L and M cone input into spectral sensitivity functions: a reanalysis

In order to better understand the nature of long-wavelength (L) and middle-wavelength (M) cone input into spectral sensitivity functions and determine the reliability with which it is possible to predict L:M cone inputs, we developed analytical methods to determine confidence intervals for L:M cone input for spectral sensitivity functions or data transformed to cone-contrast space. Spectral sensitivity functions measured by direct heterochromatic brightness matches are dominated by the L/M opponent channel over most of the spectral range. For detection of large/ long test stimuli, spectral sensitivity functions show a characteristic "notch" at the adapting wavelength, with the L/M opponent channel dominating most of the spectral range. Flicker increment threshold (FIT) spectral sensitivity functions display many of the characteristics of the luminance flicker mechanism described by Stromeyer et al. (1987). [Vision Research, 27, 1113-1137]. Previous modelling of FIT spectral sensitivity functions proposed a 2:1 L:M cone input for most of testing conditions. We show that FIT spectral sensitivity functions are dominated by L cones but show L cone suppression under bright red adapting fields. For the fitted spectral sensitivity functions or simulated data sets, we found small confidence intervals for L:M cone input into the L/M opponent channel and conclude that it is possible to reliably predict L:M cone input ratios. However, for similar data sets of additive spectral sensitivity functions, we found large confidence intervals for L:M cone input ratios and conclude that it is not possible reliably predict L:M cone input into the L/M non-opponent channel using available spectral sensitivity functions.