Evidence Accumulation and Decision Processes

Starns, Jeffrey J.; Heathcote, Andrew · 2024 · Oxford University Press eBooks

DOI: 10.1093/oxfordhb/9780190917982.013.33

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 examines the role of decision processes in memory research, arguing that memory is an interpretive process where retrieved information serves as evidence for a judgment rather than a direct record. The authors contend that understanding memory requires analyzing how this evidence is mapped onto explicit responses, a function performed by decision processes. The text reviews the shift from accuracy-based models, such as Signal Detection Theory (SDT), to evidence-accumulation models, which integrate both accuracy and response time (RT) data to provide a more comprehensive account of cognitive performance. The authors detail two primary evidence-accumulation frameworks: the Diffusion-Decision Model (DDM) and the Linear Ballistic Accumulator (LBA). The DDM assumes that noisy evidence is sampled sequentially over time until it reaches a threshold, with parameters including drift rate (memory strength), boundary separation (response caution), and starting point (bias). It incorporates both within-trial and between-trial variability, allowing it to fit complex RT distributions. The LBA, a simpler alternative, assumes linear accumulation with separate accumulators for each response option, where the first to reach a threshold determines the choice. Both models serve as measurement tools that disentangle memory processes from decision strategies, such as speed-accuracy tradeoffs. The paper also reviews the historical use of Receiver Operating Characteristic (ROC) curves in memory research, which were used to distinguish memory discriminability from response bias and to test theoretical models like the Dual-Process Signal Detection (DPSD) and Unequal-Variance SDT (UVSD) models. The authors note that while ROCs helped distinguish bias from discriminability, they failed to resolve debates regarding the nature of memory systems (e.g., recollection vs. familiarity) because ROC shape can be influenced by factors other than memory retrieval, such as confidence heuristics. The significance of this work lies in advocating for the adoption of evidence-accumulation models in memory research. These models offer superior explanatory power by accounting for RT data, which allows researchers to separate true memory strength from decision-related variables like response caution. This approach provides a more robust method for interpreting memory tasks, including recognition and source memory, and addresses limitations inherent in accuracy-only models that cannot distinguish between differences in memory ability and differences in decision strategy.

Key finding

Evidence-accumulation models provide a more comprehensive and theoretically robust framework for analyzing memory tasks than accuracy-only models because they simultaneously account for response accuracy, reaction time distributions, and the distinct contributions of memory strength versus decision caution.

Methodology

review

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.

• gap acceptance