A linear systems approach to the detection of both abrupt and smooth spatial variations in orientation-defined textures

Abstract

Two distinct paradigms have characterized most previous studies of texture perception: one has dealt with texture segregation, the other with the processing of texture gradients. Typically, studies of texture segregation have used stimuli with abrupt textural variations, whereas studies of texture gradient processing have used stimuli with smooth textural variations. In this study we have asked whether the mechanisms which process abrupt and smooth textural variations are the same, by considering whether a simple linear model can account for the detection of orientation modulation in micropattern-based textures with three types of modulation: sine-wave (SN), square-wave (SQ) and missing fundamental (MF). The MF waveform was constructed by removing the fundamental harmonic from a square-wave. We found a clear overall ordering of sensitivity: SQ > SN > MF. We found that sensitivity to the SQ and MF stimuli could be predicted very well from the SN data if one assumed that the r.m.s. output of a single linear channel underlay the detection of the orientation modulation. This suggests that the detection of both abrupt and smooth changes in orientation-defined textures is subserved by a common mechanism which mimics the operation of a single linear channel.