This means that the preferred and anti-preferred disparities (the stimulus disparities creating the largest, and smallest, responses respectively) are switched, and thus a different stimulus disparity is required to produce the maximum response for correlated and anticorrelated stimuli (see Fig 3c & 3d respectively). The binocular energy model predicts that the disparity tuning function should be inverted in ACRDS in comparison with CRDS. In an ACRDS, the contrast polarity of elements is reversed between the two eyes, such that bright pixels in the left eye are dark in the right eye, and vice versa ( Fig 3b). The tuning function is inverted and now has a preferred disparity at 5 arc minutes.Ī variation on these stimuli, known as anticorrelated random dot stereograms (ACRDS), can be used to test models of binocular vision. This tuning function has a preferred disparity at 10 arc minutes (d) the same first order binocular energy tuning function in response to an anticorrelated random dot stereogram. (c) an example first order binocular energy tuning function in response to a correlated random dot stereogram. In contrast to the correlated target in (a), the target circle contains anticorrelated (reversed polarity) dots in the images shown to the left and right eye. (b) Image pair for an anticorrelated random dot stereogram, where the outer annulus contains correlated (matching) dots in the images shown to the left and right eyes. (a) Image pair for a correlated random dot stereogram, where the outer annulus and the inner circle (the detection target) contain correlated (matching) dots in the images shown to the left and right eyes. Binocular cells in the visual cortex respond to CRDS with clearly-defined disparity tuning. Humans and many other species perceive depth consistent with the disparities present in the stimuli. In a classic, correlated random dot stereogram (CRDS), the positions of some of these dots will be shifted between the two eyes to create non-zero disparities ( Fig 3a), which are either ‘crossed’ (consistent with near depth) or ‘uncrossed’ (consistent with far depth). These consist of a stereo-pair of patterns of randomly positioned dots and allow us to isolate binocular cues to depth in a laboratory setting. Our understanding of how populations of neurons modelled in this way give rise to the perception of depth has been greatly enhanced by the use of random dot stereograms (RDS). This forms the input to the second order filters (blue pathway) where the pairs of monocular filters are summed, squared and combined to calculate the second-order binocular energy response. The first-order responses are squared and summed, to produce monocular energy responses for each eye. Outline of a second order binocular energy model (green and blue pathway): Odd and even monocular energy responses are calculated separately for each eye (green pathway). The squared outputs are summed across the two halves of the quadrature pair. Responses are summed across the two eyes for corresponding filters, then squared. Outline of a first order binocular energy model (orange pathway): Odd and even symmetric monocular receptive fields (white areas represent excitatory regions of the receptive fields and dark areas inhibitory regions) are combined to compute the first order binocular energy response. This broad distribution of binocular processing reflects the importance of three-dimensional information for both action and perception, and demonstrates the complexity of extracting depth from binocular signals. At higher levels of visual processing, binocular responses are found in both the dorsal and ventral streams. ĭisparity-tuned neurons in the primary visual cortex form the early stages of a broadly distributed network of binocular processing. Point 1 will appear closer and 3 further away. This uncrossed disparity arises as this point is further away than the fixation point. Conversely, there is a rightward shift of point 3 from the left to the right eye’s image. This crossed disparity arises as this point in the image is closer than the fixation point. For point 1, there is a small shift leftwards in the right eye’s image in comparison to point 1 in the left eye’s image. The point of fixation falls on the same retinal location and has zero disparity. Each eye, the left (top) and right (bottom), receives a different image depending on fixation (point 2).

5d stereogram