Abstract
Spectral sensitivity for brightness matching is broader than for luminance, a mismatch between appearance and performance. Suprathreshold chromatic brightness is thought to be a nonlinear weighted combination of achromatic (luminance) and chromatic responses; chromatic channels supplement luminance spectral sensitivity at short and long wavelengths, but have less effect on sensitivity at moderate wavelengths. This entirely reasonable theory is likely incorrect. If chromatic brightness spectral sensitivity is plotted as a function of luminance spectral sensitivity (indexed by wavelength), it is a beautiful power law. For 2 degree data, the exponent of this function is about 0.86, the amplification constant is about 1.20 and r^2 is 0.99. Since chromatic channels play no part in this calculation, and there is little variance unaccounted for, there is little scope for chromatic channels to contribute to chromatic brightness. Instead, chromatic brightness resembles a chromatically-gated amplification of the luminance response. The nonlinear nature of this amplification is that weak responses are amplified relatively more than strong ones, and these weak/amplifiable responses are at the spectral extremes. There is ample precedent for this in sensory integration: brightness perception of dim lights is preferentially amplified by auditory modulation and some cells in cat visual cortex amplify their firing rates to weakly effective lights in the presence of auditory stimulation (even though these neurons do not respond to auditory stimulation alone). In both the psychophysical and electrophysiological sensory integration, I find that the amplifying function closely resembles the power law followed by chromatic brightness. Funded by NSF#1456650.