Ali M Caron¹, Edward F Ester¹; ¹University of Nevada, Reno

Categorization describes the process of assigning unique and behaviorally relevant labels to stimuli. Invasive electrophysiological recordings in non-human primates have demonstrated that retinotopically organized brain areas including the superior colliculus, lateral intraparietal area, and frontal eye fields exhibit robust encoding of learned visual categories and are causally involved in abstract category decisions. These same regions are causally involved in the generation and control of voluntary and involuntary eye movements (e.g., microsaccades), raising the possibility that oculomotor behavior can be used to track abstract category decisions. We tested this possibility by tracking gaze position after human volunteers had learned to classify sets of continuous orientation stimuli into discrete groups based on an arbitrary, experimenter-imposed boundary. Importantly, to-be-categorized stimuli were presented foveally and dynamically to minimize voluntary eye movements. In Experiment 1 (N = 25 ) volunteers reported the category of to-be-classified stimuli as quickly and as accurately as possible. Category information could be decoded from gaze coordinates beginning ~250 ms after stimulus onset, with decoding performance peaking immediately before participants’ behavioral responses. Participants in Experiment 2 (N = 24) performed a delayed-match-to-category task that decoupled categorization from response selection by requiring participants to compare the category membership of sample and test stimuli separated by a blank interval. The category membership of the sample stimulus could be decoded from participants’ gaze coordinates beginning ~250 ms after its appearance, and decoding performance remained well above chance across the ensuing delay period. These results demonstrate that human oculomotor behavior can be used to reliably track abstract category decisions.

Acknowledgements: Funding: National Science Foundation Grant 2050833