Visual perception of travel distance for self-motion through crowds

Abstract

Humans can use visual motion to estimate the distance they have traveled. In static environments, optic flow generated by self-motion provides a pattern of expanding motion that is used for the estimation of travel distance. When the environment is populated by other people, their biological motion destroys the one-to-on correspondence between optic flow and travel distance. We investigated how observers estimate travel distance in a crowded environment. In three conditions, we simulated self-motion through a crowd of standing, approaching, or leading point-light walkers. For a standing crowd, optic flow is a veridical signal for distance perception. For an approaching crowd, the visual motion is the sum of the self-motion-induced optic flow and the optic flow produced by the approaching walkers. If only optic flow were to be used, travel distance estimates would be too high because of the approaching direction of the crowd toward the observer. If, on the other hand, cues from biological motion could be used to estimate the speed of the crowd, then the excessive optic from the approaching crowd flow might be compensated. In the leading crowd condition, in which walkers of the crowd keep their distance from the observer as they walk along with the observer, no optic flow is produced. In this condition, travel distance estimation would have to rely solely on biological motion information. We found that distance estimation was quite similar across these three conditions. This suggests that biological motion information can be used (a) to compensate for excessive optic flow in the approaching crowd condition and (b) to generate distance information in the leading crowd condition.

Figure 1.

Stimulus and reporting procedure. The upper panel shows the crowd of point-light walkers placed on the ground plane. The stripes on the ground plane were oriented such that they provide perceptive depth information but no optic flow. The lower panel shows the procedure for reporting perceived travel distance. The red line represents the reference line from which the participants started. This red reference line appeared only during the distance estimation and not during motion simulations. Participants moved the blue line along with the walker to indicate their traveled distance. The size of the walker scaled with its depth.

Figure 2.

Descriptive results and model fits per condition. The points show the average distance estimates. The error bars give the standard deviation of the average distance estimates. The solid red line indicates the fit of the leaky integration model. The black dotted line represents veridical performance.

Figure 3.

(A) Leaky fit per condition. The dotted line indicates veridical performance. (B, C) Leaky fit parameter k (B) and leak rate α (C) per condition. Asterisks denote significant differences between conditions.