A key aspect of consciousness lies in its perceptual content, our internal representations of the world (and the self) that form the basis of our subjective experience. How can a resource-limited system such as the brain generate rich meaningful conscious content of a complex and ever-changing environment? A potential solution would be to exploit the autocorrelation and predictability that exists in the natural world. Indeed an emerging view within visual neuroscience is that the majority of our visual experience is comprised of summary representations, i.e. vision is chiefly statistical in nature. While most research have focused on central tendency measures (average), the sense of dispersion or diversity (variance) has been largely overlooked. However, there are reasons to suspect that variance plays an important role in vision, since it sets the range of expected stimuli and provides a measure of the consistency of sensory signals.
Our work examining the processing of visual variance aims to shed light on the dynamics of variance processing, how variance allows the brain to make inferences and modulates the interpretation of subsequent stimuli.
So far we have established that variance judgements are influenced by previous sensory history (serial dependence) and in turn shape future judgements, and that this influence depends on the precision (reliability) of the sensory signal. In addition we have found evidence of adaptation after-effects concerning variance, indicating an efficient use of the neural code applied to this statistic. Further work will continue to explore how computations of variance aid in constructing visual awareness in the brain.