Abstract
Understanding contrast sensitivity has been in the focus of human psychophysics for more than 30 years. This effort, despite certain success, is far from closure. We present a novel functional model that provides a uniform explanation for contrast sensitivity data in low-medium range of spatial frequencies and the whole range of temporal frequencies. It also explains contrast sensitivity results for trapezoidal and triangular profiles that have been left unaccounted by extant contrast sensitivity models. The major features of the new model are: (1) the contrast sensitivity for spatial frequencies below 3 cycle/deg and all temporal frequencies is determined by a single channel tuned to 3 cycle/deg; (2) this channel is preceded by accelerating (quadratic) nonlinearity that affects the shape of contrast sensitivity function. Computer simulations show that differences between contrast sensitivity functions measured by different authors can be explained by differences in the exponent of this nonlinearity attributable to experimental conditions. The nonlinearity precedes the spatial filters and is therefore independent of any channel uncertainty in the visual system. The accelerating nonlinearity of this model is well-supported by current neurophysiological data.