| THE AMBIGUITIES OF COLOUR:
WITTGENSTEIN AND LICHTENBERG ON COLOUR AND COLOUR PERCEPTION
[Published in c Gefwert & O Lagerspetz (eds.), Wittgenstein and Philosophical Psychology. Uppsala Studies of Philosophy Vol. 55, pp. 223-235. ISBN 978-91-506-2046-7. ISSN 1402-1110.]
Ludwig Wittgenstein nourished a life-long interest in colour.1 His most systematic treatment of the topic is included in the posthumous work Remarks on Colour, in fact a series of notes that he started writing around the New Year of 1949/50 as he was visiting his family (for the last time, as it turned out) in Vienna.2
His Remarks are usually associated with Goethes Farbenlehre, or Theory of Colour.3 Wittgenstein’s interest in Goethe was, in this case, above all due to the fact that the Farbenlehre resembled his own enterprise in one specific respect: it was an attempt to treat colours from some perspective other than that of Newtonian physics. But one can also say that Goethe’s most important contribution to Wittgenstein’s thought on this score was not specifically Goethe’s at all. What above all captured Wittgenstein’s interest were two letters that Goethe received from others. One was from the painter Philipp Otto Runge (included in the Farbenlehre) and the other – the actual point of departure for Wittgenstein’s own reflections – was sent to Goethe by Georg Christoph Lichtenberg and was subsequently published.
Lichtenberg was in his lifetime (1742—1799) renowned as a physicist. Later generations remember him as the writer of journals, essays, and aphorisms, works of an acute mind that had considerable influence on subsequent German style. Wittgenstein, too, was helped in his thinking by these remarks written down a century and a half earlier.
Those who study colours theoretically must first of all realise that there is room for questions of principle about colour that cannot be answered by empirical investigations of light, reflecting surfaces, the neural system etc. In philosophy, if we ask ourselves what colour is, the question we are really asking is: what is the concept of colour? In other words, what is meant by ‘colour’? What do we mean, for instance, when we speak of two objects being the same colour, or when we speak of one colour being lighter than another?
Note that we cannot leave this conceptual investigation aside and simply proceed to science proper for empirical solutions. If a scientist says he or she has an empirical answer to the question what colour is, we will need to see what question it is the scientist has actually answered. We must investigate whether what the scientist calls ‘colour’ is the same as what we would call ‘colour’ in non-technical contexts. A negative answer will not imply that the scientist’s results are falsified. However, those results will apply only to some cases but not others. Other questions will remain.
One of Wittgenstein’s central philosophical insights is precisely that the analysis of what a concept means involves a description of its role, of how the concept is used. An analysis of colour concepts must, among other things, pay attention to our ways of distinguishing between colours, as well as looking at the circumstances in which we would say, on the contrary, that two colours are the same.
Alas – here, as very often in philosophy, the fish will be the last to understand that they live in water. The ambiguities of colour escape us in spite of their familiarity – or perhaps just because of it.
Wittgenstein’s philosophical interest in colour arose originally from his study of logic. Colours, too, have a kind of logic, an order that strangely enough gives the impression both of logical necessity and of natural necessity at once (See TLP 6.3751; cf ‘Some Remarks on Logical Form’, Wittgenstein 1993). Art teachers and colour scientists frequently illustrate this logic in graphic renderings such as colour circles or colour solids. Wittgenstein’s manuscript Philosophical Remarks of 1930 includes his own version (Wittgenstein 1975). – The colour circle demonstrates ‘necessary’ relations between colours; for instance, it shows that green lies between blue and yellow, and that green and red as well as blue and orange are complementary colours.
Runge gives us an example of such logic in his letter to Goethe (reproduced in the Farbenlehre, Goethe 1989, 264-271, and in turn commented by Wittgenstein in his Remarks on Colour I.21). He claims that it is impossible to imagine reddish green. Indeed, the expression ‘reddish green’ gives us the same feeling as hearing someone talk about a southwesterly north wind (Goethe 1989, 267). This as opposed to, say, bluish green. Runge demonstrates the point with the help of a colour circle (ibid., 266-267). The colour circle allows for a direct passage from green to blue, while you must go through another colour in order to get from green to red. Every sector that might include the colour ‘reddish green’ is as it were already taken up by some other colour such as brown.4
Runge also notes another necessity that cannot be adequately demonstrated by means of the colour circle. There are opaque and transparent colours. There can be transparent red but no transparent white (Goethe 1989, 268). If ‘white wine’ were properly white it would look like milk. This point gives in turn rise to a large number of paragraphs in Wittgenstein’s Remarks.5
These statements about colours resemble geometrical and mathematical propositions in the sense that they are not just statements about specific individual objects. They deal with colours as such, just as geometry deals with extended bodies as such, not with individual triangles and cubes. They are ‘timeless’ (I.1). Wittgenstein speaks of colour geometry. It describes so to speak the behaviour of ideal colours in ideal space.
And Wittgenstein wonders what kinds necessity are involved. Does colour geometry deal with how we speak of colours and think of them, or does it express some facts of nature of some kind? Could there be an alternative colour geometry that allows for reddish green or transparent white (cf I.9; I.32, I.46)? These questions have a central place in Jonathan Westphal’s Colour: A Philosophical Introduction, a work that uses Wittgenstein’s Remarks as its starting point (Westphal 1991).
However, in the Remarks on Colour, the ‘timeless’ propositions of colour geometry are contrasted with descriptions of colour judgments in a natural context.
Contrasts between colour geometry and the natural perception of colour in fact constitute a main theme in the Remarks on Colour (as stressed by McGinn, 1991). Natural perception of colour involves ambiguities that colour geometry does not account for. For instance: is white in bad light the same as grey in good light? Both ‘yes’ and ’no’ might be correct in the right circumstances. The point can be generalised. Wittgenstein speaks of the indeterminateness of our concept of sameness of colour (I.56; cf I.59, I.49).
Lichtenberg’s letter of October 1793 to Goethe seems to have given Wittgenstein a crucial impulse. The correspondence deals with the effect of illumination on colours, including white. Lichtenberg suggests that Goethe should examine what it is we call white. In other words he proposes a conceptual investigation, as opposed to Goethe’s physiological and psychological approaches. Lichtenberg suggests a definition of ‘white’. White is absolutely, optically white, perhaps the colour of the snow one would see on the highest peak of the Alps on a bright day. Few people have actually seen it. The ‘white’ we usually see is impure – but we call it white. We recognise this colour in all kinds of illumination:
For instance, I consider this sheet of paper to be white, in the deepest darkness, even at night in the weakest of starlight, in tallow- wax- and lamplight, in bright sunshine, at the sunset, in snow and rain, in the forest and in the wall-papered room […].
Lichtenberg now formulates an early version of the theory of colour constancy – in other words, of the assumption that we unconsciously adjust our colour perception:
We don’t, of course, notice this, because in all judgments based on visual perception, [the elements of] judgment and perception are completely amalgamated, so that later in life we find it impossible to pull them apart again; we believe at each moment that we see what is really just an inference.6
In the scientific language of today one might speak of a difference between sensation (sense data) and cognition. According to the theory of colour constancy, our colour sensations are adjusted by an unconscious cognitive mechanism that creates the final perception of colour.
The phenomena known as colour constancy may, for many useful purposes, be said to consist of two separate sets of phenomena; these may be called illumination constancy and the recognition of colour (Lopes 1999). Illumination constancy means the physiological adjustment of the visual system to prevailing lighting conditions. The eye adjusts itself to the overall illumination indoors and out of doors, in the morning, at noon and in the evening. The recognition of colour, on the contrary, has to do with local variations within the parameters of given lighting conditions. Human beings are able to distinguish between, on the one hand, genuine differences of colour and, on the other hand, contrasts caused by shadows and reflections. While illumination constancy involves physiological adjustments, our ability to distinguish between a stain and a shadow is not produced by specific physiological changes. Indeed it seems misleading to call it adjustment at all. It is rather an aspect of our general ability to orient ourselves and deal with a three-dimensional environment.
When Lichtenberg (as quoted) writes about the effect of illumination he is speaking of illumination constancy. Further down in the letter he discusses colour recognition:
Outside my window there is a white chimney, the both visible sides of which rarely get the same amount of light. Sometimes, when the one side looks yellow or bluish, I ask people of otherwise very good judgment [!] what colour the chimney is. Usually they reply that it is just as white on the one side as on the other, only that the sun is shining on the one but not on the other.
The two sides of the chimney are optically different, but even people ‘of otherwise very good judgment’ will think that they are the same colour. According to Lichtenberg, this is because ‘judgment and sensation’ are ‘amalgamated’. Lichtenberg seems to think that the two sides of the chimney are actually of different colours. Or perhaps he means that we ought to think so if we, by ‘colour’, meant the optical stimulus that actually meets the eye.
The example demonstrates the ambiguity ‘the same colour’. In most contexts – but not in all – we would probably say that the two sides of the chimney are the same colour, but that they look different in sunshine.
On the other hand, is it correct here to say that they look different in the first place? In one sense they do: for instance, the two sides of the chimney would be represented by different colours in a colour ‘photo (cf III.56). In another sense, they do not. For instance, what we see would not make us believe that the chimney is not white on both sides. The chimney looks exactly the way a white chimney ought to look given the lighting conditions. (On the other hand our judgment of the illumination is in itself influenced by our belief that the two visible sides of the chimney are of the same colour.)
Our ability to make these judgments demonstrates something crucial about colour: it shows us that we do not, in a natural context, judge the colours of objects by seeing whether they match a timeless model (such as Lichtenberg’s ideal white). Rather we judge the colours of objects in a context of illumination and shadows.
The context-dependent (or ‘ecological’) character of colour is shown nicely in the context of a phenomenon that furnished the actual reason why Lichtenberg wrote to Goethe. He was commenting on an essay on coloured shadows that Goethe had sent him.
The first scientific experiments on coloured shadows were reported by Otto von Guericke (1602—1686), physician and the burgomaster of Magdeburg, who described what is today known as the Guericke phenomenon. An excellent modern description is included in Scientific American 1986 (Brou et al. 1986).
In fact it is not difficult to find instances of coloured shadows in your surroundings once you know where to look for them. But you can also produce coloured shadows at home, quite industrially if you like (in the Farbenlehre, Goethe included several descriptions of his experimental arrangements). The main principle is that two light sources of different colours should illuminate the same area; for instance a candle (orange) and sunlight (neutral) indoors on a rainy day. (For variation, you can substitute electric lights, using the colours of your choice). Place the candle close to a white wall that receives indirect sunlight from the window. Place your hand between the candle and the wall and project a shadow. Now place another hand on the other side of the candle to cast another shadow. The two shadows should now be differently coloured. On areas where candlelight is blocked but natural light is unimpeded, the shadow will be blue. Where natural light but not candlelight is blocked, the shadow will be orange (or brown). Where both lights are blocked the shadow is black.
Scientists have disagreed about whether the coloured shadow should really be called coloured; according to some, the colour is an illusion. This is because the blue shadow represents a physical stimulus (or spectral power distribution of electromagnetic radiation) that would be perceived as grey if the surrounding illumination was neutral. The orange environment, however, makes us expect a brownish shadow. In relation to the surrounding areas, the shadow has an under-representation of orange. Under-representation of orange equals over-representation of the complementary colour, i.e., of blue. Thus the shadow is blue because it is less orange than grey should appear in the overall illumination.
The question whether the shadow is really coloured is a typical conceptual question. It is not settled by collecting more data; rather one will need an investigation of the concept of ‘real colour’. In the case of tangible, three-dimensional objects we typically distinguish between their real and apparent colours. The reason is no doubt the fact that you can study such objects in various kinds of illumination. A blue object in orange light would look the same as the blue shadow in the candle experiment. They would also have the same colour in a photograph. But you can move the blue object outdoors, or move a light closer to it, and study its real colour. You cannot move the shadow or study it in a better light.
This reasonably motivates the conclusion that the distinction between real and apparent colour does not apply to shadows; in other words, a coloured shadow simply has the colour it looks like having. However, in discussions of coloured shadows it is frequently contended that coloured shadows are illusions, physiologically produced and comparable to after-images.7 The idea that the blue shadow is not really blue but rather grey (while the blue object next to it is truly blue) seems to stem from the contention that the same physical stimulus, the same spectral power distribution of light, must always result in the same visible colour. However, as we have already seen, colour constancy means that such a principle cannot be upheld in other contexts.
There will be a simple answer to the question whether the blue shadow is really blue, if we accept that ‘blue’ means the same both for shadows, lights, and objects. ‘Blue’, I suggest, means the over-representation of blue or, in other words, the under-representation of orange in the given milieu. (‘Blue’ is, then, like ‘big’. ‘Big’ means comparative, not ‘absolute’ size. ‘Blue’ means comparative, not ‘absolute’ colour.)
In this way, coloured shadows turn out to be a special case of the general phenomenon of colour constancy (see Brou et al. 1986). Or rather, both are illustrations of the same general point: namely, that the colours of one’s surroundings are judged by means of comparisons. We cannot identify the colour of a given spot conclusively by looking at the coloured spot in isolation. We rather identify its colour in relation to the overall ‘colour situation’ in the surroundings.8 The coloured shadow emerges in a case where two different colour environments are superimposed on one another.
A coloured object darkens the incident light by absorbing parts of it. The absorbed parts of the spectrum correspond to the complementary of the colour that the object has. To see the colour of an object is, from this point of view, to see the under-representation of its complementary in the object. To express it in a schematic way, ‘blue’ means ‘bluer than …’ (or perhaps rather ‘less orange than …’) just as ‘dark’ means ‘darker than…’ and ‘big ‘ means ‘bigger than …’. And the point of comparison consists of the visible, coloured surroundings with its lights and shadows.
A colour chart is, in this perspective, not a context-free representation of colours. It rather represents an attempt as far as possible to control the context within which colours are viewed and compared. The surrounding surface is ‘neutral’ and the colour samples should be studied in ‘normal’ illumination. But here the very definitions of normality and neutrality are based on practical and conventional considerations. When colour scientists register correlations between perceived colour and wavelength (spectral power distribution of light) they are dealing with colours in specific, standardised conditions rather than with colour in the abstract and in general.
By ‘standard conditions’, colour scientists generally mean a normally sighted person looking at an opaque, flat, matt surface in indirect light of a composition conventionally defined as neutral. The spectral composition of ‘neutral’ light is standardised by means of international agreements. Very often this is the composition known to colour scientists as D65, which approximates the spectral power distribution of daylight in Northern Europe. Moreover, it is generally required that the surface should take up exactly two degrees of the visual field, and that it should be surrounded by a specific shade of grey.9
Standard conditions defined by colour scientists are not ‘normal’ if, by that word, we mean what is usual, typical or common. The experimental setup is in fact not supposed to approximate the conditions that obtain for colour vision in most natural settings – even less, to constitute a model for all possible cases of colour perception. The point is to create controlled, reproducible conditions for comparisons between colour samples – not to work towards a general answer to the questions, “What is colour?” or, “What is the relation between colours and physics?”
This should give pause to those philosophers (and some colour scientists) who would treat the experimental results as blueprints for future reductionist accounts of colour. While law-like connections between physical stimuli and perceived colour can be established in conditions especially designed for the purpose, questions remain about what we can say about the status of colour judgments outside those conditions. Rather than privileging cases of just one type, philosophers need to examine their relation to other cases.
Going back to Wittgenstein’s remarks.
Wittgenstein highlights the contrast between colour geometry, where colours are treated as independent of each other, and normal colour perception where the colours of individual spots are defined in relation to the whole of one’s surroundings. In this context, Wittgenstein considers the question of how the colours of a natural visual scene are reproduced in paintings. A discussion of painting illustrates the contrasts between the two ways of identifying colours and talking about them.
Ideally speaking, in colour geometry, it is true that white is the lightest colour. On a flat surface in homogeneous illumination – as in the French Tricolour – white will be the lightest part of the surface. But it does not follow that white objects are always the lightest parts of a picture. Wittgenstein imagines a painting where a white sheet of paper receives its illumination from a blue sky. The sky will have to be painted lighter than the piece of paper (I.2).
The painted canvas can be called an interface, a point of transition from colour geometry to the natural perception of colour.10 An artist’s palette represents colour geometry in an almost pure form. Colours are placed there, in good order against a neutral, flat surface. But when the artist dips his brush in paint and places a speck of colour in the right place the effect will depend on where on the canvas the pigment is placed. What in colour geometry is defined as grey will perhaps become a white area placed in a shadow, blue placed in orange light, or a highlight on a polished grand piano (cf III.95). The question what colour this spot in the painting is, will be ambiguous. One can think of ‘colour’ either as the specific pigment placed on a piece of fabric (grey) or as the colour of the object depicted in the painting (white, blue, black, etc) (cf III.77-78).
Wittgenstein remarks that Rembrandt did not use gold paint in his Man with a golden helmet (III.79). (This famous painting in Dresden was then attributed to Rembrandt and said to be a portrait of Rembrandt’s brother. Later studies have identified the real painter as one of his students.) The colour of ‘this spot’ is in one sense brown and, in another sense, gold. The painter has created the impression of gold by means of a combination of various brown, black, and yellow pigments. In this journey from the palette to the canvas, yellow – or brown or black! – turns to gold.
Wittgenstein’s interest in the transition from the palette to the canvas may, too, have been stimulated by his reading of Lichtenberg.
Lichtenberg writes, as we have seen, about the difficulty of distinguishing between ‘judgment’ and ‘sensation’ in colour judgments. He continues, immediately after the passage already quoted (I include some of that passage again):
[I]n all judgments based on visual perception, [the elements of] judgment and perception are completely amalgamated, so that later in life we find it impossible to pull them apart again; we believe at each moment that we see what is really just an inference. This is the reason why bad portrait painters smear the faces all over with flesh colour. They can’t imagine that a human face may have blue, green, yellow, and brown shadows. And the cuffs are painted in such a spotless manner that only their placement and outline betrays that the stain of chalk that the artist has thrown there is meant to be a cuff.
Lichtenberg now proceeds to the remark about the white chimney quoted earlier. The homogeneity or sameness of two samples of white in “judgment” is contrasted with the difference of “perception” when we see the two samples without preconceived notions. The successful painter must abstract away from his or her “judgment” and just try to see a visual pattern. – Hence, Lichtenberg says:
[F]or the same reason, the technique of colouring is studied more easily from the works of great masters rather than from nature, because the colour is already there on the canvas, divorced from judgment, and it can be studied like any other piece of coloured fabric, in all kinds of light and in all kinds of angles against it. But here judgment must first be divorced from perception, which not everyone can do.
Wittgenstein’s remark in I. 60 of the Remarks might be directly inspired by Lichtenberg’s discussion.11 He imagines that we actually cut up a painting in pieces and use the pieces as a pointillist jigsaw puzzle. The position of a given fragment in the jigsaw puzzle will decide whether it will be seen as part of a grey object, or as a piece of a white object in shadow, or as a highlight, or still something else (also see III.53).
Here the discussion is connected to a well-known issue within the philosophy of perception. Wittgenstein suggests the jigsaw puzzle as an analogy of how a visual scene is actually, in the last analysis, mediated to the perceiving subject. We can try to think of our visual field as consisting of a large number of tiny spots, each with a specific colour. The background of this thought experiment lies in the Phenomenalist project which constitutes an important motif in the empiricist tradition. The idea is that our visual sensations really consist of exactly this – coloured spots; then we (or rather our brains) interpret our sensations as indicating an array of objects in three-dimensional space. A strict description of our visual sensations would simply give us the colours of the individual spots but not impose a description of objects on them (cf I.61).
Wittgenstein of course deeply disagrees with the Phenomenalist project. If the colours of objects cannot be identified independently of the surroundings, then the same thing should apply to the coloured bits of mosaic. Even if one were to look at each of them in isolation from the others, the bits would be seen in some surroundings (see III.108; III.58). However, the idea of a visual mosaic required precisely that the colour of each piece should be identified regardless of everything else around it (also see III.68-69; III.262). But exactly this is impossible. For this reason (too), Phenomenalism cannot give us a coherent theory of visual perception.
Remarks on Colour was not published in a shape that its author would have approved of. Nevertheless the work is much in line with Wittgenstein’s philosophy in general. One can recognise his striving never to look at a phenomenon in isolation. A colour stands out, ‘shines’, in its surroundings, just as eyes only smile in a face (see I.55).12
Such holistic emphasis represents a permanent tendency in Wittgenstein’s work. It cannot, for instance, simply be traced to the influence of Lichtenberg. At the same time Lichtenberg seems to have struck a philosophical chord that found a resonance in Wittgenstein.
In conclusion, this discussion suggests that our colour judgments never simply concern patches of colour in abstraction. It will be misguided from the start to look for some reductivist account of colour in terms of some physical property of an isolated coloured patch or in terms of some isolated spectral property of light. This is in fact not news to experimental colour research. There, the context dependence of colour constitutes a central research topic, with a range of applications in graphic, electronic, textile and other industries. In that research, the trick is not magically to make a pigment’s surroundings vanish but rather to establish conditions where they are known and fixed. Thus the context relativity of colour judgments lies at the centre of colour science. It remains for philosophers to recognise its implications to the concept of colour.13
Brou, Philippe, Sciascia, Thomas R., Linden, Lynette & Lettvin, Jerome T. (1986): The Colors of Things. Scientific American 255:3, 80-87.
Goethe, Johann Wolfgang von (1989): Zur Farbenlehre. Sämtliche Werke Bd 10. Carl Hauser Verlag, München.
Lee, Alan (2005): Colour and Pictorial Representation. The British Journal of Aesthetics Vol 45, 49-63.
Lichtenberg, Georg Christoph (1992): Briefwechsel. Band IV. C.H. Beck, München.
Lopes, Dominic M. McIver (1999): Pictorial Color: Aesthetics and Cognitive Science. Philosophical Psychology Vol 12, 415-428.
McGinn, Marie (1991): Wittgenstein’s Remarks on Colour. Philosophy Vol 66, 435-453.
Rydefalk, S. & Wedin, M. (1997): Literature Review on the Colour Gamut in the Printing Process. PFT-rapport nr 32. Forskningsområde–Tryckkvalitet. STFI, Stockholm.
Westphal, Jonathan (1991): Colour. A Philosophical Introduction. Second edition. Basil Blackwell, Oxford.
Wittgenstein, Ludwig (1996): Familienbriefe. Hrsgg. von Brian McGuinness, Maria Concetta Ascher, Otto Pfersmann. Hölder-Pichler-Tempsky, Wien. – Here abbreviated as F.
--- (1975): Philosophical Remarks/ Philosophische Bemerkungen. Blackwell, Oxford.
--- (1978): Remarks on Colour. University of California Press, Berkeley and Los Angeles. Here abbreviated as ROC.
--- (1993): Some Remarks on Logical Form. In Ludwig Wittgenstein, Philosophical Occasions 1912-1951, s 29-35. Hackett, Indianapolis.
--- (1922): Tractatus Logico-Philosophicus. Routledge & Kegan Paul, London. Here abbreviated as TLP.