Week 4 Motion, Depth, Form: Cormack Wolfe Ch 6, 8 Kandell Ch 27, 28 Advanced readings: Werner and Chalupa Chs 49, 54, 57.

Week 4 Motion, Depth, Form: Cormack Wolfe Ch 6, 8 Kandell Ch 27, 28 Advanced readings: Werner and Chalupa Chs 49, 54, 57 What is V1 doing? Early idea: edge detectors basis for more complex patterns Later ( s) spatial frequency channels any spatial pattern can be composed of a sum of sinusoids Late 90s to now: Main idea about V1 is that it represents an efficient recoding of the information in the visual image. Images are not random. Random images would require point-by-point representation like a camera. Images have clusters of similar pixels and cells designed to pick this up. Cells extract information about spatial variation at different scales (clusters of different sizes). Can think of receptive fields as basis functions (an alphabet of elemental image components that capture clustering in local image regions) Approximating an image patch w basis functions V1 striate cortex LGN Thalamic nucleus The outputs of 64cells in the LGN can be coded with only twelve V1 cells where each cell has 64 synapses The neural coding library of learned RFs Because there are more than we need - Overcomplete (192 vs 64) - the number of cells that need to send spikes at any moment is Sparse (12 vs 64). Defining visual areas: Retinotopic responses Anatomical projections (cell properties) Note old simplistic view: One area, one attribute Is not true. Areas are selective in complex and poorly understood ways Note the case of Mike May. More complex analysis of image properties in higher visual areas (extra-striate) Major sub-division into dorsal and ventral pathways Mike May - world speed record for downhill skiing by a blind person. Lost vision at age 3 - scarred corneas. Optically 20/20 - functionally 20/500 (cf amblyopia) Answer to Molyneuxs question: Mike May couldnt tell difference between sphere and cube. Improved, but does it logically rather than perceptually. (cf other cases) Color: an orange thing on a basket ball court must be a ball. Motion: can detect moving objects, distinguish different speeds (structure from motion). Note: fMRI shows no activity in Infero-temporal cortex (corresponding to pattern recognition) but there is activity in MT, MST (motion areas) and V4 (color). Other parts of brain take over when a cortical area is inactive. Cannot recognize faces. (eyes, movement of mouth distracting) Cant perceive distance very well. Cant recognize perspective. No size constancy or lightness constancy/ segmentation of scene into objects, shadows difficult. Vision most useful for catching balls and finding things if he drops them. Fellerman and Van Essen 85 Hippocampus Connections are bi-directional Cells in MT are sensitive to motion in particular directions. Cells are also tuned for particular speeds MT lesions lead to deficits in motion perception Often only transient loss however Methods for measuring motion sensitivity: %motion and direction range Direction range sample randomly from directions over eg 90 deg range The Aperture Problem Cells in V1 can only detect motion orthogonal to the receptive field. Output is ambiguous. MT is thought to resolve this ambiguity by combining motion from different V1 cells. Integration of features (corners) is also used. A.When the eyes are held still, the image of a moving object traverses the retina. Information about movement depends upon sequential firing of receptors in the retina. B. When the eyes follow an object, the image of the moving object falls on one place on the retina and the information is conveyed by movement of the eyes or the head. Two ways of perceiving motion. MT MST Output of cells goes to brainstem regions controlling pursuit eye movements. Motion of the body in the world introduces characteristic motion patterns in the image. MST is sensitive to these patterns. Lesions in monkey MST lead to deficits in pursuit eye movements. Right occipito-parietal lesions in human leads to similar deficits in pursuit eye movements. Motion of animate agents Output to pursuit system Optic flow patterns dorsal ventral x.html MST has input from the vestibular system. Thus the cells have information about self motion from sources other than the flow field. Many cortical areas have inputs from eye movement signals as well, even as early as V1. Presumably this is responsible for the ability of the visual system to process image information independent of image motion on the retina. Pursuit movement combines eye and head