Abstract
Purpose: Texture discrimination is widely believed to rely on mechanisms that detect discontinuities within neural images (Landy & Bergen, 1991, Vision Research). An alternative view is that texture discrimination is “region based" meaning that sensitivity to a texture difference is not significantly affected by the proximity to the two textures involved (e.g., Gurnsey & Laundry, 1992, Canadian Journal of Psychology). To examine this question we compared the relative discriminability of circular and annular texture regions embedded in a larger background textures.
Method: Foreground and background textures comprised filtered noise. The filters have 1.25 octave bandwidths and centre frequencies that ranged from 3.65 cpd to 4.74 cpd in seven equal logarithmic steps. In one condition the foreground frequency was fixed at 3.65 cpd and the frequency of the background texture was varied. In a second case the frequency of the background texture was fixed at 3.65 cpd and the frequency of the foreground texture was varied. The disparate texture was contained within either a circular region or an annular region. The area of disparate texture within the circular region was 33% greater than in the annular region, whereas the annular region had 50% more boundary locations than the circular region. Nine subjects performed a 4AFC in which they were to indicate the location of the disparate region.
Results: Threshold was calculated as the frequency difference that elicited 72% correct responses. Thresholds were significantly lower for circular texture regions than for annular texture regions F(1, 8), p < 0.001. In neither case was there an asymmetry and there was no interaction of region type with foreground/background.
Conclusions: These results suggest an important role region-based processes in texture discrimination. Comparisons of texture properties appear to be performed over relatively large distances and are not restricted to the vicinity of the texture discontinuity.