Learning Cognitive Flexibility: Neural Mechanisms of Adaptive Switch Readiness
Anthony W. Sali1, Christina Bejjani1, Tobias Egner1; 1Duke University
Individuals adapt their readiness to switch task sets (“cognitive flexibility”) according to environmental demands, such that the cost associated with switching tasks is reduced when a switch is likely (e.g. Dreisbach & Haider, 2006). However, the neural mechanisms underlying this form of cognitive control learning remain unknown. To this end, we used functional magnetic resonance imaging (fMRI) and a task-switching paradigm in which the frequency of switches varied over time, ranging from 25% to 50% to 75% switch-likelihood, to probe the neural mechanisms involved in learning to match switch-readiness to changing contexts. Behaviorally, participants demonstrated smaller task-switch costs in blocks when switching was highly likely relative to blocks when switches were rare, and a temporal difference reinforcement learning model successfully accounted for both within-subject and between-subject variability in performance. A subsequent model-based fMRI analysis revealed that activity within left inferior parietal and frontal cortex increased with the magnitude of trial-by-trial unsigned switch prediction error, thus serving as a potential neural substrate of learning. Furthermore, activity within bilateral dorsal parietal cortex was associated with signed switch prediction errors such that activity was highest when switching was the most surprising. Together, our results suggest that components of the frontoparietal control network code for moment-by-moment deviations from task-switch predictions and provide a potential substrate of learned adjustments in switch-readiness.
Topic Area: EXECUTIVE PROCESSES: Goal maintenance & switching