Pouring, scooping, bouncing, rolling, twisting, and rotating: Does spontaneous categorical perception of dynamic event types reflect verbal encoding or visual processing?

Abstract

What we see encompasses not only lower-level properties (such as a ball's shape or motion) but also categorical events (such as a ball bouncing vs. rolling). Recent work demonstrates that such categorical perception occurs spontaneously during passive scene viewing: observers are better able to identify changes in static or dynamic scenes when the change involves different "visual verbs" (e.g., pouring vs. scooping), even when the within-type changes (e.g., across two different scenes of pouring) are objectively greater in magnitude. Might this occur as a part of visual processing itself, even without explicit verbal encoding? To find out, we discouraged verbal labeling via explicit instructions, a concurrent verbal suppression task, or both. In all cases, we continued to observe robust cross-event-type advantages for change detection, while carefully controlling lower-level visual features-in contrasts including pouring versus scooping, bouncing versus rolling, and rotating versus twisting. This suggests that we spontaneously see the world in terms of different "visual verbs" even without explicit verbal labeling.