William Kemball-Cook¹, Sean R. O'Bryan¹, Joo-Hyun Song¹; ¹Brown University

Pupillometry is a well-established tool for studying arousal and cognitive effort. Research shows that task-evoked pupillary response amplitudes increase with visual working memory (WM) load. However, it is currently unclear whether pupil diameter (PD) is also sensitive to the type of information stored in visual WM (e.g., spatial or feature-based), which may depend on different underlying neural substrates. Here, we directly compared task-evoked PD between spatial and feature-based WM tasks using a within-subjects design to understand if responses differ between the two task types, consistent with a representation-specific view of WM. We analyzed four different PD metrics: baseline, amplitude (relative to baseline), dilation velocity, and time-to-peak. Subjects (N = 74) indicated whether a probe matched either a color or spatial location presented in the encoding display while their pupil diameter (PD) was continuously recorded. WM load was manipulated by varying the number of items in the encoding display from 3-7. Critically, the tasks were visually identical and differed solely in their instructions. Our results revealed that dilation velocity and baseline were sensitive to task type, with consistently larger responses in the feature-based task regardless of WM load. Conversely, amplitude and time-to-peak were sensitive to WM load irrespective of task type, showing larger responses with increasing set sizes. Interestingly, time-to-peak demonstrated an ability to track load demands beyond the typical 4-5 item WM capacity plateau observed with amplitude. Overall, these findings suggest that different components of the task-evoked pupillary response can be used to index representation-specific and representation-general WM processes. Moreover, these results underscore the value of analyzing the pupillary response via separable components rather than a unitary measure (e.g., baseline-corrected amplitude), providing a deeper understanding of cognitive processes involved in working memory tasks and other tasks which require cognitive control.

Acknowledgements: NSF BCS-1555006