Abstract
PURPOSE. Points I(x,y) and I(−x,y) are symmetrically placed with respect to the y axis. Zero-mean images possess the property of bilateral symmetry about the y axis if for all (x,y), I(x,y) = I(−x,y), and the property of anti-symmetry if for all (x,y), I(x,y) = −I(−x,y). Mancini, et al. (2001, ARVO) measured sensitivity to symmetry and anti-symmetry as a function of check size for patterns comprising black and white checks. Symmetry and anti-symmetry elicited similar sensitivities for large checks. As check size decreased sensitivity to symmetrical stimuli improved modestly whereas sensitivity to anti-symmetrical stimuli decreased dramatically. We wondered if this pattern of results would hold for the more general case in which stimuli contain a wider range of grey-levels.
METHOD. Symmetrical and anti-symmetrical stimuli were created with checks having intensities drawn from a normal distribution. Check widths ranged from 0.037 to 0.297 of visual angle and all patterns were windowed within a circular aperture of 9.5 in diameter. For each check size, noise masks were created by drawing from the same normal distribution and added to the stimulus. Stimulus contrast (c) and mask contrast (1 - c) summed to 1. Four subjects performed a 2IFC task. One interval contained random noise the other contained symmetry (or anti-symmetry) with some level of added noise; exposure durations and ISIs were 333 ms. Threshold was defined as the stimulus contrast (c) eliciting 82% correct responses.
RESULTS. In most cases thresholds were unmeasurable for the anti-symmetrical stimuli; i.e., even with c = 1 subjects rarely obtained 82% correct responses. For symmetrical stimuli thresholds ranged from 0.63 to 0.70 with no significant effect of check size.
CONCLUSION. Anti-symmetry is extremely difficult if not impossible to detect in the general case. With binary stimuli (Mancini et al., 2001) subjects may be able to detect anti-symmetry by employing attentional strategies.