Keywords
lightness illusions, ON and OFF cells, fixational eye movements, simultaneous contrast, perceptual fading
Abstract
A neural model of lightness computation driven by fixational eye movements is described and used to simulate various lightness phenomenon, including the Staircase Gelb illusion and its variants, simultaneous contrast, the Chevreul illusion, and perceptual fading of stabilized images. The model provides a precise account of the lightness matches from several experiments, with an overall error of only 1.5%. In the model, spatial maps of transient ON and OFF cell activations—produced as the eyes traverse the visual scene—are sorted by eye movement direction in visual cortex. At a subsequent processing stage, the activations within these maps are summed across space and eye movement directions by hypothesized ON and OFF spatial integrator cells, whose outputs are differenced, then network-normalized, to construct a neural map of the perceived reflectances of the surfaces within the visual scene. Finally, the reflectance map undergoes a leaky temporal integration, which results in a lightness model in which the percept depends on an average of the netword computations generated across many individual eye movements. A key model assumption is that ON and OFF cells have different neural gains: smaller for ON cells than for OFF cells. When combined with a spatial integration mechanism that takes into account realistic assumptions regarding cortical magnification, the ON/OFF gain asymmetry leads to a model in which the perceived dynamic range of the surfaces lightnesses depends on both the contrast polarities and spatial arrangement of local contrast in the image in a manner that closely mimics perceptual data.
Start Date
15-5-2024 11:00 AM
End Date
15-5-2024 12:00 PM
Included in
Computational Engineering Commons, Computational Neuroscience Commons, Psychological Phenomena and Processes Commons
Explaining the Staircase Gelb illusion, simultaneous contrast, and perceptual fading of stabilized images with a neural model driven by fixational eye movements
A neural model of lightness computation driven by fixational eye movements is described and used to simulate various lightness phenomenon, including the Staircase Gelb illusion and its variants, simultaneous contrast, the Chevreul illusion, and perceptual fading of stabilized images. The model provides a precise account of the lightness matches from several experiments, with an overall error of only 1.5%. In the model, spatial maps of transient ON and OFF cell activations—produced as the eyes traverse the visual scene—are sorted by eye movement direction in visual cortex. At a subsequent processing stage, the activations within these maps are summed across space and eye movement directions by hypothesized ON and OFF spatial integrator cells, whose outputs are differenced, then network-normalized, to construct a neural map of the perceived reflectances of the surfaces within the visual scene. Finally, the reflectance map undergoes a leaky temporal integration, which results in a lightness model in which the percept depends on an average of the netword computations generated across many individual eye movements. A key model assumption is that ON and OFF cells have different neural gains: smaller for ON cells than for OFF cells. When combined with a spatial integration mechanism that takes into account realistic assumptions regarding cortical magnification, the ON/OFF gain asymmetry leads to a model in which the perceived dynamic range of the surfaces lightnesses depends on both the contrast polarities and spatial arrangement of local contrast in the image in a manner that closely mimics perceptual data.