Feature-based attention modulates pupil responses by target similarity in a rapid dynamic attention task

Pupil size is modulated by many factors including cognitive (e.g. mental effort, arousal) and non-cognitive influences (e.g. luminance). In the present study, we investigated the extent to which feature-based attention modulates pupillary responses to oriented Gabor stimuli as a function of spatial attention and target similarity. This task rapidly presented two continuous RSVP streams of target and non-target elements (1 Hz) simultaneously to the left and right of fixation. Subjects (n=39) were cued to pay attention to one of the streams at the beginning of the 4-minute task while ignoring the other stream. Target cues were presented at random intervals (every 4-16 sec) to either shift attention to the other stream (90 deg. targets) or stay (0 deg. targets). Subjects maintained fixation and responded to shift/stay cues with a keypress to track which stream (left or right) they were attending. Distractor stimuli varied according to 4 levels of similarity to targets: 6, 19, 32, and 45 deg difference from targets. We found that target stimuli in the attended stream evoked a canonical-shaped pupil dilation response that peaked between 1-2 sec post stimulus onset with a mean amplitude of 2% signal change from baseline. Targets in the non-attended stream also evoked a pupil response that was about half strength (<1% signal change). Importantly, we found evidence for feature-based modulation of pupil responses to non-target stimuli as a function of target similarity, but only in the attended stream. Attention induced a significant dilation response (0.25%) for near-targets (difference <= 6 deg), and constriction (-0.25%) for far-targets (difference > 6 deg), in comparison to stimuli in the non-attended stream, reflecting a suppression of far-targets and sharpening of the tuning-curve profile. These results demonstrate that even in a rapid, dynamic task scenario, pupillary responses reflect systematic modulations of feature-based attention on visual processing.