Automating adaptive control with item-specific learning

Cognitive control, as a goal-directed, or top-down, process, overcomes the automatic tendency to produce an action primed by the environmental stimuli. Recent empirical evidence suggests that cognitive control can be instantiated automatically by stimuli in a bottom-up manner based on prior associations between the stimulus and a particular control state. Memory-guided control explains how control is adaptive without recourse to the idea of a homunculus. This chapter reviews a mixed collection of research lines, each focusing on a distinct component of cognitive control, to establish the universality of automating control via item-specific stimulus–control (S-C) learning. The reviewed evidence suggests that S-C learning is a unique form of learning and can be distinguished from classical stimulus–response learning. Proponents of S-C learning assume that it inherits the benefit of both automatic and controlled processing to be fast, effortless, flexible and generalizable. Yet, these assumptions await to be checked in future research.