Abstract
Two classes of models have been proposed to account for the processing of texture modulations. In one class of model abstract representations such as feature or texton maps are assumed to underlie detection. In the second class of model texture modulations are processed by an energy mechanism which compares the spectral energy within narrow orientation and spatial frequency bands between different texture regions. Our textures contained concurrent modulations of orientation and spatial frequency. Through adaptation we determined the orientation and spatial frequency tuning of the luminance filters involved in their detection. Results indicate that such concurrent modulations in texture are processed jointly by mechanisms that employ luminance filters tuned to orientations and spatial frequencies at which the different texture regions differ most in their spectral energy. These orientations and spatial frequencies do not correspond to those predicted by feature or texton processing models.