Skip to main content


Computational descriptions of visual stimuli

Trainee Achievements

Computational descriptions of visual stimuli

Trainees Santos and Ross (Psychology) collaborated with Affiliate Gnang (Computer Science) to produce computational descriptions of visual stimuli used in behavioral experiments. Gnang and Santos analyzed the properties (velocity distributions) of random-dot kinematograms, where dots continually disappear and reappear in random locations, creating the percept of global motion without local motion of individual dots. These displays were used to show that anticipatory smooth pursuit eye movements can be elicited with global motion. Ross and Gnang (who models temporal properties of scenes using mathematical tensor analysis) analyzed the frequency of "scene cuts" in video clips and used this information to predict the occurrence of eye movement made to inspect the videos, finding that scene cuts inhibit eye movements.