Abstract
Purpose: Previous studies on the role of spatial phase in perceptual grouping may be complicated by the co-varying of various local features. Here we investigated effects of phase coherence on the detection of contour deformation defined by the changes of local position or orientation features. Methods: Stimuli were linear and circular contours composed of Gabor patches with carrier frequency of 3 cpd. Contour deformation was introduced by sinusoidally modulating a vertical line or the radius of a circle. Position-defined deformation (P-Pattern) was created by placing patches at peaks and troughs of the modulation, so that changing modulation only changed their positions but not orientations. Orientation-defined deformation (O-Pattern) was created by placing patches at zero-crossings. Circular contours had 1.5 deg mean radius and 4 cycles of modulation; Compatible stimulus parameters were used for linear contours. Deformation thresholds were measured for 3 carrier phase manipulations: in-phase, phase reversal of alternating patches, and random phase. Results: Subjects showed higher sensitivity for detecting O-Patterns than P-Patterns. The mean thresholds of 3 normal subjects were 0.45±0.07% and 0.62±0.29%, respectively, for detecting in-phase linear and circular O-Patterns, and 0.77±0.36(SD)% and 0.97±0.54%, respectively, for detecting in-phase linear and circular P-Patterns. Phase reversal significantly elevated the threshold for detecting linear P-Patterns by 40% (p < 0.02). Randomizing the phase further disrupted the detection of linear P-Patterns, with threshold increased by 60% (p < 0.011). In comparison, thresholds for detecting both linear and circular O-Patterns were not affected by either phase reversal or random phase. Conclusion: While local spatial phase change may affect the grouping of local position features, orientation linking can occur between cells with different phase sensitivities and may take place at a more global level of processing.