Please use this identifier to cite or link to this item:
|Title:||The spatial resolutions of stereo and motion perception and their neural basis|
|Abstract:||Depth perception requires finding matching features between the two eye’s images to estimate binocular disparity. This process has been successfully modelled using local cross-correlation. The model is based on the known physiology of primary visual cortex (V1) and has explained many aspects of stereo vision including why spatial stereoresolution is low compared to the resolution for luminance patterns, suggesting that the limit on spatial stereoresolution is set in V1. We predicted that this model would perform better at detecting square-wave disparity gratings, consisting of regions of locally constant disparity, than sine-waves which are slanted almost everywhere. We confirmed this through computational modelling and performed psychophysical experiments to test whether human performance followed the predictions of the model. We found that humans perform equally well with both waveforms. This contradicted the model’s predictions raising the question of whether spatial stereoresolution may not be limited in V1 after all or whether changing the model to include more of the known physiology may make it consistent with human performance. We incorporated the known size-disparity correlation into the model, giving disparity detectors with larger preferred disparities larger correlation windows, and found that this modified model explained the new human results. This provides further evidence that spatial stereoresolution is limited in V1. Based on previous evidence that MT neurons respond well to transparent motion in different depth planes we predicted that the spatial resolution of joint motion/disparity perception would be limited by the significantly larger MT receptive field sizes and therefore be much lower than the resolution for pure disparity. We tested this using a new joint motion/disparity grating, designed to require the detection of conjunctions between motion and disparity. We found little difference between the resolutions for disparity and joint gratings, contradicting our predictions and suggesting that a different area than MT was used.|
|Appears in Collections:||Institute of Neuroscience|
Files in This Item:
|Allenmark12.pdf||Thesis||3.65 MB||Adobe PDF||View/Open|
|dspacelicence.pdf||Licence||43.82 kB||Adobe PDF||View/Open|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.