Two completely different sources suggest the superiority of the projective model: On the one side, a projective model can predict the coordinates of the adjustments within the changed context better than a linear or affine one. It has to be noted, that none of these models describes the data as well as an unrestricted model and that this finding can be shown clearly only for the pure context-effects, however.
The context-effects are masked by greater and unsystematic memory-effects, which can be explained by the forming of prototypical representations of the target colors in memory (see Siple & Springer, 1983):
One of the reasons for the remarkable effects of memory on the reproduction of the colors can be found in the apparatus used. The transmission of the LCTF is unstable in the course of time, although great effort was taken to minimize the amount of those fluctuations. Therefore, the luminance of the monochromatic ray is not always the same even with identical adjustments. Since the stimulus presented does vary somewhat, the subjects may form a prototypical and stable representation of the colors in memory, which can differ from the color presented originally.
On the other hand, none of the behavioral measures indicates that the production of matches to highly saturated colors poses a problem to the subjects. This is valid for the investigation of the number of actions taken as well as the amount of time needed for the completion of the matches. Furthermore the matches to highly saturated stimuli are rated especially well by the subjects. So there clearly seems to exist a match for each stimulus, which can be predicted only by a projective mapping.
Although there appear some problems using the newly designed apparatus, the use of a LCTF proves to be of great value nevertheless: It allows for the production of a relatively large beam of monochromatic light (diameter 2 cm) at sufficiently high levels of luminance, whose spectrum can easily be modified by the use of a computer. No mechanical parts have to be driven.