Skadi Gerkensmeier¹, Joan Danielle K. Ongchoco¹; ¹University of British Columbia

In perception, visual processing strives towards order, both in space (via object segmentation) and in time (via event segmentation). In physics, however, systems strive towards chaos, and the reverse—maintaining order—costs energy. Here we asked how this fundamental law of entropy manifests in visual processing. Might there be an asymmetry in the effort it takes to attend to order vs. chaos? Inspired by physical systems, we turned to a curious phenomenon: the pendulum wave. It consists of a row of pendulums, swinging with identical amplitudes but monotonically increasing frequencies. Launched from the same start point, the pendulums synchronize and desynchronize over time, resulting in momentary percepts of “order” (where the pendulums look like they are moving as a single wave) and of “chaos” (where the coherent global motion seems to disappear). We tracked attentional effort across continuously transitioning pendulum waves using a multiple object tracking task. Observers tracked “pendulums” (rendered as Gabor patches) in a pendulum wave animation that was transitioning from either order to chaos or chaos to order. Observers reported whether a probe appeared on/off a tracked pendulum. Because tracking performance decays over time, we presented sliding windows of the full transition and presented probes in the early, middle, or late parts of each window. We found that tracking performance decreased over time, but this was influenced by the progression and direction of the transition. In particular, the transition from order-to-chaos disrupted performance early in the progression of the pendulum wave (compared to its chaos-to-order counterpart)—and chaos showed a stabilizing effect on performance throughout the rest of the transition. Thus, in perception, as in physics, the law of entropy seems to structure visual processing: while the loss of order can be disruptive, the resulting chaos may free up attentional resources, allowing for more flexible attention allocation.

Acknowledgements: Supported by a Natural Sciences and Engineering Research Foundation of Canada Discovery Grant