Task Switching: A PDP Model

Gilbert, Sam J.; Shallice, Tim · 2002 · OpenAlex-citations

DOI: 10.1006/cogp.2001.0770

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 presents a Parallel Distributed Processing (PDP) computational model to explain the cognitive mechanisms underlying task-switching costs, specifically addressing the debate between "task carryover" and "exogenous control process" theories. The research is motivated by conflicting empirical findings regarding why reaction times are slower when switching between tasks compared to repeating them. While some theories attribute this cost to executive control processes that reconfigure the cognitive system, others argue it results from the involuntary persistence of the previous task set and associative priming. The authors aim to clarify this debate by testing whether a task carryover account, augmented with associative learning, can sufficiently explain empirical data without invoking additional executive control stages. The model simulates switching between word reading and color naming using Stroop stimuli. It builds upon earlier Stroop models by incorporating two key mechanisms: first, the persistence of activation and inhibition in task-controlling representations from one trial to the next, implemented via a "squashing" parameter that retains a proportion of the previous trial’s state; second, associative learning, where stimuli form temporary connections with task sets based on recent co-activation, allowing stimuli to evoke previously associated tasks. The architecture consists of separate pathways for word and color processing, with task demand units receiving top-down control inputs and bottom-up inputs from stimuli. Connection weights between stimuli and task demands are updated via Hebbian learning after each trial but reset to ensure effects persist only for the immediate subsequent trial. The model successfully simulates a large body of empirical data, including the "paradoxical" asymmetric switch costs where switching into a dominant task (word reading) incurs a larger cost than switching into a nondominant task (color naming). This asymmetry is explained by the model as resulting from the persistence of inhibition applied to the dominant task during the nondominant task, which must be overcome on switch trials. Furthermore, the model replicates item-specific switch costs, where reaction times are slower for stimuli recently associated with the previous task, supporting the associative-task set inertia hypothesis. Crucially, the model demonstrates that these carryover effects alone can account for the dramatic improvement in reaction time from the first to the second trial in a task run, a finding previously argued to require exogenous control processes. The model’s behavior remains robust even when parameters are set to random values within plausible ranges. The significance of this work lies in its support for the task carryover account of switch costs. By demonstrating that a model based on persisting activation states and associative learning can fit empirical data—including findings previously seen as problematic for this explanation—the authors suggest that complex executive control processes may not be necessary to explain task-switching costs. Instead, switch costs may primarily reflect interference from the previous task set and stimulus-evoked priming. This provides a unified computational framework that clarifies the theoretical debate, suggesting that the differences between competing theories may be matters of emphasis rather than fundamental mechanistic disagreement.

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_with_issues.

Topics

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

• task switching
• executive function