Forecast gap · what you predict vs what you feel
Peak-End Rule
The peak-end rule is the tendency to judge past experiences primarily by their most intense moment (the peak) and how they ended, rather than by the total duration or average quality of the experience.
The forecast reaches for the dashed ceiling; reality settles at the solid line. The hatched shortfall is invisible in advance — you brace for the whole column and feel a fraction of it.
Why the forecast misses
The peak-end rule arises from how the 'remembering self' constructs a compressed summary of experiences, as distinct from the 'experiencing self' that lives through each moment (Kahneman, 2000). Memory appears to extract prototypical snapshots using the representativeness heuristic, averaging only the most affectively intense moment and the final moment. Duration neglect occurs because these snapshots dominate actual experience. The underlying mechanism is thought to be an efficient compression strategy that reduces complex temporal experiences to simple evaluative summaries for future decision-making.
Competing account — The accessibility model (Robinson & Clore, 2002) suggests that immediate reports reflect experiential knowledge (biased toward peak and end), but long-delayed reports become reconstructed and may be influenced by beliefs and individual differences. Some boundary conditions exist: the peak-end effect weakens for very long positive experiences (holidays) and in multiparous mothers recalling labor pain 2 months post-delivery.
Forecasts on record
A majority of participants chose the longer trial (90 seconds total) over the shorter one (60 seconds), despite experiencing more total pain. The gentler ending made the memory of the longer experience less aversive. This demonstrates both the peak-end rule and duration neglect.
Kahneman, Fredrickson, Schreiber & Redelmeier · 1993 · Undergraduate participants
Global post-procedure ratings were predicted almost perfectly by the average of the peak pain moment and the final minute's pain (correlations in the 0.7-0.9 range). Total procedure duration, which varied from 4 to 69 minutes, predicted global ratings barely at all.
Redelmeier & Kahneman · 1996 · 682 colonoscopy patients
The modified procedure was associated with a significantly lower memory of peak pain, less painful global memories, and an increased likelihood of returning for follow-up screening. The intervention of adding a gentler ending saved lives by increasing compliance.
Redelmeier, Katz & Kahneman · 2003 · Colonoscopy patients in randomized trial
The average of peak and end pain was the best predictor of recollected pain 2 days (r=.588) and 2 months (r=.509) after delivery. Delivery duration was uncorrelated with recalled pain (r=-.008 and .012). Mothers with epidurals (less painful ending) recalled less pain than matched mothers without epidurals.
Chajut, Caspi, Chen, Hod & Ariely · 2014 · n = 320 women in childbirth
Evidence — robust. The effect has replicated across multiple domains including cold-pressor tasks, colonoscopies, labor pain, aversive sounds, and consumer experiences. The 2003 colonoscopy randomized trial demonstrated practical impact by increasing patient return rates. Chajut et al. (2014) extended the finding to a real-world, hours-long, emotionally complex experience (childbirth), showing the effect is not limited to brief laboratory procedures. Some boundary conditions have been identified: duration neglect is softer for long positive experiences (Kemp et al., 2008), and the effect weakens with prior experience of the event type.
Where the gap cost something
medicine · 2003
Colonoscopy patient return rates
In a randomized trial, patients who received a modified colonoscopy with a gentler extended ending were significantly more likely to return for follow-up screening than those who received the standard procedure. The longer procedure was remembered as less unpleasant, demonstrating that modifying the peak and end shapes patient behavior and medical compliance.
medicine · 2014
Labor pain and epidural analgesia memory
Women who received epidural analgesia during childbirth recalled significantly less pain 2 months post-delivery than matched women without epidurals, even when total pain experienced was similar. The less painful ending created by the epidural improved the memory of the entire labor experience, consistent with peak-end rule predictions.
To forecast honestly
Design experiences by deliberately engineering both the emotional peak and the ending, even at the cost of the middle. For painful experiences (medical procedures, customer complaints), ensure the ending is as gentle as possible even if it extends duration. For pleasant experiences, create a clear positive peak moment and a strong closing. Collect moment-by-moment feedback rather than just global ratings to avoid being misled by peak-end evaluations.
The 2003 colonoscopy randomized trial showed that extending the procedure with a gentler ending significantly improved patient memories and increased return for follow-up screening. This demonstrates that manipulating the end experience directly changes both memory and behavior.
Catch it in the act
Watch for situations where someone evaluates an experience based on 'how it ended' or 'the worst/best moment' while being unable to accurately gauge the total duration or average quality. Common in travel reviews, medical procedure ratings, restaurant evaluations, and employee assessments of meetings or projects.
Related forecasting errors
- Duration NeglectchildDuration neglect is the corollary finding that total experience duration contributes almost nothing to retrospective evaluation. It is a direct consequence of the peak-end rule's snapshot model of remembered utility.
- Experiencing Self vs. Remembering SelfparentThe peak-end rule is a key empirical foundation for Kahneman's distinction between the experiencing self (moment-by-moment utility) and the remembering self (compressed evaluative summary).
- Recency Effectmechanistically-linkedThe end-weighting component of the peak-end rule shares mechanism with the recency effect, but the peak-end rule uniquely shows that peak intensity also matters independently and that duration is neglected.