Poster D107, Monday, March 26, 8:00-10:00 am, Exhibit Hall C
Investigating the cost of cognitive effort
Ceyda Sayali1, David Badre1; 1Brown University
People tend to avoid cognitively effortful tasks. In general, effortful tasks tend to be associated with long response times and higher error rates. Harder tasks also tend to recruit cognitive control systems. In prior work, we associated activation of frontoparietal network (FPN) and default mode network (DMN) regions during execution of a task with avoidance of those tasks. Here, we test which regions or networks in the brain predict learning of effort costs. We parametrically manipulated the level of effort by increasing cognitive control demands across tasks. In the scanned Learning phase, participants associated virtual card “decks” with tasks of a certain effort level. In the Test phase outside the scanner, participants made selections between pairs of effort tasks. We fit a reinforcement learning model that assumes costs acquired during learning influence decisions during Test. Across alternative models, we computed the cost of effort from either the average time-on-task, error rate, or task-switching probability of an effort level. The error rate model explained effort selections better than time-on-task and task-switching probability. Consistent with prior work, FPN activity increased with increasing effort execution. Expected cost and negative prediction errors positively correlated with FPN, DMN, and bilateral caudate. These results indicate that control and reward systems of the brain track learning of effort costs.
Topic Area: THINKING: Decision making