Vestibular facilitation of optic flow parsing

Abstract

Simultaneous object motion and self-motion give rise to complex patterns of retinal image motion. In order to estimate object motion accurately, the brain must parse this complex retinal motion into self-motion and object motion components. Although this computational problem can be solved, in principle, through purely visual mechanisms, extra-retinal information that arises from the vestibular system during self-motion may also play an important role. Here we investigate whether combining vestibular and visual self-motion information improves the precision of object motion estimates. Subjects were asked to discriminate the direction of object motion in the presence of simultaneous self-motion, depicted either by visual cues alone (i.e. optic flow) or by combined visual/vestibular stimuli. We report a small but significant improvement in object motion discrimination thresholds with the addition of vestibular cues. This improvement was greatest for eccentric heading directions and negligible for forward movement, a finding that could reflect increased relative reliability of vestibular versus visual cues for eccentric heading directions. Overall, these results are consistent with the hypothesis that vestibular inputs can help parse retinal image motion into self-motion and object motion components.

Figure 1. Schematic of the experimental design.

A) Side-view illustrating the task with a heading of 0° (straight forward). The subject experiences self-motion and synchronized movement of the object (dashed circle) either up or down. The subject’s task is to indicate which direction the object moved in the world. B) Close up of the pattern of image motion on the display for heading  = 60° and downward object motion in the world (from panel E). Variables v_s and v_o represent the independent components of image motion associated with the self-motion and object motion, respectively (horizontal and vertical white arrows). Note that the object motion component (v_o) is equal in all examples shown here (C-F), but the angle of deviation (d) is not because the self-motion component (v_s) depends on heading direction. (C)-(F) The experiment was conducted at four heading directions: 0°, 30°, 60°, and 90°. The optic flow associated with each heading direction (as displayed on the screen) is illustrated in each panel and each inset shows a top down view of the self-motion trajectory. As heading eccentricity increases, the focus of expansion (FOE) is displaced further from the center of the display. The resultant image motion associated with the object is also visible in these panels to the left of fixation.

Figure 2. Summary of discrimination thresholds.

Each panel shows the data from a different subject. Error bars represent 95% confidence intervals. Subjects S1-S6 participated in the main experiment, so visual-only (blue bars) and combined (red bars) thresholds were measured at all heading eccentricities. Subjects S4-S8 participated in the retinal speed (RS) control experiment (green bars). Note that subjects S7 and S8 were only tested with the 90° heading in the eye movement control experiment (lateral motion).

Figure 3. Comparison of visual-only and combined thresholds.

Percent decrease in combined threshold relative to the visual-only threshold (computed as ( σ _v - σ _c)/ σ _v; subjects S1-S6) for all four heading angles. The decrease in threshold depends on heading angle, with the smallest decrease for 0° heading and the largest decrease for 90° heading.

Figure 4. Summary of eye movement analysis.

Each row summarizes data from one subject. Only left eye (LE) velocities were used for these analyses; conducting the same analyses using right eye velocities yielded similar results. Left column shows histograms of mean eye velocities from all trials for both the Visual-only (blue) and Combined (red) conditions. Right column shows Visual-only (blue) and Combined (red) thresholds as a function of mean eye velocity, along with regression lines fit to these data (see text for details).