
This view of perception raises several obvious issues: given a set of sensory facts, what criteria are used to pick one model over another, or to seek out that model which best explains the sense data? How do we upgrade a current model, or index to a set of models? What kinds of perceptual models do we construct? Are there generalized principles that underlie the form of these models? Under what conditions will different individuals share the same models, and hence be able to communicate intelligently with one another?
An important concept underlying several of the above issues is the notion that most of the events in the world that we recognize and understand follow laws and exhibit regularities. These regularities lie at the heart of the models we construct. Hence, using the tools of mathematics, psychophysics and computer science, the Richards laboratory explores the relation between our representations of the world and the regularities of Nature that underlie them.
Richards, W., Seung, S. and Pickard, G.(2006) Neural Voting Machines. Neural Networks 19, 1161 - 1167.
Richards, W., McKay, B. and Richards, D. (2002) The probability of collective choice with shared knowledge structures. Jr. Math. Psych. 46, 338-351.
Knill, D. and Richards, W. (Eds) (1995). Perception as Baysian Inference. Cambridge, England: Cambridge University Press.
Richards, W. (1988). Natural Computation. Cambridge: MIT Press.