Abstract
Using scleral search coils, we recorded the initial torsional ocular following responses (tOFRs) that were elicited in 3 human Ss when textured patterns occupying a circular area (diameter, 32º) in the frontal plane were centered on, and briefly rotated (200 ms) about, the optic axis. With random-dot patterns, the angular speed giving best responses averaged 166º/s and the associated latencies averaged 80–90 ms, which was ∼15–20 ms longer than the latency of the horizontal and vertical OFRs to optimal motion along the cardinal axes. The initial tOFRs were quantified by measuring the changes in torsional eye position over the 80-ms period commencing 70 ms after stimulus onset (open-loop measures). Excluding one half of the field (upper, lower, right or left) indicated that motion was more effective in the lower hemifield than in any other hemifield. When polar 1-D angular sinusoidal luminance gratings were rotated in successive ¼-wavelength steps at 10-ms intervals, tOFR measures showed a Gaussian dependence on log angular wavelength (r2[[gt]]0.97; mean peak and SD: 22.9º and 0.55 log10 units, respectively) and a dependence on contrast that was well described by the Naka-Rushton equation (r2[[gt]]0.97; mean c50 and n: 5.5% and 2.2, respectively). When these polar gratings were reduced to 2–15 equally-spaced concentric annuli each with the same radial width (0.5º), same angular wavelength (22.9º) and same contrast (range, 2%–32%), all tOFR measures were well fit by a single monotonic function (r2[[gt]]0.97) that showed only minor saturation when plotted as a function of the product, log(A*Cn), where A is screen coverage, C is Michelson contrast and n is a free parameter (mean n=2.1±0.4SD). This is very different from the divisive normalization we reported recently for the horizontal and vertical OFRs, and consistent with the hypothesis that the tOFR here is determined by the total motion energy in the stimulus.