Humans are remarkably efficient even when performing novel tasks. This ability for Rapid Instructed Task Learning (RITL) has recently been intensely researched.  We hypothesized that RITL requires proactive control, which according to Braver’s Dual Mechanisms of Control framework, is based on advance loading of task-related information into working memory. Here we propose a unified experimental framework that enables systematic testing of predictions that stem from this hypothesis. Participants will learn many new task rules, each in a separate experimental miniblock comprising of (a) instructions, (b) cuing which sub-set of the newly instructed rules remains relevant; (c) a NEXT phase in which participants press a fixed key to advance the screen and is where automatic rule activation is assessed; and (d) GO – the rule implementation phase. In two large-scale behavioral projects, we will systematically manipulate rule type and novelty; rule type; NEXT phase expectancy; GO phase length/reward; and individual differences in relevant cognitive abilities. These manipulations enable us to rigorously investigate: the effects of amount of preparation, the ability to shift rule representation mode when rules are deferred, and the removal of rule information from working memory. Using fMRI, we will contrast two hypotheses concerning the anterior-to posterior axis in prefrontal cortex organization: abstract/concrete or novel/familiar. Additionally, we will examine the hypothesis that low-ability individuals cannot use anterior prefrontal representations when preparing to execute a to-be-deferred novel rule, and must rely on more posterior prefrontal representations, which are less efficient in shielding performance from interference.
Supporting grant:
Integrated Framework for Studying Proactive Control in Rapid Instructed Task Learning BSF2015186 (Meiran, Cole and Braver, Co-PIs)