Eyeing Uncertain Rewards: Pupil diameter tracks task-related arousal and error feedback in voluntary task-switching

Abstract

The brain's control mechanisms monitor for changes in environment or internal goals and adjust accordingly. When environments are stable and rewards are stable, control mechanisms favor stability. When environments and rewards change, then control must be increased on the current goal to avoid distraction or update to allow for flexibility. When new information is presented (e.g. a reward to be won) or the environment changes, the brain must flexibly adapt to this change. When the environment changes and no longer favors current goals, new goals are selected. Voluntary task-switching (VTS) is an experimental paradigm that investigates the components underlying cognitive flexibility. The majority of VTS studies have examined behavior using consistent reward schemes. In this paper, we examine the effects of uncertain reward as revealed by eye-tracking. Previous work shows pupillary measurements correspond to task-related cognitive effort. Additionally, previous studies suggest both pupil dilation and switch rates increase when reward increased relative to previous than when it remained low. We investigated dynamics of reward (reward transition) and found target-locked phasic pupil dilation was highest when reward increased relative to previous. We also investigated the effects of task alternation and found that target-locked pupil dilation increased for switch trials than repeat trials, suggesting greater effort expenditure. Although we found no effect of reward transition on voluntary switch rates, we found increased reward is associated with greater flexibility and decreased reward associated with decreased flexibility, echoing previous findings. Our results support evidence that pupillary dynamics reveal motivational dynamics (possibly mediated by LC-NE activity). There was no evidence for reward modulation of gaze activity, neither did reward influence dwell time on the target stimuli compared to the irrelevant stimuli. While, we did not observe a prediction error, dilation was increased when participants saw error feedback. Taken together, results show pupil dilation registers anticipation of uncertain reward, tracks task-related effort, and generates a signal of surprise to error feedback. These results reveal the arousal and attentional mechanisms, known to be taxed in voluntary task-switching.