Masked detection of gratings: the standard model revisited

According to the standard model of masked detection of gratings, the observers's task is to decide whether the response of some critical mechanism is reliably increased by the addition of the test stimulus to the masker. To test this conception, experiments were performed in which the contrast levels in the masker alone and test-plus-masker intervals of a forced-choice trial were randomly perturbed on each presentation. The perturbations in the two intervals of a trial were either independent, positively correlated, or negatively correlated. Such perturbations had the expected effect on the detectability of 10 c/deg test on a 10 c/deg masker. However, they had no effect at all on the detectability of 2 or 8 c/deg test stimuli on the same masker. These results contradict the standard model of masking, but are compatible with the assumption that the test stimulus is detected by some changes it produces in the local spatial features of the phenomenal appearance of the masker.