The cognitive and neural architecture of sequence representation.

Keele, Steven W.; Ivry, Richard B.; Mayr, Ulrich; Hazeltine, Eliot; Heuer, Herbert · 2003 · Psychological Review

DOI: 10.1037/0033-295x.110.2.316

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 proposes a theoretical framework distinguishing two neurocognitive systems for sequence learning, identified through serial reaction time (SRT) experiments. The authors argue that human cognition relies on a dorsal system (involving parietal and supplementary motor cortex) and a ventral system (involving temporal and lateral prefrontal cortex). The dorsal system supports implicit, unidimensional learning, associating stimuli within a single dimension regardless of attention. In contrast, the ventral system supports both implicit and explicit learning by forming associations across dimensions, but it requires attentional selection of relevant signals. This distinction explains why dual-task interference disrupts sequence learning: secondary tasks do not merely tax limited capacity but disrupt the coherence of attended signals required for the ventral system’s cross-dimensional associations. The theory is grounded in behavioral and neuroimaging evidence. Behavioral studies, such as those by Curran and Keele (1993), demonstrated that while single-task conditions allow for greater sequence learning (engaging both systems), dual-task conditions restrict learning to a residual level consistent with the unidimensional dorsal system. Notably, even subjects who were explicitly aware of the sequence showed reduced learning under dual-task conditions, indicating that awareness does not protect against this interference. Neuroimaging data from positron emission tomography (PET) studies by Grafton et al. (1995) and Hazeltine et al. (1997) confirmed distinct neural architectures for these conditions. Dual-task learning activated a dorsal network, including the left supplementary motor area, parietal cortex, and primary motor cortex. Single-task learning activated a ventral network, involving the right temporal lobe, inferior prefrontal cortex, and lateral premotor cortex. These neural sets were largely nonoverlapping, supporting the existence of two dissociable learning systems. The findings imply that the lack of correlation between attended signals, rather than limited cognitive resources, is responsible for dual-task effects on learning. The ventral system’s reliance on attentional gating means that unattended or uncorrelated secondary information prevents the formation of cross-dimensional associations. Conversely, the dorsal system operates automatically on uninterpreted stimuli within a single dimension, remaining robust against such interference. This model reframes the debate between implicit and explicit memory, suggesting that the critical distinction lies in computational capability—specifically, the ability to integrate information across dimensions—rather than in awareness alone. The theory has broad implications for understanding procedural versus declarative memory, the neural basis of complex sequential skills like speech, and the roles of the basal ganglia versus the hippocampus in learning.

Key finding

Sequence learning is supported by two distinct neural systems: a dorsal pathway for implicit, unidimensional learning that is resistant to dual-task interference, and a ventral pathway for multidimensional learning that requires attention and is disrupted by secondary tasks.

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.

• dual task multitasking