Abstract
Motivated by perception of reddish green and bluish yellow in retinally stabilized equiluminant images, Billock et al. (2001) created a softwired disruptable winner-take-all (WTA) model of cortical color opponency. This toy model was grounded in population dynamics and had variables representing neural response strength. A more realistic neural model is desirable. Wilson's model of spiking cortical neurons was employed and the simulated cortical neurons were coupled using inhibitory synapses and were driven by wavelength-selective geniculate cell spike rates from DeValois et al. Two candidate spiking WTA networks had different dynamic characteristics - each has potential applications to neural modeling. The WTA networks produce r-g and y-b opponent spike rates, which are converted into perceptual responses using Naka-Rushton-like functions derived from DeValois et al. and behavioral data. The resultant sensitivity-based opponent responses resemble Werner & Wooten's opponent average observer data.
Meeting abstract presented at the 2016 OSA Fall Vision Meeting