Updating our theories of perceiving: from predictive processing to radical enactivism

Radically enactive accounts of perceiving directly and diametrically oppose their representationalist rivals. This is true even of the most radical predictive processing theories of perception which embrace some enactivist assumptions yet retain some commitment to representationalism. Which framework should we prefer? This chapter seeks to make headway on this question by focusing on the special explanatory challenge that a certain class of perceptual illusions poses to predictive processing theories of perception. The perceptual illusions in question, of which the Müller-Lyer is the paradigm, reveal that what we see can systematically fail to update in light of what we know. We review and reject two prominent PP attempts to address this challenge—one conservative, one radical. We find both kinds of PP proposal wanting, for different reasons. In the end, we propose an alternative, simpler radical enactive, RE, explanation of the full pattern of effects of perceptual illusions: it is that our basic modes of perceiving take the form of contentless, non-inferential habits that are distinct from, and come before and below, our capacities for contentful perceptual judgement. We give reasons for thinking that this RE proposal can adequately and elegantly account for the full set of empirical findings about our patterns of response to perceptual illusions of the sort under scrutiny in this chapter.