Chapter 2

On the Difference between Empirical and

Conceptual Issues

If psychologists are to discuss conceptual issues with the necessary competence

there are certain philosophical techniques with which they need to be familiar.
In the next two chapters we shall try to indicate what these techniques are and
the kinds of thing which they can achieve.
We claim no originality for any philosophical ideas which these chapters
contain. Indeed, as far as philosophy is concerned, the points which we shall be
making are commonplace; and while it may be true that philosophers sometimes
disagree among themselves as to how and when particular arguments should be
used (for example, the argument that such-and-such is ‘not what we say in
ordinary language’), anyone trained in this area of philosophy at least knows
what techniques are available. Our purpose, therefore, is not to offer any
distinctive contribution to philosophy but simply to set out the requisite
background knowledge for the benefit of readers who are unfamiliar with the
relevant literature. Among useful expository books may be mentioned in
particular those of White (1967), Warnock (1969), Flew (1971), and Bontempo
and Odell (1975). In addition many of the discussions to be found in Wisdom
(1953) make fascinating reading.
As we shall see later, disagreement on philosophical issues (in our sense) is not
impossible, but careful reflection is needed in that case as to what the
disagreement is about; and there are in fact many philosophical insights which
when stated are obvious. We are not, therefore, setting ourselves up as members
of any particular philosophical ‘school’ or as defenders of any particular
philosophical ‘thesis’.
Some may say that the kind of philosophy which we are doing is ‘empiricist’ in
character. This description, however, seems to us unsatisfactory. An
‘empiricist’, as we understand the term, is one who attaches some special kind of
importance or primacy to what philosophers have called ‘empirical statements’
(see below) as opposed to statements of other kinds; and it is hard to see what
sort of importance or primacy is intended or how such a claim could be justified.
It will in fact be a central thesis of this book that language can be used for many
different purposes, and we know of no good arguments for saying that empirical
statements (however defined) deserve some sort of privileged status.
 11

It may also be said that we are advocating a special type of philosophy often
known as ‘linguistic’ philosophy. Here, too, we question the value of the label. It
is true that some philosophers in recent decades, in their discussion of
philosophical problems, have made a self-conscious attempt to study certain
aspects of the functioning of language. The appropriateness of the term
‘linguistic’, however, is itself a philosophical question. It is not easy to find
indisputable grounds for distinguishing ‘linguistic’ philosophy from other kinds
of philosophy; and indeed, as Price (1945) has aptly pointed out, a tra¬
ditional philosopher who believes himself to be discussing ‘the nature of self
is not necessarily doing anything very different from those more recent
philosophers who might claim that they were carrying out a linguistic analysis of
sentences containing the word ‘I’. In any case, whether the words ‘linguistic
philosophy’ are helpful or not, it needs to be made clear that those who make use
of the techniques described in the next two chapters are not thereby committed
to any particular set of philosophical views. Indeed to say ‘If you are a linguistic
philosopher you must believe so and so’ is extremely naive whatever the ‘so and
so’ may be. It is true that implicitly we are making the claim that the techniques
in question can be exploited to good purpose, or at any rate that it is justifiable
to consider what can be done with them; but we do not do so in any spirit of
philosophical dogmatism, nor are we committed to the view that exploitation of
these techniques is the only right way to do philosophy.

We shall begin by calling attention to the need (emphasized in the quotation

from Berkeley at the beginning of our book) for distinguishing things that are
different. Not only is chalk different from cheese, and a tiger from a lion, but of
special interest for present purposes is the differences between different types of
statement. (It is of course true that words can be used for many purposes other
than ‘stating’ something, e.g. ‘Go away!’, ‘I agree’, ‘Alas!’, etc.; but for present
purposes this point need not concern us.) Thus the statements ‘There is some
cheese in the larder’ and ‘There are hundreds of lions in the game reserve’ are like
each other in respects in which both are unlike ‘50 + 50 = 100’. Many philo¬
sophers have attempted to classify statements into types. Thus Plato {Republic,
vii, 529) emphasizes that the truths of mathematics are different from the
discoveries which result from common-sense observation, while Aristotle
{Nichomachean Ethics, I, iii, 4) says: ‘It is the mark of a well-trained person to
require, in each kind of inquiry, just so much exactness as the subject admits of:
it is equally absurd to accept plausible argument from a mathematician and to
demand logical proof from an orator.’ Many centuries later Hume pointed out
that ‘experimental reasoning concerning matter of fact and existence’ was
different from ‘abstract reasoning concerning quantity and number’ and that
both alike were different from the speculative metaphysics of the schoolmen {An
Enquiry Concerning Human Understanding, xii, 3); and soon afterwards Kant,
in the Critique of Pure Reason, showed how statements about God, freedom,
and immortality were different from the statements of the physicist, e.g. ‘Bodies
fall when their supports are taken away’, and different again from statements
12

about the nature of space and time, e.g. ‘Two straight lines cannot enclose
a space’.
It is widely agreed nowadays that statements, along with biological
specimens, illnesses, personality traits, types of weather, and much else, do not
all fit tidily into particular classificatory schemes. There seems rather to be a
complex web of similarities and differences. Statements—like things and
people—can be like one another in some respects and not in others; and because
of the diversity of use to which language can be put, it is perhaps wise to bear in
mind the words of J. L. Austin (1962, p. 3) who suggests that philosophers
should‘abandon the . . . worship of tidy-looking dichotomies’. If anyone doubts
the value of this warning, let him attempt to classify under a small number of
heads the following utterances: ‘There’s Helvellyn’, ‘There’s the door!’, ‘Boys
will be boys’, ‘Arsenic is poisonous’, ‘Jealousy is cruel as the grave’, and ‘All
animals are equal but some animals are more equal than others’.
This does not mean, however, that no classifications are possible at all; and
our next task is to attempt to characterize the various differences which
philosophers have had in mind when they have distinguished empirical state¬
ments from conceptual ones. It is important, of course, not to ascribe to this
distinction a greater importance than it deserves, but we hope that the discus¬
sions which occur later in the book will do something to establish its utility.
To illustrate the distinction let us consider the following statements:
(la) ‘Smith is intelligent’, (lb) ‘To be able to remember digits correctly is not
necessarily a sign of intelligence’; (2a) ‘Smith is poorly motivated’, (2b) ‘Motives
are not causes’; (3a) ‘These tomatoes are red’, (3b) ‘Tomatoes are vegetables’;
(4a) ‘Memory declines in old age’, (4b) ‘There is no such entity as "'the
memory” ’; (5a) ‘Public opinion has veered round in support of the Prime
Minister’, (5b) ‘There is no such entity as ‘‘public opinion” ’.
Now there are, of course, many respects in which the (a)-statements differ
from each other. Thus (la) and (2a) are statements about particular people,
whereas (4a) is a statement about memory in general; moreover (3a) can be
checked by a single observation, whereas (la), (2a), and (5a) all require a series
of observations. Similarly the (b)-statements differ from each other in a variety
of ways. In the discussion which follows, however, we shall be concerned not
with these differences but with the ways in which the (a)-statements are like each
other and are different from the (b)-statements.
For our purposes the most important point about the (a)-statements is that
they all admit of being verified or falsified as a result of observation by the
senses. To find out if Smith is intelligent or poorly motivated one observes him;
similarly, to find out if certain tomatoes are correctly described as ‘red’ one
observes them; to find out what happens to memory in old age one observes
elderly people, and to find out if public opinion has veered round in support of
the Prime Minister one observes a whole range of popular reactions when the
Prime Minister’s policy is discussed. Such statements are called ‘empirical’
because they are based on observation—on what we experience, empeiria being
the Greek word for ‘experience’.
A further important point about the (a)-statements is that they admit of what
 13

is called ‘operational definition’. In other words their truth or falsity can be

determined definitively as the result of a specifiable set of procedures or
‘operations’. For example, to find out if Smith is intelligent one might go
through the operations of giving him a specified intelligence test, and if he
responds in certain ways then, provided the operational definition is agreed, it
follows as a matter of logic that he is intelligent, since to be intelligent just is to
respond in those specified ways.
We may note in passing that in ordinary language the operations implicit in
words or expressions such as ‘intelligent’, ‘poorly motivated’, etc., are not
specified to the last detail. To adapt a term used by Waismann (1945), ordinary
language, unlike most technical languages used in science, is‘open-textured’. In
this particular case, for example, there are all kinds of different things that
Smith might do which would entitle one to say that he was intelligent or poorly
motivated, and similarly with many other adjectives of ordinary speech, e.g.
‘forgetful’, ‘cheerful’, ‘honest’, etc. This lack of specificity is for many purposes
an advantage: one can indicate a range of things which Smith is liable to do or
has a propensity to do without being tied down to particular examples. There
must, of course, be something about the situation which gives grounds for the
use of such adjectives, since otherwise one would not be using them
meaningfully; but that ‘something’ can be different on different occasions. In
contrast there may be contexts, particularly those in which a scientific
investigator carries out experiments which he wishes to be repeatable, where the
operations need to be formulated in detail and with full precision. The
importance of defining terms operationally may well vary from one context to
another.
The (b)-statements, in contrast, do not necessarily predict that a specifiable
occurrence will or will not take place, but they are concerned rather with the
question, ‘What counts as such and such? Thus in the case of (lb) there may be
no dispute over the fact that Smith is able to remember digits correctly: the issue
is whether ability to remember digits correctly counts as ‘being intelligent’.
Similarly in the case of (2b) there may be no dispute as to whether Smith was or
was not poorly motivated; the issue here is whether, or in what sense, actions
carried out from a certain motive, e.g. jealousy, can count as being ‘caused’ by
that motive. Similarly it is being stipulated in (3b) that tomatoes should count
as—or be classified as—vegetables, while in (4b) and (5b) the claim is that if
certain happenings take place these on their own (without reference to any kind
of enduring ‘entity’) should count as instances of memory or of a change in
public opinion.
The (b)-statements have been characterized as ‘conceptual’ because in an
important sense they are concerned with what philosophers call ‘concepts’. Our
particular examples dealt in fact with the concepts of intelligence, motivation,
vegetable, memory, and public opinion. This does not mean, as is sometimes
mistakenly supposed, that the issues simply relate to the correct or incorrect use
of certain English words. This can be shown by pointing out that the same issues
would arise in the case of words of equivalent meaning in any language at all; it
makes sense to speak of English words but it is difficult to see what sense could
14

be attached to speaking of ‘English concepts’. Still less are conceptual

discussions concerned with choice of words on aesthetic or stylistic grounds.
They are concerned rather with what has been called the ‘logical behaviour’ of
concepts, that is, with what is entailed (in the logicians’ sense), or ought to be
entailed, by their use. Thus one is making a conceptual point if one asserts that
‘Smith is an uncle’ entails ‘Smith is male’, since one is indicating what is entailed
by the concept being an uncle. What has traditionally been called ‘logic’ can
helpfully be regarded as the study of the logical behaviour of a restricted group
of concepts, of which the most important are if, all, some, and not.
There is a complication here, however, in that the truths of logic are normally
said by philosophers to be necessary truths, in contrast with the contingent
truths afforded by empirical statements. A necessary truth, on this showing, is
one whose opposite involves a self-contradiction; thus it would be self¬
contradictory to assert that Smith was an uncle if at the same time one denied
that he was male. It does not follow, however, that we have now entered an area
where the truth has been established beyond any possibility of doubt, since it is
hard to see how the notions of‘doubt’ and ‘certainty’ are applicable in a context
of this sort. Indeed, if a person professed to dispute or doubt whether ‘A is an
uncle’ entails ‘A is male’ one would simply conclude that he had misunderstood
the meaning of one of the key terms, viz. ‘uncle’, ‘entails’, or ‘male’. Moreover,
while there may admittedly be situations in which a statement about the logical
behaviour of a concept has only to be made to be accepted as obvious, there may
also be situations (as we shall see more fully later in the chapter) where such state¬
ments are mistaken. It is therefore misleading to say that all conceptual state¬
ments involve necessary truths, even though some of them are not open to
serious dispute.
Conceptual issues are important, we suggest, precisely in so far as correct
classification is important. Not only do conceptual errors result in unnecessary
muddle, but correct conceptual formulations can lead to all kinds of important
insights. For example, once it is appreciated that there are many different
happenings which can lead us to say that a person is intelligent, there is no longer
any temptation to look for the ‘true nature’ or‘real meaning’ of intelligence or to
lament that we have not so far discovered it. More important, many major
scientific advances have arisen as the result of proposals for reclassification.
Sometimes this involves the coining of new technical terms and sometimes the
use of existing words in a new way. Thus HjO, impedance, and neutron are
examples of the former, while the Copernican revolution in astronomy, which
involves the claim that the earth, no less than Venus and Mars, should be
classified as a planet, is an example of the latter. Similarly Darwin’s great insight
can be viewed as an example of conceptual innovation: he is saying that what
happens in nature should count as a case of selection. In addition, though his
view has not received universal acceptance, Freud could be interpreted as stipula¬
ting that when a child sucks at the breast this should be classified as a case of
sexual activity.
There is, of course, an important difference between asserting that X is a case
 15

of Y and stipulating, or proposing, that X should be regarded as a case of Y. The

person who asserts that a tomato is a vegetable would in most contexts be
assumed simply to be indicating what is already implicit in the concept of a
vegetable; in contrast, since an infant’s sucking at the breast is not by definition a
case of sexual activity, one is here confronted with a proposal that the concept of
sexuality be extended so as to include infantile sucking at the breast. Within
conceptual statements, therefore, it is helpful to distinguish two kinds, viz.
(i) those which report on the logical behaviour of concepts, and (ii) those which
commend proposals for conceptual change. The former may be said to involve
conceptual analysis, the latter conceptual revision. Since the boundaries at
which a concept ceases to be applicable are not always clearly defined, there
may, of course, be situations where one is unsure if the speaker is reporting on
existing boundaries or stipulating where a boundary should be drawn. For
example, if someone says, ‘To be able to repeat digits correctly is not necessarily
a sign of intelligence’, one is unsure whether this should be counted as part of an
attempt to analyse the existing concept of intelligence or as a proposal for
revising it. There may even be situations where a speaker or writer believes
himself to be doing the former when he is unwittingly doing the latter.
In Chapter 5 we shall argue that operant psychology involves a series of
proposals for conceptual revision.
It is important to bear in mind that conceptual issues can arise not only in
science but also in matters of law. Thus, in the case of (3a), although it is at
present a trivial matter whether a tomato is classified as a fruit or as a vegetable,
if in fact vegetables but not fruits became subject to a special new tax it would
clearly be in the interests of tomato sellers to obtain a ruling that a tomato is a
fruit. This, of course, is not just ‘a dispute over words’, though a discussion
about the meaning of the word ‘vegetable’ is part of what is involved. A court of
law, however, does not simply make an arbitrary decision; if it had to decide a
matter of this sort it would have to take a number of factors into account,
including past rulings and relevant similarities and differences. Flew (1951,
pp. 3^) cites a case which involved a dispute as to whether or not a flying boat
counted as a ‘ship’; and one can see how it would be open to one counsel to point
out the similarities between flying boats and ships in the standard sense and to
the other counsel to point out the differences. The final ruling was in fact that a
flying boat is not a ship. Wittgenstein has noted that uncertainty where
boundary lines come is not necessarily due to ignorance, since in some cases‘We
do not know the boundaries because none have been drawn’ (1953, section 69).
It follows, of course, that the answer to any question of the form ‘Is an X a Y?’
may sometimes be ‘Yes and no’, since Xs may be like standard cases of Y in some
respects and not in others. When all the similarities and differences are agreed
there can be no further argument, and in these situations one needs to bear in
mind a further dictum of Wittgenstein (1953, section 79), viz.‘Say what you like,
as long as it does not stop you from seeing how things are.’ It is important to
note, however, that he then adds: ‘And when you have seen this there is plenty
that you will not say.’
16

This ‘say-what-you-like’ formula is one to which we shall be returning in later

chapters, since there are many conceptual issues in psychology where it is
applicable. In the meantime it may be helpful for purposes of illustration to cite
an example from a different area in the philosophy of science. There are obvious
pressures from astronomy which indicate that ‘The earth goes round the sun’ is a
more helpful statement than ‘The sun goes round the earth'. It does not follow,
however, that it is merely foolish or obscurantist to prefer the latter; the person
who says this may simply be making an intentional choice to continue to use the
sense of‘going round’ to which he has been accustomed; and if he is interested,
for instance, in siting the windows of his house he will not be misled. It would, of
course, be quite a different matter if he were in charge of organizing a journey
into space. The great merit of the ‘say-what-you-like’ formula is its recognition
that a form of words may be adequate in one context and not in another. On this
showing, an appropriate answer to the question, ‘Does the sun really go round
the earth?’ is ‘Yes and no’; and as long as one has appreciated all the reasons for
saying ‘yes’ and all the reasons for saying ‘no’ there is nothing more to discuss.
In lighter vein, here is a masterly description of the reactions of some 11-year-
olds to the conceptual claim that ‘mathematics is a language’.

‘A language! You mean they actually talk in it?’

‘Well, not people like us, perhaps. But learned professors and old geezers
like that can understand one another in maths, even if they don’t all speak
English’.
The idea of mathematics as a language was something that Temple found
hard to resist. He gave Martin-Jones a friendly punch in the ribs and said: ‘I
say, let’s be two learned professors talking in maths’.
‘Righto’. Martin-Jones took up his cue and spoke in tones of bogus
importance. Good morning. Professor Temple! a^ T 2ab T b^ don’t you
think?’. . .
‘Sir, please, sir, can you speak maths, sir?’
Mr Wilkins paused in the act of reaching for the chalk. ‘Can I speak
whatT
‘Darbishire says mathematics is a language. He’s bonkers, isn’t he, sir?’
‘So it is a language, in a way’, Mr Wilkins agreed. ‘It’s a means of
communication’.
Temple was interested. ‘Wow! Would you say something in
mathematics, please, sir? Say Good afternoon, it’s a lovely day to-day” in
algebra!’
Mr Wilkins tut-tutted in despair. ‘You silly little boy! You can’t say
things like that’.
‘No?’ Temple sounded disappointed. ‘Well, in that case it isn’t really a
language, is it, sir? (From Leave it to Jennings by Antony Buckeridge,
Collins, 1963, pp. 77-8 and 105)

This is a typical ‘yes-and-no’ situation; and on those occasions where we have

 17

reservations about whether X is a Y we sometimes say that it is not a ‘genuine

case of Y’, is not ‘strictly speaking a case of Y’, ‘is not a Y in the full sense’, or (like
Temple in the above passage) ‘is not really a Y’.
Now it is sometimes suggested that philosophers are arrogant and dogmatic
people who announce that it is forbidden to talk in certain ways. There is
nothing necessarily arrogant, however, in commending fresh classifications or
proposing that new boundaries should be drawn, and unless such proposals are
made without giving supporting reasons it seems scarcely fair to make
accusations of dogmatism. Often, too, it may simply be that after taking
seriously a particular argument one no longer wishes to talk in certain ways;
and it may even be possible that those who none the less persist in such talk have
failed to see the force of the argument in question.
In many cases conceptual analysis involves making a comparison between
sentences which are grammatically similar but logically different. Thus ‘Tame
tigers growl’ is grammatically similar to ‘Tame tigers exist’, since both sentences
contain adjective, noun, and verb. The two are logically different, however,
since ‘Some tame tigers growl and some do not’ is an intelligible sentence,
whereas ‘Some tame tigers exist and some do not’ is a logically impossible
combination of words. Similarly, ‘There is no such entity as “public opinion’’ ’ is
grammatically akin to ‘There are no such things as centaurs’; but whereas the
latter is an empirical matter and relates to what one will or will not discover after
searching, the former does not involve searching at all but is a statement about
the interchangeability of concepts. A particularly helpful notion in this
connection—and one which will play an important part in the present book—is
that of‘translation’. We are thinking here not oftranslation in the familiar sense,
in which one translates from, say, French into English, but of transformation of
a particular expression into one which is logically equivalent. Thus those who
assert that there is no such entity as ‘public opinion’ are in effect making the
conceptual claim that sentences purporting to be about public opinion require
translation. For example, ‘Public opinion has veered round in support of the
Prime Minister’ would need on this view to be rewritten as ‘There are more
people who now believe in the Prime Minister’s policy than there were
previously’. Mabbott (1947, pp. 151-152), from whom we have taken this
example, says: ‘The work of “analysis” consists not of analysing a non-existent
entity but of rewriting the sentences so that the misleading subject is replaced by
real subjects.’ Similar translation would be needed of sentences about the
growth of inflation or changes in fashion.
We suggested earlier that conceptual statements did not carry any predictions
as a result of which their truth or falsity could be straightforwardly checked.
This now needs qualification. If we look at our data in new ways we are often
enabled to make sense of other facts which hitherto had seemed quite separate.
To quote a well-known example, the concept of attraction exhibits the relation¬
ship between two such apparently disparate facts as the falling of an apple from
a tree and the movement of tides. (There is a particularly interesting discussion
of this example in Berkeley, Principles of Human Knowledge, sections
18

103-105.) In addition, however, if we take seriously a particular concept-for

instance that of attraction—we may well be able to discover facts additional to
those which we knew already. Conceptual proposals are thus not unrelated to
particular facts, since the more new facts we discover and the more sense we are
able to make of those facts which we already know, the greater our degree of
confidence that the conceptual proposal which led to the discoveries is a helpful
one. A particular perspective or approach is seldom decisively refuted by a single
experiment, but if experimental results continue to be indecisive or
uninteresting that approach tends gradually to be abandoned, while in other
cases a concept turns out to be so important that it gains universal acceptance.
Thus the phrenologists’ concept of philoprogenitiveness has been entirely
abandoned, whereas the concept of natural selection is never called in question.
Breakthroughs in the history of science are regularly associated with conceptual
innovation.
To bring out further the difference between conceptual and empirical state¬
ments it may be helpful to consider the ways in which each can be wrong. In the
case of empirical statements the matter is straightforward. Thus a person might
make the empirical claim that it is raining and be mistaken. (There would, of
course, be a conceptual problem if there were just occasional drops and one was
unsure if this would really count as ‘raining’, but this kind of problem arises in
any situation where there are marginal cases.) Similarly a person can be wrong
over the empirical generalizations which he makes. For example, in criticizing a
colleague’s research one does not normally dispute the particular facts which he
claims to have discovered, since in most contexts one assumes that he has not
miscalculated or misreported, but it may sometimes be possible to put forward
alternative explanations of the same facts and thus show that some of his
generalizations are faulty.
Disagreement over conceptual issues, however, may sometimes be a more
complex affair. Here, for illustration purposes, is a homely example. A kindly
lady was introduced to a very young child. She asked his age, and on hearing
that he was 3 years old she said, ‘He isn’t 3 years old; he is 3 years young.’ We call
this example homely because the statement ‘He is 3 years young’ is unlikely to
make people’s hackles rise as do many conceptual statements about the merits of
behaviourism, linguistic philosophy, etc. In other respects, however, it is a
typical conceptual statement: it predicts nothing about the behaviour of clocks
or calendars which is not also predicted by ‘He is 3 years old’ and does not contra¬
dict the latter in the way in which ‘He is 4 years old’ contradicts it. Like many
other conceptual statements it involves a proposal that one should talk
differently—not, of course, for aesthetic or stylistic reasons but because in
talking differently one is thereby viewing the situation differently: this particular
proposal invites one to reflect carefully on what standards can reasonably be
expected from a child aged 3. In agreeing with such a conceptual statement,
therefore, one is likely to use such expressions as ‘Yes, that fits’, or‘Yes, I see
what you mean’. If one disagrees, it is because one rejects the proposed ‘view’ of
the situation. In the present case one might wish to oppose what one regarded as
‘mollycoddling’ of 3-year-olds, and in that case one’s reply to ‘He is 3 years
 19

young’ might be, ‘Fiddlesticks! The sooner he learns to behave like a rational
human being the better!’
Sometimes conceptual disagreements can be serious and bitter. For example,
if X believes that Y’s way of viewing a situation is wrong he is likely also, at least
some of the time, to be opposed to Y’s practices; and if Y has made it his life’s
work to promote such practices it is scarcely surprising, simply from a considera¬
tion of the logic of the situation, that heated discussion should sometimes
follow. Clarification of what the dispute is about—and especially the
distinguishing of conceptual issues from empirical ones—can sometimes do
something to ease the tension, but if X commends a particular view of a situation
and Y commends a different view one cannot as a matter of logic expect that
either reasoned argument or appeal to evidence will necessarily give decisive
grounds for preferring the one to the other. It is perhaps worth adding that
where there is disagreement over fundamentals a common result is likely to be
misperception of what conceptual proposals one’s ‘opponent’ (as one sees him)
is making. Sometimes, therefore, there is misunderstanding as well as
disagreement, along with the irritation of feeling that the ‘opponent’ has not
taken the elementary trouble of finding out what one is trying to say. It is our
experience that some of the criticisms of both Freud and Skinner have been
based on such misunderstanding; and in general it is of course unprofitable to
argue about the merits of conceptual proposals whose significance one has
misunderstood.
There are also situations where a person can be wrong in asserting that certain
concepts are interchangeable. For example, if he were to assert that statements
of the form ‘A is the cause of B’ are always equivalent to statements of the form
‘Whenever an event of type A occurs it is regularly followed by an event of type
B’, someone else might show, by producing counter examples, that this
equivalence did not always hold. In general, it is not impossible to challenge
conceptual statements, but disagreements about them are in important ways
quite different from disagreements on empirical matters. Those who profess to
disagree with a particular conceptual proposal are in effect saying that the
alleged new ‘insight’ is misguided or inappropriate, that it emphasizes the wrong
things at the expense of the right ones, or simply that the concepts in question do
not in fact function in the way described.
Many of the statements in this book will be conceptual in character, though
we shall not hesitate to make empirical and other kinds of claim where these
seem relevant. In addition we shall at times be pointing out that particular issues
are or are not empirical or conceptual—a procedure which can in effect be
regarded as a ‘higher-order’ type of conceptual discussion. In general the book
can be regarded as an exercise in second-order psychology or‘meta-psychology’,
i.e. a critical examination of some of the concepts used by psychologists (and by
operant psychologists in particular) in the course of their research. We shall try
to show that conceptual errors not only generate Unnecessary disputation and
false inferences but may actually lead psychological research in unprofitable
directions. In the next chapter we shall set out some of the techniques by means
of which such errors can be detected.
 \
Chapter 3

Conceptual Analysis and Conceptual Revision

It was pointed out in the last chapter that many conceptual issues hinge on the
question of whether an X should count as or be classified as a Y—whether a
flying boat should count as a ship, for instance, or whether a tomato should
count as a vegetable. Those conceptual statements which involve reference to
existing boundaries were said to be examples of ‘conceptual analysis’, while
those conceptual statements which call for the redrawing of boundaries were
said to be examples of ‘conceptual revision’. It was also pointed out that
conceptual revision can occur either by the invention of a new word or by the
stipulation that an existing word should be used in a new way.
Now whether one’s intention is to analyse concepts or to revise them, in either
case it is necessary to consider what philosophers have called the ‘ordinary use’
of words. Since this phrase has given rise to misunderstanding some clarification
seems called for as to why ‘ordinary use’ has been considered important.
It needs to be emphasized in the first place that philosophical discussion in
this area is not concerned with any kind of survey of people’s speech habits.
Indeed, even if these habits were entirely changed this would not make any
difference to the logical behaviour of the concepts involved. Thus, to take an
extreme case, even if people consistently used the word ‘some’ where now we use
the word ‘all’ and vice versa, this would make no difference to the logical
behaviour of the concepts now expressed by the words ‘some’ and ‘all’; it would
mean only that we had to interchange the two words in talking about this
behaviour. Some years ago one of the authors was present at a discussion in
which an outraged professor of philosophy attempted to meet objections to his
thesis by saying, ‘I don’t care tuppence if that is not how bus conductors talk.’
This is good repartee but is unlikely to have disposed of the objection. Surveys
can, of course, provide many interesting empirical truths: they indicate, for
instance, what are the differences between spoken and written English; they can
reveal the extent to which people speak grammatically and fluently or the
reverse, and they can indicate the number of people who, like Mr Jingle in
Pickwick Papers, talk in short staccato bursts. Such results, however, would
have no particular conceptual or philosophical significance. Those who study
the logical behaviour of concepts are concerned not with speech habits but with
distinctions and classifications. ‘Our ordinary words’, says Austin (1962, p.3),
‘are much subtler in their uses, and mark many more distinctions, than
philosophers have realised.’

i
 21

When Austin speaks here of ‘ordinary words’ the intended contrast is

presumably not between ordinary and technical use but between the use of a
word for familiar and clearly defined purposes and its use in certain philo¬
sophical contexts where there has been some unwitting departure from existing
methods of classifying. ‘It disperses the fog’, says Wittgenstein (1953, section 5),
‘to study the phenomena of language in primitive kinds of application in which
one can command a clear view of the aim and functioning of the wprds.’ In
another passage he says: ‘When philosophers use a word . . . one must always
ask oneself: is the word ever actually used in this way in the language-game
which is its original home? What we do is to bring words back from their meta¬
physical to their everyday usage’ (1953, section 116).
As a first example we should like to cite Austin’s comments on the philo¬
sophical expression ‘directly perceive’.

1. First of all, it is essential to realise that here the notion of perceiving

mdirectly wears the trousers—‘directly’ takes whatever sense it has from the
contrast with its opposite: while ‘indirectly’ itself (a) has a use only in
special cases, and also (b) has different uses in different cases—though that
doesn’t mean, of course, that there is not a good reason why we should use
the same word. We might, for example, contrast the man who saw the
procession directly with the man who saw it through a periscope-, or we
might contrast the place from which you can watch the door directly with
the place from which you can see it only in the mirror. Perhaps we might
contrast seeing you directly with seeing, say, your shadow on the blind; and
perhaps we might contrast hearing the music directly with hearing it relayed
outside the concert-hall. However, these last two cases suggest two further
points.

2. The first of these points is that the notion of not perceiving ‘directly’
seems most at home where, as with the periscope and the mirror, it retains
its link with the notion of a kink in direction. It seems that we must not be
looking straight at the object in question. For this reason seeing your
shadow on the blind is a doubtful case; and seeing you, for instance,
through binoculars or spectacles is certainly not a case of seeing you
indirectly at all. For such cases as these last we have (quite distinct contrasts
and different expressions—‘with the naked eye’ as opposed to ‘with a
telescope’, ‘with unaided vision’ as opposed to ‘with glasses on’. (These
expressions, in fact, are much more firmly established in ordinary use than
‘directly’ is.)

3. And the other point is that, partly no doubt for the above reason, the
notion of indirect perception is not naturally at home with senses other than
sight. With the other senses there is nothing quite analogous with the‘line of
vision’. The most natural sense of ‘hearing indirectly’, of course, is that of
being told something by an intermediary—a quite different matter. But do I
hear a shout indirectly, when I hear the echo? If I touch you with a barge¬
pole, do I touch you indirectly? Or if you offer me a pig in a poke, might I
22
feel the pig indirectly—through the poke? And what smelling indirectly
might be I have simply no idea! (Austin, 1962, pp. 15-17; reproduced by
permission of Oxford University Press)

Now it is no part of Austin’s argument to suggest that ordinary use should never
be changed. In effect, however, he is inviting his audience to become more
sensitive to the many distinctions and nuances which are to be found if we study
language with care; and, perhaps more important, he is suggesting, at least by
implication, that lack of such care can result in downright bad philosophy. The
danger, in his view, is not in conceptual innovation as such but in unwitting
innovation, since this can result in false contrasts and misleading analogies.
This is a point to which we shall return later in this chapter, when we shall cite
examples of the kinds of insight which the techniques of conceptual analysis can
achieve. Before doing so, however, it may be helpful, particularly to readers un¬
familiar with work in this area, if we say more about the techniques themselves.
It is admittedly somewhat unsatisfactory to give ‘potted versions’ of philo¬
sophical methods which have taken many years to evolve, but at the risk of over¬
simplification we should like to offer three rules of thumb which can be used in a
variety of ways whenever the logical behaviour of concepts is under discussion.
The rules in question are: (a) apply Wittgenstein’s ‘polar principle’; that is to say,
ask what is being excluded, (b) ask what would count as a paradigm case of the
correct application of a concept; and (c) where a general statement is made, ask
it it is true in a particular case.
(a) Since concepts distinguish, they divide material in at least a two-fold way.
Thus if one characterizes something as being ‘X’ one is thereby excluding the
thesis that it is ‘not-X’. As an exercise the reader may like to consider what is
excluded by the following terms: higher, graduate, slow, bachelor, forgetful. It
makes no sense to speak of a distinction which does not distinguish anything,
and if a person is said to understand a concept this implies that he also under¬
stands what is being excluded; indeed one cannot group without excluding as
well as including.
The polar principle, then, invites us to ask the question, ‘As opposed to what?’
When confronted with a particular statement, there may be contexts— and, as
will be seen, they are often highly sophisticated ones—where it is profitable to
ask the speaker, ‘What exactly is being excluded?’ or ‘What would it be like if
you were wrong?’
This must not be taken simply as a request that the speaker should clarify his
meaning. If this were so, the need to invoke the polar principle would disappear
if only people expressed themselves clearly! It may indeed be true that an
obscure statement may become less obscure if one has worked out what would
be involved if its opposite were the case; but there is more to the polar principle
than this: one could perhaps describe it as a device for exhibiting to a
sophisticated thinker that there is something odd or paradoxical about what he
is saying.
It is difficult to give examples which are not controversial; but for present
 23

purposes it is perhaps more important to find a ‘live’ issue where the polar
principle can be applied than to avoid complications or controversy. With this
requirement in mind, therefore, let us consider the thesis that man is merely a
highly complicated kind of machine. To apply the polar principle in this case
would be to invite the defender of this thesis to consider standard uses of
machine and mechanical’ and to ask himself what it would be like not to be a
machine. ‘As things are’, we might say to him, ‘there are two words, “man” and
machine , both of which we know how to use appropriately. By the same token
we can distinguish those situations where a person responds, as we say,
mechanically”, e.g. by saying “Yes, yes” in a monotonous voice, from those
situations where his responses are not mechanical. Are you then saying that
responses which we thought were not mechanical are really mechanical after all?
In that case, please indicate what characteristics a response would have to have
for you to be justified in saying that it was not mechanical.’
Here is a second example. The evidence from psychotherapy suggests, or is
believed to suggest, that the same defence mechanisms and the same apparently
neurotic weaknesses are to be found in seemingly healthy people as are found in
neurotic people. It immediately becomes tempting to say that ‘really’ (whatever
this means) we are all neurotic. To apply the polar principle in this case would be
to ask. What, then, is it like not to be neurotic?’ This last example is
controversial for a variety of reasons. There would of course be no logical
objection to saying that all of us show signs of neuroticism some of the time, and
there is the further complication that the word ‘neurotic’ is ‘open-textured’ (cf.
p.l3) in the sense that there are no precise formulated rules as to what exactly
shall count as ‘being neurotic’. For our purposes, however, the example is
convenient: it is a live one (since many readers will be aware of the pressures
which might lead a psychotherapist to say, ‘We are all neurotic’), and it exhibits
an important characteristic of the polar principle, viz. that it is used as a way of
critically examining certain claims which arise when knowledge has reached a
certain degree of sophistication.
Now those who say that man is merely a machine and those who say that we
are all neurotic are clearly making claims which are intended to be exciting and
important. One function of the polar principle, however, is to exhibit the pos¬
sibility that there is something deceptive in this challenge. Words, as we have
already indicated, hunt in pairs, and if one member of the pair correctly occurs
in a sentence it is at least meaningful, even if false, that the other member should
be substituted. In these two examples, however, it is impossible as a matter of
logic that any such substitution should occur, since the familiar distinctions
between ‘man’ and ‘machine’ and between ‘neurotic’ and ‘normal’ are being
disallowed. The words ‘machine’ and ‘neurotic’ are thus both being used without
a polar term, which means that no distinction is being drawn.
To employ a term in common use among philosophers, use of the word
‘neurotic’ is parasitic upon there being a distinction between ‘neurotic’ and
‘normal’. Unless we were aware of such a distinction the claim that we are all
neurotic could not startle us in the way that it does. In general, the use of one
24

member of a pair of terms is parasitic upon there being a polar or contrasting

term; and if no such contrast is intended then the use of the original term is
misleading. In this particular case, if one met a paradigm case of a normal
person and someone said that he, too, was neurotic, one could rightly ask, ‘What
in that case does “normal” mean?’ Indeed, if there had never been any grounds
for distinguishing neurotic from normal behaviour, then it is impossible as a
matter of logic that anyone would ever have learned to use the word ‘neurotic’
correctly. From the fact that plenty of people have learned to use the
‘normal-neurotic’ distinction correctly, however, it follows that there must have
been grounds—even if not satisfactory ones—for initially making it. In applying
the polar principle, therefore, one is in effect asking people to reflect on some
initial distinction.
On this showing, of course, the passage from Austin, quoted above, is an
example of skilful use of the polar principle; he was offering a challenge to those
philosophers who spoke of ‘direct perception’ to say what indirect perception
would involve.
(b) The second ‘rule of thumb’ is that one should consider how a concept is
used in a so-called ‘paradigm case’, i.e. in a typical or standard case where it is
correctly applied; one can then see the extent to which a new situation where this
concept is used approximates to the standard. For example, if a person
smiles and makes welcoming remarks to a stranger, then—provided the context
is such that no hypocrisy is involved—this is a situation par excellence where one
would be entitled to say that the person was friendly. The concept offriendliness
entails precisely such behaviour, and to be friendly just is to act in these or
similar ways. It makes sense to doubt whether a person regularly so behaves, but
if he does it would be impossible as a matter of logic to deny that he was friendly.
Indeed, if someone attempted to deny this one’s rejoinder might well be, ‘If that
is not being friendly then I do not know what “being friendly” means.’ Similarly
if one met a paradigm case of a normal person and someone said that he, too,
was neurotic, one could rightly ask, ‘What in that case does “normal” mean?’;
and if one met what was manifestly a human being and was told that he was a
machine, one could legitimately say, ‘If this person is not a human being then I
do not know what “human being” means.’
(c) The third ‘rule of thumb’ may be formulated as follows: to bring out what
is involved by a statement one should consider a particular consequence and ask
if this is what the speaker intends. Even before the time of Wittgenstein this
technique had been systematically used by G. E. Moore as a way of exhibiting
the oddity of some of the things which were being said by his philosophical
contemporaries (see, for instance, Moore, 1959, especially pp.32-59). If we now
try to follow Moore’s lead in the case of our present two examples, the argument
would be: ‘You say that man is merely a machine. It follows, therefore, that this
is true of all men. In that case, let us consider the particular man. Smith, and ask
if he is a machine.’ Similarly it might be argued: ‘You say that we are all neurotic.
Let us begin by asking you to justify your implied claim that Smith is neurotic.’
It is not, of course, being suggested that general statements as a class are
 25

vulnerable in this way. Thus there is nothing self-defeating in saying that all
human adults are over 3 feet tall. In this case we know exactly what is being
excluded, and there would be no temptation whatsoever, if we found a person
under 3 feet tall, to say that ‘really’ he was over 3 feet after all. The technique
which we have described is applicable only when standard cases of being an X
are in some way being disallowed.
What has been said so far, however, does not do justice to the full complexity
of our two examples. If it did, one could reject both ‘Man is merely a
complicated machine’ and ‘We are all neurotic’ as mere blunders or aberrations,
which clearly they are not. It is possible, however, to reinterpret them by saying
that each is in effect a proposal for conceptual innovation. On this showing what
is involved is a plea for the abandonment of a distinction—in the one case the
distinction between human nervous systems and artefact ‘nervous systems’, in
other case the distinction between ‘normal’ and ‘neurotic’ responses in a therapy
situation. It is not for us to defend either plea; but it is perhaps worth pointing
out that distinctions can be useful in one context and not in another.
As a final illustration we should like to quote a remarkable passage from
Hippocrates, in which he points out that in the classification of diseases a
contrast between ‘divine’ and ‘human’ is unnecessary. After indicating that the
heart and diaphragm (‘phrenes’) ‘have nothing to do . . . with the operations of
the understanding’ but that these involve the brain, he makes the following
comments on epilepsy:

The disease called the Sacred arises from causes as the others, namely those
things which enter and quit the body, such as cold, the sun, and the winds
which are ever changing and never at rest. And these things are divine, so
that there is no necessity for making a distinction, and holding this disease
to be more divine than the others, but all are divine and all human.’ (From
The Genuine Works of Hippocrates, tr. F. Adams, London, 1894, vol. II,
p. 857)

This is a quite remarkable anticipation of the polar principle!

Our next task will be to cite some further examples of the way in which the above
philosophical techniques can be used. These examples will, we hope, be ones
which psychologists can recognize as relevant to some of their own theoretical
problems. Our intention, however, is not to try to solve all the philosophical
questions raised, but simply to give the reader the chance to see the techniques in
action and thus have a preview of the use to which they will be put later in this
book.
In the left-hand column we have therefore set out a series of statements in
which certain key words have what Wittgenstein would call a ‘metaphysical’
usage, while in the right-hand column we exhibit the same words when they are
used ‘in the language-game which is [their] original home’. Our purpose is not to
make claims about the truth or appropriateness of either class of statement but
only to set them out side by side so as to exhibit the differences.
26

‘Metaphysical’ usage Usage in ‘original language-game’

(1) ‘We can never really know that ‘Are you quite sure there is a table in
there is a table in front of^us.’ your study?’

(2) ‘We can never study learning itself ‘I know that clocks tick and strike
(or intelligence or memory or and that their hands move, but I
attitudes), only their outward know nothing about the
manifestations.’ mechanisms inside.’

(3) ‘Sensations are the raw data on ‘1 was aware of strange sensations’; ‘I
which our perceptions are built.’ felt a tingling sensation in my toe’;
‘This house is built of brick.’

(4) ‘My toothache is private to me.’ ‘This road is private.’

(5) ‘You cannot step twice into the ‘Is this the Ouse again? If so, 1
same river.’ paddled in it yesterday.’

Now it is worth noting at the outset that the statements on the left become
important only after we have reached a certain degree of sophistication. (l), for
example, is a sceptical conclusion in the sense that it seems to imply some kind of
limitation to human knowledge. Yet, as Berkeley pointed out, the plain man is in
no danger of being worried by such alleged limitations. ‘We see the illiterate bulk
of mankind that walk the high road of plain common sense... for the most part
easy and undisturbed . . . They complain not of any want of evidence in their
senses, and are out of all danger of becoming sceptics’ {Principles of Human
Knowledge, Introduction, section 1).
Problems arise, in Berkeley’s view, only in more learned types of discussion.

Bid your servant meet you at such a time, in such a place, and he shall never
stay to deliberate on the meaning of these words: in conceiving that
particular time and place, or the motion by which he is to get thither, he
finds not the least difficulty. But if time be taken, exclusive of all those
particular actions and ideas that diversify the day, merely for the
continuation of existence, or duration in the abstract, then it will perhaps
gravel even a philosopher to comprehend it. {Principles of Human
Knowledge, Part I, section 97)

Berkeley was aware that apparently sceptical conclusions are not always the
result of ‘the obscurity of things, or the natural weakness and imperfection of
our understandings’ {Principles of Human Knowledge, Introduction, section 2).
Often what is involved is unnecessary disputation because in a subtle way we
have been hoodwinked by words. ‘We have first raised a dust, and then
complain, we cannot see’ {Principles of Human Knowledge, Introduction,
section 3).
Indeed (2), like (1), is just the kind of sceptical claim against which Berkeley’s
 27

argument might have been directed. ‘What you call the empty forms and outside
of things’, says Philonous, ‘seems to me the very things themselves’ {Third
Dialogue between Hylas and Philonous). To put Berkeley’s point in another
way, there is a difference between the situation where one is genuinely looking at
outward manifestations, e.g. the moving of the hands of a clock in contrast with
the moving of the cog wheels inside, and the situation where no such contrast is
involved. Similarly, to apply the same kind of argument to (2), if a person claims
to be studying learning or intelligence or memory or attitudes it is gratuitous to
suppose that, lying behind or beyond what he observes, is something else which
no one could ever conceivably observe, and that it is these hidden things to
which the words ‘learning’, ‘intelligence’, ‘memory’, and ‘attitude’ refer; and it
follows that the sceptical lament that such entities are for ever beyond the reach
of our knowledge is unjustified. It is true, of course, that on the basis of a
subject’s responses it is possible in principle to make inferences to events inside
his body, just as one might make inferences about the mechanisms of a clock
from studying the movements of its hands. Any statements made on the basis of
such inferences, however, would be physiological in character, whereas it is
plain that the concepts learn, remember, intelligent and attitude have been
correctly applied through the ages by people who knew none of the relevant
physiology; it follows, therefore, that if one is using these terms in their ordinary
sense nothing physiological is predicted by them.
It is, of course, open to an innovator to use such terms in a new way; for
example, if it were found that in all situations where a subject could be said to
have learned something a particular physiological statement, e.g. about changes
occurring at the synapse, were always true, there would be a case for saying that
learning was ‘really’ a set of changes occurring at the synapse. This, however,
would involve an innovation and would not be a correct account of the word
‘learn’ as it is at present understood. In point of fact the search for the ‘real’
nature of learning or the ‘real’ meaning of the word ‘learn’ is almost certainly
misguided. It is a mistake, as Berkeley pointed out, to suppose ‘that every name
hath, or ought to have, one only precise and settled signification’ {Principles of
Human Knowledge, Introduction, section 18). The same point is emphasized by
Wittgenstein, who cites as an example the word ‘game’.

Don’t 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, relationships, and a whole series of them at that. To repeat:
don’t think but look!—Look for example at board-games, with their multi¬
farious relationships. Now pass to card-games; here you find many corres¬
pondences 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. Or is there always winning and losing, or competition
between players? Think of patience. In ball games there is winning and
28

losing; but when a child throws his ball at the wall and catches it again, this
feature has disappeared. Look at the parts played by skill and luck; and at
the difference between skill in chess and skill in tennis. Think now of games
like ring-a-ring-a-roses; here is the element of amusement, but how many
other characteristic features have disappeared! And we can go through the
many, many other groups of games in the same way; can see how
similarities crop up and disappear.

He concludes that: ‘We see a complicated network of similarities overlapping

and criss-crossing ... I can think of no better expression to characterise these
similarities than “family resemblances’” (Wittgenstein, 1953, sections 66-67).
By the same argument, therefore, it is not necessarily correct to suppose that
those situations where someone is said to have learned something will all have
some single characteristic in common; on the contrary there may be a variety of
‘family resemblances’. The same may also be true in the case of those situations
where a person remembers something, displays intelligence, or exhibits a hostile
attitude. It is a mistake in the first place to assume that there is always some
unique defining characteristic, and it is doubly mistaken to suppose that the
events which we observe are in some way outward manifestations of some ‘inner’
occurrence which constitutes the ‘real’ meaning of the words ‘learn’,‘remember ,
‘intelligence’, and ‘attitude’. In some contexts, as Wittgenstein points out, it may
be helpful to ask what particular examples taught us the meaning of a word,
since in that case ‘it will be easier for you to see that the word must have a family
of meanings’ (1953, section 77).
Examples (3) and (4) are, we believe, self-evident without detailed discussion.
It is sufficient to point out that, in the left hand column, the familiar words
‘sensation’ and ‘private’ when used in this sophisticated context have undergone
a subtle change of meaning.
With regard to (5), the claim that one cannot step twice into the same river is
attributed to the Greek thinker, Heraclitus, and is believed to have been part of
his general emphasis on the phenomena of change. The statement is an
interesting one because, if taken literally, it is manifestly false, since one might
step into the Ouse at 10 in the morning and again at 12 noon. If Heraclitus then
answered that in that case it was no longer the same river, he would in effect be
saying that there are no situations where one is entitled to speak of ‘the same
river’ or indeed entitled to use the word ‘same’ at all. The polar term to ‘same’ is
‘different’, use of the one being parasitic on use of the other; and we might
therefore interpret Heraclitus as saying that in a true account of the world this
pair of terms would not be needed.
Now the fact that one can formulate correct sentences in the right-hand
column does not establish that the sentences in the left-hand column are simply
blunders. What it establishes is the existence of differences. To try, for example,
to infer the nature of learning from a study of people’s behaviour is in important
ways unlike inferring the mechanisms of a clock from the movements of its
hands. Similarly, to say ‘You cannot step twice into the same river’ is in an
 29

important respect unlike saying ‘You cannot step twice into the Ouse’. If the
speaker has not noticed the difference he is misleading himself, and once he has
seen ‘how things are’ (cf. p. 15) he may decide that he no longer wishes to talk in
that way.
There may be occasions, however, where the conceptual claim is made with
full awareness of its implications. For example, a person who says ‘We can never
really know that there is a table in front of us’ might defend his statement in some
such way as this: ‘I quite admit that in the ordinary sense of “know” it is correct
to say that we sometimes know that there is a table in front of us. But 1 am
inviting you to change your standards of what constitutes knowledge.’ He might
plead, in support of this, that, in contrast with empirical truths, the truths of
mathematics and logic have a special character of necessity or indubitability,
and that this is the area where knowledge in the full sense is to be found. This is
perhaps one of those philosophical questions to which the answer is ‘yes and no’
(cf. p.l5); if one has seen all the reasons for revising our standards of what
constitutes knowledge and all the reasons for keeping them as they are, then ex
hypothesi there is nothing more to discuss.
To appeal to the ordinary use of words does not therefore commit one to
saying that conceptual boundaries should never be changed; and indeed
different investigators—legislators, physicists, psychologists, and many
others—may need to reclassify for all kinds of different reasons. What is
important is to be able to recognize revisionist proposals when one meets them.
If one fails to appreciate that a particular word has undergone a shift in meaning
one may fail to distinguish things which are different; and indeed psychology
books at the present time all too often contain unwitting or unexamined
proposals for conceptual innovation. To examine how a word functions ‘in the
language-game which is its original home’ will not tell us what revisions ought to
be made but it will force us to think carefully about what it is that we are revising.