Arif Ahmed Wittgensteins Philosophical Investigations A Readers Guide 2010

WITTGENSTEINS PHILOSOPHICAL
INVESTIGATIONS
Continuum Readers Guides
Continuums Readers Guides are clear, concise and accessible introductions to
classic works of philosophy. Each book explores the major themes, historical
and philosophical context and key passages of a major philosophical text,
guiding the reader towards a thorough understanding of often demanding
material. Ideal for undergraduate students, the guides provide an essential
resource for anyone who needs to get to grips with a philosophical text.
Readers Guides available from Continuum:
Aristotles Nicomachean Ethics Christopher Warne
Aristotles Politics Judith A. Swanson and C. David Corbin
Berkeleys Principles of Human Knowledge Alasdair Richmond
Berkeleys Three Dialogues Aaron Garrett
Deleuze and Guattaris Capitalism and Schizophrenia Ian Buchanan
Deleuzes Difference and Repetition Joe Hughes
Derridas Writing and Difference Sarah Wood
Descartes Meditations Richard Francks
Hegels Philosophy of Right David Rose
Heideggers Being and Time William Blattner
Heideggers Later Writings Lee Braver
Hobbess Leviathan Laurie M. Johnson Bagby
Humes Dialogues Concerning Natural Religion Andrew Pyle
Humes Enquiry Concerning Human Understanding Alan Bailey and
Dan OBrien
Kants Critique of Aesthetic Judgement Fiona Hughes
Kants Critique of Pure Reason James Luchte
Kants Groundwork for the Metaphysics of Morals Paul Guyer
Kuhns The Structure of Scientific Revolutions John Preston
Lockes Essay Concerning Human Understanding William Uzgalis
Lockes Second Treatise of Government Paul Kelly
Mills On Liberty Geoffrey Scarre
Mills Utilitarianism Henry West
Nietzsches On the Genealogy of Morals Daniel Conway
Nietzsches The Birth of Tragedy Douglas Burnham and
Martin Jesinghausen
Platos Republic Luke Purshouse
Platos Symposium Thomas L. Cooksey
Rousseaus The Social Contract Christopher Wraight
Sartres Being and Nothingness Sebastian Gardner
Spinozas Ethics Thomas J. Cook
Wittgensteins Tractatus Logico-Philosophicus Roger M. White
WITTGENSTEINS
PHILOSOPHICAL
INVESTIGATIONS
A READERS GUIDE
ARIF AHMED
Continuum International Publishing Group
The Tower Building 80 Maiden Lane
11 York Road Suite 704
London SE1 7NX New York, NY 10038
www.continuumbooks.com
Arif Ahmed, 2010
All rights reserved. No part of this publication may be reproduced or
transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage
or retrieval system, without prior permission in writing
from the publishers.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
ISBN: HB: 978-0-8264-9263-0
PB: 978-0-8264-9264-7
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress.
Typeset by Newgen Imaging Systems Pvt Ltd, Chennai, India
Printed and bound in Great Britain by CPI Antony Rowe Ltd,
Chippenham, Wiltshire
v
CONTENTS
Preface vi
1. Context 1
2. Overview of Themes 8
3. Reading the Text 11
Section 1. The Augustinian picture 11
1.1. Unity of function 12
1.2. Ostension 16
1.3. Referentialism 22
1.4. Sentences are combinations of names 34
Section 2. Family resemblance and the ideal
of precision 40
2.1. Family resemblance 40
2.2. Vagueness 45
2.3. The nature of philosophy 61
2.4. PI 1347: The general form of the proposition 67
Section 3. Meaning and understanding 69
3.1. The cube 70
3.2. Dispositions 76
3.3. Understanding and guidance 84
3.4. The paradox 94
Section 4. Privacy 107
4.1. Private language 108
4.2. The inner and the outer 129
4. Reception and Influence 150
5. Guide to Further Reading 154
Notes 156
References 161
Index 165
vi
PREFACE
This book is aimed at second- and third-year undergraduates
who are coming to Wittgenstein for the first time. It doesnt cover
all of Philosophical Investigations, but it does cover everything
in it that you would expect to find in any course on the later
Wittgenstein: that is, the criticisms of the Tractatus, the material
on rule-following, and the material on private language.
I wrote the book in the summer of 2009 but it is based upon
lectures that I gave, in the Michaelmas terms of 2007 and 2008,
to those Cambridge students who were taking the Special
Subject paper (Wittgenstein) in Part II of the Philosophy
Tripos. I am grateful to those students for their comments
and questions. And I am grateful to Prof. Michael Luntley of
Warwick University for helpful comments on Chapter 1.
Quotations from Wittgenstein use single quotes and double
quotes exactly as he does.
1
CHAPTER 1
CONTEXT
If you are going to get anything at all out of Wittgensteins
Philosophical Investigations then you must come to it with some
knowledge of his Tractatus Logico-Philosophicus. That earlier
work is by far the single most important influence upon this later
one; for instance, the first 100-odd sections of Philosophical
Investigations constitute a direct and extended attack upon the
central theses and presuppositions of the Tractatus. Here I say
something about the background and content of that earlier
work. The first of these tasks involves a brief discussion of the
two most important philosophical influences upon it: Russell
and Frege.
RUSSELL
In the late nineteenth century the dominant position on the
British philosophical scene was idealism. A crude sketch of that
position is this: one can never have direct access to reality itself;
on the contrary human experience and thought themselves
interpose a sort of distorting lens between us and it, so that one
could never know reality as it is in itself but only as it appears to
us. This view derived ultimately from the transcendental idealism
that Immanuel Kant had first stated in his Critique of Pure
Reason (1782).
Russell rejected idealism; the nature of his alternative view
is set out with his usual clarity in The Problems of Philosophy
(Russell 1912). For anyone who is completely new to this whole
area, that book is probably the best place to begin.
Russell thought that we could have direct knowledge of reality;
the relation by which we achieved this he called acquaintance
(Russell 1912: 23). In particular that relation connects a knowing
subject to the immediate objects of its sensory knowledge: these
are not enduring physical objects like tables, mountains, and the
like but sensations themselves: colour-patches, sounds, smells,
WITTGENSTEINS PHILOSOPHICAL INVESTIGATIONS
2
and the like. These Russell called sense data (ibid. 4). Acquaintance
also connects the knowing subject with universals or concepts
like whiteness, diversity and brotherhood (ibid. 28).
He extended this account to the problem of intentionality:
that is, the question of how thoughts and language could
be about objects other than themselves. Russell thought that
acquaintance gives the answer: Every proposition which we
understand must be composed wholly of constituents with which
we are acquainted (ibid. 32). Thus the relation of aboutness
becomes that of identity: your thoughts just are the things
they are about; and they are your thoughts by virtue of your
acquaintance with them. And you here denotes not your body
but the non-sensory subject of acquaintance, which can however
also be its object, since one is probably acquainted with oneself
(ibid. 278).
But surely many propositions do not describe items of sensory
acquaintance or universals for instance, most of ones everyday
thoughts or utterances. If, for instance, I have the thought that
Henry Kissinger was an astute diplomat I am surely thinking
about a particular object Henry Kissinger with which I am
not acquainted, since (a) he is an enduring flesh-and-blood
object and not a sight or a smell; and anyway (b) I have never
met him.
Russells answer may be presented in two stages. The first stage
notes that although Henry Kissinger appears to correspond to
some element of my thought when I think that Henry Kissinger
was an astute diplomat what really occurs there corresponds
more precisely to an associated definite description, that is, a
phrase of the form The so-and-so (ibid. 28). So the thought
that I express in English by the sentence Henry Kissinger was an
astute diplomat is more precisely rendered by one of the form
The so-and-so was an astute diplomat, in which The so-and-so
should get replaced by whatever description I associate with
Henry Kissinger, for example, The last Secretary of State in
the second Nixon Administration.
The second stage is to note that sentences containing definite
descriptions may themselves be replaced with sentences that
contain only expressions for universals. Russell himself provided
the manual for making such replacements with his famous
theory of descriptions. That theory translates a sentence of the
CONTEXT
3
form The F is G into this: At least one thing is F, no more than
one thing is F, and whatever is F is G or in formal notation
xy ((Fy y = x) & Gx). So the next stage in our rewriting of
that sentence, which supposedly expressed a thought about
Henry Kissinger, is as follows: At least one person was the last
Secretary of State in the second Nixon Administration, at most
one person was, and whoever it was was an astute diplomat. All
reference to Henry Kissinger has now dropped out; a suitable
repetition of these stages will also eliminate all reference to
Nixon. The resulting sentence contains only reference to univer-
sals; and this reveals the true contents of the thought or
proposition that the initial sentence so misleadingly expressed.
We can thus see that after all that proposition does contain only
objects of my acquaintance, that is, universals. This process of
translating sentences into a form that reveals the true contents
of the underlying thought is what Wittgenstein called its analysis
(TLP 3.23.201).
FREGE
The formal analysis of definite descriptions to which I have just
alluded would not have been possible without the seminal work
of Russells German contemporary Frege. It was he who intro-
duced what is essentially the same quantifier-notation that is
taught to all first-year philosophers and mathematicians to
this day; doing so enabled him to give a systematic account of
arguments of a complexity that was well beyond the scope
of traditional logic. In particular it allowed for a systematic
treatment of arguments involving mixed quantification, that is,
those in whose formal renditions a universal quantifier occurs
within the scope of an existential one or vice versa. Everyone
is taller than someone is a simple instance of this; so too are
sentences in which definite descriptions take the position of the
grammatical subject, for example, The last Secretary of State in
the second Nixon Administration was an astute diplomat.
Here I should also mention three more of Freges many
important contributions to philosophy.
The first was the distinction between sense and reference.
A simple way to think about names is to suppose that they just
mean what they stand for, or as I shall sometimes say refer to or
denote (cf. PI 1
1
). But Hesperus and Phosphorus were names
4
for the Evening Star and the Morning Star before it was realized
that both were identical (to the planet Venus). So the statement
Hesperus = Phosphorus appears to be both true and inform-
ative. But if the names Hesperus and Phosphorus just mean
what they stand for then this sentence appears to be saying no
more than what we could learn from the triviality Hesperus =
Hesperus. Frege concluded that the meaning of a name was not
just its reference: it further contained an element called its sense,
which was some proprietary way of presenting its referent (Frege
1892: 57). This sense could be encoded in a definite description, so
that the sense that Henry Kissinger carries for me is that of the
last Secretary of State in the second Nixon Administration. It is
the fact that Hesperus and Phosphorus have different senses
that explains the informativeness of Hesperus = Phosphorus.
Now as a doctrine about simple names Wittgenstein rejected
this view in the Tractatus. He thought that ordinary English
or German names could be analysed away via the theory of
descriptions as already outlined; and further that any names
that survived this process of analysis would be simple in the
sense of meaning just what they denoted (TLP 3.2023.203).
But he accepted the corresponding view about sentences, for
he agreed with the second Fregean point that we must distinguish
the referents of its elements from the information that a sentence
conveys. This informational content is the sense of the sentence;
so Hesperus = Hesperus plainly has a different sense from
Hesperus = Phosphorus even though the elements of the two
sentences refer to the same thing. In the Tractatus Wittgenstein
identified the sense of a sentence with its truth-conditions: that is,
the totality of possible situations that would make it true. (This
follows from TLP 4.4.) So even though the two sentences seem
to say the same thing about the same thing, The last Secretary
of State in the second Nixon Administration was an astute
diplomat and The person who accompanied Nixon to China
was an astute diplomat actually convey different senses, since
there is a possible situation for example, one where it was
Mrs Nixon and not Kissinger who went with Nixon to China
in which the first sentence is true and the second sentence is
false. The sense of a sentence is the proposition or thought that
it expresses; it is what we judge to be true when we make any
judgement at all.
CONTEXT
5
The third Fregean point was the famous context principle: not
to ask after the meaning of a name or any other word except as
it occurs in a sentence. What this means in practice is that if we
know the meaning of every sentence in which a sign occurs,
then we know the meaning of the sign itself there is no further
question to be asked about its meaning. Wittgenstein was in
agreement with this in both the Tractatus and Philosophical
Investigations (TLP 3.3, PI 49). In the Tractatus he took the
constituents of the propositions to be not words but variables
ranging over all the propositions that contained them: that is to
say, they were functions that took as values a range of propositions
(TLP 3.313). For instance, the proposition Brutus killed Brutus
contains the function x killed y, a function that takes, for
example, the value Brutus killed Caesar when x takes the value
Brutus and y takes the value Caesar.
As the preceding discussion has already indicated, some of the
most important Tractarian doctrines were either elaborations
upon or reactions to certain claims of Frege and Russell. Let us
now set those doctrines in their immediate context, that is, the
theory of meaning of the Tractatus itself.
THE TRACTATUS
In the Tractatus Wittgenstein was concerned to answer the
question of intentionality: how can thought and language
manage to be about something external to themselves? He
accepted the referentialist theory that names mean just what
they denote; he was therefore also committed to the Russellian
programme of analysing everyday sentences to reveal the thoughts
they express as combinations of propositions containing only
such simple names as meet this demand.
Several further questions arise at this point. First, what sorts
of things did he think that the names of a fully analysed
language denoted? Second, how did these names combine to
form meaningful sentences? Third, how did these sentences
combine to form the sentences of everyday language?
What dictated his answer to the first question was a certain
departure from Russells epistemological concerns. Among
Wittgensteins most fundamental ideas at this point was that
our thoughts have a definite sense: each thought exhaustively
divides the space of possibilities into those that make it true and
6
those that make it false. But in order to say anything about such
possibilities, there must in all of them be something for the
thought to be about. So what the names in the thought denote
are things that exist in all possible worlds.
It is obvious that an imagined world, however different it
may be from the real one, must have something a form in
common with it. (TLP 2.022)
These objects which together constitute the form of the world
must therefore be indestructible. And since whatever is composite
could be decomposed, that means that they are completely simple
entities. Wittgenstein does not say anything further about their
nature in the Tractatus; nor does he say whether they are identical
to the objects of Russellian acquaintance though we have no
reason to think that they are.
His answer to the second question was the famous picture
theory of meaning. A sentence consists in an arrangement of
names like symbols on a map; that two names stand in a relation
(e.g. that Jane and John stand to the left and the right of the
word loves) tells us truly or falsely that a certain state of
affairs obtains in reality (e.g. that Jane loves John): TLP 3.1432.
According to this pictorial theory of meaning a language can
never say what its terms refer to without presupposing another
one; and if my thought is itself in a language (that language
which alone I understand TLP 5.62) then this means that the
formal similarity between my language and the reality it describes
can never be explicitly grasped. However language itself is in a
sense an inarticulate signal about that form: it is as it were the
medium through which the world communicates its nature to us.
For by figuring in the subject-matter of a language with such-
and-such names in it, the world reveals to us that it contains
such-and-such objects.
But this is something that can only be shown. It cannot be said
(TLP 4.126). Philosophy itself, therefore, and in particular the
doctrine of the Tractatus, falls into a kind of nonsense when we
try to express it in propositional form (TLP 6.54). Philosophy is
not a body of doctrine but an activity.
It is evident how this contributed to the mysticism for which
the Tractatus is famous. And that there are limits to what can be
CONTEXT
7
said was a theme that haunted Wittgenstein throughout his career.
In the later work it re-emerges in the treatment of ostension (PI
2835) and of rule-following (PI 857, 138201). But there is a
difference. What was mysterious in the earlier work was located
in the hidden nature of reality as revealed by language. But in
the later work and this is why it is so disorienting it is the
everyday that begins to seem mysterious. Let us now turn to it.
8
CHAPTER 2
OVERVIEW OF THEMES
Four main themes run through Philosophical Investigations: the
idea that meaningful sentences are combinations of names;
the idea that what can be said at all can be said precisely; the
idea that meaning and understanding are mental processes that
accompany speech and writing; and the idea that we can talk
and think about our inner sensational states quite independently
of what is going on around us. This section briefly outlines his
treatment of each one.
We saw that in the Tractatus Wittgenstein had taken thoughts
to depict possible situations by combining the names that occur
within them. In Philosophical Investigations he disputes this view
not directly, but by attacking a much broader syndrome of
which it is but one symptom. That syndrome may be called
the Augustinian conception of meaning, because Wittgenstein
begins the work with a quotation from St Augustine that illus-
trates many of its aspects. He then proceeds to attack these ideas.
Perhaps most important among them is the preconception that
also underlay the Tractatus: namely, that there is one single
phenomenon of linguistic meaning and that this demands one
single theoretical explanation. We shall discuss his attack on this
and other Augustinian views in Section 1.
It may be that languages do not all fall under one pattern but
rather that they form a family in which different similarities
obtain between different types of linguistic activity. In that case
language would be a family resemblance concept. But then
calling something a language is not saying anything definite! But
who says that every sentence can or even should say something
definite? The Tractatus had insisted that every sentence say
something precise so that its truth or falsity was determined for
every possible situation; otherwise it does not express a thought
at all. Wittgenstein now argues as we shall see in Section 2
that that is such a narrow conception of language and of thought
OVERVIEW OF THEMES
9
as to be inadequate to most of our actual speech, writing and
thought. The apparent crudity of language does not conceal
perfectly precise thoughts whose content philosophy may then
uncover; indeed it is not the business of philosophy to uncover
anything at all. We conclude Section 2 with a brief discussion of
this picture of the philosophical enterprise.
Nothing is hidden: Wittgenstein argues in PI that what gives
words meaning is not some relation between the simple signs
concealed within our language and the simple objects hidden
beneath the complexity of the world; it is rather our overt use
of the words of our actual language. But it can easily seem that
this is wrong because it can easily seem that some hidden psy-
chological process must accompany the meaningful use of words;
otherwise what distinguishes our speech from a meaningless if
highly structured pattern of sound? Again, when you understand
a word for the first time it seems as though what has gone through
your mind then, and not how you subsequently use the word, is
its meaning; for otherwise how could the former be what guides
the latter? Wittgenstein shows that no such accompanying
process could be what we want meaning and understanding
to be. These passages are apt to produce a kind of intellectual
dizziness in the reader, for it starts to seem impossible that
anyone should ever mean anything. We conclude Section 3 by
separating this feeling from Wittgensteins own view before
isolating what really is disquieting about the latter.
We have seen that Russell thought that one could refer to
the objects of ones immediate sensory acquaintance; this was
something that he had in common with many empiricists. It is
very easy to suppose that one could set up a means for doing
so that proceeded in complete indifference to ones public use
of ones words; so that there might be such a thing as a hidden
or internal sphere of meaning that was quite independent of
public language. Wittgensteins famous argument that this is
not so is known as the private language argument: in Section 4
we shall discuss both it and what it is embedded in: namely, a
complete reordering of the intuitive way of distinguishing ones
inner sensational life from ones outer bodily activities.
I said that the Tractatus idea, that certain truths can only be
shown, resurfaces in various particular sequences of PI. That is
true; but perhaps its most important legacy is rather the overall
10
effect that PI will have on an intelligent and attentive reader.
In a sense nothing has changed: one is inclined to go on acqui-
escing in ordinary commonsense ways of describing reality,
so long as these do not attempt to stray into philosophy. But in
another sense everything has changed: the whole world seems to
take on a different aspect.
But this feature is not something that I can attempt here to
analyse or explain; if you want to appreciate it you will have
to read PI itself. In any case it probably has as much to do with
the power of Wittgensteins writing as with the quality of his
arguments, and of course it is only the arguments that concern
me here.
11
CHAPTER 3
READING THE TEXT
SECTION 1. THE AUGUSTINIAN PICTURE
Wittgenstein begins with the following quotation from
St Augustines Confessions:
When they (my elders) named some object, and accordingly
moved towards something, I saw this and grasped that the
thing was called by the sound they uttered when they meant
to point it out. Their intention was shewn by their bodily
movements, as it were the natural language of all peoples: the
expression of the face, the play of the eyes, the movement of
other parts of the body, and the tone of voice which expresses
our state of mind in seeking, having, rejecting, or avoiding
something. Thus, as I heard words repeatedly used in their
proper places in various sentences, I gradually learnt to under-
stand what objects they signified; and after I had trained my
mouth to form these signs, I used them to express my own
desires. (PI 1)
He says that these words contain a particular picture of the
essence of language. It is this: the individual words of language
name objects sentences are combinations of such names (PI 1).
That picture is common to many philosophical theories of
language that differ over both the nature of the naming relation
and its objects. One such theory was Lockes; another appears in
the Tractatus. What all such theories have in common is that
they offer uniform accounts of the meanings of words and of the
sentences containing them. Where they differ is over (a) the kinds
of uniformity that they impose upon language and (b) the items
that they treat as the true names.
(a) If one says that the individual words of a language name
objects, what is one implying that they have in common?
Wittgenstein considers three candidates for this role: (i) that they
12
all have the same function; (ii) that they all get introduced in the
same way; or (iii) that they all mean just what they stand for.
(b) Some theories apply the uniform account to our natural
languages, for example, English: they treat their expressions as
having a uniform function or as being introduced in the same
way. For instance, Lockes theory of meaning is that ordinary
English words name ideas. Other theories concede that expres-
sions of natural language typically do not have the uniformity
postulated by the theory but insist that this unity may be revealed
by an analysis of such languages. For instance, Wittgenstein
had held in the Tractatus that analysis would reveal meaningful
sentences of English or German to be truth-functional syntheses
of atomic sentences, the latter being simply combinations of
names (TLP 4.221, 5).
In sections 164 of Philosophical Investigations Wittgenstein
attacks various combinations of the positions described under
(a) and (b). Initially he considers natural languages; but from PI 39
onwards he turns to the idea that one might defend some of the
theses under (a) by restricting them to analysed languages. We
start like Wittgenstein with the idea that all the words of a natural
language function in the same way.
Before doing so it is worth introducing one piece of terminology
that appears again and again in the course of the work. Many of
Wittgensteins examples involve very simple languages or linguistic
procedures, for example, languages consisting entirely of orders
(PI 2), of names for colours on a grid (PI 48), or the procedure
of inferring what a man is going to do from his expression of a
decision (PI 632). Wittgenstein calls these mini-languages
language-games. And he also uses the term language-game
for the language together with the actions that are typical
causes and effects
1
of its utterances (PI 7bd).
1.1. Unity of function
At PI 1d he asks us to imagine a man who takes to the grocers a
slip marked five red apples; the shopkeeper takes the slip and
opens the drawer marked apples; then he looks up the word
red in a table and finds a colour sample opposite it; then he
says the series of cardinal numbers I assume that he knows
them by heart up to the word five and for each number he
READING THE TEXT
13
takes an apple of the same colour as the sample out of the
drawer. It is in this and similar ways that one operates
with words.
What the example shows is how differently the shopkeeper
operates with the words five, red and apples. When he sees
the word apple he opens a drawer; but when he sees the word
red he looks at a colour sample, and so on. But then what hope
is there for a uniform account of the meanings of those words?
And now look at the much greater variety in our operations
with words. How likely is it that the words of any language
approaching the complexity of English could all get their mean-
ings through a single mechanism?
One of the ways in which Wittgenstein makes vivid the difference
between the actual functioning of language and that predicted by
some preconceived model of it is to imagine a language to which
the latter is adequate. That is what he does in PI 2 (and he later
describes it as the method of s2: PI 48). There he imagines that:
A is building with building-stones; there are blocks, pillars,
slabs and beams. B has to pass the stones, and that in the
order in which A needs them. For this purpose they use a
language consisting of the words block, pillar, slab,
beam. A calls them out; B brings the stone which he has
learnt to bring at such-and-such a call. Conceive this as a
complete primitive language.
It is easy to see why one might say of this primitive language
that all of its expressions were names, if one meant by this that
they all functioned in the same way. For their uses are uniform
(modulo the objects they denote): B does the same with the
referent of slab when he hears Slab! as he does with the referent
of beam when he hears Beam!, that is, he passes it to A. But
that reason for imputing uniformity to the language in PI 2 is
completely absent from PI 1d. There is no clear sense in which
the grocer does the same with the referent (whatever it is) of five
when he sees the word on the slip as he does with the referent of
red when he sees that word on the slip.
One might of course manufacture such a sense. One might
insist that the grocer is using five to signify the number five and
14
red to signify the colour red. But that is a merely verbal manoeu-
vre: saying that in each case the word signifies the object does
not reveal their similarities but conceals their differences.
Imagine someones saying: All tools serve to modify
something. Thus a hammer modifies the position of a nail,
a saw the shape of a board, and so on. And what is
modified by a rule, a glue-pot, and nails? Our knowledge
of a things length, the temperature of the glue, and the
solidity of a box. Would anything be gained by this
assimilation of expressions? (PI 14)
The point of the question is as obvious as its answer.
But then what tempted us into thinking that all words function
uniformly in the first place? Wittgenstein thinks that there is
a specific explanation: the uniform appearance of written and
spoken expressions that makes them seem to get their meanings
in the same way (PI 11b).
It is like looking into the cabin of a locomotive. We see
handles all looking more or less alike. (Naturally, since they
are all supposed to be handled.) But one is the handle of a
crank which can be moved continuously (it regulates the
opening of a valve); another is the handle of a switch, which
has only two effective positions, it is either off or on; a third is
the handle of a brake-lever, the harder one pulls on it, the
harder it brakes; a fourth, the handle of a pump: it has an
effect only so long as it is moved to and fro. (PI 12)
In this analogy words correspond to handles: they all look
similar. Thus the word red is written in the same alphabet as
the word five, it is composed in a linear script, and so on. And
the different uses of words, like five and red, correspond to the
different functions of the handles.
But he also has a general explanation: it arises from a tendency
that also drives scientific inquiry, namely, the urge to find a single
explanatory theory that unites disparate phenomena. Newtons
theory of gravitation is a spectacular example of what that tend-
ency can produce: it explains how quite disparate phenomena
(apples falling to the Earth and planets orbiting the Sun) arise
READING THE TEXT
15
from the operation of a single mechanism (the force of gravity).
In the present case the disparate phenomena are the many
varieties of linguistic expression; the explanatory theory is
any that tries to explain their meaning as arising through the
operation of a single mechanism (here the mechanism of refer-
ence). Wittgenstein thought that philosophy ought not to aspire
in this direction (PI 109): we shall discuss the reasons for this
in due course (2.3).
How convincing is Wittgensteins attack? A brief examination
of what they actually say will make it clear why his likely opponents
on this point are unlikely to find the example at PI 1d very con-
vincing. The most likely response will be that that example gets
only a superficial description in the passage in question. There
are inner activities which accompany or precede the grocers outer
ones, and when we look at the totality of his actions we can see
that what he does with five corresponds to what he does with
red in just the way that what B does with Slab! corresponds to
what he does with Beam!
Thus consider Lockes version of the thesis. On Lockes view the
primary purpose of words is to name ones ideas (Essay III.ii.8)
roughly, private mental images. In general these ideas govern ones
application of words to external objects by functioning as
templates with which to compare the latter (Essay IV.iii.2). If
a post-box matches in a certain way the private idea that one
associates with the word red then one should apply red to
the post-box too. If a fruit matches in a certain other way the
private idea that one associates with the word apple then one
should apply the word apple to that fruit too. Since Locke
applies the same account to number (Essay II.xvi), his theory
implies enough uniformity in the uses of five, red and apples
for there to be some point in ascribing them all the same
function. For in each case what the grocer does with the word
is to compare external objects with the associated idea. This
both precedes and explains his application of it to the external
object. Nothing in the passages that we have considered has
anything to say about any of this.
But Wittgenstein was certainly aware of the point. Indeed
it gets raised immediately after the example itself: he has the
interlocutor ask But how does he know where and how he is
to look up the word red and what he is to do with the word
16
five? one tempting answer to the first question being that he
finds the colour on the table that matches the idea he associates
with red. But he does not address it any further here. He returns
to the thought that comparison with ideas explains the external
application of our words at PI 73 (discussed at 2.2.3 (a) (iii)). He
attacks the thought that words might be made to stand for the
subjects of such comparisons at PI 258 (discussed at 4.1.2).
1.2. Ostension
Another uniformity that philosophers have supposed to exist in
ordinary language is that its words are learnt by ostension: the
person who is teaching you the word points to or otherwise
indicates the referent of the word while saying it (PI 6b). Such a
picture is naturally associated with the classical empiricist view
that every word stands for an object of immediate sensory (or
introspective) awareness. But it is also evident in the passage
from St Augustine. Wittgensteins view is that there it is the effect
of excessive concentration on a small class of expressions:
If you describe the learning of language in this way you are,
I believe, thinking primarily of nouns like table, chair,
loaf , and of peoples names, and only secondarily of the
names of certain actions and properties; and of the remaining
kinds of word as something that will take care of itself. (PI 1)
What do you point to on this theory in order to teach somebody
the meaning of, for example, if? Maybe there is an answer to
this question (cf. the if-feeling discussed at PI II, vi); but in any
case Wittgenstein appears to have a general argument against
the thesis that anyone might have acquired his entire linguistic
capability via ostension.
That argument appears at PI 2835. There, Wittgenstein
initially appears to be raising a doubt as to whether ostensive
definition is possible at all. Suppose that I am trying to teach
someone the word red. I point to a tall cylindrical post-box and
say: That is called red . He might be forgiven for thinking that
red means post-box, or tall, or cylindrical.
And he might equally well take the name of a person, of which
I give an ostensive definition, as that of a colour, of a race,
READING THE TEXT
17
or even of a point of the compass. That is to say: an ostensive
definition can be variously interpreted in every case. (PI 28)
Nor could I remove that possibility of misunderstanding by
saying: That colour is called: red , for the same possibility
of misunderstanding will arise when I try to give an ostensive
definition of colour (PI 29a). Neither would it help for me to
point to several objects (say, a post-box, a fire-engine and a
tomato) while saying All of these are called: red. For the pupil
might take red to be a predicate true of all and only post-boxes,
fire-engines and tomatoes.
But these examples do not show, and are not intended to show,
that there could never be a successful ostensive definition. For
the fact that a possibility of misunderstanding exists does not
mean that the definition was somehow defective in any case,
that is, including cases where that possibility was not realized.
The definition is successful as long as it does have the intended
effect regardless of whether or not it might have had another
(cf. PI 87bc). But they do show that for an ostensive definition
to be successful the pupil must already be prepared to take it in
the right way he must, for example, understand that I meant a
colour when I said That is called red (or, for example, that
I meant a property of a surface when I said These are all called
colours ).
Now what does this taking it in the right way consist in?
Wittgenstein writes:
Whether the word number [or colour] is necessary in an
ostensive definition depends on whether without it the other
person takes the definition otherwise than I wish . . . And how
he takes the definition is seen in the use that he makes of the
word defined. (PI 29)
Now the last sentence does not equate how you take the defini-
tion with how you go on to use the word so defined; it merely
says that we can see the former in the latter (cf. determined by
at PI 139b, discussed at 3.1; exhibited in at PI 201b, discussed at
3.4.3). But it is clear enough that in general nothing other than
ones consequent use of the word is relevant to how one takes
the definition, that is, what one takes the word so defined to
18
mean; and indeed Wittgenstein does make an equivalent equation
slightly later in the text in a passage that explicitly connects it
with ostensive definition:
For a large class of cases though not for all in which
we employ the word meaning it can be defined thus: the
meaning of a word is its use in the language.
And the meaning of a name is sometimes explained by
pointing to its bearer.
2
(PI 43)
So we take how you take the definition to mean no more or less
than: how you go on to use the word.
But this interpretation of taking it in the right way is open to
an objection. Surely the objector will say whether or not the
pupil takes the definition in the right way is at best of causal
relevance to whether or not he goes on to use it in this or that way.
Surely taking the definition in the right way is an experience
concurrent with the giving of the ostensive definition. Such an
experience might be, for instance, that of attending to the colour
or attending to the shape of the ostended object. So for the pupil
to take in the right way the ostensive definition That is red
is for him to attend to the colour (and not, say, the shape) of the
ostended object while the ostensive definition is going on, however
he then and perhaps consequently goes on to use the word.
Wittgenstein makes two replies to this objection.
(i) There is no one thing that one does in all cases of attending
to the colour. On the contrary there are many different activities
that we call attending to the colour. Consider various occasions
for dong so: here are Wittgensteins examples:
Is this blue the same as the blue over there? Do you see any
difference?
You are mixing paint and you say Its hard to get the blue
of this sky.
Its turning fine, and you can already see the blue sky
again.
Look what different effects these two blues have.
Do you see the blue book over there? Bring it here.
This blue signal-light means . . .
Whats this blue called? Is it indigo? (PI 33)
READING THE TEXT
19
Wittgenstein himself emphasizes the fact that one performs
different bodily actions on different such occasions screwing
up ones eyes, framing the colour with ones hand, and so
on but the examples establish a similar diversity among the
accompanying experiences.
But whether we are concerned with bodily actions or experi-
ences, it is not clear why this diversity matters particularly in
light of Wittgensteins doctrine of family resemblance (see 2.1).
For an objector might reply that there is one type of process
or experience that occurs in all of these cases: namely, that of
attending to the colour. The fact that so many diverse processes
are acts of attention to the colour should not stop us from saying
that they all have that in common, any more than the evident
disparities between board-games, card-games, ball-games and
Olympic games should stop us from saying that they all have
something in common, namely, that they are all games (cf.
PI 66). So why could not a whole range of experiences that
differed in diverse other ways not all be experiences of attending
to the colour? (Just as a whole range of experiences that differ
in diverse ways could still be visual experiences of a face. A
Martian might see nothing in common among them. Does that
make us wrong to do so?)
It would strengthen Wittgensteins case here to emphasize what
he says in another discussion of a related topic (PI 171b). For
the fact is that on many occasions of attending to the colour,
nothing that one feels could be a characteristic experience of it
indeed one often has no special experiences at all (other than
of the colour itself). When you were reading these last few
sentences you were presumably attending to the content of the
words and not to their typeface. But where were the experiences
of attending to the content?
If that is correct then the present objection to PI 33 is beside
the point: what distinguishes taking the definition [e.g. of a
colour-word] in the right way from taking it otherwise cannot
be something that we experience exactly in the first case, since
no experience is always present in that case. So the important
point against this idea is not the actual diversity, along some
dimensions, of the experiences we have when we attend to the
colour: that diversity is compatible with their all being experiences
of the postulated type. The important point is that one can take
20
the definition in the right way and feel nothing at all that even
could be such an experience.
(ii) In any event Wittgensteins really decisive point is at PI 34.
Suppose that some psychological state or process does obtain
or occur on all and only occasions of attending to the shape,
say. Cant the pupil misinterpret it? So that, for instance, he
attends to the shape (i.e. he has that experience) while I say This
[pointing] is called: a circle and now he treats circle as the
name of a colour.
But suppose someone said: I always do the same thing
when I attend to the shape: my eye follows the outline and
I feel . . .. And suppose this person to give someone else
the ostensive definition That is called a circle , pointing
to a circular object and having all these experiences
cannot his hearer still interpret the definition differently, even
though he sees the others eyes following the outline, and
even though he feels what the other feels? (PI 34)
This argument illustrates a pattern that appears again and again
in Philosophical Investigations. It is as follows: in cases where
we are tempted to think that a psychological expression, say
meaning a cube or pointing at the colour or expecting a
loud report, names a particular type of inner state or process,
just imagine that a process of that type gets accompanied by
or indeed produces behaviour that is characteristic of some
incompatible psychological state, say that of meaning a triangular
prism, or pointing at the shape, or expecting a quiet report.
When we think about these cases we will say that the supposed
inner state or process was, after all, quite inessential to the
psychological state of which it had seemed the very essence.
So if we reject this temptation, we see that the possibility of
their misunderstanding shows not that ostensive definitions are
impossible but only that they cannot operate in an intellectual
vacuum. They require a certain background capacity on the part
of the pupil. And that indispensable background is this: that the
pupil must be prepared to respond, for example, to the definition
of the word red by using the word so defined as we standardly
use colour-words (e.g. applying it to both or neither of two
chromatically indistinguishable objects in the same light).
READING THE TEXT
21
Wittgenstein summarizes this conclusion as follows:
And now, I think, we can say: Augustine describes the learning
of human language as if the child came into a strange country
and did not understand the language of the country; that is,
as if it already had a language, only not this one. Or again: as if
the child could already think, only not yet speak. And think
would here mean something like talk to itself . (PI 32)
If this really is the correct conclusion then Augustine was wrong
to place ostensive definition at the foundation of all language. In
order that an ostensive definition be successful the pupil must
already have a language of some sort; therefore no finite pupil
can have relied solely upon ostension.
But is it the correct conclusion? It is certainly true that in
order to take an ostensive definition in the right way the pupil
must have some capacities that he did not acquire by exposure to
ostensive definition. But why must those capacities be specifically
linguistic?
Here is a possibility. A certain kind of animal can mimic
sounds that its elders make; its drive to do so in a given situation
S depends upon the observable similarity already obtaining
between S and those other situations in which it has heard that
sound. Suppose that the sound is the same as the English word
red; suppose that on all and only the occasions O when a child
of this species has heard its elders make that sound, a red object
has been conspicuous to all parties. And suppose that in con-
sequence the child utters the sound Red when and only when a
red object is conspicuous to it. Hasnt the child learnt Red by
ostensive definition without already being able to do anything
like speak another language? Isnt he now able to do something
that has as much claim to count as a rudimentary report (on
colours in his environment) as have moves in the language-game
of PI 2 (conceived now as a complete language: PI 6a) to count
as rudimentary orders?
3
Of course the child would have to come to the learning situation
equipped with capacities that ostensive definition could not give it.
In particular it would have to come pre-equipped with a similarity
metric. That is, it would have to have been true of it all along that
given any situation S in which a red object was conspicuously
22
present and any situation T in which only a blue one was, it
classed S as more similar than T to the learning situation O
in which red got introduced to it. But there is no reason to
suppose that this is impossible. Indeed the fact that animals and
children as well as human adults make inductive inferences in
some directions and not others shows that some such metric is
widespread. For instance, somebody who burns his hand in the
fire will thereafter avoid fire and not, say, cats: this shows that in
some sense he treats subsequently encountered fires as more
similar than subsequently encountered cats to the initial fire.
Moreover it shows that such a metric is prelinguistic: one could
have it and animals do have it in the absence of any language
at all.
In short the objection to Wittgenstein is this: why cant the
similarity metric that we need anyway for inductive inference be
a non-linguistic basis for our taking ostensive definitions right?
4

And if it is, then couldnt there in principle be a language-game
all of whose terms were taught, and learnt, by this primitive
form of ostension?
Now it is true that plenty of expressions in our language
neither are nor could be learnt by ostension without some
further linguistic capacities. Theoretical terms like electric
charge, mass, inflation and even solidity are good examples:
no mere perceptual encounter with, for example, a solid body
could impart to me the physical theory in which the concept
of solidity finds a place (cf. Evans 1980: 26970). And it is
plausible that speaking a language is a necessary condition on
possessing that sort of theoretical background. But it is an
empirical fact that our language contains terms of this sort
and not only the sort of term that could be learnt by ostension
by someone who has no language; and the argument that terms
of the former sort cannot be learnt by ostension is quite different
from Wittgensteins. It therefore seems to me that Wittgenstein has
failed to establish the anti-Augustinian conclusion advertised at
PI 32 and quoted above.
5
1.3. Referentialism
A third kind of uniformity that all words are names might seem
to impose upon language is this: that all words refer to objects;
and the object exhausts the meaning of the word, so that two
READING THE TEXT
23
words that have the same reference ipso facto have the same
meaning. And if a word has no bearer then it has no meaning.
1.3.1. In ordinary language
Taken as a doctrine about ordinary language this is as clearly false
as any doctrine in the area. The first example that Wittgenstein
directs against it is as follows:
Let us first discuss this point of the argument: that a word has
no meaning if nothing corresponds to it. It is important to
note that the word meaning is being used illicitly if it is used
to signify the thing that corresponds to the word. That is
to confound the meaning of the name with the bearer of the
name. When Mr N. N. dies one says that the bearer of the
name dies, not that the meaning dies. And it would be non-
sensical to say that, for if the name ceased to have meaning it
would make no sense to say Mr N. N. is dead. (PI 40)
As it stands the argument might seem too hasty: it assumes that
the relation of reference unlike, say, the relations of precedence
or causation can only hold between a name and something that
exists contemporaneously with it. For if we drop that assumption
we can say that N. N. continues to refer to N. N. even after the
latter has died, and so somebody who thinks that the meaning of
N. N. is the object that it denotes can maintain that Mr N. N. is
dead remains meaningful after becoming true.
A further example that gets around this objection exploits a
refinement of the primitive builders language-game as it appeared
in PI 2. We are to imagine that the tools A uses in building bear
certain marks. When A shews his assistant such a mark, he brings
the tool that has that mark on it (PI 15). These marks may be
regarded as the names of the objects that bear them.
But has for instance a name which has never been used for a
tool also got a meaning in that game? Let us assume that
X is such a sign and that A gives this sign to B well, even
such signs could be given a place in the language-game, and
B might have, say, to answer them too with a shake of the
head. (One could imagine this as a sort of joke between
them.) (PI 42)
24
Even the name X has some sort of meaning; but not only is
there not anything to be its bearer (as is true also of Mr N. N.):
there never was. And it is very easy to imagine names and other
expressions of ordinary language that clearly have as much use
as X, and which equally clearly have no bearers (e.g. names
from fiction like Sherlock Holmes, or names like Atlantis and
Vulcan that are associated with conditions that turned out to be
unsatisfied). It is therefore false that bearerless expressions of
ordinary language are meaningless.
1.3.2. In the Tractatus
But another application of this doctrine is not to ordinary
language but to what in the Tractatus Wittgenstein had called
its analysis. Wittgenstein then believed that thoughts made a
determinate claim about reality: this means that if I have a
thought, say that Napoleon conquered Russia in 1812, every
possible situation makes that thought either true or false. But
how can this be? After all Napoleon might never have lived (if,
for example, his parents had never met). So how, in a possible
situation where he never did live, could my thought be so much
as a thought let alone true or false if there is nothing there
(i.e. no Napoleon) for it to be about?
In the Tractatus his answer had been that the universe consists
of objects (TLP 2.021). The possible configurations of these
objects exhaust all possibilities: every possibility for our world is
therefore one where all actual objects exist (TLP 2.0222.023,
2.0271). Those objects are therefore indestructible, that is, neces-
sarily existent; since all that is composite might be decomposed,
it follows that the objects are simple (TLP 2.02). A thought
describes a possibility by describing some possible configuration
of simple objects: these possible configurations are states of
affairs (TLP 2.0272). A thought describes a state of affairs in an
essentially pictorial manner (TLP 2.1, 3). It consists of names
for the objects that figure in that state of affairs (TLP 3.202);
the name simply means what it refers to (TLP 3.203); and the
configuration of names depicts the possible configuration of the
objects named (TLP 3.21).
Our initial question was how could my (false) thought, that
Napoleon conquered Russia in 1812, be so much as a thought
about possible situations in which Napoleon never existed in
READING THE TEXT
25
which, therefore, there is apparently nothing for it to be about?
And the answer is that the thought is really not about Napoleon;
or if you like that it is about him, but only by virtue of being
about those simple and indestructible objects of which he is
composed. Those objects exist in all possible situations; it is only
that in ones where they have certain arrangements Napoleon
himself does not exist. So there is always something for my
thought to be true about; that is why it can be true or false of
every possible situation.
In order to apply this account to everyday languages Wittgenstein
proposed that those languages in fact conceal the structure of
the thoughts that we use them to express (TLP 4.002); what they
present as simple (Napoleon) are really objects of great com-
plexity. It is only through the analysis of ordinary language that
we may reveal the perfectly precise content of those thoughts; so
that the elements of a fully analysed sentence correspond to
those of a thought (TLP 3.23.201).
For instance, sentences of ordinary language contain simple
elements that appear to name entities that are complex and hence
destructible. For example, Cleopatras Needle appears to name
a large and complex object that might one day cease to exist and
in any event might never have existed. (Note that Cleopatras
Needle is not a description but a name: what it denotes is not
a needle and was never Cleopatras.) So a sentence such as
Cleopatras Needle is made from granite does not mirror the
structure of the thoughts it expresses. Full analysis of it will
reveal a simple sentence, or a complex combination of simple
sentences, in which names of simple and hence indestructible
objects are combined in a way that depicts a possible combination
of those objects.
To pursue this example for merely illustrative purposes:
Cleopatras needle is made from granite might get rendered
as N
1
N
2
N
3
. . . N
k
G. Here N
1
. . . N
k
names the atoms that
compose Cleopatras needle; G names the relation that atoms
stand in when they compose something granite; and the concat-
enation of names of atoms on the left with that of a universal on
the right is a way of saying that the universal applies to the atoms
(cf. TLP 3.1432).
This example is more definite than the Tractatus ever was
about the kind of sentence that analysis ultimately reveals.
26
But independently of that, it does illustrate the crucial point that
what appear to be names in ordinary language get eliminated
in analysis: Cleopatras Needle has vanished altogether from
N
1
N
2
N
3
. . . N
k
G. This gave Wittgenstein room in the earlier
work to maintain that all the elements of analysed sentences
are expressions that mean exactly what they denote (what he
there called names) while conceding that this was not so for
everyday, or unanalysed language. For only the latter contains
expressions like Cleopatras Needle or Mr N. N., that is, ones
that mean something whether or not their referents exist.
Whereas the elements of a fully analysed language, like N
3
or
G, refer only to what could not have failed to exist.
We shall focus upon two points that Wittgenstein now
makes against his earlier position. First, he questions the very
intelligibility of the idea that the world is made up of absolutely
simple objects; second, he denies that the analysed version of an
ordinary-language sentence expresses the same meaning more
perspicuously. We shall consider these in turn.
1.3.3. Simplicity
Wittgenstein says that it makes no sense to speak absolutely of
simplicity and compositeness:
What are the simple constituent parts of a chair? The bits
of wood from which it is made? Or the molecules, or the
atoms? Simple means: not composite. And here the point
is: in what sense composite? It makes no sense at all to speak
absolutely of the simple parts of a chair . . . But isnt a chess-
board, for instance, obviously, and absolutely, composite?
You are probably thinking of the composition out of
thirty-two white and thirty-two black squares. But could we
not also say, for instance, that it was composed of the colours
black and white and the schema of squares? And if there are
quite different ways of looking at it, do you still want to say
that the chessboard is absolutely composite? (PI 47a, d)
Now these examples show only that we lack a clear conception of
what components a complex object has. This is not to show that
we have no absolute concept of simplicity or compositeness at all.
The Wittgenstein of the Tractatus might reply that the existence
READING THE TEXT
27
of more than one possible decomposition of the chessboard
shows only that we can take various routes to the final analysis,
just as somebody who thought that words were ultimately com-
posites of letters could still admit that there is nothing to choose
between saying (a) that logic is composed of lo and gic and
(b) that it is composed of log and ic.
But it is appropriate for Wittgenstein to complain against
his earlier self that the Tractatus does nothing to explain an
absolute notion of simplicity to somebody who does not already
possess one. Now you might propose that we treat the Tractarian
motivation for simplicity as definitive of it. More precisely: the
reason for thinking that there are simples in the first place is
that the universe must be composed of indestructible, that is,
necessarily existent items. So could we perhaps take simple in
the Tractatus system to mean indestructible?
Doing so would make it clear enough what simple means;
but it would still leave it quite unclear whether or not anything is
simple. It is hard to see how a particular could be indestructible:
it appears straightforward enough to imagine the non-existence
of any particular and so there must be a presumption against its
necessary existence. What about universals?
Something red can be destroyed [says the interlocutor], but
red cannot be destroyed, and that is why the meaning of the
word red is independent of the existence of a red thing.
Certainly it makes no sense to say that the colour red is torn up
or pounded to bits. But dont we say The red is vanishing?
And dont clutch at the idea of our always being able to bring
red before our minds eye even when there is nothing red any
more. That is just as if you choose to say that there would still
always be a chemical reaction producing a red flame. For
suppose you cannot remember the colour any more? (PI 57)
It is not clear what it means to say that red exists if it doesnt
mean that something red exists. In what sense, then, can universals
be called indestructible?
But then what was wrong with the train of thought that took
us to the allied notions of simplicity and indestructibility in the
first place? Recall what it was: in order for our thoughts to be
true or false in all possibilities, there must be something in each
28
of those possibilities for those thoughts to be about. Wittgenstein
discusses this line of thought in the following passage:
What the names in language signify [says the interlocutor]
must be indestructible; for it must be possible to describe the
state of affairs in which everything destructible is destroyed.
And this description will contain words; and what corresponds
to these cannot then be destroyed, for otherwise the words
would have no meaning. I must not saw off the branch on
which I am sitting.
One might, of course, object at once that this description
would have to except itself from the destruction. (PI 55)
This immediate objection encapsulates a serious problem with
the original train of thought. It was being supposed that in order
for an actual thought or description meaningfully to say some-
thing about a counterfactual situation, there must be something
in the counterfactual situation to make the thought or descrip-
tion meaningful. But that is simply false. When we assess a
description with respect to a non-actual situation we require
only that the description is meaningful; we do not require that
the description would have been meaningful if that non-actual
situation had obtained, as follows from the fact (to which
Wittgenstein is ironically alluding) that we can certainly
describe possible situations in which the description would
not even have existed.
There is another way of being tempted into thinking that the
preconditions of language are in some sense necessary. Suppose
that it is necessary for certain coloured things to exist in order
that certain colour-words have a meaning. Suppose in particular
that the word sepia depends for its meaning upon the existence
of a certain unreproducible standard sample of that colour,
with which we compare candidates for that predicate; this may
be because to the unaided eye sepia objects are chromatically
indistinguishable from objects of other shades. Then in one
sense the sentence There is a sepia sample cannot be false: for if
it were false, then the expression sepia occurring within it would
have no meaning; and so that sentence would be not false after
all but meaningless. And similarly one might think that the
objects that our language depends upon for its meaning are
READING THE TEXT
29
indestructible, simply because a sentence asserting the non-
existence of any of them would be meaningless if true.
But of course to say that There is a sepia sample is true
if meaningful is not to say that the sepia sample itself exists
necessarily. It is only to say something about the presuppositions
of the language in which it is stated:
We can put it like this: This sample is an instrument of the
language used in ascriptions of colour. In this language-game
it is not something that is represented, but is a means of
representation . . . What looks as if it had to exist, is part
of the language. It is a paradigm in our language-game;
something with which comparison is made. And this may
be an important observation; but it is none the less an
observation concerning our language-game our method
of representation. (PI 50)
Much the same could be said about the objects of the
Tractatus.
1.3.4. Analysis
At PI 604 Wittgenstein argues as follows: first that it is unclear
whether the analysis envisaged in the Tractatus really yields
something with the same meaning as what it operates upon; and
second that in any case it does not reveal that meaning more
perspicuously.
In connection with the first point he starts by simply denying
that analysis reveals more clearly the thought that you had all
along and which you originally expressed by the unanalysed
sentence.
When I say: My broom is in the corner, is this really a
statement about the broomstick and the brush? Well, it could
at any rate be replaced by a statement giving the position of
the stick and the position of the brush . . . Then does someone
who says that the broom is in the corner really mean: the
broomstick is there, and so is the brush, and the broomstick
is fixed in the brush? If we were to ask anyone if he meant
this he would probably say that he had not thought specially
of the broomstick or specially of the brush at all. And that
30
would be the right answer, for he meant to speak neither of
the stick nor of the brush in particular. (PI 60)
As it stands this answer invites the reply that even if you didnt
think specially of the brush or of the broomstick, still they were
in some sense latent in what you meant all along. Compare: if
I give somebody the order to count to 1,000, it is likely that
I wont think specially of the step from 625 to 626; and yet if he
goes straight from 625 to 627 it would be right for me to say
that he had deviated from what I had all along meant him to do
(cf. PI 187). We shall return to this point (see 3.3).
In any case Wittgenstein continues with a discussion of the
order Bring me the broom. He introduces letters (a) and (b) to
denote language-games in which (a) whole instruments for
example, brooms or (b) their parts for example, brushes are
given names for use in orders. Thus (a) might contain the order
Bring me the broom; (b) might contain the order Bring me
the broomstick and the brush which is fitted onto it. Now,
must the analysed order in (b) have the same meaning as the
unanalysed order in (a)? Wittgenstein invites us to imagine
a case where it is not so clear:
Suppose for instance that the person who is given the orders
in (a) and (b) has to look up a table co-ordinating names and
pictures before bringing what is required. Does he do the same
when he carries out an order in (a) and the corresponding one
in (b)? Yes and no. (PI 62)
Presumably yes because he brings the same object in both cases;
presumably no because this action has a different precedent in
each case.
The relevance of this point whether he does the same to the
one at issue whether the orders mean the same relies upon an
extension of the equation at PI 43 between meaning and use
(discussed at 1.2). There he had said that in a large range of cases
the meaning of a word is its use in the language; he now implic-
itly supposes that the meaning of a sentence too is its use in
the language, where the use of an order covers things like: what
somebody does when he obeys it.
Now a defender of the Tractatus neednt grant this assump-
tion or at any rate its applicability to the question of analysis.
READING THE TEXT
31
He might maintain that orders, questions and descriptions, were
never the intended subjects of Tractarian analysis. The intended
subjects of analysis are instead the senses that are common to all
of them. For instance, if we take the description Caesar crossed
the Rubicon. and the question Did Caesar cross the Rubicon?,
it is plausible that they each have something in common as well
as something not in common. What they have in common is
their sense: the assumption (PI 22) or truth-condition, Caesars
crossing the Rubicon; and what they do not have in common is
that the question is asking whether the truth-condition obtains
whereas the description is saying that it does. By similar reason-
ing we could make a case that an order shares something with
both a question and a description: Shut the door! shares with
Is the door shut? and The door is shut. what might be called
the obedience-condition of the first sentence, the affirmation-
condition of the second and the truth-condition of the third.
And the defender of the Tractatus will continue what con-
cerned me all along was how thoughts could be about possible
situations in which there seemed to be nothing for them to be
about. And this question arises for questions and orders as much
as for descriptions. But if we say that the thought that Napoleon
conquered Russia in 1812 is really about indestructible objects,
then both the question Did Napoleon conquer Russia in 1812?
and the description Napoleon conquered Russia in 1812. must
be about the same indestructible objects, because one of them
is asking just what the other is asserting. So what I was really
trying to analyse all along was the sense of our thoughts.
And it follows he will conclude that you cannot criticize
me on the grounds that an analysed order has a different use
from its unanalysed version. For this does not show that the
analysed sense is different from the unanalysed one. The only
thing that could show a difference in the sense of the two
analysed and unanalysed orders would be a difference in the
obedience-conditions of the order in (a) and the order in (b).
But as we can see, these are perfectly alike: to bring someone
the broom is to bring him the broomstick fixed in the brush.
There is a brief and unsatisfactory response to this at PI 22,
where Wittgenstein rejects the idea that every assertion contains
an assumption, which is the thing that is asserted. He objects that
the idea arises from the possibility of separating the assumption
32
from the rest of the assertion, so that instead of Caesar crossed
the Rubicon one might write It is asserted: that Caesar crossed
the Rubicon, where that Caesar crossed the Rubicon is the
assumption or sense of the assertion. But, he continues, if this
possibility of rewriting is the basis of that idea it is a bad one:
We might very well also write every statement in the form of
a question followed by a Yes; for instance: Is it raining?
Yes! Would this shew that every statement contained a
question?
Of course the answer is no.
But then this possibility of rewriting need not have been the
motivation for the idea in the first place. Isnt it rather that
the separation of the sense of a sentence from the other factors
that generate its use is an obvious prerequisite for obtaining a
systematic explanation of that use? It is part of my knowing how
to speak and write English that I know how to use both the
description Caesar crossed the Rubicon and the question Did
Caesar cross the Rubicon? Isnt it likely or at least a good work-
ing supposition that the best explanation of these two abilities
will appeal to something that they have in common namely, my
knowledge of their sense? As we shall soon discuss Wittgenstein
didnt think that philosophy should try to explain anything
(see 2.3); but apart from this quite generally applicable stance
it is hard to see what motivates his opposition to the Tractatus
on this fundamental point.
The other line of objection to analysis appears at PI 634,
where he raises the question of whether something essential to
the thought might not be lost by the very process of isolating its
constituents (We murder to dissect). He gives the example of
people who have names not for coloured shapes (as at PI 48)
but for certain combinations of them, so that, for instance, a
rectangle which is half red and half green (i.e. a red square to
the left of a green one) is called U and one that is half green,
half white is called V. And he compares this language-game
with another (i.e. that at PI 48) into which, for example, U
would get translated as RG, where R denotes red squares and
G denotes green ones.
READING THE TEXT
33
Presumably what has been lost in translation is the unity for its
speakers of the experience that their language described using the
single expressions U and V. For there is a difference between
the bare experience of seeing a red square adjacent to a green
square and the gestalt experience of seeing a red-and-green
rectangle. And the proposed translation of U as RG effaces
this difference. We have similar experiences too: Think of the
cases where we say This arrangement of colours (say the French
tricolor) has a quite special character (PI 64). Or again, when
we hear music we often hear a chord as a unity of different notes:
it is not at all the same thing as merely hearing one thing at the
same time as another, as happens when, for example, one hears
somebodys mobile phone go off in the middle of a concert. But
why could not the analysis also capture this unity? For instance,
why could one not analyse the gestalt experience recorded by U
as, say, R*G? Here the sign * is a conventional indication of
the unity of what we may still regard as the phenomenological
elements of the experience.
In the face of that objection another of Wittgensteins
examples would seem more to the point:
I could imagine someone saying that he saw a red and yellow
star, but did not see anything yellow because he sees a star
as, so to speak, a conjunction of coloured parts which he
cannot separate. (RFM VII-65)
So an analysis of a red-and-yellow star experience (see the
illustration in RFM) might get rendered in his language as TS,
where S denotes a star-shape and T denotes the composite
red-and-yellow chromatic experience. It would now seem to
falsify his experience to translate its description as RYS or even
R*YS for that makes it seem as though yellow was already
part of the experience; but the fact that he might notice the
yellow on closer inspection clearly neednt mean that it was part
of his experience all along (cf. PI 171b). But the author of the
Tractatus might reply that that only shows that analysis can reveal
unnoticed features of the thought and we have already seen
(in connection with the broom example) how he might further
defend that.
34
Finally, it is worth mentioning that in his pre-Tractatus Note-
books Wittgenstein toyed with a conception of analysis that gave
much greater weight to the concerns raised at PI 634 than is
evident from the finished work. He wrote on 18 June 1915:
It seems to me perfectly possible that patches in our visual
field are simple objects, in that we do not perceive any single
point of a patch separately; the visual appearances of stars
even seem certainly to be so. What I mean is: if, e.g., I say that
this watch is not in the drawer, there is absolutely no need for
it to FOLLOW LOGICALLY that a wheel that is in the watch
is not in the drawer, for perhaps I had not the least knowledge
that the wheel was in the watch, and hence could not have
meant by this watch the complex in which the wheel occurs.
And it is certain moreover that I do not see all the parts of
my theoretical visual field. Who knows whether I see infinitely
many points? (NB 645)
The entries from around this period offer a quite different vision
of the simple objects lying at the limits of analysis than what
you would gather from the Tractatus itself. According to the for-
mer, what distinguishes terms for simple from terms for complex
objects is the role of those terms within the language, not the
metaphysical complexity or compositeness of what they denote,
just as PI 50 proposes. So that if, for example, the complexity of
Socrates is irrelevant to our use of Socrates in the language (e.g.
in inferences involving Socrates), the term Socrates may be
said to denote a simple in that language; and if we replaced it
with an analysis say, The composite object consisting of
Socratess head stuck on to Socratess body we should no lon-
ger be speaking the same language, just as PI 64b proposes. It is
striking that Wittgensteins criticisms of the Tractatus concep-
tions of simplicity and analysis appear to be recommending, or
at least to be allowing for, a reversion to ones that he had enter-
tained so far beforehand.
6
1.4. Sentences are combinations of names
The final element of the Augustinian picture that I discuss here
is the idea that sentences are combinations of other expressions.
It is not explicit in the quotation from the Confessions at PI 1 but
READING THE TEXT
35
it is present in Wittgensteins gloss upon it:
These words, it seems to me, give us a particular picture of the
essence of human language. It is this: the individual words in
language name objects sentences are combinations of such
names. (PI 1)
We have already seen reason to doubt that all words are names in
any of three senses; so we have reason to doubt that all sentences
are combinations of names. But can we say that in general they
are combinations at all, that is, of any sorts of word? As with
referentialism we shall consider first the application of that idea
to everyday language, and then to the thoughts behind it, of
which Tractarian analysis was supposed to reveal the structure.
It is clear enough that many everyday sentences are not
combinations of expressions at all but rather consist of a single
word. This is indeed quite obviously true of all sentences of the
imaginary language at PI 2 (here conceived as a complete
language): its only sentences are Beam!, Slab!, and the like.
And while plenty of English sentences are combinations, plenty
of them are not. Think of exclamations alone . . . Water! Away!
Ow! Help! Fine! No! (PI 27).
But then what about the thoughts that these sentences express
are they essentially composite? In the Tractatus he had thought
so: according to it a thought is a proposition with a sense (TLP 4);
but a proposition can only have a sense by picturing a possible
situation (TLP 3), and:
It is only in so far as a proposition is logically segmented that
it is a picture of a situation.
(Even the proposition, Ambulo, is composite: for its stem
with a different ending yields a different sense, and so does its
ending with a different stem.) (TLP 4.032)
We can take two further points away from this section. What
we can gather from his comment on Ambulo is that what it is
for a proposition to be composite, to be a combination of other
expressions, is for it to have something of semantic significance
in common with other propositions indeed Wittgenstein goes
so far as to present each component of a proposition as the set
36
of propositions that contain it (TLP 3.313.318; cf. TLP 5.5261b).
And what we can take away from TLP 4.032a in conjunction
with the picture theory is that this kind of composition is a
necessary condition on meaningfulness.
But then what about the language of PI 2? None of its
sentences appear to have anything of semantic significance in
common with any other sentences of it, for example, Beam!
and Slab! have no such thing in common.
7
From the perspective
of the Tractatus this would have to be one of those cases in
which the structure of everyday language conceals the structure
of the thoughts that it is supposed to be expressing (TLP 4.002d).
According to it the sentences Beam! and Slab! would have to
be expressing thoughts that did have a common element, for
example, I want you to bring me a beam; and I want you to bring
me a slab.
Wittgensteins discussion of these matters at PI 1921 covers
both points. On the first point he is in partial agreement with
his earlier self. He agrees that meaning a sentence as composite
or simple is not so to speak a local property of the utterance
but rather to do with the language in which it belongs. You
cannot tell whether, for example, the utterance Bring me a slab
is meant as one long word or as four short ones by attending
solely to it; instead you must look at the other expressive resources
of English.
I think we shall be inclined to say: we mean the sentence as
four words when we use it in contrast with other sentences such
as Hand me a slab, Bring him a slab, Bring two slabs,
etc.; that is, in contrast with sentences containing the separate
words of our command in other combinations. (PI 20)
But note that this agreement is not over the compositeness or
otherwise of its meaning but only over whether the sentence Bring
me a slab is meant as composite or simple or as a single word.
What Wittgenstein says here could equally be applied as a criterion
for whether or not, for example, an intrinsically meaningless
pattern of sound in a composition is a chord or a single note we
may settle it by asking whether its elements recur in other com-
binations in other music of its genre. He is not giving a criterion
for whether the thought behind the sentence Bring me a slab, or
READING THE TEXT
37
that behind the sentence Slab! in the language of PI 2, is itself
composite or simple. So he is not here agreeing with the Tractatus
line that these thoughts must themselves be composite.
It is worth noting that in the course of expounding this view of
the compositeness of sentences he attacks a faulty understanding
of it, one of which the Tractatus is entirely innocent but which is
still of interest because of its analogy with other views that do
appear in the earlier work. The view that I have in mind is the
idea that using the (composite) sentence in contrast with others
is a matter of accompanying ones overt utterance with some
unobservable mental process. Wittgenstein effectively criticizes
that idea by trying to get more definite about it; for as soon
as one tries to be specific about what this accompaniment is
supposed to be one finds oneself describing something that
plainly does not happen:
But what does using one sentence in contrast with others
consist in? Do the others, perhaps, hover before ones mind? All
of them? And while one is saying the one sentence, or before, or
afterwards? No. Even if such an explanation rather tempts us,
we need only think for a moment of what actually happens in
order to see that we are going astray here. (PI 20)
To say that you mean a sentence as composite is not to describe
an inner process that accompanies the utterance; rather it is a
comparison of that sentence with others that occur in the language
that you then used. The mistake that Wittgenstein here criticizes
does not occur in the Tractatus; but other variations upon its ground
bass do occur there. That ground bass is the idea that meaning,
understanding, intending, attending and thinking, and so on are
inner processes that go on alongside their verbal expressions.
Wittgensteins criticisms of this ideas various realizations runs
through Philosophical Investigations like a scarlet thread.
And one realization of it that does occur in the Tractatus is that
way of understanding 4.032a that makes it compatible with the
language of the builders at PI 2. Remember that speakers of that
language cannot have meant the sentence Slab! as composite,
at least not by the criterion that we have just seen proposed at
PI 20; for it has nothing in common with Beam! or any other
sentence in the language. But according to the Tractatus builder
38
A nonetheless means something complex when she says Slab!
That is to say, perspicuous representations of what she meant by
Slab! and Beam! would between them show a common element,
for example, what I want you to bring me a slab shares with
I want you to bring me a beam.
But how do you do this: how do you mean that while you say
Slab!? Do you say the unshortened sentence to yourself ?
And why should I translate the call Slab! into a different
expression in order to say what someone means by it? And if
they mean the same thing why should I not say: When he
says Slab! he means Slab! ? . . . But when I call Slab!,
then what I want is, that he should bring me a slab! Certainly,
but does wanting this consist in thinking in some form or
other a different sentence from the one you utter? (PI 19)
The error is different from the one that PI 20 attacks but the
form of the criticism is the same: if we look at what actually
goes on, we are not aware of any composite act of meaning
that accompanies the utterance of a one-word sentence. Why
say then that you really mean something composite?
Well, one reason might be a desire to explain what the
Tractatus emphasizes at 4.024.027: the fact that a proposition
can communicate something new. More specifically, how can
we explain the fact that the capacity to understand certain
propositions brings in its wake the capacity to understand a
whole range of others, unless we suppose that understanding
a proposition means grasping something composite and hence
also grasping what arrangements of its elements can express?
We can illustrate this point with a slight elaboration upon the
builders language of PI 2. Suppose that in this new language
A calls out Slab! (or Beam! or . . . ) in a low-pitched voice
when she wants B to bring her a slab; and she calls out Slab! in
a high-pitched voice when she wants B to bring her anything
but a slab. Let us write Slab + for the first, low-pitched, type
of utterance, and Slab for the second, high-pitched, type of
utterance. Now in accordance with the way that Philosophical
Investigations presents it, we may say that for the builders
assist ant to understand the order is for her to respond to it in
READING THE TEXT
39
the right way. So it is logically possible for somebody to under-
stand, say, Slab but not Slab +, or to understand both such
orders but not to understand either Beam or Beam +. And
such understanding could be imparted by explicit training in the
use of (i.e. in the correct response to) an expression, or it could
be left to the builders assistant to pick it up by watching others.
Now suppose we discover that whenever a builders assistant
has achieved by any means an understanding of the orders
Beam + and Slab , she has also and without any further
instruction come to grasp the orders Beam and Slab +. This
phenomenon would cry out for explanation: it is a very simple
instance of the fact that we can understand the sense of a
propositional sign without its sense having been explained to us
(TLP 4.02). And isnt the most obvious explanation just this:
that the builders assistant has a separate grasp of (a) the words
Beam and Slab and (b) their pitch; and it is through a recomb-
ination of these principles that her grasp of both Beam + and
Slab gives rise to a grasp of Beam and Slab +?
In that case we should have a reason for appealing to a certain
complexity in her thoughts to explain certain regularities in her
behaviour. Such complexity need not be that of any conscious or
introspectible process of the sort that PI 19 and PI 20 attack;
rather, the notion of thought that we need is a theoretical one
that we accept only because of the behavioural data that it
explains. I am not saying that that is how it is with human beings;
but I am saying that it is an empirical question whether it is so,
and that there is at least prima facie reason to think that it is so
for languages that go beyond that of PI 2 to the extent and in the
direction that my elaboration of it has indicated.
Let me conclude this section by remarking upon a theme that
connects it with the ones on ostension and functional unity: the
idea that certain mental or intentional occurrences are or involve
inner processes accompanying their visible or audible expres-
sion. So far we have seen four phenomena to which it has initially
attractive application: (a) the meaningful use of linguistic expres-
sions (1.1); (b) attending to the colour of something (rather than,
for example, its shape) (1.2); (c) meaning a sentence as compos-
ite rather than simple (1.4); (d) meaning something composite
when uttering a simple sentence (1.4).Wittgenstein diagnoses
40
one source of this attraction in the following summing-up of his
discussion of ostension:
And we do here what we do in a host of similar cases: because
we cannot specify any one bodily action which we call point-
ing to the shape (as opposed, for example, to the colour), we
say that a spiritual [geistige: mental, intellectual] activity cor-
responds to these words.
Where our language suggests a body and there is none:
there, we should like to say, is a spirit. (PI 36)
A proper understanding of this remark requires some understand-
ing of his doctrine of family resemblance to which we now turn.
SECTION 2. FAMILY RESEMBLANCE AND THE IDEAL
OF PRECISION
2.1. Family resemblance
Recall our discussion of PI 604 (1.3.4). There Wittgenstein had
tried to raise doubts about the identity of two pairs of language-
games, or two ways of playing each of two language-games. The
first pair consisted of (a) an unanalysed and (b) an analysed
language of orders in which one had to obey an order by cor-
relating names with pictures (PI 62). The second pair consisted
of (c) an analysed language-game of the sort described at PI 48,
that is, one in which we concatenate letters such as R and G to
describe the coloured squares that we see before us; and (d) an
unanalysed language-game in which atomic expressions like
U and V are used to characterize apparently composite pairs
of squares, for example, U signifies a red square adjacent to a
green one. The intuitive or at least Tractarian idea was that the
very same meanings that get expressed in (a) and (d) also get
more perspicuous expression in (b) and (c). And we have seen
Wittgensteins reasons in PI for wanting to deny this.
We have also glanced at his reason in the Tractatus for wishing
quite firmly to maintain it. That reason was basically a certain
conception of the essence of language (i.e. its propositions). The
essence of a proposition is that it describes how things stand.
It now seems possible to give the most general propositional
form: that is, to give a description of the propositions of any
sign-language whatsoever in such a way that every possible
READING THE TEXT
41
sense can be expressed by a symbol satisfying the description,
and every symbol satisfying the description can express a sense,
provided that the meanings of the names are suitably chosen.
It is clear that only what is essential to the most general
propositional form may be included in its description for
otherwise it would not be the most general form . . .
The general form of a proposition is: This is how things
stand. (TLP 4.5)
Now to say that things stand in a certain way is to pick out a
possible way in which things can stand; so two propositions
say the same thing if they pick out the same possible ways in
which things can stand. That is: we can identify the meanings of
propositions p and q just in case they would be true in just the
same circumstances and false in just the same circumstances
(this follows from TLP 4.4). And we can extend the account
straightforwardly to orders as already sketched (1.3.4): orders
O1 and O2 express the same sense just in case they would be
obeyed in the same circumstances and disobeyed in the same
circumstances. And from this account of the essence of language
there follows a straightforward verdict on the cases (a)(d).
An order in (a) has the same meaning as the corresponding one
in (b) because they have the same obedience-conditions; and the
description U in (d) has the same meaning as RG in (c) because
they have the same truth-conditions.
With this in mind we can ask whether Philosophical Investigations
offers any rival account of the essence of language that justifies its
alternative verdict on cases (a)(d). Wittgenstein puts the demand
for such an account into the mouth of an imaginary accuser:
You take the easy way out! You talk about all sorts of language-
games, but have nowhere said what the essence of a language-
game, and hence of language, is: what is common to all these
activities, and what makes them into language or parts of lan-
guage. So you let yourself off the very part of the investigation
that once gave you yourself most headache, the part about the
general form of propositions and of language. (PI 65)
At PI 65b Wittgenstein admits the justice of this comment, and
this opens a sequence in which he elaborates upon why, contrary
to what he thought in the Tractatus, no such account is possible.
42
For it seems as though an account of the necessary and suffi-
cient conditions of language must be possible; otherwise, what
explanation is there for our applying of the word language in
just the cases where we do apply it? This question, which hovers
in the background of the discussion around PI 65, plays upon a
very important theme. It is the idea that to understand an expres-
sion is to be in some mental state, or to have some mental item
within ones purview, that guides ones use of that expression.
Wittgenstein is going to argue from a number of directions that
things are more complicated than that. For first, there need not
be any such inner guide to ones use of an expression; and sec-
ond, even in cases where there is, there are still many different
ways in which it can be said to guide ones use. I shall return to
this second point (3.1.1) but let us now consider an instance of
the first that has immediate bearing.
What Wittgenstein argues is that it need not be the case that
we apply a word in all cases because of one necessary and suffi-
cient condition that those cases have in common. To establish
this he considers the word game:
Consider for example the proceedings that we call games. I
mean board-games, card-games, ball-games, Olympic games,
and so on. What is common to them all? Dont say: There
must be something common, or they would not be called
games but look and see whether there is anything
common to all. For if you look at them you will not see
something that is common to all, but similarities, relation-
ships, and a whole series of them at that. To repeat: dont
think, but look! Look for example at board-games, with
their multifarious relationships. Now pass to card-games;
here you find many correspondences with the first group, but
many common features drop out, and others appear. When
we pass next to ball-games much that is common is retained,
but much is lost. Are they all amusing? Compare chess
with noughts and crosses. (Etc. PI 66)
The example is compelling; and even if by some feat of ingenuity
someone managed to find a non-trivial necessary and sufficient
condition for being a game, it would still be clear that that
was not why we applied the word game in all the cases where
READING THE TEXT
43
we do apply it. Rather the explanation that Wittgenstein
appears here to be gesturing at is this: the reason we apply
the expression game to certain activities is because of their
bearing a similarity or perhaps some other relationship to
other things that we already call games. And one might give
more or less similar explanations for the use of many words.
8

Wittgenstein calls these relationships family resemblances and
writes that the items falling under them form a family (PI 67a);
I shall also say that concepts that resemble game in this respect
are family resemblance concepts.
But if I dont know the necessary and sufficient conditions
for somethings being a game then how am I supposed to explain
to someone else what a game is? Say, a child who has heard his
brother say Lets play a game and who wants to know what
game means; or to a native Francophone who knows enough
English to understand an explanation of that word but not yet
the word itself ? Well, what do we actually do in these cases? we
give the pupil examples of games and we say: This and similar
things are called games (PI 69). And the vagueness inherent in
the word similar is not a vague way of gesturing at something
that I know more precisely but for some reason cannot say
(PI 71b): for there is nothing more to know about what game
means. And if the explanation works, so that my pupil now
uses the word game as I do (allowing for independent differ-
ences of opinion between us), why should we not say that I have
told him precisely what I meant by it?
9
I shall return to the issue
of vagueness at 2.2.
Now as well as to game Wittgenstein applies this point to
language, language-game and proposition: there need be no
one thing that all languages or propositions have in common,
for there need be no one such thing that explains our pattern of
applying these words. We see that what we call sentence and
language has not the formal unity that I imagined [in the
Tractatus], but is the family of structures more or less related to
one another (PI 108).
But just how telling is this point against the Tractatus? Suppose
that Wittgensteins earlier self agreed that our ordinary use of the
word language, like our ordinary use of, say, matter, is some-
thing of a mess. He might still insist that what he was discussing
in the Tractatus was not the language of everyday speech quite
44
possibly no one thing answers to that but to a somewhat more
regimented successor-concept that picks out a particular kind of
language-game of the utmost importance: the language-game
of describing how things actually stand. One can certainly agree
with Philosophical Investigations that it is not essential to all
language-games that they describe how things stand in some
respect (cf. PI 363); but why should it not be essential to a very
important and interesting subset of them: a subset moreover that
includes the language of science? And if we do restrict ourselves
to that subset, why can we not say that on the criteria that are
appropriate for them, the very meanings that sentences of the
language-games (a) and (d) express obscurely get clear expression
in the language-games (b) and (c) respectively? But this was
the question that prompted Wittgensteins discussion of family
resemblance in the first place. So it looks as though we might
accept the main conclusions of that discussion while also holding
on to the Tractarian ideas that they were supposed to dislodge.
Of course this is not to say that the Tractatus has given a
correct account of how even some limited sector of language
achieves what it seems to. Indeed there are still many serious
criticisms of its detail to come. But it is to say that considera-
tions of family resemblance do not undermine the Tractatus
project itself; they only restrict its scope.
I return finally to the point that concluded chapter 1. Recall
Wittgensteins denial at PI 34 that what distinguishes the act of
attending to somethings shape from, for example, attending to
its colour say, in the course of giving an ostensive definition
is some accompanying experience. His criticism, which seemed
decisive, was that even if such an experience took place in him,
somebody who went on in consequence to use the defined term
as a colour term would have been attending all along to the
colour and not to the shape. This point is so straightforward and
compelling that it raises the question of how things could even
have seemed otherwise. Why was the idea of an accompanying
experience even prima facie attractive?
At PI 36 Wittgenstein says that we do here what we do in a
host of similar cases: because we cannot specify any one bodily
action which we call pointing to the shape . . . we say that a spiritual
activity corresponds to these words. PI 33 has already empha-
sized the fact that we cannot find any one bodily action (recall
READING THE TEXT
45
the list of ways in which one can attend to the colour blue, PI 33);
but the reason that this makes us postulate a single type of spiri-
tual activity behind all of these behavioural ones is precisely
because we think that they must all have one thing in common.
But now we see that they need not: attending to the shape might
be a family resemblance concept. That is to say, it might correctly
describe different bodily actions on different occasions; but it does
not owe this to any one spiritual action that unites all such cases.
We have already seen four cases of temptation to postulate
a spiritual activity behind the bodily ones that seem to express it.
These were (a) the meaningful use of linguistic expressions;
(b) attending to the colour of something (rather than, for
example, its shape); (c) meaning a sentence as composite rather
than simple; (d) meaning something composite when uttering
a simple sentence. And we have seen how consideration of
family resemblance explains the temptation in case (b) or
rather how our failure to consider it does.
It is not so clear that such a failure is what explains the temp-
tation in case (d): what lay behind that was rather the Tractatus
demand for compositeness as a necessary condition upon mean-
ingfulness. But in cases (a) and (c) the explanation is plausible.
Case (a) was the temptation to say, of the grocer of PI 1, that his
overt use of apple was the tip of an iceberg: behind all of its
various uses is an act of mental comparison that explains it (e.g.
comparison with an idea). And one reason for thinking this is
the fear that otherwise there would be no one explanation for all
of his various uses of the word apple. But why should there be
just one? And in case (c) we suffer from two illusions: first we
think that meaning it as composite must be something that
goes on when he says it (here it is occurrences of meaning it as
gerund or gerundive that fool us); and then finding no one
bodily action that goes on when you mean a sentence as, for
example, four words rather than one. The second illusion arises
from a failure to see that instances of meaning it as composite
might form a family.
2.2. Vagueness
2.2.1. Vagueness and family resemblance
We can best introduce Wittgensteins notion of vagueness by
means of a passage in which he appears to confuse it with
46
family resemblance:
How should we explain to someone what a game is? I imagine
that we should describe games to him, and we might add:
This and similar things are called games . And do we know
any more about it ourselves? Is it only other people whom we
cannot tell exactly what a game is? But this is not ignorance.
We do not know the boundaries because none have been
drawn. To repeat, we can draw a boundary for a special
purpose. Does it take that to make the concept usable? Not at
all! (Except for that special purpose.) (PI 69)
A vague concept is one whose application is not everywhere
determined by ones grasp of it: thus, for example, the concept
blue is vague because, although somebody who grasps it will
typically know that the sky is blue and that grass is not, there
may well be objects that he will not classify in either category
even though he knows everything relevant to settling it (i.e. he
is looking at them in good light, his eyes are functioning well,
etc.). We may call objects of this latter sort borderline cases;
so the existence or at least possibility of borderline cases can be
taken as the defining feature of vague concepts. In this passage
Wittgenstein appears to be suggesting that game is vague. (We
do not know the boundaries because none have been drawn.)
But we should distinguish family resemblance from vague-
ness. Family resemblance is neither necessary nor sufficient for
vagueness. Not sufficient because a concept might have a family
resemblance character and yet not admit of borderline cases.
Thus take Wittgensteins own example of number:
And for instance the kinds of number form a family in the
same way [as do games]. Why do we call something a num-
ber? Well, perhaps because it has a direct relationship
with several things that have hitherto been called number;
and this may be said to give it an indirect relationship to other
things that we call the same name. (PI 67)
In fact these relationships appear to be quite different. Thus
integers count as numbers because they are what you get if you
close the set of natural numbers under subtraction; rational
READING THE TEXT
47
numbers count as numbers because they are what you get if you
close the set of integers under division (except by zero); real
numbers count as numbers because they are what you get if you
close the set of rational numbers under the operation that forms
suprema from arbitrary sets of them that are bounded above;
and complex numbers are numbers because they are what you
get if you close the set of real numbers under the operation that
forms roots of arbitrary polynomials with real coefficients. But
each of these relationships has a perfectly definite extension
there are no borderline numbers in that sense in which there
might be borderline blue objects. So the concept of number is
not a vague concept.
10
Or again, one might imagine that the concept of game
was circumscribed in such a way that there could not be any
borderline games; and yet in such a way that it still included
board-games, card-games, Olympic games and the rest. In that
case it would be true to say that games form a family but false to
say that no boundaries have been drawn. And although this is
not how things actually are, its possibility suffices to establish
what the actual situation with numbers also shows: that family
resemblance does not suffice for vagueness.
Nor does vagueness suffice for family resemblance. The concept
of a tricolor admits of straightforward analysis: it applies (let
us say) to flags that carry a certain arrangement of a red, a white
and a blue vertical stripe. And it is because we deem them to have
that arrangement that we apply it to just the flags that we do. So
tricolor is not a family resemblance concept. On the other hand
it is a vague one since red stripe and blue stripe clearly admit
of borderline cases: if any concepts are vague then they are.
Hence so too is the concept that is compounded from them
in the way that tricolor is: it is easy to imagine a borderline
tricolor. So a concept can be vague without being a family
resemblance concept; and a concept can be a family resemblance
concept without being vague.
2.2.2. Vagueness and analysis
But vagueness is as relevant as family resemblance to the discus-
sion at this point, though for a slightly different reason. The
discussion at PI 604 concerned the language-games (a)(d)
(defined at the start of this chapter). The whole issue between
48
Wittgenstein and his interlocutor concerned the identities of
these language-games. More precisely, it was over whether the
differences between the uses of sentences in (a) (Bring me
the broom) and (b) (Bring me the broomstick which is fitted
in the brush) constituted a difference in their meanings; and
equally over whether the differences between the uses of sen-
tences in (c) (RG) and (d) (U) constituted a difference in their
meanings. And we may try to settle this issue in the manner of the
Tractatus: find the essence of language and propositions and then
apply it to the present case. We have seen that Wittgenstein rejects
that because he thinks that language-games form a family.
But the aim of comparing language-games (c) and (d) had
been to illustrate a particular sort of point about analysis:
To say . . . that a sentence in (b) is an analysed form of one
in (a) readily seduces us into thinking that that the former is
the more fundamental form; that it alone shews what is meant
by the other, and so on. For example, we think: If you have
only the unanalysed form you miss the analysis; but if you
know the analysed form that gives you everything. But may
I not say that an aspect of the former is lost upon you in the
latter case as well as the former? (PI 63)
The point about (c) and (d) was that something of the gestalt
aspect of the experiences recorded in the unanalysed (d) (U)
is lost in translation to the analysed (c) (RG). But there is
another way to illustrate the point at PI 63, for another aspect
that may also be lost in analysis is the essential vagueness of the
unanalysed expressions; that this is so is one of the things that
Wittgensteins discussion of vagueness aims to establish.
In the Tractatus the process of analysis was supposed to lead
to the elementary propositions out of which the analysandum
was truth-functionally constructed (TLP 4.4). We might for
instance analyse P (The broom is in the corner) as O1 & O2 &
O3, where O1, O2 and O3 are elementary propositions (as it
might be, The broomstick is in the corner, The brush is in the
corner, The broomstick is fixed in the brush). The analysis
is supposed to preserve truth-conditions, so P is true in just
those situations in which (O1 & O2 & O3) is true, that is, in
just those situations in which each of O1, O2 and O3 is true.
READING THE TEXT
49
Now the elementary propositions are supposed to be maximally
precise, for elementary propositions are pictures, and
What constitutes a picture is that its elements are related to
one another in a determinate way . . . The fact that the ele-
ments of a picture are related to one another in a determinate
way represents that its elements are related to one another in
the same way. (TLP 2.14, 2.15)
Hence every possible world is such as to make an elementary
proposition definitely true or such as to make it definitely false.
The same therefore applies to all truth-functional combinations
of elementary propositions; and if the programme of analysis in
the Tractatus is well-conceived then that means all propositions.
So if that programme is well-conceived, every proposition draws
a definite boundary around the possible situations that make
it true: to use language is therefore also to express a thought
of complete precision. A proposition has one and only one
complete analysis. What a proposition expresses it expresses in
a determinate manner (TLP 3.253.251).
But Wittgenstein now argues that the vagueness of our language
is essential to it. It is not that our language contains vague means
of expressing precise thoughts; it is that it contains vague means
of expressing vague thoughts. To eliminate the vagueness is to
lose an aspect of what you meant all along.
At PI 7088 he illustrates this point with various examples of
vague orders, vague descriptions and vague concepts. Among
these is the description Moses did not exist. Wittgenstein agrees
with Russell that names may be defined by means of definite
descriptions, that is, expressions of the form The so-and-so.
Thus we might define Moses by the definite description The
man who led the Israelites through the wilderness, or by the
definite description The man who lived at that time and place
and was then called Moses , or by any of many other such
definite descriptions (PI 79a).
Each such definition could be regarded as contributing to the
analysis of Moses did not exist, that is, to a specification of just
those possible situations in which it is true and just those in
which it is false. Thus Wittgensteins first illustrative definition
would make Moses did not exist true in just those situations in
50
which no one man led the Israelites through the wilderness,
even if some one man did do all of the other deeds that are
commonly credited to Moses. And his second illustrative defini-
tion would make Moses did not exist true just in case no one
man who lived at that time and place was called Moses, even if
someone called Schmoses did all of the deeds that are credited
to Moses (PI 79a).
But spelling out these consequences of such definitions is
enough to show them up as unsatisfactory; for none of these
paraphrases, nor anything as definite as any of them, seems
to capture what I mean by Moses did not exist. Wittgenstein
therefore offers another definition: I shall perhaps say: By
Moses I understand the man who did what the Bible relates of
Moses, or at any rate a good deal of it (PI 79b). So by Moses
did not exist, according to this definition, I shall mean roughly
this: No one man did very many of the things that the Bible
relates of Moses. That is a better account of what I mean.
But of course it makes my meaning vague because it gives no
answer to the question How many of those things? Suppose
it were discovered, for example, that there was a man whom
Pharaohs daughter rescued from the Nile, who called forth a
plague of frogs upon Egypt and who was called Moses, but
that neither he nor anyone else led the Israelites through the
wilderness or brought down the Ten Commandments, and so
on. Am I to say that Moses did not exist is true or that it is
false? Here we have a case that the definition simply does not
decide, for nothing in it settles whether or not that hypothetical
situation is one in which the man called Moses did very many of
the things that the Bible relates of him. So what I meant by
Moses did not exist was not something that drew a definite
boundary around all of the possible situations that make it true.
And any attempt to replace it with something that does draw
such a boundary will falsify what I meant.
If someone were to draw a sharp boundary I could not
acknowledge it as the one that I too always wanted to draw,
or had drawn in my mind. For I did not want to draw one
at all. His concept may then be said to be not the same as
mine, but akin to it. The kinship is that of two pictures, one
of which consists of colour patches with vague contours, and
READING THE TEXT
51
the other of patches similarly shaped and distributed, but
with clear contours. (PI 76)
This beautiful metaphor is here applied to concepts but the point
clearly transposes to propositions. The points on the canvas repres-
ent possible situations and the patches represent regions of points
corresponding to possible situations that make a given proposition
true. The point is that you can draw no sharp boundary around
such regions if the proposition is Moses did not exist.
2.2.3. Objections to Wittgensteins account
But there are reasons for thinking either (a) that this must be an
illusion because one cannot mean something vague; or (b) that
illusion or not, it is pernicious because one should not mean
something vague. The reasons that Wittgenstein discusses for
(a) are as follows: (i) that if one means something vague then
one doesnt really know what one means; (ii) that a vague
concept is unusable; (iii) that the mental samples that we
associate with words eliminate all vagueness in their application.
For (b): that exactness is always preferable to inexactness. Let
us consider these in turn.
(i) We should distinguish this objection to vagueness from a
similar-sounding objection to family resemblance that we have
already discussed. That other objection was that you do know
what you mean by, for example, game but for some reason you
cannot say it. But the present objection is that you do not know
what you mean by a vague concept. The objector says: If the
concept game is uncircumscribed like that, you dont really
know what you mean by game. Wittgenstein answers with a
rhetorical question:
When I give the description: The ground was quite covered
with plants do you want to say I dont know what I am talking
about until I can give the definition of a plant? (PI 70)
By itself this is decisive clearly in whatever sense of knowing
what one is talking about that is relevant to linguistic under-
standing, one does know what one is talking about but it is
worth saying a little more about what motivates the objection.
One way to get clearer on what somebody means in cases of
avoidable unclarity is to ask him to specify the possible situations
52
that would make his statement true. For instance, if a witness to
a car accident says in court I was driving faster than usual it
would be reasonable to ask him whether, say, 60 mph counts for
him as faster than usual, and if not whether 70 mph so counts,
and so on. This is in effect a way of getting him to say more pre-
cisely than he initially did what he meant all along, by specifying
of certain possible situations for example, one in which he was
driving at 60 mph, or one in which he was driving at 70 mph
whether or not each of them makes his sentence true. If his
responses to enough such questions are consistent then we shall
find it natural to call them clarifications of what he had meant
all along by faster than usual.
But if that is how we ascertain what someone really means,
then it is also natural to think that somebody who cant answer
such clarifying questions doesnt know what he means. Certainly
this response would be reasonable in some cases. Suppose I say
The ground was quite covered with plants. Then you ask me
Do spider plants count as plants?, Do geraniums count as
plants?, and so on. But to each question I dont answer yes
or no; I just say I dont know whether they are plants, I just
mean that it was quite covered with plants. If for enough species
of plants (or other living things) I cannot say whether or not it
counts as a plant as I initially meant that word, it will be safe to
conclude that I didnt know what I meant when I said that the
ground was quite covered with plants.
But while it is true that a sufficiently porous concept, as we
might call it, is not really a concept at all, it doesnt follow that in
order to mean anything at all I must be able to stop all such gaps
(cf. PI 99). Consider again the example of the witness to the car
accident. The barrister asks him What exactly counts for you as
faster than usual? Does 60 mph count? Yes. Does 50 mph
count? No. Does 55 mph count? Yes. Well then, does
52.5 mph count for you as faster than usual? Here we can
imagine the witness saying that he doesnt know. But this is not
because he didnt know what he meant all along, but because he
didnt mean anything more precise than what he had already
said. To think the former would be to apply an everyday criterion
for not knowing what one meant with a nicety that is quite
inappropriate. The temptation to do so is what lies behind the
interlocutors objection at PI 70.
READING THE TEXT
53
(ii) Wittgensteins discussion of the second objection is
equally brief and rhetorical; but again there is a good deal
more to be said.
Frege compares a concept to an area and says that an area
with vague boundaries cannot be called an area at all. This
presumably means that we cannot do anything with it. But
is it senseless to say: Stand roughly there? (PI 71)
Of course it is not senseless; of course we can do something with
it. But this does not really deal with the best reason for thinking
that vague concepts are unusable, which is as follows.
It is of the essence of many vague concepts because it is
essential for our use of them that their application is in optimal
circumstances assessable solely on an observational basis. To put
it more plainly: in optimal circumstances (good light, good
eyesight, etc.) I can tell that something is red just by looking;
I can tell whether you are standing roughly over there just by
looking; I can tell whether a siren is loud just by listening; I can
tell whether a rose is fragrant just by smelling, and so on. And if
I cannot tell by such means, then nothing else is going to settle it.
If a book has a cover that I hesitate to class as either red or not
red then neither looking at any other thing nor any other sort of
inquiry is going to settle it for me.
Now vague concepts that have this feature will also have
another one: that one cannot consistently apply the concept to
just one of two objects that are observationally indistinguishable,
at least along the dimension that determines its application. For
instance, one could not say in optimal circumstances of two
visible but chromatically indistinguishable balls that one was red
without agreeing that the other one was too. Doing so would be
inconsistent with the condition that you can tell whether a ball is
red just by looking. Since looking tells us the same whichever
ball we look at, if it tells us that one ball is red then it tells us that
the other ball is red too. Notice that I am not just saying that one
cannot affirm that one ball is red while denying that the other one
is. I am saying something stronger than that: one cannot even
affirm that one ball is red while withholding judgement on the
other. Even that would be inconsistent with the condition that
you can tell whether a ball is red just by looking.
54
Now among the concepts that have this feature there are some
that have this further one: that there is a finite sequence of objects,
of which the first definitely is and the last definitely is not an
instance of the concept, but such that no object in the sequence is
observationally distinguishable from its successor along the
dimension that determines the application of the concept (except
obviously the last one, which has no successor). For instance, we
might imagine a sequence of coloured balls of which the first is
clearly red and the last is clearly green. The balls are so arranged
that the reflectance properties of the surface of each ball are very
slightly different from those of its predecessor too slightly to
register as a visible difference in colour. So balls that are adjacent
in this sequence are chromatically indistinguishable. Let us call
such a sequence a Sorites sequence.
We are finally in a position to see why vague concepts that
satisfy these conditions, and many propositions that involve
them, might after all be unusable. The difficulty is that given
a Sorites sequence we can easily get a competent user of the
corresponding vague concept to contradict himself. For instance,
somebody who accepts that the first member of the sequence of
balls is red will have to say the same of the second one. But if he
says it of the second one then he will have to say it of the third
one . . . until in the end he finds himself saying of a visibly green
ball that it is red!
Or consider Wittgensteins own example. If your standing on
that is, your centre of gravitys being directly above a point on
the surface of the Earth counts as obedience to what I meant
when I said Stand roughly there, then so does your standing on
a point that is 1 nm (1 10
9
m) to the north of it. But then a
lengthy argument shows that if I say Stand roughly there,
pointing at the region around the base of Nelsons Column in
London, you could comply with my order by standing at the
North Pole. The same point applies to ethical concepts, at least
if they too admit of a continuum of intermediate cases (as
Wittgenstein seems to think: PI 77a): if it is all right to touch
somebodys elbow with your little finger then it is all right to
sleep with her. And so on.
Here we come close to the idea which Frege had, and which one
can find so hard to grasp, that the use of vague expressions is
READING THE TEXT
55
fundamentally incoherent. One may be inclined to dismiss
Freges idea as a mere prejudice if one does not reflect on
examples such as these. (Dummett 1975b: 262)
In short it is their liability to generate contradictions that makes
vague concepts unusable. Nothing in the Philosophical Investiga-
tions discussion of vagueness directly addresses this point.
However it is evident from certain of his other writings what
Wittgenstein would have said to anyone who pressed it against
him. He would have said that the possibility of deriving a contra-
diction from certain rules does not make those rules unusable if
we do not in fact derive the contradiction.
[S]uppose that there is a contradiction in the statutes of a
particular country. There might be a statute that on feast days
the vice-president had to sit next to the president, and another
statute that he had to sit between two ladies. This contradiction
may remain unnoticed for some time, if he is constantly ill
on feast-days. But one day a feast comes and he is not ill.
Then what do we do? I may say, We must get rid of this
contradiction. All right, but does that vitiate what we did
before? Not at all. (LFM 210)
In real life nobody ever does come across Sorites sequences; or
if anyone does he does not base any practical conclusions upon
the application of vague concepts to them. So in real life vague
concepts are perfectly usable, even though a blindly mechanical
application of them to such sequences would be of no use to us.
It is worth making two brief points about that. First,
Wittgensteins analogy of statutes can be somewhat misleading,
for it gives the impression that any inconsistent set of rules and
axioms is not thereby unusable so long as we do not derive the
contradiction itself (cf. the exchange with Turing at LFM 220).
But that is just not true: if a set of sentences is inconsistent then
you can successively derive both a statement and its negation
from it without ever proceeding via an explicit contradiction. But
this point is immaterial to the case at hand: so long as either
vague concepts are not in fact applied to Sorites sequences, or the
results of doing so are disregarded, nothing stands in the way of
saying that such concepts are usable. After all, they are used.
56
Second, beneath the somewhat bland pragmatism that I am
here attributing to Wittgenstein there may be a deeper reason for
rejecting the inference from inconsistency to unusability. For as
we shall see, he can seem to be saying that a blind mechanical
application of whatever rules govern the employment of my
concepts is not a way of revealing something that was hidden all
along within them. So in particular the deriving of a contradic-
tion from them does not show that something was wrong with my
rules all along. For I want to say and it stops being a game
not: and now we see that it wasnt a game (RFM III-77). I
shall return to the matter at 3.4.2.
(iii) It is easy to think of the grasping of a concept in terms
of a visual metaphor. This manifests itself in everyday turns
of phrase: when you grasp, for example, the colour concept
yellow ochre you may be said to see what is common to all
of its instances. It is only slightly harder to take such turns
of phrase literally and to think that when one grasps a word
one becomes acquainted with an object of introspection. This
inner object then gets compared with outer ones in order to
settle the application of a word: in the case (e.g.) of yellow
ochre it applies to whatever outer object chromatically matches
ones inner sample. But if that is what happens when one
grasps colour terms, how can they be vague? After all, it is not
a vague matter whether or not this colour that I see here is or
is not a match for the inner object that I associate with the
term yellow ochre.
Wittgenstein is later going to attack the very idea of an inner
object of introspection, cognizance of which constitutes your
grasp of, say, a colour term (see, for example, PI 293, discussed
at 4.1.3). But at this point he answers in another way. He says:
suppose that there was such an inner object and consider how it
would work for colours that admitted of many shades (say, blue
or green). In that case there is typically no question of a chro-
matic match between the inner sample and the outer object: for
green to admit of different shades is for a range of chromatically
distinguishable items to fall under it. So the mere having of an
inner sample of green is not going to be of any use to me. It does
not settle precisely what does, and what does not, fall under the
term green as I understand it. To do that I need in addition to
have some policy of how to use it; in particular some policy of
READING THE TEXT
57
saying that green applies to something if and only if the latter
lies within such-and-such a chromatic range of my sample.
Which shade is the sample in my mind of the colour green
the sample of what is common to all shades of green?
But might there not be such general samples? Say a . . .
sample of pure green? Certainly there might. But . . . for a
slip of pure green to be understood as a sample of all that is
greenish and not as a sample of pure green this in turn
resides in the way the samples are used. (PI 73)
Far from eliminating the vagueness of green this analysis of my
grasp of that term merely relocates it: for if I do use the sample
as a sample of all that is greenish then my judgements of
whether something resembles that sample enough to be called
green itself will inevitably admit of borderline cases.
The argument so far has been that the mere presence in my
mind or for that matter on a piece of paper of a sample
associated with green is not going to tell me how to apply it.
But while this may seem to be true of that kind of sample there
may also seem to be other sorts of sample that do intimate their
application to an observer. The cases that I have in mind are those
in which you see something as something: what Wittgenstein else-
where calls cases of aspect-perception (PI pp. 1656/1934). A
good example is the drawing on PI p. 166/194 of a duck-rabbit:
an object that can appear either as a duck or as a rabbit; and here
it looks as though I am describing a difference in the sorts of
visual experience that it can produce. Or again one can see a
squiggly line as a handwritten word; and on the one hand this
appears to be a sort of experience (see examples (c) and (d) at
PI p. 169/198); but on the other hand someone who sees it as
such will regard a certain range of other patterns as resembling
it; whereas someone who just sees the squiggly line will see no
resemblance between tokens of (e.g.) the word pleasure that
are written in two very different hands. Now couldnt it happen in a
similar way that one sees a sample of green as a sample of all
that is greenish rather than as a sample of pure green one only
attends, so to speak, to the greenish aspect of it? On the
one hand this would seem to be a sort of experience. And on
the other hand it would seem to be one that directs your policy.
58
Wittgenstein admits that such a thing might happen:
Of course, there is such a thing as seeing in this way or that;
and there are also cases where whoever sees a sample like this
will in general use it in this way, and whoever sees it otherwise
in another way. (PI 74)
But he insists that this seeing of the sample in a certain way is
only contingently related to the use to which you then put it. He
does so in connection with a different example: the idea that one
might use an inner sample of a leaf as a sample of leaf-shapes
in general because of the way in which one sees it:
[T]he idea that if you see this leaf as a sample of leaf shape
in general you see it differently from someone who regards it
as, say, a sample of this particular shape. Now this may well
be so though it is not so for it would only be to say that, as
a matter of experience, if you see the leaf in a particular way,
you use it in such-and-such a way or according to such-and-
such rules. (PI 74)
Here Wittgensteins point appears to be that seeing a leaf, say, as
a sample of leaf-shapes in general is not yet to know how to
apply it: it is just an experience, and how one goes on to use it is
still up to you. On this point he appears to agree with his earlier
self (TLP 5.5423) but not with some of the things that he says in
the subtler exploration of these phenomena at Philosophical
Investigations II, xi. Consider this passage:
If I saw the duck-rabbit as a rabbit, then I saw: these shapes
and colours (I give them in detail) and I saw besides some-
thing like this: and here I point to a number of different
pictures of rabbits. This shews the difference between the
concepts.
Seeing as . . . is not part of perception. And for that
reason it is like seeing and again not like. (PI p. 168/1967)
Here it looks as though seeing something as something is less a
matter of what one sees and more a matter of how one treats it
(e.g. what one points to in order to describe what one saw).
READING THE TEXT
59
In that case, the connection between (a) seeing a sample as such-
and-such and (b) using it to classify other objects as relevantly
like or unlike it, is not adventitious but conceptual: to see an
inner sample as (e.g.) a sample of a cube, of a leaf-shape in
general or of all that is greenish, is in part to classify these
things and not those as relevantly like it: as cubes, leaf-shaped
objects or green things.
It may therefore be that the remarks on aspect-perception in
the later part of the book can after all be developed into a partial
defence of the tendency, noted at the beginning of this section,
to assimilate understanding to seeing. Instead of doing so
I merely note here that success in that project would do nothing
to save Wittgensteins main target in the material around PI 74.
That was the idea that no inner (or outer) sample can
eliminate vagueness in the application of a concept. For even
if seeing the shape of a leaf as a sample of leaf-shape in general
is somehow an experience that simultaneously guides my policy
of classification, there is still no reason to expect that it will
recommend a definite verdict in every case. On the contrary,
although I will be certainly inclined to consider the general leaf-
shape to be relevantly like some things and not relevantly like
others, borderline cases will remain too.
(b) Wittgenstein also attempts to deal with the idea that
whether or not we do in fact use vague expression, we should
not do so. He makes two important points in this connection.
The first is that greater precision of expression can often
be pointless. It all depends on what you want to do with the
proposition or concept that you are making more precise.
If I tell someone Stand roughly here may not this
explanation work perfectly? And may not [any] other one
fail too?
But isnt it an inexact explanation? Yes; why shouldnt we
call it inexact? Only let us understand what inexact
means. For it does not mean unusable. And let us consider
what we call an exact explanation in contrast with this one.
Perhaps something like drawing a chalk line round an area?
Here it strikes us at once that the line has breadth. So a colour-
edge would be more exact. But has this exactness still got a
function here: isnt the engine idling? (PI 88ab)
60
It is very easy to imagine situations in which the point of saying
Stand roughly here would not be served any better by a speci-
fication of the desired region that used colour-edges than by one
that used a chalk line (this would be the case, for instance, if the
person saying it was a bowler in a cricket team instructing one of
his fielders).
That is obviously right and calls for little comment; but the
second point is rather less plausible.
Inexact is really a reproach, and exact is praise. And that
is to say that what is inexact attains its goal less perfectly
than what is more exact. Thus the point here is what we call
the goal. Am I inexact when I do not give our distance from
the sun to the nearest foot, or tell a joiner the width of a table
to the nearest thousandth of an inch?
No single ideal of exactness has been laid down; we do not
know what we are supposed to imagine under this head
unless you yourself lay down what is to be so called. But you
will find it difficult to hit upon such a convention; at least any
that satisfies you. (PI 88de)
The argument seems to be that (a) one cannot reproach,
for example, a measurement for being inexact unless one has
a concept of the ideal of exactness. But (b) we have no such
concept that is, we do not know what would count as an
unimprovable degree of precision.
But there are two obvious responses to this. The first is that
even if both points were true of the case of measurement, it is
not clear that point (b) is true of the case that more immediately
concerns us, that is, propositions. We might say that a proposi-
tion is inexact if one cannot draw a definite boundary around its
truth-conditions, that is, if there is no one set of possible worlds
consisting of all and only those situations in which it is true, the
others comprising all and only the possible situations in which it
is false. And so a proposition may be regarded as perfectly exact if
it conforms to the ideal of the Tractatus: every possible situation
is either one that makes it true or one that makes it false: there
are no borderline cases. It may be true that in fact our thoughts
do not conform to this ideal; but the ideal is still there.
READING THE TEXT
61
And the second response is to question (a). It may seem an
odd state to be in, but there seems to be no incoherence in always
preferring more exact to less exact methods of measurement
even if you have no conception of what unimprovable precision
would be like (compare: always preferring the bigger of two
numbers). Of course one would need to grasp the relation that
holds between two methods of measurement when one is more
precise than the other; but Wittgenstein is not expressing any
doubts about that.
The importance of the general preference for precision perhaps
lies not so much in Wittgensteins treatment of it at PI 88 as in his
regarding it as symptomatic of a certain conception or rather,
as he now thinks, misconception of the nature and task of
philosophy itself. We turn now to his treatment of that subject.
2.3. The nature of philosophy
Of all the subjects of which Philosophical Investigations treats,
this is probably the one on which its views have the most in com-
mon with those of the Tractatus. However there are also points
of disagreement, as is evident from the preceding discussion of
PI 88. Here I discuss both continuities and differences in connec-
tion with the idea that philosophy is a body of activity and not a
doctrine, aimed at revealing the nonsensicality of what philo-
sophers had traditionally taken their subject to be.
In the Tractatus Wittgenstein had maintained that philosophy
was not a body of empirical or a priori doctrine but a form of
activity (TLP 4.1114.112). Its task was not to help us to answer
questions with which science can also help us; its task was rather
to set limits to what science can say, or what he then thought
amounted to the same thing to what can be thought (TLP
4.1134.114).
In fact the claim that philosophy must at least include activities
as well as doctrine follows from this conception of its task on the
Tractatuss own theory of meaning. Philosophy can only state the
limits of thought and language by means of propositions that
generalize over all propositions, including themselves. But
No proposition can make a statement about itself, because a
propositional sign cannot be contained in itself (that is the
whole of the theory of types). (TLP 3.332)
62
For instance, suppose that some proposition asserted that all
propositions are pictures (TLP 4.021). If that proposition is to
have the generality that is necessary for setting a limit to what
can be thought, it must apply to itself. So it must be saying
among other things that the proposition that all propositions are
pictures is itself a picture. Now consider the two occurrences
of the functional sign x is a picture in the proposition that the
proposition that all propositions are pictures is itself a picture.
Any sign, Wittgenstein says, contains a syntactic specification of
all the meaningful substitutions for x, that is, a specification of
what signs can be substituted for x (compare the sign n + 2 = m,
where the use of the expressions n and m might conventionally
be taken to indicate that only Arabic numerals can be substi-
tuted for them). But this means that the sign cannot itself be
among those substitutions:
In logical syntax the meaning of a sign should never play a
role. It must be possible to establish logical syntax without
mentioning the meaning of a sign: only the description of
expressions may be presupposed. (TLP 3.33)
The reason why a function cannot be its own argument is that
the sign for a function already contains the prototype of its
argument, and it cannot contain itself. (TLP 3.333)
And it follows that if the proposition that the proposition that all
propositions are pictures is itself a picture says anything at all,
the two occurrences in it of x is a picture must be occurrences of
different signs, since the first sign cannot contain itself, whereas
the second sign clearly can because in this propositional sign
it does contain the first sign. Since therefore they admit of
different substitution instances they must be different signs.
(This is the linguistic counterpart of TLP 2.0213.) But then our
proposition is not making a suitably general claim: it does not
say of itself what it tries to say of everything else, any more
than The only thing that deters hardened criminals is a long
sentence is a long sentence says of The only thing that deters
hardened criminals what the latter says of the only thing that
determines hardened criminals. So our proposition has misfired:
if it made a claim at all it was not the one that was intended.
READING THE TEXT
63
Thus in order to achieve the aim of philosophy we must resort
not to stating the limit of thought but to showing others that
their assertions have crossed that limit.
The correct method in philosophy would really be the follow-
ing: to say nothing except what can be said, i.e. propositions
of natural science i.e. something that has nothing to do with
philosophy and then, whenever someone else wanted to say
something metaphysical, to demonstrate to him that he had
failed to give a meaning to certain signs in his propositions.
(TLP 6.53)
This activity of demonstrating that he had failed to give a
meaning to certain signs is what philosophers who have read
the Tractatus ought to be doing to anyone who tries to say
something metaphysical.
Now in Philosophical Investigations he rejects that earlier reason
for thinking that these truths cannot be stated. For he no longer
thinks that the meaning of a sign must be laid down once and for
all. He now thinks on the contrary that there may be continuities
between different patterns of use, so that by extending the use of
a sign to cover new cases one might still be keeping faith with
its old meaning. This might happen, for example, when people
who initially applied the word pain only to living things then
began applying it to dolls (cf. PI 282); or when people who
initially thought that only visible operations on paper counted as
calculations then began to recognize an activity of calculating
in the head (cf. PI 364, 3856). Equally then, somebody who
had used the word proposition to apply only to a limited range
of sentences might with similar fidelity to what he already meant
begin to apply it to a wider range of sentences. (I discuss the
application of this point to sensation ascriptions at 4.2.6.)
But although he no longer has this reason for doing so, he
continues to accept that philosophy is the activity of showing
people who scratch their heads over metaphysical doctrines that
they really are nonsensical:
The results of philosophy are the uncovering of one or
another piece of plain nonsense and bumps that the under-
standing has got by running its head up against the limits
of language. (PI 119)
64
If one tried to advance theses in philosophy, it would never
be possible to debate them, because everyone would agree to
them. (PI 128)
The reason for recommending this activity is that the philosoph-
ical talk it is directed against is itself an activity in which language
really has no function what appears to be meaningful (because
grammatically well-formed) sentences are truly meaningless
because they have no use at all:
When philosophers use a word knowledge, being, object,
I, proposition, name and try to grasp the essence of
the thing, one must always ask oneself: is the word ever actually
used in this way in the language which is its original home?
What we do is to bring words back from their metaphysical
to their everyday use. (PI 116)
So what the philosopher ought to do is this: faced with someone
who wonders whether, for example, all words are names, or
whether space is necessarily three-dimensional, one does not
endorse or oppose that thesis, one reminds him of our ordinary
use of the expressions word, name, space, and so on and then
he sees that his sentence is really meaningless. The work of the
philosopher consists in assembling reminders for a particular
purpose (PI 127).
What are we to make of this? It can certainly appear that philo-
sophical claims of any great generality are just meaningless.
Continuing with the example All words are names: if somebody
said that in a philosophical context it would not be clear at once
just what he meant by it. And Wittgenstein is surely correct that
this is because he is using the word names in a way that is very
distant from its non-philosophical uses. So somebody who had
only learnt to use the word name in non-philosophical contexts
(What are the names of these children?; Everybody whose
name begins with S step forward) might well find himself at a
loss in face of the general Augustinian thesis.
But why should the correct response to this be to give up on
the philosophical claim? Why couldnt one admit that without
further explanation one only has a hazy idea of what it means,
READING THE TEXT
65
but then instead of giving up try to give it a meaning whose truth
one could then test?
Obviously one couldnt give it any meaning one liked, or at
least not if one wished to maintain that one was expressing a
philosophical thesis. If I announce that by All words are names
I am going to mean what everyone else means by Pigs fly, and
then conclude that All words are names is false, nobody is going
to think that I have said anything of relevance to philosophy.
But in contrast with this case there are ways of making a word
precise that can be said to keep faith with our possibly hazy
everyday understanding.
Thus scientific enquiry frequently gives words some precise
meaning that emphasizes just one of the many criteria that
unsystematically settle everyday usage depending on the purpose
at hand. For instance, somebody who wishes to measure changes
in the income of a nation cannot be content with the ordinary
word income because of its vagueness; he has to replace it with
some such specific notion as GDP, that is, some measurable
variable that captures at least some of the things about income
that made the initial question worth caring about. Of course
that variable may not encompass everything that concerned him
about income in the first place, and it may encompass other
things too; but then he is free to devise other variables that will
improve his coverage, for example, GNP. One could not object
that by so defining income our economist has departed from
ordinary usage to such an extent as to make his enquiry point-
less (which is what happened in the case of Pigs fly). On the
contrary what he has done is to isolate the point (or better: one
of the points) of asking after changes in income.
Now why can we not do the same in philosophy? Return to
the thesis that all words are names. Thus baldly stated it means
nothing: but why cant we simply specify the things that we might
mean by names in this context and then go on to test the result-
ing theory? In fact that was just what I tried very perfunctorily
to do: in chapter 1 I suggested three relatively precise meanings
for or consequences of that thesis all words have the same
function, all words are learnt by ostension, all words mean just
what they denote and tried to learn in light of Wittgensteins
own remarks which were true and which were false. Why doesnt
66
that count as a way of advancing philosophical theses and also as
a way of settling them?
One objection to this idea of philosophy is that it would
give many philosophical theses an empirical interpretation.
For instance, the sentence all words are names might get
interpreted as the empirical thesis that we learn all words by
ostensive definition. Or the thesis that space is necessarily three-
dimensional might get interpreted as a question that only an
empirically supported physical theory can answer. But (the
objector will continue) philosophical theses are not empirical
(and therefore typically contingent) theses but rather a priori
truths (if they are truths at all) discoverable from the armchair
(cf. PI 109).
But why think that? Only on a very narrow conception of
philosophy could one say that it made no empirical claims. To
take four examples that are very distant from one another in
time as well as doctrine: Aristotle (in De Anima), Berkeley (in the
New Theory of Vision), Marx (in Capital) and Russell (in The
Analysis of Mind) were all engaged in philosophical enquiries
that resulted in empirical (and simultaneously philosophical)
claims. Who is Wittgenstein to tell them that they are not doing
philosophy? Or again, why should we not say that fundamental
metaphysical questions concerning space and time are in fact the
province of physics as well as philosophy? Who is Wittgenstein
to tell us that Newtons theory, or Einsteins, were not advances
that did more for our philosophical understanding of space, time
and matter than centuries of purely a priori theorizing?
In connection with time Wittgenstein remarks:
Augustine says in the Confessions [What, then, is time? If
nobody asks me, I know well enough what it is; but if I am
asked what it is and try to explain, I am baffled.] This could
not be said about a question of natural science (What is
the specific gravity of hydrogen? for instance). Something
that we know when no one asks us, but no longer know when
we are supposed to give an account of it, is something that we
need to remind ourselves of. (PI 89)
We feel as if we had to penetrate phenomena: our investigation,
however, is directed not towards phenomena, but, as one might
READING THE TEXT
67
say, towards the possibilities of phenomena. We remind our-
selves, that is to say, of the kind of statement that we make
about phenomena. Thus Augustine recalls to mind the differ-
ent statements that are made about the duration, past, present
or future, of events. (PI 90)
I confess to finding this attitude incomprehensible. I myself
dont know what time is whether or not you ask me; but if I knew
its role in an empirically justified physical theory I should know
well enough what it was. And the same goes for space, and colour,
and the mind, and language.
It seems to me that the distance between Wittgensteins
conception of philosophy and this rival conception of it is so
great that perhaps they ought not to be reckoned rivals at all.
The truth is that here we have two completely different activities:
trying to answer very general questions about man and the
universe while frankly admitting that empirical findings will
often be relevant to this; and reminding ourselves a priori of the
ordinary uses of words to stop ourselves from getting enmeshed
in confusions. Both activities are legitimate; and in the end it
doesnt much matter which one gets to be called philosophy.
He himself once said:
If, e.g., we call our investigations philosophy, this title, on
the one hand, seems appropriate, on the other hand it has
certainly misled people. (One might say that the subject we
are dealing with is one of the heirs of the subject which used
to be called philosophy.) (BB 28)
But the subject that tries to find answers to specific philosophical
questions or at least to artificial formulations of them is
another heir, and it is the one that most philosophers in the
Western tradition pursue today. What justifies the continued study
of Wittgenstein for them is that intentionally or not much of
what he had to say sheds light upon their problems too.
2.4. PI 1347: The general form of the proposition
The picture of language that had held captive the author of the
Tractatus appeared at 4.5 of that book: The general form of a
proposition is: This is how things stand. Now this expression
68
This is how things stand is something that he had got from
everyday language and nowhere else (PI 134); and Wittgenstein
now applies the method of PI 116 by asking: what is its use in
everyday language?
Well, it is true that This is how things stand has a kind of
generality that enables it to stand in for any proposition: we
can always say, He said p; that is, he said that that is how things
stand; and what follows the semicolon here just repeats what
preceded it, whatever p is (PI 134b). But the fact that this sort
of generality attaches to This is how things stand does not
mean that its use at Tractatus 4.5 tells us anything about what a
proposition is. That is: if an alien who visited Earth and asked
to be shown which bits of language expressed propositions, it
would not do tell him They are the ones of the form This is
how things stand.
The reason it wouldnt do is that you need already to know
which bits of language express propositions or better, which
ones you are prepared to call propositions before you can say
which ones are of the form This is how things stand. It is as if
somebody were to explain what an Arabic numeral is by saying
that any Arabic numeral can be meaningfully substituted for the
2 in 121. It is true, but it doesnt specify any means of telling
whether anything is an Arabic numeral: rather all it says is that
given the Arabic numerals of which 2 is one the game that
we play with them licenses this substitution.
The same point applies to a related idea: that the essence of
a proposition its making a claim on reality is captured by
saying that it is the sort of thing that can be true or false
as opposed, say, to an exclamation, a sneeze or a hat. Again
Wittgenstein argues that this only seems to tell us anything. For
truth is not a special relation that propositions have to reality; it
is rather that p is true is just a way of saying whatever p does;
and p is false is just a way of saying what ~p says (PI 136). But
we cannot decide what belongs to that class by first assembling a
collection of words and then asking whether truth and falsity
somehow fit that assembly, as one jigsaw piece might fit another;
it is rather that we only predicate true and false of what we
call a proposition (ibid.).
So while both being true or false and being of the form: This
is how things stand do indeed specify necessary and sufficient
READING THE TEXT
69
conditions on propositionhood, they do not specify any feature
common to all propositions in virtue of which we call them
propositions; nor do they specify any feature that we could
identify prior to having identified the propositions themselves.
Two things follow from this: first, that proposition may well
express a family resemblance concept. At any rate the existence
of these necessary and sufficient conditions on propositionhood
does nothing to show otherwise. That is why Wittgenstein says:
But havent we got a concept of what a proposition is, of what
we take proposition to mean? Yes, just as we also have a
concept of what we mean by game. (PI 135)
The second thing that follows is that the supposed insight behind
the Tractatus was not really that but merely a move within our
grammar. It had seemed both profound and mysterious that
there are such things as propositions, that is, items that are
somehow able to do so much as lay a claim upon reality, to reach
right up to it and say this is so (PI 95). But saying that that
is what they do is just offering a paraphrase, not describing a
deep connection in reality. It is not that somehow we have
invested certain spoken or written items with a magical power
that nobody quite understands. Instead, it is just of the things
that we call propositions that we feel able to say that they make
a claim on reality, reach up to it, and the rest. Wittgensteins
later accusation of the solipsist can with greater justice be
directed at Tractatus 4.5: You interpret a grammatical move-
ment made by yourself as a quasi-physical phenomenon which
you are observing (PI 401).
SECTION 3. MEANING AND UNDERSTANDING
Wittgenstein pivots upon the notion of fitting discussed at PI
1367 to begin an extended discussion of understanding and
meaning that constitutes the heart of Philosophical Investigations.
Somebody might say that the use of a word fits its meaning;
and you would say this if you thought that the meaning of a
word is not the use itself (as proposed at PI 43) but rather some-
thing that guides the use. And that is surely a very natural thought
if you reflect upon what happens when you suddenly understand
something. For you seem to grasp it in a flash, and what we
70
grasp in this way is surely something different from the use
which is extended in time (PI 138). It is this idea of grasping
something in a flash that will preoccupy Wittgenstein in the next
50-odd sections.
3.1. The cube
But what does occur to you when you grasp something in a flash?
Wittgenstein first considers the suggestion that it is a mental
picture (PI 139c). Before discussing what he says about this it
is worth noting one distinctive and interesting feature of many
of his discussions of mental pictures or inner images: when-
ever he considers their function he thinks that it makes no
difference whether he considers inner or outer pictures.
We could perfectly well, for our purposes, replace every
process of imagining by a process of looking at an object
or by painting, drawing or modeling and every process of
speaking to oneself by speaking aloud or writing. (BB 4;
cf. PI 141b)
Thus what he says, for example, about the inner image of a cube
at PI 139 applies at least as well to a drawing of a cube on a piece
of paper. The strategy is really just a way of taking seriously
the picture of inner images or activities. For such inner objects,
states, events and processes are modelled upon their outer
counterparts; both the point and the ultimate downfall of these
models is that we are transposing an essentially third-person
situation into some kind of mental interior (Moran 2003: 2).
3.1.1. The picture and its projection
Wittgensteins actual discussion of the mental image proposal
centres upon the example of the word cube. Could it be that
when I suddenly come to understand the word cube, what
occurs to me is an inner picture of a cube? At PI 139 he does not
deny that one could call this picture the meaning of the word
cube, so that the meaning has come before my mind when I
understood cube for the first time. What he denies is that just
one pattern of use fits the picture in the way that just one jigsaw
piece fits another.
READING THE TEXT
71
Why not? Because if one use of the word cube can be said
to fit the picture that then came before my mind, so too could
another.
Perhaps you say: Its quite simple; if that picture occurs to
me and I point to a triangular prism for instance, and say it
is a cube, then this use of the word doesnt fit the picture.
But doesnt it fit? I have purposely so chosen the example that
it is quite easy to imagine a method of projection according
to which the picture does fit after all. (PI 139)
By a method of projection here Wittgenstein means something
like this: suppose that one had some method of drawing imagin-
ary projection lines connecting certain points on the surface of
the pictured cube to points on the surface of a physical object.
(Here one really has to think of a picture on paper, or even a
three-dimensional model.) If by some application of that method
the points of the pictured cube can be matched one-to-one with
points on the physical object, we shall apply the word cube to
the latter. And now his point is that nothing in the picture itself
fits any particular way of drawing such lines. For instance, if
I always drew lines from the vertices of the pictured cube to the
vertices of the physical object, then obviously Id only apply the
word cube to objects which themselves had eight vertices, so Id
apply the word to cubes
11
but not to triangular prisms. But what
if I practiced instead a method of drawing lines from each face
of the pictured cube to the vertices of the physical object? In that
case a triangular prism would fit the picture that I had associ-
ated with cube, for it has six vertices and a cube has six faces.
There is an obvious way to modify the idea that might seem to
get around this objection. This is to suppose that the inner
picture associated with cube depicts not only the cube itself but
also an appropriate method of projection. Thus imagine that the
inner picture associated with cube consisted of two pictured
cubes, together with lines of projection passing from the verti-
ces of the one to the vertices of the other. Wouldnt this rule out
my applying cube to triangular prisms?
But does this really get me any further? Cant I now imagine
different applications of this schema too? The answer is that
you can: nothing in this more complicated picture tells you how
72
to project it onto one or another application of the word cube
(PI 141). For instance, what is there in the picture that settles
how I am to compare the pictured lines of projection with the
imaginary ones that connect one of the pictured cubes to a
physical object? If each imaginary projection line is meant to
have the same origin as a pictured projection line (i.e. a vertex
of one of the cubes) then of course I will have a method of com-
parison that rules out my applying cube to a triangular prism.
But if each imaginary projection line is meant to have, say, an
origin that is equidistant from four coplanar origins of pictured
projection lines (i.e. a centre-point of one of the cubes faces),
then my method of comparison will not rule out my so applying
it. So this amendment does not get us anywhere: it looks as
though my use of the word cube cannot after all be said to fit
or not to fit a picture that comes before my mind.
It is worth mentioning that this position resembles that in the
Tractatus: or more precisely, what Wittgenstein says here about
pictures is close to what he says in the Tractatus about sentences
or propositional signs, which he thought were pictures. The mere
presence before ones inner eye (or outer eye) of a propositional
or pictorial sign cannot amount to an understanding of that
picture because the picture cannot for the reasons that we
have just rehearsed contain its own method of projection. But
it is also the view of the Tractatus that to understand the picture
in one way or another one must also perform a special kind of
psychological activity: one must think its sense; doing this is what
he calls projection (TLP 3.11). To do so is to have a thought, that
is, really to mean one thing rather than another by the picture.
A propositional sign, applied and thought out, is a thought
(TLP 3.5).
12
One has then turned the propositional sign into a
propositional symbol, that is, a sign together with its method of
application (TLP 3.32).
We shall return to the Tractatus idea, that understanding a
picture in some way is a psychological process that accompanies
it, at 3.3; and at 4.2.3 we shall consider the further Tractatus idea
that there must be a self or soul to carry out this process.
3.1.2. Extremely general facts of nature
People who have never been troubled by what is now troubling
us have always found it quite natural to say that one can grasp
READING THE TEXT
73
somebodys meaning in a flash, that the meaning of a word can
come before ones mind, and all the rest. Does Wittgenstein want
to prohibit these natural turns of phrase or if not to prohibit
them then at least to insist that they are no more literally true
than The sun sets?
The answer to both questions is no. In response to the
foregoing argument about projection he has the interlocutor
say: Well yes, but then maynt an application come before
my mind ? It may [Wittgenstein replies]: only we need to get
clearer about our application of this expression. Later in the
same section he continues:
Can there be a collision between picture and application? There
can, inasmuch as the picture makes us expect a different use,
because people in general apply this picture like this. (PI 141)
But if there can be a collision then there can also be a fit; so
contrary to the impression that PI 139 can give, Wittgenstein is
not arguing that in no sense of fit can ones use of the associ-
ated word be said to fit or fail to fit a picture in ones mind (or on
paper). What he is arguing against is a certain misconception of
what that fitting amounts to; the ordinary speaker is immune
from this criticism in so far as he does not have that illegitimate
sense in mind. Let us consider first what the illegitimate sense of
fitting is, and then what the legitimate sense of fitting is.
The illegitimate idea is that the picture somehow forces an
application upon us (PI 140), so that a use fits the picture not in
virtue of its being a normal response to the picture (i.e. a typical
way of using it) but independently of how we normally respond to
it: so that, for example, if the same picture were to occur to a
Martian in connection with the word cube, the same application
of that word would thereby be laid down for him too (cf. PI p. 46
n. 2 (b)/54 n. (b)). It is against this conception of fitting that
PI 139 was directed; and what it showed was that the application
of the word cube does not fit the associated picture in that rather
demanding sense, because there are other processes, besides
the one we originally thought of, which we should sometimes be
prepared to call applying the picture of a cube (PI 140).
But there is a second and less demanding sense in which a use
may be said to fit a picture. He describes this at PI 141 in the
74
words already quoted. The use that fits a picture is the use that we
expect on the basis of peoples actual application of the picture.
They might have applied it in another way or in no coherent way
at all. But as a matter of fact when you say to somebody Bring
me something looking like this while showing him a picture of a
cube, he will bring you a cube. That is why we can say truly that
somebody who diverges from this normal case somebody who
brings, for example, a triangular prism is using the word cube
in a way that does not fit that picture.
Wittgenstein had already illustrated this sense in which one
thing can fit another:
What about learning to determine the subject of a sentence
by means of the question Who or what . . .? Here, surely,
there is such a thing as the subjects fitting this question; for
otherwise how should we find out what the subject was by
means of the question? We find it out much as we find out
which letter of the alphabet comes after K by saying the
alphabet up to K to ourselves. Now in what sense does L fit
on to this series of letters? (PI 137)
L fits on to that series of letters in the sense that it comes next
in our alphabet, that is, we customarily so arrange the letters of
our alphabet that L immediately succeeds K. It is in the same
sense that the use can be said to fit the meaning that comes
before your mind when you say cube to yourself.
Let us now return to the issue that motivated this line of
enquiry. It was an objection to the idea that meaning is use: no,
it was said, meaning is not use but something prior to use that
your use fits. And the reason for this was supposed to be that
the meaning can come before your mind in a flash; but how can
the use, which is extended in time, come before your mind in
such a flash?
Wittgenstein seems to me to have taught us two things about
this. First, he is agreeing that the meaning can be a picture that
comes before your mind, but that this will only happen when the
application of the picture comes before your mind too. And
that is what will happen if you are prepared to use the picture as
it is typically used. So the application, which is extended in time,
can come before your mind after all. Understanding really is
READING THE TEXT
75
something that makes its appearance in a moment (PI 151). But
how can the use, which is extended in time, appear to you all at
once? I shall return to that in a moment.
The second thing that he has taught us is this: the picture is
something that the use can in a sense fit or fail to fit, but that is
only because it has the use it does, that is, because people will
(e.g.) fetch a cube when they are shown a picture of one and
asked to fetch what is in the picture. So independently of
the first point we can insist that use remains prior to meaning
in the order of explanation. A use does not fit a picture because
of the meaning that the picture already had; rather, a picture
has the meaning that it does because of the use that we typically
give it.
Let me conclude this section by briefly emphasizing two points
that are important for an understanding of Wittgensteins over-
all approach to problems of meaning and understanding. The first
point is that his discussion at PI 13941 illustrates the extent to
which the existence of a concept relies upon contingent regular-
ities. In particular the relation that fitting expresses would not
pick out anything at all if there happened not to be any typical
reaction to a picture. If on one day I responded to the request to
bring me something looking like this by bringing a triangular
prism, and on another day I brought a cube; if on those occasions
you responded in different ways too, and differently also from me;
if no pattern emerged over a longer period of time or over a larger
demographic cross-section: in that case we should attach no more
sense to saying that a pattern of use fits the picture than we do
to saying that it fits (e.g.) the weather.
As indicated by its exposition at PI 142 (see also p. 48/56n
and p. 195/230), this is one instance of a very widespread, and
for Wittgenstein very important, pattern of dependence of our
concepts that is, of our having words to express concepts upon
extremely general facts of nature (p. 48/56n). Other examples
in Wittgensteins work are as follows: the distinction between
certain and uncertain memories and its dependence upon the
facts about their actual reliability (OC 632); the dependence of
our arithmetical language upon general facts about the stability
of physical objects (RFM I-37); and, as we shall see, the depend-
ence of third-personal sensational ascriptions upon certain
regularities in our means of expressing the latter (4.2.5).
76
And the second point is that if these very general facts do
obtain, I dont think that Wittgenstein wants to stop anywhere
short of saying that the occurrence of a mental image on an
occasion can really be (on that occasion) the instantaneous onset
of understanding. But how can the application, which is
extended in time, occur to you all at once? The answer is that
for something to be the appearance before your mind of the
application, for a mental event to count as that, is something
that depends on very general facts that are spread out over time;
but that event need not itself be spread out over time. Compare:
a daub of paint on a canvas can be a painted smile; but its being
one depends upon what is painted elsewhere on the canvas. (In
another painting, that very daub would not have been a smile.)
And yet it does not follow that the daub itself is spread all over
the canvas.
3.2. Dispositions
Wittgenstein now turns to a new example. Suppose that we are
trying to teach a pupil to use the system of Arabic numerals by
showing him, and getting him to copy, initial segments of the
number series in that notation: 0, 1, 2, 3, . . . . So we write down
such sequences, and he writes them down after us, and when he
gets it wrong we correct him. In all probability he will if he is a
normal human being eventually master the system, that is, he
will acquire our ability to write down indefinitely long segments
of the number series.
How do we tell that he has mastered it? Well, by looking at
what he does: if he continues the series correctly enough times
without prompting, and if he can be made to see his errors
for what they are and also to correct them in that case we shall
say with reasonable confidence that he has mastered the system.
But his mastery of the system cannot, it seems, consist in his
having continued the series correctly up to, say, the hundredth
place on some particular occasion. For his mastery of it extends
beyond the hundredth place. Wittgenstein gives these two very
clear expressions of that idea:
Perhaps you will say here: to have got the system (or, again, to
understand it) cant consist in continuing the series up to this
or that number: that is only applying ones understanding.
READING THE TEXT
77
The understanding itself is a state which is the source of the
correct use. (PI 146)
Your idea, then, is that you know the application of the rule
of the series quite apart from remembering actual applica-
tions to particular numbers. And you will perhaps say: Of
course! For the series is infinite and the bit of it that I can
have developed finite. (PI 147)
Clearly the importance of this example is that it presents another
temptation to distinguish how you mean something from how
you use it, and more particularly: to distinguish ones under-
standing of a series from his actual development of it, because
the former is infinite (or indefinite) and the latter is finite; so
to understand the series must be a matter of being in a certain
sort of mental state from which all our acts spring as from a
reservoir (BB 146).
Now a reservoir is there whether or not one is drawing
water from it. And at PI 148 Wittgenstein asks whether our
knowledge of the application of the rule is present when it
is not being exercised or even thought about. And an obvious
reply would be that this knowledge or understanding is like
ones knowledge of the ABC, or the multiplication tables: such a
state is always present, because, it is thought, that state is a type
of disposition.
3.2.1. A-dispositions and B-dispositions
In discussing Wittgensteins treatment of this possibility at PI
149 we must distinguish two things that the word disposition
might mean. When one says that an object (say, a heated glass
bowl) has a disposition to crack if placed in cold water, one
might mean simply this: that if it were placed in cold water then
it would crack. A heated glass bowl could have that disposition
without ever being in cold water; but even if it is in cold water at
some time it will have the property even at those other times
when it is not. So for something to have a certain disposition in
this sense is just for it to satisfy a certain condition about what
would happen to it in certain hypothetical circumstances. Let us
call this sort of feature an A-disposition.
Now for all that we have said about A-dispositions, two objects
might differ only over them: there might, for example, be two glass
78
bowls of exactly the same temperature and molecular structure,
and in fact indiscernible in every respect except that if you were
to put them both in cold water, one but not the other would crack.
It would be very surprising if something like that did happen,
because an object that has an A-disposition normally has it in
virtue of some property that is not itself just a condition upon its
behaviour in hypothetical circumstances, but an actual feature
of the object. For instance, bits of salt have A-dispositions to
dissolve (i.e. if you were to put this salt in water it would form
a solution); but they do so in virtue of an underlying chemical
structure that explains their dissolution. This actual feature of an
object that underlies some one of its A-dispositions is often called
the basis of that A-disposition: here I shall call it a B-disposition.
3.2.2. The first objection: PI 149
Let us now return to Wittgensteins discussion of the idea that
understanding the series, knowing ones multiplication table,
and the like, are what he calls dispositions. He writes:
If one says that knowing the ABC is a state of the mind, one
is thinking of a state of a mental apparatus (perhaps of the
brain) by means of which we explain the manifestations of
that knowledge. Such a state is called a disposition. (PI 149)
He is here using disposition to denote the state that explains
ones tendency to utter the ABC in response to the stimulus of
a query; that is, he is using it to denote what I am calling a
B-disposition. The postulated state corresponds to the molecu-
lar structure of the salt, not its hypothetical behaviour in water.
And the proposal is that we identify the pupils knowledge of the
ABC (or understanding of the series) with the state of the mind
or brain that explains its manifestations in his behaviour.
Wittgenstein criticizes the idea as follows:
But there are objections to speaking of a state of the mind
here, inasmuch as there ought to be two different criteria
for such a state: a knowledge of the construction of the
apparatus, quite apart from what it does. (PI 149)
What could this mean?
READING THE TEXT
79
The difficulty seems to be that we cannot specify the state in
question except as whatever it is that explains the pupils reciting
the ABC correctly (or: whatever it is that explains his being able
to continue the series on any occasion). But why is that a diffi-
culty with this realization of the reservoir idea? Wittgenstein
does not say: but here is a proposal.
Suppose that we made an exhaustive list of all the B-dispositions
that could underlie an A-disposition towards the appropriate
behaviour in this case, the A-disposition that the pupil pos-
sesses if and only if he would, if asked, correctly recite the ABC.
These would be properties of his brain or mind: records of how
things actually are with him. The property of enjoying a red
after-image might be one such item; the property of having
neuron N firing at a rate R might be another. Then I am taking
the proposal to be that if knowledge of the ABC is a state of the
person then it is identical with some (possibly very complex)
combination of properties on that list: whichever combination,
that is, that best explains the A-disposition of interest.
Wittgensteins objection is that knowledge of the ABC cannot
be so identified because our sole criterion for settling whether
someone possesses that knowledge is the behavioural manifesta-
tion and not the state that explains it in normal cases. Suppose
that we found that among the human beings that we initially
test, all and only those in brain-state N
1
were able to recite the
ABC at will. Let us now see what happens if we identify the
understanding with that brain-state (exactly the same points
apply to any other B-dispositional state of the mind). Suppose
that we now find a group of people in whom brain-state N
1
does
not underlie their A-disposition to recite the ABC if asked:
people from this second group who are in N
1
frequently get it
wrong. Rather, among people of this group it is possession of
the brain-state N
2
that underlies it. I think we should all agree
that among people from this second group it is those in brain-
state N
2
and not those in brain-state N
1
who know their ABC.
But the people in N
1
in the second group are in the same state as
people from the first group who do know their ABC; knowledge
of the ABC cannot therefore be a matter of ones being in any
such B-dispositional state.
We may illustrate the point with a familiar contrast. We use
the terms red and blue to denote surfaces that are A-disposed
80
to produce impressions of red or blue upon normal observers
in a range of normal conditions. If the surfaces that actually
produce an impression of red (e.g. the surfaces of tomatoes)
had instead produced an impression of blue, those very surfaces
would have been not red, but blue (i.e. that is how actual people
should describe this circumstance). By contrast we use the term
water to denote whatever chemical actually falls from the sky as
rain, is clear and drinkable, and so on. That chemical is actually
H
2
O; but and here is the contrast with red and blue if that
very stuff, that very H
2
O had been opaque and poisonous, it
would still have been water. So in the case of water, we have
knowledge of the construction of the apparatus [from hydrogen
and oxygen], quite apart from what it does [fall from the sky, fill
lakes, etc.]. So being composed of water is a state of something
but being red is not; nor, for the same reason, is ones knowledge
of the ABC, or ones understanding of a word, or ones being
able to continue a numerical sequence.
Of course our use of understanding knowing the ABC, and so
on might have been different: those words might have resembled
water in just the respect that they actually resemble red. In that
case they would have been concepts that picked out states.
To see what sort of a difference that would involve, we need
only compare the psychological concepts that we apply to human
beings with the analogues that we apply to machines. A very
revealing because very straightforward example is the
concept of reading, by which Wittgenstein means not anything
involving linguistic understanding but such processes as transcrip-
tion and its converse of reading out loud; also sight-reading a
musical score and turning a typescript into handwriting. (PI
156a). More generally we may call reading any activity that
involves deriving a reproduction in one format of an original
that is in another (PI 162).
Let us consider the difference between (a) the grounds on
which we say that a child who is learning some such activity has
started to read and (b) the grounds on which we might say that a
machine that is under construction say, a pianola has started
to read. This example makes the difference very clear:
Human beings or creatures of some other kind are used by
us as reading-machines. They are trained for this purpose.
READING THE TEXT
81
The trainer says of some that they can already read, of others
that they cannot yet do so. Take the case of a pupil who has
so far not taken part in the training; if he is shewn a written
word he will sometimes produce some sort of sound, and here
and there it happens accidentally to be roughly right. A third
person hears this pupil on such an occasion and says He is
reading. But the teacher says: No, he isnt reading; that was
just an accident. But let us suppose that this pupil con-
tinues to react correctly to further words that are put before
him. After a while the teacher says: Now he can read! But
what of that first word? Is the teacher to say I was wrong,
and he did read it or: He only began really to read later
on? When did he begin to read? (PI 157)
It is the mark of the concept as applied to human beings that
the question has no answer: the only facts that are relevant to
settling it are facts about what the pupil did; and if they do not
settle it then nothing remains to be settled. (But surely it is a
matter of logic that either he was reading or he wasnt! I shall
discuss this objection at 4.2.7.)
That is not how things stand with a mechanical reading device,
for example, a pianola. This is a device that automatically plays
music if you feed into it a metal plate on which the score is
engraved. Suppose that while somebody is constructing it we
keep feeding it with such plates. Now and again it emits a note
that happens to be on the metal plate, but this is just an accident.
A third person hearing this says Now it is reading music. But
we say No, it isnt reading; that was just an accident. But let
us suppose that the machine continues to respond correctly to
further scores that are fed into it. After a while we say: Now it
can read! But what of that first note?
In this case, and in contrast to the case of the pupil, there is
an answer to the question whether it was reading that note. For
in the case of a pianola, it is not only what it does but also its
internal construction by which we settle whether or not it is
reading. If the right wires were connected when we fed it that
first note, it was reading even then; but if they were not, it was
not reading.
When we speak of a pianolas being able to read we are allowing
for the identification of this state in some other way than the
82
pianolas behaviour a knowledge of the construction of the
apparatus, quite apart from what it does. But when we say it of
a human being we do not allow for that possibility:
But in the case of the living reading-machine reading meant
reacting to written signs in such-and-such ways. This concept
was therefore quite independent of that of a mental or other
mechanism. (PI 157)
That is why being able to read is not really a type of mental state
at all. And the concepts of knowing the ABC and understand-
ing the series do not pick out states either. We cannot identify
such knowledge or understanding with any B-disposition.
3.2.3. The second objection: PI 158
At PI 158 Wittgenstein appears to allude to another reason to be
sceptical about identifying the ability to read with a B-disposition:
what he says there is clearly just as applicable to the cases of
understanding a series and knowing the ABC. He has just written
that The change when the pupil began to read was a change in
his behaviour; and it makes no sense here to speak of a first
word in his new state. Now he has the interlocutor reply:
But isnt that only because of our too slight acquaintance
with what goes on in the brain and the nervous system? If we
had a more accurate knowledge of these things we should see
what connexions were established by the training, and then
we should be able to say when we looked into his brain: Now
he has read this word, now the reading connexion has been
set up. And [W. now replies] it presumably must be like
that for otherwise how could we be so sure that there was
such a connexion? That it is so is presumably a priori or is it
only probable? And how probable is it? Now, ask yourself:
what do you know about these things? But if it is a priori,
that means that it is a form of account which is very convincing
to us. (PI 158)
It looks as though Wittgenstein is here raising a new objection to
the idea that reading and the same goes for understanding
might be a state of somebodys brain. The objection appears
READING THE TEXT
83
to be this: there is no reason to suppose a priori that any such
state underlies the behaviour characteristic of somebody who
can read in Wittgensteins sense. To put it in the terminology of
the preceding sections: there is no reason to suppose that any
neural B-dispositions underlie the behavioural A-dispositions
that interest us, for example, the disposition to convert a written
text into the pattern of sounds that we conventionally associate
with it.
I said at 3.2.1 that it would be very surprising if we found
A-dispositions that were not as it were underwritten by any
B-dispositions.
13
But Wittgenstein considered it a real possibility
in the psychological sphere. That is, he thought that the behavi-
oural dispositions by whose manifestations the pupil shows that
he can read, or has mastered a series, might have no explanation
in terms of his brain-states. A famous sequence of remarks from
his late writings on the philosophy of mind illustrates what
alternative he prefers; here are two extracts from it.
No supposition seems to me more natural than that there is
no process in the brain correlated with associating or with
thinking; so that it would be impossible to read off thought-
processes from brain processes. I mean this: if I talk or write
there is, I assume, a system of impulses going out from my
brain and correlated with my spoken or written thoughts. But
why should the system continue further in the direction of the
centre? Why should this order not proceed, so to speak, out
of chaos? (Z 608)
I saw this man years ago: now I have seen him again, I recog-
nize him, I remember his name. And why does there have to
be a cause of this remembering in my nervous system? Why
must something or other, whatever it may be, be stored up
there in any form? Why must a trace have been left behind?
Why should there not be a psychological regularity to which
no physiological regularity corresponds? If this upsets our
concept of causality then it is high time that it was upset.
(Z 610; cf. Z 609, 61113)
These remarks do nothing to refute the presumption in favour of
such a correlation; but what they do make clear is that it is really
84
an empirical matter whether or not any particular psychological
regularity corresponds to a physiological one.
Note the difference between this objection to the identification
of understanding with a state, and the objection at PI 149. The
objection of PI 149 as I am interpreting it here is that under-
standing is not a state or B-disposition of the mind or the brain,
in fact not a state at all, because given ones A-dispositions,
whether one understands is of necessity quite independent of
ones B-dispositions. This might be true even though ones actual
B-dispositions explained ones A-dispositions, just as the actual
facts about its surface reflectance might explain an objects
colour. Whereas the present objection is that there might not be
any B-dispositions in the brain to explain ones A-dispositions:
so no brain-state is even a candidate for being the state of under-
standing, quite independently of the first objection.
The effectiveness or otherwise of this second objection depends
so heavily upon the success and prospects of empirical research
into the brain that further comment upon it would be out of
place here. I shall only say that the first objection seems to me to
be quite effective enough to make this second one otiose.
3.3. Understanding and guidance
3.3.1. A dilemma
At PI 151 Wittgenstein introduces another example. A writes
down a series of numbers and B is trying to detect the principle
of the series, that is, the rule for any step in the series given all the
preceding ones. For instance if A writes down 1, 2, 3, . . . then
the principle of the series is that each element is got by adding 1
to the preceding one. Or if A writes down 1, 4, 9, 16, . . . then
the principle of the series is that for each natural number n, the
nth element of the series is the square of n. What interests
Wittgenstein is the question of what happens when B suddenly
grasps the principle of a series: the example he gives begins 1, 5,
11, 19, 29, . . .. And the reason for his interest is obvious: this
example, like those at PI 139 and PI 141, is a case in which an
episode of understanding seems to contain the whole future
development of the series (or the whole use of a word) within
itself and thereby to guide ones use.
Wittgenstein first imagines B experiencing the kind of thing
that many people actually observe to be going on when they
READING THE TEXT
85
suddenly understand the principle behind a series. For instance
B might simply try out certain formulas until the right one occurs
to him. Or he might experience a feeling of tension while
all sorts of vague thoughts go through his head. Then he asks
himself what the series of differences is; and on seeing that it is
4, 6, 8, 10 he says Now I can go on. Or perhaps he just has a
sensation as of a light quick intake of breath and continues
the series.
But to which of these types of experience are we referring
when we speak of the onset of Bs understanding? It looks
as though it cant be any of them. For instance, it cant be the
occurrence to B of the formula that is, the appearance in his
mind or visibly on paper of the sign a
n
= n
2
+ n 1 because it
is perfectly imaginable that the formula should occur to him and
that he should nevertheless not understand (PI 152). This form
of argument should be familiar from PI 34 (1.2) and PI 139
(3.1.1) and its plausibility is evident. So too is the possibility of
generalizing it to cover all of those other experiences that might
have occurred when B understood. So it looks as though the
onset of Bs understanding is not a type of experience.
We are trying to get hold of the mental process of under-
standing which seems to be hidden behind those coarser and
therefore more readily visible accompaniments. But we do
not succeed; or, rather, it does not get as far as a real attempt.
For even supposing I had found something that had hap-
pened in all those cases of understanding, why should it be
the understanding? (PI 153)
But if understanding is not an overt process, neither can it be
a covert process. For how can the process of understanding
have been hidden, when I said Now I understand because
I understood?! And if I say it is hidden then how do I know
what I have to look for? (ibid., cf. PI 147a).
Wittgenstein has therefore reached an intolerable dilemma.
On the one hand there appears to be no special experience
of understanding because any such experience could equally
accompany misunderstanding. But on the other hand if the
onset of understanding does not impinge upon my awareness as
such, what justifies Bs saying that he does understand? Surely
86
one is often enough justified in saying that, for instance in cases
when one has understood.
He resolves this tension in the very next section. The state-
ment Now I understand does not describe any type of process.
So awareness of any such type of process was never going to be
sufficient to justify anyones saying Now I understand! There
are instead particular circumstances, which justify me in saying
I can go on when the formula occurs to me (PI 154).
Now what exactly does this mean? In order to get clear on this
point Wittgenstein begins an extended discussion of reading in
the sense introduced at 3.2.2, and we turn now to this.
3.3.2. Reading, derivation and guidance
The discussion of reading at PI 15678 in fact serves three
purposes. One of these is to illustrate in connection with a rela-
tively simple psychological verb (read as against understand)
how such terms can fail to denote any type of process, conscious
or otherwise. The second is to illustrate how one can be none
the less justified in attributing that expression to oneself, and in
particular how it is the circumstances of the utterance that can
do the justifying.
But a third purpose has to do with the nature and not only the
simplicity of reading. It is natural to suppose that when one is
reading a text out loud one does not merely utter the words at
the same time as looking at the text; rather ones speech is derived
from or guided by the printed letters. This notion of derivation is
also a component of the more complex notion of understanding.
Someone who thought that the understanding was a sort of
reservoir from which our acts spring is going to think that as
well as a state of understanding there is an act of deriving ones
applications, for example, of a word from that state; or an experi-
ence of having that state guide ones applications of the word.
So by arguing that derivation and guidance do not themselves
denote any type of process, Wittgenstein will not only have
established the same for reading; he will also have further
undermined that misconception of understanding itself.
At PI 162 he suggests that we define reading as follows: one
reads when one derives the reproduction from the original. Now
what is this process of derivation? One can certainly give examples
of cases that seem clearly enough to involve it. For instance,
READING THE TEXT
87
suppose that we have taught someone a rule for turning a printed
text into a handwritten one. To this end we might give him a
table consisting of two columns, each row of which matches a
printed letter on the left with its cursive equivalent on the right.
And he shews that he is deriving his script from the printed
words by consulting the table (PI 162).
But it is possible to interpret the table in more than one way.
It might happen that instead of matching a printed letter with
the cursive equivalent to its right, the pupil matches a printed
letter with the cursive one that is one below it and to the right:
so he matches A with b, C with d, . . . and Z with a (i.e. in
accordance with the second schema drawn at PI 86). Surely he
would still count as deriving his script from the printed words,
although possibly not in the way that we had originally intended.
Or it might happen that his method evolves by the rule that if on
one day he writes (e.g.) c for B, then on the next day he writes
d for B, and so on. Or perhaps that he so evolves each time
he comes to the same letter, so that if one B in a text gets
transcribed as d, the next one gets transcribed as e.
The point of this sequence of examples is that at no point in it
do we seem to lose anything that was essential to the derivation
that was clearly going on in the first case. And yet it is clear
that the further along the sequence we go, the more closely the
procedures resemble random ones. And we can imagine extend-
ing the sequence so that eventually we get procedures that are
random; and yet no clear dividing line separates the procedures
that are random from those that involve derivation (PI 163).
Wittgenstein concludes from this that deriving is a family
resemblance concept; so too therefore is reading.
For certainly [the initial case] was a special case of deriving;
what is essential to deriving, however, was not hidden beneath
the surface of this case, but this surface was one case out of
the family of cases of deriving.
And in the same way we also use the word to read for a
family of cases. And in different circumstances we apply
different criteria for a persons reading. (PI 164)
Now that may all be true but it surely is not what the example
showed. The example alluded to a sequence of examples of
88
which the first definitely was a case of derivation and the last
definitely was not; the point of it was that each step seemed to
have as good a claim to count as an instance of derivation as
the preceding one. But this relies not on the family resemblance
character of the concept derive but on its vagueness: for all that
it tells us there may well be just one feature of derivations that
makes us call them that all it says is that if so then that feature
admits of borderline cases.
But even if he had shown that what we call derivations form
a family, that by itself would not be enough to show that there
was no one such type of process (which was one of the claims
about understanding that he was trying to illustrate). It would
only show that many processes that differed in a variety of other
ways could all belong to this one type (cf. the discussion of PI 33
on attending to the colour at 1.2).
In fact the example can seem doubly irrelevant. For the other
claim about understanding that Wittgenstein may be trying to
illustrate is that I am justified in saying that I have understood
the sequence, not because of any particular kind of experience
but because of the particular circumstances in which I say it. But
the sequence of examples does not illustrate that point either,
because they are all described from a third-person perspective:
they concern the circumstances in which we are justified in
saying that somebody else has derived the script from the text.
So even if it is true that we are justified, in saying of the pupil
that he has derived from the text, by the special circumstances
in which we say it, nothing follows as to whether he is similarly
justified in saying that of himself. For what justifies a third-
personal claim might be very different from what justifies
the corresponding first-personal one, as Wittgenstein himself
emphasizes elsewhere (4.2.5). But Wittgenstein does discuss
the first-personal perspective on reading very shortly afterwards,
so let us now turn to that.
Are we aware of a difference between a case of genuinely
reading a sequence of letters say the sequence A B O V E
and a case of uttering the sequence while looking at the letters
(as a man might do who could speak but not read English)? It
seems as though we are in particular it is tempting to suppose
that we are aware not just of the coincidence of our utterance
with our seeing of the corresponding letters; we are aware of the
READING THE TEXT
89
latter somehow guiding the former (PI 170b), or as Wittgenstein
also puts it, a feeling of causal connection (PI 169a) or of
influence (PI 169c):
We imagine that a feeling enables us to perceive as it were a
connecting mechanism between the look of the word and the
sound that we utter. For when I speak of the experiences
of being influenced, of causal connexion, of being guided,
that is really meant to imply that I as it were feel the move-
ment of the lever which connects seeing the letters with
speaking. (PI 170)
But do we have any such feeling?
Against the idea that we have any feeling of causal connection
Wittgenstein writes:
Causation is surely something established by experiments, by
observing a regular concomitance of events for example. So
how could I say that I felt something that is established by
experiment? (PI 169)
Without supplementation this argument can easily seem most
implausible. The fact that we can establish something by one
sort of procedure (an experiment) doesnt exclude its being
detected by another (feeling it). We can establish that some-
body has an upset stomach by performing certain experiments;
but he can tell it in another way too by feeling it. So of course
we can say that you feel something that may also be established
by experiment. Where is the difficulty?
14
One way to make the argument look more plausible would be
to emphasize the point about the regular concomitance of events.
On a regularity account (and on many other accounts) of what
it is, the obtaining of a causal connection has consequences
that are spread over time and space, whereas one cannot feel
something of that extent but only how things are with one
here and now. So one cannot feel the obtaining of a causal con-
nection because it is spread over space and time in a way that the
contents of ones feeling cannot be. Of course one can see some
things and events at distant times and places, for example, stars
and sunsets; but that is not the same thing as seeing the whole
90
panoply of correlations (past and future) that we are supposing
a causal connection between two types of event to involve.
But if that is the argument then it is implausible, if only
because it appears to confuse the content of an experience with
its cause. And in fact there is a sort of experience that can easily
seem to be of just such an extended regularity. What I have in
mind is this (the example appears at RFM IV-39): if by looking
at a token of the word Bismarck one sees that it has eight letters,
what one also sees is that that type of word has eight letters,
that is, that every one of its tokens, past, present and future and
wherever they may be, has eight letters. Why does this not count
as experiencing a regularity? Similarly then, when one reads a
new word out loud, why could it not be part of ones experience
that this type of sound goes with that type of printed pattern?
Still, Wittgenstein has a further and better argument against
the idea that what justifies one in saying that one was guided is a
type of experience that distinguishes the presence of guidance
from its absence. Like many of his best arguments it appears
to involve phenomenological introspection. It is that when you
observe what actually happens in such cases you will notice that
there is in fact no experience at all that could be the experience
of guidance:
Make some arbitrary doodle on a bit of paper And now
make a copy next to it, let yourself be guided by it. I should
like to say: Sure enough, I was guided here. But as for what
was characteristic in what happened if I say what happened,
I no longer find it characteristic.
But now notice this: while I am being guided everything is
quite simple, I notice nothing special; but afterwards, when I
ask myself what it was that happened, it seems to have been
something indescribable. Afterwards no description satisfies
me. Its as if I couldnt believe that I merely looked, made
such-and-such a face, and drew a line. But dont I remember
anything else? No . . . (PI 175)
I think we should take this as making an introspective case for
the empirical conclusion that no such experience of guidance
exists; taken as such it seems to me as decisive as Humes argument
against any special experience of a self (Treatise I.iv.6).
READING THE TEXT
91
It is worth noting that in addition to this rather straightfor-
ward point Wittgenstein appears to have a subtler one in mind.
He says:
When I look back on the experience I have the feeling that
what is essential about it is an experience of being influenced,
of a connexion as opposed to any mere simultaneity of
phenomena: but at the same time I should not be willing to
call any experienced phenomenon the experience of being
influenced. (This contains the germ of the idea that the will
is not a phenomenon.) (PI 176)
This argument is not simply a report that he has looked within
himself for an experience of the because and come up short,
just as one might fail, for example, to find an experience of
acidity when tasting wine. It is rather a claim that nothing could
be that experience.
But why not? In Philosophical Investigations he does not
say. One reason might be that to experience the because is
to experience a regularity, and he might have thought for the
reason already criticized that nothing could be an experience of
a regularity. Another line of thought is this: he had once thought
that the content of both experience and linguistic description
had to be contingent: Whatever we see could be other than it is.
Whatever we can describe at all could be other than it is (TLP
5.634). Now the traditional conception of causation involves neces-
sary connection: if an event A is a sufficient cause for another
event B then A could not have occurred without B. Putting these
two ideas together we reach the conclusion that to experience the
because is to experience a necessary connection, that is, just the
sort of experience that the Tractatus had ruled out.
Whether or not it was what Wittgenstein had in mind, this line
of thought faces the same objection as its predecessor: when
you see that the word Bismarck has eight letters, why are you
not seeing not only that things are but also that they must be a
certain way, and hence experiencing a necessary connection as
well as a regularity? And this is to say nothing of the dubious
association between causation and necessary connection. But in
any case the straightforward introspective argument appears to
me already to have established the point at which PI 176 was
92
aimed that is, that there is no experience of guidance so
I shall leave the matter there.
But if there is no such experience then what was the source
of our mistake? Why did we think that there must be one?
Wittgensteins answer at this point sounds rather Kantian. It is
as though we impose the concept of guidance upon a stream
of sensational input in which it does not originate:
I should like to say I experience the because. Not because
I remember such an experience, but because when I reflect
on what I experience in such a case I look at it through the
medium of the concept because (or influence or cause or
connexion). (PI 177)
But what does it really mean to look at something through the
medium of the concept?
Here Wittgenstein does not say; but I think that what he means
is not at all Kantian. Instead, looking at it through the medium
of the concept means: failing to reflect that guidancewhich he
runs together with influence and other such words at PI 175b
expresses a family resemblance concept. If one thinks that there
must be some one thing that makes us apply the concept guid-
ance in all the cases where we do apply it, then of course one is
bound to postulate a typical experience of guidance even if most
of the time you do not remember one. Wittgensteins discussion
is aimed at loosening the grip of that preconception: thus in
addition to the direct argument just cited he gives a list of vari-
ous activities that count as guidance (PI 172). The upshot is
supposed to be that not only do we no longer take guidance to
be a unitary kind of mental experience at all; but also, we are
free of the conceptual prejudice that had attracted us to that
view, as it were in spite of ourselves.
But then if it does not describe anything you experience, what
does justify an assertion that one is reading, that one has derived
the script from the printed text, or that one understands the
principle behind the series? Wittgenstein doesnt although
he very well could use the example of guidance to illustrate the
point. Instead he returns to the example that PI 151 introduced.
Recall that A writes down a series of numbers and B is trying to
detect the principle of the series, that is, the rule for any step in
READING THE TEXT
93
the series given all the preceding ones. For example, if A writes
down 1, 2, 3, . . . then the principle of the series is that each
element is got by adding 1 to the preceding element. What
justifies B in saying Now I understand? Here is the answer:
It is clear that we should not say B has the right to say the
words Now I know how to go on, just because he thought
of the formula unless experience shewed that there was a
connexion between thinking of the formula saying it, writing
it down and actually continuing the series. And obviously
such a connexion does exist. (PI 179)
What experience shews is not some necessary connection but only
a statistical association. It is because of our experience that most of
the time, when B or somebody like him has the thinking of the
formula experience,
15
he continues the series correctly, we are right
to call B justified in saying that he knows how to go on.
Now to say what justifies an assertion is not to say what it
describes, and Wittgenstein immediately blocks that interpreta-
tion of him. In particular then, he denies that Now I understand
is Bs way of saying Now I am having an experience that I know
empirically to lead to the continuation of the series. For as we
saw in connection with guidance, B might be justified in saying
Now I know how to go on even if nothing at all occurs in his
mind (PI 179c). And it would be quite misleading to call Bs words
in that case a description of either a mental state or anything else.
One might rather call them a signal; and we judge whether it
was rightly employed by what he goes on to do (PI 180).
Nor, from Bs own point of view, is his assertion Now I know
how to go on or Now I understand in any sense derived from
or otherwise supported by his awareness that he can usually
continue the series in circumstances in which he typically says
that. For one doesnt argue like this: I have always been able to
continue the series when the formula has occurred to me in the
past, so it will happen now too. If anything, an experience such
as the occurrence of the formula to B is the cause and not the
ground of his certainty that he can continue the series (PI 325).
We might illustrate the point with the following analogy.
A dog that has repeatedly got its dinner upon the sounding of
a bell, and only then, will soon acquire the habit of salivating
94
whenever it hears one. We might call the salivation justified,
not because it describes an empirical regularity connecting the
bell and the dinner, or even because the dog knows but cannot
speak of such a regularity it may show no other signs of expecting
dinner but simply because the regularity exists. Similarly, it is
an extremely general fact of nature that when B and people like
him say Now I understand the series, they are in fact able to
continue the series; and it is the existence of this regularity
whether or not B believes in it that justifies B.
But then B may not be able to cite any reason for saying that he
can continue the series: so is he really justified in saying that he
understands? I think Wittgenstein would regard that question as
one of those cases in which ordinary language goes on philosophi-
cal holiday (PI 38). On the ordinary sense, that is, the ordinary use
of the word justified, B is justified. A good ground is one that
looks like this (PI 486). I shall return to this point at 3.4.3.
The feeling can persist that when ones understanding of the
series makes its first appearance, it isnt something that just happens
typically to precede its successful development, that is, as a matter
of empirical fact. Instead the whole series seems contained within
ones momentary understanding in some other and more instanta-
neous way. The same feeling arises when we contemplate the giving
of an order, say, to continue a series: it is as though the whole
series were somehow contained within that order, or whatever lay
behind it, in a more intimate way than simply its being typically
associated with some and not other patterns of behaviour. After
all, one can order someone to continue a particular series, and to
mean that very series, whether or not he then carries it out.
We are as it were surprised, not at anyones knowing the
future, but at his being able to prophesy at all (right or wrong).
As if the mere prophecy, no matter whether true or false, fore-
shadowed the future . . . (PI 461). It is the attempt to get clear
on this foreshadowing that ushers in the most celebrated and
controversial material in the book.
3.4. The paradox
3.4.1. The possibility of deviance
Let us return to the example of PI 143 (introduced at the start
of 3.2) and imagine that the pupil has mastered the series of
natural numbers. And let us now write +n for the series
READING THE TEXT
95
0, n, 2n, . . ., so that +2 abbreviates the series 0, 2, 4, 6, . . .
and +3 abbreviates the series 0, 3, 6, 9, . . .. And suppose that
the pupil appears to understand the series +2 perfectly well
because he has passed all our tests of his ability to write down
initial segments of it. Suppose finally that in these tests the pupil
was never (as it happens) asked to extend the series +2 (or any
other series) beyond 1000.
Now we ask him to continue the series +2 beyond 1000 and,
instead of 1000, 1002, 1004, 1006 he writes 1000, 1004, 1008,
1012. Then we shall say that he has got the series wrong. More-
over, what he has done is to write down something contrary to
what we meant when we gave the order. And now it looks as
though, for that to be the case, we shall have to say that there was
something going on when we meant it that anticipated all of the
elements of +2 beyond 1000, not only 1002, 1004, 1006, but
also 1866, 1868, 1870 and countless others that equally never
occurred to us at the time of giving the order.
But there may have been nothing in the order or in our minds
at the time that could possibly have shown him that. If we say,
You were meant to add two: look how you began the series! he
may reply, Yes, isnt it right? I thought that was how I was meant
to do it!; if we say, You are supposed to do the same thing at
every stage! he may reply, Yes, wasnt I doing the same thing?;
if we show him a table correlating each step with the number to
be written at that step, he may read it in some deviant manner
(cf. those at PI 86, PI 163) that legitimates his own practice; and
if we supplement the table with arrows, he may so interpret them
as to do the same. Here it looks as if the order were beginning
to stammer. As if the signs were precariously trying to evoke
understanding in [him] (PI 433).
The point can now be made very general. Any stage of the
series +2 or of any other sequence that he has not encoun-
tered before might be the starting-point for some deviance on
the part of a perfectly rational man who has received exactly the
same training as those from whose continuation he is deviating.
Thus there might be a pupil who began the series +3 0, 3,
9, . . . but on passing 57 wrote 66, 75, 84, that is, appending the
continuation of what we should call the series +9.
And there is no reason to restrict the point to numerical
sequences. A similar deviancy might enter into any novel
96
application by a rational person of any word: there might, for
example, be a rational person who has been trained as we have
in the use of the word cube and who has so far applied it in
just the way that we should expect, but who now applies it not
to newly encountered cubes but to newly encountered triangu-
lar prisms.
Finally, there is no reason for the deviancy to begin with a
novel application: somebody might use the word green for green
objects that he has observed hitherto and blue ones that he
observes from now on, including those that he has already seen
and classified. (He will say that they have turned green if they
were blue all along; he will say that they have remained green if
they have changed from green to blue between what we should
call the two classificatory eras.) In short, for any application of
any word, a man who has been trained just as we have been
trained might dispute this application with us, however obvious
it may seem; and there need be nothing we can point to in our
common prior training or in the words agreed prior use or in
anything else of which we are both aware to which he will
acknowledge that his deviant application is unfaithful.
This means that we were wrong after all to suppose that the
rule or order contains its future applications in some special
way; that is, in some way of which the merely statistical connec-
tion described at PI 179 falls short. For if it did, then we ought
to be able to point to something in it that would convince any
deviant of his error. But there is nothing in it that might do that:
given anything that was going on at the time of giving the order,
we can imagine a rational pupil being perfectly aware of it and
yet persisting in his deviance.
Similarly neither a sign nor any mental accompaniment of it
carries within itself the particular meaning that it seems to intim-
ate to us. Rather, what makes it a sign of this rather than that is
how we normally respond to it. That is what justifies my respond-
ing to it in one way rather than another on this occasion.
Let me ask this: what has the expression of a rule say
a sign-post got to do with my actions? What sort of
connexion is there here? Well, perhaps this one: I have
been trained to react to this sign in a particular way, and
now I do so react to it.
READING THE TEXT
97
But that is only to give a causal connexion; to tell how it
has come about that we now go by the sign-post; not what
this going-by-the-sign really consists in. On the contrary;
I have further indicated that a person goes by a sign-post only
in so far as there exists a regular use of sign-posts, a custom.
(PI 198)
Wittgensteins treatment of these examples thus uproots the
deep and important misconception that meaning and under-
standing are mental states or occurrences that somehow contain
within themselves some particular pattern of usage. But it has
seemed to some that Wittgenstein had intended these examples
to point at something even more radical: namely, that under-
standing and meaning are themselves an illusion. I turn now to
this reading.
3.4.2. Meaning-facts
The possibility, that a rational deviant of the sort described
might yet have missed nothing of what went on when we gave
the order +2, raises the question: how then did the person who
gave the order manage to mean the sequence +2 rather than
anything else? Indeed when he or you, or I develops any
sequence, or for that matter obeys any rule at all, in what sense
can we be credited with fidelity to what we meant all along?
Shouldnt we rather compare the announcement of a rule say,
the formula of a series with the first stroke of paint on a blank
canvas? That is, it is an illusion to suppose that any rule for their
use, or indeed any prior use at all, ever distinguishes correct from
incorrect patterns of use of words or of anything else. We still
have to choose at every stage whether to continue the pattern in
one way or another.
Wittgenstein can certainly seem to be denying that a rule ever
can lay down an application in this way. On the contrast between
thinking of understanding as a kind of insight and thinking of
it as a kind of choice, he writes:
What you are saying, then, comes to this: a new insight
intuition is needed at every step to carry out the order +n
correctly. To carry it out correctly! How is it decided what
is the right step to take at any particular stage? . . . It would
98
almost be more correct to say, not that an intuition was
needed at every stage, but that a new decision was needed at
every stage. (PI 186)
And shortly afterwards he makes explicit the general form of the
problem that provoked this talk of decision:
But how [the interlocutor asks] can a rule shew me what
I have to do at this point? Whatever I do is, on some inter-
pretation, in accord with the rule. (PI 198)
This is the famous rule-following paradox that dominates
Kripkes notorious reading of Wittgenstein.
Kripkes account of the paradox is not a thesis about what
Wittgenstein actually meant; instead it presents thoughts that
Wittgensteins argument provoked in Kripke (Kripke 1982: 5).
But (quite apart from its intrinsic interest) it is still worth
examining in order clearly to distinguish Wittgensteins own
line from it.
According to Kripke (pp. 9, 21) the considerations that I have
just mentioned (about deviant usage) give rise to a question that
I shall put in terms of Wittgensteins example, not his: what fact
was there about me, at the time of giving the order to continue
the sequence +2 beyond 1000, in virtue of which I meant all
along that its recipient was to write, say, 1868 after 1866, or
10004 after 10002 (or any other pair that I did not explicitly
mention at the time of giving the order)? In the course of a very
penetrating discussion Kripke examines and rejects certain
candidate facts: facts about my verbal training (pp. 1516), facts
about my dispositions to respond in one way or another
(pp. 2237), facts about the relative simplicity of the competing
interpretations (pp. 389), facts about my introspectible mental
state (pp. 4150) and sui generis facts about my mind (pp. 514).
He concludes that there is no such fact. Therefore a rule cannot
tell me what to do, and there is no such thing as meaning any-
thing by a rule. That is Kripkes interpretation of the following
celebrated passage:
This was our paradox: no course of action could be determined
by a rule, because any course of action can be made out to
READING THE TEXT
99
accord with the rule. The answer was: if any action can be
made out to accord with the rule, then it can also be made out
to conflict with it. And so there would be neither accord nor
conflict here. (PI 201)
Kripke takes this to be Wittgensteins conclusion and he draws
the obvious and dramatic corollary. There can be no such thing
as meaning anything by any word. Each new application we
make is a leap in the dark; any present intention could be
interpreted so as to accord with anything we may choose to do.
So there can be neither accord, nor conflict (p. 55).
Hence we do not speak of any facts when we say, for example:
You meant the series +2 when you ordered the pupil to write
down the series. But according to Kripke, that does not mean
that we should drop such talk altogether. Rather, we should note
that even if a sentence like that describes no facts, it can still
possess assertibility conditions. These are rules defeasibly licens-
ing its assertion under certain conditions. Those conditions are
essentially comparative: you are (defeasibly) licensed to assert,
for example, Jones means addition by his sign + if his use of
that sign in the cases where he has applied it matches your use
of + or addition (pp. 8990). Talk about meaning is therefore
based essentially on a comparison in a way that inverts common-
sense. It is not that you and Jones both use + in the same way
because you have both attached the same meaning to it; rather,
we are entitled to say that you have attached the same meaning
to + if you both use it in roughly the same way. It is in this
essentially comparative dimension of meaning-attributions that
Kripke locates the kernel of Wittgensteins private language
argument (pp. 87113).
Before assessing this interpretation of Wittgenstein it is worth
mentioning that it offers us a deeper account of his response to
the difficulties concerning vagueness that I mentioned at 2.2.3 (a)
(ii). Recall the difficulty: the rules governing our use of vague
predicates may well make it possible to construct a finite sequence
of objects in such a way that some such predicate definitely
applies to the first and not to the last object in the sequence,
while simultaneously insisting that one apply the predicate to an
object if and only if one applies it to whatever lies next to it in
the sequence. This would show the rules to be inconsistent; and
100
the pragmatic attitude that I there attributed to Wittgenstein was
that this does not make the concept unusable so long as our use
of the rules never in fact confronts us with a contradiction.
Kripkes interpretation reveals the possibility of a different
response. For if the rules for using a predicate in fact do not
determine any one pattern of use rather than any other, then
until we actually find a real Sorites sequence and by making suc-
cessive applications of the rules arrive at the contradiction, that
contradiction is not even implicit in those rules. Thus the rules
governing a vague predicate would be as harmless as because
as impotent as those governing any other expression.
Turning back to Kripke, I think it is fairly certain that
Wittgenstein is not arguing in the way that Kripke presents him.
One fairly clear indication of this is the remark at PI 136a: if
as it says p is true is just another way to say p, then so is
p expresses a fact or It is a fact that p. But then the
propriety of He meant +2 stands or falls with that of It is
a fact that he meant +2; and so once we are entitled to say
the former even if only in accordance with Kripkes own
assertibility conditions there is simply no room for denying
the latter. In short Wittgensteins own theory of truth commits
him to rejecting any sort of distinction of the sort that Kripke
wants to draw, between those assertoric sentences that really do,
and those that only seem to, state the facts.
16
Note in addition that according to Kripkes reading there is
no conflict between what occurs to one when one understands
a word and whatever one then goes on to do with it. But
Wittgenstein denies just this. Recall his discussion of the
idea that what comes before ones mind when one understands
a word is a picture. He writes: Can there be a collision between
picture and application? There can . . . (PI 141c). Or again:
But I already knew, at the time when I gave the order, that he
ought to write 1002 after 1000. Certainly; and you can also
say you meant it then . . . (PI 187)
So certainly Wittgenstein, unlike Kripke, does not deny that you
mean or intend one thing and not another when you state a rule;
nor does he deny that the auditors action his application of a
word, for instance, or his continuation of a sequence can
READING THE TEXT
101
accord with what was meant, or that it can conflict with what
was meant. And as we have seen, what justifies our saying or
your saying that you meant one thing rather than another, is
the customary use of the expression.
3.4.3. Something animal
This line of thought faces an objection that may be introduced by
means of the second paragraph of PI 201, where Wittgenstein
replies to Kripkean scepticism about the possibility of meaning
anything:
It can be seen that there is a misunderstanding here from the
mere fact that in the course of our argument we give one
interpretation after another; as if each one contented us at
least for a moment, until we thought of yet another standing
behind it. What this shews is that there is a way of grasping
a rule which is not an interpretation, but which is exhibited
in what we call obeying the rule and going against it in
actual cases.
What he means by giving one interpretation after another is
different descriptions of ways that one could take the rule: so,
for instance, one interpretation of +2 would be Add 2 to
each number in the sequence to get the next one. Another inter-
pretation of it would be Add 2 to each number less than 1000
to get the next one. Add 4 to each number greater than or equal
to 1000 to get the next one. And of course we can think of indef-
initely many others. (Do not confuse this use of interpretation
with that at PI 34.)
But the fact that we can think of all these interpretations does
not show that we never understood the rule in one way rather
than another, or that it never tells us what to do. The rule does
tell us what to do, because there is a way that we do in fact
respond to the rule. In general we do in fact respond to the rule
by adding 2 to each number to get the next one; we do in fact
say that somebody who responds otherwise is going against it,
and so on; and that is why the rule tells us to do that and not
something else.
It is important to be clear about what Wittgenstein does mean,
and what he does not mean, by what we call obeying the rule
102
and going against it in actual cases. He is not saying that what
you mean by a rule on an occasion is altogether settled by what
you call obeying the rule, or by what the intended auditor calls
that, in this case on the first alternative it would be possible to
mean +2 by bububu (a nonsense expression, cf. PI p. 16n/18n);
on the second alternative it would be impossible ever to go
against a rule.
What he is saying is rather that what exhibits ones grasp of
a rule is what is generally called obeying the rule and going
against it, that is, by myself at other times or by other people,
possibly at the same time, but in any case in ordinary circum-
stances. Suppose that by my utterance of +2 I meant that he
ought to write, for example, 1868 after 1866 but that nothing
in my inner or outer behaviour, or in his, settled this. (I never
mentioned those numbers, and I never thought of them, and
neither did he.) Then what exhibits the fact that I did mean him
to write 1868 after 1866 is the fact that other people have
continued the series in that way. It is an instance of what they
call obeying the rule; just as writing, say, 1870 immediately
after 1866 is an instance of what they call going against it;
and they do so in ordinary circumstances, for instance, in the
classroom.
You will immediately object that when I said +2 I meant: not
only certain segments of the sequence that have not occurred to
me but which have occurred to others. I also meant segments
of the sequence that never have occurred to me or to anyone
else. For instance, when I said +2 I meant that he should write
10004 after 10002, but even if nobody had ever written down
or thought of those numbers I should still have meant just what
I actually did mean; and so I should even in that situation have
meant that he should write 10004 after 10002. So how can what
we call obeying the rule and going against it in actual cases
exhibit my having meant just that by +2?
The answer is that you can exhibit something that points
beyond the examples that you explicitly mention or that anyone
has ever mentioned. We can do this by the use of and so on or
ellipsis (PI 208g, NB 49). Teaching by example may also be an
exhibition of that sort. For instance, if I wish to explain the word
game to somebody I may describe a few games and then I might
add This and similar things are called games (PI 69). Here the
READING THE TEXT
103
expression and similar things is performing the function that
might in the explanation of other words have been performed by
and so on or by . . .. Or again I may explain the word same to
someone by example.
In the course of this teaching I shall shew him the same
colours, the same lengths, the same shapes, I shall make
him find them and produce them, and so on. I shall,
for instance, get him to continue an ornamental pattern
uniformly when told to do so. And also to continue pro-
gressions. And so, for example, when given: . .. ... to go
on: .... ..... ...... . (PI 208)
As well as being one itself this passage describes two further
exhibitions of meaning by example: one exhibits what one
means by same by showing the pupil examples of same F;
and then he exhibits what he has understood by giving other
examples. It is in this and similar ways, then, that ones meaning
may be exhibited in what we call obeying the rule and going
against it in actual cases.
But mustnt this exhibition of what you meant still miss out
on something? Let us suppose that up to this point nobody has
ever concerned themselves with what number follows 1866 in
the sequence +2. But when you gave the order +2 you certainly
meant that the pupil should write 1868 after 1866. Wittgenstein
says that that is what our judgement about actual cases
exhibits; this is exhibited to me, or to a competent pupil, not
by your having mentioned that particular example, but by your
having mentioned other examples: say, by your having written
998 after 996, or by your having written 212 after 210, and
so on. But the complaint is: how can you have exhibited that
you meant just that by your talk of other examples? As the
interlocutor puts it: [D]o you really explain to the other person
what you yourself understand? Dont you get him to guess the
essential thing? You give him examples, but he has to guess
their drift, to guess your intention (PI 210).
The answer to this objection is first of all to expand it. For if
it applies at all it does not only apply in cases where the sequence
that one is supposed to have in mind reaches beyond all of the
segments of it that one has actually mentioned, written down or
104
otherwise made explicit (if only to oneself).
17
Suppose, for instance,
that you did somehow write down in a legible fashion the entirety
of the sequence +2, so that nobody who read it could be in any
doubt that you had 1868 after 1866. Your pupil would still be
in as much need of guesswork as before. For he would still have
just as much need to guess at the method of projection from
your script to his: what transformation is to count as writing
down the same as what you have written down is it the numbers
that you have written down, or what you get by adding 0 to the
first, 2 to the second, 4 to the third . . .?
That is wrong; where we have gone wrong is through applying
an absurdly severe conception of what exhibiting ones meaning
involves. What the objector is insisting upon is that in order
to exhibit that I mean one thing I must eliminate all possible
alternatives because my hearer is otherwise reduced to guessing
between the uneliminated ones. But according to Wittgenstein it
is not really guessing to which one is reduced. The passage just
quoted from PI 210 continues:
Any explanation which I can give myself I give to him too.
He guesses what I intend would mean: various interpretations
of my explanation come to his mind, and he lights on one
of them. So in this case he could ask; and I could and should
answer him.
Here interpretration presumably has the sense recommended
for it at PI 201, that is, the substitution of one expression of
the rule for another. So to say that somebody guesses what you
meant means that various alternative descriptions of rules that
you might mean actually occur to him. Even if indefinitely many
continuations are compatible with the examples that you have
shown him (as will always be the case), he is not guessing unless
a number of them occur to him and he chooses from among
them. So if in fact no interpretation occurs to him, if instead he
merely continues the sequence as you wished, then he has not
had to guess.
Now this sense of guess almost certainly preserves something
that belongs to the everyday use of the word guess but which
the interlocutors more philosophical usage in PI 210 appears
to have discarded: namely, the intuition that when in ordinary
READING THE TEXT
105
cases the pupil learns by example, he is not guessing. In this
case our language has gone on what Wittgenstein would call a
holiday (PI 38). We feel as thought we had made the exciting
discovery that one can only ever guess at anothers meaning. But
that is not an everyday use of the word guess.
But although unfaithful to the intuitive extension of the word
guess, the interlocutors use of it at PI 210 does legitimately
highlight something in its intuitive intension. This is the idea that
one is guessing if one does not reach ones conclusion through a
rational process. And it is reasonable to describe many learning
situations as being of that sort. For instance, suppose again that
the pupil is asked to continue the sequence +2 in the circum-
stance that nobody has ever developed it past 1866. Then what
explains his now writing 1868 rather than, for example, 1870
after 1866? He cannot point to any reason; or rather, if he
points to his teaching and the order itself as a reason, he will not
be able to say why he takes that teaching as a reason for writing
1868 rather than as a reason for writing 1870.
Now nothing is gained by insisting that that the pupil is or
that he is not guessing in such cases. What matters is what
Wittgenstein agrees upon: that we do eventually run up against
this giving out of reasons. And he seeks quite rightly to
downplay its practical significance.
How can he know how he is to continue a pattern by himself
whatever instruction you give him? Well, how do I know? If
that means Have I reasons? the answer is: my reasons will
soon give out. And then I shall act, without reasons. (PI 211)
When someone whom I am afraid of orders me to continue
the series, I act quickly, with perfect certainty, and the lack of
reasons does not trouble me. (PI 212)
Now it is certainly possible to overreact to this point, that the
pupil does not have a reason for supposing you to mean that he
should write, for example, 1868 after 1866. The overreaction
would be to think that we ought always to doubt or hesitate over
any continuation of any order simply because we lack the slightest
justification for one continuation rather than another. But that is
wrong: the absence of any reason for doing A rather than B, and
106
vice versa, is not in itself a reason for not doing either. To use a
word [or continue a pattern] without a justification does not
mean to use it without right (PI 289). Nor do we in fact have any
doubts, at least not most of the time, for it is not the case that
we are in doubt because it is possible for us to imagine a doubt
(PI 84).
18
But behind that relatively complacent practical point lies a
more unsettling truth about human nature. For if the drift of
PI 21012 is correct then it seems that my understanding
of somebody else is better described as animal or instinctive
response than as rational. The fact is that the pupil, and other
English speakers, just are inclined to continue the order +2
by writing 1868 after 1866, and to regard any other continuation
as wrong.
If I have exhausted the justifications I have reached bedrock,
and my spade is turned. Then I am inclined to say: This is
simply what I do. (PI 217)
When I obey a rule, I do not choose. I obey the rule blindly.
(PI 219)
What is especially disturbing about this position is that it appears
to introduce an element of instinct into what had seemed to be a
paradigm of rationality: the continuation of a mathematical
sequence. That is not for a moment to impugn the confidence
that we rest in such continuations, but only to point out that it
arises not from any sort of rational apprehension but rather is a
matter of blind behaviour.
I want to conceive it as something that lies beyond being
justified or unjustified; as it were, as something animal.
(OC 359)
Naturalism is the philosophy that sees us as just one animal
among others rather than as beings that reason has endowed
with a special sort of insight; it is an important and general truth
that has ethical and political as well as metaphysical implications.
Wittgensteins greatest achievement in Philosophical Investigations
was to show just how deeply it reaches.
READING THE TEXT
107
SECTION 4. PRIVACY
What follows the treatment of rule-following, meaning and
understanding in Philosophical Investigations is an equally
subtle and searching discussion of sensations and sensational
language.
Part of the interest of that topic arises from the fact that sensa-
tional states and processes toothaches, afterimages, kinaesthetic
sensations etc. have long presented a range of problems to tra-
ditional philosophy. Are they identical with physical states and
processes or otherwise reducible to them? How can anyone know
that anyone else has sensations at all? And even if he knows that,
how can he know that anyone elses sensations are like his own?
For that matter, what explains the special certainty with which
he judges his own sensations? What is it to own a sensation?
Wittgenstein does not directly address all of these questions. But
he does give us reasons for doubting the reality of the problems
that they appear to express.
Another part of its interest lies in the fact that many philo-
sophers have placed sensations at the foundation of language.
For instance, we have seen that Locke took words primarily to
stand for ones private ideas, and one plausible interpretation
of idea in this context is sensory image. Indeed we saw that
that thought of Lockes was what lay behind an objection (still
unanswered) to what seemed to follow from the shopping trip
example at PI 1 (1.1). For one might have insisted that the
example did not establish any very great diversity among the
uses of the words five, red and apples, on the grounds that if
we broaden our view of this example to consider the role that
inner states play in it, we see a greater uniformity in their uses than
superficial consideration could reveal. Also in Wittgensteins own
time the logical positivists had thought that one could (Carnap
2003: s61) or must (Ayer 1971: ch. 2) analyse meaningful state-
ments into statements about ones immediate sensory states.
For sensational states to play the role that they are supposed
by these philosophers to have it must be possible for their owners
to speak of them without any knowledge of the external or bodily
criteria on which their third-personal attributions normally rely;
indeed this must be possible whether or not they have such cri-
teria. A language in which its owner refers to sensations that lack
all such criteria is known in the literature as a private language;
108
and we shall look first at two arguments that Wittgenstein
presents against its possibility.
Finally, it looks as though private sensational states are a
counterexample to some of Wittgensteins own remarks on
meaning and understanding. We have already seen that there
need be nothing going on in your awareness at the time, for
example, of the picture of a cubes occurring to you that will rule
out some unintended interpretation of cube; that occurrence
does not therefore force a particular use upon you (PI 140).
Rather, the fact that pictures of cubes customarily have these
and not those other uses is what justifies our calling this event
the occurrence before your mind of an application (PI 141c;
cf. PI 179, 198).
But consider the case of an inner sensory state. Suppose that
on some occasion I decide to use a word S to denote sensations
of the same type as this one and here I focus on some particu-
lar sensation, say my experience of a particular shade of red that
I have just noticed. By doing so, it seems that I could (at least in
principle) set up an association between the word S and that
very shade, so that it is the image of that shade that occurs to me
when I say S? to myself. Now couldnt that be a case in which a
certain conscious occurrence what we might call a private
ostensive definition has forced a use upon me? Perhaps indeed
that is how our actual colour-words get their meaning. How
is he to know what colour he is to pick out when he hears
red? Quite simple: he is to take the colour whose image
occurs to him when he hears the word (PI 239).
So the private ostensive definition appears to put the
application of a word before your mind quite independently
of any other regularities in the use of the image on which
you are then focusing your attention. The first argument that
we shall consider is directed against the possibility of such
a definition.
4.1. Private language
4.1.1. Preliminaries
Wittgenstein first states his target in the following passage:
A human being can encourage himself, give himself orders, obey,
blame and punish himself; he can ask himself a question and
READING THE TEXT
109
answer it. We could even imagine human beings who spoke
only in monologue . . .
But could we also imagine a language in which a person
could write down or give vocal expression to his inner experi-
ences his feelings, moods, and the rest for his private
use? . . . The individual words of this language are to refer
to what can only be known to the person speaking; to his
immediate private sensations. So another person cannot
understand the language. (PI 243)
The passage actually appears to confuse two quite different
features that languages might have and which would justify
the title of privacy; only one of these is of interest to us. In
particular, the penultimate sentence says that the words of this
language (a) refer to what can only be known to its speaker. And
the last sentence says that those words (b) can only be understood
by that speaker. And not only this but also that feature (b) is
a consequence of feature (a). Let us call a language with feature
(a) epistemically privileged; let us call a language with feature
(b) idiolectic.
Edward Craig has shown quite convincingly (Craig 1997) that
a language may be epistemically privileged without being
idiolectic (so Wittgensteins final so is out of place). Consider
that for someone else to understand an epistemically privileged
language it suffices for him to have a true belief about the
meanings of its words: knowledge is not necessary. Imagine
that Bobby uses apple to denote apples whereas his brother
Johnny, whose voice is identical, speaks a variant of English in
which apple denotes tomatoes. One of them rings me up and
without saying his name asks me to fetch him five red apples.
Then he hangs up. I dont know whether the token of apples
that I heard on the telephone refers to apples or to tomatoes;
still, I guess that it refers to apples, and as it happens I am
right. In that case it would be correct to say that I, who did
not know what my interlocutor meant, had understood the
instruction. It would also be correct to say that the person who
was giving it to me had successfully communicated with me
on that occasion.
In general then, one might understand someones words with-
out knowing what they mean; it is therefore hard to see why
110
any epistemically privileged language must also be idiolectic.
And this is also true of the special case at hand: if, for example,
I do not and perhaps cannot know whether your word red
refers to what I call blue (cf. the famous Lockean fantasy:
Essay II.xxxii.15) it will suffice for communication between
us that I guess right. In the case at hand therefore, the two
criteria come apart at any rate, epistemic privilege does not
entail idiolect.
Now Wittgenstein later gives a third characterization of his
target: he states that what concerns him is a language whose
terms refer to states or events that have no natural external
(i.e. behavioural or otherwise bodily) signs by which others may
judge of their occurrence. That is what he states just before the
first argument that I am going to examine (PI 256). For instance,
pins and needles would not qualify as a referent of a term in a
private language if it has characteristic causes (e.g. sitting in
one place for a long time); nausea would not qualify if it has
characteristic effects (e.g. vomiting); and toothache would not
qualify because it has characteristic causes (e.g. a rotten tooth)
and characteristic effects (e.g. clutching ones jaw).
If on the other hand you thought, as a Lockean should
19
, that
we could not judge of somebodys pins and needles, nausea, and
so on by these outer signs, then you would allow terms referring
to pins and needles, nausea and toothache among those of
Wittgensteins target language, on this third characterization.
And you would presumably also hold that English, or a sector
of it, was a private language in this third sense, containing as
it does terms for pins and needles, nausea and toothache. So
an argument against this third sort of language would also be an
argument against a Lockean interpretation of English.
Three different kinds of language might therefore count as
Wittgensteins target: (a) A language whose terms refer to items
known only to its speaker; (b) A language whose terms can only
be understood by its speaker; (c) A language whose terms refer
to sensational states and processes that have no external mani-
festations by which anyone other than their possessor can judge
of their occurrence. I shall take Wittgenstein to be arguing
against a language of type (c) on the grounds that this makes for
the most interesting interpretations of what he says.
READING THE TEXT
111
4.1.2. S in the diary
One can gather the gist of the first argument from PI 258, which
runs as follows:
Let us imagine the following case. I want to keep a diary about
the recurrence of a certain sensation. To this end I associate it
with the sign S and write this sign in a calendar for every
day on which I have the sensation. I first want to remark
that a definition of the sign cannot be formulated. But
still I can give myself a kind of ostensive definition. How?
Can I point to the sensation? Not in the ordinary sense.
But I speak, or write the sign down, and at the same time
I concentrate my attention on the sensation and so, as it
were, point to it inwardly. But what is this ceremony for?
for that is all it seems to be! A definition surely serves to
establish the meaning of a sign. Well, that is done precisely
by the concentration of my attention; for in this way I impress
on myself the connexion between the sign and the sensation.
But I impress it on myself can only mean: this process
brings it about that I remember the connexion right in the
future. But in the present case I have no criterion of correct-
ness. One would like to say: whatever is going to seem right
to me is right. And that only means that here we cant talk
about right.
Although the detailed interpretation of this passage is a matter
of some controversy, it is possible to be fairly definite as to its
aim and drift.
The quoted section does not make it explicit, but it is clear
from the preceding ones that Wittgenstein is here attacking the
idea that one can by ostensive definition establish the terms of
what I am calling a private language. That is, S is supposed to
refer to a sensation that has no natural external expression.
20

The aim of the section is then to show that one cannot establish
the meaning that is, the reference of the signs of a private
language by means of a private ostensive definition. To show
this would perhaps not show that a private language was imposs-
ible, but it would make it difficult to see how anyone could learn
its terms: difficult, therefore, to accept the Lockean view that
112
natural languages like English have a private sector. It would
also be to refute the supposed counterexample that I mentioned
at the end of this chapters introduction. Private ostensive defini-
tion was supposed to be a means of presenting the application
of a word to its user quite independently of any regularities in the
use of the associated image; as such it would be a counterexample
to Wittgensteins doctrine that the way something is meant on
an occasion depends on its customary usage. If successful the
argument that we are now considering would show that the
private ostensive definition would not after all endow the word
S with any meaning at all.
The broad drift of the argument is also clear and may be set
out as follows. The private diarist wishes to introduce a word S
for the type of sensation that he is presently having. And he
tries to do it by ostensive definition. But ostensive definition is
not just a matter of uttering the defined expression S in the
presence of the type of sensation that you are intending to name.
Nor is it enough to utter S while concentrating your attention on
that type. What is needed in addition is that the definition brings
it about that the diarist remembers the connection right the
connection, that is, between sign and sensation. But a private
ostensive definition cannot bring that about because it supplies
no criterion of correctness for my use of S. So the supposed
private ostensive definition was not really a definition of S at
all. It was an idle ceremony.
But we cannot assess whether the argument thus outlined
achieves its aim without examining its details. And with respect
to these there is scope for great variety of interpretation. Settling
which of these interpretations is the one that Wittgenstein really
had in mind would be beyond the scope of this book, even if
I were able to do it. So instead I shall simply present the three
main interpretations of the argument and say something about
their plausibility as arguments.
Consideration of the details of PI 258 naturally gives rise to
the following five questions:
(a) What is it to remember the connection right?
(b) Why must a successful ostensive definition bring it about
that you remember the connection right?
(c) What is a criterion of correctness?
READING THE TEXT
113
(d) Why must a successful definition supply a criterion of
correctness?
(e) Why does a private ostensive definition fail to meet
condition (b) and/or condition (d)?
The three interpretations that I shall consider may be character-
ized roughly as follows: (i) verificationism if you name a type
S you have to specify evidence for its recurrence, and the private
linguist cannot do this; (ii) the meaning-check if you name a
type S then you have to specify a way of checking later what
you meant by it all along, and the private linguist cannot do this;
(iii) sortalism if you name a type S then you have to specify
what type of thing you are naming S, and the private linguist
cannot do this. Let us now consider these three interpretations
in turn.
(i) Verificationism. According to this interpretation of PI 258
the answers to our questions (a)(e) may be stated as follows.
(a) To remember the connection right is to write S in ones
diary for just those days on which the diarist experiences a sensa-
tion of the type that he initially named S. (b) A successful
ostensive definition must bring this about because if one uses S
in some very different way then it has not established a connec-
tion between sign and sensation. (c) A criterion of correctness is
some means of verifying that (more generally, of checking
whether) what one is now experiencing is correctly called S,
that is, of the same sensational type as that initially defined.
(d) A criterion of correctness is necessary because in order to use
the term S consistently with its definition, as required by (b),
one must be able to decide whether one has done so. (e) The
private ostensive definition fails to meet the condition (d) (and
therefore also (b)) because in the private case one has no means
of checking whether or not one is applying the word S consis-
tently with its definition.
All of these elements are clearly present in the best statement
of this interpretation: Norman Malcolms. Imagining a private
ostensive definition of pain, he writes:
Let us suppose that I did fix my attention on a pain as I
pronounced the word pain to myself. I think that thereby I
established a connection between the word and the sensation.
114
But I did not establish a connection if subsequently I applied
that word to sensations other than pain or to things other
than sensations, e.g. emotions. My private definition was a
success only if it led me to use the word correctly in the future.
In the present case, correctly would mean consistently with
my own definition . . . Now how is it to be decided whether I
have used the word consistently? What will be the difference
between my having used it consistently and its seeming to me
that I have? Or has this distinction vanished? Whatever is
going to seem right to me is right. And that only means that
here we cant talk about right . (Malcolm 1954: 68)
That passage exhibits points (a)(d). As for the final point (e),
Malcolms view is that one cannot verify that one is using S
correctly on the sole basis of ones impression that one is doing
so. Instead one must have some independent check of that
impression itself. And I cannot check my impression against
another impression because that would not be an independent
check. As if someone were to buy several copies of the morning
paper to assure himself that what it said was true (PI 265,
quoted in Malcolm ibid.). But in the case of the private sensa-
tion there is no further check, for there are no external signs
of S (e.g. typical bodily causes or behavioural effects) by whose
occurrence or otherwise one can confirm or disconfirm ones
present impression that it is now occurring.
But first, how plausible is point (d)? Why should we not say:
it is one thing to use an expression consistently with ones
definition but quite another to know that one is doing so. Why
could the former not happen without the latter? The following
example of how it could happen is due to Strawson:
[A man] might simply be struck by the recurrence of a certain
sensation and get into the habit of making a certain mark in
a different place every time it occurred. The making of the
marks would help to impress the occurrence on his memory.
One can easily imagine this procedure being elaborated into a
system of dating. (Strawson 1954: 44)
Or again, a child could be trained or, we may as well imagine,
might magically acquire a habitual tendency to utter the sound
READING THE TEXT
115
S whenever it feels hot; we could think of this behaviour as a
primitive sort of report just as the behaviour of a thermometer
could be thought of as a primitive sort of report on its temper-
ature. Nothing in this story requires that the child have any
means of checking that it is reporting the same sensational type
as the one that we initially connected with S; but still, why
should we not call this reporting?
Malcolms response to Strawsons example is to say that the
mere uttering of a sound in the presence of, say, a cow, does not
even make that sound into a word, let alone a word for cows.
The sound might refer to anything or nothing. What is
necessary is that it should play a part in various activities,
in calling, fetching, counting cows, distinguishing cows from
other things and pictures of cows from pictures of other things.
If the sound has no fixed place in activities (language-games)
of this sort, then it isnt a word for cow. (Malcolm 1954: 96)
But why must a word or sound play a role in all of the activities
that Malcolm mentions or, for that matter, in any of them?
Why does not Strawsons very light sketch of its use suffice to
indicate how S could belong to a language-game of reporting
ones sensations? Consider again the very simple builders
language-game at PI 2. There is nothing wrong with saying
that beam is a name in that language for beams (cf. PI 37). So
the habitual uttering (by A) of Slab! in the presence of a desire
for a slab, and the habitual fetching (by B) in the presence of
As utterance, can make slab a name for slabs: why then cannot
the mere utterance of S in the presence of S do the same? One
could insist if one wished that this pattern of usage is too simple
to count as an instance of our fully fledged language-game of
reporting; but to insist that therefore it cannot count as a report
at all is to draw a sharp line around the concept of reporting that
an objector to Wittgenstein will refuse to acknowledge.
21
Nor is it clear that we should accept the verificationists
point (e). It is one thing to say that no publicly available occur-
rence may function as an independent check upon whether
one has had S again; it is another thing to say that nothing at
all can perform this role. Why should one not be able to check
ones impression, that this is S again, against another private
116
occurrence? For instance, it may be that whenever I have had
sensation S in the past I have also and simultaneously had
sensation T. So the presence of T on a later occasion may
confirm my impression, on that occasion, that what I am now
having is S; and the absence of T on another such occasion
may disconfirm my impression, on that other occasion, that
what I am now having is S.
The defender of Malcolm will immediately reply that exactly
the same problem arises for T itself: how am I to confirm my
impression, on any occasion, that I am now having T? (Cf. PI 265:
checking the same newspaper twice.) I cannot now appeal to
a third type of private occurrence, because that will lead to an
infinite regress, and I cannot appeal to S itself, because that
would be circular. So I will have to appeal to some publicly
available correlate of T, and so the manoeuvre only delays
Malcolms victory.
But the difficulty with this reply is that it presupposes that
in order for T to be an independent check on S I must have an
independent check on T itself. If that assumption were true
the unavailability of a private language would be the least of
my problems: for it applies equally to the public world. I seem
to recall that the word desk applies to this object in front of me,
but how can I check? Well, I know that desks have a flat top
and four legs. But how can I check whether I am correct to apply
flat to the object in front of me? Well, desks are usually
flat, and what is in front of me is a desk, so . . . That would be
circular. So I must appeal to some third thing: and so again we
have an infinite regress. Ayer, to whom this objection is due,
summed it up as follows:
[U]nless there is something that one is allowed to recognize,
no test can ever be completed: there will be no justification
for the use of any sign at all. (Ayer 1954: 256)
22
And this applies to the private as well as to the public sector of
our language.
(ii) The Meaning-Check. The second interpretation gives the
following answers to our five questions. (a) To remember the
connection right is to remember what sensation-type occurred in
the definition of S. (b) It is necessary to remember the connection
READING THE TEXT
117
right because in order to mean S by S one must know what S
means; and one does not know what S means unless one
remembers its definition. (c) A criterion of correctness is a
means of checking ones memory of the definition of S. (d) A
criterion of correctness is necessary because without it one
does not remember its definition; and so by (b) one does not
know what S means. (e) In the private case one cannot check
ones memory of the definition because one has nothing with
which to compare it.
The principal difference between the meaning-check inter-
pretation and the verificationist one, from which all the other
differences flow, is over question (a). The difference is that
the meaning-checkers answer is necessary but not sufficient for
the verificationists: in order to use S consistently with the
definition on an occasion, you must be able to remember the
definition; but one can remember the definition correctly with-
out using S correctly, for one might choose to write something
misleading or not to write anything at all (Kenny 1975: 192;
Canfield 2001: 384).
In consequence the meaning-checkers answer to question
(b) is different from the verificationists. But it is at least as
implausible. It relies on the commonplace but for all that
mistaken equation of understanding S with knowing the
meaning of S . We have already seen (4.1.1) that we can place
pressure upon that connection in the interpersonal case: in order
to understand another I do not need to know but only to have
a true belief about what he means. But the same point applies
to the intrapersonal case: as long as I so much as guess right
how I defined S, I understand it well enough.
Alternatively, my true belief as to what I meant might be the
upshot of a causal process whose reliability qualifies that belief
as a form of knowledge, at least on a sufficiently externalist
conception of our knowledge of the past. That would not be
a counterexample to (b); but it would be a counterexample
to (d). Thus Kenny, to whom the meaning-check interpretation
is due, writes:
Suppose next that the private-language speaker says By S I
mean the sensation I named S in the past. Since he no longer
has the past sensation he must rely on memory: he must call
118
up a memory-sample of S and compare it with his current
sensation to see if the two are alike. But of course he must
call up the right memory. Now is it possible that the wrong
memory might come at his call? If not, then S means what-
ever memory occurs to him in connexion with S, and . . .
whatever seems right to him is right. If so, then he does not
really know what he means. (Kenny 1975: 192)
Of course it is possible that one should misremember which sen-
sation one named S in the past. But why should the possibility
of being wrong rule out the actuality of knowledge? as if every
doubt merely revealed an existing gap in the foundations (PI 87).
After all, if one thinks of knowledge as something that a reliable
but uncheckable process can produce, then of course one can
still remember the definition right, and hence know what it was,
whether or not one can check ones memory. And if the meaning-
checker should now insist upon so meaning the word knowledge
that knowledge requires an independent check is analytic, we
are back to the original objection to his point (b).
(3) Sortalism. Both verificationism and the meaning-check
interpretation take it for granted that the argument is basically
epistemological: that what cripples the private diarists attempt
to mean something is his not being able to know something
for instance, whether S is correctly applied in the present
instance. The third interpretation rejects this assumption. Instead
the argument is taken to reveal a more basic flaw in the private
diarists procedure. It is not that there is something that he cant
know, but that there is nothing that he can. More precisely: his
attempt at a private ostensive definition has failed to specify any
correct use for the sign S, whether or not he could later come to
know about it. Even if God had looked into the private linguists
mind he could not have seen there what the sign S was supposed
to mean (cf. PI p. 185/217). The reason for this is that the private
ostensive definition has not established which, of the many types
of thing that the initial occurrence of S falls under, is the type or
sortal that S is meant to be denoting.
In terms of our five questions, this sortalist interpretation may
be characterized as follows. (a) To remember the connection
right is subsequently to use S in accordance with the pattern of
correct use that the initial definition settles. (b) An otherwise
READING THE TEXT
119
satisfactory definition that does not bring this about has not
established the meaning of the sign. (c) A criterion of correct-
ness is not a means for the diarist to tell what the correct use
is; it is what makes this rather than that pattern of use correct.
(d) A successful definition must supply such a criterion because
otherwise there is no right or wrong use of the sign S. (e) But
this cannot happen in the private case because the mere con-
centration of ones attention upon S while saying S does not
make any one pattern of subsequent use more correct than any
other, for it does not specify what type of thing S is supposed
to denote.
In the two clearest and most interesting statements of sortalism
due to McGinn (1997: 1314) and Stroud (2001): for a briefer
account see Glock 1996: 312 (a) and (b) are stated very briefly
or are only implicit; but this is of little importance since it is
(c)(e) that are doing all the work. This is Stroud:
The speaker [at PI 258] is imagined to give himself an ostensive
definition that serves to establish the meaning of a certain
sign by impressing on himself the connexion between that
sign and a sensation he has at the time he utters it or writes it
down . . . It is supposed to be a connection that establishes
how the sign is to be correctly used in future. But a word can
be connected with one and the same thing in many different
ways. To say that in the present case I have no criterion of
correctness is to observe that nothing has so far been done
to determine which future applications of that sign will be
correct and which will not. (2001: 154)
On this interpretation of criterion of correctness he continues:
The point of there being no criterion of correctness cannot
be simply that there is no test that the speaker can rely on to
tell him that his application of the sign is correct. There is so
far nothing for any such test to be a test for. Nothing has been
done to fix what being correct is or would be. (Ibid.)
And on the question of what exactly is undetermined about the
future application of S he writes: The overall role of the sign
has not been specified: there is so far no post at all at which it
120
is stationed. It is just a mark written or a sound uttered on a
certain occasion (ibid.).
The words post and stationed in that last remark are allusions
to their occurrence in PI 257, which itself is an application to
the private sector of the discussion of ostension at PI 2835
(discussed at 1.2). There Wittgenstein had argued that one could
not introduce a name for a colour, a numeral, a point of the
compass, and so on unless there was something in virtue of
which one had introduced it as a name for one kind of thing
or another. But what makes it true that the private linguist has
introduced the name for, say, the sensation-type that encompasses
S rather than for some other type that does so? For sensation
is a word of our common language, not of one intelligible to me
alone. So the use of this word stands in need of a justification
which everybody understands (PI 261).
Be that as it may, it is not clear what logical or conceptual
difficulty this version of the argument raises for the private
diarist. In particular it is not clear what grounds there are for
accepting its point (e). Let the diarist utter the sound S in the
presence of a sensation on one occasion, and let this occasion
be so impressed upon his memory that whenever in future he has
something of the same sensational type, he then writes S in his
diary. There is certainly no need for him to be able to say, in
defining S, that it is the name for a sensation; all that is needed
is that given his similarity space he responds in similar ways to
experiences of the same sensational type. And I cannot see
any logical incoherence in this possibility (though I have no idea
how likely it is).
You might object that wed have no justification for saying
that he was employing S as the sign for a sensation. Of course if
we is supposed to mean other human beings that is obviously
true, for the private language was defined to be one whose terms
referred to something that had no external manifestation, so
nobody else could tell when it was occurring. To us, it would
look as though he was simply writing S in his diary on random
days. But that surely does not mean that he is not in fact using
the sign S as a sign for a sensation. If God had looked into his
mind then he would be justified in so describing it.
The objector might have something else in mind. As we shall
see (4.2.5), it was Wittgensteins view that our everyday words
READING THE TEXT
121
for sensation are tied up with our natural (i.e. bodily) expres-
sions of sensations (PI 256). So, the objector will continue, we
cannot say that S is a sign for a sensation because all sensation
words stand for states that have natural bodily expressions,
and this S has none. (That seems to be the point of a remark at
Stroud 2001: 1545.)
But this argument faces two objections. First, it is hardly clear
that our everyday words for sensations are tied up with that is,
presumably, require for their understanding an appreciation
of their natural expressions. This is something that, for
example, the Lockean will contest; and if the Wittgensteinian is
going to defend it then he had better not employ the sortalist
version of the private language argument (or his defence will be
circular). And second, even if our everyday words for sensations
are so tied up with their natural expressions, it need not be true
that all sensation words are so tied up in any possible language,
real or imaginary, natural or artificial. So even in that case the
argument would have no logical but at best only inductive force.
It therefore appears that on all of these interpretations the
private language argument fails. But we should not give up on
it. In addition to these three main interpretations of it, there
are others whose distance from these three and each other
make one suspect that many treasures (i.e. arguments) remain
concealed within PI 258 and its environs. It is too soon to despair
of finding one that is actually sound.
23
4.1.3. The beetle in the box
Whether or not the argument at PI 258 establishes any of
the things that I said Wittgenstein there had an interest in
showing, the famous beetle-in-the-box argument at PI 293 at
least seems to be a clearer and more decisive demonstration
of one of them.
We have already seen that on what I have called the Lockean
conception, words of our everyday language refer primarily
to ideas: that is, objects of quasi-sensory introspection that
function as inner samples. Thus with the word red I associate
(on this view) an inner sample of red; and if I wish to know, for
example, whether this apple is red, what settles the matter for me
is a comparison of the inner sample with the outer object. And
we have seen how this idea presents a challenge to Wittgensteins
122
claims about the functional diversity of our words. If one can
as Locke thought extend this account to all general terms
of, say, English (blue, man, horse, sacrifice, etc.), then one
could reasonably insist against Wittgenstein that the superficial
diversity in their uses conceals a deeper commonality.
If the Lockean view is right then the words of our everyday
public language refer to entities that are in principle quite
unknowable to anyone else: at least, nobody can have any
evidence for what anybody else has got. If my word red refers
to my inner sample of red, and if your word red refers to your
inner sample of red, who is to say whether your inner sample is
of the same chromatic type as mine?
It may seem at first that we can infer this from the agreement
between us in our use of the word red, at least in central cases.
For instance, we agree that this tomato which we are observing
from the same angle, in the same light, and so on is red, and
we agree that that apple is not red but green. However on the
Lockean principle, this agreement might equally be explained in
at least two other ways.
(i) If you hold a Lockean theory of perception as well as a
Lockean theory of meaning, you will say that what each of us
immediately perceives of the apple or tomato is not it but the
sensory idea that it produces in each of us (Essay II.ix). So it
may be that there is a systematic permutation in the objects of
your perception relative to the objects of mine: what produces a
green idea in you might produce a red idea in me, while what
produces a red idea in you produces a green idea in me. Given
that we learn the words green, red, and so on by being shown
objects that are typically said to have those colours, it is there-
fore likely that you will associate with the word red an idea of
the same type as the one that I associate with the word green,
and vice versa. In that case you would expect our uses of the
words red and green to match just as they actually do. And
can we rule out that case?
(ii) Wittgenstein himself was correct to say at PI 73 that the
inner sample cannot determine ones use of a word by itself: one
also needs to use it in a certain way; or, in the terms of PI 139, to
have settled upon a method of projection. So even if you associ-
ate with red an idea of the same chromatic type as the one that
I associate with green, and vice versa, this difference might still
READING THE TEXT
123
not show up in our uses of red and green. For we might also
differ, and in a precisely compensating way, over our methods of
projection: so that your lines of projection extend from the idea
that you associate with red to tomato ketchup, London buses
and post-boxes; whereas mine extend from an idea of the same
chromatic type which, however, I associate with green
to grass, lettuce and the caps worn by Australian cricketers.
And conversely, they extend from the idea that I associate with
red to tomato ketchup, and so on, and from the idea that you
associate with green to grass, lettuce, and so on. That is just to
say that we associate the same inner ideas with different outer
objects, and conversely that we associate different inner ideas
with the same outer objects, but in such a way that none of
this shows up in our use of the associated words. Indeed this
hypothesis would predict just the agreement on overt verbal
judgements that we actually observe; on what grounds, then, can
we rule it out?
So the Lockean view makes the primary referents of each
persons words unknowable to anybody else: at least, it is not
something that they can rely upon any evidence to settle. In the
terminology of 4.1.1 it makes language out to be epistemically
privileged. Only its utterer knows what an utterance of red
denotes, for only he knows what idea he associates with it. It is
as if when I uttered the word I cast a sidelong glance at the
private sensation, as it were in order to say to myself: I know all
right what I mean by it (PI 274). It is this aspect of the Lockean
view that Wittgensteins argument fastens upon. It attempts to
show that the words of a communal language could not refer to
entities that are private in this way.
This is the argument.
Suppose everyone had a box with something in it: we call it a
beetle. No one can look into anyone elses box, and every-
one says he knows what a beetle is only by looking at his
beetle. Here it would be quite possible for everyone to have
something different in his box. One might even imagine such
a thing constantly changing. But suppose the word beetle
had a use in these peoples language? If so it would not be
used as the name of a thing. The thing in the box has no place
in the language-game at all; not even as a something: for the
124
box might even be empty. No, one can divide through by
the thing in the box; it cancels out, whatever it is.
That is to say: if we construe the grammar of the expression
of sensation on the model of object and designation the
object drops out of consideration as irrelevant. (PI 293bc)
The intention of the analogy is fairly clear. The beetle corresponds
to the inner or private idea that we are each supposed to associ-
ate with, for example, some sensation word of English, to which
latter the word beetle corresponds. And the point is that the
public use of the expression beetle, whatever it is, does not
depend upon what is in anyones box. For its public use would
proceed undisturbed if each of us had a different object in his
box; indeed it might proceed undisturbed if some of us had
nothing in the box. So in its public use the word beetle in some-
bodys hand or mouth does not refer to his beetle; and the
analogous conclusion goes for sensation words and indeed
all other words of an everyday language like English, that is,
one that has many speakers.
Let us first distinguish this point from what the argument
does not show. It does not show nor, I think, does it attempt to
show that the words red, green, and so on could not have a
private meaning in addition to their public one. But what it does
try to show is that if we focus solely upon the public usage of
these words, that is, the use upon which we all agree, we can
eliminate all reference to private objects in the explanation of it.
That would be enough to defeat the Lockean challenge to PI 1
that I mentioned at 1.1. It would show that we should not see the
public uses of different words as superficially dissimilar tips of
similar icebergs; rather, those tips are each quite disconnected
from whatever lies beneath the surface.
But does it show this much? There can be no question, I think,
that the argument illustrates pretty clearly that one may vary
ones inner objects without varying the public use, in some
sense of use, of the associated word. But what do we mean
here by use?
Wittgenstein nowhere makes it precise, but I think that in
order properly to assess this argument we must do so. At any
rate we must make precise what counts as the same use. Let
us say, then, that a persons public use of a word is the totality
READING THE TEXT
125
of his dispositions to utter, and to assent to, or dissent from,
sentences containing that word, in response to given sensory
stimulations. For instance, it belongs to my public use of the
word red that I am disposed to assent to the query Red? when
confronted with a visibly salient ripe tomato; and it further
belongs to my public use of that word that I am disposed to
say Red! when queried about its colour shortly thereafter.
(In the terminology of 3.2.1 I am speaking here of a persons
A-dispositions, not his B-dispositions.) And just as you would
expect, we shall say that two people have the same public use for
a word if and only if the use that one of them has for it is the
same as the use that the other one has for it.
This is a highly artificial and certainly not very Wittgensteinian
definition of public use and same public use. Moreover, it
makes it very unlikely that two people will ever have the same
public use for a word. For it makes ones use of a word depend
in part upon what would usually be reckoned collateral informa-
tion as opposed to knowledge of that words meaning. For
instance, in response to the question Was Lyndon Johnson the
thirty-sixth President of the United States? A might be disposed
to say Yes whereas B is disposed to say No; on the present
definition it follows that A and B have different public uses for
thirty-six.
But neither of these points is going to matter. For one thing, my
artificial version of public use does no disservice to Wittgenstein
in the present context. If the beetle in the box cannot be made to
establish its intended conclusion by means of this conception of
sameness of public use, then I do not see how it can be made to
work at all. And second, the fact that my definition makes it very
unlikely that two people will ever have the same public use for
a word makes no real difference either. It means that one of
the premises that I attribute to Wittgenstein is going to seem
very strong; but it will induce a compensating weakness in the
other premise, so the overall soundness of the argument will not
be affected.
So on this conception of sameness of public use, the argument
does show that two people might have the same public use for an
expression while associating it with different private objects. For
instance, A and B might associate quite different private objects
with the word toothache. But still they might both have exactly
126
the same public use for the word. For instance, they could and
would both go to the dentist complaining of toothache, and he
could and would treat them appropriately in response, for example,
by extracting the rotten tooth. What is in the box cancels out.
But it does not show that just any variation of the private
object would be immaterial to ones use of the word. Suppose
that somebody associated the same inner object and the same
method of projection with the words green and red. In that
case the inner object would not cancel out: there would have to
be a difference between his and our usage of at least one of the
words red and green. For he would apply red to ripe tomatoes
if and only if he applied green to ripe tomatoes; whereas we
apply red only to ripe tomatoes and green only to unripe ones.
What the argument shows, then, is that we can preserve the
public use of words while varying the inner object ad libitum so
long as we make compensating variations elsewhere. In particular
the variation must assign different inner objects (or, if you like,
different inner-objects-plus-methods-of-projection) to words that
were initially assigned to different ones, and it must for the same
reasons assign the same inner objects to words that were initially
assigned to the same ones. Such a variation is called an injection
of the inner objects. So we can put it like this: Wittgenstein has
shown that an injection of the inner objects associated with our
public words is immaterial to their public use.
But this is not yet the intended conclusion of the beetle-
in-the-box argument. The intended conclusion is the one stated
at PI 293c: that none of these words actually denote any inner
object. So does this claim follow from the premise that an
injection of the inner objects associated with our public words
is immaterial to their public use?
It does not follow logically, but it does follow on the additional
assumption of a premise that Wittgenstein shows every sign of
accepting (short of actually stating it). This premise can be put
as a slogan: no variation in reference without variation in use.
Wittgenstein actually comes fairly close to endorsing it with
the remark that the word name is used to characterize many
different kinds of use of a word (PI 38b; my emphasis). But this
does not quite count as an endorsement of what I mean, if only
because Wittgenstein nowhere gives to same use the precise
meaning that I have assigned it for present purposes.
24
In any
READING THE TEXT
127
case it seems quite clearly to be in the spirit of Wittgensteins
general idea that semantic phenomena are not independent of
but arise wholly from ones use of words. More precisely, the
premise is this: there could not be two people who had the same
public use for words that yet had different references. We may
term this premise semantic behaviourism.
25
But PI 293b has shown that there could be two people
who had the same public use for a word (like beetle) that was
associated with different inner objects. It follows from our two
premises that that word cannot refer to either associated inner
object. Since both word and object were arbitrary, we can
generalize the conclusion of this argument: no word of a public
language refers to any associated inner object. It therefore seems
that the argument has successfully established the conclusion
that it states at PI 293c.
But one might ask whether it has thereby pointed out any
special difficulty about reference to private objects. After all, we
could make the very same point about public objects; Quine has
done so.
26
Let us suppose that every object has a complement.
The complement of an object O is the object O* that is composed
of the entirety of the universe (all of space and time) except for
that object (so O** = O). Now imagine two people A and B:
A associates his words with ordinary physical objects tomato
with tomatoes, Fido with Fido, dog with dogs, and so on
whereas B associates those words with their complements
tomato with complements of tomatoes, Fido with Fido*,
dog with the complements of dogs, and so on. Clearly, replac-
ing all objects with their complements is an injection of those
objects, in the sense of injection defined earlier.
Now on certain simplifying assumptions B will assent to (e.g.)
Fido is a dog if and only if Fido* is a complement of a dog,
that is, if and only if Fido is a dog, that is, if and only if A will
assent to the same sentence given the same evidence. What
happens is that the variation in B on what A associates with
Fido exactly compensates for the variation in B on what
A associates with dog. And I believe that this point can be
generalized; so it is possible that A and B should have the
same public use for all of their words. That is, one may vary
by injection a set of outer objects while the public use of the
associated words proceeds quite undisturbed.
128
Applying now the premise of semantic behaviourism (which
has no special application to inner objects) and generalizing
the result as in the private case we may conclude that no word of
a public language refers to any associated outer object. The outer
object cancels out in just the same way as the inner one.
So Wittgensteins argument, at least on this construal of it,
does not after all reveal any special problem with the Lockean
idea that ones words denote associated private objects. Now
it may be that the problem resides in my reading of the
argument. In particular and for the purpose of assessing it,
I assigned to the notion of sameness of public use a definition
that Wittgenstein would quite possibly not have accepted.
Could he have had in mind some other notion of public use
than the brute behavioural propensities with which I there
identified it?
Quite possibly he did; but it is not clear how that would help
him. Whatever notion of use he may have had in mind,
the argument at PI 293b appears to show quite effectively that
variation of associated objects is compatible with sameness
of use in my sense as well as his. So PI 293b does successfully
establish my reading of the first premise, whatever else it
establishes.
And whatever notion of use he may have had in mind is surely
one that cuts no finer than mine: two people who have the same
public use for a word in my sense will have the same public
use for it in any sense of that Protean word. It follows that if
semantic behaviourism is true on his interpretation of same
public use then it is true on mine as well. So if the second
premise of his argument is correct, whatever it may be, then so
is the second premise of my reading of it.
So if Wittgensteins beetle-in-the-box argument establishes
the semantical irrelevance of inner objects then it establishes the
semantical irrelevance of outer ones too. Hence it establishes no
special difficulty for a Lockean interpretation of our language
on which words denote associated private or inner objects. What
is wrong with the beetle-in-the-box argument is not that it fails
to establish its conclusion. It is that one could travel much
further in its direction; from the perspective that one then attains,
Wittgensteins limitation of his argument to the private sector
looks arbitrary.
27
READING THE TEXT
129
4.2. The inner and the outer
The two passages that this chapter has so far focused upon are
embedded within a much larger discussion of a variety of mis-
conceptions that affect philosophical thinking about sensations
and indeed the whole picture of our mental life. Those miscon-
ceptions coalesce into what might with some historical license be
called the Cartesian picture, although in using this label one
should bear in mind that it has probably shaped the thoughts of
many people who were not Descartes to a greater extent than it
ever informed the work of that great philosopher, and that
among these people were philosophers with whom he was not
otherwise in intellectual sympathy.
Having outlined this picture we shall discuss various points
at which Wittgenstein attacked it: (a) the idea that their bearer
has a special epistemic authority over his sensational states;
(b) the idea that we are inevitably in a state of ignorance as to
somebody elses sensational states; (c) the idea that their bearer
is a self or soul that is distinct from ones body; (d) the idea that
we come to understand third-personal sensational attributions
by a transference of our understanding in the first-personal case.
I conclude with a discussion of Wittgensteins alternative to (d).
4.2.1. The Cartesian picture
The basic Cartesian picture is that from the point of view of any
human individual, for instance me, the world is open to being
metaphysically factored into two logically independent compon-
ents. On the one hand there is the physical or external world
(though it is slightly misleading to call it external as it includes
the interior of my body as well as my brain). At least in principle
everyone has equally good observational access to this world. It is
true that in practice there are some parts of it that I am in a better
position to observe than anyone else. For instance, I can feel the
goings-on in my muscles better than anybody else can see them, so
I am in a better position to observe a cramp in them than anyone
else. But at least in principle another observer could observe that
cramp by looking at my muscles; in fact in principle somebody
elses nerves might be connected to my muscle, in which case that
other person would be as well-placed as me to feel the cramp.
On the other hand there is the mental or internal world. This
is the world of mental states, processes and events: sensations,
130
feelings, moods but also thoughts, beliefs and preferences. In
principle I have better access to my own mental states than any-
body else. In particular my sensations are objects of direct
internal observation, by which I have infallible access to them. I
cannot be mistaken about my sensations. My know ledge of them
therefore has a security to which my know ledge of the merely
external world can never attain.
But nobody else can observe them directly. This is not because
they are hidden within my skull: even if my head had been made
of glass, so that other people could see everything that was going
on in my brain, they would still not be observing my sensations.
So other people have to make an inference about my sensations
upon the basis of what they do observe: in practice this is my
behaviour.
And what is this I that observes these sensational states? It is
not my body or any mere state of my body. It is a spiritual entity
called the self that directly observes just the sensations that
it owns. There are many such selves, each one corresponding
to a person, and each one has its sensations that it observes
directly and which no other self observes directly. But the self
can understand what it is for another self to have a sensation
by concentrating on its own. Joness having toothache, Smith
might say to himself, is just another occurrence of this [here
concentrating on his own toothache], only an occurrence that
Jones owns, not me.
This picture is not some medley of ideas that Descartes
happened to throw together. There are indisputable facts that
make it very natural for many people (at least for many
Westerners) to think in terms of it without ever making it
explicit. Wittgensteins treatment of the picture is an attempt
to extract it by these roots.
4.2.2 The epistemic asymmetry
We have seen that one important component of the Cartesian
picture is the idea that I know, for example, whether I am feeling
a headache right now; everybody else has to infer it. In connec-
tion with this epistemic asymmetry Wittgenstein writes:
In what sense are my sensations private? Well, only I can
know whether I am really in pain; another person can only
READING THE TEXT
131
surmise it. In one way this is wrong, and in another nonsense.
If we are using the word to know as it is normally used (and
how else are we to use it?), then other people very often know
when I am in pain. Yes, but all the same not with the certainty
with which I know it myself! It cant be said of me at all
(except perhaps as a joke) that I know I am in pain. What is it
supposed to mean except perhaps that I am in pain?
Other people cannot be said to learn of my sensations only
from my behaviour, for I cannot be said to learn of them.
I have them.
The truth is: it makes sense to say about other people that
they doubt whether I am in pain; but not to say it about
myself. (PI 246)
Let us start with my knowledge of my own sensations. Wittgenstein
asks what it might mean to say that I know that I am in pain
other than that I am in pain. But why should we think that these
two statements mean the same? After all, they have different
uses. I might say I am in pain to the dentist when he is extract-
ing my tooth as a way to indicate that he should administer more
anaesthetic. But Id only say I know that I am in pain! if the
dentist started to question my first statement: Are you sure that
you feel pain?, and so on.
Wittgenstein thinks that I know I am in pain does not mean
anything else because knowledge-statements are not true in cases
where it makes no sense to doubt. He writes elsewhere that
I know . . . may mean I do not doubt . . . but does not
mean that the words I doubt . . . are senseless, that doubt
is logically excluded. (PI p. 188/p. 221)
And as he says at PI 246c it seems that this is one of those
cases in which the expression of doubt is senseless or logically
excluded. At least it is so in this sense: somebody who sincerely
said I doubt whether I am in pain would thereby reveal nothing
except his misunderstanding of one of those words: he might
think, for instance, that pain names a certain type of bodily
damage.
But in that sense of logical exclusion, utterances of It is raining
on a plainly sunny day are logically excluded too: somebody
132
who sincerely said that should be taken to have misunderstood
his words. But then the fact that a statement is logically excluded
cannot rule out the sensefulness and indeed truth of its negation.
It is not raining would on that occasion say something so
obvious as to make its utterance pointless; but for all that it
would say something true. Why then should I know that I am
in pain not say something true as well?
Another reason for thinking that I know that I am in pain
cannot mean anything other than I am in pain is that I am
in pain is not something that one can either know or believe,
because it is not a description of anything at all but simply a
means of attracting attention disguised as an assertion. This
may be part of what Wittgenstein had in mind when he wrote
that the verbal expression of pain replaces crying (PI 244; cf.
PI 290b). Obviously it does not make sense to affix I believe . . .
or I know . . . before a cry; perhaps then we ought to say
the same about the statement got by affixing either before I am
in pain.
But we have good reason to think that this is something that
one can believe. The point of attributing beliefs to people is to
explain their behaviour; but there are plenty of things that I can
do that make it pointful to attribute that belief to me. Suppose
that walking into a room I learn that at least one other person in
it is in pain; and I infer (i.e. form the belief) that at least two
people in the room are in pain. Couldnt it be a good explanation
of this that I already believe that I am in pain? In that case, I am
in pain expresses what someone might be said to believe; why
then shouldnt it express what someone might be said to know?
Let us turn now to other peoples knowledge of my pain.
Wittgenstein said that if we are using the word to know as it
is ordinarily used (and how else are we to use it?), then other
people very often know when I am in pain. As an empirical
report on ordinary usage I dare say that this is true. It is certainly
much easier to think of likely scenarios in which an ordinary
speaker of English would say Johnny knows that I am in pain
than it is to think of ones in which he or she would say I know
that I am in pain.
But why think that what people say frequently must be taken
literally, or that, if so taken, it must be true? Imagine someone
arguing like this: If we are using the words to rise and to set
READING THE TEXT
133
as they are normally used (and how else are we to use them?),
then the sun rises and sets every day. The natural response to
this would be that there is a way of using the words to rise and
to set that does not slavishly adhere to the sprawling mass of
their ordinary usages but attempts to extract a concept from
some subset of their literal uses (as suggested at 2.3).
In fact doing so is itself no departure from ordinary usage. For
most ordinary but educated speakers will acknowledge such a
literal core: confronted with the astronomical evidence they will
concede that of course the sun doesnt really rise and set, and
that this is just a manner of speaking. But then why should we
not say the same about knowledge of other minds? After all,
confronted with the philosophical evidence, most ordinary
speakers will concede: Of course one doesnt ever really know
what another person is feeling when his tooth is rotten: I just
assume that he is feeling the same as I do. Wittgensteins point
therefore relies on there being some weakness in that philosophical
evidence, to which we now turn.
The traditional problem of our knowledge of other minds
arises directly from the Cartesian picture itself. According to
that picture, Jones can never observe my pain; he can only ever
infer that I am in pain from premises about my behaviour whose
truth he does observe: A. A. is exhibiting pain-behaviour, there-
fore he is in pain. But then the question arises: with what right
does Jones make that inference? After all he has never observed
it to have been confirmed by any of its instances: that is, he has
never been in a position to observe, on the one hand my pain-
behaviour, and on the other hand my pain. It is only observations
of that sort that would justify his concluding that very often A.
A.s pain accompanied his pain-behaviour, thereby warranting
the problematic inference. I am the only person who is in a
position to make observations of that sort. But that is not going
to be of any help to Jones: for even if I made those observations,
my report of the results A. A. is in pain on n% of the occasions
on which he exhibits pain-behaviour will not confirm the
problematic inference for Jones without the help of an equally
problematic inference from A. A. reports that he was in pain on
an occasion to A. A. was in pain on that occasion. For how
can Jones rule out the hypothesis that I call pain what he calls
pleasure (see 4.1.3)?
134
Of course what Jones can observe on the Cartesian picture is
his own case: and he can summarize the results in the form Jones
is in pain on m% of the occasions on which he exhibits pain-
behaviour. But however great m is, this might seem to be very
weak inductive grounds on which to extend the inference from
others pain-behaviour to their pain. It would be as rash as it
would be for one of the characters in PI 293 to infer, from the
observation of a black stag beetle in his own box, that everyone
elses box contained a black stag beetle.
There is therefore a reasonable case for thinking that Jones
does not really know whether I am in pain. This case certainly
recommends divergence from the ordinary use of the word
know, but not an arbitrary divergence. It is rather that it
combines principles about knowledge that ordinary speakers
would in any case acknowledge for instance, that if an infer-
ence yields knowledge then it must have a sound evidential
base with premises taken from the Cartesian picture for
instance, that Jones cannot observe my pain and seems to
establish that on a sense of know that respects these principles,
there is indeed no knowledge of other minds.
The appeal to the ordinary use of know at PI 246 therefore
does nothing to legitimize others knowledge of my sensations.
However Wittgenstein does raise a further difficulty with the
argument that such knowledge is impossible. This objection
strikes at the heart of Cartesian picture. It is that that picture,
and the scepticism about other minds that follows from it,
rest upon a misunderstanding of the ordinary concept not
of knowledge but of pain. I shall discuss this point at 4.2.4;
but a proper explanation of it requires a brief discussion of
Wittgensteins views on the self.
4.2.3. The self
On the Cartesian picture your understanding of I am in pain
involved a correlation on the one hand between pain and an
inner object, and on the other hand between I and an inner
subject your self, the owner of these states. You are in pain
whenever that object stands in the right relation to that subject.
Apart from Descartes himself the most important writer on
this subject is Hume, who notoriously pointed out that when he
enters into his own experience he never encounters any such
READING THE TEXT
135
owner but only the sensations themselves (Treatise I.iv.6).
Wittgenstein seems to have endorsed this reasoning in the
Tractatus:
If I wrote a book called The World as I found it, I should have
to include a report on my body, and should have to say which
parts were subordinate to my will, and which were not, etc.,
this being a method of isolating the subject, or rather of
showing that in an important sense there is no subject; for it
alone could not be mentioned in that book. (TLP 5.631)
But the reader should note that in the Tractatus he also believed
in a metaphysical self (TLP 5.641) that was not found in experi-
ence: rather, it was the thing whose mental acts were what infused
its symbolic thought-processes with meaning: the thing that
somehow correlates names with their referents by means of the
method of projection (TLP 3.11). Language is my language the
language that I infuse with meaning; so language itself is a sort
of immanent reflection of a thinking subject. That was why he
felt able to write:
The limits of my language mean the limits of my world . . .
The world is my world: this is manifest in the fact that the
limits of language (of that language which alone I under-
stand) mean the limits of my world. (TLP 5.6, 5.62)
28
By the time of the Investigations Wittgenstein had partially
revised his earlier attitude. On the one hand, he no longer agreed
that a metaphysical self was a necessary condition of meaning:
as we saw at section 3 the act of meaning or act of projection
was a fiction: what turns a dead sign into a meaningful symbol is
its use and not a hocus-pocus which can be performed only by
the soul (PI 454). On the other hand, he still sympathized with
the Humean argument expressed in the passage quoted from
TLP 5.631, or at any rate with its conclusion. That is, he thought
that when I say I am in pain there is no thing to which I attribute
that property:
When I say I am in pain [says the interlocutor], I do not
point to a person who is in pain, since in a certain sense I have
136
no idea who is. And this can be given a justification. For the
main point is: I did not say that such-and-such a person was
in pain, but I am . . . . . Now in saying this I dont name any
person. Just as I dont name anyone when I groan with pain.
Though someone else sees who is in pain from the groaning.
(PI 404)
I is not the name of a person (PI 410)
Why is I not the name of a person? One reason might be the
Humean thought to which I just alluded: that there is no self or
soul if by that we mean an immaterial subject of sensations. But
this argument is not quite sufficient: for it leaves it open that I
is the name of my body and that that is a person. On the other
hand, that proposal cant be right either, because if I were the
name of my body then I am feeling such-and-such could not be
uttered truly by someone who felt such-and-such outside of his
body, for instance in anothers body (PI 302a, PI 409) or at the
end of a stick (PI 626).
29
What then is the function of I in the context I am in pain?
It exploits the fact that the mouth that speaks and the hand that
writes are causally connected with unique closeness to a discrete
organic body. This allows for the existence of conventions by
which the behaviour of the mouth and hand can be especially
closely associated with, for example, ailments of that body . So
when a mouth says I am in pain we know which body needs
treatment the body that is attached to that mouth. Thus the
function of the word I in I am in pain is to attract peoples
attention to a particular body.
Now we have already seen that this connection need not be
invariable. One might certainly feel pain in some distant part of
the world from ones mouth; indeed that was the reason for
doubting that I names the body that feels pain. But were this
contingent fact never to obtain we should have no occasion to
speak of my pains or to say I am in pain. Were it the case that,
somewhat like the three witches in the story, we all took turns to
speak through a single mouth in central London, to which we
were each connected by radio receivers and radio transmitters,
this mouth would never say, for example, I am in pain or speak
of my pains at any rate there would be no point in its doing
READING THE TEXT
137
so. Instead it would say things like There is pain at 36 Ramsey
Street, Glasgow and then people would attend to the bodies in
that place. Here we are imagining certain very general facts of
nature to be different; doing so makes intelligible the formation
of very different concepts from ours (PI II, xii: p. 195).
But given that the facts of nature are as they are we can see
that the first-person present tense sentence I am in pain serves
the same function as a cry to draw peoples attention to a
certain body. And that is Wittgensteins account of it: I am in
pain does not describe a cry but replaces one (PI 244).
Bearing all of this in mind, let us now turn to the concept
of pain, and in particular to our expression of it by means of
third-personal sensation ascriptions.
4.2.4. Third-personal sensation ascriptions:
The negative account
What underlies the Cartesian idea about epistemic asymmetry is
an idea about what I am doing when I attribute sensations to
somebody else. This idea can be expressed as a thesis about how
we come to grasp such third-personal ascriptions. It is that each
of us initially grasps what it is to be in pain, say: that is, we
initially master the first-personal avowal of pain. And then we
come to grasp its third-personal counterpart via the reflection
that for Johnny to be in pain is for him to have just what I have
when I am in pain. On this very natural account there is nothing
especially mysterious about the move from an understanding
of the first-personal to an understanding of the third-personal
sensation language. It is just a special case of the more general
competence that we have and which we exhibit in other cases
of moving from an understanding of what is mine to an
understanding of what is anothers.
Thus Billy might first learn the difference between things
that are and things that are not his toys: he is in a position to
exercise certain rights over his toys that he is not in a position
to exercise over other toys. And then he comes to understand
that certain toys are and certain toys are not Johnnys toys by
a process of abstraction of himself from the concept of his toys:
namely, what he learns is that Johnnys toys are just those toys to
which Johnny stands in the very same proprietary relations that
he (Billy) stands to his toys.
138
According to the Cartesian the step from my pain to Johnnys
pain proceeds in essentially the same way. To understand sensation
terms is in the first instance to understand that they denote inner
objects to which you stand in a special relation these are my
pains, my afterimages, and so on; and one extends ones under-
standing to the third-personal case when one learns that Johnnys
pain denotes objects that are inner to Johnny, that is, objects to
which Johnny stands in the same relation as you do to your pains.
But recall our discussion in the preceding section. We saw
there that the function of I in the case of first-personal avowals
was not to name any thing but rather to draw peoples attention
to the mouth (and associated body) that says it. Now if that is
the function of I in contexts like I am in pain then it is evident
that mastery of the latter does not demand acquaintance with
or any other knowledge of some relation between a type of
inner object (the pain) and an inner subject (the self that has
it), from which latter one may then abstract so as to move from
a grasp of I am in pain to a grasp of Johnny is in pain. In order
that I am in pain function in the manner just sketched, it is
necessary only that mastery of it involve (roughly) a disposition
to utter it when and only when certain bodily events occur in the
body that is attached to your mouth. But this means that mastery
of I am in pain does not give you the resources to move by
abstraction to mastery of Johnny is in pain. For you have no
idea at all of what you are supposed to be abstracting from.
Thus Wittgenstein writes:
If one has to imagine someone elses pain on the model of
ones own, this is none too easy a thing to do: for I have to
imagine pain which I do not feel on the model of the pain
which I do feel. That is, what I have to do is not simply to
make a transition in imagination from one place of pain to
another. As, from pain in the hand to pain in the arm. For
I am not to imagine that I feel pain in some region of his
body. (Which would also be possible.) (PI 302)
And this is the point of the famous comparison with 5 oclock
on the sun:
But if I suppose that someone has a pain [says the interlocutor],
then I am simply supposing that he has just the same as I have
READING THE TEXT
139
so often had. That gets us no further. It is as if I were to
say: You surely know what It is 5 oclock here means; so
you also know what Its 5 oclock on the sun means. It means
simply that it is just the same there as it is here when it is 5
oclock. The explanation by means of identity does not
work here. For I know well enough that one can call 5 oclock
here and 5 oclock there the same time, but what I do not
know is in what cases one is to speak of its being the same
time here and there. (PI 350)
According to the Cartesian picture I observe both pain and
myself when I am in pain; what I am to do is then to suppose
that for Johnny to be in pain is for another subject to have that
same relation to that same object as I do when I am in pain.
Against this we can see in the preceding passages the con-
sequences of Wittgensteins rejection of the subject of pain:
since the I in I am in pain doesnt refer to any subject there
is no reason to think that we can understand Johnny is in
pain by regarding it as simply a variation upon the subject
but not upon the object of I am in pain. That is the point of
the comparison with 5 oclock on the sun: since our under-
standing of temporal expressions of the form It is 5 oclock
at x extends only to cases where x denotes a place on the
surface of the Earth, there is no reason to think that we can
understand It is 5 oclock on the sun by regarding it as simply
a variation upon the location but not upon the time of, for
example, It is 5 oclock at Greenwich.
4.2.5. Third-personal sensation ascriptions:
The positive account
How then do we understand third-personal ascriptions of
sensation? Wittgenstein writes that an inner process stands
in need of outward criteria (PI 580) and it is by the criteria of
pain its natural and involuntary behavioural manifestations
that we tell whether to ascribe it.
It is worth briefly saying something at this point about
Wittgensteins notion of a criterion. Let us distinguish between
two sorts of evidence for a given type of state of affairs. One
sort of evidence depends, for its status as evidence, upon an
observed correlation between itself and what it is evidence for;
140
that is to say, it would cease to count as evidence were we to find
out that a relatively high proportion of cases in which the
evidence was present were ones in which the state of affairs
was absent.
An example of this sort of evidence is the barometer, whose
prediction of rain is only defeasibly evidence for rain. If we
observed on enough occasions that the rain did not succeed the
barometers prediction of it we should eventually cease to rely
upon the barometer as a means for predicting rain. The fact that
this could happen illustrates the sense in which the barometers
behaviour is only defeasibly evidence for future rain.
But another sort of evidence does not depend upon its status
as evidence for something upon such an observed correlation
between it and the thing it is evidence for; its status as evidence
is a priori (or, as Wittgenstein says at PI 354, founded upon
a definition). Let us consider the status of pain-behaviour as
evidence for pain. What sort of evidence could make us doubt
that pain-behaviour (among humans) really was evidence for their
being in pain? The answer is: none. Instead we should doubt the
status as such of the evidence that indicated (e.g.) that there was
no pain in the presence of pain-behaviour, or that there was pain
in its absence.
Evidence of the first sort is what Wittgenstein means by a
symptom. Evidence of the second sort is what he means by
a criterion. (He contrasts the two at PI 354.) Note well that
a criterion for a state of affairs may itself be defeasible evidence
for that state of affairs, that is, not sufficient for it;
30
what is
indefeasible is its status as evidence.
And it is clear that pain-behaviour is a criterion of pain. For
given that our concept of pain is as it is nothing could count as
evidence against that evidential link. Such evidence might look
like this: when you hit somebody with a hammer he does not cry
out or make any antagonistic response; and he does nothing to
avoid hammer blows in the future. On the other hand when
you gave him strawberries for the first time he winces and then
retaliates (by hitting you), and he does exhibit strawberry-
avoiding behaviour thereafter. Nobody who observed this case
would conclude that his pain-behaviour was not after all evid-
ence of his pain. Rather, he would conclude that what hurts us
does not hurt him (and vice versa).
READING THE TEXT
141
Here then is Wittgensteins account of third-personal sensation
ascriptions. A speaker learns them by learning their criteria.
What you learn is the kind of behaviour that counts as pain-
behaviour (and in the case of other sensational states you learn
their criteria, which might, for example, be their causes or their
effects). And your mastery of the term pain involves being able
to distinguish pain-behaviour from other sorts of behaviour,
and treating the former as defeasible grounds for asserting that
the other person is in pain. To master the ascription of pain
on these grounds is to master our concept of pain, at least in
third-personal contexts.
Obviously there is a good deal more to be said about that
account, but we are now in a position to say something more
about the problem of our knowledge of other minds. In the
passage that distinguishes criteria from symptoms Wittgenstein
makes an analogous point in connection with visual impressions.
He writes:
We say, for example: Experience teaches that there is rain
when the barometer falls, but it also teaches that there is rain
when we have certain sensations of wet and cold, or such-
and-such visual impressions. In defence of this one says that
these sense-impressions may deceive us. But here one fails to
reflect that the fact that the false appearance is precisely one
of rain is founded on a definition. (PI 354)
The point here is not that our sense-impressions may lie, but
that we understand their language. (And this language like
any other is founded on convention.) (PI 355)
The application to other minds ought to be clear. The problem
was supposed to be that Jones cannot have any reason for think-
ing that I am in pain because he cannot observe my pain. The
answer is that it is part of the concept of pain that Jones can have
such reasons, namely, my pain-behaviour. And somebody who
sincerely doubts whether my pain-behaviour is at least grounds
for ascribing pain to me has also failed to grasp the concept.
That is because pain-behaviour is criterial for pain.
It is clear that this account of third-personal sensation ascrip-
tions completely upsets the Cartesian picture that gave rise
142
to the problem in the first place. But it is also clear that
it threatens to cut certain very natural lines of thought that
would otherwise continue to nourish the Cartesian conception.
Because Wittgensteins treatment of these lines of thought
connects these topics with other central themes in both the
Tractatus and Philosophical Investigations, I shall conclude by
considering two.
4.2.6. Pain is pain
The most obvious objection to Wittgensteins account is that it
has omitted something utterly crucial to our third-personal
ascriptions of pain. We have seen that one does not imagine
someone elses pain on the model of ones own (PI 302). But
Wittgensteins account seems to go too far in the opposite
direction. He seems to imply that one imagines someone elses
pain on the model of something completely different from ones
own. For we have two things: the pain that I feel when I say that
I am in pain, and the pain-behaviour that Jones shows when
I say that he is in pain. These two things are entirely different;
so it looks as if my grounds for making statements of the form x
is in pain are irremediably disjunctive.
So on Wittgensteins account of it, it is starting to look as
though the word pain is simply ambiguous between its first- and
third-personal uses. So nothing would be lost from my language
if I instituted different words for each. And Wittgenstein does
seem willing to draw this startling conclusion.
If I were to reserve the word pain solely for what I had
hitherto called my pain, and others L. W.s pain, I should
do other people no injustice, so long as a notation were
provided in which the loss of the word pain in other con-
nexions were somehow supplied. Other people would still
be pitied, treated by doctors and so on. It would, of course,
be no objection to this mode of expression to say: But look
here, other people have just the same as you! (PI 403)
Of course it is correct to say that they have just the same as me
(because it is correct, at least in English, to say they have pain
and correct to say I have pain) but that does not mean that
something in the objects themselves the intrinsic nature of
READING THE TEXT
143
their pains and our pains forces upon us a notation that
employs the same expression for both.
But does this not make pain comparable to bank? Of course
the ambiguity of pain would not quite be the same as that of
bank, because it is primarily in one kind of context (the first-
personal one) that pain has one meaning and it is primarily in
another kind of context (the third-personal one) that pain has
another meaning the one we learn by coming to appreciate its
criteria. On the other hand just one occurrence of bank can
have different meanings (I visited the bank). Perhaps a better
comparison would be between the word pain and the word
March (the name of the month and the order): the contexts in
which it takes one meaning have little to do with those in which
it takes another; still it is correctly described as a mere accident
of our language that the word March can in these contexts
denote quite different kinds of thing. Given Wittgensteins
semantics for pain, must we not then describe it most implaus-
ibly as a mere accident of (probably all) human languages that
any expression that has the first-person use of the English word
also has its third-person use?
One central thesis of the Tractatus entailed a positive answer to
this question. According to that book the grammatical behaviour
of a word was dictated by the nature of the simple object that it
denoted, and to know the object is to know all of its possible
occurrences in states of affairs (2.0123). Now the possibility of the
notation of PI 403 would indicate to the author of the Tractatus
that somebody could know what pain in that language denoted,
that is, what pain was, without knowing that other people, too,
might be in pain. But that means that the word pain as it occurs
in English has different modes of signification, and so belongs to
different symbols (TLP 3.323); and in the case of such accidental
ambiguities we must make use of a sign-language that excludes
them by not using the same sign for different symbols (TLP
3.325). That is, if we can use a language like that of PI 403 then
from a logical point of view we must.
In Philosophical Investigations Wittgenstein addresses this ques-
tion in a sequence that includes the following statement of it:
But how can I decide what is an essential, and what an
inessential, accidental, feature of the notation? Is there some
144
reality lying behind the notation, which shapes its grammar?
(PI 562)
A little later he raises the question in a way that has direct
application to pain. And he returns a preliminary answer:
Why the same word? In the calculus we make no use of this
identity! . . . But what does it mean here to speak of
making use of the identity? For isnt it a use, if we do in
fact use the same word? (PI 565)
But that is quite unsatisfactory. You might as well say that we
make use of the identity of the word for the depositing and
lending institution on the one hand and for the sloped ground by
a river on the other for dont we use the same word (bank) for
each of them? There is then another suggestion:
And now it looks as if the use of the same word or the same
piece [i.e. in a game] had a purpose if the identity is not
accidental, inessential. And as if the purpose were that one
should be able to recognize the piece and know how to play.
(PI 566)
So the point of our using the same word pain in both its first-
personal and third-personal uses is that somebody will be able
to pick up its use in one context (say, the third-personal) because
he recognizes it as the word that occurs in another (say, the first-
personal).
If one expunges from this suggestion the idea of a designing
consciousness as connoted by the word purpose one is left with a
fairly plausible suggestion. The point is that we use the same word
because someone who understands the word pain in its first-
personal context will, because of that use of it, naturally extend it
to the third-personal context in such a way that he ascribes what
he called pain to other people on the grounds that we actually
have for doing so (though for all that he is not logically compelled
to do so). That is: the extended similarity space that is centred
upon our feelings of pain is such that we are more likely to extend
our word for these feelings to people who are injured or cry out
than we are to those who exhibit other types of behaviour.
READING THE TEXT
145
This is an interesting empirical conjecture that cannot be
settled in the armchair; if confirmed it would indicate one
reason that one might have for regarding the use of the same
word in first- and third-personal cases neither as forced upon us
by the nature of things nor as a logically undesirable accident.
Rather, the facts about how we extend the uses of words that
have been taught to us in some limited sphere would be one of
those facts of nature by which we explain the formation of the
associated concept (PI p. 195/230).
4.2.7. God sees it
There is a very general objection to Wittgensteins procedure here,
one that applies not only at this point but also to his discussion
of language-games, family resemblance, reading, understanding
and meaning. That is because it latches onto a feature of the
method of Philosophical Investigations that runs through the
whole work and which distinguishes its approach very sharply
from that in the Tractatus.
If somebody wanted to know the meaning of a sentence
like Jones was reading at such-and-such time we have seen
how Wittgenstein would respond. He would offer an account
of when we are justified in saying that he was reading. We are
justified in saying it when he has said the written words right
(i.e.) 50 times consecutively (PI 157). In the recently introduced
terminology we can say that his doing so is a criterion for his
reading.
Again, we can be justified in saying that somebody who gives
the order Write down the sequence +2 means the sequence +2
and not some deviation from it. What justifies it is what he
counts as obeying the rule and going against it in actual cases
(PI 201). We may say that this too is a criterion of what he meant
by the order.
And the same pattern is evident in his account of third-personal
sensation ascriptions. The account tells us what justifies such
ascriptions, that is, their criteria. In the case of pain, these are
behavioural, but they need not be so long as they are outer. For
instance, there might be a sensation for which the criterion was a
physiological event, such as a rise in blood pressure (PI 270).
In all of these cases we are not told the truth-conditions of the
statements in question. We are only told when we are justified in
146
making them. Contrast statements whose truth-conditions we
do give. A statement of the form x is a square is true just in case
xs replacement denotes something that has four equal sides that
meet at right angles. We are here being told what claim upon
reality the statement, that something is square, makes. And this
is just what his later accounts of reading, meaning, understand-
ing, and third-personal sensation ascriptions appear to avoid.
We have already seen that in the Tractatus the statement of
truth-conditions was the one and only way to explain a proposi-
tion: for the essence of the proposition is to lay a claim upon
reality, to say that this is how things stand (TLP 4.5). It should
also be evident that the Cartesian account of third-personal
sensation ascriptions meets this demand. On that account you
can explain to me the meaning of Johnny is in pain to me by
saying, You surely know what it is for you to be in pain. Johnny
is in pain means simply that he has what you have when you
are in pain (cf. PI 350). Whatever the deficiencies of this account
it at least aspires to that Tractarian ideal; that is, it attempts
to specify what it would take for Johnny to be in pain, not just
the sort of thing that would justify my saying so. Whereas the
account that we are now considering does nothing of the sort: it
makes it appear as though somebody who was able to recognize
pain-behaviour knew all that there was to know about the
meaning of Johnny is in pain.
Well, it may be said, there is nothing particularly surprising
or objectionable about that. The whole point of Philosophical
Investigations was to break the spell of a highly unified concep-
tion of language, on which the analysis of a sentence conforms
in all cases to a single ideal. When the spell is broken we see that
there are indeed many different kinds of meaningful sentence, so
too many different kinds of explanation of a sentences meaning.
Why then should anyone be surprised or concerned that the
correct explanation of Johnny is in pain is so different in form
from that of Such-and-such is square?
The answer and this is the objection is that logic itself
seems to dictate otherwise. Either he is in pain or he isnt is an
instance of the theorem of classical logic known as the law of
excluded middle; so is Either he was reading that first word or
he wasnt (recall the example at PI 157); so is Either he meant
+2 or he didnt. And it seems that the application of the law in
READING THE TEXT
147
these cases shows that there is more to the ascription of reading,
meaning, understanding and pain than Wittgenstein allows. For
the law seems to demand that there be a fact one way or another,
even if we dont know what it is. We have to rely on these external
signs of pain, or understanding; but the statement itself reaches
into a hidden region of reality that in any case makes it true or
makes it false.
But nothing in Wittgensteins explanations of these statements
shows what sort of fact this could be: all it tells us is what justi-
fies us in those cases in which we can tell one way or another
whether, for example, he is in pain. Wittgensteins account there-
fore seems superficial. It seems that logic itself forces upon us a
conception of these statements meaning that is beyond the reach
of the sort of explanation that he would have us accept.
In two passages, one of great beauty, Wittgenstein gives
powerful expression to this concern.
Here it happens that our thinking plays us a queer trick. We
want, that is, to quote the law of excluded middle and to say:
Either such an image is in his mind, or it is not; there is no
third possibility! (PI 352)
A picture is conjured up which seems to fix the sense unam-
biguously. The actual use, compared with that suggested by
the picture, seems like something muddied. Here again we get
the same thing as in set theory: the form of expression we use
seems to have been designed for a god, who knows what we
cannot know; he sees the whole of each of those infinite series
and he sees into human consciousness. For us, of course, these
forms of expression are like pontificals which we may put on,
but cannot do much with, since we lack the effective power
that would give these vestments meaning and purpose.
In the actual use of expressions we make detours, we go by
side roads. We see the straight highway before us, but of course
we cannot use it, because it is permanently closed. (PI 426)
This passage also contains the seeds of Wittgensteins reply. We
must distinguish between the logical law itself, which is simply a
convention of our language, and the picture that it suggested to
us. The picture is of a region of reality that outruns its visible
148
part: the inside of someones mind, say, or the infinite continua-
tion of the decimal expansion of . But that picture is just a
picture: it tells us nothing about the actual use of the expression
but just stands there like an irrelevant or merely ornamental
illustration to a text.
Thus PI 352 continues:
The law of excluded middle says here: It must either look
like this, or like that. So it really and this is a truism says
nothing at all, but gives us a picture. And the problem ought
now to be: does reality accord with the picture or not? And
this picture seems to determine what we have to do, what to
look for, and how but it does not do so, just because we do
not know how it is to be applied . . .
Similarly when it is said Either he has this experience, or
not what primarily occurs to us is a picture which by itself
seems to make the sense of the expressions unmistakable:
Now you know what is in question we should like to say.
And that is precisely what it does not tell him.
31
The metaphysical picture, of a region of reality that outruns
its visible part, is inert: once it is distinguished from the law
of excluded middle it makes no difference to anything that any-
one actually says whether or not we accept that picture. So we
should of course accept the law itself; but we should not
think that it reveals any incoherence in the way we actually
employ third-personal sensation ascriptions, which is what
Wittgenstein takes himself to be describing.
Nor does this mean that the law of excluded middle should
itself become an object of philosophical scrutiny, as though
philosophers could settle whether or not it is true, in cases where
the evidence is equivocal, that either Jones was in pain or he
wasnt. To do so would be to have some independent means of
deciding whether Jones was in pain is either true or false in
cases where I have no means of telling which. But there is no
independent means for deciding that. Given Wittgensteins
conception of truth, It is true that Jones was in pain or it is
false that Jones was in pain is just another way of saying Either
READING THE TEXT
149
Jones was in pain or he wasnt and so cannot itself be evaluated
by some prior method (cf. PI 136a, discussed at 2.4).
Philosophy simply puts everything before us, and neither
explains nor deduces anything (PI 126). We have seen some
reasons for dissatisfaction with that general attitude at 2.3. But
given Wittgensteins conception of truth its application to the
present objection is entirely appropriate.
150
CHAPTER 4
RECEPTION AND INFLUENCE
Since its publication in 1953 Philosophical Investigations has been
widely recognized as a great work. Strawsons review called it a
treatment, by a philosopher of genius, of a number of intricate
problems, intricately connected (1954: 22), and Feyerabend wrote
that considered as a contribution to traditional philosophy
rather than as a criticism of that enterprise it was a great
achievement (1955: 149). More recently Michael Dummett has
written that its first 100-odd sections almost all compel assent;
although there may be large questions about how one should go
on from there, it is almost impossible to read those paragraphs
and maintain any reservations about this definitive treatment of
the topics with which they deal (1981: 239).
Its most notable early critic was Russell, who wrote that
The late Wittgenstein . . . seems to have grown tired of serious
thinking and to have invented a doctrine which would make
such an activity unnecessary. I do not for one moment believe
that the doctrine which has these lazy consequences is true. I
realize, however, that I have an overpoweringly strong bias
against it, for, if it is true, philosophy is, at best, a slight help
to lexicographers, and at worst, an idle tea-table amusement.
(Russell 1959: 21617)
While it is easy to sympathize with Russells puzzlement over why
anybody who adhered to Wittgensteins conception of philosophy
would attach much importance to it, it is difficult not to consider
this otherwise unfair. There is a good deal in Philosophical Investi-
gations that makes a substantial contribution to our understanding
of reality and in particular the nature of thinking, meaning
and understanding while making none at all to lexicography.
It is possible that Russells judgement was biased by a lumping
151
together of Wittgenstein with Ryle and the Oxford ordinary
language school under Austin (see Ayer 1984: 1336).
Unlike the Tractatus, which from around 1925 dominated the
thinking of that group of scientists and philosophers known
as the Vienna Circle whose members included Carnap, Schlick
and Neurath and which thereby in consequence of the 1930s
diaspora had a deep if indirect influence on the positivist
legacy in the United States, there was never any school that
drew its main inspiration from Philosophical Investigations.
It did however exert a strong influence upon a number of
individual philosophers; and, in the minds of those who were
both insightful enough to absorb Wittgensteins later philosophy
and strong-minded enough not to find it overwhelming, it
fostered some of the most original and important philosophical
work since Wittgensteins. Of these I shall mention three.
1
Michael Dummetts work on logic and anti-realism proceeds
from a Wittgensteinian premise among others. What he took
from Wittgenstein was the idea that there can be no private
semantic ingredient of a public language (see 4.1.3): if two indi-
viduals agree completely about the use to be made of a statement
then they agree about its meaning (Dummett 1975a: 216, 226).
He contended further that not every aspect of linguistic use was
relevant to the meanings of the expressions involved; rather,
some uses were so to speak definitive of those expressions: they
laid down the meanings, and further patterns of use might be
criticized for their infidelity to those meanings. In particular
the patterns of inference that characterize classical logic were
open to such criticisms: there is nothing in our overt application
of true and false to simple mathematical statements that
distinguishes those notions from the notions of proof and
refutation; nothing, therefore, that justifies the further classical
assumption that every statement is true or false; and so, finally,
no justification (of the familiar truth-tabular sort) for the
classical law of excluded middle (Dummett 1975a: 2256).
We have seen that Wittgenstein himself would have rejected
the idea that linguistic use was subject to criticism: it is for
philosophers to describe linguistic use, not to criticize it. This
difference between him and Dummett on this point can perhaps
be traced back to Wittgensteins insistence that the notion of
152
truth has no content except as a device of disquotation (see 3.4).
For if Wittgenstein had been right about that, it would have been
futile in any case to demand any justification of, for example,
Either he was reading or he wasnt in terms of the conditions
under which its disjuncts were true (see 4.2.7). Of course
Dummett rejects that redundancy theory of truth (Dummett
1959: 47).
We have already had occasion to consider Kripkes seminal
monograph (Kripke 1982) probably the most influential work
by an American philosopher to have been primarily inspired by
Philosophical Investigations. In it Kripke develops a sweeping
scepticism about meaning on the basis of the remarks on rule-
following that we considered in Section 3. According to it there
simply is no such thing as meaning, understanding, following a
rule, and so on. But he also presents what he calls, following
Hume, a sceptical solution to these doubts: it is still legitimate
for us to talk about somebodys meaning one thing rather than
another by his words. Only we must not think that in doing so
we are describing any facts. The scepticism itself, though not
the solution, has notable points of continuity with the work
of W. V. Quine on the indeterminacy of translation and the
inscrutability of reference. The main difference, as Kripke him-
self points out (1982: 57), is that Quines arguments proceed on
the assumption that no inner processes can endow meaning
upon otherwise meaningless signs; whereas among Wittgensteins
achievements are his arguments for this premise (see 3.3.2).
But we have also considered two differences between Kripkes
presentation of Wittgenstein and what the latter actually wrote.
It was of course Wittgensteins view that there are such things
as meaning and understanding and he was concerned only to
combat certain misconceptions of them. More importantly
Wittgensteins conception of truth strongly suggests that he
didnt think we could draw any line among assertions to separate
those that do from those that do not state facts; consequently
he would have denied that there was anything sceptical after all
about Kripkes sceptical solution (see 3.4.2).
John McDowells work deploys a number of recognizably
Wittgensteinian ideas in a wholly different way, in particular
certain strands of that romantic side of his work which we had
occasion to notice at PI 634 (see 1.3.4). Consider, for instance,
153
the idea of exhibition at PI 201: we saw that you can exhibit
your meaning to someone else, but he has to be someone who
belongs to your community in the sense of sharing your natural
reactions. McDowells own epistemology is a development of
this: on it, one can literally hear, for example, what somebody
in ones own community is thinking (e.g. that he wants you
to continue the sequence +2); whereas somebody outside of
that community can not: shared membership in a linguistic
community is not just a matter of matching in aspects of an
exterior that we present to anyone whatever, but equips us to
make our minds available to one another, by confronting one
another with a different exterior from that which we present
to outsiders (McDowell 1984: 253).
2
And this is just one of
many Wittgensteinian ideas that McDowell has taken up and
developed in his own way; others include the private language
argument and the Wittgensteinian conception of philosophy as
an essentially reactive, not constructive, activity. McDowell 1994
is the fullest presentation of the resulting synthesis.
As these philosophers have illustrated Wittgensteins later work
has had a powerful and largely beneficial effect on the modern
field; what they also illustrate to more varying degrees is that
this effect is most likely to be achieved, not through slavish
adherence to his own philosophical views or through an attempt
to practice philosophy in the manner that he recommended, but
through a critical engagement with those views in an attempt to
discriminate and to extract whichever of his doctrines make a
contribution to our philosophical knowledge. And that is the
spirit in which you too should approach that work.
154
CHAPTER 5
GUIDE TO FURTHER READING
In addition to the works mentioned in the main text, the reader
may find the following works helpful.
CONTEXT
The best short study of the Tractatus remains Ramseys review
of it (Ramsey 1923). But that is very difficult at a first reading.
Mounce 1989 and Kenny 1975 are both excellent introductions.
SECTION 1. THE AUGUSTINIAN PICTURE
Baker and Hackers monumental study of the Investigations
combines section-by-section textual exegesis with scholarly
essays on textual and philosophical questions raised by the text.
A new edition of its first volume has recently been issued (Baker
and Hacker 2004): chapter 1 of it is an excellent discussion of
this material. Michael Luntley makes a very interesting case for
an unorthodox reading of Wittgensteins attitude towards
Augustine: see his forthcoming and the references therein.
SECTION 2. FAMILY RESEMBLANCE AND THE IDEAL
OF PRECISION
Bambrough 19601961 is an attempt to apply Wittgensteins doc-
trine of family resemblance to the problem of universals. Forster
(forthcoming) is a thorough and insightful critical study of
Wittgensteins discussion. Wittgensteins doubts about the very
possibility of philosophy, doubts that I have isolated from my
interpretation of the rest of the work, are for some commentators
its interpretative key. I have largely ignored this kind of reading
because I do not consider it suitable for an introduction of this
sort. But Stern 2004 is an excellent, sensitive reading of Philo-
sophical Investigations in this Pyrrhonian light.
GUIDE TO FURTHER READING
155
SECTION 3. MEANING AND UNDERSTANDING
Heal 1989 chapter 9 discusses Wittgensteins views on proof and
meaning, as does Wright 1980 (which is as much a work on
Dummetts views as it is on Wittgensteins). Miller and Wright
2002 is a helpful collection of essays on Kripkes work on rule-
following.
SECTION 4. PRIVACY
The Postscript to Kripke 1982 contains lucid discussion of
Wittgensteins views on other minds. The last 15 pages of BB
contain helpful discussion of Wittgensteins views on the self;
and his Notes for a Lecture on Private Experience and Sense
Data, available in PO, contain very helpful background to the
Private Language Argument.
156
NOTES
1. CONTEXT
1 A note on the text. References in which PI precedes a number
are to section numbers of Part I of Philosophical Investigations. An
occasional succeeding letter indicates a paragraph within a section;
thus PI 1d refers to the fourth paragraph of section 1. I have used
the third edition. The pagination in that edition is slightly different
from earlier editions, so if I refer to a page number in the third edition
(which will happen when I want to refer to a remark in Part II or a
note in Part I) I also mention the corresponding page in the first
edition (1953), which matches that in the second edition.
It will soon become evident to anyone who reads it that at least
some of the sentences in PI do not express the views or attitudes of
Wittgenstein himself but those of an imaginary interlocutor. Where
the context does not make it clear I will use square brackets to indi-
cate the quoted sentences of which this is true.
References to the Tractatus will also be by section number. As
with references to Philosophical Investigations a succeeding letter
indicates a paragraph within a section.
3. READING THE TEXT
1 Causes and effects is my crude interpretation of language and the
actions into which it is woven (PI 7d).
2 The reader will notice that Wittgenstein mentions exceptions to this
rule and may be curious to know which ones he allowed. One sort of
exceptional case is that in which the meaning of a word is said to be
something that you experience (so that, for example, if you repeat a
word enough times it starts to sound meaningless). Wittgenstein
discusses cases of this sort at PI pp. 1825/21417. PI does not make
clear what the other exceptions are supposed to be.
3 Wittgenstein discusses a similar case at PI p. 160/187. There he says:
If you trained someone to emit a particular sound at the sight of
something red, another at the sight of something yellow, and so
on for other colours, still he would not yet be describing objects
by their colours. Though he might be a help to us in giving a
description. A description is a representation of a distribution
in spaces (in that of time, for instance).
But this looks dogmatic. At any rate no reason is given for deny-
ing it the status of a report on something; or for denying that the
NOTES
157
whole ostensive teaching plus consequent habitual response is
some kind of language-game.
4 This is roughly the account of ostension in Quine 1969: 1214.
5 Someone might object that I have failed to distinguish what
Wittgenstein calls ostensive definition from what he calls ostensive
teaching (PI 6b). The form of training that I have described in con-
nection with animals is only ostensive teaching; what distinguishes
an occasion of ostensive definition is that the pupil is able to ask
the name of the thing or type of thing that is then being defined.
So all that my example shows is that ostensive teaching does not
presuppose that the pupil can already speak a language; but what
Wittgenstein aimed to establish was the weaker thesis that ostensive
definition presupposes this.
. My two related replies are (a) that if we make it part of the
definition of ostensive definition that the pupil be able to ask some-
things name then of course it will be true that I gave X a successful
ostensive definition of red entails X could already speak a
language. But if this is all that Wittgenstein had aimed to show
then his argument loses all of its interest; (b) that neither Augustine
nor any of the empiricists who might reasonably be taken to be
targets at this point have relied upon any such distinction: for them,
the idea that ostension lay at the bottom of language was the idea that
ostensive teaching did so. (Augustine nowhere speaks of asking his
elders the names of the objects before him.) In short the distinction
between ostensive teaching and ostensive definition can only save
Wittgensteins argument at the cost of making it irrelevant.
6 For further discussion of this Notebooks material see Sullivan
2003.
7 It is true that in the written version of the language they have
the letter a in common, but this is not something of semantic
significance in the manner of (say) the ending o in Ambulo. In the
Tractatus it is our meaningful use of signs that turns them from
semantically inert to semantically significant elements of language
what he there called symbols. See TLP 3.262, 3.3263.328.
8 See, for example, the illuminating discussion of why we call facial
features friendly at BB 145.
9 PI 78 is puzzling in this connection, for in its comparison of how
the word game is used with how a clarinet sounds it appears to be
suggesting that both are things that one can know without being
able to say them. On the criterion of being able to say that is
operative in the text that seems to be simply wrong though
evidently on that criterion one can only say how the word game
is used to the right kind of audience. See further 3.4.3.
10 See also the helpful discussion in Fogelin 1987: 1334.
11 Of course Id also apply it (wrongly) to all cuboids, parallelepipeds,
and so on, but this makes no difference to the present point.
12 For further discussion see Hacker 1999; Stern 1995: 679.
NOTES
158
13 Cf. Evanss remark on this deep conceptual prejudice Evans
1980: 2767.
14 Wittgenstein does immediately concede that observation of regular
concomitances is not the only way to establish causation. He may
have had in mind Russells discussion in The Limits of Empiricism.
See Rheess note to Cause and Effect: Intuitive Awareness in
PO p. 370.
15 Isnt it wrong to call thinking of the formula an experience?
Wittgenstein has already shown (see 1.2) that there is no experience
of attending to the shape or attending to the colour. Couldnt he
argue similarly that there is none of thinking of the formula?
And indeed he does seem to endorse a similar view of what happens
when one thinks of a person (PI 691).
I think that in this passage he must mean thinking of the formula
to denote a type of experience. Otherwise the fourth sentence of PI
179, which speaks explicitly of such an experience, would be quite
irrelevant. And I think that he avoids any inconsistency because
when he speaks in this passage of thinking of the formula he just
means, for example, that B has a mental image of the sign for the
formula (or for that matter an outer experience of that sign). This
would explain his interpolation of the words saying it, writing it
down in the quoted passage.
16 Kripkes own, somewhat confusing, discussion of PI 136a is at p. 86;
I discuss it in Ahmed 2007: 1426.
17 For instance it would apply to ones knowledge of the ABC
(PI 149).
18 He takes a comparable attitude towards other forms of scepticism,
for example, that about other minds (PI 284, 2889, 303), induction
(PI 472) and the external world (OC 36971).
19 Locke himself falls short of being ideally Lockean in this respect.
At Essay II.xxxii.15 he says that there are many reasons for thinking
that other people feel the same chromatic sensation as I do when,
for example, we look at a marigold: but that being besides my
present Business, I shall not trouble my reader with them.
20 For instance at PI 256 he writes: But suppose I didnt have any
natural expression for the sensation, but only had the sensation?
And now I simply associate names with sensations and use these
names in descriptions. And at PI 257 he writes: What would it be
like if human beings shewed no outward signs of pain (did not
groan, grimace, etc.)? Then it would be impossible to teach a child
the use of the word tooth-ache. Well, lets assume the child is a
genius and himself invents a name for the sensation!
21 Two further objections might seem to apply at this point.
The first point is that merely saying cow in the (visible) presence
of cows is not enough to make cow denote cows because it faces an
underdetermination problem: to say cow in the visible presence of
a cow is to say cow in the visible presence of the surface of a cow.
So what makes it true that cow refers to cows rather than to their
NOTES
159
surfaces? I reply that every language-game, however complex, faces
an equally devastating underdetermination problem. The private
diarist faces no special difficulty here.
The second point is directed against my claim that if somebody
who is engaged in as simple a language-game as the builders can
be said to have names for her instruments, then so can the private
diarist. The difference between the builders and the private diarist,
it will be claimed, is that B has other responses to Slab! than just
the mechanical fetching of a slab. It may be, for instance, that B will
shake her head if she sees nothing slablike in the vicinity, and shrug
her shoulders if she sees, for example, items that might be slabs but
might equally be beams.
But there is no reason why the private diarist should not play a
similarly complicated language-game. It may be, for instance, that
he writes S in his diary-entries for days on which he definitely did
not have S. And it might be that the boundaries of the type are
indefinite, so that his behaviour is like that of someone whose use
of an inner sample is as described at PI 73. In that case he may,
for example, write S? in his diary to describe days on which he
experienced a borderline case.
22 See also Blackburn 1984: 299300.
23 Notable among these are Wright 1986, Canfield 2001 and of
course Kripke 1982, which however places no special emphasis on
PI 258, arguing instead that it is simply a special case of a more
general argument that had already been made by PI 202 (Kripke
1982: 3).
24 Note that that precise meaning for same public use is what makes
the first premise that a permutation of the inner objects associated
with our public words is immaterial to their public use very strong;
but it is also what makes this second premise very weak. And PI
293b does in fact establish that strong first premise; that is why
I said that the present interpretation of same public use gives
Wittgensteins argument the best possible chance.
25 This term is usually reserved for Quines doctrine, best stated by
him at 1968: 289: [T]here are no meanings, nor likenesses nor
distinctions of meaning, beyond what are implicit in peoples
dispositions to overt behaviour. As long as the two are not confused
there is no harm in appropriating the term in the present context;
doing so also highlights the closeness between the two philosophers
on this point.
26 Here I sketch a version of the argument at Quine 1981: 1920.
27 At this point it may be worth mentioning Freges argument that
if everyones words referred to his private ideas then there could
be no verbal expression of a dispute between two people. Freges
view seems to have been that if you say Nettles are always green
and I say Nettles are not always green, and if we each mean
some different private sensation-type by green, then we are not
really disagreeing any more than two people are disputing whether
NOTES
160
a ten-pound note is genuine, where each meant the one he himself
had in his pocket and understood the word genuine in his own
particular sense (Frege 1967: 29). But it suffices for genuine dis-
agreement between two parties that there be sufficient agreement
between them as to what sorts of evidence would settle the
truth-value of the disputed sentence. More generally the fact that
disputants refer with the same words to different things does
not forestall their framing disagreements in a public language:
for instance, a nominalist mathematician could certainly express
disagreement with a Platonist one over a mathematical theorem
stated in the usual terms.
28 The treatment of solipsism in the Tractatus to which the quoted
passages belong is notoriously difficult and its interpretation
highly contentious. The present interpretation of it is based upon
Kripkes reading, for more details of which see Kripke 1982: 1312
n. 13.
29 You might retort that in that case the place where I felt it counts
in virtue of just that fact as part of my body. But then we can
retort back that in that case the claim that I refers to my body
is equivalent to the tautology that I in I feel such-and-such refers
to whatever it is that feels something when I feel it; but to say that that
is what I refers to is not to say anything about what I refers to.
30 This is not the only interpretation of Wittgensteins term; its most
prominent rival (McDowells) differs from it on just this point.
See McDowell 1982, esp. p. 380.
31 McDowell 1978: 303 n. 16 is a helpful commentary on this passage.
4. RECEPTION AND INFLUENCE
1 These are certainly not the only three philosophers of whom
this is true. Among the most important of the remainder is
G. E. M. Anscombe, whose seminal Intention (1957) applied
Wittgensteinian insights to problems to do with intention,
action, decision and the will.
2 The connection with the romantic idea of a national character
is obvious. Contrast Hume: Would you know the sentiments,
inclinations and course of life of the Greeks and the Romans?
Study well the temper and actions of the French and the English:
You cannot be much mistaken in transferring to the former most
of the observations which you have made with regard to the latter.
161
REFERENCES
WORKS BY WITTGENSTEIN
BB The Blue and Brown Books. Oxford: Blackwell, 1960.
LFM Lectures on the Foundations of Mathematics: Cambridge
1939. Ed. Cora Diamond. Chicago: University of Chicago
Press, 1975.
NB Notebooks 191416. Ed. G. E. M. Anscombe and
G. H. von Wright, trans. G. E. M. Anscombe. Second
edition. Chicago: University of Chicago Press, 1984.
OC On Certainty. Ed. G. E. M. Anscombe and G. H. von
Wright, trans. G. E. M. Anscombe and D. Paul. Oxford:
Blackwell, 1969.
PI Philosophical Investigations. Ed. G. E. M. Anscombe and
R. Rhees, trans. G. E. M. Anscombe. Third edition.
Oxford: Blackwell, 2001.
PO Philosophical Occasions 19121951. Ed. J. Klagge and
A. Nordmann. Indianapolis: Hackett, 1993.
RFM Remarks on the Foundations of Mathematics. Ed. G. E. M.
Anscombe, R. Rhees and G. H. von Wright, trans.
G. E. M. Anscombe. Oxford: Blackwell, 1991.
TLP Tractatus Logico-Philosophicus. Trans. D. F. Pears and
B. F. McGuinness. London: Routledge, 1961.
Z Zettel. Ed. G. E. M. Anscombe and G. H. von Wright,
trans. G. E. M. Anscombe. Second edition. Oxford:
Blackwell, 1981.
OTHER WORKS
Ahmed, A. 2007. Saul Kripke. London: Continuum.
Ayer, A. J. 1954. Could there be a private language? Proceedings of the
Aristotelian Society, Supplementary vol. 28: 6376. Reprinted in
G. Pitcher, ed., Wittgenstein: The Philosophical Investigations: 25166.
London: Macmillan.
. 1971. Language, Truth and Logic. Second edition. Harmondsworth:
Penguin.
. 1984. Wittgenstein. Harmondsworth: Penguin.
Baker, G. P. and Hacker, P. M. S. 2004. Wittgenstein: Understanding
and Meaning. Part I: Essays. Oxford: Blackwell.
Bambrough, R. 19601961. Universals and family resemblances.
Proceedings of the Aristotelian Society 61: 20722.
Blackburn, S. 1984. The individual strikes back. Synthese 58: 281301.
Canfield, J. V. 2001. Private language: The diary case. Australasian
Journal of Philosophy 79: 37794.
REFERENCES
162
Carnap, R. 2003. The Logical Structure of the World. La Salle, Ill.: Open
Court.
Craig, E. J. 1997. Meaning and privacy. In R. Hale and C. Wright, eds,
Blackwell Companion to the Philosophy of Language: 12745. Oxford:
Blackwell.
Dummett, M. A. E. 1959. Truth. Proceedings of the Aristotelian
Society 59: 14162. Reprinted in his Truth and Other Enigmas (1978):
124. London: Duckworth.
. 1975a. The philosophical basis of intuitionistic logic. In H. E. Rose
and J. C. Shepherdson, eds, Logic Colloquium 73: 540. Amsterdam:
North-Holland. Reprinted in his Truth and Other Enigmas (1978):
21547. London: Duckworth.
.1975b. Wangs paradox. Synthese 30: 30124. Reprinted in his
Truth and Other Enigmas (1978). London: Duckworth.
. 1981. Frege and Wittgenstein. In I. Block, ed., Perspectives on the
Philosophy of Wittgenstein: 3142. Oxford: Oxford University Press.
Reprinted in his Frege and Other Philosophers (1991): 23748. Oxford:
Clarendon Press.
Evans, G. 1980. Things without the mind. In Z. van Straaten, ed.,
Philosophical Subjects: Essays Presented to P. F. Strawson: 76116.
Oxford: Clarendon Press. Reprinted in his Collected Papers (1985):
24990. Oxford: Oxford University Press.
Feyerabend, P. 1955. Wittgensteins Philosophical Investigations. Philo-
sophical Review 64: 44983. Reprinted in G. Pitcher, ed., Wittgenstein:
The Philosophical Investigations: 10450. London: Macmillan.
Fogelin, R. 1987. Wittgenstein. Second edition. London: Routledge.
Forster, M. Forthcoming. Wittgenstein on family resemblance concepts.
In A. Ahmed, ed., Wittgensteins Philosophical Investigations: A
Critical Guide. Cambridge: Cambridge University Press.
Frege, G. 1967. The thought: A logical enquiry. Trans. A. M. and
M. Quinton, in P. F. Strawson, ed., Philosophical Logic: 1738.
Frege [1892] 1960. On sense and reference. In P. Geach and M. Black,
eds, Translations from the Philosophical Writings of Gottlob Frege:
5678. Oxford: Blackwell.
Glock, H.-J. 1996. A Wittgenstein Dictionary. Oxford: Blackwell.
Hacker, P. M. S. 1999. Naming, thinking and meaning in the
Tractatus. Philosophical Investigations 22 (2): 11935.
Heal, J. 1989. Fact and Meaning: Quine and Wittgenstein on Philosophy
of Language. Oxford: Blackwell.
Hume, D. [1739] 1949. Treatise of Human Nature. Ed. with an analytical
index by L. A, Selby-Bigge. Oxford: Clarendon Press.
Kenny, A. J. P. 1975. Wittgenstein. Harmondsworth: Penguin.
Kripke, S. A. 1982. Wittgenstein on Rules and Private Language. Oxford:
Blackwell.
Locke, J. [1694] 1979. Essay Concerning Human Understanding. Ed.
P. H. Nidditch. Oxford: Clarendon Press.
Luntley, M. Forthcoming. Whats doing? Activity, naming and
Wittgensteins response to Augustine. In A. Ahmed, ed.,
REFERENCES
163
Wittgensteins Philosophical Investigations: A Critical Guide.
Cambridge: Cambridge University Press.
McDowell, J. 1978. On The reality of the past . In C. Hookway
and P. Pettit, eds, Action and Interpretation: Studies in the Philosophy
of the Social Sciences: 12744. Cambridge: Cambridge University
Press. Reprinted in his Meaning, Knowledge and Reality (1998):
295313. Harvard: Harvard University Press.
. 1982. Criteria, defeasibility and knowledge. Proceedings of the
British Academy 68: 45579. Reprinted in his Meaning, Knowledge
and Reality (1998): 36994. Harvard: Harvard University Press.
. 1984. Wittgenstein on following a rule. Synthese 58: 32563.
Reprinted in his Mind, Value and Reality (1998): 22162. Harvard:
Harvard University Press.
. 1994. Mind and World. Harvard: Harvard University Press.
McGinn, M. 1997. Wittgenstein and the Philosophical Investigations.
London: Routledge.
Malcolm, N. 1954. Wittgensteins Philosophical Investigations. Philo-
sophical Review 63: 53059. Reprinted in G. Pitcher, ed., Wittgenstein:
The Philosophical Investigations: 65103. London: Macmillan.
Miller, A. and C. Wright, eds. 2002. Rule-Following and Meaning.
London: Acumen.
Moran, R. 2003. Authority and Estrangement. Princeton: Princeton
University Press.
Mounce 1989. Wittgensteins Tractatus: An Introduction. Chicago:
Chicago University Press.
Quine, W. V. 1968. Ontological relativity. Journal of Philosophy 65:
185212. Reprinted in Ontological Relativity and Other Essays (1969):
2668. New York: Columbia University Press
. 1969. Natural kinds. In N. Rescher, ed., Essays in Honour of Carl
G. Hempel: 523. Dordrecht: Reidel. Reprinted in Ontological
Relativity and Other Essays (1969): 11438. New York: Columbia
University Press.
. 1981. Things and their place in theories. In his Theories and Things:
123. Harvard: Harvard University Press.
Ramsey, F. P. 1923. Critical notice of the Tractatus. Mind 32: 46578.
Russell, B. [1912] 2001. The Problems of Philosophy. Oxford: Oxford
University Press.
. 1959. My Philosophical Development. Woking: Unwin.
Stern, D. 1995. Wittgenstein on Mind and Language. Oxford: Oxford
University Press.
. 2004. Wittgensteins Philosophical Investigations: An Introduction.
Cambridge: Cambridge University Press.
Strawson, P. F. 1954. Review of Wittgensteins Philosophical Investiga-
tions. Mind 63: 7099. Reprinted in G. Pitcher, ed., Wittgenstein: The
Philosophical Investigations: 2264. London: Macmillan.
Stroud, B. 2001. Private objects, physical objects and ostension.
In D. Charles and W. Child, eds, Wittgensteinian Themes: Essays in
Honour of David Pears: 14362. Oxford: Oxford University Press.
REFERENCES
164
Sullivan, P. 2003. Simplicity and analysis in early Wittgenstein.
European Journal of Philosophy 11: 7288.
Wright, C. 1980. Wittgenstein on the Foundations of Mathematics.
London: Duckworth.
. 1986. Does Philosophical Investigations I, 258 suggest a cogent
argument against private language? In J. McDowell and P. Pettit, eds,
Subject, Thought and Context. New York: Oxford University Press.
165
INDEX
acquaintance (Russellian) 13,
6, 9
Ahmed, A. 158n. 16
analysis 3, 4, 12, 242, 29, 315,
4751
Anscombe, G. E. M. 160n. 1
anti-realism 151
Aristotle 66
assumption 31
Ayer, A. J. 116
Baker, G. P. 154
Bambrough, R. 154
bank 1434
beetle-in-the-box 1218
Berkeley, G. 66
Bismarck 901
Blackburn, S. 159n. 22
blood pressure 145
The Blue and Brown Books
(BB) 67, 70, 77, 155
brain 789, 824, 12930
calculations 63
Canfield, J. V. 159n. 23
Carnap, R. 107, 151
Cartesian picture, the 12930,
139, 141
causal connection 23, 8991, 97
chromatic words 20, 28, 54,
567, 108, 1223
complement 127
compositeness 26, 34, 367, 45
Confessions (St Augustine) 11
context principle (Frege) 5
contradiction 556
Craig, E. J. 109
criteria 13941
vs. symptoms 1401
cube, the 70
decision 12, 98
derivation 868
Descartes, R. 12930, 134
see also Cartesian picture
descriptions, Russells theory
of 24
dispositions 7684
duck-rabbit 57
Dummett, M. A. E. 55, 150, 152
Einstein, A. 66
epistemic asymmetry, the 1304
Evans, G. 22, 158n. 13
excluded middle, the law
of 14651
exhibit 1013, 153
experience
of attending to something 1819
of being influenced 91
of guidance 86, 90, 92
of meaning something 156n. 2
of the self 90
facts 72, 75, 99
family resemblance 8, 19, 407,
51, 69, 878, 92, 145, 154
and vagueness 457
Feyerabend, P. 150
fitting 69, 735
Fogelin, R. 157n. 10
Forster, M. 154
Frege, G. 1, 4, 534, 159n. 27
functional unity 12, 39
INDEX
166
games 19, 423, 47, 51, 102
gestalt 33, 48
Glock, H.-J. 119
God 145, 147
guess(ing) 1035
Hacker, P. M. S. 154, 157n. 12
Heal, B. J. 155
Hookway, C. 163
Hume, D. 90, 1346, 160n. 2
I 134, 136, 1389
see also self, the
ideas 12
inductive inferences 22
inexact 5960
inner objects 126
inner process 139
inner sample 56, 589, 1212
interpretration 101, 104
justification 1056, 116, 120
Kant, I. 1, 92
Kenny, A. J. P. 154
Kissinger, H. 24
knowledge 64, 77, 79, 82, 109,
1314, 141
Kripke, S. A. 98100, 152, 155,
158n. 16, 159n. 23,
160n. 28
language-game 12, 48, 115, 123,
159n. 21
Lectures on the Foundations of
Mathematics (LFM) 55
Locke, J. 1516, 107, 121
Luntley, M. 154
McDowell, J. 152, 160n. 30,
160n. 31
McGinn, M. 119
Malcolm, N. 11315
Marx, K. 66
memories 75
Miller, A. 155
Moran, R. 70
Moses 49, 50
names 36, 8, 1113, 16, 226,
28, 345, 49, 645
naturalism 106
necessary connection 91, 93
Neurath, O. 151
Newton, I. 14, 66
Nile 50
Nixon, R. 24
Notebooks 191416 (NB) 34, 102
number 467
obedience-conditions 31, 41
On Certainty (OC) 75, 106
ostension 1622, 3940, 65
ostensive definition 1618, 202,
44, 66, 108, 11113, 11819
private 108, 11113, 118
ostensive teaching 157n. 3,
157n. 5
ostensive training vs. ostensive
teaching 157n. 5
pain 113, 1313, 136, 1425
paradox (rule-following) 94106
perception 569
philosophy 61
the nature of 617
pianola 802
Pitcher, G. 1623
private language 10828
projection
method of 71, 122, 135
quantifier 3
Quine, W. V. 157n. 4, 159n. 25,
159n. 26
Ramsey, F. P. 154
reading 802
INDEX
167
reference 15, 23, 111, 124, 152
referentialism 2234
in ordinary language 234
simplicity and
compositeness 269
in the Tractatus 2434
Remarks on the Foundations of
Mathematics (RFM) 33,
56, 75, 90
Rhees, R. 158n. 14, 161
rule-following 107
Russell, B. 12, 5, 9, 49,
66, 150
St Augustine 11, 16, 157n. 5
Augustinian picture 1140, 54
scepticism 101, 134, 152
Schlick, M. 151
seeing as 58
self, the 1347
semantic behaviourism 1278,
159n. 25
sensation 111, 1201, 130
first-personal avowals 138
third-personal
ascriptions 13745
sense 310, 13, 312
vs. reference 3, 4
sense data 2
sepia 289
sequence, numerical 80, 95, 106
set theory 147
similarity space 120, 144
simplicity 269
Socrates 34
Sorites (sequence) 54, 55, 100
sortalism 113, 11819
spirit 40
Stern, D. 154, 157n. 12
Strawson, P. F. 11415, 150
Stroud, B. 119
Sullivan, P. 157n. 6
Sun, 5 oclock on it 139
symptoms 1401
Tractatus Logico-Philosophicus
(TLP) 1, 39, 1112,
2441, 435, 489, 58,
603, 679, 72, 91, 135,
1423, 1456, 151, 154,
157n. 7
objects 3, 6, 9, 246
picture theory
(of meaning) 6, 36
propositional forms 679
solipsism in 160n. 28
symbol 72, 157n. 7
vs. sign 72, 135, 143
syntax 62
types, theory of 61
tricolor 33, 47
truth 8, 66, 68, 100, 1489, 152
truth-conditions 4, 31, 41, 48, 60,
1456
understanding 69
grasping in a flash 6970
and guidance 8494
universals 2, 3, 27, 154
use
meaning as 18, 30, 6970
public use 125
vagueness 4551, 57, 88
and analysis 4751
verificationism 113, 118
water 778, 80
will 91, 160n. 1
Wright, C. 155, 159n. 23, 1624
yellow ochre 56
Zettel (Z) 83

Arif Ahmed Wittgensteins Philosophical Investigations A Readers Guide 2010

Uploaded by

Copyright:

Available Formats

Arif Ahmed Wittgensteins Philosophical Investigations A Readers Guide 2010

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Arif Ahmed Wittgensteins Philosophical Investigations A Readers Guide 2010

Uploaded by

Copyright:

Available Formats

WITTGENSTEINS PHILOSOPHICAL

You might also like