Assessment of Theory

A Historical Assessment of Land's Colour Theory

While not an official biography, the following is reprinted with the kind permission of the Royal Society.

by F.W. Campbell, F.R.S.†

Land’s rather dramatic demonstrations to the public of his views on colour theory and his lack of references to earlier work in the field generated excessive criticism from the ‘colour establishment’. To them, the dramatic demonstrations that made him appear in the popular mind as an iconoclast challenging accepted ideas were merely vivid examples of the phenomenon of colour constancy that had in essence been described 150 years earlier. Thus, Thomas Young had noted in 1807 that white paper still looks white even when illuminated by the red light from a fire, and earlier still, Gaspard Monge (1746-1818) had demonstrated that patches of colour may dramatically change their appearance as the result of the colouration of their surroundings. The perception of colour was not simply a matter of the physical nature of light, and our sensation of the hue of reflecting surfaces is much more stable -- less dependent on the illuminant -- than we are normally aware. This is analogous to brightness constancy, the processes of light and dark adaptation that make us unaware of the absolute luminance of a surface unless we use an exposure meter to adjust our camera. Land’s contribution was to formulate a precise model -- the retinex theory -- that could relate these perceptual phenomena to neural events in the brain.

Why did Land not quote early workers and try to relate them to his own research? He certainly had a very extensive personal library that contained many books on visual science. He had also studied them, for in conversation he often quoted early workers in discussion and he had an extensive knowledge of the current literature. For example, he was very impressed with the recent advances in cortical neurophysiology by Hubel & Wiesel and Zeki and he often flew to London to sit in on Zeki’s experiments. It might be that aphorisms, and dominated his approach to science.

Today we stand on the threshold of understanding neurophysiologically how the visual cortex achieves this colour constancy that is so essential for our survival in our world of frequently changing colour temperature. Maybe the visual cortex has a global colour temperature device that measures the ratio of activity of the short and long wavelength cones and corrects the resulting perception of surface colours. Maybe the diffuse stray light from the cornea and lens (14%) is sampled by the cones and enables the visual cortex to correct for the global spectral properties of the illuminant, but this is speculation and has not been investigated.

Land knew that he could not achieve this with his colour film so he fitted his cameras with a flash lamp and daylight film. If the objects of interest were within the range of his high colour temperature xenon flash, their colour rendering would be acceptable under most circumstances. He once told me that if flesh tones are correctly rendered, a photograph is almost always acceptable. Readers may have noticed that the best way of adjusting the colour control on their TV set is to do so when a face is present on the screen.

There is one unequivocal statement that can be made about Land’s excursion into visual science: he stimulated a widespread research interest in colour and brightness constancy. By 1992 many visual scientists had become involved in research in this field using modern techniques and theories. An excellent review of this upsurge of interest is given by Kingdom & Moulden (1992). It contains 59 references, almost all of them recent. For the reader with an extensive knowledge of colour science, McLaren (1986) should be consulted. It contains the following conclusion: although the Retinex Theory has not yet been subjected to the essential stage in the development of any scientific theory, i.e. experiments designed to prove it wrong, it is tempting to consider its effects on colour vision theory and on the practice of colorimetry should it emerge successfully from such experiments.

A reader who wishes to learn more about the processing of colour in the visual cortex would gain much insight by reading the article by Semir Zeki (1992). The drawbridge between the psychology of perception (psychophysics) and neurophysiology is rapidly being lowered. Land deserves some credit for his prescience that the retina and cortex were both involved and for his many ingenious experiments that aroused so much interest in scientists from other disciplines.