An illustrative guide to expressing cognitive theories using evidence accumulation modelling

Strickland, Luke; Boag, Russell J.; Stevenson, Niek; Heathcote, Andrew · 2026 · Crossref

DOI: 10.3758/s13428-026-02970-w

archive: archived pipeline: cataloged verified

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

Summary

This tutorial addresses the challenge of connecting cognitive process theories to complex behavioral data using evidence accumulation models (EAMs). While EAMs effectively decompose choice and response time distributions into latent decision parameters, existing guidance often focuses on simple experimental designs. This paper provides a detailed guide on how to instantiate complex cognitive theories by mapping EAM parameters to experimental designs using an augmented linear model language within the R package EMC2. The authors argue that directly estimating theory-driven parameters, rather than computing contrasts post-hoc, allows for more precise hypothesis testing, better prior specification in Bayesian frameworks, and the integration of trial-by-trial dynamics. The methodology utilizes hierarchical Bayesian models implemented in EMC2, specifically focusing on the linear ballistic accumulator model as a primary example, though the techniques apply to other race models. The tutorial demonstrates this approach through two case studies. The first instantiates the Prospective Memory Decision Control (PMDC) theory, which explains how individuals perform planned actions during an ongoing task. The PMDC model maps cognitive processes such as "capacity sharing" and "cognitive control" to specific EAM parameters. Capacity sharing is quantified by decomposing ongoing-task accumulation rates into "processing quality" (difference between match and mismatch rates) and "urgency" (sum of rates). Cognitive control is divided into proactive control, measured by threshold adjustments in prospective memory blocks, and reactive control, measured by inhibitory or excitatory effects on accumulation rates when target stimuli are encountered. The second example illustrates how humans integrate advice from automated decision aids into their choices. The findings demonstrate that cognitive theories can be directly embedded into EAMs by defining parameters that represent theoretical constructs, such as the difference in thresholds between conditions or the inhibition of ongoing-task processing. By using linear model language, the authors show how to create hypothesis-focused parameters that capture these effects directly during estimation. This approach allows for constraints that align with theoretical expectations, such as restricting reactive inhibition to positive values. The tutorial further shows how distinct theories, such as those for prospective memory and human-automation interaction, can be unified into a single EAM framework. This unified model was applied to real experimental data, demonstrating the practical utility of the method for analyzing complex designs where multiple cognitive mechanisms operate simultaneously. The significance of this work lies in providing a robust, generalizable workflow for computational cognitive modeling. By moving beyond simple dummy-coded parameterizations, researchers can more directly test theoretical predictions about cognitive mechanisms like attention, memory, and control. The ability to estimate theory-specific parameters directly facilitates the incorporation of dynamic processes, such as learning or trial-by-trial adjustments, and strengthens the Bayesian inference process by allowing priors to be placed on meaningful psychological constructs. This guide bridges the gap between abstract cognitive theories and the statistical modeling of behavioral data, offering a template for future research in decision-making and cognitive control.

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.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

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

• signal detection theory

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Theoretical Contribution: computational model