Why simulations of colour for CVD observers might not be what they seem

摘要

A common task in universal design is to create a 'simulation' of the appearance of a colour image as it appears to a CVD observer. Although such simulations are useful in illustrating the particular problems that a CVD observer has in discriminating between colours in an image, it may not be reasonable to assume that such a simulation accurately conveys the experience of the CVD observer to an observer with normal vision. Two problems with this assumption are discussed here. First, it risks confusing appearance with sensation. A colour appearance model can more or less accurately predict the change in appearance of a colour when it is viewed under different conditions, but does not define the actual sensation. Such a sensation cannot be directly communicated but merely located on a scale with other related sensations, hi practice we avoid this epistemological problem by asking observers to judge colour matches, relations and differences, none of which requires examination of the sensation itself. Since we do not truly know what sensation a normal observer experiences, it seems unscientific to suppose that we can do so for CVD observers. Secondly, and following from the above, the relation between stimulus and corresponding sensation is established as part of neural development during infancy, and while we can determine the stimulus we cannot readily determine what sensation the stimulus is mapped to, or what the available range of sensations is for a given observer. It is suggested that a similar range of sensations could be available to CVD observers as to normal observers.