The mechanism of human color vision and potential implanted devices for artificial color vision

Abstract

Vision plays a major role in perceiving external stimuli and information in our daily lives. The neural mechanism of color vision is complicated, involving the co-ordinated functions of a variety of cells, such as retinal cells and lateral geniculate nucleus cells, as well as multiple levels of the visual cortex. In this work, we reviewed the history of experimental and theoretical studies on this issue, from the fundamental functions of the individual cells of the visual system to the coding in the transmission of neural signals and sophisticated brain processes at different levels. We discuss various hypotheses, models, and theories related to the color vision mechanism and present some suggestions for developing novel implanted devices that may help restore color vision in visually impaired people or introduce artificial color vision to those who need it.

Keywords: artificial vision; color perception; color vision; restoration of sight; retina; visual cortex.

Figure 1

A summary of color processing in the color visual system. The figure illustrates the anatomy (the middle figure), physiology (A–D), and perception (E–H) of color vision. Three types of visual cone cells (L, M, and S) capture information about different light frequencies based on their relative light absorption (A). Cells within the LGN compare the cone signals (B). Some cells' receptive fields are excited by L cone cells and are suppressed by M cone cells (forming “red-green” cells); others are excited by S cones and suppressed by (L + M) cones (forming “blue-yellow” cells), forming the basis for chromatic opponency (E). In V1, specialized “double-opponent” cells compare cone ratios in specific visual spaces with those in adjacent areas (C), probably constructed from LGN color cells, and they perform color contrast calculations (F). The color signal then proceeds to V2, where specific hues are represented in color bands spanning the thin stripes of V2 (D), potentially contributing to our perception of hue (G). In V4, more cells will be associated with color constancy (H). Processing color information involves multiple layers and is a complex process. Introducing coding strategies into artificial vision may lead to achieving controlled artificial color vision for the blind in the future.

Figure 2

Normalized spectral sensitivity of retinal rod and cone cells. Data sourced from Nathans (1999). The peak absorption of the spectra by the three classes of cones, L, M, and S, is 560, 530, and 425 nm, respectively, whereas the peak absorption of the rods is around 500 nm. The figure illustrates the normalized spectral absorption peaks of the four cells. However, in fact, rods will only respond in very dark environments and depolarize in normal light intensity.

Figure 3

The structure of the retina (from Kim, 2020). The retina is primarily composed of three layers of neurons. The first contact with light is the ganglion cells, followed by the bipolar cells and, finally, the photoreceptors, including the cones and rods. However, photoreceptors are the first to respond to light. Interconnecting these three types of cells is two types of interneurons, namely, horizontal cells and amacrine cells. These cells have distinct roles and functions: photoreceptors receive light stimulation and convert it into electrical energy, eliciting nerve impulses; bipolar cells facilitate information transmission between photoreceptors and RGCs predominantly; horizontal cells and amacrine cells regulate the transmission of information from photoreceptors to bipolar cells and from bipolar cells to RGCs, respectively; RGCs are accountable for processing and integrating information, conveying electrical signals via nerve axons to the relay cells of the LGN, eventually reaching the visual cortex to engender vision. Created with BioRender.com.

Figure 4

Visual stimulation is encoded by the response latency of retinal ganglion cells (Gollisch and Meister, 2008). (A) A photograph of a swimming salamander larva is projected onto the salamander retina. The ellipse in the upper right corner illustrates the receptive field of a salamander ganglion cell. During each stimulation trial, the image is slightly shifted based on the grid of dots. This allows for recording the responses of the depicted ganglion cell to all pixels in the photograph. (B) The firing activities of the depicted ganglion cell are illustrated, with each dot representing a spike at receptive field locations along the column, indicated by the arrows in (A). (C) A visual image is reconstructed using a gray-scale plot based on the spike latency in each trial. (D) Corresponding gray-scale plot of the spike counts.

Figure 5

Inhibitory feedback from horizontal cells to cones. A recording from a turtle cone reveals that illuminating this cone with a small spot of light elicits a hyperpolarizing response. The subsequent application of an annulus to illuminate the surrounding receptive field induces hyperpolarization in the surrounding cones, leading to hyperpolarization in their postsynaptic horizontal cells. This alteration in inhibitory feedback from horizontal cells to the central cone elicits a depolarizing response in that cone. Cone response adapted from Burkhardt et al. (1988).

Figure 6

Models of double-opponent V1 neurons. (A) A schematic representation of receptive fields is depicted for a 2D plane of L-M opponent cells, exhibiting side-by-side spatial antagonistic regions and opponent cone weights. Weights above the horizontal plane indicate “ON” states, wherein an increase in light elicits an increase in response; conversely, weights below the horizontal plane signify “OFF” states, where a decrease in light leads to a decrease in response. The left panel illustrates the organization of the 2D receptive field, while the right panel presents a hypothetical spatial sensitivity profile. (B) A 3D schematic of the aforementioned receptive field model. (C, D) Two-dimensional maps (derived from subspace reverse correlation) depict the sensitivity of this cell to L (A) and M (B) cone isolation patterns (Johnson et al., 2008). Pseudo-color mapping indicates excitation to increases in red and excitation to decreases in blue. Fixation points within the visual field are denoted by stars and open circles to facilitate comparisons between L cone and M cone maps. At the star locations, L cone maps exhibit decreasing excitatory responses, whereas M cone maps show increasing excitatory responses; conversely, at locations labeled with open circles, the pattern is reversed.

Figure 7

Color-specific band in the second cortical visual area (V2). (A) Regions corresponding to peak activity in response to different colored stimuli, each tested independently, are delineated on the surface image of the brain. (B) The correlation between neuronal responses recorded by multi-unit electrodes and optical signals in V2 color bands reveals a significant correlation between the two signal types. Adapted from Xiao et al. (2003).

Figure 8

Relationship between color discrimination ability and eccentricity angle. (A) Schematic structure of an eyeball. (B) Density distributions of cones and rods along the retina (blind spot ignored). (C) Recognition accuracy of yellow words on blue background and blue words on yellow background with eccentricity angle. (D) Recognition accuracy of Chinese characters with various colors is assessed against different backgrounds and horizontal eccentricities. (E) Recognition accuracy of digits with various colors is assessed against different backgrounds and horizontal eccentricities. Adapted from Lin et al. (2023).

Figure 9

Schematic diagram of the “eye in eye” based on reflection. The device consists of a concave reflector and a micro-display encapsulated in a container. By utilizing the integrated micro-display and concave mirror, the device precisely projects colored light into the fovea of the retina.

Figure 10

Schematic diagram of the flexible micro-display device. The flexible micro-display is positioned in front of the retina. It utilizes an external image-receiving module capable of receiving video information from the Internet or local cameras. After compression and encoding via the image-processing module, the processed image is transmitted to the intraocular implantable micro-display screen.