Attention as a Solution Class to Optimization under Constraint

Many contemporary scholars have called for abandoning the term "attention," arguing that it has been applied so broadly that it can no longer refer to a single underlying mechanism. This article asserts that attention need not refer to one mechanism, but rather comprises a solution class to the cognitive system's most fundamental problem: how to maximize performance objectives, given inherent biological and environmental constraints. With this perspective, separate solutions may be implemented simultaneously across multiple functional domains, such as perception, decision-making, and learning. Here, a taxonomy of attentional solutions was extracted through focused review of findings from model-based cognitive neuroscience, a field predicated on mathematical formalization and simultaneous consideration of behavioral performance (i.e. objectives) and process-level neurophysiological markers (i.e. biologically-constrained resources). The goal of this review is thus to reify attention while embracing its plurality, offering a framework whereby attentional solutions can be proposed, evaluated, and falsified in the context of system-level objectives and constraints.