Expectation, Retrospection & Evaluation Bias

← Back to Domain-Specific Families

Abstractions about hindsight, expectancy, duration neglect, idealized baselines, hedonic adaptation, covariate shift, and biased evaluation over time.

12 abstractions in this family — domain-specific abstractions that sit near one another in structural-signature space (k-means over structural-signature embeddings). Each is shown with its short description.

• Choice-Supportive Bias — Explain why we misremember past decisions in our own favor: dissonance reduction rewrites the stored memory so the chosen option looks better and the rejected ones worse than the evidence actually showed at the time of choice.
• Covariate-Shift Blind Spot — The deployment failure in which a model's input distribution P(X) drifts outside its training support while P(Y|X) holds — and goes undetected because monitoring watches lagging outcome metrics instead of the immediately-available input signal.
• Crespi Effect — Read operant behavior after a reward change as a discrepancy-driven transient rather than an absolute-value response: animals shifted up overshoot same-reward controls (elation) and animals shifted down undershoot them (depression), calibrated to the signed gap between delivery and a recalibrating expectation.
• Declinism — Believe the past was better and the future will be worse — even against the measured data — via a self-reinforcing loop of rosy retrospection on the past, negativity- and recency-weighted evaluation of the present, and naive extrapolation forward.
• Duration Neglect — The judgment bias in which the length of a bounded experience contributes negligible weight to its retrospective evaluation, which is instead built from a representative peak and the final moments (the peak-end rule) — so remembered utility systematically diverges from the integral of what was lived.
• End-of-History Illusion — Explain why people at any age acknowledge large past-decade change yet forecast far less future change — an asymmetry in which vivid episodic contrasts underwrite recall while the future self, lacking any episodic record, is built by minimal edits to the present self.
• Exaggerated Expectation — Lift the expectation-realization gap out as a third measurable quantity carrying a stable sign — so a forecaster's systematic, directional miss across a class of events is diagnosed as a defect in the forecasting machinery, not random noise.
• Expectancy Disconfirmation — Compare perceived performance with a prior expectation and use the signed discrepancy—positive, zero, or negative—as a primary input to satisfaction, so the same performance can be evaluated differently when its reference changes.
• Gold-Standard Erosion — Recognize that a model scored against a mutable reference label can show stable metrics while its real validity silently degrades, because the answer key — not the model — has drifted away from the construct it once operationalized.
• Hedonic Treadmill — The pattern in which subjective well-being rises or falls after a life change but drifts back toward a temperamentally stable baseline over weeks to years, because a gain-control mechanism recalibrates affective evaluation against recent experience.
• Hindsight Bias — The tendency, once an outcome is known, to inflate its perceived prior probability and misremember one's own judgment as closer to it than it was, because outcome knowledge is folded into the memory of the prior situation and reconstructs rather than retrieves it.
• Scoring Rule — Evaluate a probabilistic forecast after its outcome by mapping the report–outcome pair to a numeric loss or reward, with propriety governing whether truthful distributions are optimal in expectation.