We propose a new psychometric model incorporating noise as well as stimulus effects, based on recent findings that noise can improve human perception via a mechanism of stochastic resonance (SR). This model assumes that the psychometric function can be regarded as a bivariate function of noise and stimulus intensities. The algorithm of the PSI Bayesian adaptive estimation method is modified so that it is applicable to our new model. In computer simulations, our new procedure successfully estimates the bivariate psychometric function within a few hundred trials. We also demonstrate several examples in which the procedure is applied to actual psychophysical experiments.
展开▼
