A painted path deceives. Its pigments absorb some wavelengths and send others toward the eye, yet the reds, blues, and yellows that seem to sit on the surface arise only after neural circuitry translates reflected light into experience. Light arrives first. Color appears later.
That delay is the trick. Phototransduction does the real work: cone cells in the retina convert photon energy into electrochemical signals, while the visual cortex compares their outputs through opponent-process coding. No pigment reports itself. Three cone classes respond broadly rather than labeling discrete hues, so perceived color emerges from relative activity, not from a single wavelength stamped with a name. The visual cortex behaves like a graphics processor running color constancy, correcting shifts in illumination so surfaces keep a workable identity.
That correction is useful, not innocent. It lets a route remain legible as clouds pass or shade moves, but it also means the mind can smooth differences that a spectrometer would register without hesitation. The brain edits. A shadow may make the same paint seem altered even when its material composition has not changed. Machine vision faces the same trap. As cameras infer illumination from incomplete scenes, they may separate reflectance from appearance more effectively, exposing human color as a fast, private computation.
