Its easier to track moving items if they are collinear with their direction of motion

Todd S. Horowitz, Jeremy M. Wolfe, & Jennifer S. DiMase


How can the visual system track multiple moving objects through time? We propose that attending to trajectories is facilitated by mechanisms similar to the association fields proposed for contour extraction (e.g. Field, Hayes, & Hess, 1993 Vision Research 33, 173-93). Here we investigate one prediction of this spatiotemporal association field hypothesis: objects aligned with their heading should generate stronger forward associations and better tracking than stimuli whose motion is orthogonal to their orientation.

Stimuli were gabor patches (s.f. = .5 cycles/°) on a grey background. There were ten items on each trial. Five items were cued at the start of the trial, and five observers tracked them for ten seconds. Items bounced off display edges and each other. Velocity varied inversely with distance to other items. In separate blocks, items were oriented either collinear or orthogonal to their headings. When motion ceased, observers identified the cued items.

As predicted, accuracy (mean; sem) was significantly higher (t(5) = 5.82, p < .005) in the collinear condition (74.7%; 3.9%) than in orthogonal condition (67.1%; 2.7%).

Spatiotemporal association fields may form the basis for a theory of trajectory prediction in multielement visual tracking.