How Would the World Look if It Looked as if It were Encoded as an Intertwined Set of Probability Density Distributions?

ogy. Leaving this difficult issue aside, now consider the psychological evidence that suggests vision is indeterminate, evidence which may fit nicely with the suggestion that visual processing uses probabilistic coding. Two lines of empirical evidence show that visual perception is an ongoing process which involves repeated sampling of the environment. Such a structure to visual perception fits naturally with a probabilistic interpretation: we take repeated samples in order to update our best estimate of the way the world is. First, our experience of the visual periphery is highly indeterminate (Cohen and Dennett, 2011). We are able to experience parts of the world in a determinate manner through our ability continuously to gain different perspectives through action (Findlay and Gilchrist, 2003). Second, both inattentional blindness (Mack and Rock, 1998) and change blindness (Rensink et al., 1997) paradigms suggest that our experience of the world can lack basic details. Both of these lines of evidence fit nicely with an understanding of visual experience as probabilistic. The indeterminacy of the visual periphery can be interpreted as a probabilistic representational format. Similarly, the missing details as revealed in inattentional and change blindness experiments reveal that our generative models are more successful with the gist of a visual scene and offer only vague estimates about the details. In short, both the general a priori structure of perception and recent evidence in perceptual psychology converge on the theme that visual experience involves indeterminate implicit anticipations. This theme fits quite well with the suggestion, explored by Clark, that the brain actively predicts A commentary on