Neuroscience of Attention

This chapter explores “predictive coding” models, which challenge classic theories of perception and brain function. By incorporating details of both the connectivity between brain areas and the levels of microcircuitry within cortical regions, these models suggest a radical new way to conceive of perception and cognition. Whereas classic models assume that feed-forward, or bottom-up, processing is mainly responsible for our perception, predictive coding theories suggest that top-down models determine our perception, with bottom-up processing simply correcting errors in those models. Neuroscience evidence is presented for the abundance of top-down connections, the efficiency of neural coding, the role of expectancy in attention, and how the balance of top-down and bottom-up processing is related to the dysfunctional attention processes in some clinical groups. The allocation of attention is thought to be a dynamic and changing process wherein top-down hyper-priors are integrated with current priors that are being continually updated within and across levels. According to such models, attention affects the expected precision (reliability) of bottom-up information and the likelihood that this information will be used to update the current top-down models. Predictive coding theories that are opening new ways of thinking about the neural mechanisms that drive our attention are discussed.