An information-theoretic account of cognitive effort

Cognitive effort is a significant phenomenon, yet current psychological research remains entangled in several unresolved challenges. These include debates over the existence of the ego depletion effect, difficulties in accounting for diverse moderators that appear to mitigate this effect, and the dual-aspect of effort. This paper addresses these challenges by drawing on the active inference approach to cognitive effort, within a broad information-theoretic framework. This paper extracts a definition of cognitive effort from active inference, as the information cost of deviating from habitual to goal-directed mental actions, formally quantified by the Kullback – Leibler divergence. From this perspective, experiments cited as evidence against the existence of ego depletion effect are interpreted as invalid, and the dual-aspect of effort is reframed as a trade-off between accuracy and complexity in a capacity-limited agent. The ego depletion effect is then understood as a consequence of constrained complexity, and the diverse moderators are interpreted as reducing the estimated complexity. I conclude that active inference provides a satisfactory account of cognitive effort, which disentangles key challenges and deepens our understanding of cognitive efforts and its manifestations in everyday life.