We are grateful to Editorial Board member Rob Smith for his suggestion to dedicate a Special Issue to the topic of short wavelength-sensitive (S) cones and S-cone pathways. We also thank most warmly the authors who contributed these reviews and the other experts in the field who reviewed the manuscripts. This Special Issue is a particularly appropriate way to celebrate the twenty-fifth anniversary of the founding of Visual Neuroscience because it highlights the breadth of topics covered by articles in the journal. The topics range from the first steps of vision in the eye to responses in extrastriate cortex and the properties of human color vision. Here, we summarize the thrust of these reviews and highlight the important unanswered questions raised by our comprehensive survey of S-cones and S-cone pathways.
David Hunt and Leo Peichl consider the S-cones from an evolutionary perspective. They review an extensive literature on amino acid sequences of S-cone pigments, opsins, in vertebrates and conclude that the S-cones are truly primordial, having appeared before other photoreceptor (rod and cone) classes in the vertebrate evolution. The authors describe how the marsupials and eutherian mammals express a specific (SWS1) class of opsin, with peak sensitivity ranging from ultraviolet (ultraviolet sensitive, UVS) to violet wavelengths (violet sensitive, VS). The UVS or VS pigment in most species is expressed specifically in “true” S-cone cells, which have a relatively low spatial density and, typically, some sort of gradient in their density. In some species (including three well-known laboratory model species: rats, mice, and guinea pigs), however, the UVS or VS pigment is coexpressed with long-wavelength sensitive (LWS) class pigments in individual cones, often in a restricted region of the retina. The authors propose that coexpression of S-opsin is a result of incomplete suppression of a default developmental pathway, which in turn reflects the deep evolutionary history of S-cones. In other mammals, S-cones are only present in restricted regions of the retina or, in some nocturnal and aquatic mammals, absent entirely. The authors discuss the evolutionary significance of these findings and also of the ultraviolet sensitivity of some mammalian S-cones. They conclude that the primary function of true S-cones in mammals is for color vision, because they provide input to neurons with responses of opposite polarity to different wavelengths of light, called “color opponent” cells.
Dennis Dacey, Jill Crook and Orin Packer describe the retinal circuits that give rise to color opponent responses in S-cones, bipolar cells, and retinal ganglion cells of primates. They review the evidence that S-OFF responses originate from large sparse and giant sparse (melanopsin-containing) ganglion cells in primates and, in macaques, also from a subset of central OFF midget ganglion cells. Their main focus is on the neural circuits that provide inputs to the small bistratified ganglion cells, so named because they have dendrites branching in two different strata of the inner plexiform layer. These are the best-understood color-selective ganglion cells, and they are excited by increments in S-cone stimulation. The authors show that the “blue-on/yellow-off ” characteristic of these cells arises from S-ON-type excitation by S-ON (also called “blue-cone”) bipolar cells and L/M-OFF-type excitation by bipolar cells that are depolarized by decrements in long (L) and medium (M) wavelengths. They summarize the evidence that the antagonistic receptive field surrounds of the two types of bipolar cells cancel, leaving the small bistratified ganglion cells with receptive field centers in which S-ON and L/M-OFF responses can both be elicited (also called “Type II” receptive fields). The excitatory S-ON and L/M-OFF inputs are counteracted by the inhibition arising in two different places. The S-cone inhibition arises mostly in the outer retina but strong L/M inhibition arises from amacrine cells in the inner retina. The authors point out that the relatively weak inhibition of S-ON inputs may be a first step in the amplification of signals from S-cones, which enables them to make a stronger contribution to color perception than expected from their sparse distribution. Another possible contribution to the amplification of S-cone signals is that each S-cone typically contacts two S-ON bipolar cells.
Kiyoharu Miyagashima, Ulrike Grünert and Wei Li review contributions of inner retinal circuitry to the processing of signals from S-cones. They begin with the evidence for a primordial color opponent system in mammals that compares the signals from S-cones with those that are sensitive to longer wavelengths. Both L- and M-sensitive pigments are expressed in many primate species, as a result of allelic variation or gene duplication events that took place after primates diverged from other mammalian lines. This development allowed trichromatic color vision but does not seem to have changed the first step in retinal pathways transmitting S-cone signals. A single type of S-ON bipolar cell, receiving inputs only from S-cones, has been described in all mammals studied so far. In primates, there are midget bipolar cells, which receive input from single cones in the central retina. ON midget bipolar cells receive input from L- and M-cones but do not receive any input from S-cones. There may be a difference between Old and New World monkeys in the inputs from S-cones to OFF midget bipolar cells in central retina; these have been reported in macaques but are absent in marmosets.
Diffuse bipolar cells, which receive inputs from multiple cones, typically receive only sparse inputs from S-cones and, therefore, the more numerous L and M cones provide virtually all of their input. A major focus of the review is on the pathway in which S-ON amacrine cells generate S-OFF responses in retinal ganglion cells via inhibitory synapses. The S-cone amacrine cells were discovered last year in the ground squirrel retina, but retinal ganglion cells they contact have not yet been identified morphologically.
David Marshak and Stephen Mills review the literature on retinal ganglion cells that respond to stimulation of S-cones; these have been found in one marsupial (tammar wallaby) and several species of placental mammals. The small bistratified ganglion cell of primates was the first of these to be identified morphologically, and because the S-ON bipolar cell appeared similar in all mammals, it was widely assumed that a homologue would be found in other mammals, as well. Therefore, it was surprising that S-ON ganglion cells of guinea pig and rabbit retinas are monostratified, and their M-OFF responses are derived from horizontal cells in the outer plexiform layer. This mechanism also underlies the color opponency of a recently described type of bistratified S-OFF ganglion cell in rabbit retina. These cells receive excitatory input from S-OFF bipolar cells, and their color opponent responses are inherited from the bipolar cells. Thus, there seems to be three distinct mechanisms for generating color opponency in retinal ganglion cells: 1) surround responses generated in the outer plexiform layer by horizontal cells are conveyed via excitatory synapses from bipolar cells, 2) amacrine cells that receive cone-specific input make inhibitory synapses and 3) excitatory synapses from ON bipolar cells selective for one cone type and OFF bipolar cells selective for another cone type. The authors combine results from primate and non-primate mammals to generate an evolutionary scenario to account for this diversity. According to this scheme, the anscestral type was monostratified, and the bistratified types emerged independently in the ancestors of primates and rabbits.
Paul Martin and Barry Lee review the spatial and temporal properties of S-cone receptive fields in subcortical pathways. By contrast to the rapidly expanding knowledge of S-cone pathways in the retina, knowledge of the distribution of S-cone signals in subcortical pathways remains largely confined to the primate lateral geniculate nucleus (LGN). Martin and Lee review the early experimental studies of cone-opponent cells in primates and show how S-ON cells became distinguished as the best approximation to ideal color-coding (“Type II”) receptive fields. The anatomical segregation of S-cone signals to the primitive koniocellular division of the LGN, whereas red–green opponent (L/M) signals segregate to the parvocellular division of the LGN, is consistent with the recent evolution of high-resolution (foveal) vision together with the divergence of M and L cone pigments to feed parvocellular receptive fields. Furthermore, there is now established functional homology between S-ON receptive fields in cat LGN and monkey LGN, reinforcing the idea that S-cone pathways are a primitive color vision system. The authors explain the experimental niceties of detecting weak functional inputs from S-cones and show cautionary examples of the potential pitfalls of the “silent substitution” method. The authors also highlight the discrepancy between the anatomical prediction of S-OFF midget ganglion cells in central visual field and the large receptive fields of S-OFF cells recorded in LGN.
Youping Xiao reviews the next stage of S-cone pathways in the primary visual cortex (V1) and the second visual cortical area (V2). Here, we know that big transformations of the visual signal take place. Inputs from the two eyes converge, and there is also physiological evidence for the mixing of signals from the parallel afferent pathways. The segregation of S-cone signals established in LGN is preserved at the first stage of cortical processing, where S-cone signal-carrying axons from LGN terminate in more superficial layers than L/M-cone signal-carrying axons. Xiao points out that in V1 (unlike LGN), cells cannot be divided into clear groups on the basis of their S-cone weight. Whether the S-cone signals are simply dissipated among cortical fields or some recoding into a population of highly color-selective cells takes place here remains a major unresolved question. Overall, it is clear that color tuning in neither LGN nor V1 cells is clearly related to color perception. Xiao summarizes the tantalizing evidence that V1 cells are grouped according to their preference for colors perceived as “warm” or “cool” in perception, and may even form spatial clusters with distinct hue preferences.
Bevil Conway follows the S-cone signals deep into the extrastriate (“association”) cortex, where decades of research have produced compelling evidence for functional specialization of distinct cortical areas: for object recognition, movement perception, and, perhaps, even a “colouring-in” area to imbue our world with the glory of color. Yet, the contribution of S-cone signals to color computation in the cortex has been difficult to unravel. Conway makes a smart contribution to the debate, examining the presence of S-cone signals in two extrastriate areas with purportedly opposite functions. Cells in area MT, part of the dorsal “where” stream for movement detection and visually guided action, do likely get at least some S-cone input, albeit much weaker than the dominant S-cone input to blue-ON cells in LGN. This input may have more to do with navigation than color perception. For example, the S-cone signals can discriminate sunlight from shadows as a clue to scene geometry. By juxtaposing S-cone responses in MT with sharp color tuning of some cells in regions in inferotemporal (IT) cortex, Conway makes the important distinction between sensitivity (any cell which responds to S-cone contrast) and selectivity (S-cone recipient cells that extract color contrast yet ignore brightness contrast).
Hannah Smithson takes our exploration of S-cone signals to the area of human visual performance, with specific focus on how the special qualities of the S-cone pathways are manifest in visual sensation and perception. Here again, the question arises whether S-cone pathways are exclusively devoted to transmitting color signals or if the deep evolutionary history of S-cones has led to a broader role in spatial vision and object boundary detection. The established spatial asymmetry of S-ON and S-OFF pathways can be linked to greater sensitivity and higher acuity of the human vision for S-cone contrast increments compared with decrements. Smithson, in common with other authors in the Special Issue, makes a special point of explaining the fine stimulus control needed to isolate S-cone mechanisms, in psychophysics as in electrophysiology, and explains some of the trade secrets for success without artifacts. The role of S-cone signals in control of eye movements and attention capture is considered (S-cones seem to play a limited role) as are the residual visual functions surviving damage to primary visual cortex or “blindsight” (S-cones seem to play little if any role here).
Unanswered questions:
Hunt and Peichl raise our first big unanswered question: first, the relation of S-cone topography to what may seem obvious environmental signals (the blue sky and the blue of the ocean) is still not at all clear. Some aquatic mammals lack S-cones but some do not, S-pigment coexpression can be restricted to dorsal or ventral retina in different species with similar lifestyle, and the spectral tuning of UVS and VS pigments in birds and mammals remains at best tenuously related to their ecological specializations. Second, if the primordial photoreceptors were S-cones, why do S-cones make up only a minority of receptors in extant vertebrates?
The S-OFF pathways received a great deal of attention in the reviews by Crook et al. and Miyagashima et al., but this topic remains controversial. The existence of an S-OFF pathway mediated by S-OFF midget ganglion cells in the central retina of Old World anthropoids is supported by evidence from electron microscopy and psychophysics, but there has not yet been a direct anatomical demonstration of a link from the identified S-cones to the identified midget bipolar cells in central retina. This would be essential to convince the skeptics, who argue that there is very little, if any, evidence for an S-OFF midget pathway from recordings of neural activity in the LGN. A counter argument is that there are likely very few S-OFF midget ganglion cells compared with the other types, and they are expected to have atypical light responses because their receptive field surrounds are mediated by, H2 rather than H1, horizontal cells.
Many enduring mysteries remain unsolved at higher levels of processing. For example, unique hues (red, yellow, green and blue) enjoy universal use as basic color names across human societies, but cells tuned to unique hues have not been clearly identified at any stage of visual processing, and S-cone activity levels map to lime-violet percepts, not the familiar blue-yellow axis. These kinds of gulfs between color, as practiced by neuroscientists, and color as enjoyed by most humans are embarrassing but are a spur to further research.
