Red lines denote oriented edges in a natural scene

Red lines denote oriented edges in a natural scene

The Probabilistic Brain

Our visual world is filled with patterns: the grass is green, the sky is blue, far objects appear small. What is the impact of our constant interactions with this structure, over evolutionary and behavioral timescales? My doctoral and postdoctoral research in this area is driven by the possibility that our perceptions — and misperceptions — are not the result of arbitrary and complex rules, but rather simple probabilistic inference.

One topic of particular interest is the role of our prior knowledge about these statistical trends. We expect cars to stop at red lights, and are willing to risk our lives on this expectation by walking across an intersection, conferring the ability to make rapid decisions and act upon them. To understand how perception may use prior information about the statistics of the environment, I investigated two case studies: the visual orientation of 3D surfaces and of 2D lines. In both cases we have found a strong correspondence between the environmental statistics and the prior information used by observers.

SELECT PRESS & WRITINGS

Salinas, E. Prior & prejudice. Nature Neuroscience, ”News & Views” commentary, 14:943–945. (2011) [PDF]

The Probabilistic Mind. Science News feature. 180:18. Oct. 18, 2011 (2011). [PDF]

Girshick, AR, MS Landy, EP Simoncelli. Cardinal rules: visual orientation perception reflects knowledge of environmental statistics. Nature Neuroscience, 14:926-932  (2011). [PDF][Supplement]

Girshick, AR, J Burge, G Erlikhman, MS Banks. Prior expectations in slant perception: Has the visual system internalized natural scene geometry? Journal of Vision, 8(6), 77a. (2008) [Abstract]

Burge, J, AR Girshick, MS Banks. Visual-haptic adaptation is determined by relative reliability. Journal of Neuroscience, 30(22):7714-21 (2010). [PDF]

Girshick, AR, MS Banks. Probabilistic combination of disparity and texture slant information: weighted averaging and robust estimation as optimal percepts. Journal of Vision, 9(9):8. 1-20. (2009) [PDF]

Girshick, AR. Probabilistic integration of sensory information for 3D visual surface slant perception. Ph.D. thesis, Berkeley: University of California. (2007)

Girshick, AR, Burge, J, Banks, MS. Bayesian cue combination: coupling of disparity-texture information compared to coupling of visual-haptic information. Journal of Vision, 7(9), 68a. (2007)