Colour Vision Mechanisms 1
Chromatic Theory and Color Matching
Chromatic theory was first proposed by Thomas Young in the early 1800s and elaborated by Helmholtz later in the 19th century. Young suggested the existence of three types of particles in the retina, responsible for the sensations of red, green, and blue. He posited that all other hues are mixtures of these primaries.
At the time of Young's suggestion, the detailed structure of the retina was unknown and cones were not known to exist. However, his suggestion was prescient, as we now know the three types of particles are the three spectral types of cones, although he did not know their nature. Cones and other photoreceptors were later described by 19th-century neuroanatomists, and inferences about their function were validated by color matching experiments.
Color matching involves obtaining any color by adjusting the intensities of three spectral lines, typically in the blue, green, and red regions of the spectrum. For instance, narrowband lights with peaks at approximately nm (blue), nm (green), and nm (red) can be adjusted to match any color. By manipulating the relative intensities of these lights, one can match the perceived color of a surface with a distinct spectral reflectance curve.
Two spectra are considered metamers if they are physically different but produce the same subjective color sensation. For example, the broadband spectrum reflected from a piece of blue paper can be matched by a mix of three narrowband lights, even though the spectra are objectively different.
Historically, color matching was demonstrated using psychophysical methods. A broadband light source was split into narrowband spectral lights using adjustable interference filters. A narrowband spectral light (e.g., yellow at nm) would be projected onto one half of a target field, while a mixture of red, green, and blue lights was projected onto the other half. The task was to adjust the intensities of the red, green, and blue lights until the border between the two halves disappeared, indicating indistinguishable colors.
If two physically different spectra elicit the same cone signals, the subjective color will be identical. The relative cone signals, varying on a scale from 0 to 1, can be matched metrically by adjusting the intensities of the three narrowband spectral lights. The precise peak wavelengths of the red, green, and blue lights are not critical, as long as they are distributed across the spectrum to differentially affect the three classes of cones.
The success of trichromatic theory is exemplified by color matching, where any color on a device such as a computer screen is produced by mixing variable intensities of red, green, and blue primaries, which forms the basis of the RGB color coding scheme.
Opponent Colors and Color Aftereffects
While trichromatic theory suggests three primary colors, some scientists and artists consider there to be four unique hues: red, green, yellow, and blue. This concept relates to opponent colors and can be demonstrated through color aftereffects.
Fixating on a white cross within an image with altered colors results in seeing the same image in the complementary colors once the original image is removed. Eventually, the colors fade, revealing that the original image was black and white. This phenomenon, known as a color aftereffect, is related to opponent colors.
The four unique hues were postulated by Hering in the 19th century, based on subjective introspection. Hering argued that while colors like orange (yellowish-red) and purple (reddish-blue) are conceivable, it's impossible to imagine a greenish-red or a yellowish-blue. This concept leads to the idea of a color wheel with two opposing axes: green versus red and blue versus yellow.
The impossibility of a greenish-red arises from the opponent nature of these colors. If colors are plotted on a scale from to , where represents full green and represents full red, a color cannot simultaneously have both negative and positive values. Therefore, a color must lean towards either green or red, with 0 representing an achromatic state (white or grey).
Color aftereffects can be explained using the concept of adaptation in neural mechanisms observing these opponent axes. Staring at a bright green surface for several seconds stimulates the green end of the opponent mechanism, causing the neural mechanisms signaling green to become adapted. When switching to a bright white surface, the adapted green response is smaller than the red response, resulting in a biased perception towards red.
The same mechanism applies to the blue-yellow axes. Adaptation to bright blue leads to an afterimage in the yellow direction, and vice versa. The color aftereffects are evident in the example with the photographs of fruit, where greenish versus reddish and bluish versus yellowish aftereffects are observed.
Reconciliation of Trichromacy and Opponent Colors
The apparent conflict between Young-Helmholtz's trichromatic theory and Hering's opponent colors can be resolved by understanding cone signals and retinal circuitry. Opponent processing can be produced by comparing cone signals. Two chromatic channels correspond to blue versus yellow and red versus green.
The red versus green channel (L-M) compares signals from the L cone and the M cone. If the signal is stronger from the L cone, the perception moves towards red, and if it's stronger from the M cone, it moves towards green. This corresponds approximately to the green-red opponent axes.
For the blue-yellow channel, there is no yellow cone. However, summing the signals of the M and L cones results in a peak sensitivity in the yellowish end of the spectrum. The blue-yellow axis is S versus L+M. The yellow end of this axis means stronger signaling from L+M compared to S, and the blue end is stronger signal in S compared to L+M.
Additionally, there is an achromatic channel, black versus white, which corresponds to the luminance channel. The luminance channel integrates the input of the M and L cones, disregarding differences between them, but simply taking the sum of many L cones and using that as a signal for luminance.
Neural Evidence for Opponent Processing
Evidence for color opponent neurons includes inputs of opposite signs from cones of different spectral classes. Recordings from retinal ganglion neurons since the 1960s have shown such effects, with similar results in parvocellular neurons in the LGN.
Stimulation with different intensities at different wavelengths shows that wavelengths above approximately nm elicit net excitation in L+ cells, with excitation growing as wavelength increases. Conversely, moving towards the green wavelengths results in progressive inhibition. M+ and L- cells have more or less the opposite profile of response.
A classic example is blue-yellow potency, originally described as blue-yellow cells with a single, large receptive field where all regions of the field receive excitation from S cones and inhibition from the summed M and L cones. More recently, more variations of S cone inputs combined with various others have been found. Essentially, there is an L versus M axis (L+ and L-) and an S+ versus L+M axis corresponding roughly to blue-yellow.
A wider stimulation of the entire receptive field with white light results in little response compared to stimulating only the centre, which gives a clear excitatory response. This illustrates the classic centre-surround receptive field organisation. Stimulating with an annulus of light just on the surround would result in inhibition.
Stimulating the entire receptive field with long wavelength light stimulates the L cones more than the M cones. The L cone feeding the centre will be strongly excited, while the M cones feeding the surround will be only weakly excited, resulting in net excitation. Conversely, stimulating with short to middle wave light stimulates the M cone significantly more than the L cone, resulting in weak excitation of the L cone and strong inhibition from the M cone, leading to net inhibition.
In dichromatic mammals, the orange and green blobs (representing M and L cones) would be the same. Centre-surround circuitry remains. If it's an on-centre cell, there will be an excitatory pathway via the bipolar cell from a single cone in the centre of the receptive field to the major ganglion cells. Multiple cones in the surrounding region of the receptive field have an inhibitory pathway, again via bipolar cells.
In primates, a mutation resulted in the split of a single M or L cone class into two cones: the M cone and the L cone. Input in the foothill region, at least to the centre of the receptive field from a single cone. Cones randomly either M or L regarding the surround. The input may come from a mixture of M and L cones. Given at least a single M cone in the centre leading to excitation, even if there is a mixture of M and L cones in the surround, the net wavelength sensitivity is going to be different from that of a pure M cone centre, producing wavelength opponency.
Humans and some other primates have both the M versus L cone opponent pathway via the midget ganglion cells and the short wave cones opposed to middle/long wave cones. The midget ganglion cells project to the parvocellular layers of the LGN, where M versus L processing can be found. The koniocellular layers receive input from a different class of retinal ganglion cells.
These wavelength opponent subsystems maintain synaptic terminations in different layers or sublayers of the primary visual cortex, indicating the preservation of wavelength opponency through the LGN and up to the cortex, potentially serving the encoding of colour cues.
This synthesis of the original trichromatic theory with opponent processing, based on four unique hues, seems comprehensive. However, complexities related to color perception make the story less complete, particularly when considering color constancy - which is addressed in the next segment.