Abstract
PURPOSE. Mancini et al. (2001, ARVO) measured the detectability of symmetrical patterns defined by checks that were of the same polarity (SP) or opposite polarity (OP) across a vertical axis of symmetry. Stimuli comprised black and white checks having widths of 0.037 to 0.297 degrees visual angle windowed within a circular aperture 9.5 degrees in diameter. The proportion (p) of SP checks matched across a vertical axis of symmetry ranged from 0 to 1. Patches with p = 1 are perfectly symmetrical (i.e., SP), those with p = 0 are anti-symmetrical (i.e., OP) and those with p = 0.5 are random. Results showed that SP and OP stimuli elicited similar thresholds for large checks. However, OP thresholds increased dramatically as check size was reduced, whereas SP thresholds did not. In fact, for the smallest check sizes, performance on the OP stimuli did not exceed chance even when p = 0. This suggests that the mechanism responsible for encoding OP stimuli works only on low frequency components of the display. We therefore asked if OP stimuli composed of large checks could be rendered undetectable if low frequencies are removed from the display.
METHOD. Stimuli were constructed from black and white checks having widths of 0.297 degrees visual angle. The proportion of matching elements was varied as before to determine threshold. Stimuli were filtered with a high pass Butterworth filter having frequency cutoffs of 0 to 24 cycles/patch. An adaptive procedure was used to find threshold deviations (82% correct in a 2IFC) above and below p = 0.5.
RESULTS. Thresholds were higher for the OP stimuli than SP stimuli for all cutoff frequencies. There was a general trend for thresholds to increase modestly as low frequencies were removed from both SP and OP stimuli. However, high pass filtering did not render the OP stimuli impossible to detect.
CONCLUSION. The hypothesis that only low frequency channels contribute to the detection of OP stimuli must be rejected.
Supported by NSERC and FCAR Grants to Rick Gurnsey