1. Neuroscience
Download icon

Understanding neural signals of post-decisional performance monitoring: An integrative review

  1. Kobe Desender  Is a corresponding author
  2. K Richard Ridderinkhof
  3. Peter R Murphy
  1. Brain and Cognition, KU Leuven, Belgium
  2. Department of Neurophysiology and Pathophysiology, University Medical Center Hamburg-Eppendorf, Germany
  3. Department of Psychology, University of Amsterdam, Netherlands
  4. Amsterdam center for Brain and Cognition (ABC), University of Amsterdam, Netherlands
  5. Trinity College Institute of Neuroscience, Trinity College Dublin, Ireland
Review Article
Cite this article as: eLife 2021;10:e67556 doi: 10.7554/eLife.67556
4 figures

Figures

The Pe as a signature of post-decisional evidence accumulation.

(A) Schematic of pre- and post-decisional evidence accumulation. In the pre-decisional period of a two-alternative forced-choice task, noisy sensory evidence accumulates over time until the decision variable (DV; black trace) reaches one of two bounds (a or -a), corresponding to the two available choice alternatives. After the DV reaches a boundary and triggers a choice, evidence accumulation continues during the post-decisional period (informed by the same sensory evidence and/or other sources of ‘error evidence’). In the depicted example, post-decisional evidence can either further confirm and thus be consistent with the choice made (cyan trace), it can be inconsistent with the choice just made (purple trace) or not really informative about the preceding choice (marine blue trace). In the other panels, we focus on how such post-decisional accumulation can explain different expressions of performance monitoring. (B) Error detection. Post-decisional accumulation can produce error detection by assuming that participants impose another bound, represented here by the transition between the green and red area (left panel). Only when the post-decisional accumulated evidence ends up above that additional boundary (i.e. in the red area) will participants indicate that they made an error. Consistent with this, the Pe gradually ramps up preceding the detection of errors, whereas it is diminished in amplitude or even absent for undetected errors (right panel). (B) Reproduced from Figure 2 Murphy et al., 2015. (C) Decision confidence. Post-decisional accumulation can produce graded confidence judgments, by assuming the accumulated post-decisional evidence is compared against multiple discrete criteria, represented here by the six colored areas (left panel). The indicated level of confidence then directly depends on the category into which the DV falls. Consistent with this, Pe amplitude has been shown to scale inversely with reported level of confidence (right panel). (C) Reproduced from Figure 3 Boldt and Yeung, 2015. (D) Adaptive modulation of next-trial speed-accuracy tradeoff. The DV furnished by the post-decisional accumulation process can serve as the basis for altering subsequent decision policy. Specifically, depending on the accumulated evidence that a preceding choice was incorrect, the decision boundary can be changed to instantiate a more cautious policy on the subsequent trial and thus decrease the probability of consecutive errors (left panel). Consistent with this, the Pe positively predicts the height of the decision boundary on the following trial (right panel). (D) Replotted from Figure 8 Desender et al., 2019a.

Morphological characteristics of the Pe.

(A) Schematic of post-decisional evidence accumulation for fast, medium and slow responses when time-locked to error commission (left panel) or to error detection (right panel). (B) Average Pe amplitudes separately for fast, medium, and slow detection RTs when time-locked to error commission (left panel) or to error detection (right panel). (B) Reproduced from Figure 2 Murphy et al., 2015.

Appendix 1—figure 1
The relation between confidence and accuracy and between error detection and accuracy for the model in which post-decisional accumulation was a continuation of pre-decisional accumulation (top panel) and for the model that selectively accumulates evidence that conflicted with the initial choice (bottom panel).
Appendix 1—figure 2
Both models provide qualitatively similar predictions concerning three key findings concerning confidence.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)