The neural correlates of self-regulated behavior: Endogenous task switching

In constantly changing environments, we are forced to dynamically re-organize our behavior in order to reach desired goals. Often, there is no obvious way to adapt to these environments, so different alternative courses of action have to be explored. Once a successful strategy is found, it is exploited until the environment changes and the need for behavioral adaptation arises again. This type of behavior requires endogenous and strategic switches between different behavioral sets. However, the neural implementation of these processes remains unclear. We use multivariate pattern classification of functional MRI data to investigate endogenous adaptations to external task demands in a modified task switching paradigm. More specifically, subjects were free to choose between three different task sets on a trial-by-trial basis. Difficulty varied independently in the three tasks, increasing for the chosen task set and decreasing for the non-chosen task sets. This created an exploration-exploitation dilemma in which subjects had to find the time point at which it is optimal to leave the current task set and switch to a different, possibly easier, one. Strategic choices lead to constantly low difficulty while poor choices lead to high difficulty. Therefore, to keep difficulty low, subjects were forced to constantly monitor the task environment and adapt their behavior accordingly. Behavioral results show that subjects cope efficiently with the task after some training. They track the task environment, keeping difficulty low through strategically placed switches between task sets. Crucially, potential confounds such as task preferences or learning effects do not influence choice behavior. While subjects performed the task we acquired functional MRI data. Using a time resolved decoding approach, we are able to decode whether subjects will stay in the current task or switch away from it from frontopolar cortex. This area is known to be involved in free choice processes. Furthermore, we find predictive information about decisions in posterior parietal cortex, part of a fronto-parietal network associated with cognitive control functions. These results suggest that endogenous behavioral adaptations share a common neural substrate with free choice processes and cognitive control functions - such as the preparation and reconfiguration of task sets. Additional experiments will further illuminate how neural representations of choice-relevant task features (i.e. difficulty) interact with representations of actual choices. This will help us understand how humans strategically adapt their behavior to cope with dynamically changing environments.