Photoreceptors have high energy demands and densely packed mitochondria through which light passes before phototransduction. Old world primates including humans have three cone photoreceptor types mediating color vision with short (S blue), medium (M green), and long (L red) wavelength sensitivities. However, S-cones are enigmatic. They comprise <10% of the total cone population, their responses saturate early, and they are susceptible in aging and disease. Here, we show that primate S-cones actually have few mitochondria and are fueled by glycolysis, not by mitochondrial respiration. Glycolysis has a limited ability to sustain activity, potentially explaining early S-cone saturation. Mitochondria act as optical filters showing reduced light transmission at 400–450 nm where S-cones are most sensitive (420 nm). This absorbance is likely to arise in a mitochondrial porphyrin that absorbs strongly in the Soret band. Hence, reducing mitochondria will improve S-cone sensitivity but result in increased glycolysis as an alternative energy source, potentially increasing diabetic vulnerability due to restricted glucose access. Further, glycolysis carries a price resulting in premature functional decline as seen in aged S-cones. Soret band absorption may also impact on mitochondrial rich M and L cones by reducing sensitivity at the lower end of their spectral sensitivity range resulting in increased differentiation from S-cone responses. These data add to the list of unique characteristic of S-cones and may also explain aspects of their vulnerability.

This study investigated chromatic induction from inhomogeneous background patterns. Previous work showed that a background pattern detected by only S cones induced strong color shifts in a nearby test area (Monnier & Shevell, 2003). In that work, the S-cone patterns were composed with constant L- and M-cone stimulation over the entire background; in terms of L and M cones, therefore, the background was uniform. S-cone stimulation was varied over space to produce S-cone-isolated background patterns. These S-cone patterns, however, established spatial structure (the pattern) at both the receptoral level (S-cone stimulation) and the postreceptoral level (S/(L+M)). Here, these two levels of pattern representation were unconfounded to determine whether color shifts induced by S-cone patterns were due to spatial structure within an S-cone-specific neural pathway versus a pathway that combines responses from S cones and other cone types (e.g. S/(L+M)). The results showed that the induced color shifts were mediated by signals within a pathway that combines responses from multiple cone types. These results are consistent with a +s/−s spatially antagonistic neural receptive field, which is found in some neurons in V1 and V2.