Answers are grounded in this paper and cite their sources.
Humans can identify and categorize objects in a dynamic visual scene with remarkable accuracy. To achieve this, our visual system relies not only on static information but also on dynamic cues. Turning a static image upside down hinders recognition. Similar inversion effects have been found in dynamic videos of human action. However, it is unknown whether this inversion effect extends to a broader set of categories when only dynamic cues are available. Here, we used object kinematograms to investigate the inversion effect on object recognition using only dynamic cues. We extracted motion cues from the videos and projected them onto randomly moving dots to create object kinematograms that contain only dynamic information. Participants viewed upright or inverted object kinematograms from four animate and four inanimate categories in an online experiment and completed three tasks: animacy judgment, subordinate-level categorization, and object naming. Overall, inversion impaired object categorization based on motion cues, although the magnitude of the effect varied across categories. For animacy judgment, the animate category showed a small inversion effect, whereas we did not observe a significant inversion effect for the inanimate category. For subordinate categorization and object naming, we observed significant inversion effects in some animate and some inanimate categories. These findings demonstrate that inversion disrupts object recognition even when only dynamic information is available and that the magnitude of this disruption depends on the object category. This variability may reflect differences in the natural orientation statistics of objects or the physical dynamics of their motion.
No takes yet. Share an insight, caveat, or question.
Faridani et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: