摘要：SummaryCan we recover self-motion from vision? This basic issue remains unsolved since, while the human visual system is known to estimate thedirectionof self-motion from optic flow, it remains unclear whether it also estimates thespeed. Importantly, the latter requires disentangling self-motion speed and depths of objects in the scene as retinal velocity depends on both. Here we show that our automatic regulator of walking speed based on vision, which estimates and maintains the speed to its preferred range by adjusting stride length, is robust to changes in the depths. The robustness was not explained by temporal-frequency-based speed coding previously suggested to underlie depth-invariant object-motion perception. Meanwhile, it broke down, not only when the interocular distance was virtually manipulated but also when monocular depth cues were deceptive. These observations suggest that our visuomotor system embeds a speedometer that calculates self-motion speed from vision by integrating monocular/binocular depth and motion cues.Graphical abstractDisplay OmittedHighlights•Changes in optic flow speed triggers implicit adjustments of walking speed•The response is invariant with respect to the depths of objects in the scene•The invariance is not explained by temporal-frequency-based speed coding•Both binocular and monocular depth cues contribute to the invarianceBiological sciences; Neuroscience; Behavioral neuroscience; Sensory neuroscience
