Please note that some of this background information has not been updated for several years. Any updates are welcomed and will be gratefully acknowledged!
Light reflected or emitted by external objects enters the eye through the pupil. The optics of the eye form an upside-down image of those objects on the rear, inner surface of the eyeball (the retina). There, a dense carpet of photoreceptors converts light (photons) into electro-chemical signals, which are processed by neural circuits in the retina and transmitted to the brain.
What we are able to perceive is limited initially by the nature and arrangement of the photoreceptors. Human night-time vision depends on sensitive rod photoreceptors, while human day-time colour vision depends on the three types of cone photoreceptors.
The three types of cone photoreceptor are commonly referred to as red, green and blue. However, since the attachment of colour names to cones can be misleading (see Note 1), most psychophysicists prefer to call them instead the the long-wavelength-sensitive (L), middle-wavelength sensitive (M) and short-wavelength-sensitive (S), respectively. These names simply refer to the relative positions in the visible spectrum (see Note 2) in which each type is maximally sensitive.
Individual cones signal only the rate at which they absorb photons, without regard to photon wavelengths. Though changing the wavelength of a photon changes its probability of absorption, it does not change the neural effect that it has once it has been absorbed. Thus, single photoreceptors transmit no information about the wavelengths of the photons that they absorb. Our ability to perceive colour depends upon comparisons of the outputs of the three cone types, each with different spectral sensitivity. These comparisons are done by the neural circuitry of the retina.
The normal observer, with three types of cones, can match any spectral light to a mixture of three fixed-colour primary lights (one of which may have to be added to the spectral light to complete the match). The match is determined at the cone level: the total quantal catch produced by the three primaries in each of the three cone types is the same as the quantal catch produced by the spectral test light. The three functions relating the matching intensities of the three primary lights to the wavelength of the spectral light are known as the colour matching functions or CMFs.
If we knew the sensitivity of each of the three cone types to the three primary lights, we could reconstruct the cone spectral sensitivities from the three CMFs by a simple linear transformation. Those nine sensitivities, however are unknown. They are traditionally estimated (Young, 1807; Helmholtz, 1866; König & Dieterici, 1886, 1893) by assuming that the three types of dichromats -- protanopes, deuteranopes and tritanopes -- all possess reduced forms of trichromatic vision, lacking one of the three normal cone types -- the L, M and S, respectively. The missing values can be obtained by determining the locations of the confusion points of the three types of dichromats in the trichromatic chromaticity diagram (see Wyszecki & Stiles, 1982), but a better method is to measure the dichromat's cone spectral sensitivities directly.
The cone fundamentals depend on estimates of those missing values. Cone fundamentals can be thought of as CMFs of three imaginary matching lights, each of which exclusively and separately stimulates the one of the three cone types.
(Note 1) It is misleading for two main reasons. First, because stimulating a long-, middle- or short-wavelength-sensitive cone uniquely does not necessarily lead to the corresponding red, green or blue sensation. Second, because the cones are not maximally sensitive in the red, green and blue parts of the spectrum. The blue cones are most sensitive in the violet, and the red cones in the yellow-green part of the spectrum.
(Note 2) The visible spectrum spans ultraviolet, violet, blue, blue-green, green, yellow, orange, red, and infra-red.