1The brain uses many sources of information to determine distance and 3D shape.2Oculomotor cues to depth. There are two hypotheses about how these circuits work: inflow hypothesis (A) and outflow hypothesis (B).34Occlusion cues. T intersections play an important role in depth perception from occlusion.5Perspective (texture gradient) cues.6More perspective cue examples.Perspective and the vanishing point.78Shading cues to depth and shape9Flipping the previous slide over shows that the brain tends to interpret the shading cues as if the light is coming from above.10Almost all the shape seen in sand dunes (and in faces) is due to shading/shadows. Sand in the desert (and skin on a face) is fairly homogeneous material.11Depth from shadows (Daniel Kersten Lab)12Depth from shadows (Daniel Kersten Lab)1314Blur and visibility cues15The familiarity with the relative sizes of hands and heads can be used for distance judgments.16Familiarity, perspective, occlusion, and other cues.17Using natural scene statistics to optimally solve the stereo correspondence problem.18192122Binocularity cues to depth and shape. When the eyes are fixated on the green square, objects at other distances fall on non-corresponding locations in the two eyes. The brain uses those disparities to judge relative depth.23The Veith-Muller circle shows the points in space that fall on corresponding locations in the two eyes. P, Q and R project to corresponding locations.24A simple stereogram.25Random dot stereograms are useful for studying the disparity cue in isolation. They show that depth from disparity is not a matter of recognizing objects in each eye and then comparing the positions of those objects. In fact, objects can be recognized from disparity information alone.26Sharp edges are not required for seeing depth from binocular disparity.Binocular receptive fields are receptive fields measured for both the left (L) and right (R) eye. Sometimes the shape is nearly the same in the two eyes sometimes it is different. Neurons that have the same shape receptive field in both eyes will respond best to zero disparity (if the receptive fields are in the same location). Binocular receptive fields can differ in shape, or position, or both between the two eyes.2728Types of binocular disparity turning functions measured in V1 neurons.2930Disparity thresholds for different spatial frequency targets. For medium and high spatial frequencies disparity thresholds are small (down to 20 sec of arc or less).31Geometry of binocular disparity. The disparity of a point (black circles) from the fixation (convergence) point (open circle) is an “absolute disparity.” The disparity between the two points is a “relative disparity.” Neurons in V1 appear to be primarily sensitive to absolute disparity. Relative disparity is more general (i.e., independent of convergence point), and apparently is computed in higher brain areas (e.g., V2).32Humans are very sensitive to relative disparity, but sensitivity does drop off when the baseline disparity is from the point of fixation (convergence point).33Apparent depth as a function of disparity. The black dots are disparities where there is no diplopia (double vision). The diagonal line shows the locus of veridical (accurate) depth perception.34Demonstration of binocular rivalry.35Dr. Cormack’s lab has shown that the typical range of disparities in the natural environment is modest (-1.5 deg to 1.5 deg), and corresponds quite well to the range of disparity preferences that have been measured for disparity selective neurons in a cortical area known to process disparity and motion information.36Humans and macaques can detect very small disparities: down to 20 sec of arc or less. This high stereo acuity provides very good depth information at close distances. The blue line shows the smallest detectable difference in distance as function of total distance from the eyes. The red line shows the difference in distance above which diplopia (double vision) occurs.373839404142Motion cues to distance. When fixating the horizon while translating in a car or train, the nearer an object the faster it moves across the retina. When fixating a closer object while translating, the further an object from the fixated object the faster it moves across the