Abstract
Many standard models of color vision include a “luminance” mechanism that combines signals from the L and M cones additively (e.g., Boynton, 1979; Guth, 1991; Lennie, Pokorny, & Smith, 1993). Experiments that employ “equiluminant” stimuli generally assume linearity, explicitly or implicitly. The luminance mechanism is often studied using moderate to high frequency flicker or motion stimuli. However, brief, weak equichromatic flashes are also generally believed to be detected by this mechanism (Eskew, McLellan, & Giulianini, 1999; Krauskopf, Williams, & Heeley, 1982). The present experiments were undertaken to determine if the mechanism that detects “white” Gaussian blobs, on a gray background, combines the L and M cones linearly, using a new approach that was developed to study the linearity of mechanisms detecting S-cone isolating flashes (Giulianini & Eskew, 2007). A test flash is embedded in masking noise (flickering lines covering the test region). In different runs, the noise consists of L-cone modulations only, M cone modulations only, or the superposition of the two cone component noises (L&M noise). If the detection mechanism combines its L and M cone inputs linearly, the threshold elevation produced by the compound L&M noise is predictable from a specific combination of the elevations produced by the two component noises. Previous results with this method found that S cone mechanisms are nonlinear, but that the ‘green’ mechanism is remarkably linear. In the present study, we measured the masking effect of 8 different compound noises and their associated L and M components. The results show that the mechanism that detects achromatic flashes does not combine its L and M cone inputs linearly. This result, along with others using flicker detection (Stockman & Plummer, 2005; Stromeyer, Chaparro, Tolias, & Kronauer, 1997), calls into question the generality of the idea of equiluminance.