Abstract

Recent developments in AI suggest that highly complex forms of prediction implicit in human cognitive and perceptual processes can be efficiently replicated in artificial systems. Within cognitive science, it has long been accepted that natural perceptual systems incorporate some form of predictive behaviour. Conventional models of perception construe such prediction to occur either at the level of inference and detection or in the learning of sensory and motor interdependencies. In this chapter I argue that prediction as construed in these artificial systems and models of perception does not capture the true anticipatory nature of natural perceptual systems. I suggest that an understanding of the anticipatory structure of perceptual systems requires a more nuanced understanding of the steps required for the evolution of natural perceptual systems.