nucleation

LIT-1477: A systematic review of human avoidance learning: Cognition, computation, and methods

2 min read

Properties5
StatusActive — read in full 2026-10-24 (NOTE-1134) from the PsyArXiv preprint, not the closed version of record; worth reading as the only systematic map of human avoidance learning, and for the finding, rated robust, that an available avoidance response protects threat expectancy from extinction, the more so the more effective it is. Avoidance is learned without ending the warning signal and transfers by inference, against two-factor theory; but some avoidance survives visible removal of the outcome, against a purely inferential account. Healthy volunteers with shock or noise only; nothing here is about anxiety disorders.
Version3
Tagslearning-and-conditioning, psychopathology-and-treatment
Cited as “paper” byNOTE-1134 — Mancinelli & Bach — A systematic review of human avoidance learning
Cited as “source” byTHEORY-1184 — In three laboratory studies of healthy humans, offering a costless avoidance response for a stimulus never paired with the aversive outcome led people to use it, and withdrawing the response then raised their expectancy of the outcome for that stimulus; signalling of the response is a confound, THEORY-1189 — In laboratory avoidance learning in healthy humans, an avoidance response that remains available during extinction persists, in some participants for hundreds of unpunished trials, and keeps expectancy of the aversive outcome higher than in people denied it; the more effective and the cheaper the response, the slower it extinguishes, THEORY-900 — A systematic review of 116 studies of avoidance learning in healthy humans finds the evidence contradicting two-factor and other associative theories and favouring expectancy and inference-based accounts, but also finding residual avoidance after the outcome is visibly removed, which a purely inferential account does not predict, THEORY-901 — As Mancinelli and Bach's systematic review of avoidance learning in healthy humans reports it, rating both A, an avoidance response left available during extinction persists, in some participants for hundreds of unpunished trials, and keeps expectancy of the aversive outcome higher than in people denied it (nine studies); and less effective or costlier responses extinguish faster (six studies, one exception), THEORY-920 — As Mancinelli and Bach's systematic review reports it, rating it B, in three laboratory studies of healthy humans a signalled, costless avoidance response offered for a stimulus never paired with the aversive outcome was used, and withdrawing it raised expectancy of the outcome for that stimulus, though Xia et al. (2019) found no change in skin conductance or startle; every study signalled the response's availability, which may itself have read as a danger cue

Graph View

Mancinelli, F. & Bach, D. R. (2026). A systematic review of human avoidance learning: Cognition, computation, and methods. Neuroscience & Biobehavioral Reviews 190: 106916. doi:10.1016/j.neubiorev.2026.106916. Preprint: PsyArXiv, doi:10.31234/osf.io/yqcd4_v1 (2025).

Key takeaways

From the reading, NOTE-1134. Page numbers are the preprint’s.

  • Scope. 116 articles (1939–2024) of “experimental procedures involving healthy humans” (p. 7) with shock or loud noise; a systematic review under PRISMA, not a meta-analysis. Phenomena are rated A to C for robustness (Fig. 4).
  • Against two-factor theory. Avoidance is learned without ending the CS (8 studies, A); it transfers to stimuli never trained with it; it appears after Pavlovian learning alone (A).
  • Protection from extinction (A). With avoidance available, “42% (Malloy & Levis, 1988) and 47% (Williams & Levis, 1991) of participants did not extinguish a unary and costless AR by the 500th would-be-unpunished trial” (p. 18), and those allowed to avoid “retain higher US expectancies” (p. 18). More effective responses last longer, “putatively because effective ARs provide fewer opportunities to detect changes in contingency” (p. 28).
  • Avoidance spreads to the safe stimulus (B). Offering a response for a never-punished stimulus, then withdrawing it, raises threat expectancy for that stimulus (pp. 19–20); signalling is a confound.
  • Not purely inferential. Some avoidance persists after signalled outcome omission, without habits forming: “the evidence also contradicts a purely inference-based account” (p. 25).
  • Paradigms too simple. Mostly instructed, signalled button presses; the authors call for multi-step, naturalistic tasks.

Why this is here

The exchange cited it at turn U14, unnamed, as a 2026 systematic review of 116 human avoidance-learning studies giving the strongest support to expectancy- and inference-based accounts. The abstract confirms the count and the verdict for inference-based accounts against associative ones; it does not name expectancy accounts separately, and it adds that the inference-based theories are themselves incomplete.

Corrected 2026-10-24, from the reading. The review supports expectancy theory by name, which the abstract does not, but it also says the evidence contradicts a purely inference-based account. Its studies are of healthy volunteers only, so it says nothing directly about anxiety. For CLAIM-446 it supplies laboratory evidence for the claim’s identifiability step (protection from extinction) and for the closing inference from one’s own avoidance to danger (spread to the safe stimulus), and it qualifies the claim’s sentence that avoidance “does not strengthen the belief”. Filed from it: THEORY-901, THEORY-920 and THEORY-900.

Backlinks