Remembered experience is dominated by peaks and endings, inconsistent with the average of momentary experience

What it is

The peak-end rule: when people recall an episode afterward, the evaluation is driven mainly by its most intense moment (the peak) and its ending (the end), not by the average across all moments. An episode with a flat middle and a pleasant ending is remembered better than its moment-by-moment mean would predict.

Why it happens

Recall is reconstruction, not playback: memory keeps salient events (intense, emotionally tagged moments) and sequence-end information while heavily compressing the middle. This explains two common gaps—"the process was grueling but I remember it fondly" (pain front-loaded, ending resolved) and "it went smoothly yet I remember it badly" (one late snag defined the whole). For design, investment in moment-to-moment polish that never lands on a peak or an ending contributes nearly nothing to the remembered verdict.

Studying it

The validation paradigm comes from Kahneman and Redelmeier's medical-pain studies: record moment-to-moment discomfort continuously and the retrospective evaluation afterward, then fit a "peak + end" model against a "mean" model for predictive power. Interface research can reuse it: collect minute-level immediate ratings during the task and recalled evaluation after, comparing the models' explanatory power. Design experiments manipulate the ending (normal, delightful, frustrating) and measure how far recall moves.

Where it stops holding

The effect is strongest for short, single episodes; long-horizon cumulative evaluation approaches a weighted average (peak-end of each episode, then averaged). Individual differences are large—emotionally stable participants recall closer to the mean. "Peak" means most intense, not best: a negative peak damages recall as much as a positive one helps. And the rule never endorses neglecting the middle: a mid-episode disaster becomes the peak itself.

Applying it

• Prioritize design resources at two points: the peak moment (the key result reveal) and the ending (completion screen, closing feedback).
• Audit every flow's endpoint: close with confirmation, thanks, and a next step—never let a flow end on an error toast or in silence.
• Use the immediate-rating curve to locate peaks, then amplify positive ones and defuse potential negative ones.

Related

• Same group: B5.12.1 Experience splits into anticipated, momentary, episodic, and cumulative time scales, and conclusions cannot move across scales · B5.12.3 A single measurement captures momentary or episodic experience and cannot be extrapolated to long-term experience · B5.12.4 First-use experience and long-term experience often call for opposite optimization directions
• Nearby: A3 Memory · P1 Emotional Design
• Search terms: peak-end rule · remembered experience · moment-to-moment