Vernier step acuity and bisection acuity for texture-defined form

Abstract

Using as the stimulus a texture pattern of short lines, we compared positional acuity thresholds for an orientation-texture-defined (OTD) boundary and a luminance-defined (LD) boundary. Texture lines had different orientations but the same luminance on either side of the OTD boundary, and different luminances but the same orientation on either side of the LD boundary. For the LD boundary, both vernier step acuity threshold and bisection acuity threshold were inversely proportional to the number of texture lines per degree (i.e., the pattern's spatial sampling frequency) over the entire 1.9–59 samples/deg frequency range investigated, though thresholds were considerably lower than the distance between adjacent lines. For the OTD boundary, both thresholds were inversely proportional to spatial sampling frequency (though thresholds were again considerably less than the distance between adjacent lines) but only for sampling frequencies below 20 samples/deg. For sampling frequencies below 20 samples/deg, the ration between positional acuity thresholds for OTD and LD boundaries was approximately constant (3.5:1 for vernier acuity and 1.4:1 for bisection acuity). As sampling frequency was increased beyond 20 samples/deg both vernier and bisection acuity thresholds for OTD boundaries rose steeply. Both thresholds fell to a minimum near 20 samples/deg. For vernier step acuity the minimum threshold was 2.3 and 2.4 min arc (two observers), and for bisection acuity 1.7 and 1.9 min arc. We propose that these minimum thresholds approach a physiological limit of positional acuity for an OTD boundary, and that the limit is determined by a balance between the progressive improvement of positional acuity caused by increasing the frequency of spatial sampling vs the progressive reduction in visibility of the OTD boundary caused by the associated reduction in the length of texture lines. These physiological limits are far higher than the corresponding limits for sharp-edged high-contrast LD targets (2–5 and 1–5 sec arc, respectively). For an OTD boundary the effect of orientation contrast on vernier step acuity threshold approximated a square root law, while the effect of orientation contrast on bisection acuity approximated a linear law. Observers can combine positional information carried by texture contrast with positional information carried by luminance contrast. As to the combination rule, our findings are consistent with probability summation between independent channels.