Abstract
A plaid is a combination of two (or more) gratings or spirals with differing orientations and contrasts. Both grating- and spiral-composed plaid patterns were adapted to investigate how the human visual system combines information from orientation-selective channels to form a coherent percept. The subthreshold summation paradigm and constant stimuli method were employed to measure the detection thresholds of the component (e.g., grating) and compound stimuli (e.g., plaid), where the components employed five different contrast-combination ratios. The component stimuli included gratings, spirals, gratings-composed plaids and spiral-composed plaids. The compound stimuli had two components that had orthogonal orientations. The results showed the detection thresholds of the compound stimuli cannot be predicted by either the probability summation process or the additive summation process for all of the stimulus patterns, regardless of whether a high threshold theory model or signal detection theory model was employed. We then developed the contrast gain control model based on the signal detection theory model to describe the psychometric function (the proportion correct against the contrast of the component or compound stimuli). The response of each oriented mechanism is the excitation of a linear filter divided by a broadband inhibitory input. The proportion correct is determined by a nonlinear combination of the responses of those mechanisms. The contrast gain control model explained the data well compared to the model without inhibition terms. In addition, for those gratings- or spiral-composed plaids as component stimuli, the psychometric functions could be described from the interactions in the gratings- or spiral-sensitive channels. Thus, a plaid pattern is mediated by a combination of orientation-selective mechanisms, and the divisive inhibition is required to describe the detection performance.