Luminance was introduced by the CIE as a photometric analog of radianc
e. This implies that an additive spectral-luminosity function characte
rizes the human observer. In practice, many different spectral-sensiti
vity functions characterize human vision, although few produce the add
itive spectral-luminosity function V(lambda), which is suitable for us
e in practical photometry. Methods that give rise to additive spectral
-sensitivity functions that most resemble V(lambda) tend to have in co
mmon the use of spatial or temporal frequencies that will discriminate
against signals from the short-wavelength-sensitive cone pathways or
against signals in other chromatic pathways. Some of the difference am
ong results obtained with different techniques seems to reflect the ex
tent to which the methods can bring about changes in the state of chro
matic adaptation, but it also seems likely that not all tasks tap the
same postreceptoral mechanisms. Psychophysical evidence is equivocal r
egarding the nature of the postreceptoral mechanisms: some evidence su
ggests just three mechanisms, one of which has a spectral sensitivity
that is like V(lambda); other evidence suggests the existence of multi
ple mechanisms with different spectral sensitivities. Physiological re
cordings from neurons in the macaque's visual pathway suggest that the
properties of the magnocellular system may be sufficient to account f
or spectral-sensitivity functions measured with the techniques of hete
rochromatic flicker photometry, minimally distinct border, and critica
l flicker fusion. These are the psychophysical methods that yield spec
tral sensitivities that are most like V(lambda). Other methods of meas
uring spectral sensitivity seem more likely to depend on signals that
travel through the parvocellular system.