Josh McDermott continues his lecture on motion perception for MIT's course 9.35, Perception, Spring 2024. Building on the previous session, he examines how the visual system converts raw input from retinal photoreceptors into judgments about how objects and scenes move, tackling the ambiguities inherent in reconstructing motion from a sequence of two-dimensional images. The lecture works through the computational problems the brain must solve to distinguish object motion from self-motion, and how neural mechanisms might resolve the inherent uncertainty in visual signals. As part of an ongoing MIT OpenCourseWare series, it assumes familiarity with earlier lectures on early vision and builds toward a fuller account of how perception constructs a stable picture of a moving world from noisy, incomplete sensory data.