Josh McDermott teaches this lecture from MIT's 9.35 Perception course, covering how the visual system infers distance and depth from a two-dimensional retinal image. He works through the standard cues the brain relies on, including binocular disparity, motion parallax, occlusion, relative size, and perspective, explaining how each supplies information about how near or far objects are. The lecture addresses why depth perception is a genuinely hard inference problem, since the retina itself only registers a flat projection of the world, and discusses how the brain combines multiple imperfect cues into a single estimate of spatial layout. Running about 38 minutes, it fits within MIT OpenCourseWare's Spring 2024 offering of the course and assumes familiarity with earlier lectures on the visual system.