Abstract
The horizontal vergence eye movements (Vg) of 2 human Ss were recorded while horizontal disparities were applied to vertical square-wave gratings lacking the fundamental (0.04–0.5c/°). Dichoptic viewing was by Wheatstone stereoscope. Initially, a central fixation spot was seen against a grey background, then these were replaced with the grating pattern (52° wide × 43° high; contrast: 0.4, 0.8; duration: 200ms) with a crossed (uncrossed) disparity that was of the fundamental wavelength. Thus, the luminance features (peaks, troughs) and the 4n + 1 harmonics (n=integer) had crossed (uncrossed) disparity, whereas the 4n − 1 harmonics—including the strongest Fourier component (3rd harmonic)—had the reverse disparity (cf., Adelson & Bergen, 1985). The earliest Vg were reversed, i.e., crossed disparities caused divergence and uncrossed disparities caused convergence, and had the ultra-short latencies characteristic of the responses to the 3rd harmonic alone (pure sinusoid): 75–80ms. These data are consistent with the notion that the earliest Vg are mediated by a disparity-energy mechanism. However, the missing fundamental is a very complex stimulus and we also tried simpler ones. For example, we examined the Vg when disparity was applied to vertical grating patterns consisting of a sum of two sinusoids of frequency 3f and 4f (“3f+4f”), which created a repeating pattern with a “beat” frequency, f (Hammett, Ledgeway & Smith, 1993). The disparities were of the “beat” wavelength, so the feature-defined disparity was crossed (uncrossed), whereas the disparity of the 3f component was uncrossed (crossed) and that of the 4f component was zero. Initial Vg were again reversed, i.e., in the direction of the 3rd harmonic, and again shared the same ultra-short latency. Other compound stimuli included 3f+5f and 3f+7f and all yielded Vg consistent with energy-based detection, though we cannot rule out an additional (much weaker?) contribution from a correspondence-based mechanism.
Supported by the National Eye Institute Intramural Research Program