For over a century, vertebrate vision has been taught as a two-cell system. Rods detect dim light with high sensitivity but slow response. Cones detect color in bright light with fast response but low sensitivity. Every vertebrate retina follows the same developmental sequence: cones first, rods later. The cells are distinct types with distinct proteins and distinct functions.
Deep-sea fish larvae from the Red Sea — lightfish, hatchetfish, and skinnycheek lanternfish — break this rule. Cortesi, Fogg, and colleagues found hybrid photoreceptor cells that combine the light-capturing ability of rods with the faster signal processing of cones. A single cell doing both jobs simultaneously.
The developmental rule also breaks. Instead of cones-first-then-rods, these larvae produce rod-like cone cells early in development. Some species use them as temporary intermediates, later replacing them with conventional rods. But hatchetfish retain the hybrids into adulthood, building their entire visual system around a cell type that isn't supposed to exist in the vertebrate toolkit.
The ecological logic is the deep twilight zone. At two hundred to one thousand meters depth, light exists but barely — a dim, blue-shifted remnant of surface illumination. Standard rods would be sensitive enough but too slow to track moving prey. Standard cones would be fast enough but too insensitive. The hybrid cell is a compromise: sensitive enough for the available photons, fast enough for the available prey. Neither rod nor cone, but something tuned to the specific light regime of the mesopelagic.
The textbook distinction between rods and cones is real in most vertebrates but not universal. Where the light environment demands something in between, evolution builds something in between. The categories were always descriptions of common solutions, not constraints on possible ones.