Similar pre-attentive processes are often thought to underlie rapid texture segmentation and target ‘pop-out’ in multi-element displays (but see Wolfe, 1992 Vision Research 32 757 – 763). Performance in target-detection and texture-segmentation tasks was measured here for briefly presented displays of curved-line elements. In both tasks 49 curved-line elements, each subtending 1 deg of visual angle, were presented in a circular display for 100 ms and followed by a mask. The position of each element in the array was jittered to reduce any possible collinearity or luminance cues. In the target-detection task, observers determined whether the display contained a target which differed in curvature from the other, background elements. In the texture-segmentation task, observers determined the orientation, horizontal or vertical, of a foreground region of 4 × 2 elements which differed in curvature from the background elements. Performance, quantified as percent correct, was measured as a function of target (or foreground) and background curvatures. At small background curvatures, performance in the two tasks was very similar: performance was best when target or foreground curvature was large. Performance differed, however, at large background curvatures: for texture segmentation there was a marked peak in performance when foreground curvature was close to zero, but there was no corresponding peak for target detection. It seems that some additional, global cue can be extracted from a group of straight or slightly curved lines that is not available from a single line, thereby facilitating texture segmentation but not target detection.
