What Happens After a Fast Versus Slow Error, and How Does It Relate to Evidence Accumulation?

Damaso, Karlye A. M.; Williams, Paul G.; Heathcote, Andrew · 2022 · Computational Brain & Behavior

DOI: 10.1007/s42113-022-00137-2

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper investigates the cognitive mechanisms underlying post-error adjustments in decision-making, specifically addressing why responses are sometimes slowed and other times sped up after an error. While the traditional view posits that errors trigger increased response caution (post-error slowing) to improve accuracy, recent findings show inconsistent effects, including post-error speeding and accuracy decreases. The authors aim to disentangle these effects by distinguishing between two types of errors—fast "response-speed" errors and slow "evidence-quality" errors—and by controlling for confounds like fatigue using a robust trial-pairing method. The study analyzes data from two recognition memory experiments where participants were instructed to prioritize either speed or accuracy. Experiment 1 featured a short response-to-stimulus interval, while Experiment 2 included a longer interval due to a post-decision confidence rating. The authors fitted the data using two evidence accumulation models: the Drift Diffusion Model (DDM) and the Linear Ballistic Accumulator (LBA). Using hierarchical Bayesian estimation and the Deviance Information Criterion for model selection, they compared how well each model accounted for post-error effects on response time and accuracy. The results indicate that the LBA provided a superior fit to the data compared to the DDM. The nature of post-error effects depended heavily on the experimental conditions. In Experiment 1 (short interval), post-error speeding and slowing were driven entirely by changes in non-decision time (encoding and response execution), with no effects on decision thresholds or evidence accumulation rates. In contrast, Experiment 2 (long interval) revealed that post-error effects also influenced the rate of evidence accumulation and the amount of evidence required for a decision. Specifically, under speed emphasis, fast errors were followed by slowing, whereas under accuracy emphasis, slow errors were followed by speeding. These findings challenge the classic account that post-error slowing is solely due to increased response caution. Instead, the results suggest that post-error adjustments are multifaceted, involving both non-decision processes and decision parameters, depending on task constraints. The distinction between error types is critical: fast errors may trigger adaptive caution, while slow errors may trigger "recklessness" or speeding, particularly when participants are frustrated or prioritize speed. The study highlights the importance of using robust analytical methods and flexible models like the LBA to accurately characterize the latent cognitive variables driving post-error behavior.

Key finding

Post-error adjustments are driven by changes in non-decision time under short inter-trial intervals but involve changes in evidence accumulation rates and thresholds under longer intervals, with the linear ballistic accumulator providing a more accurate account of these effects than the drift diffusion model.

Methodology

modeling

Sample size: 93

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via author_sweep_intake on 2026-05-28.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• human error taxonomy