Dogs fail to recognize a human pointing gesture in two-dimensional depictions of motion cues

•Dogs underwent a pointing task, where demonstrators had progressively poorer visual information.•Stimuli were, in order: a real person, a video, a silhouette, and three types of point light displays.•Dogs only responded above chance level to the real-life and the black and white video.•Presenting the point-light displays first did not improve dogs’ performance.•Dogs seem unable to recognise humans in two-dimensional depictions of human motion cues. Few studies have investigated biological motion perception in dogs and it remains unknown whether dogs recognise the biological identity of two-dimensional animations of human motion cues. To test this, we assessed the dogs’ (N = 32) responses to point-light displays of a human performing a pointing gesture towards one of two pots. At the start of the experiment the demonstrator was a real-life person, but over the course of the test dogs were presented with two-dimensional figurative representations of pointing gestures in which visual information was progressively removed until only the isolated motion cues remained. Dogs’ accuracy was above chance level only with real-life and black-and-white videos, but not with the silhouette or the point-light figure. Dogs’ accuracy during these conditions was significantly lower than in the real-life condition. This result could not be explained by trial order since dogs’ performance was still not higher than chance when only the point-light figure condition was presented after the initial demonstration. The results imply that dogs are unable to recognise humans in two-dimensional depictions of human motion cues only. In spite of extensive exposure to human movement, dogs need more perceptual cues to detect equivalence between human two-dimensional animations and the represented living entity.