Shared Mechanisms Drive Ocular Following and Motion Perception

Abstract

How features of complex visual patterns are combined to drive perception and eye movements is not well understood. Here we simultaneously assessed human observers' perceptual direction estimates and ocular following responses (OFR) evoked by moving plaids made from two summed gratings with varying contrast ratios. When the gratings were of equal contrast, observers' eye movements and perceptual reports followed the motion of the plaid pattern. However, when the contrasts were unequal, eye movements and reports during early phases of the OFR were biased toward the direction of the high-contrast grating component; during later phases, both responses followed the plaid pattern direction. The shift from component- to pattern-driven behavior resembles the shift in tuning seen under similar conditions in neuronal responses recorded from monkey MT. Moreover, for some conditions, pattern tracking and perceptual reports were correlated on a trial-by-trial basis. The OFR may therefore provide a precise behavioral readout of the dynamics of neural motion integration for complex visual patterns.

Keywords: eye movements; ocular following; pattern motion; perception–action.

Figure 1.

A, Example plaids with a 1:1, 1:4, and 1:16 contrast ratio made by adding two gratings with different orientations and contrasts. Blue and red arrows indicate component motion directions; the purple arrow shows pattern direction. B, Trial timeline.

Figure 2.

Single observer data, rotated as described in Materials and Methods so that eye movements in the direction of the high contrast component are labeled as component and eye movements 90° away as orthogonal. A, Component and (B) orthogonal speed over time for the 1:1 contrast ratio. Traces show averages across trials for each presentation duration. Blue lines indicate analysis time windows for early OFR (dark blue, 95–135 ms), late OFR (medium blue, 155–195 ms), and tracking (light blue, 275–315 ms). C, Component and (D) orthogonal speed for each contrast condition (gray shades; red line indicates grating), for the two longest duration conditions. E, Average eye movement trajectories for each contrast ratio.

Figure 3.

Comparison of contrast-dependent biases in OFR and perceptual estimates across observers (n = 8). A, Eye movement direction over time relative to the stimulus onset for each contrast ratio. Traces show averages across observers and shaded areas represent ±1 SD. B, Reported motion direction for each contrast ratio and presentation duration. Shaded areas represent ±1 SD. C, Contrast ratio functions for early (dark blue), late (medium blue), and tracking (light blue) phases of the OFR. Dots and error bars show mean ± 1 SD across observers. Vertical lines on the baseline indicate r₅₀. D, Contrast ratio functions for perceived direction for each presentation duration. Same conventions as in panel C.

Figure 4.

Comparison of biases in perception and tracking. Orange colors denote presentation durations; blue outlines are OFR phases. Large circles are averages across observers per comparison between perception and tracking (e.g., dark orange circle with dark blue outline shows comparison between the shortest presentation duration and the early OFR phase). Gray lines connect individual observer data and show consistent time-shifted sensitivity across observers. Note, because we compare perceptual responses for the two longer presentation durations to the same tracking response, gray lines connect the average of the two longer presentation durations with the individual data points for the shorter presentation durations.

Figure 5.

Trial-by-trial correlations between perceptual reports and OFR for all observers. Top row, early OFR (95–135 ms); middle row, late OFR (155–195 ms); bottom row, tracking (275–315 ms). Data for the tracking phase were pooled from the 250 and 500 ms presentation times. Dashed lines are identity lines.

Figure 6.

Schematic diagram summarizing key findings and proposed mechanisms. Plaids and gratings are processed via a motion integration stage providing a shared pattern motion signal for OFR and perception. Two noise sources are proposed: a common sensory noise source shared by perception and eye movements and a private source for each effector system. N, noise; size of the circles illustrates magnitude of intrasubject trial-by-trial variability.