Pupil Size Automatically Tracks and Predicts the Complexity of Visual Information

Stimulus complexity is a fundamental constraint on the human visual system to summarize scenes into manageable representations. Although the pupil is known to reflect cognitive load, its role in processing visual complexity remains largely unexplored. In four experiments exploring orientation ensemble perception, we found that pupil size serves as a dynamic indicator of stimulus complexity, which was measured by entropy. Experiment 1 showed that pupillary constriction decreased (resulting in a larger pupil) as orientation distributions became more dispersed, a change that tracked increases in behavioral adjustment errors. Experiment 2 revealed that these pupillary adjustments occurred automatically during passive viewing and correlated with individual differences in perceptual sensitivity. Experiment 3 showed that these automatic pupillary adjustments emerged even when participants never performed any task and purely passively viewed the bar arrays throughout the entire session. Finally, Experiment 4 demonstrated that the pupil adjusted predictively in response to cues signaling upcoming stimulus complexity. Collectively, our findings suggest that the pupil plays an active, predictive role in visual summarization: it dilates to gather more information from complex stimuli, thereby optimizing the encoding of summary statistics at the earliest stage of the visual hierarchy.