Behavior often drops after rewards are withdrawn

What it is

Post-reward decline: when an extrinsic incentive stops, behavior often falls not just back to its pre-incentive baseline but below it — and below a never-incented control group. The activity collapse after a promotion ends and the participation crash when a streak bonus is cancelled follow this pattern. Decline is the temporal face of the problem: the incentive sustained behavior while active and punished it after removal.

Why it happens

Three forces pull down simultaneously. First, attribution history: during the reward period the person explained the behavior as "for the payoff"; that explanation survives the payoff, so "why continue with no reward?" becomes an honest question. Second, damaged interest: whatever interest was crowded out during the reward period does not return with the reward — the undermining happened while the reward was live; withdrawal merely exposes it. Third, price anchoring: the reward set a price for the behavior, and withdrawal marks it down to zero — an activity that once "earned points" now continues at no price, looking cheaper than one never priced at all. Large, expected, per-act promotional incentives load all three mechanisms at maximum.

Studying it

• Paradigm: post-cessation measurement is part of the free-choice paradigm (the free period after rewards settle is the withdrawal window); field studies use baseline–incentive–withdrawal phase designs, comparing withdrawal-phase behavior against baseline with a no-incentive control to separate seasonality from true decline.
• Variables: manipulations include incentive duration, contingency structure, and withdrawal style (abrupt vs. graded); outcomes are behavior frequency at successive post-cessation time points and time to return to baseline.
• Methodological cautions: novelty effects and natural decay masquerade as decline in field data — a control condition is mandatory; withdrawal dates often coincide with external events (terms, holidays, redesigns) and must be offset or modeled; how deep the drop is carries more information than whether it drops.

Where it stops holding

Decline is not fate. If skill growth or social bonds formed during the incentive period, part of the behavior survives — what survives is carried by growth and relationships, not by the reward. Short, small, gratitude-framed incentives decline shallowly; large, expected, per-act settlement declines deeply. Treating decline as a certainty, or betting it away, both miss the mechanism.

Applying it

• Treat every incentive program as one that must ship with an exit strategy: a defined withdrawal point, a withdrawal style, and a post-withdrawal support (competence feedback, community, habit prompts).
• Withdraw gradually rather than abruptly; replace reward feedback with competence feedback before the cutoff and watch retention through the transition.
• Track "behavior retention in week N after withdrawal," not "peak participation during the incentive" — the latter is flattering and misleading by construction.
• To validate: keep a non-settling control group for any new incentive and plot both behavior curves; the post-withdrawal gap between curves is the incentive's net cost.

Related

• Same group: P2.01.1 Intrinsic motivation comes from the activity itself · P2.01.2 Extrinsic rewards can crowd out intrinsic motivation · P2.01.4 Autonomy, competence, and relatedness are the three sources of intrinsic motivation · P2.01.5 Informational feedback does not crowd out; controlling feedback does · P2.01.6 Tasks with no intrinsic interest have no crowding-out problem · P2.01.7 Unexpected after-the-fact rewards harm less than promised ones
• Nearby: P3.10 Freedom to exit · P2.14 Reciprocity and social pressure
• Search terms: reward cessation · post-reward decline · incentive withdrawal