A Skill‐Based Approach to Modeling the Attentional Blink. (17th July 2020)

• Record Type: Journal Article
• Authors: Hoekstra, Corné; Martens, Sander; Taatgen, Niels A.
• Is Part Of: Topics in cognitive science. Volume 12:Number 3(2020)
• Page Start: 1030
• Page End: 1045
• Publication Date: 2020-07-17
• Subjects: Attentional blink -- PRIMs -- ACT‐R -- Skill‐based modeling -- Cognitive model -- Instruction learning -- Skill‐based approach -- Cognitive architectures; Cognitive science -- Periodicals; Cognitive Science -- Periodicals; 153.05
• DOI: 10.1111/tops.12514
• Languages: English
• ISSNs: 1756-8757

Abstract

Abstract: People can often learn new tasks quickly. This is hard to explain with cognitive models because they either need extensive task‐specific knowledge or a long training session. In this article, we try to solve this by proposing that task knowledge can be decomposed into skills. A skill is a task‐independent set of knowledge that can be reused for different tasks. As a demonstration, we created an attentional blink model from the general skills that we extracted from models of visual attention and working memory. The results suggest that this is a feasible modeling method, which could lead to more generalizable models. Abstract : People can learn to perform new tasks very quickly by making use of lower‐level skills they have developed when learning previous tasks. Hoekstra, Martens, and Taatgen model this process, showing how a system trained on simple tasks (visual search and two working memory tasks) can then quickly learn to perform the attentional blink task, and it ends up making the same sorts of errors as people do.