Abstract
Perceptual decisions arise from visual input interacting with ongoing brain activity, yet the specific contribution of spontaneous brain activity to visual perceptual decision making has not been established. Here, we review two decades of work looking at how trial-to-trial variability in the amplitude of brain oscillations impacts subjective and objective aspects of visual perception. Surprisingly, computational models of perceptual decision making have only recently been applied in this context, providing an opportunity to unify decades of data into a common decision-making framework. A synthesis of this data along with new experiments from our lab reveals several novel conclusions: 1) Trial-to-trial fluctuations in the power of prestimulus alpha (~ 10 Hz) oscillations consistently modulate perceptual decisions of the same physical stimulus 2) The mechanisms by which this happens are sensory in nature, as reduced alpha leads to enhanced early visual responses 3) In detection tasks, lower prestimulus alpha power increases both hit rates and false alarm rates equally, leading to no change in sensitivity (d’). 4) In discrimination tasks lower prestimulus alpha power does not change accuracy but, 5) enhances subjective reports of confidence and visibility. We propose a detection theoretic model that can qualitatively capture all of these empirical results. The model assumes that prestimulus alpha equally impacts the distributions of sensory evidence as well as noise and that criteria for detection and confidence/visibility do not update to accommodate sensory changes. The implication of this is that subjective aspects of perception (visibility, detection, confidence) readily dissociate from objective discrimination performance (d’) because our subjective estimates do not adapt to moment-to-moment changes in alpha-related fluctuations in cortical excitability.