment, undergraduate students were paid a dollar for each three-dimensional puzzle they solved. The experimenter then left the room and the money stopped. The students who had been paid put the puzzles aside; the students who had never been paid carried on working on them. The payment had not added to their interest. It had displaced it.

The second study is the marker-pen experiment. Lepper, Greene and Nisbett chose children who already spent a lot of free time drawing with felt-tip markers, then had them draw under one of three conditions: a reward was shown in advance and could be won by drawing, the same reward arrived as a surprise afterwards, or the children got only verbal feedback and no reward. A week or two later the markers were back on the table with nothing on offer. Only the children who had expected the reward had changed. They spent about half as much free time drawing, and their drawings were rated lower in quality. Expecting the reward, not receiving it, was what did the damage.

The effect is narrower than the slogan

A 1999 meta-analysis by Deci, Koestner and Ryan pooled 128 laboratory experiments. With the reward types mixed together the effect looked modest, which is why other reviewers doubted it. Split apart, the picture was sharper: tangible rewards that were expected and tied to doing the task reliably reduced later free-choice motivation, while verbal rewards such as praise tended to raise it. An earlier 1994 meta-analysis had reached a more sceptical verdict by pooling reward types together. The practical point that survives both is that "rewards" is too broad a category to argue about as one thing.

Two limits are worth keeping. Most of the evidence comes from short-term laboratory tasks, so how far the effect carries into real-world incentives such as salaries and scholarships is still contested. And the researchers describe their classroom prescriptions as recommendations drawn from the theory, not as experimental results in their own right.

The self-determination continuum: four kinds of extrinsic motivation

The idea that holds the two halves together is internalisation: taking in a reason for acting until it feels like your own. Ryan and Deci arrange extrinsic motivation along that process, from reasons a person merely complies with to reasons they fully endorse.

The last row is the one people miss. Integrated regulation is the most autonomous form of extrinsic motivation, and it shares a lot with intrinsic motivation: it is volitional and unconflicted. It is still extrinsic, because the behaviour is done for its instrumental value, for an outcome separate from the act. In Ryan and Deci's own terms, the more a person internalises the reasons for an action, the more their extrinsically motivated actions become self-determined, but that does not turn them into intrinsic motivation.

Two corrections follow. First, the continuum is not a ladder climbed over a lifetime: a person can take up a new behaviour at any point on it, and the theory does not claim that everyone passes through each stage. Second, extrinsic is not a synonym for controlled. Identified and integrated regulation are extrinsic and largely autonomous at the same time.

The phrase for how the difference feels from the inside is perceived locus of causality, from the researcher Richard deCharms: behaviour feels internally caused when it matches your own values, and externally caused when it is done under pressure. That is a neighbouring idea to the locus of control Julian Rotter measured, but a different one. Rotter's construct is an expectancy about whether outcomes follow from what you do; perceived locus of causality is about why you are acting.

When rewards help and when they hurt

Put the studies together and the guidance is conditional rather than anti-reward. A reward is most likely to damage interest when it is tangible, expected in advance and tied to doing the task. When it arrives as a surprise, or as verbal feedback, it usually does not.

• Expected versus unexpected: a reward promised before the task starts is the kind that undermines interest. One that arrives afterwards as a surprise generally leaves interest intact.
• Tangible versus verbal: money and prizes carry more risk than praise. Deci's first experiment also tested verbal rewards, and those raised later interest rather than lowering it.
• Controlling versus informational: the same event can be read as pressure, do this and you get paid, or as information, that was well done. The controlling reading threatens autonomy; the informational one supports competence.
• Performance contingency: a reward tied to how well someone did can carry information about their ability, which is why performance-contingent rewards do not always behave like completion-contingent ones.
• Initial interest: where there is little interest to erode, a reward can usefully start engagement. It is rewarding tasks people already enjoy that tends to backfire.

A 2022 review of the literature reached a compatible conclusion: extrinsic reinforcers reduce intrinsic motivation mainly when their delivery is contingent on task performance, and the two motivations behave as partially distinct, interacting processes rather than as simple opposites.

Economists Roland Bénabou and Jean Tirole offered a model that fits the same facts. A reward is also a signal, because it tells the person something about the task and about what the person offering it thinks of them. Paying a child to read can suggest that reading is the sort of thing one would not do for free, or that the adult assumed an incentive was needed. That is why they describe incentives as weak reinforcers in the short run and negative reinforcers in the long run.

What this looks like in the classroom and at work

The advice that follows from the theory is about proportion, not prohibition. Where a task is already interesting, the thing to lighten is evaluation and the pressure around it, and the thing to protect is choice. Where a task holds little interest, an incentive can start engagement that interest may later take over.

• Use informational feedback, such as specific praise about what was done well, for tasks people enjoy. Save tangible rewards for tasks they do not.
• Give a reason. Identified and integrated regulation grow when a person can see why a task matters to something they already value.
• Keep some choice over how the work is done, because autonomy is one of the three needs the theory rests on.
• Do not expect a reward to be neutral just because it is generous. If it reads as control, its size does not settle the question.

The flip side of a controlled task is one where nothing the person does changes the outcome. That is the condition behind learned helplessness: when responses stop controlling what happens, effort falls away. Autonomy and contingency are two ways of saying the same thing, that acting matters.

Learning the vocabulary

The terminology here is a small set of near-synonyms: intrinsic and extrinsic, the four regulation types, overjustification, autonomy and competence. Terms this close together interfere with each other when you reread them in a list rather than test yourself on them, which is why our guide to memorising psychology terminology treats them as a set. In Psychology World you can save terms to a named list and drill that list with flashcards, multiple-choice questions once it has four or more words, or typing, which is the mode that really tests recall.