Your visual system is a layered machine that predicts motion
The visual system is the physiology behind visual perception: it detects, transduces, and interprets visible light to build images and a mental model of space. Tied to the eye and its brain partners, it handles depth, motion, pattern recognition, and visually guided movement in humans and, more broadly, vertebrates.
Cornea and lens refract a tiny image onto the retina, where rods and cones fire electrical pulses that leave via the optic nerve through the optic canal. At the optic chiasm many fibres cross, then most terminate in the lateral geniculate nucleus. Before forwarding to primary visual cortex V1, the LGN gauges object range and stamps velocity tags that anticipate movement; some fibres already branch toward V2 and V3.
V1 begins edge detection within about forty milliseconds, later folding in global organization around one hundred milliseconds, and drafts a bottom-up saliency map for gaze. V2 shares much of that job while parsing illusory contours, comparing left–right inputs for depth, and separating figure from ground; it talks to V1 through V5 and the pulvinar, a hub for saccades and attention. V3 tracks global motion direction and speed; V4 spots simple shapes; V5 and V6 divide self-motion from object-versus-background motion.
Downstream, the inferior temporal gyrus recognizes complex objects and faces and, with the hippocampus, helps mint memories. Pretectal nuclei modulate pain, REM, and accommodation; the Edinger–Westphal nucleus helps pupil control. The pathway is therefore less a camera cable than a stacked prediction engine: optics shrink the world, the LGN tags trajectories, and cortical areas V1–V6 stagger in time to turn photons into actionable space.
Much of the editing starts in the eye itself. Roughly 130 million photoreceptors, including about 120 million rods per eye, funnel their signals into only around 1.2 million ganglion-cell axons, so the retina compresses heavily before anything reaches the brain. Its outputs split into distinct populations: M cells tuned to depth and quick to adapt, P cells attentive to colour and shape, K cells with very large colour-sensitive fields, and others steering eye movements. A 2006 University of Pennsylvania estimate put the human retina's throughput near 8,960 kilobits per second, against about 875 for a guinea pig. And in 2007, work with people lacking rods and cones showed that light-sensitive ganglion cells, peaking at 481 nanometres, contribute to perception as well.
Source: Visual system
