Moonlight is miserly. Against leaf litter, a fox's eyes can flash when a beam strikes them, yet that shine is not the hunt itself. Behind the retina sits the tapetum lucidum, a reflective layer that sends unabsorbed photons back through the photoreceptors. Think of it as a photon-recycling cache, although its real mechanism is optical reflection. A moving outline matters.
This trick is ingenious. Light that would otherwise leave the eye passes the retina twice, increasing the chance of absorption by photopigment molecules within rod cells. No photons appear. The second pass can soften image detail, but it raises the odds that a rod will register a dim edge or a brief shift in contrast. Rods favor sensitivity. Neural circuits compare successive retinal signals, allowing motion to stand apart from a background that barely changes.
The payoff is real. A fox is not seeing darkness as daylight; it is extracting change from a starved signal through a dilated pupil, a rod-rich retina, and returned light. Motion is cheap information. An ear twitch or a vole crossing a dark patch alters the retinal pattern faster than the background does. Engineers building low-light machine vision should study the bargain: use optical recirculation to feed sensors, then let temporal processing reject what stays still. Future cameras could pair that logic with event-based sensors that flag change before a full image is formed.