Cited by 4,027Neural and Behavioral Psychology StudiesNeurotransmitter Receptor Influence on BehaviorFree Will and Agency

The neural basis of human error processing: Reinforcement learning, dopamine, and the error-related negativity.

Holroyd, Clay B., Coles, Michael G. H.·2002·Psychological Review

Abstract

The authors present a unified account of 2 neural systems concerned with the development and expression of adaptive behaviors: a mesencephalic dopamine system for reinforcement learning and a "generic" error-processing system associated with the anterior cingulate cortex. The existence of the error-processing system has been inferred from the error-related negativity (ERN), a component of the event-related brain potential elicited when human participants commit errors in reaction-time tasks. The authors propose that the ERN is generated when a negative reinforcement learning signal is conveyed to the anterior cingulate cortex via the mesencephalic dopamine system and that this signal is used by the anterior cingulate cortex to modify performance on the task at hand. They provide support for this proposal using both computational modeling and psychophysiological experimentation.

Integrity signals

Registry

Crossref · 10.1037/0033-295x.109.4.679

Cites

280 of 280 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The ACC is a site of computation of prediction errors encoding the difference between expected and obtained outcomes of responding.

    “Moreover, studies using fMRI and electrophysiological (error-related negativity) methods in humans, as well as single-unit electrophysiology in experimental animals, have identified the ACC to be a site of computation of prediction errors encoding the difference between expected and obtained outcomes of responding [115–117].”
  • The functional meaning of FRN has shifted from a purely negative feedback detector toward a broader index of reinforcement-learning signals, reward prediction error, action-outcome evaluation, and reward positivity modulation.

    “its functional meaning has shifted from a purely “negative feedback detector” toward a broader index of reinforcement-learning signals, reward prediction error, action–outcome evaluation, and reward positivity modulation [117,118,126,127,128,129].”
  • Classical reinforcement-learning models propose that unexpected unfavourable outcomes produce negative prediction-error signals that are conveyed through dopaminergic and medial-frontal systems and used to update future behaviour.

    “Classical reinforcement-learning models propose that unexpected unfavourable outcomes produce negative prediction-error signals that are conveyed through dopaminergic and medial-frontal systems and used to update future behaviour [117,118].”
  • In the broader ERP literature, response-locked performance monitoring is usually described as a sequence in which early medial-frontal activity detects conflict, error likelihood, or worse-than-expected outcomes, whereas later positive components support more conscious, evaluative, and metacognitive appraisal of the action that has just been executed.

    “In the broader ERP literature, response-locked performance monitoring is usually described as a sequence in which early medial-frontal activity detects conflict, error likelihood, or worse-than-expected outcomes, whereas later positive components support more conscious, evaluative, and metacognitive appraisal of the action that has just been executed [113,116,117,185,186,187].”
  • Earlier components such as ERN are often interpreted in terms of rapid conflict detection, reinforcement-learning signals, or medial-frontal alarm mechanisms, whereas later positive components appear more closely related to conscious appraisal, confidence, and adaptive updating.

    “Earlier components such as ERN are often interpreted in terms of rapid conflict detection, reinforcement-learning signals, or medial-frontal alarm mechanisms, whereas later positive components appear more closely related to conscious appraisal, confidence, and adaptive updating [113,116,117,187].”

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Holroyd, C. B., & Coles, M. G. H.
Affiliations
University of Illinois Urbana-Champaign (US)
Venue
Psychological Review
Published
2002-10
Volume
109
Issue
4
Pages
679-709
ISSN
  • 0033-295X (ISSN-L)
  • 1939-1471
OpenAlex
W2135572147
Cited by
4,027
References
280
Citations by year
2026: 125 · 2025: 172 · 2024: 145 · 2023: 138 · 2022: 152 · 2021: 193 · 2020: 188 · 2019: 188 · 2018: 220 · 2017: 219 · 2016: 218 · 2015: 246 · 2014: 244 · 2013: 219 · 2012: 204
FWCI
30.66
Type
journal article
Language
en
Open access
closed
Locations
  • Psychological Review (publishedVersion)
  • PubMed (publishedVersion)
  • Ghent University Academic Bibliography (Ghent University) (submittedVersion)
Funders
  • National Institutes of Health (5r01mh041445-13, 5f31mh011530-04)
  • National Institute of Mental Health (MH41445, MH11530)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural and Behavioral Psychology Studies
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neurotransmitter Receptor Influence on Behavior
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Free Will and Agency
Keywords
error processing, error-related negativity, reinforcement learning, dopamine, anterior cingulate cortex, negative reinforcement, adaptive behavior, event-related potentials, reaction time task, computational modeling
MeSH terms
Analysis of Variance, Brain Mapping, Dopamine, Humans, Models, Neurological, Motor Cortex, Probability Learning, Reinforcement, Psychology

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review