Purpose. We studied how phase and orientation of local elements can affect the detection of global patterns.
Methods. The local elements were Gabor patches with spatial frequencies of 4 or 6.6 cpd. Each global pattern consisted of 1-8 Gabor patches aligned vertically. In a global pattern, the local elements can be (1) collinear and in-phase with each other, (2) collinear and counter-phase to their immediate neighbors, and (3) orthogonal to the global orientation. We used a 2AFC paradigm to measure the detection thresholds for the global pattern.
Results. If local elements and global patterns had the same orientation, the detection threshold decreased as the number of local elements (n) increases from 1 to 3 with slopes of about -0.5 (4 cpd) and -0.3 (6.6 cpd). No further threshold decreases were observed as n increased from 3 to 8. In fovea, the phase configuration of the local elements had no significant effect on the detection threshold. If local elements were orthogonal to the global orientation, there was much less threshold reduction as n increased.
In 5o peripheral, the phase configuration of the local elements had strong influence on the detection threshold. Basically, the detection threshold was determined by the number of in-phase local elements.
Conclusions. We confirmed that visual system sees elongated patterns best (Polat & Tyler, 1997 ARVO). There is considerable spatial summation among local elements of the same orientations. In fovea, this summation process is insensitive to phase. There is little summation across different orientations. However, in peripheral, this summation is phase sensitive. The results suggests that the spatial summation in fovea involves a full-wave rectification process while in peripheral involves a half-wave rectification process.