Abstract
Individual differences in human L and M cone fundamentals occur because of differences in wavelength of photopigment peak spectral sensitivity (lambda max) and differences in photopigment optical density. Although Rayleigh matching can be used to estimate these parameters, changes in lambda max and optical density are confounded when a single reference wavelength is used (Thomas & Mollon, 2004). Here we evaluate the use of Rayleigh matching with multiple reference wavelengths to estimate these parameters. We simulated Rayleigh matches using a mixture of two narrowband primaries (560 and 670 nm peak power) and 15 narrowband reference lights (peaks between 570 and 640 nm). Mixing ratio and reference intensity at the match were determined by computing cone excitations and bringing these into agreement via an interleaved staircase procedure with simulated forced-choice redder/greener and lighter/darker comparisons. We varied cone fundamentals using the Asano et al. (2016) model of population variation in lambda max and optical density. We then fit the model parameters to best account for each simulated observer’s matches and computed the mean RMSE (20 observers) between simulated and recovered cone fundamentals. For comparison, we also found the RMSE between simulated and standard cone fundamentals. Our procedure recovered L and M cone fundamentals well. The RMSE between simulated and recovered fundamentals was 0.00046, compared to 0.0076 obtained by using the standard fundamentals. The RMSE obtained from simulations with the best single reference wavelength (620 nm) was 0.0026. Our results indicate that Rayleigh matching with multiple reference wavelengths can resolve ambiguity about L and M cone fundamentals that occurs when a single reference wavelength is used. This could improve estimation of individual observers’ L and M cone fundamentals, although experimentation that accounts for the precision of real observer matches will be required to validate the approach.