ColoursBarry Maund

dialectica Vol. 60, N° 3 (2006), pp. 245–268

The Illusory Theory of Colours: An Anti-Realist Theory

Barry Maund†

ABSTRACT
Despite the fact about colour, that it is one of the most obvious and conspicuous features of the
world, there is a vast number of different theories about colour, theories which seem to be
proliferating rather than decreasing. How is it possible that there can be so much disagreement
about what colours are? Is it possible that these different theorists are not talking about the same
thing? Could it be that more than one of them is right? Indeed some theorists, e.g. Leo M.
Hurvich, D. L. McAdam and K. Nassau, say that the term ‘colour’ is used to identify a range
of different properties, e.g. pigments, properties of light, and sensations. Such a view has its
attractions, but it raises the question of what it is that unites these various concepts – what is it
that would make them all concepts of colour? What is it that justifies using the same terms,
‘yellow’, ‘blue’, ‘pink’, mauve’, and so on?
This paper aims to address this question, arguing that its answer supports the conclusion
that the best theory of colour is a form of anti-realism: the Illusory theory of colours. There are
two parts to this thesis, one negative, the other positive. The negative part is that there are no
colours, as they are ordinarily conceived. The positive part is that, nevertheless, the world is
such that ‘it is as if there are such colours’. Such a theory has important implications. One is
that it doesn’t fall neatly into the usual taxonomy of philosophical theories. In particular, it does
not deserve the label ‘eliminativist’. Another is that it allows some space for the views expressed
by Hurvich, McAdam and Nassau, but not quite in the sense that they intend.

Colour, on the face of it, is one of the most obvious and conspicuous and unmiss-
able features of the world. The visual world, the world we see, is a world populated
by coloured objects: it is a world with a rich tapestry of colours. Despite this fact
about colour, however, there are a vast number of different theories about colour,
theories that seem to be proliferating rather than decreasing.
How is it possible that there can be so much disagreement about what colours
are? Is it possible that these different theorists are not talking about the same thing,
and that the opposition is more apparent than real? Could it be that more than one
of them is right? Leo M. Hurvich begins the second chapter of his Color Vision
by first asking ‘What is color?’ and then asking a series of questions: Is colour
something that inheres in objects themselves? Is it related to the light falling on
an object? Is it a photochemical event that occurs in the receptor layer of the eye?
Is it a neural brain-excitation process? Is it a psychical event? Hurvich’s answer
is ‘Color is all these things’.1

†
Department of Philosophy, Arts Building, The University of Western Australia, 35
Stirling Highway, Crawley WA 6009, Australia; Email: [email protected]
1
Hurvich 1981, 2.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350
Main Street, Malden, MA 02148, USA
246 Barry Maund

D. L. McAdam and K. Nassau are other well-known colour theorists who,

while not being so ecumenical as Hurvich, nevertheless say that the term ‘colour’
is used to identify different properties, e.g. pigments, properties of light, and
sensations.2 I have some sympathy with the kind of view expressed by Hurvich,
McAdam and Nassau, but there is also something that makes me want to resist.
It concerns the question of what it is that unites these various concepts – what is
it that would make them all concepts of colour? What is it that justifies using the
same terms, ‘yellow’, ‘blue’, ‘pink’, mauve’, and so on?
It seems to me that answering those questions is the key to answering the
question of what is the right theory of colour. I shall defend this claim, while
arguing that the best theory of colour is a form of anti-realism: the Illusory theory
of colours. There are two parts to this thesis, one negative, the other positive. The
negative part is that there are no colours, as they are ordinarily conceived. The
positive part, is that, nevertheless, the world is such that ‘it is as if there are such
colours’. Such a theory has important implications. One is that it doesn’t fall
neatly into the usual taxonomy of philosophical theories. In particular, it does not
deserve the label ‘eliminativist’. Another is that it allows some space for the views
expressed by Hurvich, McAdam and Nassau, but not in the sense that they intend.

Part I
1. Realism and anti-realism about colours
Alex Byrne and David Hilbert, in Readings in Color, Vol. 1, set out a taxonomy
of philosophical theories of colour: eliminativist, dispositionalist, physicalist and
primitivist theories of colour.3 These theories are taken to be rivals, though the
last three are said to have something very important in common: they are all realist
theories of colour. (In making this claim, Byrne and Hilbert are taking ‘realist’ to
contrast with ‘imaginary’ or ‘false’ or ‘fictional’, rather than ‘idealist’ ‘subjectiv-
ist’ or ‘instrumental’. I would have thought that those forms of dispositionalism
where the disposition is perceiver-dependent are non-realist, in some strong
sense.) The four theories cannot be understood as rivals, however, unless we
understand clearly what the subject matter is that we are talking about, the subject
that we are trying to give an account of. I think that identifying this subject matter
is more complicated than is often supposed, and that in identifying the proper
subject matter, there are a number of important distinctions that need to be made.
One reason for making this clear is given by Van Fraassen, who points out that,
according to one form of anti-realism, the sentences of a theory are false if
interpreted literally, but true if interpreted non-literally.4 On this account, some
2
McAdam 1997, 33; K. Nassau 1998, 3.
3
Byrne and Hilbert 1997, xi–xii.
4
Van Fraassen 1980/1998, 356–58.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 247

anti-realists are offering a proposal for how the sentences should be interpreted,
and do not fit neatly into the Byrne-Hilbert package.
To specify an attitude or position as anti-realist (or realist) requires specifying
what it is one is anti-realist about. Usually, one is anti-realist with respect to
certain sentences, terms or theories, e.g., sentences of theories about astronomy,
atoms, electromagnetic fields or societies. In the case of colour, I take it that one
central target is the set of colour-terms, red, blue, mauve, etc., that comprise the
colour vocabulary of ordinary competent colour perceivers, who use colour lan-
guage. It is not arbitrary that this colour vocabulary should be our target, for such
terms are at the heart of a wide set of colour-principles and conceptual practices
pertaining to colour.5
There is, however, more than one way of being an anti-realist with respect to
a set of existing terms. One way is to argue that, given the way these terms operate,
the way they are understood fits an anti-realist description. With respect to colour
terms, this seems to be the position of those philosophers such as M. Dummett,
Gareth Evans and John McDowell, who argue, as an analytic thesis, that colour
terms apply to dispositions of objects to appear in a certain way.6
There is, however, quite a different way of being an anti-realist, e.g., in the
way discussed by B. Van Fraassen. Following this approach, one says, in effect,
to the people using a theory and its related terms, something like the following:
‘Look, I know that you believe that the sentences in the theory, literally under-
stood, to be true, but there is no need to hold those beliefs. There is a way of
understanding the statements that you want to make so that it is not necessary to
believe that they are true, and yet you can achieve all that you could (should)
possibly want to achieve, with your theory’. This being said, there are, as Van
Fraassen also points out, two ways for the anti-realist to go:
(1) The terms and sentences in the theory should not be understood literally,
but their meaning replaced by a new meaning: statements ostensibly
about electromagnetic fields should be interpreted as statements about
dispositions to affect measuring instruments, and other macroscopic
devices; statements about the Resurrection of Christ’s body should be
interpreted as statements about the profound changes made in the lives
of people closely related to Christ, and so on.
(2) The sentences in the theory should be understood literally, but should be
treated not as true, but rather ‘as if they are true’.7
Or possibly, s/he might combine both approaches.

5
See Broackes 1992; Johnston 1992; Maund 1995.
6
Dummett 1979, 28; Evans 1980, 272; McDowell 1985, 111–12.
7
Van Fraassen 1980/1998, 357–58.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

248 Barry Maund

Let me apply some of the points raised here to the specific case of colour,
illustrating some of the problems. I have defended a certain theory of colour.
The theory is that our visual perception has a certain character. The world we
visually perceive is a world of coloured things, of skies, oceans, ground, objects,
faces, clothing, fruit and so on. We perceive these things as having a variety of
colours. Our visual experiences represent things as being yellow, purple, brown,
golden, and so on. There is a common character to these colours. The properties
purport to be properties of a certain type: to be objective properties which are
sensuous, qualitative, manifest properties possessed by skies, lemons, oceans
etc.
The thesis is that objects do not have the colour properties that our experi-
ences represent them as having. In an important sense, our colour experiences
are illusory. There are no colours in the world, so conceived. This theory is
sometimes referred to as an ‘error theory’, and sometimes described as commit-
ted to ‘colour eliminativism’. Both of these labels are, however, misleading.
They are appropriate, if at all, only if qualified. The label ‘error theory’ is mis-
leading because the implication of calling something an error, at least in an
unqualified sense, is that it should be eradicated or eliminated. It is true that, on
my theory there are errors in perception, but the theory would be better thought
of as an ‘illusion theory’. The point about illusions is that many of them are
harmless, and others are actually beneficial: they serve a valuable function, e.g.
the illusions of the theatre or cinema. Colours are a case in point, or so I argue.
They are illusions to be celebrated, to be enjoyed, fostered, shared, the subject
of poetry, and so on. They are no more to be eliminated than are the illusions of
the theatre and cinema, and of the novel. (Here I go against Plato in his attack
on the poets.)
A similar point can be made about the label ‘colour eliminativism’. I would
be unhappy with the suggestion that, according to my theory, there are no colours,
i.e. that there are no colours tout court, or no colours period. I think that a
qualification needs to be inserted. What I claim is that the world contains no
colours, as traditionally conceived, i.e., as traditionally understood and as repre-
sented in perception. That leaves open the possibility that there might be a non-
traditional conception of colour, say, a technical one. Nor does it imply that the
traditional conception should be eliminated – because, for one thing, given the
point about illusions above, that concept may have a central role to play in how
the illusion works, and for another it might otherwise be valuable. Indeed, one of
the possibilities is the one outlined by Van Fraassen: that one should treat the
sentences about colour, firstly as literally understood, and secondly, ‘as if they are
true’. There is a vast difference between saying of a sentence ‘it is false’ and
‘it is as if it is true’, even if with both statements, the sentence is false. This

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 249

discussion, it seems to me, throws light on some of the discussions of colour and
its problems by well-known figures in the scientific tradition.
The thesis that I defend with respect to colour is best thought of as an Illusory
theory rather than to describe it as an ‘error theory’ or as ‘colour eliminativism’.
It is important, that is, to stress the illusory character of colour experience rather
than its making errors. One point about illusions is that they can be beneficial and
have wonderful, or at least pleasing side effects. A second point is that the illusions
can have important functions to play, either biologically or psychologically. A
third point, which may be involved with the second, is that even illusions can be
involved in the acquisition of the truth. That is to say, truth and error can be
entwined to the advantage of the former. A number of perceptual examples
illustrate this last point: the bent stick illusion; the phenomena of perceptual
constancies; (iii) the experience of pain; (iv) the images in mirrors.

(1) The Bent Stick: The way in which truth can be cloaked in error is
illustrated by the hoary example of the bent stick in water. The straight
oar admittedly does not look straight, but the way it looks still carries
information that one can extract. For although the oar looks bent, there
is a specific degree of bentness that it looks: it does not look any old
way. The point is neatly captured by Edmond Wright in a discussion of
how illusions are often misinterpreted in debates between direct and
indirect realists:
An expert on refractive indices walks by a row of beakers each illustrating the so
called ‘illusion’ of the Bent Stick and confidently reports as he taps them one by
one ‘Water, petrol, benzene, acetone, alcohol’ (Wright 1996, 34).

The stick does not merely appear bent, it appears bent to a certain degree. The
fact that it is bent to a certain degree means that we can draw an inference about
the density of the liquid in which the stick is placed.

(2) Pain: The experiences of pain are both illusory and truthful. That we
know the pain is not in the tooth may lead us to take a tablet, but our
belief that there is something wrong in our tooth leads us to the dentist.
But unless the pain is felt in the tooth, I won’t be led to the dentist. In
other words, my true belief that something is wrong in the tooth is
brought out by the illusion – that the pain is in the tooth, i.e. the pain is
represented as being in the tooth. The point of pain experiences is that
the illusion leads the sufferer naturally and immediately to do something
about the place at which the pain is felt: where the shoe pinches, the bee
stings, etc. The experiences of pain are illusory in that pains are repre-
sented as being in places in which they are not, and thereby provide us

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

250 Barry Maund

with the means for forming the true belief that there is damage in those
parts of the body.
(3) Mirrors: In the case of mirrors, I have the illusion of there being a
tortured soul remarkably like myself the other side of the painted glass,
a certain distance away. That illusion can tell me something truthful: that
my tie is straight, my grin is crooked and my eyes are watery.
The point about colour illusions is that even if objects do not have colours,
they appear to have them, and for many purposes, that is good enough. For those
purposes, it is sufficient that objects appear to be coloured: they do not have to
actually have them. The point about colours is that for many purposes, the colours
serve the purposes, by being a sign – for the perceiver – and of course, the illusion
can function as well, as a sign, as the real thing.

2. Problems with colour: the scientific tradition

There is a famous remark in the writings of David Hume:
Sounds, colours, heat and cold, according to modern philosophy are not qualities in
objects, but perceptions in the mind (D. Hume 1738/1911, Bk III, part I, Section I,
177)8
What Hume says is literally false – at least it is tempting to think so. There are
passages in the texts of Descartes and Locke, to take two of the philosophers that
he had in mind, in which they seem to speak of colours as secondary qualities, as
powers to cause sensations or sensory ideas of colours, and the secondary qualities
are qualities of physical bodies. Their official position would seem to be that there
are two kinds of colour: colour as a sensory quality, inherent in sensations or in
sensory experiences, ‘colour-as-it-is-in-experience’ (or ‘colour-as-it-is-in-sensa-
tion’), and colour as a quality of physical bodies, ‘colour as-it-inheres-in-physical-
bodies’.9 Colours, as they inhere in physical bodies, are secondary qualities,
powers to induce experiences that have colours of the other, subjective kind. Some
physical bodies are yellow, in the secondary quality sense, since they have the
power to cause, in the right circumstances, experiences that have yellow-as-it-is-
in-experience.10
I do not think that this view fully captures the views of Descartes and Locke.
Even if it did, however, we might well feel that Hume still has a point. For at the
basis of the distinction Descartes and Locke make between the two kinds of
colour, is a more fundamental thesis about colour terms and colour concepts. For
8
See also Bk I, IV, IV, 216.
9
Descartes 1954, A & G, LXVIII–LXX; Locke 1961, BK II, Ch.VIII.
10
This view has been challenged by Alexander 1985, esp. Ch. 8, who claims that, for
Locke and for Boyle, colours are not secondary qualities, which are physical textures, but Ideas
in the mind.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 251

them, there are also two kinds of ways in which colour terms and concepts are
used. The primary application is to the sensory quality, to colour-as-it-is-in-
experience, but there is a secondary application as well. Colour terms are also
applied to physical bodies, but in a way that is derivative on the first. Descartes,
for example, says ‘when we say we perceive colours in objects, it is really just
the same as though we said (my emphasis], that we perceived in objects something
as to whose nature we are ignorant but which produces in us a very manifest and
obvious sensation, called the sensation of colour’ (Descartes 1954, LXX). The
proper use of colour terms when applied to physical bodies should be understood,
following Robert Boyle, as following a principle of metonymy, e.g, as in the use
by waiters in a cafe: ‘the egg-burger did not leave a tip’; ‘the ham sandwich spilt
the tomato sauce’, ‘the two fried eggs made a pass at me’. (I borrow the example
from George Lakoff.) According to the principle of metonymy, as applied to
colours, physical bodies have colours, but only in a derivative sense. The statement
that a ball is yellow is elliptical for the statement that the ball has a power, to
induce experiences that are experiences of yellow. The primary sense of colour
terms is to apply to the colour in the experience. Hence the physical objects no
more have colours than do ham sandwiches have desires or make passes, and we
can understand the point of the Humean view.
Therefore, though Hume is not sensitive to the complexities of the Descartes-
Locke position, we might feel, nevertheless, that he is making an important point.
The undeniable implication of their position is that, with respect to the primary
sense of colour terms, physical objects do not have colours, a view that is surely
counter-intuitive. What brings the point out more clearly is the often-quoted
passage from Galileo:
I think that tastes, odours, colours and so on, are no more than mere names so far
as the object in which we place them is concerned, and that they reside only in the
consciousness. Hence if the living things were removed, all these qualities would
be wiped away and annihilated (Galileo 1623/1957, 274).
We might note also that many modern scientists express themselves in the Humean
way by saying that colours are not in the world, e.g. E. H. Land, S. Zeki, Rolf
Kuehni, Alain Chrisment.
Zeki:
The nervous system, rather than analyze colours, takes what information there is in
the external environment … and transforms that information to construct colours,
using its own algorithms to do so. In other words it constructs something which is
a property of the brain, not the world outside (Zeki 1983, 764).
Kuehni:
The present outline implies that color experience is entirely subjective. The world
out there is not colored. Colors are symbolic representations of localized spectral

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

252 Barry Maund

power structures in comparison to neighbouring different spectral power structures

(Kuehni 2001, 24).

Alain Chrisment:

Color is all around us and it conditions us. So what is color?

There is no simple answer to this, since color is not a physical reality … the color
is in the brain rather than in the material. Color is an interpretation by the cortex,
of sensations detected by the eye (Chrisment 1998, 5).

The Humean interpretation is supported by the fact that the use of colour terms
in the secondary quality sense is a sophisticated use. Although Descartes gives
the secondary quality account when, as quoted above, he says ‘it is just as though
we said, . . . ’, he concedes, as does Locke, that the naive perceiver, or the
philosophically-innocent, is unlikely to think this way. Instead, he or she thinks
of physical bodies as having the sort of quality that is the same as the colour-as-
it-is-in-experience. Descartes and Locke say that we are subject to a tendency
to make a hasty judgement, without reflection, or from habit that we fell into in
our youth:

So we easily fall into the mistake of judging that the feature of objects that we call
colour is something just like the colour in our sensation; i.e. of thinking that we
clearly perceive something which in fact we do not perceive at all (Descartes 1954,
195–96).

It is because we are prone to make such mistaken judgements, Descartes and

Locke argue, that many of us feel that the philosophical/scientific account of
colour is so implausible. Once we acquire the philosophical understanding to
correct that mistake, the account should no longer be seen as counter-intuitive.
(Old habits, however, die hard.)
It is important to note that on this account, Descartes and Locke are not
rejecting the ordinary concept, and replacing it by a new. Their account of what
is required to truly ascribe colour terms to physical bodies presupposes that the
old concept is retained. It is crucial both in the primary application of colour
terms, to subjective qualities, and in the secondary application, to physical
bodies.
Thus, on my reading of the Cartesian-Lockean position, it is quite complex.
There is a primary application of colour terms such that they apply to experiential
qualities, qualities inherent in sensations or experiences, and a secondary, deriv-
ative, metonymic sense, in which colour terms are legitimately applied to physical
bodies. It is tempting but mistaken to say that the first use is a naive, unreflective,
‘vulgar’ use, and the second a sophisticated, reflective, philosophical use. (We
need another term besides ‘naive’ and ‘sophisticated’, to capture the attitude one

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 253

can adopt when one becomes sophisticated, of acknowledging the legitimate, if

restricted, role of the naive attitude.)
The view, rather, is that there is a primary fundamental use of colour terms,
which is legitimate and proper, when restricted to experiential qualities. There is,
however, a mistaken and illegitimate use: when the term is applied, naively and
unreflectingly, to physical bodies. What is mistaken is the way the term is under-
stood, when so used. There is, however, a secondary, derivative application of
colour terms, to physical bodies, which requires a more sophisticated understand-
ing. Both the primary and the secondary applications, properly understood, are
‘philosophical’.
On the Descartes-Locke position, sentences such as ‘bananas are yellow’, and
‘that dress is claret-red’ can be understood in different ways. In the ordinary
understanding of ‘yellow’, such sentences are false. However, there is another way
of understanding the sentences – as elliptical for ‘bananas have the power to
appear yellow, in appropriate circumstances’, ‘that dress has the power to appear
claret-red, in normal circumstances’. These sentences should be understood as
operating according to a principle of metonymy, as presupposing the ordinary
concept of colour, of yellow, or of claret-red. Accordingly, the sentence ‘bananas
are yellow’, understood in the literal way, is false, while understood in the met-
onymic way, is true. This account provides an up-to-date expression of the 18th
century slogan that one should aim to ‘speak with the vulgar and think with the
learned’. My comment on that slogan is that it should be modified so as to take
account of the fact that the learned should combine being learned with being
‘vulgar’.
We might note that, on this account, there is no need to revise the practice of
uttering sentences such as ‘that dress is claret-red’ or ‘ripe bananas are yellow’.
We simply have to take them metonymically, not literally. So understood, the
sentences are true. And we can leave the ‘vulgar’ to think as they do: we should
learn ‘to speak with the vulgar and think with the learned’.
We should note that this form of dispositionalism allows for two versions.
According to one, objects may be said to have a ‘true’ colour – it has the
disposition to appear in a certain way, under normal conditions and to normal
observers. According to the second version, colours are treated as relativised, to
circumstances and to observers, and no disposition is picked out as privileged.
Given the complexities of this position, the original framework set out by
Byrne and Hilbert needs modification. The Descartes-Locke position does not fit
neatly into that framework, being partly realist and partly eliminativist. I think,
however, that it is better described as a version of anti-realism, with respect to the
colours of physical bodies. I think, moreover, that it points the way to a different
version of anti-realism.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

254 Barry Maund

3. A reconstruction of the traditional scientific view

The Cartesian-Lockean position, as I have presented it, is quite complex. The

position has not been popular in philosophical circles. Given its complexities,
however, there are different reasons why we might feel dissatisfied. Many of us,
for example, will disagree with its account of the primary application of colour
terms, but we might have different conceptions of what the right account is. And
even if we agree on what the right account should be, we might disagree on
whether physical bodies have colours, in that sense. Or then again we might think
that the position is hopelessly confused.
I think that Descartes and Locke are right in their instincts, but wrong in the
way they formulate their thesis. In particular, I think they go wrong in how they
characterise the primary use of colour terms. I think that Descartes and Locke
mischaracterise the conflict with the scientific/philosophical account of colour and
the common-sense account. The problem, as I see it, is that once we understand
the primary quality/secondary quality distinction central to the scientific tradition,
it seems to us that, while intuitively colour is one of the primary qualities, equally
with shape and number, nevertheless, within the scientific tradition, colour is
excluded from the list of primary qualities. Intuitively, colours seem to be an
objective part of the world, to be intrinsic, qualitative features of the objective world.
I think we should reject the scientists’ account of the natural/folk concept of
colour, construing it instead in such a way that colours are taken to be perceiver-
independent, intrinsic, qualitative features of physical surfaces, volumes and other
physical entities such as skies, rainbows and flames. This is the kind of colour
that our visual experiences represent objects as having. The Descartes-Locke
position therefore should be reframed so as to adopt this formulation of the folk
concept, and to argue that no instances of this concept are physically actualised.
According to this way of thinking, Descartes and Locke were right that, given
the natural (naive, pre-reflective) concept of colour, we can conclude that objects
do not have colours. Their way of characterising the concept, however, is at fault.
It should not be described as ‘colour as it is in experience’. It would be better
termed ‘colour-as-we-see-it’. The ordinary concept is more plausibly construed
as a concept of a certain kind of property: a perceiver-independent, intrinsic,
qualitative feature of physical surfaces (i.e. it is not a dispositional property either
to affect light or to appear to observers).
If we do reconstruct the scientific position in the way I have suggested, that
still leaves us with the traditional problem – of finding a place for colours in the
scientific account of nature. The answer to that, it seems to me, is that Descartes
and Locke are substantially correct.
There are two kinds of ways in which colour terms can be used, or understood.
The primary application of colour terms is to physical bodies, to bodies in

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 255

objective space, where the colour concept is a concept of an intrinsic, sensuous,

qualitative feature of those bodies. Sentences such as ‘bananas are yellow’, and
‘that dress is claret-red’ can be understood in different ways. In the ordinary
understanding of ‘yellow’, such sentences are false (i.e. understood literally, the
sentences are false). However, there is another way of understanding the sentences
– as elliptical for ‘bananas have the power to appear yellow, in appropriate
circumstances’, ‘that dress has the power to appear claret-red, in normal circum-
stances’. These sentences should be understood as operating according to a prin-
ciple of metonymy, as presupposing the ordinary concept of colour, of yellow, or
of claret-red. Accordingly, the sentence ‘bananas are yellow’, understood in the
literal way, is false, while understood in the metonymic way, is true.
We can, if we like, say that there are two senses of colour terms, two colour
concepts, the secondary quality sense/concept, and the primary quality sense/
concept, provided that we acknowledge the priority of the primary quality sense.
Physical bodies do not have colours in the primary sense, but they do have colours
in the secondary quality sense, for there really are important dispositional prop-
erties, for objects really do have powers to appear in the ways distinctive of the
various colours. Having said that, I want to make it clear that the position that I
have just outlined is compatible with either taking the dispositions to appear to
be relative to conditions and perceivers, or to be privileged, i.e. to be those
dispositions that are picked out by specification of normal conditions and perceiv-
ers. (Though I favour the former.)
I claim we can characterise the way yellow things appear, i.e. the way some-
thing is represented as yellow. It turns out, however, that no physical bodies have
such properties, i.e. they do not have the yellowness they are typically presented
as having. The fact that there are no actual instances of such properties, however,
does not mean that that particular concept of yellow should be eradicated or
rejected or modified, though it might be supplemented. It still serves important
purposes for it is still the case that lots of things are represented as having that
property.
The fact that things are represented in this way is significant for a number of
reasons. One is that it is central to the social purposes of colours, the purposes
embedded in the social/conceptual practices pertaining to colours. The second
reason is that if in the way perception works, physical things are represented in
that way, then we had better have an account of such representations, of the way
things appear, if we are to have an adequate account of perception.
There is a third reason. The fact that there is a certain way coloured things are
represented as being allows us to exploit the principle of metonymy that Boyle
explicitly refers to, and Locke and Descartes imply. Physical bodies can be said
to be yellow in the sense that they have the power to cause perceptual states that
represent them as yellow. They are not yellow in the literal sense, but there is a

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

256 Barry Maund

metonymic sense in which they are yellow, and that is important to know about.
There is a distinctive way that bananas and lemons look, and that is different from
that which is characteristic of tomatoes and cherries, and it is important to know
of each.
The principle of metonymy, moreover, can be extended – at least in principle.
If we can identify in different types of objects, say physical surfaces, certain types
of light modifying features, say reflectances, that have a distinctive causal role in
the way coloured things appear, then there is reason for calling them ‘yellow’ in
a metonymic sense. I say that this might be done, because there might be large
practical difficulties that prevent it being done, or which make it not a very useful
thing to do. I take it that, in this way, we can give expression to the ideas separately
stated by Hurvich, McAdam and Nassau (see Introduction), that the term ‘colour’
is used to identify different properties, e.g. pigments, properties of light and
sensations, except that I think that things are more complicated than they suggest,
and I offer a different account of ‘different senses’ of the term ‘colour’.
There is a disanalogy in the application of the principle of metonymy between
the case of colour and the case of the cafe, which makes its application to colours
far more significant, and far more useful. The usefulness of the principle in the
case of colours can be explained by thinking of a hypothetical extension of the
cafe case. It might turn out, for example that the cafe chef was mixing ingredients
into the food in such a way that in ham sandwiches, there was something that
made people lustful, in eggs something that made them clumsy, and in cheese
something that made them irascible. In such a case, the situation would be more
analogous to the colour case. In the case of colours, there is something in the
various objects which causes the different kinds of appearances, and that makes
it important to identify the various objects by reference to the colour terms, even
though the use is metonymic.

4. Phenomenological and phenomenal qualities

The principle of metonymy can be further extended to apply to phenomenal
qualities, qualities intrinsic to our experiences of colour. The colours that our
perceptual experiences represent objects as having, are of a type that I call
‘phenomenological qualities’, using the term in contrast to ‘phenomenal quali-
ties’. Phenomenological qualities are qualities that objects appear to have. If they
are qualities, they are qualities of objects in a three-dimensional objective space,
and in time. They are not presented as qualities of experiences. When I perceive
my lemon tree in the garden, I perceive the tree to contain yellow lemons against
a green background. The experience represents the lemons as being yellow, i.e.
as having a quality, which is a phenomenological quality. Phenomenal qualities
are qualities inherent in, and intrinsic to, experience. This is a matter of stipulation.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 257

They are qualia, in full-bodied, full-blooded, Australian Shiraz, variety. Paradigm

examples of phenomenal qualities are found in feelings of pain and pleasure – the
stabbing painful quality, the dull, throbbing painful quality, etc.
Though a phenomenal quality is a quality inherent in one’s experience, it may
be part of the phenomenological quality. The phenomenal quality is presented in
experience but it is not experienced as a phenomenal quality. It is construed as an
objective quality of a physical body in objective space. It is taken to be instantiated
in a body in that space. (In David Hume’s terms, one conflates the phenomenal
quality with an objective quality.) It is tempting to say that the phenomenal quality
is projected onto the physical object; Tempting but misleading. If the quality is
projected, it is projected into the content of the experience, the representational
content. We have a ready example, of this sort of situation. When we experience
pains, we experience them in our bodies: the dull, throbbing pain behind the eyes,
the sharp pain in the tooth, the ache in the left knee. The head, the tooth, the knee,
are all represented as having a certain quality, i.e. they have a phenomenological
quality. Which quality is it? The head behind the eye is represented as having a
dull throbbing pain in it, the tooth as having the sharp, painful quality, etc.
When I feel pain in my knee, then I experience my knee as having pain in it.
I experience the pain in my knee, just as surely as I experience pressure in my
knee, caused by the table against which the knee rests. If the table has a nail
protruding then I feel the pain in my knee just as I feel the pressure there. My
feeling pain in the knee is an experience with a certain representational content.
The pain is part of the content. The only proper account of this content, it seems
to me, is that there is a certain phenomenal quality, the feeling of pain, which is
part of the representational content.
It is helpful in this context to refer to Nelson Goodman’s example of repre-
sentation by exemplification.11 A painter may represent the red colour of a brick
house, and the yellow colour of the road, by exemplifying the red and the yellow,
respectively, in the picture he is painting. Likewise two actors may represent two
characters in a play as kissing, by actually kissing each other. They could of course
simulate the kiss, but actors usually don’t. The characters are represented as
kissing by exemplification. Likewise a film could portray John Malkovich looking
surprised by John Malkovich’s being surprised.
Experiences of colour, I submit, work in a similar way to experiences of pain.
When I experience a lemon as yellow, I experience it as being the phenomeno-
logical colour. This quality is illusory. The lemon does not have it. However, there
is a phenomenal quality, characteristic to the yellow phenomenological quality:
call it yellow*. Appeal to this phenomenal quality forms part of the explanation
why the phenomenological quality is the way that it is. In these cases there are

11
Goodman 1976, 52–67; Goodman and Elgin 1988, 19–23.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

258 Barry Maund

representations but although we can make a distinction, conceptually, between the

vehicle and the content of representation, it is the case, nevertheless, that the
vehicle has certain qualities that form part of the content. (For further discussion,
see Maund 2003.)12
I agree with Michael Tye and Tim Crane that phenomenological qualities
are qualities objects are represented as having.13 In the case of colour, they are
sensuous, intrinsic, objective, qualitative, manifest features. They are objective
properties, properties of objects in objective space and time. Phenomenal qualities
are qualities intrinsic to the experience – they are full-blooded, full-bodied qualia.
Qualia, in my view, are not ineffable, qualities about which we cannot be mis-
taken. Qualia are not the properties thought to be automatically identified by the
use of the phrase ‘what it is like’, applied to our perceptual experiences. They are
phenomenal items used to explain why our experiences have the phenomenologi-
cal character that they do.
This distinction between the phenomenological and phenomenal qualities
allows us to see a way forward. We can modify the principle of metonymy as
proposed by Boyle, which is implicit in the scientific tradition. Taking the phe-
nomenological concept as the primary one, we can identify both a phenomenal
quality, say yellow* which corresponds to the phenomenological colour yellow,
and a property of physical bodies, a secondary quality, call it ‘yellowsq’. Both
senses of ‘yellow’, yellow* and yellowsq can be understood as following the
principle of metonymy.
I should explain how my account differs from that of Michael Tye, even
though, in important respects they are the same. Tye acknowledges that perceptual
experiences have phenomenal character, but argues that this character is explained
in terms of the representational content of the experience, and not in terms of
intrinsic properties of the experiences. My claim is that he is assuming a false
dichotomy, namely: that features of the experience are either:
• features intrinsic to the experience
• features intrinsic to the content of the experience
but not both. I claim that there are some features that are both.
(Note: in using the expression ‘intrinsic to the experience’ I mean to cover
both aspects of the experiencing, and features of phenomenal items, if any, that
are given in experience. There is a philosophical dispute about whether perceptual
experiences should be given an act-object analysis or an adverbial analysis. I am
deliberately trying to bypass this issue. I should add that in my view, which I have
defended, this dispute is an empty one, with little metaphysical significance.)

12
Maund 2003, Ch. 9.
13
Tye 2000, Ch. 3, esp. 45–48 ; Crane 2001, 137–144.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 259

Part II
1. Introduction
In the first part of my paper, I set out a thesis that I wish to defend The Illusory
Theory of Colours, or The Virtual Properties Theory of Colour. In that part, I was
attempting to spell out what the thesis is. In this part, I shall briefly provide some
arguments for it. The thesis that I wish to defend, it is important to note, is a
version of Anti-Realism. The point of insisting on this is that there are two parts
to the thesis: a negative part, and a positive part.
Negative Part:
(i) Physical bodies do not have colours, colours as ordinarily understood;
(ii) Sentences about colours, literally understood, are false.
Positive Part:
(i) We should retain the ordinary concept of colour
(ii) We should treat the sentences, literally understood, ‘as if they are true’.
It is important to remember that the thesis has both the positive and the negative
parts. This is what makes it a version of anti-realism. It also means that arguing
for it will require a neat balancing act: there is a tension between the two parts.
Nevertheless, the balancing act can be carried off: it is simply an implication of
any anti-realist thesis that there is this sort of difficulty. I am arguing for three things:
(1) that there is what we might call ‘an ordinary understanding of colour’,
an ordinary concept(s) of colour: a ‘folk concept(s) of colour’;
(2) that there are no colours, so understood;
(3) that the ordinary concept of colour has an important role to play.

2. Challenges to the argument

The thesis that I wish to defend, it is important to stress, has two parts: the first
is that the scientific tradition is right in holding that physical bodies do not have
colours, as normally and ordinarily conceived, although generally thinkers in this
tradition have been mistaken, or sometimes confused, in their understanding of
what the normal conception is. The second part to the thesis is that the ordinary
concept should be retained. This second part is an expression of an anti-realist
position, that sentences involving colour terms may be understood literally, but
treated ‘as if they are true’.
The argument that nature does not contain colours, as ordinarily and naturally
conceived, depends on our getting clear about what the ordinary conception is.
Once we are clear about that, I argue, the conclusion follows that there are no

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

260 Barry Maund

instances in the world of such colours – or at least, there is no good reason to

think that there are. One way to challenge this argument is to reject the formulation
of the ordinary conception. A realist or objectivist about colours might hold, for
example, that the way the ordinary concept of colour functions, it specifies colours
as those objective properties in the world that have a certain causal role, e.g. that
appear in characteristic ways. Putting it another way, the way our names for colour
operate is by identifying or picking out certain properties, e.g. those that occupy
a certain causal role in the perceptual identification of colours.
That is one way to challenge the account offered by the scientific tradition.
There is, however, a different sort of challenge. It is to reject the emphasis that I
have placed on establishing what the natural or ordinary concept of colour, some
would call it ‘the folk concept’ of colour, is. There are two very different ways in
which this rejection might go. One way, e.g. as proposed by Jonathan Westphal,
is to say that the task of saying what the nature of colour is, what colours are
essentially, is a very different task from giving an account of what the concept of
colour is. The latter task involves conceptual analysis, but the former requires
empirical investigation. I shall argue that this sort of response does not work. First,
it involves a false dichotomy. Second, it fails to acknowledge the central impor-
tance of the ordinary conception of colour.
The second way of rejecting the argument for the ‘virtual colour’ thesis is to
admit that the ordinary conception of colour is such that it has no actual instan-
tiations, but all the same, it does not follow that there are no colours. Mark
Johnston, for example, claims that the traditional concept of colour is a ‘cluster
concept’, incorporating a wide range of different components.14 If it should turn
out that not all of these components are instantiated together, it need not matter.
Providing enough of the components are found, or enough of the right ones, we
can and should maintain that there really are colours.
For the most part, I am in agreement with Johnston. It seems to me that he is
agreeing with my position, which is: there are no colours as traditionally con-
ceived, but there is reason to say that there are colours if we revise the concepts.
And I agree that this makes sense roughly because first there is a set of purposes
for having the traditional concept, and which is such that there are no properties
which ground these purposes, and second, some of the purposes can be served if
we revise the concept by construing colours as dispositional properties. However,
I go beyond Johnston in the following crucial respect. He overlooks the possibility
that we should revise the traditional concept by introducing not just one new
concept, but two (at least). My argument is that some of the old purposes can be
served by one concept, and other purposes by the other. (It was this point that
formed the basis of an early paper of mine ‘Colour: a Case for Conceptual

14
Johnston 1992/1997, 137–39.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 261

Fission’.) Indeed some purposes can be served by a concept for which there are
no actualisations. For some purposes, it is enough for it to be true that ‘it is as if
the concept is exemplified’.
We have a ready-made example that illustrates the point. Take the concepts of
mass in Newtonian Mechanics and Relativistic Mechanics. In Newtonian mechan-
ics, we had a concept of mass, whereas in Relativistic Mechanics, we have two
concepts of mass. The right thing to say about this situation is that the world does
not contain mass as the Newtonians conceived it: there are no instances of New-
tonian mass. The world does contain mass however, two kinds of mass, rest mass
and relativistic mass. The situation is similar to that of colour. There are no colours
as traditionally conceived. There may be colours, as differently conceived, but
they are different kinds of colour. Also just as with Newtonian mass, we retain
the concept, even though there are no Newtonian masses, since for many purposes
the world is ‘just as if there are such masses’, so with colours: the world is as if
there are colours, traditionally conceived.
This last step is vital. Although the traditional concept is not actualised, there
is point in retaining it, just as there is point in retaining the concepts satanic,
angelic, even if one does not believe in Satan and angels (and likewise with
concepts of witch and star). The point is that the colour concept, in the traditional
sense, still serves a purpose, even if not actualised. Very roughly, it is because the
concept serves a valuable purpose, in characterising the way things look, that it
is important. It is significant to say that something looks yellow, in the traditional
sense, even if nothing is yellow, in that sense.
A third way of challenging the argument that I am presenting would seem to
be that offered in a recent article in Behavioral and Brain Sciences, by Alex Byrne
and David Hilbert.15 These authors in defending a realist view of colour according
to which colours are reflectances of a certain type, state that while the problem
of colour realism concerns various salient properties that objects visually appear
to have, ‘it does not concern, at least in the first instance, colour language or colour
concepts. The issue is not how to define the words “red”, “yellow”, and so on.
Neither is it about the nature of the concept “RED” ’. They go on to say that the
problem of colour realism is primarily a problem in ‘the theory of perception, not
a problem in the theory of thought or language’. But here they are operating with
a false dichotomy. What they overlook is that the problem is about both perception
and thought, i.e. about perception, and more specifically that perception which
operates, at least in part, through concepts. This point is valid even for animals,
which do not have linguistically expressed concepts. Such animals can be thought
of as having ‘practical concepts’ or ‘working concepts’. (See Campbell 1994;
Maund 2003, ch. 2) There is a more specific point to make about the special case

15
Byrne and Hilbert 2003.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

262 Barry Maund

of colour. Byrne and Hilbert say that the problem of colour realism concerns
various salient properties that objects visually appear to have’. But those proper-
ties are the ones that our ordinary understanding of colour, our folk concept,
addresses. Our ordinary understanding of colour targets those properties: (see the
next section.)

3. Perceptual saliency and the folk concept of colour

I have stressed the need to clarify the ordinary, pre-reflective understanding of
colour – to specify what has been referred to as the ‘folk concept’ of colour, I
think with affection, by Frank Jackson and David Lewis, for example. This
conflicts with the view of Byrne and Hilbert who claim that colour realism is an
issue related to colour perception, and that is quite separate from the issue of
how colour terms operate, and of how colour concepts are to be specified. I
disagree, but even if Byrne and Hilbert were right in their positive claim, this
would mean that there were two problems of colour realism. For I cannot see
how we can have an adequate theory of colour that did not address the question
of what the ordinary conception of colour is, and whether there are instantiations
of the concept. (There may of course be other interesting issues.) But in any
case, it seems to me that Byrne and Hilbert are not right, for the two issues are
intimately related.
This point can be illustrated by reference to the point introduced at the begin-
ning of this paper. I raised the question then of whether the wide range of theories
about colour were competing, or whether they were in some sense, complemen-
tary. For the different positions to be rival theories of colour they need to be talking
about a common subject matter. We need, that is, to identify the colours, the things
we are talking about, prior to employing the various philosophical theories on
offer, and prior to assessing them.
There is no doubt that colours are properties that have a conspicuous percep-
tual saliency; that they are, as Byrne and Hilbert state, ‘salient properties that
objects visually appear to have’, but to say this is not enough to specify what
colours are. There is a simple and obvious answer to the question of identifying
the properties of colour. It is an answer that acknowledges that, for a start, colours
are properties for which (a) there are competent colour-perceivers who perceptu-
ally identify and recognize, objects as having colours; (b) there is a flourishing
colour vocabulary, e.g. colour names; (c) there are properties for which there are
obvious paradigms: tomatoes are red, skies (in Perth) are blue, lemons (ripe) are
yellow, grass is green, and so on.
This answer recognises that there is a vigorous set of conceptual practices
concerning colour concepts. These practices include the use of colour names but
they involve far more. (That is to say, the term ‘conceptual practices’, embraces

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 263

a wide range of practices, linguistic and non-linguistic, that cover a wider range
than the having of thoughts and the talking about colour.) For one thing, they
involve the exercise of a range of capacities, e.g. to make certain colour matches:
to identify and recognise, by looking, the colour of objects. For another thing,
they include the capacities to use colours for a variety of purposes, social and
personal. One important role colours play is in ‘practical epistemology’, i.e. as
signs for the identification and re-identification of physical objects. In this
capacity, they serve not only as natural signs, but as social, conventional signs,
e.g. as badges, uniforms, for ceremony etc. But there is an even more significant
role for colours. In addition, they may also be said to have a ‘life of their own’.
That is, they are used in social life to amuse, to entertain, to delight, to shock, to
impress, to astound, to warn, to attract, to be enjoyed, and so on, in contexts
having to do with pageantry, ceremonial, courtship, painting, lighting, plays,
clothing, dining, drinking, and so on. In the visual arts, in paints, in design, in
lighting, in fashion, in industry, as well as a whole range of other practices,
colour is important.
The existence of these colour conceptual practices is significant, for a number
of reasons. One is that they allow us to identify a subject matter for theories of
colour (to be theories about). Secondly, close analysis of these practices allows
us to identify a range of colour-principles that colours, at least on the face of it,
obey. These principles are embedded either explicitly or implicitly, in the prac-
tices. Their significance is that they provide certain constraints upon any adequate
theory of colour. To say this is not to say that the principles are unassailable or
irrevocable. It is to say, rather, that the principles need to be addressed by any
adequate theory of colour. One way they might be addressed is to argue that they
merely appear to be true, and to explain why they appear so. Another way is to
show that the principles lend themselves to an interpretation that differs from the
orthodox one, and can be defended.
One highly significant fact that is revealed by the study of the conceptual
practices governing colour, is that the fact that colours are perceptually salient,
visually conspicuous, in the way that they are, lies at the heart of these conceptual
practices. It is because colours are perceptually salient in the way that they are
that they serve the purposes that they do, and the conceptual practices take the
form that they do. Accordingly, we do not have to accept the contrast Byrne and
Hilbert make between giving an account of the colour perception and given an
account of concepts of colour. Furthermore, having the wealth of relevant con-
ceptual practices in mind, it is possible to recognise that, in talking of the percep-
tual saliency of colours, we are recognising a set of perceptual facts about colours:
that we can identify a range of paradigms; that there are competent colour-
perceivers who perceptually identify and recognise objects as having colours, and
who can make colour-matches; that colours are properties that play a causal role

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

264 Barry Maund

in the perceptual identification and recognition of colours, and in the identification

of paradigms of colour; that there is a distinctive and characteristic way that, for
each colour, things that have the colour, look. This last point is particularly
important: there is an important property that many objects share: of looking
yellow, of looking pink, of looking blue, and so on.
The perceptual saliency of colours allows us to identify further general prin-
ciples concerning colour. One of the most striking things about colours is that
they can be taken collectively so as to form a structured array of properties, with
a characteristic structure. Colours as a group can be placed in systematically
ordered arrays, which can be captured in psychological colour spaces, constructed
according to one or other of the various colour systems, as in the Swedish Natural
Colour System, the Munsell Colour System. These spaces are based upon match-
ing (and other perceptual) judgements of colour perceivers: X matches Y; X is
closer to Z than to R; X contrasts more with A than with Z; In contrast to X, R
is between A and Z. On the basis of such judgements colours can be placed in a
colour space. The dimensions of the colour spaces will vary not only according
to the system, but according to whether we are considering surface colours,
volume colours, aperture colours, etc.; but in each case, colours collectively obey
certain structural principles.
There is an additional aspect to colours, i.e. to the sorts of properties identified
in the ways outlined above, which is grounded in their perceptual saliency.
Colours are properties that we, as competent colour perceivers, are trained to
identify, recognise and teach others. Having such capacities, it is possible to
describe the colours that we are competent in identifying, as properties which
have a certain general character: they are properties presented in perception as
objective, intrinsic, qualitative, sensuous features of bodies in physical space. The
features are intrinsic, not relational and in particular, not perceiver-dependent.
They are objective, in the sense of being intrinsic features of physical objects.
They are sensuous, qualitative features, whose nature is manifest.
The properties so far identified, are general ones, which hold either of colours
generally, or of colours taken as group. It is possible, however, to identify more
specific principles. First, there is a range of principles concerning the way
coloured objects appear and the circumstances under which the way they appear
change, e.g. related to the effect of the surround of the object seen, changes in
illumination, and the (limited) phenomena of colour-constancy. Second, there is
a body of putative causal colour truths:

• lemons (bananas, wheat, . . . ) turn from green to yellow as they ripen,

• insects with red stripes tend to be venomous,
• red sky at night, shepherds’ delight; red sky at morning, sailors’ warning;
• acids turn blue litmus paper red.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 265

Summarising, we can classify the various principles pertaining to colour as

follows:
1. Colours are Perceptually Salient:
(a) there are distinctive sets of paradigms for colours;
(b) colours are perceptually identified and recognised by competent
colour perceivers;
(c) colours have a distinctive characteristic way they appear;
(d) colours have a causal role, vis-à-vis the perception and identification
of specific objects as red, yellow, blue, and so on; and with their use
as paradigm examples in teaching colour language;
2. Structural Principles: colours can be arranged systematically in ordered
arrays. That is, a set of internal relationships hold between the range of
colours.
3. Phenomenological Principles: Colours, as presented in perception, are
intrinsic (i.e. non-relational), features of objects, feature which have a
certain character. They are (a) intrinsic, i.e., non-relational; (b) objective
perceiver-independent; (c) sensuous, (d) qualitative and; (e) their nature
is manifest.
4. Social/Epistemological Role: One of the central roles colours play is that
of serving as a sign, either as a natural sign, for the presence of objects
or for properties of objects, such as spatio-location, danger, invitation,
and so on; or as a conventional sign, in a social context.
5. Specific Causal Principles:
(a) there is a range of principles concerning the way coloured objects
appear and the circumstances under which the way they appear
change, e.g. related to the effect of the surround of the object seen,
changes in illumination, and the (limited) phenomena of colour-
constancy.
(b) there is, however, a body of putative causal colour truths covering
colours of objects, e.g. ripening lemons turn from green to yellow;
white houses are cooler than black ones.

4. The argument that there are no colours traditionally conceived

Colours, on the face of it, are properties that satisfy this range of principles.
Colours, as they are ordinarily conceptualised (colours as specified by the ‘folk
concept’ of colour), are properties of this kind. I say, ‘on the face of it’, that
colours satisfy this range of principles. The problem is, however, that once we try
to take into account the results of the scientific analysis of colour, it seems that
there are no properties that fulfil the requirements. In particular, there are no

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

266 Barry Maund

properties that both play the right causal role, in the perception of colour, and
which satisfy the structural principles (nor that have the right sensuous, qualitative
character).
There are, that is to say, no colours that are intrinsic, non-relational, perceiver-
independent properties and which satisfy the requirements of any three-
dimensional colour solid. None that is, that allow us to make sense of the way in
which we perceive and identify and recognise colours. In particular, the colours,
which the surfaces of physical objects look to have, collectively form a group with
the 3 + 1 structure, based on the 4 unique colours and the black/white pair. There
are objective, intrinsic properties, which play a causal role in the experience of
such colours, but they do not satisfy the structural relationships. We may be able
to hold on to the objective characterisation of colour if we give up this last
requirement, e.g. if we say that it is one thing for objects to be red, and another
for them to look red, and that it is the way they appear that satisfy the requirement,
and not the colours themselves, but that is to give up the ordinary characterisation
of colour. It is to say that there are colours but not as we had originally conceived
them. That can be done, but we must recognise the price. We have agreed that
there are no colours as ordinarily conceived. In the second place, even if we accept
that, there will still be a need for the original concept of colour, or something like
it, for we shall still need to characterise the way coloured things look, i.e. in terms
of something’s looking red, looking yellow, and so on. Not only that, but there is
a crucial need for the folk concept of colour, for the way things appear is crucial
to the role that colours play in our social lives, i.e., to the role that makes colours
so important to us.
Alternatively, we might weaken the notion of ‘causal role’ so that, for example,
colours can play a causal role without being the productive cases of perception.
An example borrowed from Fodor, in another context, is this: tall parents cause,
by and large, the birth of tall offspring, even though it is not the tallness per se
that causes the offspring to be tall, but some genetic feature that is the basis for
the tallness. Likewise, it might be true that red things cause the perception of them
as red, even if it some feature of each such thing, which is the causal basis for
the redness, that does the causing.
A plausible way for how colours might serve such a causal role can be found
if colours were construed as dispositional properties, pure or mixed. Mixed dis-
position: for X to be yellow, is to have some feature by virtue of having which,
X causes the perceiver to see it as yellow (under the right sort of conditions). But
to accept that account of colours would be to reject a crucial part of the Phenom-
enological aspect to colours, according to which colours are intrinsic, objective
properties of objects.
On the face of it, it seems hard to see how colours, construed as intrinsic,
objective, qualitative, manifest properties can play the right causal role. That they

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

 The Illusory Theory of Colours 267

can play such a role seems to be the view of John Campbell, who defends what
he calls the Simple View of Colours.16 But I find it hard to see how this can be.
There is an important difference between the tall parent case and the colours case.
Tallness is a property that, let us say, has in people a specific causal basis. It is
however, not only a property distinct from the causal basis, but one whose
instances can be known to obtain quite independently from being the effect of the
causal basis. It can be perceived, for example, and it can be measured. In the case
of colour, on the other hand, things are different. The only way we know about
the colour is by it being perceived. This raises the question of how we know that
the object, in addition to having the causal basis, has the colour as well. We might
put the problem this way. Suppose that the simple, qualitative, manifest property
is supervenient on say spectral reflectance. Suppose that overnight things changed:
the simple feature that was previously there, slipped away. How would we know?

Conclusion
Given that with our ordinary understanding of colour, the folk concept of colours,
we seem to be committed to an inconsistent set of principles, how should we
proceed? My proposal is that we should adopt an anti-realist practice. We should
treat sentences ostensibly about coloured physical bodies, ‘as if they are true’, or
at least, certain central statements. For many purposes that is appropriate. The
point is that the colour concept, in the traditional sense, still serves a purpose,
even if not actualised. Very roughly, it is because the concept serves a valuable
purpose, in characterising the way things look, that it is important. It is significant
to say that something looks yellow, in the traditional sense, even if nothing is ever
yellow, in that sense.
This proposal does not mean that we should stop inquiring into the causes of
why we perceive colours the way that we do. Nor does it stop us from introducing
new concepts pertaining to colour, e.g. dispositional properties, and possibly
‘objectivist’ concepts, that function in the metonymic way outlined in the first part
of this paper. What makes the anti-realist position attractive is that it combines
the point that there are no colours, traditionally conceived, while acknowledging
that there is reason to retain the traditional concept. The reason is that the standard
conceptual practices operate because central to them is the way that colours
appear. For all practical purposes, it is not necessary that objects have colours: it
is sufficient if ‘it is as if they have them’.

References
Alexander, P. 1985, Ideas, Qualities and Corpuscles, Cambridge: Cambridge University Press.

16
Campbell 1993/1997, 177–78.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica

268 Barry Maund

Byrne, A. and Hilbert, D. R. (eds.) 1997a, Readings on Color, Vol. I: The Philosophy of Color,
Cambridge, MA: MIT Press.
Byrne, A. and Hilbert, D. R. (eds.) 1997b, Readings on Color, Vol. II: The Science of Color,
Cambridge, MA: MIT Press.
Byrne, A. and Hilbert, D. R. 2003, ‘Color Realism and Color Science’, Behavioral and Brain
Sciences, 26, 1, pp. 3–21.
Broackes, J 1992/1997, ‘The Autonomy of Colour’, in: A. Byrne and D. R. Hilbert, eds., 1997a, pp.
191–226; originally in: D. Charles and K. Lennon, eds. 1992, Reduction, Explanation, Realism.
Campbell, J. 1993/1997, ‘A Simple View of Colour’, in: A. Byrne and D. R. Hilbert, eds., 1997a,
pp. 170–190; originally in: J. Haldane and C. Wright, eds., 1993, Reality, Representation and
Projection.
Campbell, J. 1994, Past, Space and Self, Cambridge, MA: MIT Press.
Chrisment, A. 1998, Color and Colorimetry, Paris: Editions 3C Conseil.
Crane, T. 2001, Elements of Mind, Oxford: Oxford University Press.
Descartes R. 1954, Philosophical Writings, trans. G. E. M. Anscombe, and P. Geach, London:
Nelson.
Dummett, M. 1979, ‘Common Sense and Physics’, in: G. Macdonald, ed., Perception and Identity,
London: Macmillan, pp. 1–40.
Evans, G. 1980, ‘Things Without the Mind’, in: Z. van Straaten, ed., Philosophical Subjects, Oxford:
Oxford University Press, pp. 76–116.
Galileo, G. The Assayer, in: Stillman Drake ed. and trans. 1623/1957, Discoveries and Opinions of
Galileo, Garden City, N.Y: Doubleday, pp. 229–280.
Goodman, N. 1976, Languages of Art, 2nd edn., Indianapolis, IN: Hackett.
Goodman, N. and Elgin, C. Z. 1988, Reconceptions in Philosophy and Other Arts and Sciences,
London: Routledge.
Hume, D. 1738/1911, Treatise of Human Nature, London: Dent.
Hurvich, L. M 1981, Color Vision, Sunderland: Sinauer.
Johnston, M. 1992/1997, ‘How to Speak of the Colors’, in: A. Byrne and D. R. Hilbert, eds., 1997a,
pp.137–76.
Kuehni, R. G. 2001, ‘Color: what could it be?’, Symposium: What is Color?, AIC Congress
Rochester.
Land, E. H. 1983, ‘Recent Advances in Retinex Theory and Some Implications for Cortical
Computations: Color Vision and the Natural Image’, Proceedings of the National Academy of
Sciences, 80, pp. 5163–9.
Locke, J. 1961, Essay Concerning Human Understanding, ed. J. Yolton, London: Dent.
McAdam, D. L. 1997, ‘The Physical Basis of Color Specification’, in: A. Byrne and D. R. Hilbert,
eds., 1997b, pp. 33–63.
Maund, B. 1981, ‘Colour: A Case for Conceptual Fission’, Australasian Journal of Philosophy, 59,
pp. 308–22.
Maund, B. 1995, Colours: Their Nature and Representation, Cambridge: Cambridge University
Press.
Maund, B. 2003, Perception, Chesham: Acumen.
McDowell, J. 1985, ‘Values and Secondary Qualities’, in: T. Honderich, ed., Morality and Objec-
tivity, London: Routledge and Kegan Paul, pp. 110–27.
Nassau, K. 1998, ‘Fundamental of Color Science’ in: K. Nassau, ed., Color for Science, Art and
Technology, Amsterdam: Elsevier, p.3.
Tye, M. 2000, Consciousness, Color and Content, Cambridge, MA: MIT Press.
Van Fraassen, B. 1980/1998, ‘Arguments Concerning Scientific Realism’ in: J. Kourany, Scientific
Knowledge, 2nd edn., Belmont:Wadsworth, pp. 355–368, originally in: Van Fraassen 1982, Sci-
entific Image, Oxford: Clarendon, ch.2.
Wright, E. 1996, ‘What It Isn’t Like’, American Philosophical Quarterly, 33, pp. 23–42.
Zeki, S. 1983, ‘Colour Coding in the Cerebral Cortex’, Neuroscience, 9, 4, pp. 741–81.

© 2006 The Author. Journal compilation © 2006 Editorial Board of dialectica