Vision as Bayesian inference: analysis by synthesis?

Alan Yuille, Daniel Kersten

Research output: Contribution to journalArticlepeer-review

387 Scopus citations

Abstract

We argue that the study of human vision should be aimed at determining how humans perform natural tasks with natural images. Attempts to understand the phenomenology of vision from artificial stimuli, although worthwhile as a starting point, can lead to faulty generalizations about visual systems, because of the enormous complexity of natural images. Dealing with this complexity is daunting, but Bayesian inference on structured probability distributions offers the ability to design theories of vision that can deal with the complexity of natural images, and that use 'analysis by synthesis' strategies with intriguing similarities to the brain. We examine these strategies using recent examples from computer vision, and outline some important imlications for cognitive science.

Original languageEnglish (US)
Pages (from-to)301-308
Number of pages8
JournalTrends in Cognitive Sciences
Volume10
Issue number7
DOIs
StatePublished - Jul 2006

Bibliographical note

Funding Information:
We would like to acknowledge helpful feedback from the reviewers and the TICS editors. This work was supported by ONR N00014-05-1-0124 and NIH RO1 EY015261.

Fingerprint Dive into the research topics of 'Vision as Bayesian inference: analysis by synthesis?'. Together they form a unique fingerprint.

Cite this