Abstract
Detection of global organization in an array of line segments is a measure of extrastriate integration of pattern information. Coherence threholds can be tested for different complexities of pattern organization; detection of parallel texture requires simpler, more local mechanisms than for concentric configuration. Measures of spatial integration (Braddick et al ECVP 2002) show larger integration area for the latter, consistent with use of higher-level receptive fields. Here we address whether pattern complexity leads to differences in temporal integration. Stimuli were arrays of line segments. In a central circular region, a variable % of segments were aligned either parallel to a common orientation or to concentric circles. Subjects had to discriminate between 0% and other coherence values, in a 2AFC task. Processing time was limited to display intervals varying between 27 and 1067 ms, with a random mask preceding and following each interval. Coherence thresholds for each duration were determined from an adaptive sequence to determine threshold, and critical duration estimated from fitting a double linear function. Critical duration for detecting the coherence of concentric patterns (mean = 196 ms) was consistently shorter than for parallel patterns (mean = 299 ms). Thus, despite requiring more complex spatial operations, information for detecting concentric organization is integrated over a shorter temporal period than that for parallel organization. There must exist a rapid pathway for processing concentric patterns, perhaps representing an intermediate stage in global processing of specific forms including faces. The slower integration of parallel pattern information suggests that this cannot simply reflect a lower stage in a pattern-processing hierarchy. These results will be discussed in relation to the brain areas activated by pattern coherence, as indicated from functional imaging experiments.
Medical Research Council grant G7908507