Proceedings of the Annual Meeting of the Cognitive Science Society

There is a tendency to criticize traditional attention research and argue for abandoning the term attention. This paper identifies the main critique as reification, circularity, false dichotomy, and one–many problem, and offers a systematic response. Circularity refers to treating attention as both explanans and explanandum. As argued, the underlying assumption of the reification critique concerns the specification of neural processes of attention. Drawing on Marr's distinction between levels of analysis, I argue that attention is a phenomenon at the computational level, generated by neural processes, which helps address these two critiques. To address false dichotomy and one–many problem, I adopt the active inference approach and offer a definition of attention as the rule-governed filtering of inputs. This definition justifies the conceptual distinction between intention and attention, attention and preattention, and reshapes the endogenous–exogenous distinction. It also explains how various types of attention fall under a single definition.