Retrospective Evaluations of Gambling Wins: Evidence for a ‘Peak-End’ Rule

This study explored retrospective evaluations of temporally extended outcomes. Participants, primarily non-gamblers or recreational gamblers, played a fruit machine that presented sequences of payouts and then retrospectively evaluated their experiences. Experiments 1 and 2 replicated a robust “peakend” effect on estimation judgments and choices, with participants exhibiting biases in favor of sessions with high peakend values. Experiment 3 used modified stimuli that reduced the payout structure to wins and non-wins (no extreme values) and found no effect; we suggest this may arise from the influence of sensitivity to affect. Analysis across conditions found a consistent underestimation of total winnings and frequency of wins, in support of memory-based evaluation strategies. Consequences of these findings for gambling judgments and decisions are discussed.