Proceedings of the Annual Meeting of the Cognitive Science Society

Decades of research have established that while people’s performance suffers when they need to quickly switch between tasks, they can reduce these performance costs the more time they have to prepare. Two major theories have attempted to explain how people actively prepare for tasks over time, debating whether these task state transitions are discrete or gradual. We attempted to bring clarity to this debate by developing new statistical methods for single-trial modeling of task state transitions, which we use in a task that combines the strengths of cued and predictable task-switching. We found that participants’ behavior was best explained as a hybrid between discrete and gradual transitions. Over the preparation period, participants discretely transitioned from an unprepared state into a dynamic, increasingly prepared state. These findings provide a new account of cognitive flexibility, paving the way for mechanistic models of task-switching.