Strategic preparation in task-switching paradigms: which task set to prepare and when?

What it means to be mentally ''prepared'' is multifaceted, variously encompassing psychophysiological states that may depend on prior practice and planning, and current attentional and sensorimotor readiness. All of these depend on the nature of predicted future events. One domain in which preparation has been intensively investigated is task switching. We provide a rational analysis, in which questions of what to prepare and when come to the fore. Specifically, this analysis seeks to resolve the tradeoff between the mental costs associated with initiating and maintaining states of preparation, and the performance costs of sloth and inaccuracy associated with failing to be prepared when the target arrives. We show that a simple model derived from this analysis can explain a range of task-switching phenomena, including the residual switch cost, the mixture property of reaction times implied by the failure-to-engage hypothesis, and changes in behaviour when the predictability of key experimental variables (stimulus timing and task identity) are manipulated. We then build a slightly more complex model, motivated by some of the many observations about foreperiod effects in the performance of single tasks, which captures various subtle patterns of inaccurate performance. We end by discussing the benefits and limitations of our primarily top-down approach.