Anne Sutter

Spatial frequency channels and perceptual grouping in texture segregation

The literature suggests that both spatial frequency components and grouping processes affect texture segregation. The experiments we report investigated texture segregation in a three-part (tripartite) pattern in which each part contains equal numbers of two different squares that are arranged in vertical stripes in the top and bottom parts and in a checkerboard in the center part. We found: (1) A difference in sign of contrast yields strong texture segregation. (2) Texture segregation is a negatively accelerated function of the ratio of the Rayleigh contrasts of the low-luminance and high-luminance squares. (3) Size and contrast are not independent attributes but can cancel each other. High spatial frequency differences are not sufficient to cause segregation in the presence of contradictory information from the low spatial frequency channels. (4) Hue is a weak feature relative to contrast. Hue differences are not sufficient to cause texture segregation in the presence of contradictory contrast information. The results of the experiments support the argument that the higher order processes in texture segregation have access to information corresponding to the outputs of the spatial frequency channels. They are qualitatively consistent with the hypothesis that tripartite segregation is primarily a function of spatial frequency components and not of grouping processes.

Contrast and spatial variables in texture segregation: Testing a simple spatial-frequency channels model

Observers were shown patterns composed of two textures in which each texture contained two types of elements. The elements were arranged in a striped pattern in the top and bottom regions and in a checked pattern in the center region. Observers rated the degree to which the three regions were seen as distinct. When the elements were squares or lines, perceived segregation resulting from differences in element size could be canceled by differences in element contrast. Minimal perceived segregation occurred when the products of the area and the contrast (areal contrasts) of the elements were equal. This dependence of perceived segregation on the areal contrasts of the elements is consistent with a simple model based on the hypothesis that the perceived segregation of the regions is a function of their differential stimulation of spatial-frequency channels. Two aspects of the data were not consistent with quantitative predictions of the model. First, as the size difference between the large and small elements increased, the ratings at the point of minimum perceived segregation increased. Second, some effects of changing the fundamental frequency of the textures were not predicted by the model. These discrepancies may be explained by a more complex model in which a rectification or similar nonlinearity occurs between two stages of orientation- and spatial-frequency-selective linear filters.

Investigating simple and complex mechanisms in texture segregation using the speed-accuracy tradeoff method.

Several recent models of texture segregation have proposed two mechanisms: simple, linear channels (first-order, Fourier mechanisms) and complex channels (second-order, non-Fourier mechanisms). We used the speed-accuracy tradeoff (SAT) method to examine the time course of texture segregation processing in simple and complex channels. The stimuli were texture patterns designed to segregate primarily as a result of activity in one set of channels or the other. We presented subjects with textures that were checked or striped arrangements of either Gaussian-blob or Gabor-patch elements. Subjects were required to identify the orientation of a rectangular texture region embedded in a background field of a different texture. A range of contrasts and a control task were used to equate visibility of the Gabor and Gaussian textures. SAT functions were obtained by requiring subjects to respond within 200 msec after an auditory cue. We found that when segregation depended primarily on simple channels, performance was faster than when it depended primarily on complex channels: the 75% correct level was reached 100-200 msec sooner and this extra speed was reflected both in smaller delay and higher rate parameters.

Lightness differences and the perceived segregation of regions and populations

A striking finding reported by Beck, Sutter, and Ivry (1987) was that, in textures composed of regions differentiated by the arrangement (checks and stripes) of two texture elements (light and dark squares), a large lightness difference between the squares could fail to yield segregation between the regions, whereas a smaller lightness difference could sometimes yield strong segregation. In the experiments reported here, we compared the segregation of striped and checked arrangements of light and dark squares into regions with the segregation of two randomly interspersed populations of light and dark squares into subpopuiations. Perceived lightnesses are the same for a given set of squares, whether they are arranged in regions or in intermixed populations. Perceived population segregation is approximately a single-valued function of the lightness differences of the squares, but perceived region segregation is not. The reason for the difference between population segregation and region segregation may be that region segregation is mediated by detectors’ having large oriented receptive fields (large bar detectors) that are sensitive to the fundamental spatial frequency and orientation of the texture region as defined by the arrangement of the squares (Beck et al., 1987; Sutter, Beck, & Graham, 1989). These detectors cannot be responsible for population segregation, because the light and dark squares are distributed randomly throughout these patterns and therefore do not define a consistent arrangement of any particular spatial frequency or orientation. The light and dark squares in the population patterns fall equally on excitatory and inhibitory regions of large bar detectors. A plausible explanation for population segregation is to suppose that the segregation is the result of similarity grouping of the light and dark squares.

Effect of spatial scale and background luminance on the intensive and spatial nonlinearities in texture segregation

Perceived segregation between element-arrangement textures is affected both by spatial scale and background luminance. The effects on the spatial nonlinearity are consistent with the proposed structure for complex (second-order) channels. The effects on the intensive nonlinearity are not consistent with an early, local nonlinearity but are consistent with either (i) a relatively early, local, nonlinearity occurring before the spatial frequency channels but after a sensitivity-setting stage, or (ii) inhibitory interaction among channels modeled as a normalization network. Thus the texture intensive nonlinearity comes after sensitivity to spatial frequency and background luminance has been determined. For six of seven observers, the texture intensive nonlinearity was compressive by 10% contrast for both increments and decrements (at high background luminance, large spatial scale.

A comparison of the dynamics of simple (Fourier) and complex (non-Fourier) mechanisms in texture segregation.

Models of texture segregation frequently feature two processing mechanisms: simple, linear channels (1st-order, Fourier mechanisms) and complex channels (2nd-order, non-Fourier mechanisms). Using texture patterns designed to segregate primarily as a result of activity in one set of channels or the other, we employed the method of cued response to obtain speed-accuracy tradeoff (SAT) functions measuring the time course of texture segregation processing in simple and complex channels. Here, both simple-channel and complex-channel patterns are composed of Gabor-patch texture elements, thus equating input to simple channels and the first stage of complex channels. Subjects were required to identify the orientation of a rectangular texture-region embedded in a background field of a different texture. SAT functions were obtained by requiring subjects to respond within 200 ms after an auditory cue. We found that: (1) when segregation depended primarily on activity in simple channels, performance was faster and better than when it depended primarily on complex channels; (2) in contrast to a previous study (Sutter, A., & Graham, N. (1995). Investigating simple and complex mechanisms in texture segragation using the speed-accuracy tradeoff method. Vision Research, 35, 2825-2843), simple-channel (Fourier) patterns composed of two textured regions were just as easily segregated as simple-channel patterns in which one of the regions was blank instead of textured; (3) performance with complex-channel patterns composed of diagonally oriented Gabor-patches was considerably worse than performance with complex-channel patterns composed of vertically and/or horizontally oriented Gabor-patches; (4) among simple-channel patterns containing only one region of texture (background-only or rectangle-only), there were minimal differences in performance; and (5) as in previous experiments, there were large individual differences in the segregation of complex-channel (non-Fourier) patterns. All of the above results can be explained within the framework of the simple- and complex-channels model of texture segregation.

Non-linear processes in perceived region segregation: orientation selectivity of complex channels.

Models incorporating linear spatial‐frequency‐ and orientation‐selective channels explain many aspects of visual texture segregation. The inability of such models to fully explain texture segregation results, indicates that non‐linear processes are also involved. One non‐linearity that has been suggested is complex channels consisting of two stages of linear filtering separated by a rectification‐type non‐linearity (much like cortical complex cells). Here we further demonstrate the usefulness of complex channels in explaining texture segregation results and investigate the orientation‐selectivity of the first stage of such complex channels. Our results suggest that the first stage is much more selective for orientation than are lateral geniculate nucleus cells, but that the first‐stage orientation bandwidth is rather wide with some interaction occuring between perpendicular orientations.