Paid for What You Already Loved: Self-Determination Theory and the Crowding Out of Motivation

The Hook: The Break Nobody Was Watching

Picture a plain laboratory room in the late 1960s. On the table lie SOMA puzzles — seven angular wooden pieces to be assembled into three-dimensional figures shown on cards. It is one of those tasks people find interesting from the inside: you try, you fail, you turn a piece over, it fits, and you want the next one.

An experimenter sets the task, lets the person work, and then leaves the room for eight minutes on some pretext. The puzzles stay on the table. So do magazines — Time, The New Yorker, Playboy. The participant is alone, nobody expects anything of them, they may do whatever they like.

They are being watched. Through a one-way mirror, a stopwatch records how many of those 480 seconds they voluntarily spend on the puzzles. That number is the entire measurement. In the literature it is called the free-choice measure, and it is an attempt to operationalize something that questionnaires cannot capture, because people are polite when asked: the willingness to keep doing something with no external reason whatsoever.

Edward Deci added one element. One group was paid a dollar for every puzzle solved during the middle of three sessions. In the third session they were told the budget was exhausted and there would be no more money. Then came the break with the one-way mirror again.

The group that had been paid in between spent noticeably less time on the puzzles in that final free period than before. The control group, which had never seen money, declined less.

That, in all its plainness, is the finding that set off one of the longest and most bitter disputes in motivation psychology. Money for an activity you already enjoy can damage the enjoyment. Not raise it, not leave it untouched — damage it.

Five decades later the picture is more complicated and more interesting than either the popular version ("bonuses destroy motivation") or the economic counterposition ("more incentive, more performance") claims. Deci's original experiment was, as his own intellectual heirs now write openly, statistically underpowered, and its main effect only marginally significant. A 1994 meta-analysis declared the effect negligible. A 1999 meta-analysis declared the 1994 meta-analysis methodologically flawed and found the effect at a respectable size. Then a 2014 meta-analysis covering more than 200,000 people showed that the question had been posed wrongly all along — and that the answer depends on whether you mean the quantity or the quality of a performance.

This article walks the road from that laboratory room with the one-way mirror to the question of how you build a bonus scheme, a set of OKRs, or a review process that does not destroy precisely what it is meant to promote.

Part 1: What 1971 Actually Showed — and What It Did Not

The Three Experiments

Deci's 1971 paper (Journal of Personality and Social Psychology, 18(1), 105–115) consists of three investigations, and they must be kept carefully apart, because they carry different evidential weight.

Experiment 1 — the laboratory. Twenty-four people in two groups, three sessions at least a day apart. Session 1 was a baseline with no mention of reward, with the eight-minute free period as the initial measurement. In Session 2, the experimental group received one dollar per puzzle solved; the control group received nothing. In Session 3 nobody was paid, on the grounds that funding had covered only one session. Dependent variable: seconds of voluntary puzzle engagement, Session 1 versus Session 3.

The result pointed in the expected direction — the paid group dropped — but it was only marginally significant. Twelve people per group is, by today's standards, hopelessly underpowered for a medium effect. It is notable, and speaks well of the field, that this assessment does not come from critics but from a retrospective by Richard Ryan and colleagues, that is, from the center of the same research tradition.

Experiment 2 — the field. This is the underrated one. Deci observed two four-person teams at a student newspaper who wrote headlines and worked non-overlapping shifts. Four weeks of normal operation as baseline. Then one team was paid 50 cents per headline for three weeks while the other was paid nothing. Payment was then stopped and both teams observed for three more weeks, with a follow-up in the first two weeks of the next semester. Time per headline was recorded covertly — as was the number of absences.

The unpaid team got steadily faster over the weeks, as one would expect for a practiced task. The previously paid team showed almost no further efficiency gains once payment ended, and a trend toward more absence. This is not a laboratory break but real work over months — and the effect appears not as a collapse but as development that fails to happen. That pattern should look familiar to anyone who has watched teams over longer stretches of time.

Experiment 3 — the praise. Almost identical to Experiment 1, but with verbal rewards instead of money: feedback such as "You did very well in completing the task; many participants did not complete it." Result: verbal reward did not undermine intrinsic motivation. The control group declined across sessions; the praised group held its level.

This third experiment is the most important and most frequently overlooked part of the paper. It shows that the issue is not "reward" as a category. Money and praise are both rewards, and they work in opposite directions. Anyone who reads Deci's paper as "rewards are harmful" has skipped a third of it.

The Quality of the Evidence

I think it matters not to skip this point, because it is what puts the rest of this article on honest footing: Deci's 1971 work is a beginning, not a demonstration. Three small studies, one of them marginally significant, one with eight people in total, one whose sample size is not even reported and which may have confounded subgroups of arts and technical students. If this were the only finding, the correct response would be: interesting hypothesis, please replicate.

Which is exactly what happened over five decades, across more than a hundred experiments. That is why the story is worth telling — not because the original study was strong, but because it was a weak study whose finding held. This is the inverse of textbook classics whose original study looked strong and whose finding did not hold; how that difference arises is traced in The Peak of Not Knowing: Why Psychology's Most Famous Effect May Be Just a Statistical Mirage through the opposite case.

Part 2: The Drawing Study — How the Effect Got Its Name

The Design

Two years after Deci came the study that lodged the effect in collective memory: Mark R. Lepper, David Greene, and Richard E. Nisbett, Undermining children's intrinsic interest with extrinsic reward: A test of the "overjustification" hypothesis (Journal of Personality and Social Psychology, 28(1), 129–137).

The setup is exemplary in its clarity. At the Bing Nursery School on the Stanford University campus, preschoolers aged 40 to 64 months were first observed unobtrusively in their usual environment: how much of their free play time did they spend, of their own accord, with felt-tip pens and paper? That value is the baseline, and it is decisive — what is measured is not drawing as such, but the change in voluntary drawing.

Children with identifiable initial interest were then taken individually to another room and assigned to three conditions (55 children in the sessions, 51 in the analysis):

• Expected award: the children agreed to draw in order to earn an award.
• Unexpected award: the children drew without knowing about an award and received one afterward.
• No award: the children drew; there was no award, neither announced nor given.

The award was the "Good Player Award": a colored 3 × 5-inch card with spaces for the child's name and school, a large gold star, and a red ribbon. No money, no toy — a piece of cardboard with a star.

Seven to fourteen days later, classroom observation resumed, across no more than four consecutive school days.

The Result

Share of free play time spent with the drawing materials:

The group that had expected an award drew about half as much in free play as the other two groups. The planned contrast of expected award against the remaining conditions was significant, F(1, 48) = 6.19, p < .025; the individual comparisons likewise (expected vs. unexpected: t = 2.32; expected vs. no award: t = 2.05; p < .025 in each case). Within groups, only the expected-award group declined significantly against its own baseline (t = 2.61, p < .02); the other two did not change meaningfully. Because the variances differed substantially (Fmax = 5.39, p < .01), the analysis was run on log-transformed data — a detail showing this was not naive averaging.

Almost in passing, the paper reports a second finding that I find regularly missing from popular accounts: the quality of the drawings produced during the session was lower in the expected-award condition (mean 2.18) than with an unexpected award (2.85; t = 3.01, p < .01) or no award (2.69; t = 2.08, p < .05). Children who expected an award for drawing not only drew less voluntarily later — they drew worse at the time. That side finding anticipates what would become the headline result in meta-analysis forty years on.

Why "Overjustification"

The name of the hypothesis comes from attribution theory and is more precise than it sounds. The idea: people — four-year-olds included — explain their own behavior by reaching for the most plausible available cause. Someone who draws because they want to draw attributes the interest to themselves. Someone who draws and receives a certificate with a gold star suddenly has two reasons, and the external one is the more salient. The internal reason is devalued, because it is no longer needed to explain the behavior. When the certificate later disappears, the reason that had become load-bearing disappears with it.

This is a cognitive explanation, not a behaviorist one, and that is exactly where the dynamite lay. In the learning theory of the time, reinforcement was by definition something that increases behavior. Here was a finding in which reinforcement lowered behavior durably. That the explanation runs through self-attribution rather than through reinforcement itself ties the finding closely to The Price of Justification: Cognitive Dissonance from Festinger to Neuroscience, which is likewise about people adjusting their attitudes to the explanations available for their own behavior.

Part 3: The Theory — Every Reward Has Two Faces

Cognitive Evaluation Theory

The individual findings grew into a theory called Cognitive Evaluation Theory (CET), today treated as a sub-theory of the broader Self-Determination Theory (SDT). At its core is a claim about needs: intrinsic motivation rests on the basic psychological needs for autonomy (experiencing oneself as the originator of one's own action) and competence (experiencing oneself as effective). Events that satisfy these needs tend to increase intrinsic motivation; events that thwart them tend to decrease it.

The decisive theoretical step is the distinction between two aspects that every reward carries simultaneously:

• The controlling aspect: a reward can be experienced as steering one's behavior. Autonomy is thwarted, and intrinsic motivation falls.
• The informational aspect: a reward can signal competence — it says you did something well. The competence need is satisfied, and intrinsic motivation can rise.

Every reward has both aspects. Which one dominates depends on the design and — the theory stresses this explicitly — on the interpersonal style in which it is administered. The same bonus can be recognition or training-to-command, depending on how it is announced, justified, and handed over.

This resolves the apparent contradiction within Deci's own data. Money per puzzle solved is predominantly controlling: it turns the puzzles into a means. Praise for a good solution is predominantly informational: it says something about one's ability and steers nothing. Both are rewards; they hit different needs.

The Three Needs and the Move to SDT

The broader Self-Determination Theory adds a third basic need, relatedness, and replaces the binary of "intrinsic versus extrinsic" with a continuum of regulatory quality: from external regulation (I do it because someone demands it), through introjected regulation (I do it because otherwise I feel bad) and identified regulation (I do it because I consider it important), to intrinsic motivation (I do it because the activity itself carries me).

This refinement has practical consequences. Most of the work we do is not intrinsically motivated and never will be. Documenting an access matrix is, for almost nobody, an activity they would seek out during a free period behind a one-way mirror. So the relevant question is rarely "intrinsic or not," but: is the activity identified, because the person understands and accepts what it is for — or external, because otherwise someone complains? In practice the gap between those two is wider than the gap between identified and intrinsic regulation.

It is the same structure described at the team level in Psychological Safety: The Hidden Force Behind High-Performing Teams: not satisfaction as a goal, but a condition under which certain behaviors become possible at all.

Part 4: The Dispute — How Two Meta-Analyses Reached Opposite Conclusions

1994: The Effect Is Negligible

In 1994, Judy Cameron and W. David Pierce published a meta-analysis in the Review of Educational Research titled Reinforcement, Reward, and Intrinsic Motivation: A Meta-Analysis (64(3), 363–423). Their verdict: the undermining effect is minimal, empirically barely supportable, and of no significance for educational practice. They went further and recommended abandoning Cognitive Evaluation Theory.

This was no marginal remark. The Review of Educational Research is a high-ranking review journal, and the message had traction in education policy: reward schemes in schools are harmless.

1999: The Counter-Meta-Analysis

Five years later, Edward Deci, Richard Koestner, and Richard Ryan answered: A Meta-Analytic Review of Experiments Examining the Effects of Extrinsic Rewards on Intrinsic Motivation (Psychological Bulletin, 125(6), 627–668). It included 128 studies from 94 published articles and 19 unpublished dissertations — the latter deliberately, to limit publication bias. 101 studies used a free-choice behavioral measure; 84 used a self-report measure of interest and enjoyment.

The central effect sizes (Cohen's d; negative values mean undermining):

Three things about this table deserve attention.

First, it confirms the theoretical prediction in detail, not merely in sign. Tangible, expected rewards undermine. Verbal feedback enhances — and at a similar magnitude, just with the opposite sign. Read the "all rewards" row as the headline finding and you are averaging two opposing effects into a muted residue. Precisely this conflation was the chief methodological charge against the 1994 analysis.

Second, the gradation across contingencies is instructive. Engagement-contingent reward ("you get money for doing this") undermines most strongly (−0.40), performance-contingent reward least (−0.28). This fits the theory: only performance-contingent reward contains any information about competence, so it carries an informational component that counteracts the controlling one. It still undermines.

Third — and this is rarely discussed — the behavioral measure and the self-report diverge substantially. On self-reported interest, the effects of tangible rewards run from −0.07 to −0.17, that is, close to zero. The undermining shows up far more strongly in behavior, when nobody is watching, than in what people say about it. I am of the opinion that this is the single most practically important number in the whole table: an incentive scheme that damages intrinsic motivation will, in all likelihood, not show up in an employee survey.

The analysis also found an age effect: tangible rewards tended to be more detrimental for children than for college students, and verbal rewards less enhancing for children — the free-choice effect was d = 0.11 (not significant) for children versus d = 0.43 for college students.

2001: The Reply

Deci, Koestner, and Ryan summarized the dispute in 2001 in the very journal in which Cameron and Pierce had published: Extrinsic Rewards and Intrinsic Motivation in Education: Reconsidered Once Again (Review of Educational Research, 71(1), 1–27). Their argument has three strands: the 1994 analysis was seriously flawed methodologically; tangible rewards do have a substantial undermining effect; and the evidence supports the very Cognitive Evaluation Theory whose abandonment had been recommended.

What one can learn from this dispute holds regardless of whose conclusion one eventually accepts: two meta-analyses over overlapping literature can reach opposite conclusions if they aggregate differently. A meta-analysis is not a neutral tally. Every decision about which studies belong in which bucket is a substantive decision — and it co-determines the result. Throw tangible and verbal rewards into one bucket and you get zero; separate them and you get two opposing effects of respectable size each. The number is correctly computed either way.

Part 5: The Turn — The Question Had Been Posed Wrongly

183 Samples, 212,468 People

In 2014 came the paper I regard as the real advance: Christopher P. Cerasoli, Jessica M. Nicklin, and Michael T. Ford, Intrinsic Motivation and Extrinsic Incentives Jointly Predict Performance: A 40-Year Meta-Analysis (Psychological Bulletin, 140(4), 980–1008). 183 independent samples, N = 212,468, from 154 sources (28 of them unpublished), spanning school, work, and physical performance.

The authors' starting point is a provocation: the debate over the undermining effect had become a distraction. It had spent forty years settling whether incentives lower intrinsic motivation, and in doing so had neglected the more obvious question — how strongly intrinsic motivation actually predicts performance, and in what relationship it stands to incentives when it does.

The overall finding: intrinsic motivation predicts performance with a corrected population correlation of \(\rho\) = .26 (80% credibility interval .16 to .36). Across a base that large and heterogeneous, that is a substantial relationship.

The Decisive Split: Quality Versus Quantity

Then comes the distinction that changes everything. The authors separate quality from quantity performance:

It becomes starker still in the meta-analytic regression, where intrinsic motivation and incentives enter together as predictors:

Read the last row twice. For quality performance, intrinsic motivation carries \(\beta\) = .35 — and incentives carry \(\beta\) = .06, which is to say practically nothing. For quantity performance the ratio reverses: incentives .33, intrinsic motivation .24.

This is the cleanest empirical statement of an intuition everyone in practice already has: you can pay people to do more. You can only very partially pay them to do it better. Count tickets and you get tickets. Want code quality and you need people who want code quality.

The Crowding-Out Pattern

The third major finding concerns how incentives and intrinsic motivation affect each other. The authors distinguish whether incentives are tied directly or indirectly to performance:

Where incentives are directly and visibly bound to performance, intrinsic motivation becomes a weaker predictor. The authors explicitly call this a "crowding out." This is not the same finding as the undermining effect (there, motivation itself falls; here, its predictive power falls), but it is the same phenomenon viewed from another side: where the external incentive steers the behavior, the internal drive contributes less to the result.

Note the sample count: k = 8 samples for the indirect coupling is thin. I am of the opinion that this particular comparison should be read as a strong hint, not as a settled finding — it is the most weakly supported number in this article.

The Needs at Work

Worth adding is Anja Van den Broeck, D. Lance Ferris, Chu-Hsiang Chang, and Christopher C. Rosen, A Review of Self-Determination Theory's Basic Psychological Needs at Work (Journal of Management, 42(5), 1195–1229), a meta-analysis across 99 studies with 119 samples. Corrected correlations, excerpt:

Two things stand out. Competence is the strongest predictor of pure task performance (.40) — people who experience themselves as effective produce more. Autonomy dominates wherever experience and staying are at stake: job satisfaction .69, burnout −.60, turnover intention −.38. Competence alone keeps nobody in the company (−.07).

The authors are fair toward their own theory: the three needs explained little variance in external and introjected motivation, which they explicitly flag as a challenge to SDT. Their overall verdict is that the needs "mostly fit the criteria" for basic psychological needs — with qualifications.

Part 6: What the Evidence Does Not Say

Six limitations worth holding in mind before turning this material into a decision.

1. It is about activities with initial interest. The undermining effect presupposes that there is intrinsic motivation available to be undermined. Lepper and colleagues deliberately selected children with identifiable interest in drawing. For an activity nobody does voluntarily, there is nothing to crowd out — here incentives are not harmful but the only available mechanism.

2. The effect sizes are medium, not dramatic. d = −0.34 for tangible rewards is a medium effect. It does not mean payment "destroys" motivation. Popular reception — above all in the management literature — has routinely turned medium effects into catastrophe scenarios.

3. Correlation remains correlation. Cerasoli et al. and Van den Broeck et al. are meta-analyses of predominantly correlational studies. That intrinsic motivation relates to quality performance at \(\rho\) = .35 leaves open how much of that is causal and in which direction. The reverse path is equally plausible: someone who delivers good work experiences competence and develops more interest as a result.

4. The behavioral measure is contestable. Free choice during a laboratory break is a short, artificial situation. Reading "less time with the puzzle = less intrinsic motivation" is natural, but not compelling; a person might simply conclude that the task is finished.

5. Timing is not yet settled. A more recent experimental series by Yuxia Liu and colleagues (Frontiers in Psychology, 13, 853879) reports that immediate external rewards tend to support intrinsic motivation rather than depress it, compared with delayed ones. These are four small experiments with 347 analyzed participants in total, with individual effects just reaching significance (for example t(56) = 2.018, p = .048, d = 0.54) and no pooled effect size. I am of the opinion that this is a serious hint at an underestimated moderator, but not yet a basis for decisions.

6. The age effects are inconvenient. Tangible rewards harm children more; verbal praise helps them less (d = 0.11, not significant, versus d = 0.43 for college students). Transferring findings from student samples to schools — or the reverse — transfers too much.

Part 7: Transfer — Incentives in Software and Organizations

The evidence can be condensed into a decision grid you can hold against your own systems.

Four transfers strike me as particularly robust:

Metrics-as-incentives couple quantity, not quality. The regression row "quality: incentives \(\beta\) = .06" is the empirical basis for something known as folklore in software development: what you measure and reward, you get — in quantity. Story points, closed tickets, deployment counts, test coverage in percent. Every one of those can be raised without the product getting better. That averaged metrics can systematically hide what matters is described for the technical case in The Long Tail of Latency: Tail Latency and Why Averages Lie in the Cloud; here there is the added twist that the metric does not merely misrepresent the behavior, it changes it.

Code review is the place with the best ratio. Reasoned, competence-related feedback is the only form of reward in the table with a robust positive effect on both measures (+0.33 / +0.31). A review that explains why a solution is good is therefore not niceness overhead but the most effective intervention available — and it costs nothing but attention. That the same feedback only lands if the environment permits it is the connection to Psychological Safety: The Hidden Force Behind High-Performing Teams.

Autonomy costs are the hidden price of process. Autonomy correlates with burnout at −.60 and with turnover intention at −.38. Every additional approval step, every requirement without a rationale, every tool that is mandated rather than chosen pays into that account. This is not an argument against process — compliance requirements are not negotiable. It is an argument for justifying process, because that is precisely the difference between external and identified regulation. The theory explicitly stresses that external constraints such as deadlines and budgets need not undermine autonomy if they are introduced in an autonomy-supportive way.

Voluntariness is a resource you can spend. Open-source contributions, internal tooling, documentation someone writes on the side, knowledge sharing — these are the classic cases of an activity with initial interest. Turn them into a target system ("everyone contributes two internal tools per quarter") and you have constructed exactly the configuration of expected, engagement-contingent reward with d \(\approx\) −0.40. That the initial interest returns afterward is not guaranteed.

The Central Takeaway

The practical lesson: incentives reliably steer the quantity and barely the quality of work — if you want quality, you must not overwrite the interest that is already there, but feed it autonomy and competence feedback.

Three concrete steps follow directly from the evidence:

1. For every incentive scheme, ask: quantity or quality? If the goal is quantity — throughput, coverage, unit count — incentives are the right tool (\(\beta\) = .33). If the goal is quality — architectural decisions, security analysis, user experience — they are nearly ineffective (\(\beta\) = .06) and risk crowding out what actually carries the result. On this reading, an incentive on quality is not a weak lever but the wrong one.
2. Separate feedback from reward. Verbal, specific, competence-related feedback is the only form with a robust positive effect. It loses that character the moment it is announced and made conditional ("whoever manages this gets named at the all-hands") — at that point information becomes control. Praise after the fact, concretely reasoned, unannounced.
3. Do not trust surveys to detect the damage. The gap between d \(\approx\) −0.34 in behavior and d \(\approx\) −0.07 in self-report is the most important methodological warning this field produces. To find out whether an incentive scheme is crowding out motivation, you must observe behavior in the absence of incentive: what happens in the time nobody bills? Who contributes to things that earn no points? No questionnaire answers those questions.

A Question to Reflect On

The finding with the longest reach is not the undermining effect but its asymmetry: incentives act on what is countable and are blind to what counts. At the same time, everything we can steer in an organization is countable — otherwise we could not steer it.

From that follows an uncomfortable question: In which parts of your work does quality hold up not because a system secures it, but because individual people produce it of their own accord — and what in your target system, your bonus model, or your reporting is in the process of replacing that drive with a metric?

Cross-References in the Vault

• The Marshmallow Promise: How a Candy Test Redefined Willpower for Half a Century – and Was Then Put to the Test Itself – the same preschool campus, the same decade, the adjacent question: how reward and self-regulation relate, and how both findings were refined by later scrutiny.
• The Price of Justification: Cognitive Dissonance from Festinger to Neuroscience – the theoretical neighbor: both effects rest on people adjusting their attitudes to the explanation available for their own behavior.
• Desirable Difficulties: Why Effortless Learning Deceives and Memory Lives on Effort – the counterpart in learning: there, effort raises the yield; here, a badly placed reward destroys the willingness to expend it.
• Psychological Safety: The Hidden Force Behind High-Performing Teams – the team-level version of the same logic: a condition that enables behavior rather than an incentive that compels it.
• The Peak of Not Knowing: Why Psychology's Most Famous Effect May Be Just a Statistical Mirage – the methodological contrast: a textbook finding that fails statistical scrutiny, while the undermining effect survives it across 128 studies.
• The Gentle Push: Why a Pre-Ticked Box Achieves More Than Any Argument — and Whether It Is Permissible to Nudge People Toward Their Own Good – the intervention side: if behavior depends on how the situation is designed, then every incentive design is a normative decision.
• Losses Loom Larger: How Prospect Theory Dethroned Rational Man – and Why Its Most Famous Principle Is Now on Trial Itself – the second great fracture in expected-utility theory: there, the valuation of outcomes; here, the valuation of reasons.
• The Long Tail of Latency: Tail Latency and Why Averages Lie in the Cloud – the metrics problem in technical form: what you average disappears; what you reward grows.
• The Predictive Brain: Predictive Processing and the Illusion of Perception – one layer deeper: why expectation shapes not only behavior but perception itself.
• The 38 Witnesses Who Never Were: The Bystander Effect Between Laboratory Finding and Video Evidence – the related case of a debate in which two solid bodies of evidence appear contradictory only because they measure different things.

Sources

• Deci, E. L. (1971): Effects of externally mediated rewards on intrinsic motivation. Journal of Personality and Social Psychology, 18(1), 105–115. DOI: 10.1037/h0030644
• 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. DOI: 10.1037/h0035519
• Cameron, J., & Pierce, W. D. (1994): Reinforcement, reward, and intrinsic motivation: A meta-analysis. Review of Educational Research, 64(3), 363–423. DOI: 10.3102/00346543064003363
• 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.
• Deci, E. L., Koestner, R., & Ryan, R. M. (2001): Extrinsic rewards and intrinsic motivation in education: Reconsidered once again. Review of Educational Research, 71(1), 1–27. DOI: 10.3102/00346543071001001
• 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. DOI: 10.1037/a0035661
• Van den Broeck, A., Ferris, D. L., Chang, C.-H., & Rosen, C. C. (2016): A review of self-determination theory's basic psychological needs at work. Journal of Management, 42(5), 1195–1229. DOI: 10.1177/0149206316632058
• Liu, Y., Yang, Y., Bai, X., Chen, Y., & Mo, L. (2022): Do immediate external rewards really enhance intrinsic motivation? Frontiers in Psychology, 13, 853879. DOI: 10.3389/fpsyg.2022.853879

← All articles