In June 1983 the Swedish Council for Coordination and Planning of Research, through their comitte for Future Oriented Research, arranged an interdisciplinary international conference at Lund University, devoted to the theme: “An inventory of present thinking about parts & wholes.” The following article was my contribution to the discussions.

Colour presents an everlasting challenge to anyone wanting to understand it. This is due to the fact that colour, being a sense-quality, is something immediately given, in its very essence irreducible. You could tell a lot about “red” to a colour-blind person but you could never explain what it is like to see red. This can only be experienced.

However, for the inquiring mind there are many adventurous paths to follow through the realm of colour. I will put down one such line of thought deriving from my own efforts to understand colour.

Looking back I consider that my interest in these matters began when I got in my hand a copy of J.W. Goethe’s “Zur Farbenlehre”. Being myself a physicist I was provoked by Goethe’s polemics against the way Newton had treated the problem of light and colour in his famous “Opticks”. Now, Goethe was an artist and made a most impressive attempt to introduce an artistic method of inquiry into the natural sciences, much to the horror of the orthodox scientists. This meant, among other things, that he advocated what can somewhat loosely be called. a holistic approach against the atomistic conceptions of the Newtonian school. Goethe opened my eyes to the limitations of the established way of discussing the phenomena of colour – but, on the other hand, I still wanted to make myself clear about the central question: What is the physics of colour? To what extent is it meaningful to use physical concepts in colour science?

On a fine summer-day you may be lucky to enjoy the beauty of the variegated flowers in the grass of the meadow. As you look closely at one of these flowers the question may suggest itself:

“Why is this buttercup yellow?” (1)

This seems to be a perfectly reasonable question concerning the nature of colour. It can be answered in many ways. Let us start with the most sceptical:

“Well, it isn’t yellow, strictly speaking. The “yellow” is illusory – I mean, it’s something, you imagine. In the physical reality there are no colours. (2)

If you accept this kind of answer, you have to take refuge in physiology, trying to find out how the eye and brain work, in order to arrive at a scientific explanation of colour phenomena. That is, you accept the hypothesis that the sense qualities can be regarded as “products” of the sense organs, which reflect or “reveal” their functioning. A hypothesis which, as I learnt from the philosophers, is far from logically self-evident.

There is also another argument against (2). Colour, as we know it from common experience, is something “out there”. It belongs to the environment, as a property of the objects. Colour makes it possible for us to orient ourselves, to identify and recognize the various objects, and it tells us about materials. To say that colour has nothing to do with physics, then, would be to retreat too far. Colour has to do with the play of radiant energy and material bodies in space and time. Hence there must be physical aspects of colour. Having arrived at this point in the argument I decided to start from scratch and try to answer the question: What is the physical correlate of perceived colour?


In passing by, it is worthwhile to consider that there is at least one undeniable coupling between physics and perception. Suppose we have two small adjacent fields, A and B, each emitting its light into the eye of an observer.

Now, if the two stimuli A and B are physically identical (i.e. there is no known measurements by which they could be discriminated) then they do also have the same colour. The perceptual experience that the two fields have identical quality is explained by the fact that they are physically identical. Physiological details concerning the mechanism of colour vision need not enter the argument (Actually a colour-blind person would come to the same conclusion concerning the identity of the two fields). In fact, you can’t imagine any mechanism capable of systematically distinguishing stimuli. which are physically identical!


But let us return to the example with the buttercup. Another possible answer to question (1) is:

“Because the petals reflect exactly those wavelengths which belong to the yellow part of the spectrum of sunlight” (3)

This answer roots in the idea that the phenomenon of colour corresponds to a property of light radiation which the physicists nowadays express through wavelengths. Thus, each particular hue is uniquely related to a particular. wavelength, the dictionary for translation being given by the spectrum. On closer scrutiny, I found this idea untenable or at least misleading in several respects. First, if you study how the various colour sensations correspond to various spectral compositions of light you find that, for instance, one and the same yellow is obtainable in several ways. In terms of the experiment above, this means that the two adjacent fields may look identical, even if the two stimuli A and B are physically different.

As is nowadays verified on any colour television or computer screen, you can make a pure yellow by adding red and green. That is, a yellow light need not contain any wavelength from the yellow region of the spectrum. You can also design two lights, giving rise to identical sensations (e.g. when displayed as A and B in the experiment above) but still having no single wavelength in common.

The figure shows three examples of spectral energy distributions corresponding to one and the same colour, i.e. the yellow colour of a lemon-peel seen in daylight.

As a matter of fact the colour sensation of a light stimulus does not depend on the presence or absence of any particular wavelength in it.Accordingly, the physical correlate to colour is not wavelength per se, but rather the distribution of energy over the visual range of wavelengths. Moreover, it is not the distribution as such, but the general shape of it which determines colour. If the spectral curve on the whole slopes one way, the light is yellow, if it slopes the other way, it is blue. If the curve is concave upwards the light is red, if downwards, green. Mathematically all this can be beautifully modelled in terms of linear algebra, and I once spent some time playing around with this. However, the essential point for the present discussion is the conclusion that the appropriate physical correlate to colour is found on a high level of complexity, despite the fact that the perceived colour may be simple and homogeneous, as for instance yellow. In colour theory there is no reason to regard monochromatic light, or light of a specific wavelength, as more elementary than, say, white sun light.

Generalizing: It is no inherent property of a natural phenomenon to be “simple” or “complex” — this is dependent on the way we choose to tackle the problem conceptually and formally.


I can’t resist making one more comment on this, before we come to the next great turning point of the argumentation. You never meet spectrally pure light in nature, so why expect that the highly artificial spectrum should determine colour perceptions? It seems more appropriate to put the initial question the other way round, asking:

“Why is spectral light at the wavelength 580 nm yellow to most observers, looking into a spectroscope”

and directly adding the answer:

“Because such light happens to have the same chromatic valency as the light reflected from the sunlit petals of buttercups, cowslips and the like, which are — as we all know — yellow” (4)

Which is a more elaborate version of the typical “Wittgensteinian” answer:

“Because this is what yellow is”(5)


But this reply, however laconic it may be, doesn’t mark the end of this short eventful history. There remain some irritating questions, one being the following:

“Do you really see the light reflected from the petals? To me, as a naive observer, there is no yellow light. It’s just the petal which is yellow!”

This question turns our attention in a new direction. It has since long been regarded as a fundamental and quite puzzling fact of visual perception that the visual world is so stable in spite of the incessantly varying conditions of observation. The so called “colour constancy” means that the objects around us keep their respective colours relatively unchanged in spite of the fact that the spectral composition of the light illuminating them may vary appreciably during the day.

If the light reflected from an object surface was the determinator of the perceived colour of the surface, this colour would constantly change, not unlike a chameleon, following the variations in the illumination. But it doesn’t.

So, dissociating ourselves from the idea that colour should be a property of light, let us preferably look Upon it instead as a property of the objects. This is the hypothesis of “colour as reflectance“.

The role of light is not to be visible, but make visible. This gives us a new answer to the question concerning buttercups:

“Because it is a property of these petals to absorb light at wavelengths below 500 nm more strongly than above.” (7)

This new point of view has an important consequence: If colour corresponds to reflectance (which is an invariant property of the object) then there cannot exist any unique correlation between wavelength composition and colour. Logically, colour cannot at the same time correspond to a specific property of material bodies and a specific property of light radiation.

But still, of course, it is the array of light reaching our eyes which makes it possible for us to perceive the reflectances of the variegated objects around us. (If it is dark we don’t see any colours, not because the objects have no colours but because one of the main conditions for seeing is not fulfilled). The optical array must contain some invariant structure, presumably on a high level of complexity (from the physical point of view) which carries information about the optical properties of the object surfaces.

But, as always when the question is about information, the individual element has meaning only in relation to a certain context. Thus it can be said that each local colour of a scene is related to the total layout of chromatic valencies over the field of view. As far as concerns light in colour perception: the whole governs the parts.

The hypothesis of “colour as reflectance” is attractive but it doesn’t hold true in general. Colour constancy is highly dependent on circumstances, even on the attitude of the observer. Theoretically, as far as I have found in my efforts to formalize the matter, for the correct working of an optical instrument, capable of direct determination of a layout of surface reflectances, you have to assume the validity of certain restrictions as to occurring reflectances. Further, preferably, the light detectors should sample three separate channels over the spectral range. Our eyes don’t, since the sensitivities of the retinal cones overlap to a significant degree.

I have given arguments earlier in this essay to show that the eye cannot properly be regarded as a coarse spectrometer. But it is no ideal reflectance-meter either.’ Colour seems to have its own way of life! in some cases it corresponds to surface reflection of the materials, in other cases it corresponds to wavelength of radiation — in most cases to neither of these. The two extreme alternatives, “colour as reflectance” and “colour as wavelength” are complementary aspects of one and the same whole: the interplay of light and matter on the scene of space and time.


Nevertheless we have reached an important insight. Colour is perceived as belonging to things. Colour is usually colour of something. The primary question in this discourse concerned the colour of buttercups: But the object in question need not be material. This is beautifully demonstrated by the coloured shadows, appearing whenever you have two illuminations of different quality falling from different directions towards an object, for instance as illustrated below.

Say that the transparent piece of glass is red. Then the shadow c will be red, and the shadow b will be blue-green.

If you should like to analyse a coloured shadow, you can ‘t take it out of the scene where it appears. It has no autonomous existence. Suppose you get the idea to isolate it by looking at it through a black tube. Then, again, there is no shadow: What you see through the tube is only a part of the surface onto which the shadow falls. But you don’t see the shadow: It is defined by its border. It is “shadow” only in relation to the surround which is not-shadow. A shadow is not a material object, but it is one of those objects which can have colour. I think it fair to say that a shadow is unitary, it is not composed of some sort of infinitesimal colour elements which add up to form the shadow.

The same is true even when we speak of the colour of a material object: it is the object as “an object” which has the colour, not the atomic constituents of it. Colour doesn’t exist point-wise. Colour always belongs to wholes.


When Goethe was so violently arguing against the use of such concepts as “Rays of Light”, serving to establish some sort of one-to-one correspondence between different aspects of the visual process regarded as a chain of events, I think this was out of a fundamental insight into the nature of colour. In one of the paragraphs of the Farbenlehre he says: “Vor allen Dingen erinnern wir uns, dass wir imReiche der Bilder wandeln.” Colour belongs to the visual world — the world of the eye — which consists of images, pictures, things, shapes, gestalts, i.e. entities of a holistic character.

So the first answer (2) wasn’t totally wrong, after all. Colour is something we perceive imaginatively. But on the other hand this is true of the rolling balls, the swinging pendulum, the yardsticks, clocks and gauges of the physicist. You need a lot of imagination to be able to grasp what physics is all about, and to see the point in the various simple experiments and what they tell us about the lawfulness of nature. The physical world-picture, if it is at all a picture, is not only a mathematical structure but it is an imagination. To be sure, there must be a place for colour in it, as well as material bodies and all other “things” making up the furniture of earth.

I think I had better stop here. Much more could be said concerning the enigma of colour. But I hope I have been able to show that to grasp colour phenomena one has to make use of holistic as well as atomistic concepts. It seems that our faculty of vision comprises both the act of analysis and that of synthesis — thus demanding both types of concepts if it should be completely understood.

I have the feeling that in the last analysis one arrives at a similar conclusion in any science.

© Pehr Sällström 2006.

Article Source :

Originally published in “An Inventory of Present Thinking about Parts & Wholes”, Vol.1, p.113 (1983) issued by the SWEDISH COUNCIL FOR PLANNING AND COORDINATION OF RESEARCH