Purchase this article with an account.
Nicolas Davidenko, Yeram Cheong, Jacob Smith; Motion pareidolia: illusory perception of coherent apparent motion in random noise. Journal of Vision 2015;15(12):3. doi: 10.1167/15.12.3.
Download citation file:
© 2017 Association for Research in Vision and Ophthalmology.
We report a novel class of visual illusion we term motion pareidolia, in which the sequential presentation of random textures gives rise to illusory percepts of coherent apparent motion. In 3 experiments, we presented naive observers with sequences of random pixel arrays refreshing randomly every 400ms. In Experiment 1 (n=92), these sequences began with a "real" apparent motion prime (e.g. a fixed random texture shifting up and down by 4 pixels), which degenerated into randomness across the first 12 frames. Observers were instructed to press a button when they noticed that the primed motion pattern changed or stopped. Across conditions, 74% of observers reported that the primed motion persisted for at least 3 random frames following the prime, and often for 20 or more random frames (see Supplementary materials). In Experiment 2 (n=50), we tested whether these illusory percepts could arise spontaneously. We presented observers with 6-frame sequences depicting one of 4 motion patterns (alternating: up-down or right-left; or same-direction: up-up or right-right). Across trials we varied visual noise from 0% (pure signal) to 100% (pure noise) and instructed observers to identify which of the 4 motion patterns was being depicted, or whether the display was random. Observers falsely identified motion on 41% of the pure noise trials, indicating that motion pareidolia can arise spontaneously without a direct prime. Further, observers were significantly more likely to falsely perceive alternating than same-direction motion. In Experiment 3 (n=56), we tested whether this alternating bias could result from the boundary conditions of the display region. Indeed, when the display region was ring-shaped, the alternating bias disappeared. Overall, our data demonstrate that naive observers readily perceive suggested motion patterns in random noise. To account for this phenomenon, we propose a top-down selective visual attention model that is biased to confirm expectations.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only