GDnD Wiki

GDnD Wiki

This site is for Ulster University students and staff. Enter the class access password to continue.

Overjustification Effect

4 min read

Summary

The overjustification effect is the finding that giving people an expected, external reward for something they already enjoy can reduce their intrinsic interest in it. The reward becomes the reason for the activity: when the reward stops, so does much of the activity, sometimes falling below where it started. For designers it is the central caution about extrinsic rewards (points, badges, loot, payment): a reward can add fun, or it can replace it.

Key ideas

  • The activity was already rewarding. The effect concerns intrinsically motivated behaviour. Rewarding a task nobody wanted to do is a different situation.
  • Expected, tangible, contingent rewards do the damage. “Do this and you’ll get that” reframes play as work.
  • Controlling vs informational rewards. Cognitive evaluation theory (part of self-determination-theory) distinguishes rewards experienced as controlling (pressure to comply; they undermine autonomy) from rewards experienced as informational (feedback that you have become competent; they can support intrinsic motivation). The same badge can be either, depending on framing and context.
  • The free-choice measure. Researchers measure intrinsic motivation by what people do when nothing is at stake: after the rewarded session, does the person keep playing when left alone?

Evidence

  • Lepper, Greene and Nisbett (1973), “Undermining children’s intrinsic interest with extrinsic reward”: preschool children who liked drawing with felt-tip markers were split into groups. Those who were promised a “Good Player” certificate for drawing later spent less free-play time drawing than children who received no reward or an unexpected one. The unexpected-reward group was not harmed.
  • Deci (1971) introduced the free-choice paradigm with puzzle-solving: participants paid for solving puzzles spent less time on the puzzles during an unobserved free period than unpaid participants.
  • Deci, Koestner and Ryan (1999), a meta-analysis of 128 experiments in Psychological Bulletin: expected, tangible rewards contingent on doing or completing a task reliably reduced free-choice intrinsic motivation. Positive verbal feedback tended to enhance it (at least for college students). Unexpected rewards and rewards not contingent on the task had little effect.
  • The course lecture series presents the effect as “paying people to play can kill the fun” and contrasts controlling rewards (“do this and you’ll get that”) with informational ones (“look how good you’ve got”) (course lecture series, see source-player-psychology-lectures).

In practice

Common places the effect shows up in games:

  • Daily login rewards and streaks: players log in to protect the streak, not because they want to play. When the streak breaks, many stop entirely.
  • Achievements for things players would do anyway: an achievement for exploring a map players already enjoy exploring can turn curiosity into checklist completion.
  • Gamified learning or fitness: points for reading or running can raise activity in the short term and lower it when the points stop (see gamification).
  • Tournament prizes and public leaderboards: the prize becomes the point; players who cannot win stop caring.

Design moves that reduce the risk:

  1. Make rewards informational. Frame them as feedback on growing skill (“new personal best”, “you cleared it without taking damage”) rather than as payment.
  2. Prefer unexpected or occasional rewards for activities that are already fun, rather than announcing a fixed payment for every action.
  3. Reward with more play. Rewards that open new content or abilities (access and facility in reward-types) feed back into the activity rather than replacing it.
  4. Check the free-choice signal in playtests. After a rewarded session ends, watch who keeps playing when nothing is at stake (see playtesting).

Implications

  • “Just add points” is not a design. Extrinsic rewards need a reason beyond engagement numbers.
  • The effect is strongest where intrinsic interest is already high, which is exactly where game designers usually want to add rewards.
  • It connects reward design to ethics: systems that keep players through controlling rewards rather than enjoyment drift towards dark-patterns.

Open questions

  • The size and generality of the effect have been debated. Cameron and Pierce (1994) argued from their own meta-analysis that the negative effects were limited; Deci, Koestner and Ryan’s 1999 analysis was in part a response. Designers should treat the effect as a reliable risk, not a law.
  • How far do findings from short laboratory tasks transfer to long-running games with layered reward systems? Arguably, games combine controlling and informational rewards so tightly that the balance shifts over a player’s lifetime.

Graph View

Backlinks