Grazer Philosophische Studien

63 (2002) 123-140

SOMETHING ABOUT MARY*

Alex BYRNE
Massachusetts Institute of Technology, USA

Summary
Jackson’s black-and-white Mary teaches us that the propositional content
of perception cannot be fully expressed in language.

In one of the earliest and best replies to Frank Jackson’s knowledge

argument, Horgan claimed, in effect, that the argument illegiti-
mately draws a metaphysical conclusion – that physicalism is false –
from an epistemic premise – that physically omniscient Mary would
not know everything. Horgan’s response has become standard. And,
as it happens, I think it is correct.1 But although physicalism survives
unscathed, Jackson’s thought experiment does hold an important
lesson.2 It suggests, I shall argue, that the propositional content of
perception is ineffable, in the sense that it cannot be fully expressed
in language.
This conclusion will be reached by considering a puzzle about
perception posed by a slightly modified version of Jackson’s
thought experiment. As will become clear, the puzzle has affinities
* Earlier drafts were improved thanks to Michael Glanzberg, Benj Hellie,
David Hilbert, Jim John, Sarah McGrath, Jim Pryor, Susanna Siegel, Daniel
Stoljar, Mike Thau, Ralph Wedgwood, and Steve Yablo. Ancestors of the paper
were given at ANU (RSSS), and University College London; I am grateful to the
audiences there and at Celje for helpful discussion.
1. Moreover, I think Mary’s physical omniscience and logical acumen would
not enable her to know much of anything expressible in everyday vocabulary
(Byrne 1999).
2. Horgan himself thinks a repaired version of the knowledge argument does
threaten physicalism (Graham and Horgan 2000).
 124

with Kripke’s puzzle about belief and Molyneux’s problem, al-

though these connections will not be explicitly traced.3
Before explaining the puzzle, some preliminary remarks are
needed.

1. Mary’s new belief, and the content of perception

The new belief. When Mary leaves her room and sees a ripe tomato
for the first time, she learns, Jackson tells us, a fact about experi-
ences (1986, 293). And commentators almost invariably follow
Jackson in taking this to be the fundamental intuition at work in the
knowledge argument, either to be defanged, or explained away.
Horgan, for instance, formulates Mary’s new knowledge as follows:

Seeing ripe tomatoes has this property,

where ‘this property’ is used to designate the colour-quale that is
instantiated in her present experience. (1984, 151)

Now I don’t want to deny that Mary learns something about the ex-
perience of seeing ripe tomatoes. But recall G. E. Moore’s much-
cited observation that “[w]hen we try to introspect the sensation of
blue, all we can see is the blue: the other element is as if it were di-
aphanous” (1903, 450).4 Surely Mary also learns – or at least comes
to believe – something about tomatoes.5 “Good grief,” we can imag-
ine her saying, “so ripe tomatoes have this property; I daresay it’s the

3. Thau (1998) proposes a solution to the puzzle, which he integrates with a

lengthy defense of Millianism. As will be apparent in section 3 below, I disagree
with Thau’s solution, but I am greatly indebted to his discussion. I should empha-
size that I only consider an early formulation of Thau’s view that was not in-
tended for publication. A more developed and doubtless somewhat modified ver-
sion will appear in Thau forthcoming.
The puzzle is also discussed in important papers by Nida-Rümelin (1995,
1998). Ignoring good advice from David Chalmers and Sven Walter, I only read
these after it was too late to take them into account.
4. Shoemaker (1990, 112-3) first pointed out its importance.
5. Horgan would now emphasize just this feature (Horgan and Tienson
2000).
 125

property red – fancy that!”

So Mary seems to acquire a belief about tomatoes, moreover one
that she would naturally express using a demonstrative to refer to the
property red. It would be nice, though, to set tricky issues about
demonstrative belief aside. Fortunately we can do that by noting that
there is another belief that Mary seems to acquire: the tomato ap-
pears a distinctive way to her, and she comes to believe that it is that
way. Here the property red is not the subject of predication, picked
out by ‘this property’; rather, it is the property predicated.

The content of perception. Perceptual experiences, like the familiar

propositional attitudes, are representational or intentional mental
states. And, also like the propositional attitudes, they have proposi-
tional content, which specifies the way the world seems to the sub-
ject. If the world is this way, the experience is veridical. If the world
is not this way, the experience is some kind of illusion. At any rate, it
is widely accepted that perceptual experiences have propositional
content, and here it will be assumed without argument.

2. The puzzle explained

It will be useful to focus exclusively on a particular example. Partly

to get off the restricted diet of tomatoes, and partly in honor of G. E.
Moore, let it be the experience of looking at a “blue bead” (Moore
1903). This experience is an event: when it occurs, the world appears
a certain way to its subject. Setting aside a worry about whether the
bead is represented as a bead, and temporarily glossing over the fact
that the bead looks to be a specific shade of blue, this much seems
right: the content of the experience is, inter alia, that the bead is blue.
Although the experience represents the bead as blue, that doesn’t
mean that the experience itself is blue, as Moore saw very clearly.
(Good job too, because the experience isn’t blue.)
Mary has never seen any blue objects. In fact, nothing has even
looked blue to her. She has fully functioning color vision. Moreover,
Mary knows a lot about the colors, and in particular she knows a fair
amount about the color blue: that the sky and sapphires are blue, that
ultramarine is a shade of blue, and so on. We need not suppose that
 126

Mary is as knowledgeable as Jackson describes her – indeed, per-

haps she’s a scientific ignoramus. And maybe Mary has even seen a
few objects with colors other than blue. One day she learns she will
be shown a blue bead. Imagine that Mary is confident about the reli-
ability of her visual system. What will happen when Mary is shown
the bead? Intuitively she will acquire the belief that the bead is a cer-
tain highly distinctive way; as I shall cautiously put it, Mary will ac-
quire a belief about the color of the bead.
If Mary really does acquire such a belief, this is very puzzling.
When she sees the bead, vision supplies her with the information
that the bead is blue. (It also supplies her with other information, for
instance about the bead’s shape; but this does not seem to be rele-
vant.) However, Mary already believes that the bead is blue. So why
would she acquire a belief about the bead’s color? For vividness,
imagine that Maud is Mary’s visual system. When Mary sees the
bead, Maud tells Mary that the bead is blue. However, this was en-
tirely to be expected, because Mary believes that the bead is blue and
that Maud is a reliable guide to Mary’s environment. Mary, then,
ought to be entirely unimpressed by Maud’s old news. But that’s
wrong: the colorful condition of the bead as revealed by Maud will
come as a big surprise.
Setting out this reasoning a little more formally:

1. Before seeing the bead, Mary believes that the bead is blue.
2. Mary’s visual experience represents that the bead is blue.
3. If someone believes that an object is blue, and then has a visual
experience that represents that the object is blue, then (various
qualifications aside), she will not acquire a belief about its color.
Therefore:
4. When she sees the bead, Mary will not acquire a belief about its
color.

The complement clauses should be read opaquely. Thus if Mary

only believes that the bead has such-and-such surface spectral reflec-
tance, then premise 1 is false, even if blueness is the property of hav-
ing such-and-such reflectance.
The “various qualifications” in premise 3 need not detain us: the
person must identify the object she sees as the one that she believes
 127

is blue; she must not take herself to be the victim of some kind of il-
lusion; her visual experience must not also represent the bead as be-
ing some other color; and so on.
The difficulty is that this argument has plausible premises, an im-
plausible conclusion, and it’s valid (or, more exactly, would be if we
filled in a few details). The puzzle is to explain either why one of the
premises is false, or why we should embrace the conclusion.

3. Some suggestions

Well, why not embrace the conclusion?

First Suggestion: embrace the conclusion

According to this suggestion, our intuitions to the contrary notwith-
standing, Mary will not acquire any belief about the color of the
bead.
Since Mary certainly seems to acquire something, a proponent of
this suggestion has to equip Mary with some surrogate that we have
mistakenly taken to be a new belief about the color of the bead. Per-
haps the surrogate is some other belief. But what might that be?
Suppose Mary believes – as she would put it – that George W.
Bush is in town. Mary’s aunt Maud (not to be confused with her vi-
sual system of the same name) now says to Mary, “Dubya is in
town”, to which Mary replies, “Fancy that!” Mary evidently ac-
quires a belief. Suppose, though, that Uncompromising Millianism
about names is true – the semantic content of a name is just its refer-
ent, and coreferential names are substitutable salva veritate in prop-
ositional attitude contexts. Then, since Dubya = George W. Bush,
Maud apparently told Mary what she believed already, that George
W. Bush was in town. So Mary ought to be entirely unimpressed by
Maud’s old news. Isn’t this a big problem for Uncompromising
Millianism? No. Even if Mary believed all along that Dubya is in
town, there are many candidates for the belief that she acquires when
Maud says to her, “Dubya is in town”. An obvious one is a belief
about the representation Maud uses to express the proposition that
Dubya is in town: for instance, that someone called ‘Dubya’ is in
town.
 128

If this is a promising analogy for the perceptual case, it suggests

that when Mary sees the bead she acquires a belief about the repre-
sentation Maud uses to convey the information that the bead is blue.
For instance, Mary might acquire a belief about mental paint: “Men-
tal properties of the experience that represent the [blueness of the
bead]” (Block forthcoming, 9).
However, part of the point of contemporary insistence on the “di-
aphanousness” or “transparency” of experience is that one is not
aware of perceptual representations, unlike linguistic ones. When
one sees a blue bead, one is aware of what is represented, not of what
does the representing. This way of developing the first suggestion is
therefore not phenomenologically plausible.6
Following Lewis (1983) and Nemirow (1980), perhaps Mary
simply acquires a cluster of abilities, which Lewis characterizes as
the “abilities to remember, imagine, and recognize” experiences of
colors (1988, 289). We can add the ability to recognize colored ob-
jects by sight to the list. Insofar as it makes sense to recognize an ex-
perience as of a blue object, one does it by concentrating on the col-
ors in the scene before the eyes. If Mary acquires the ability to recog-
nize visual experiences as of blue objects, she therefore also ac-
quires the ability to recognize objects as blue by sight.
It is very plausible that Mary acquires the ability to recognize ob-
jects as blue by sight. However, it is very hard to maintain this while
denying that Mary acquires a belief about the color of the bead. Con-
sider two scenarios. (A) Mary believes that the bead is blue; we don’t
show her the bead, but instead a blue ball, without telling her its
color. Despite believing that the bead is blue, Mary won’t recognize
the ball as being the same color as the bead. (B) Mary believes that
the bead is blue, sees the bead, and then we show her the blue ball,
again without telling her its color. Because Mary acquires the ability
to recognize objects as blue by sight when she sees the bead, this
time she will recognize the ball as being the same color as the bead.
Because Mary in (B) recognizes the ball as being the same color
as the bead, she must have remembered something about the bead’s
color. What is that she remembers? Not something that she could re-

6. Admittedly this is a bit quick (for the reason why, see Block forthcoming,
7-9); because of space limitations, it will have to do.
 129

member in (A) – in particular, that the bead is blue – because other-

wise in that scenario she would recognize the ball as being the same
color as the bead. Mary therefore remembers something, and so be-
lieves something, about the color of the bead in (B) that she doesn’t
in (A). She acquires this belief when she acquires the ability; hence
Mary acquires a belief about the color of the bead when she sees it.

Second Suggestion: deny premise 1

According to this suggestion Mary doesn’t believe that the bead is
blue, at any rate before seeing it. Neither does she believe, for in-
stance, that the sky is blue. In other words, using ‘concept’ in a stan-
dard and unobjectionable sense, Mary lacks the concept BLUE.7
This is because one needs to have certain kinds of experiences in or-
der to possess the concept BLUE – experiences that Mary doesn’t
have before seeing the bead. To be sure, Mary utters sentences like
‘The sky is blue’ before she sees any blue objects, but Mary is not
thereby expressing the belief that the sky is blue; rather, she is ex-
pressing the metalinguistic belief that the sky has a property called
‘blue’, or something of the sort.

7. In this unobjectionable sense, an instance of ‘S possesses the concept F’ is

true iff a corresponding instance of ‘S fs that … F …’is true (where ‘fs’is replaced
by a propositional attitude verb like ‘believes’). On another usage, concepts are
modes of presentation or Fregean senses. In order to “possess a concept” in this
second sense (according to Peacocke 1992), the subject has to meet the concept’s
“possession condition”, which “states what is required for full mastery of [the]
concept” (29). Clearly it does not follow that if S possesses the concept F in the first
sense she possesses it in the second (someone who rejected Fregeanism entirely
could of course hold that people possess concepts in the first sense). And in fact
Peacocke’s own theory allows that S may believe that … F … without having “full
mastery” of the concept F, and so without possessing it (1992, 27-33).
This last observation suggests the following way of responding to the puzzle.
Suppose it could be made out that Mary lacks “full mastery” of the concept
BLUE, but that despite this she does believe that the bead is blue (following
Burge 1979). On this view, premise 1 is true, but because of Mary’s lack of con-
ceptual competence there seems to be some prospect of denying premise 3. How-
ever, surely Mary does not lack “full mastery” of the concept BLUE. (In Burge’s
terminology, Mary does not have an “incomplete understanding” of the “no-
tion”.) We need not suppose that Mary is under any misconception about the
boundaries of the color, for instance (cf. Burge 1979, 81-2).
 130

Anyone attracted by this account will say that Mary acquires the
concept BLUE when she sees the bead. Suppose she gives the name
‘B’to the distinctive property that the bead appears to have, and then
draws the conclusion, as she would put it, that “B is the property
blue”. Her way of stating the conclusion is misleading: Mary has ac-
tually concluded that the property blue is called ‘the property blue’.
It is an interesting sociological fact that philosophers are sharply
divided about whether someone in Mary’s predicament would have
the concept BLUE, which on the face of it seems to be a fairly simple
issue. Some, following a tradition probably as long as philosophy it-
self, find it obvious that she doesn’t, others find it obvious that she
does.
Ignoring some exegetical subtleties, those who hold that Mary
lacks the concept BLUE include Wittgenstein, Nagel, Williams,
Peacocke, McDowell, and Harman; those who hold that she has the
concept include Frege, Smart, Jackson, Block, Shoemaker, and
Fodor.8
Nagel, for example, writes of “a martian scientist” who “would
never be able to understand the human [concept] of [a] rainbow”,
which is “connected with a particular point of view and a particular
visual phenomenology” (1974, 443). On the other hand, Jackson ap-
parently supposes that Mary has color concepts despite lacking
color experiences: she is the world’s greatest color scientist, after all.
Physically omniscient Mary understands ‘red’ as well as anyone,
and knows she will see a red object when she is released from her
cell. The only thing she doesn’t know is what seeing red is like.
Despite the weight of tradition, it isn’t very plausible to hold that
Mary lacks the concept BLUE before seeing the bead.9 Take cases
more extreme than Mary’s – the blind and those suffering from vari-
ous forms of color blindness. Such people can use color vocabulary
perfectly well, and give every indication of believing, for example,
8. See Wittgenstein 1977, I, 13; Nagel 1974; Williams 1978, 241-52; Pea-
cocke 1984; McDowell 1985; Harman 1990; Frege 1918; Smart 1963, 80-1;
Jackson 1982; Block 1990; Shoemaker 1994; Fodor 1998. Perhaps the exegetical
subtleties are not quite so subtle: some of those classified as holding that Mary
lacks the concept may well only admit that someone blind would lack the con-
cept, for instance Peacocke and Harman.
9. Here I agree with Thau (1998, ch. 5).
 131

that the sky is blue. In fact, the acquisition and use of color vocabu-
lary by young children doesn’t seem to vary very much between the
blind and the sighted (Landau and Gleitman 1985). Evidently the
blind lack something very useful: they can’t recognize the colors by
sight. But it isn’t clear at all why this should prevent them from fully
understanding ‘blue’ or from believing that the sky is blue, any more
than someone’s inability to recognize a chiliagon by sight or another
sensory modality should prevent her from fully understanding
‘chiliagon’ or from believing that chiliagons have many sides.10
Of course, the other side of the debate does have a point. It is not
implausible at all to think that Mary does lack a concept before see-
ing the bead.11 However, this missing concept is not the concept
BLUE, for the reasons given, and therefore even if the missing con-
cept can help with the puzzle, it won’t allow us to deny premise 1. So
we should press on.

Third Suggestion: deny premise 2

According to the third suggestion, Maud is not telling Mary that the
bead is blue. There are three main ways this suggestion can be devel-
oped. First, the Fregean way: Maud is telling Mary that the bead is
blue*, where ‘blue*’ has the very same reference as ‘blue’, but a dif-
ferent sense. (This comes in different flavors, depending on the pre-
ferred conception of sense.) Second, Maud is telling Mary that the
bead is, say, blue17 – a specific shade of blue. Third, Maud is telling

10. Yablo (1995, 492) makes the interesting observation that we are tempted
to treat ‘square’ differently from ‘chiliagon’ (or ‘milliagon’, to use his example).
In the former case of ‘square’ we are inclined to say that someone doesn’t fully
understand the word if she lacks the appropriate perceptual recognitional capac-
ity. (However, this inclination is mistaken, or so I’ve suggested.) A similar con-
trast applies to ‘blue’ and ‘image blue’ (as used by the Glidden paint company to
name a fine-grained shade of blue). Despite lacking (as we all do) a recognitional
capacity for fine-grained shades, the Glidden people presumably understand
their own terminology (cf. Block 1999; see also the discussion of ‘blue17’ under
the “third suggestion” below).
11. See Crimmins 1989. According to him, and in my terminology, Mary has
the concept BLUE, but lack a concept of the same property with more demanding
possession conditions. (In Crimmins’ terminology, Mary has a idea of blue, but
lacks the concept.)
 132

Mary that the bead has some non-color property, which we may call
‘phenomenal-blue’ (Thau 1998, ch. 5).12
The following objection applies to all these ways. Take, for exam-
ple, the proposal that Maud is telling Mary that the bead is blue*.
Imagine that we introduce the expression ‘blue*’ into our language.
Surely Mary could understand it without having had special kinds of
experiences. After all, once it is agreed that Mary needn’t have spe-
cial experiences in order to understand ‘blue’, there doesn’t seem
much motivation for insisting that ‘blue*’ should be treated any dif-
ferently. (And if special experiences are required in order to under-
stand ‘blue*’, the present proposal is superfluous: we might as well
deny premise 1 instead.) So we may suppose that, before seeing the
bead, Mary understands ‘blue*’ and believes (on the basis of testi-
mony) that some things are blue*, in particular the bead she is about
to see. Intuitively, Mary will still acquire a belief about the color of
the bead when she sees it. This is the puzzle about perception all over
again, with ‘blue’ replaced by ‘blue*’. Thus no progress has been
made.
Similar remarks tell against the second and third ways: replace
‘blue*’ in the above paragraph with either ‘blue17’ or ‘phenome-
nal-blue’.

Fourth Suggestion: deny premise 3

The third premise, recall, is:

3. If someone believes that an object is blue, and then has a visual

experience that represents that the object is blue, then (various
qualifications aside), she will not acquire a belief about its color.

Taking a hint from Dennett (1995), perhaps premise 3 fails because

the content of Mary’s visual experience is very rich. Mary’s visual
12. Cf. Shoemaker 1994. Thau’s proposal is analogous to Castañeda’s re-
sponse to Frege’s puzzle: the morning star isn’t identical to the evening star
(Castañeda 1974).
Because phenomenal-blue is not a color, Thau embraces the conclusion in ad-
dition to denying premise 2. According to him, Mary doesn’t acquire a belief
about the color of the bead. She does acquire a belief, though – unsurprisingly,
one about the phenomenal-color of the bead.
 133

system Maud isn’t just telling Mary that the bead is blue, she’s tell-
ing her that it’s blue and flat in this tiny region, blue and curved in an-
other region … and so on and so on. It’s a long story, and in any real-
istic description of Mary’s initial predicament she won’t believe
anything quite so complicated. When she sees the bead, she acquires
this complicated belief about its color.
The difficulty with this attempt is that intuitively Mary will ac-
quire a belief even if we drastically impoverish the content of her vi-
sual experience. If we imagine Mary looking through a reduction
screen at a tiny blue patch, and so imagine that the content of her ex-
perience is relatively poor, it seems just as plausible as it did origi-
nally that Mary will acquire a belief about the color of the thing she
sees.13
Taking a hint from Fodor (1990), let us try something else. To
simplify matters, suppose that Mary has a “belief box” in her head,
in which expressions taken from Spanish, French and English may
be inscribed (in a special Mentalese font). And suppose that the
Mentalese predicates ‘azul’ and ‘bleu’, despite having the very
same semantic content – the property blue – have different “func-
tional roles”. In particular, ‘bleu’ is intimately linked to the produc-
tion of Mary’s utterances of sentences like ‘The bead I am about to
see is blue’, ‘The sky is blue’, etc. ‘Azul’, on the other hand, is inti-
mately linked to her visual system. When Mary sees the bead, and
has an experience with the content that the bead is blue, the sentence
‘The bead is azul’ appears in her belief box. Following Fodor, sup-
pose that beliefs are not just individuated by their contents, but also
by the functional role of their mental representations. Then Mary ac-
quires a new belief, because the sentences ‘The bead is azul’ and
‘The bead is bleu’ play very different roles in Mary’s cognitive
economy.
However, this proposal is at best an explanation of why Mary will
acquire a belief about the color of the bead if she possesses a particu-
lar cognitive architecture. But the guiding intuition in the Mary ex-
ample is that any rational person in Mary’s epistemic position who
gets information about the bead in the visual way Mary gets it and
who trusts her senses will acquire a belief about the color of the

13. See Block 1995, 277, and Thau 1998, ch. 5.

 134

bead. It is hard to see how the present proposal can explain this. Sup-
pose Mary’s Mentalese only contains ‘bleu’, linked as before to sen-
tences containing ‘blue’ and also linked to her visual system in the
manner of ‘azul’. So, when Mary sees the bead, the sentence ‘The
bead is bleu’ will appear in Mary’s belief box – if it’s not there al-
ready. Since it is there already, seeing the bead will produce no
change. Mary will not acquire a belief about the color of the bead:
she won’t express any surprise or amazement when she sees it. How,
on the present proposal, could Mary’s failure to acquire a belief pos-
sibly be irrational?

4. The limits of language

To try to break the gridlock, let us return to the Fregean proposal:

Maud is not telling Mary that the bead is blue, but rather that it is
blue*, where ‘blue*’ has the same reference as blue, but a different
sense. The objection was that the puzzle arises all over again, with
‘blue*’ in place of ‘blue’. That objection relied on two principles.
The first is that Maud’s pronouncement about the bead could be ex-
pressed in language.14 The second is that no linguistic expression re-
quires special experiences in order to be understood. The first princi-
ple allows us to introduce the expression ‘blue*’, and putting that to-
gether with the second principle gives us another version of the puz-
zle, with Mary initially believing that the bead is blue*. Could either
of these principles be resisted?
The second principle seems on firm ground. Understanding lan-
guage is in a way cognitively undemanding, as Kripke and others
have observed. In order to have the English expressions ‘Aristotle’,
‘tiger’, and ‘water’in one’s vocabulary, for example, one need not be
at all epistemically intimate with their referents. Someone not per-
ceptually acquainted with tigers, and profoundly ignorant of basic
facts about them, is not thereby prevented from fully understanding
‘tiger’. With any system of communication there’s a tradeoff be-

14. Given that Maud’s pronouncement is explained in terms of “a different

sense”, on a strict Fregean usage it immediately follows that it is linguistically
expressible. The usage in this paper is therefore not so strict.
 135

tween the amount of specialized knowledge and experience required

to master it, and the size of the community who can use it. Language
has erred on the side of inclusivity: children, couch potatoes, and
Helen Keller can all speak of topics of which they have little knowl-
edge or first-hand experience.
The first principle, though, is more open to question. And if it is
incorrect, then the puzzle can be solved as follows. Premise 2 is
false. Mary’s experience doesn’t represent that the bead is blue, but
rather that it’s … well, ex hypothesi no bit of language will do the
trick! Using ‘Ã’ to pretend to say what can’t be said, Mary’s experi-
ence represents that the bead is Ã. Since Mary trusts her senses, she
acquires the belief that the bead is Ã. This new belief is a belief
about the color of the bead, because ‘Ã’ refers to the property blue.
(So, on a transparent reading, Mary’s experience does represent that
the bead is blue – but recall that the complement clauses should be
read opaquely.) The puzzle cannot be resurrected for ‘Ã’, because
that could not be an expression of a natural language: the proposition
that the bead is à is not linguistic content.15,16 If this solution to the
puzzle is right, the content of perception, although it can be remem-
bered and believed, cannot be (entirely) expressed in language, and
is in this sense ineffable.
In fact, there is already a theory of ineffable perceptual content in
the literature. According to Peacocke’s A Study of Concepts (1992),
perceptual experiences have “nonconceptual content”, a special
kind of propositional content that is a combination of “scenario con-
tent” and “protopropositional content”. These two sorts of abstract
objects are built to Russellian specifications: a protopropositional
content is a Russellian proposition, while a scenario content is
something more complicated, but constructed along similar lines.
Language and thought, on the other hand, exclusively have contents
that are composed of Fregean senses or “concepts”.17
15. Let us not fuss about ‘the bead’: insert another squiggle, if you wish. It is
worth noting that some theories of propositions tie them too closely to language
for our purposes, for instance Schiffer’s “pleonastic” theory (1996).
16. Since formal “languages” are cheap, the proposition that the bead is Ã
could be expressed by a sentence of some language. But such a language could
not be used for communication in the manner of natural languages.
17. Cf. ‘concept’ as introduced in section 3, and fn. 7 above.
 136

However, Peacocke’s nonconceptual content, although ineffable,

isn’t quite what we want. Nonconceptual content cannot be be-
lieved, but Mary believes the ineffable proposition that the bead is Ã
when she sees it. Still, Peacocke’s discussion amply demonstrates
the important point that ineffable content need not be immune to rig-
orous theorizing. The proposition that the bead is à is not express-
ible in language, but that is no barrier to a substantive account of it.
The thought that something about perceptual experience is ineffa-
ble is hardly new. According to Frege, for instance, the “inner
world” of “sense impressions” and other “ideas” resists linguistic
expression. No sentence, he thinks, can ever fully characterize my
“sense impression of green”. One of his reasons – perhaps charitably
interpreted – seems to be this. If there is such a sentence, then the se-
mantics of language can be made to vary just by varying the “ideas”
in the minds of speakers. In particular, the inverted spectrum hypoth-
esis should be an example of “inverted semantics”: what you mean
by ‘red’ is what I mean by ‘green’, and vice versa. But this is absurd:
you and I speak the same language.18 (The later Wittgenstein may be
represented as endorsing the reasoning, but instead locating the ab-
surdity in Frege’s ineffability hypothesis, concluding that there is no
inner world of “ideas”, as Frege conceives it.19)
Another example is C. I. Lewis’s rather obscure doctrine that “the
given” cannot be “described”: “in a sense the given is ineffable, al-
ways” (1929, 52-3). Lewis also mentions the inverted spectrum, ap-
parently thinking along lines similar to Frege’s (73-6).
Peacocke is not the only contemporary proponent of ineffability.
In footnote to the passage we examined earlier in section 1, Horgan
all but explicitly says that Mary’s new knowledge is not linguisti-
cally expressible (1994, 151, fn. 5). Block has claimed, using rea-
soning of the sort we just attributed to Frege, that “phenomenal char-
acter is not expressible in English” (1996, 47). Jackson commits
himself to the view that some knowlege about qualia cannot be
stated in language (1986, 295). And Raffman has argued at length,

18. The argument is suggested by Frege 1918/1997, 334-5, together with

Frege 1879/1980, §27.
19. See the “inverted beetles” and “absent beetles” of §293, Wittgenstein
1953.
 137

using music as the main example, that a certain level of perceptual

representation is ineffable (1993).
On the position defended here, these philosophers have an impor-
tant insight.
Unfortunately, we cannot conclude by announcing that the puzzle
about perception has received a satisfying solution. For that, we
need a theory of linguistic content, and a theory of perceptual con-
tent, which together explain why perceptual content, even though it
can be the content of belief, isn’t (or isn’t entirely) linguistic content.
But I hope a mere solution is enough for one paper.

REFERENCES

Block, N. 1990. Inverted earth. Philosophical Perspectives 4, ed. J.

Tomberlin. Atascadero, CA: Ridgeview.
Block, N. 1995. How many concepts of consciousness? Behavioral and
Brain Sciences 18, 272-84.
Block, N. 1996. Mental paint and mental latex. Philosophical Issues 7, ed.
E. Villanueva. Atascadero, CA: Ridgeview.
Block, N. 1999. Sexism, racism, ageism and the nature of consciousness.
Philosophical Topics 26 (1&2), The Philosophy of Sydney Shoemaker,
ed. R. Moran, A. Sidelle, and J. Whiting. Fayetteville: University of
Arkansas Press.
Block, N. Forthcoming. Mental paint. Others on Burge: 10 Essays With
Responses From Tyler Burge, ed. M. Hahn and B. Ramberg. Cam-
bridge, MA: MIT Press. (Reference to the manuscript:
www.nyu.edu/gsas/dept/philo/faculty/block/papers/mentalpaint.pdf)
Burge, T. 1979. Individualism and the mental. Midwest Studies in Philos-
ophy 4, ed. P. French, T. Vehling and H. Wettstein. Minneapolis: Uni-
versity of Minnesota Press.
Byrne, A. 1999. Cosmic hermeneutics. Philosophical Perspectives 13, ed.
J. Tomberlin. Oxford: Blackwell.
Castañeda, H.-N. 1974. Thinking and the structure of the world. Philo-
 138

sophia 4, 3-40.
Crimmins, M. 1989. Having ideas and having the concept. Mind and Lan-
guage 4, 280-94.
Dennett, D. C. 1995. The path not taken. Behavioral and Brain Sciences
18, 252-3.
Fodor, J. 1990. Substitution arguments and the individuation of beliefs.
Meaning and Method, ed. G. Boolos. Cambridge: Cambridge Univer-
sity Press.
Fodor, J. 1998. There are no recognitional concepts – not even RED.
Philosophical Issues 9, ed. E. Villanueva. Atascadero, CA: Ridgeview.
Frege, G. 1879/1980. The Foundations of Arithmetic. Oxford: Basil
Blackwell.
Frege. G. 1918/1997. Thought. The Frege Reader, ed. M. Beaney. Oxford:
Blackwell.
Graham, G., and T. Horgan. 2000. Mary Mary, quite contrary. Philosophi-
cal Studies 99, 59-87.
Harman, G. 1990. The intrinsic quality of experience. Philosophical Per-
spectives 4, ed. J. Tomberlin. Atascadero, CA: Ridgeview.
Horgan, T. 1984. Jackson on physical information and qualia. Philosophi-
cal Quarterly 34, 147-52.
Horgan, T., and J. Tienson. 2000. Narrow content and the phenomenology
of intentionality. Presidential address given at the 26th SPP meeting.
Jackson, F. 1982. Epiphenomenal qualia. Philosophical Quarterly 32,
127-36.
Jackson, F. 1986. What Mary didn’t know. Journal of Philosophy 83,
291-5.
Landau, B., and L. Gleitman. 1985. Language and Experience: evidence
from the blind child. Cambridge, MA: Harvard University Press.
Lewis, C. I. 1929. Mind and the World Order. New York: Scribner.
Lewis, D. 1983. Knowing what it’s like: postscript to “Mad pain and mar-
tian pain”. Lewis, Philosophical Papers vol. 1, Oxford: Oxford Univer-
sity Press.
Lewis, D. 1988. What experience teaches. Proceedings of the Russellian
Society 13, 29-57. (Reference to the reprinting in Lewis, Papers in
 139

Metaphysics and Epistemology. Cambridge: Cambridge University

Press, 1999.)
McDowell, J. 1985. Values and secondary qualities. Morality and Objec-
tivity, ed. T. Honderich. London: Routledge and Kegan Paul.
Moore, G. E. 1903. The refutation of idealism. Mind 12, 433-53.
Nagel, T. 1974. What is it like to be a bat? Philosophical Review 83,
435-50.
Nemirow, L. 1980. Review of Nagel’s Mortal Questions. Philosophical
Review 89, 475-6.
Nida-Rümelin, M. 1995. What Mary couldn’t know: belief about phe-
nomenal states. Conscious Experience, ed. T. Metzinger. Exeter, Eng-
land: Imprint Academic.
Nida-Rümelin, M. 1998. On belief about experiences. An epistemological
distinction applied to the knowledge argument against physicalism.
Philosophy and Phenomenological Research 58, 51-73.
Peacocke, C. 1984. Colour concepts and colour experience. Synthese 58,
365-82.
Peacocke, C. 1992. A Study of Concepts. Cambridge, MA: MIT Press.
Raffman, D. 1993. Language, Music, and Mind. Cambridge, MA: MIT
Press.
Schiffer, S. 1996. Language-created language-independent entities.
Philosophical Topics 24, 149-67.
Shoemaker, S. 1990. Qualities and qualia: what’s in the mind? Philosophy
and Phenomenological Research 50, Supplement, 109-31.
Shoemaker, S. 1994. Phenomenal character. Noûs 28, 21-38.
Smart, J. J. C. 1963. Philosophy and Scientific Realism. London:
Routledge & Kegan Paul.
Thau, M. 1998. Belief and Perception: a unified account. Ph. D. diss.,
Princeton University.
Thau, M. Forthcoming. Consciousness and Cognition. Oxford: Oxford
University Press.
Tye, M. 1995. Ten Problems of Consciousness. Cambridge, MA: MIT
Press.
Williams, B. 1978. Descartes: the project of pure enquiry. London: Pen-
 140

guin Books.
Wittgenstein, L. 1953. Philosophical Investigations. Oxford: Blackwell.
Wittgenstein, L. 1977. Remarks on Colour. Berkeley: University of Cali-
fornia Press.
Yablo, S. 1995. Singling out properties. Philosophical Perspectives 9, ed.
J. Tomberlin. Astascadero, CA: Ridgeview.