Abstract
A number of studies have found that the supra-threshold color responses of anomalous trichromats are stronger than predicted from the reduced spectral separation of their longer wave cone pigments. These compensatory effects could occur if post-receptoral neurons amplify their gain to discount the weaker difference signal provided by the cones. However, this compensation is typically incomplete, and the factors that limit it are not well understood. We modeled the consequences of color deficiencies and gain changes within a population code for cortical color vision to assess the impact of adaptation at different stages. P cells carrying LvsM signals are also responsive to luminance (L+M) contrast, and there is evidence from both humans and primates that they do not adapt to a color vision loss. The weaker LvsM signal in anomalous trichromacy will therefore bias the distribution of cell preferences away from the LvsM and toward the L+M axis. Gain changes occurring subsequently in the cortex could restore the average responses in individual neurons, but would not undo the biases in the population response. For example, even if cells tuned to pure LvsM signals fully recovered their response, the population as a whole would still carry weaker LvsM signals. Models of these response changes show that this can lead to partial recovery of the LvsM responses in anomalous trichromats, but also predicts increased responses to luminance contrast, consistent with improved luminance sensitivity in color deficient observers (e.g. Doron et al. 2019). Gain changes that instead occurred later, e.g. after combining signals from the LvsM and S cones, predict enhanced responses to S-cone mediated signals, which are not observed. These analyses constrain the potential sites and mechanisms of compensation for a color loss, while also pointing to testable predictions about how color signals (e.g. contrast) are represented within the population.