Abstract
Recent research on voluntary task switching indicates that people use different strategies to manage multitasking constraints. In this study, we conducted two experiments to investigate individual task selection behavior and provide empirical evidence for two theoretically derived strategies-local and global-aimed at balancing time-related costs in a self-organized task switching paradigm. We implemented a delay for the stimulus required for task repetition (i.e., stimulus-onset asynchrony [SOA]), which increased with each consecutive repetition until a task switch reset the SOA. Thus, repeating a task required waiting, while switching incurred performance costs. We examined individual waiting times at which participants chose to switch tasks (switch SOA) and their task performance (switch costs) across conditions. Results revealed that some participants had switch SOAs similar to their switch costs, consistent with a local strategy. Others showed considerably smaller switch SOAs compared to their switch costs, aligning with the global strategy. These individuals likely accounted for the fact that task switching in this paradigm reduces waiting times in subsequent trials. Importantly, our visual observations of behavioral patterns were confirmed using the expectation-maximization method, a technique sometimes applied in machine learning, providing statistical support for the existence of these two strategies. Overall, the findings suggest that individuals differ in their preferred task selection strategies, with preferences remaining relatively stable across varying experimental conditions.