Abstract
Visual perception, experienced as a continuous flow, counterintuitively seems to rely on a discrete sampling mechanism. Accumulating evidence suggests that this sampling operates in an oscillatory fashion at a rate of ~ 10Hz (alpha rhythm), such that “perceptual snapshots” are taken at favorable phases of the rhythm. Less is known about the unfavorable phases of this rhythm: do they constitute gaps during which no information is sampled at all, or is information sampled with insufficient precision for the task at hand? Both alternatives lead to poor performance in simple detection and discrimination tasks, making them impossible to distinguish. By contrast, using continuous report paradigms allows estimating independent model parameters for the amount of guessing and precision. We used such a continuous report task in combination with Posner-like, exogenous (i.e. involuntary) attentional cueing. On each trial, participants reported the orientation of a Landolt ring target that was briefly flashed at one of 20 cue-target intervals spaced in steps of 41ms, covering a time window from 192 to 983ms. 3840 trials were collected for each participant (N=14) across 8 sessions. The distribution of errors in participants’ reports was fitted with Standard Mixture Models separately for each cue-target interval and separately for validly (target at the attended location) and invalidly cued trials (target at the unattended location, thus requiring reorienting). The resulting time courses of the model’s guessing and precision parameters were analyzed with a Fast Fourier Transform to identify significant rhythms. We found that in invalid trials, which required attentional reorienting, precision was modulated rhythmically at the alpha frequency (9.6Hz). This result indicates that perceptual rhythms reflect fluctuations in precision rather than a succession of snapshots and gaps.