[SK's comments]
[These are the two papers mentioned by Bambrough (1961) in a footnote:
"Of recent writings on this topic I believe that only
[these]
show a complete understanding of the nature and importance of Wittgenstein's contribution"
to
"the problem of universals".]
JOHN WISDOМ
PHILOSOPHY AND PSYCHO-ANALYSIS
(1964)
[51]
METAPHYSICS AND VERIFICATION
(Reprinted from Mind, Vol. XLVII, N.S., No. 188, 1938)
Disclaimer: This one is tough for sledding for an amateur, but I think I understand enough to recommend it unreservedly. It is dour and meandering, but still, this is REAL philosophy, like an entree that finally arrives after too long spent nibbling the stale popcorn of 'dreary' Aesthetics.
The first page notwithstanding, comments are shorter and less frequent than my usual. I wish I could claim continence and restraint, but the real reason is incomprehension.
Minimal highlighting and excisions. Footnotes excised unless they are highly relevant.
If I don't say so, the chopped-up formatting really is essential this time; nothing I've so far annotated here cries out for it quite like this one does.
Enjoy!
'THE meaning of a statement
is
the method of its verification.'
Some philosophers bring out this principle with confidence and satisfaction;
others are utterly opposed to it ...
This conflict is of the greatest importance ...
The Verification Principle
is
the generalization of a very large class of metaphysical theories,
namely
all naturalistic, empirical, positivistic theories.
While its opposite, which I venture to call
the Idiosyncracy Platitude,
is the generalization of all common-sense, realist, transcendental theories.
The verification principle is the generalization of such theories as:
A cherry is nothing but sensations and possibilities of more;
A mind is nothing but a pattern of behaviour;
There are no such things as numbers, only numerals, and the laws of logic and mathematics are really rules of grammar;
Beauty is nothing but the features in respect of which a thing is beautiful, and the feelings these arouse.
Consulting Google (yes, now in "AI Mode") . . .
The
cherry
comes from Berkeley's
Three Dialogues between Hylas and Philonous
(1713).
"Berkeley argued for the principle of esse est percipi ("to be is to be perceived"). He used the example of a cherry to illustrate that material objects do not exist independently of a mind.
The
possibilities
are J.S. Mill's:
"Mill defined matter not as a "thing" but as a "permanent possibility of sensation"."
Perhaps the point is: there simply is no way to verify any "exist[ence independent] of a mind".
The
mind
as
behaviour
is credited to Ryle (1949), but clearly the idea did not await that year or that book.
The
numerals
are said to be names-as-in-nominalism.
The
grammar
is Wittgenstein's.
I now catch myself thinking that "numbers" are a special case, but without being able to say why that should be.
And now "AI Mode" has decided that it's having a conversation rather than fielding a series of discrete queries . . .
The
beauty,
features,
and
feelings
are attributed to Ayer:
"This specific formulation mimics the style of A.J. Ayer in Language, Truth, and Logic (1936), where he argued that aesthetic and ethical judgments are "nothing but" expressions of emotion and have no literal meaning beyond the feelings they arouse."
Googs fixates on the appearance of "feelings" here and calls this a 'subjectivist' position. That is an impressively humanlike gloss, which is to say that it's not just wrong but wrong in a perniciously idealistic and dissociating way.
The aphorism as given is silent on the (non)universality of beauty. It says only that "beauty" is in respect of some features. If anything this suggests objectivism, at least in the sense of our being able to 'objectively' attribute any given sensation of 'beauty' to some discrete "feature". Whether or not this is everyone's (or anyone else's) convergent experience, the "verification" of this feature-feeling correspondence in even a single case would be a notable 'objectivist' achievement. We seem to lack even this much the vast majority of the time; even a weak sense of "verification" is too strong for most of the 'evidence' that 'objectivists' rely upon. (e.g. Carroll, p. 193: " . . . the pervasive finding that the opening of Mozart's Twenty-Ninth Symphony is beautiful . . . " By the time this statement is properly qualified, nothing is left of it.)
That is one thing. The contention that "judgments are "nothing but" expressions of emotion and have no literal meaning beyond the feelings they arouse" is getting at something else entirely. To say that these "feelings" "have no literal meaning" is to bypass entirely the obj-subj-ism flame war and to hone in on something else, something which would apply just as strongly in one case as in the other. Maybe some humans and a few bots would call this 'subjectivism', but I think that's wrong. As with Nominalism as against Realism, the terms seem to have been chosen by one side in order to tilt the rhetorical field their own way. But if the feelings have no meaning, where is the Subjectivism? What does it matter if group feelings converge or diverge if those feelings are meaningless in any case? Devoid of meaning, we can't even speak of convergence or divergence, because there is then no basis on which to compare emotional specimens. In that case, to call this position 'subjectivism' is to define it in opposition to the 'objectivism' which has also been shorn of its basis for verifying a central tenet. Perhaps that indeed is the unthinking lashing-out of the now-jaded Objectivist, but if so it is dead wrong, because the Subjectivist has taken the same haircut.
At this early juncture, if I had to guess at the topline takeaway, it's something like: such things as cherry, mind, number, and beauty cannot figure in statements such that those statements have a discernible method of verification; but such things as sensations, behaviour, numerals, and features can figure as such.
According to the idiosyncracy platitude
every sort of statement has its own sort of meaning,
and
when philosophers ask
'What is the analysis of X-propositions?'
the answer is that
they are ultimate,
that
'everything is what it is and not another thing'
(Butler, quoted by Moore on the title-page of Principia Ethica).
This principle is the generalization of theories such as:
Ethical propositions involve value predicates and are ultimate;
Psychological propositions are not reducible to physiological propositions, they are ultimate;
Mathematical propositions are necessary synthetic propositions—an ultimate sort of proposition;
Statements about nations are not to be reduced to statements about individuals, they are about a certain sort of concrete universal.
Googs attributes the first of these also to Ayer,
"who argued in his work, Language, Truth and Logic, that moral statements express emotions (emotivism) rather than objective facts, fitting with the idea that value judgments are non-natural and non-reducible ... "
Maybe "reducible" would mean something like, 'attributable to features'? And "ultimate" would mean 'not attributable to features'?
The next is Wittgenstein's. Googs quotes the Tractatus directly:
"Psychology is no more closely related to philosophy than any other natural science... Psychological propositions are not reducible to physiological propositions, they are ultimate."
The remaining are attributed to this very paper. Humph.
There are not other answers
to these metaphysical questions.
Consequently most or all metaphysical conflict finds expression in
'Shall we or shall we not accept the principle that the meaning of a statement is the method of its verification?'
and sometimes
'Is the verification principle true?'
Is this to say that by ultimate statements we mean un-verifiable statements? That is, there is no method by which they could be verified?
I do not at all wish to
[52]
suggest that we cannot get on with metaphysical questions without first dealing with this question.
...
In fact an intermediate plan is best—
first an examination of easier metaphysical and nearly metaphysical questions,
then a mention of the verification principle,
then an attack upon the more difficult theories,
then a more thorough investigation of the verification principle,
then a return to the theories. . . .¹
... 'Is the verification principle true?', ...
I myself should at once ask for the question to be put in the wider, less answer-fixing form
'Shall we accept the verification principle?'
For I believe the other form misleads us as to the general nature of the question asked.
... once its general nature is apparent
the question
'Is it true or not?'
vanishes into insignificance
while
its important metaphysical merits and demerits will have become apparent in the process.
What I have in mind
is this.
Many different things are covered by the expressions 'accepting a theory', 'holding a view'.
'Shall we become Fascists?
'Are you a Surrealist?'
'Is art the production of significant form?'
'Does 1 plus 1 make 2?'
...
To accept the theory that cancer is due to a germ is different from accepting the theory that it is due to a chemical poison.
But this is a difference in the theories,
not in the sort of theories they are;
and it is not the sort of difference I have in mind and call a difference
¹ On the whole the process of thought has been
from the more specific theories to the more general,
from the doctrine that analytic propositions are verbal to the doctrine that all necessary propositions are verbal,
and from this and such theories as those mentioned above to the verification principle, rather than deductively downwards from it.
I admit that in the writings of those supporters of the principle who are positivists ... there is to be found ground for Dr. Ewing's accusation that the procedure has been from the principle to the specific theories.
Such a procedure, once the verification principle has been recommended by the specific cases, is perfectly satisfactory in a way I shall try to explain.
...
[53]
in general nature.
A difference in general nature
is such a difference as we draw attention to when we say
'To become a Fascist is not a purely intellectual process like becoming convinced of the germ theory of cancer',
or such a difference as we indicate when we say
'A man who says that 1 plus 1 makes 2 does not really make a statement, he registers a decision'.
We often use the words 'not really a statement'¹ when we wish to draw attention to a difference in general nature,
a difference in style of functioning
as opposed to difference in subject-matter.
People are inclined to say
'The statements of fiction are not really statements and so it is silly to ask whether they are true; and even poetical statements such as "A woman is a foreign land" are not really statements, and aren't really true or false'.
Now there are differences of this sort within the range of statements which are on most occasions unhesitatingly called statements and to which it is quite usual to apply the expressions true or false.
When this is so, the differences in general nature are apt to be overlooked.
And sometimes overlooking them produces an inappropriateness in what we do when asked whether they are true.
We act like one who when asked whether it is true that 'the stars are rogues which light the wanderer home' says he doesn't know and looks up books on astronomy.
Hmm. Isn't it the questioner who is at fault there?
It is,
in my opinion,
the neglecting of this sort of difference which has prevented the solving,
the dissolving in Wittgenstein's² phrase,
of metaphysical problems and of the problem of the verification principle in particular.
...
...
When people bring out with a dashing air the words 'The meaning of a statement is really simply the method of its verification',
like one who says 'The value of a thing is really simply its power in exchange'³,
in what sort of way are they using words?
What is the general nature of their theory?
¹ ...
² Wittgenstein has not read this paper, and I warn people against supposing it a closer imitation of him than it is. On the other hand, I can hardly exaggerate the debt I owe to him ... I have put a W against some examples which I owe to him. It must not be assumed that they are used in a way he would approve.
³ Indeed one might put the verification principle in the form
'The meaning of symbols is really simply their power in prediction'.
[54]
The answer is
'It is a metaphysical theory'.
Cool.
I wonder just how this reframing might bear upon Danto's many gleeful retellings of the Principle's demise? This is the demise of an analytic or epistemological theory, not of a metaphysical one, no?
True, it is a peculiar metaphysical theory as appears from the fact that we are inclined to say:
it is not so much a metaphysical theory as
a recipe for framing metaphysical theories;
it is not a metaphysical theory, it is a mnemonic device
for getting from metaphysical theories which have been illuminating in easy metaphysical difficulties to theories which shall work in harder cases,
a mnemonic device reminding us how to meet objections to positivistic theories;
it is a recommendation to so use 'mean' that
S means the same as S' provided they are verified in the same way,
where this recommendation is not for the purpose of
metaphysically illuminating the use of 'meaning'
but for another metaphysical purpose, namely
the illumination of the use of expressions which on the recommended use of 'meaning' will be said to mean the same.
OOFDAH.
To say that
two statements,
S
and
S',
mean the same,
we must know
(among other things)
what
mean
means.
Leave aside for now
the hermetic logic of the VP;
consider
what it does
to
the use of expressions:
it stipulates a meaning-of-meaning
per which
S
and
S'
either
do
or
do not
mean the same
depending upon whether
they
are
or
are not
verified in the same way.
It is this 'altruism' which makes the verification principle a peculiar metaphysical theory.¹
I don't follow, re: altruism. We think that the VP is an "altruist" in a world of violent semantic disagreement? And it is thus only a recommendation on how to proceed, not a truth which simply must be heeded?
But it is
the likeness of the verification principle to metaphysical theories
which I now want to emphasize and explain.
It is like not only to such theories as
'A mind is really simply a pattern of behaviour',
'Goodness is a matter of causing approval',
but also to such theories as
'We never really know that what we see is real and not a dream',
'We never really know what is going on in another person's mind',
'Nothing is really the same from moment to moment',
'All words are vague'.
It is to emphasize this likeness that I call the verification principle a metaphysical theory.
I should be prepared to argue that there is nothing incorrect in calling it this.
But that is neither here nor there.
Hmm. What theory is not latently metaphysical? If they all are, or could be, depending merely on what we'd like to do with them, then there's nothing to see here.
What we are concerned with is its metaphysical nature.
And to illuminate this I say that it is a sort
¹
The non-altruistic theory
connected with it
is
'The meaning of a statement is a functional feature of it, not an object, like the Hampton Court Maze, of which it is the plan',
or
'The meaning of a statement is a matter of its uses'.
The comparison of this with the verification principle would help to deal with many conflicts in which one philosopher says that S means the same as S' while another says it does not.
For often these arise because
one is using 'meaning' in accordance with the verification principle so that only certain functions count
in estimating the meaning of a sentence,
while the other is using it so that
other functions count.
Thus
one may count only the conditional predictive functions of a sentence
so that
'You will be stone deaf to-morrow'
means the same to a deaf man as to one who can hear, ...
But the unconditional predictive power is different; ...
And if the unconditional predictive power is counted in estimating meaning then the sentences do not mean the same to a deaf man as to one who is not.
This, worked out, throws much light on the puzzles connected with
the soul's survival of the body and the ego-centric predicament.
. . . 'You, Soulie, will vacate your host tomorrow, never to return' . . .
" . . . One step a-head-a-ya . . . "
Maybe I will understand the last sentence better (at all) by the end of the paper.
I think I understand the rest, but it doesn't seem likely to be useful.
Calling this
function
seems too bounded, formal, and known to all parties. It's more like (the dreaded) 'context', which is infinite, leaky, fractal, and most importantly,
asymmetrical.
When 'context' is given its due, no statement is quite the same as any other. Then, if we think there are (or must be)
any
instances of same-meaning statements, maybe we should look for 'simple contexts', so to speak; i.e., contexts where the possibilities for meaning are drastically narrowed by circumstance. And then, if the meat of the above example is granted (I don't see how it can be denied), then how can there be any sure things at all here?
[55]
of metaphysical theory;
and for our purpose it does not matter whether it is a sort of metaphysical theory
(a) in the way that a hackney is a sort of horse, or
(b) in the way that a motor cycle is, because it is a sort of tireless horse on wheels.
If (a) my statement is correct;
if (b) it is not.
But this correctness is of no importance.¹
For I make the statement to draw attention to certain likenesses,
and whether they suffice or no for the proper application of 'metaphysical' does not affect their existence.
We greatly look forward to putting names to these likenesses.
I say that the verification principle is a metaphysical principle because I want
(1) to draw the attention of those who accept it to the deplorably oldfashioned clothes in which it presents itself.
Indeed it resembles not only positivistic theories but also the worst transcendental theories
by appearing in the disguise either of
a scientific discovery removing popular illusion,
or of
a logical equation (incorrect) from which deductions may be made.
No wonder our conservative friends cannot accept it.
I want
(2) to draw the attention of those who reject it to the fact that
because they are taken in by its disguise they fail to recognize the merits
which like other metaphysical theories it conceals.
Both those who accept it and those who reject it do not realize what they are doing because they do not notice that it is disguised.
...
we shall now see what a metaphysical theory really is
...
...
It is possible to go on in either of the two following ways:
We may examine the nature of the verification principle and thus throw light on the nature of other metaphysical theories,
or we may examine the nature of other metaphysical theories and thus throw light on the nature of the verification principle.
Let us adopt the latter plan and work from the specific to the general.
Then applying our results in a direct examination of the verification principle we shall obtain a review of the whole of metaphysics,
because the
¹ Except while it is thought to be;
like the doctor's highly coloured medicine or the reserve ratio of the Bank of England.
People believe it is important that the reserve ratio should not fall below 9 to 1, say;
and because they believe this,
it is important.
[56]
verification principle is the generalization of one set of answers to metaphysical questions
while its opposite is the generalization of the opposing answers.¹
1. What Is A Metaphysical Theory?
1. The metaphysician is a profound scientist?
Metaphysical theories were at one time presented as super-scientific discoveries.
They are still presented so as to be mistaken for these.
Consider the verification principle² itself and Wittgenstein's remark in a lecture,
'We are inclined to think that there must be something in common to all games, say, and that this common property is the justification for applying the general term "game" to the various games.'³
Hmm. Shall we presume that this
remark
was developed in the now-familiar manner? The
super-scientific
part would seem to lie in what comes next.
The now-evident fitness of the
family resemblance
concept for operationalization in empirical inquiry sits uncomfortably with the notion that it ever was "super-scientific," unless by "super" our author means 'highly potent' rather than 'above', 'over', 'beyond'.
But this way of presenting metaphysical theories went out of fashion.
Something was lost when this happened,
because, for one thing,
the scientific theory prevented people from regarding metaphysics as analysis, a matter of how words are used.
In a sense, like poetry, metaphysics is synthetic because it reveals something which is hidden in a way in which logical definitions never do.
Well,
it reveals something,
certainly.
Why should we think we know
what
exactly has been revealed, and
why
it matters?
(Asking about
poetry
and hoping we can do better for
metaphysics.)
(Is
this
what
synthetic
means?!)
It is worth seeing something of why the scientific theory of metaphysics gained such a hold.
It did so
because it is so tempting
to exhibit the metaphysician
as
an extra cautious scientist.
Don't follow. Synthetic means in-cautious.
A scientist or sophisticated person may doubt whether a rabbit has come from a hat or a table risen from the floor, when an ordinary or unsophisticated person will swear these things have happened.
Hmm. Also, the scientist who has never traveled North may declare:
"In the North, where there is snow, all bears are white."
The unsophisticate, meanwhile, will
doubt
this. The roles are reversed.
Short of certainty, doubt is always a tenable position, and usually it is more secure overall; but curiously enough, what we call 'trust' in a contemporary sociological context is essentially the elision of doubt by way of
feeling good.
This is highly irresponsible and ultimately unsustainable, and we've staked everything on it.
Why unsustainable? Because
feelings
are
(quite literally)
unsustainable; and if we do find the rationalist and the rube constantly reversing roles vis-a-vis doubt and trust, it's probably because they
feel
differently in different cases.
But if the phenomena can be regularly repeated, photographed, seen by all, smelt by dog, scared by cats, then even the scientist will say 'We now know for a fact that these things happen'.
Yep. And the rubes?
There are,
however,
people
who might still refuse to say 'We know a rabbit came', 'We know the table rose'.
A madman might refuse,
a Times correspondent might refuse,⁴
(for he never
¹ ...
³ The great value I attach to what Wittgenstein says here, appears later, pp. 80-92.
⁴ If only we had noticed how much liker is the philosopher's doubt to that of The Times correspondent than to that of the madman!
The Times correspondent is as energetic as anyone in building armaments, and the philosopher as confident as anyone when seating himself on a table.
Hence Wittgenstein's expression 'pseudo-doubt'.
[57]
says that a certain power has so many troops at such and such a place but always only that this would appear to be so).
Well . . . if nothing at all appears to the rube, then his "doubt" cannot quite be doubted, can it?
And a philosopher might refuse.
But the philosopher or metaphysician is easily distinguished from the others,
not so much by his appearance as by his subsequent talk.
He will say,
'We don't know
that the table moved because we don't know whether there was a table there at all.
We don't know
that we are not constantly being taken in by an arch-deceiver, a super-illusionist who runs without a hitch a continuous show of prodigious length: so we can never be certain.
After all there is no difference in kind between a fleeting glimpse and an hour's scrutiny.
And if seeing does not suffice, does seeing and touching or seeing, touching and smelling?
Are 5 independent witnesses required to prevent talk of infectious hallucination or would 237 be necessary?
Further observation may increase probability
but probability is not knowledge.
Knowledge man never obtains;
not from the time he reaches for a golden ball and learns it's the moon till when he dreams of Paradise and wakes in Camden Town'.
Per Googs, the latter line is from
The Autobiography of Mark Rutherford
by William Hale White.
The "golden ball" line turns up nothing pre-1938.
Of course our
philosopher
scores a few points here, even in caricature. Again, doubt is the secure position, and the obstinate one. Put it aside and things seem to work out just fine, if only we can just
feel good
and
trust.
The mistake is to think that things
always
work out, and/or that the difference between 'always' and 'almost always' isn't important.
Again,
a doctor may be sceptical of the value
of a new drug although you tell him
ten people took it and soon improved.
He says that perhaps they all changed their diets in a certain way when they took it,
or that perhaps it was suggestion;
or he says that with so few cases it may have been a coincidence.
But if thousands of people, some with faith and some without, some from cold climates, some from warm, some . . . all improve then even a very scientific doctor will admit the value of the drug.
It is only the philosopher who says
'But it may still be coincidence.
Where do you draw the line and say, "Now we know it was not coincidence, now we know that quinine has an effect"?'
Again,
consider the series of questions:
'Do flowers feel?'
'Does the amæba feel?'
'Does a worm feel?'
'Does a dog feel?'
Here the ordinary man says 'Yes', and the scientist may say 'No'.
The philosopher proceeds,
'Does a baby feel?'
'Does your friend feel?'
and he says that we do not know that they do.
Then there is the series which
begins with the policeman asking whether you are sure that this is the man you saw hanging round the garage,
and ends with the philosopher asking whether you are
[58]
sure that the woman who comes down to breakfast each morning looking like your wife is always the same woman.
Excellent.
Ostensibly the policeman is deadly serious and the philosopher is not (quite); and yet the serious, pragmatic one turns out to be a Naive Realist while the jocular, speculative one is a Radical Skeptic. There is much to chew on here, I think.
It is all this sort of thing which leads people to think of the philosopher as a super-scientist.
Well, he
is
trying to answer the same
kind
of question that a scientist might try to answer. The difference is a difference in method.
The philosopher's method does loom
supra
the scientist's, despite being more limited overall, and despite (I think?) ultimately being parasitic upon long-settled empirical work rather than free from any dependence upon it. Again, it depends on which sense of "super" is meant here, and (befitting the occasion) this is not unequivocally clear.
Nor is it only the epistemological metaphysical theories which are associated in this way with science.
For through the epistemological theories
the ontological theories also are associated with it.
Instead of asking
'Do we know that there are Siamese cats?'
we may ask
'Are there Siamese cats?'
And instead of asking 'Are there Siamese cats?' and answering
'There are none really, only dogs which look like cats,'
we may, because there are such appearances of Siamese cats as there are, ask instead
'Are Siamese cats really cats?'
or
'What are Siamese cats really?'
Similarly,
there being such appearances of chairs as there are,
instead of asking
'Are there chairs?'
and answering
'There are none, only families of sense-data',
we may ask
'What are chairs really?'
and answer
'They are really families of sense-data'.
There is an intermediate form of words
namely,
'There are cats but only in a Pickwickian sense'.
'There are chairs but only in a Pickwickian sense'.
Propositions about cats and chairs will then be said to be phenomenally true.
2.
Logical Theory of Metaphysics.
The metaphysician is a profound logician
engaged on an
a priori
science of definitions?
This new formulation 'What is a chair?' was of the utmost importance.
For though one may use this form of words 'What is an X?' in asking for fresh factual information,
like the wife of an instrument-maker who, though she knows anemometers well by sight, may suddenly ask 'Jack, what is an anemometer?',
one may also use it in a very different way,
the way Socrates explained he was using it when he asked
'What is virtue?'
From his explanation it is clear that
he was asking for a logical definition.¹
Consequently if, in order to avoid the hint of ridiculousness in the question
'Are there such things as chairs?'
we ask
'What are chairs?'
then we use a form of words which at once suggests an analogy between the philosopher and,
not the scientist,
but the logician.
Now
as science grew
and people saw better
how it is based on observation and experiment
there grew a suspicion of anyone who professed to have obtained new factual information by
¹ There are reasons for saying that
he was really wanting a metaphysical 'definition'.
[59]
anything but empirical methods.
Nowadays of course we
see
better how observation
can be selective,
experiment
can be poorly controlled,
and how we'll
never even know
about the vast majority of experiments or observations, no matter how hard we try to find out about them.
So, kindly forgive my occasional bouts of
empiricist
suspicion.
Imagine Empiricism done right!
There were sneers at the philosophers who were represented as employing armchair, a priori methods.
The deplorable affair of Hegel and the planets was not forgotten.
From Googs:
- The Dissertation (1801): Hegel sought his teaching license at Jena with this work [De Orbitis Planetarum], proposing a system for planetary distances based on Plato's Timaeus, which suggested a gap between Mars and Jupiter.
- The "Empty Space" Controversy: Just as his dissertation was published, the asteroid Ceres (the first of the belt) was discovered in that very gap, leading to widespread criticism that he used pure logic to deny empirical evidence.
Besides, philosophers could not agree, were in fact in a deadlock—at each other's throats.
To this situation came Moore and Russell,
Moore from a study of Plato and Aristotle, and Russell from the world of logic and mathematics.
They revolutionized philosophy for many of us by
reinterpreting the philosopher's 'What is an X'
as a request for logical analysis.
...
Russell wrote
'. . . every philosophical problem, when it is subjected to the necessary analysis and purification, is found either to be not really philosophical at all, or else to be, in the sense in which we are using the word, logical'.
Moore writes,
'I am not at all sceptical as to
the truth
of such propositions as "the earth has existed for many years past" . . . But I am very sceptical as to what, in certain respects,
the correct analysis
of such propositions is'.
This theory of metaphysics
provided a response
to the metaphysical challenge
'If a metaphysical theory isn't empirical, what is it?'
(How often is this club used to induce us to accept some simplifying lie.)
Like all good big ideas it seemed the simple, natural thing to say, the moment someone said it.
And it proved itself by breaking up the deadlock.
The theories of descriptions and numbers, and the theory of classes, were all pieces of logical analysis
and they worked like charms on many hitherto incurable philosophical complaints.
The proof of the unreality of space, the ontological argument for the existence of God and the extra entities in the universes of discourse all went up in smoke, though from the fictional entities there lingered still a peculiar smell.
Thus these analyses showed themselves to be what philosophers needed, what they were really asking for.
But there were little clouds upon the horizon.
(1) Unless the positivistic analyses were correct the epistemological difficulties 'How do we know that chairs and other minds and value predicates exist?' could be
[60]
answered only by
Intuitionism (special way of knowing),
Scepticism (we don't know),
Dogmatism (we know somehow — never mind how).
(2) Unfortunately in the most fundamental cases the positivistic analyses could never be got both complete and correct.
I could benefit from some specific examples of positivistic incompleteness/incorrectness. I don't understand how we could be expected, before the fact, already to know what would make a "positivistic analysis" "complete." Similarly, as we await positive evidence, I'm having trouble imagining just what would constitute correctness and incorrectness, aside from obvious cases of one piece of evidence directly contradicting another.
This encouraged people to say what Broad and sometimes Russell were inclined to say, namely that
the philosopher gave the analysis not of all the plain man does say
but of all he has any right to say.
We shall see
how these points are connected with the nature and fundamental defects of the analysis theory.
3. The metaphysician looks for
the definition of the indefinable.
3.1. The meaning of this statement.
To say that metaphysics is analysis is unsatisfying in certain respects.
To begin, with people are apt to ask,
'And what is analysis?'
... when James says that an emotion is nothing but a complex of sensations ... 'And what is a sensation?' ...
... 'A nation is nothing but the individuals which make it up' ... 'What are individuals?'
These responses are hardly complaints,
and are therefore to be contrasted with those prompted by
'A nation is a society of a certain sort',
'An instinct is an unacquired disposition involving consciousness'.
These definitions are apt to make people say,
'But that is just what I want to know.
What is a society?
What is a disposition?'
Unlike these, the question
'And what is analysis?'
is not a complaint that we have not provided the right stuff,
but a request for the same again.
It can be called a complaint against incompleteness if you like, but if so, it must be distinguished from the complaints
(a) of another sort of incompleteness and
(b) of incorrectness, to which we must now attend.
For
'Metaphysics is analysis'
is apt,
in the second place,
to provoke the challenge
'Not all analysis is metaphysics.
Now which sort is?'
We are reminded at once
of how people respond to
'Statements about nations are just statements about individuals'.
They are apt to say
'Ah! Now which statements about individuals
[61]
are about nations?
Complicated ones, no doubt.
But be more specific.
Take any statement about a nation and tell us which statements about individuals make it up'.
Now
there are two reasons
for which people make this challenge.
(1) They may make it because they hold that statements about nations are not statements about individuals nor compounds of such statements, or are so only in a very extended sense.
(2) They may be satisfied that nation-statements are individual-statements, perhaps in a somewhat stretched sense, but wish to ask what distinguishes those individual-statements which are nation-statements from those which are not—
like one who is not troubled about the sort of entity a cow is, but feels he doesn't know what the essence of the distinction between cows and other things of the same sort is, because he feels sure this is a matter of shape or colour or upper incisors or something, while yet nothing of this seems essential.
But let us attend first to the matter
of the genus or general nature of metaphysics.
I want to contend that metaphysics is not analysis or, if you like,
that it is so only in a dangerously extended sense.
The extension has been illuminating in the past, but it is an extension and as such it may, and in fact now does, mislead.
Questions of this sort
as to the general nature of
metaphysics,
analysis,
mathematics,
ethics,
poetry,
fiction
have been made more difficult by defining these subjects as respectively classes of metaphysical, analytic, fictional, etc., propositions, and then asking to what genus these classes of proposition belong.
It is better to describe mathematics, analysis, metaphysics, poetry, as sorts of game played with words as pieces,
the usual significance of which must of course be understood by the players,
and then to define metaphysical, analytic and fictional questions and sentences by reference to the purposes they serve in the game.
Well okay, somebody was privy to Wittgenstein's work-in-progress long in advance of most of the rest of the world. Not that the argument is any worse for it . . . , but it is entirely unremarkable, then, that Bambrough should find this paper to be ahead of the . . . uh, . . . game.
Now
I contend that metaphysics is not analysis,
and that metaphysical questions are not requests for analyses,
even though they are expressed in forms of words
which may be used in the analysis game to ask for analyses,
and that metaphysical answers are not analyses
though they may be forms of words
used in the analytic game to state analyses.
Of course one could say instead that the questions are requests for analyses and the answers
[62]
analyses, only all with a hidden purpose,
just as one may say that 'What is the meaning of good?' when asked by a philosopher, is a verbal question but with a hidden purpose, instead of saying that it is not really a verbal question.
To say that analytic statements are verbal is useful if one wishes to get rid of the idea that they differ from statements about words in the way that statements about dogs differ from statements about cats or statements about colour differ from statements about shape and size.
One might express this by saying that
to say that analytic propositions are verbal is useful in getting rid of the idea that they differ from ones that are not analytic in being about a new species of thing or 'in subject-matter'.
I shall use the last expression although it involves deciding arbitrarily not to describe the sort of difference there is between analytic and verbal statements as a difference in subject-matter.
The excuse for doing this is that
for every statement about abstract entities
—propositions, characteristics—
there is a verbal statement which makes the same factual claims
though its meaning is different.
...
[e.g.]
If I say that
'good'
means
'approved by the majority',
in so far as I make any factual claims they are justified by what people would say, that is, are verbal;
but my purpose, or primary intention, in saying so, is not verbal.
Suppose a Chinaman is decoding an English message, and does not know the meaning either of 'vixen' or of 'female fox', but says after investigation, '"vixen" means the same as "female fox"'.
He says this though he knows his hearer also does not know the meaning of either expression.
Suppose now Smith says '"vixen" means "female fox"', because he believes his hearer knows the meaning of 'female fox' but not of 'vixen'.
Suppose now someone says, 'A vixen may be defined as a female fox'.
The factual claims involved in the statements are the same.
But the purposes they serve are very different,
and this makes us speak differently about their meanings.
A hearer understands the Chinaman's statement though he understands neither 'vixen' nor 'female fox';
but only if he understands one
[63]
of the two does he understand Smith,
and only if he understands the meaning of both does he understand or, if you like, fully understand the philosopher.
Hence statements about propositions and characteristics such as
'He asserted the proposition that Africa is hot',
'She is chic',
can be turned into statements about words
'He uttered the sentence "Africa is hot"',
'She is what the French call "chic"',
if, and only if the verbal statements are so used that we say that a man understands them only if he understands not merely the expressions 'the sentence "Africa is hot"' and 'what the French call "chic"' but also understands the sentence 'Africa is hot' and the word 'chic'.¹
I hadn't the slightest clue what verbal statements really means here until I read the footnote . . .
It follows that
to say that
analytic statements,
such as
'Phœnixes are birds which renew themselves from their own ashes',
are verbal statements,
emphasizes their likeness to the statements of translators and decoders
and their unlikeness to scientific statements
such as 'Phœnixes do not drink before sunset'.
At the same time it slurs over a serious unlikeness between analytic statements and those of translators and decoders,
an unlikeness which is emphasized if one insists on the platitude that they are not verbal, but are of their own peculiar sort.
It's not at all clearcut that the drink statement could not equally well be analytic nor that the renew statement could not be scientific. Something of the "unconditionality" of the "stone deaf" remark seems to apply here too.
Now
I wish to emphasize the unlikeness
between metaphysical questions
and answers and analytic ones.
So I say that
metaphysics is not analysis,
metaphysical questions not requests for analyses,
and metaphysical answers not analytic definitions,
or, if you like,
not merely analytic (compare 'not merely verbal').
Of course
if
'Metaphysics is a sort of analysis'
is taken like
'A motor cycle is a sort of horse though it can neither gallop nor jump'
then it is not incorrect—how can such statements be incorrect?—and it is illuminating, but it is dangerously apt to mislead.
It is apt, for example, to mislead people into rejecting statements of the sort
'Nation-statements are individual-statements',
'Analytic statements are verbal',
as not to the purpose,
¹ (a) To deny this results in the extreme paradox 'All statements are verbal'.
(b) This is part of the explanation of the necessity of necessary statements.
For such statements connect abstract things and are therefore purely verbal
in a way in which 'He asserted Africa is hot' is not;
that is,
they are purely about the use of the expressions they connect.
And what they assert must be known to the hearer if he understands them.
Hence, if he denies them, the speaker says the hearer does not understand.
This is characteristic of necessary statements.
Logically necessary statements are checked by the actual usage of language and to this extent may be called true and false.
Metaphysically necessary statements only have excuses in the actual use of language and so can only be called 'excusable and 'inexcusable'.
This footnote should have been a paragraph in the main text.
[64]
because they are wrong in analysis.
It is apt, for example, to make people reject such a statement as 'Metaphysical questions are analytic questions' as not to the purpose in metaphysics, because it is wrong in analysis.
I do not reject
'Metaphysics is a sort of analysis'
as
not to the purpose in metaphysics because it is incorrect in analysis,
but I bring out its dangers in metaphysics by bringing out its incorrectness in analysis.
It will be noticed
that one who asserts
as an analytic proposition
that metaphysics is analysis
cannot defend it on the lines I have just indicated.
He will have to defend it on the lines that it is correct as an analysis.
In disputing this I shall be doing what I want to do, bringing out how it misleads as metaphysics.
If the metaphysician really wants analysis
it is a curious thing that
nearly every formula for giving definitions
which is submitted to him
he rejects,
either on the ground that the definitions it yields are not sufficiently profound to be called metaphysical,
or on the ground that the definitions it yields are not definitions because they are incorrect.
If when a metaphysician asked
'What is a penny?'
we were to reply
'A coin, English, etc.'
that would be no good to him.
Even if we said 'A material thing which, etc.' we should only have succeeded in stating the essence of the problem which he put in the misleadingly specific form 'What is a penny?'
On the other hand, if we reply: 'Pennies are just bundles of sensations' or, in order to avoid tiresome objections, 'Statements about pennies are just statements about sensations',
then the metaphysician complains that there are no statements about sensations which mean the same as any statement about a material thing.
The metaphysician
is like a man who,
meeting an old friend in disguise,
asks us
'Who is this?'
and then if we merely sponge the old friend's face and straighten his tie, says 'Still I don't know who it is',
while if we pull off the false beard and the wig, he says 'But this is not now the same man. The man I asked you about had a beard and a wig'.
Nothing satisfies him.
Or rather
—and the change is important—
nothing in the way of analysis satisfies him.
[65]
Now
I contend that this is because
it is not what he wants.
When a man tells me he would love to ride a horse
but, no matter what animal I offer him, says 'Not that one',
then I think he does not really want to ride.
Especially if, when offered a motor cycle, he takes it with alacrity.
Compare Carroll's bus to Baton Rouge.
Of course, you may say if you like that I did not find the right horse.
Or you may say that he wanted a perfect horse and then explain that a perfect horse is impossible, since, unless its speed was unlimited, we might at any time wish it could go faster, while if it could always go faster when we wanted, then, like an aeroplane, it would shrivel space and thus be unable to conquer distance.
And this
is
what is said
about the metaphysician—
that we have not found the analysis of what he wants analysed,
or, in other cases, that what he wants analysed is ultimate, unanalysable.
I am content to accept the latter statement but not the former.
Instead of saying
'The man who said he wanted a horse
really wanted a motor cycle'
one might say
'He wanted a horse but needed a motor cycle'
or
'What he wanted primarily, apart from his wrong idea about what he wanted, was a motor cycle'.
Now many metaphysicians, nowadays, put their requests in the form
'What is the analysis of X propositions?'
and because of this we may feel it best to say that they want analysis but need something else.
On the other hand in view of the fact that other people with no axe to grind in the way of a theory of metaphysics ask very like questions in the older form
'What is the ultimate nature of so and so?'
and are then satisfied by judicious description,
we may feel it best to say
'What metaphysicians want, or really want, is not definition but description'.
Hmm. That would make them under-scientists? Apprentices? Science begins with judicious description and proceeds from there, no?
3.2. Proof that the metaphysician looks for the definition of the indefinable.
The metaphysician's request is the limit to which a series of requests for analysis approaches.
Now requests for analysis approach nearer and nearer to requests for the impossible.
This appears if we consider for a moment what has sometimes been called 'the paradox of analysis'.
It is sometimes claimed that analysis is impossible for the following reasons:
To analyse p is to translate the sentence 'p' into another which expresses a proposition of different structure, form and elements from p.
But any proposition which has not
[66]
the form and elements of p is not the same proposition as p.
Hence
any sentence which expresses a proposition having structure different from p does not express the same proposition as 'p',
i.e, does not mean the same as 'p',
i.e. is not a translation of 'p' and hence does not provide an analysis of p.
Now
of course
analysis is possible.
It is done.
Universal affirmative propositions are not ultimate;
they can be analysed into propositions which are not universal affirmatives.
But a consideration of the paradox of analysis enables us to see on what the possibility of analysis depends.
Universal affirmatives are analysable because we do not mean by 'universal affirmatives' 'the propositions expressed by English sentences of the sort "All S is P" or any which mean the same, no matter what devices they employ'.
True, we do not mean by 'universal affirmatives' 'those propositions which are expressed by English sentences of the form "All S is P"'.
They can be asserted in French.
But we do mean by 'universal affirmatives' those propositions which are expressed by sentences using devices having a use like 'All' or 'Every' or 'Tous les'.
Consequently we do not say of a man who says 'There were lions there, but there was not one which attempted to harm him', that he asserts a universal affirmative proposition.
At the same time in the sense of 'same proposition' required by 'mean the same' in the sense of 'mean the same' required by 'translation of',
we should say of a man who says
'There were lions there, but there was not one that attempted to harm him',
that he asserts the same proposition as one who says
'All the lions there made no attempt to harm him'.
In general:
X propositions are analysable into non-X propositions,
only if the differences we require between two sentences
in order to say that one expresses an X proposition and the other does not,
do not conflict with the resemblances we require
in order to say that two sentences mean the same.
We are tempted
on the one hand
to use 'mean the same' so narrowly
that if two sentences, though they convey the same information, function otherwise very differently, e.g. express propositions of different form, then they do not mean the same.
Directly we do this, it becomes impossible to analyse propositions of one form into propositions of another.
And we are continually tempted
to define 'X propositions'
[67]
(the ones we want analysed),
as
'those which are often dressed in English sentences of kind "X", but may be dressed in any sentences which mean the same'.
Directly we do this it becomes impossible to analyse X propositions into non-X propositions.
We do not always yield to these temptations,
so analysis is possible.
Thus if I mean by 'a fictional proposition', such a proposition as you or I now make if we assert that Mrs. Bardell had fainted in Mr. Pickwick's arms,
then one can analyse fictional propositions into non-fictional as follows:
If A asserts of some author's character, S, that S had a property P then X means 'In the accepted text it is written that S had P'.
But it easily happens that
we yield to the temptations without knowing it
and then analysis becomes impossible
without our knowing it.
Thus suppose I mean by a fictional proposition
'such a proposition as an author expresses when he says of one of his characters "S had P"'.
Now here any sentence which could be said to mean the same as the one the author used would have to fulfil the same sort of imaginative, non-informative, purpose;
and 'fictional proposition' in the sense now in question is so used that any sentence used with the same sort of imaginative purpose would be said to be a fictional proposition.
Hence it is impossible to analyse fictional propositions in this sense.
That is, in this sense fictional propositions are ultimate.
Now
it is my contention that
this happens again and again in metaphysics:
that is,
that 'X proposition' is used so widely and 'mean the same' so narrowly,
that the only correct answer to the question
'Are X propositions analysable into others?'
is that they are not.
This is frequently obscured by the fact that the expression 'X proposition' has a narrower use than that given it by the metaphysician with whom we are speaking,
while nevertheless, that narrower use is so profound that definition of it is fairly called philosophical and even metaphysical.
These claims can be supported
only by looking at
metaphysical questions and answers
and the reactions people have made to these.
Let us now do this.
3.21.
Formal Puzzles:
(a) Numerical propositions.
People have asked
'What is the nature of numerical propositions?'
meaning
not mathematical propositions but such propositions as
'Two
[68]
men were murdered, three women drowned'.
Russell offers the analysis:
'Not three people are interested in mathematical logic'
may be expressed in the form:
'If x is interested in mathematical logic, and also y is interested, and also z is interested, then x is identical with y, or x is identical with z, or y is identical with z'.
And this satisfies some people.
Their being satisfied does not prove that their question was not profound enough to be called philosophical or even if you like metaphysical.
But I do want to claim that some who use the same form of words would not be satisfied, would say that they were asking a more general question involving a wider use of 'numerical proposition'.
They would say that Russell's new formulation, which comes to
'Something is interested in mathematical logic and something not identical with it is interested in mathematical logic',²
still expresses a numerical proposition, and thus does not provide an analysis of numerical propositions into non-numerical propositions but only sponges their faces.
We should say of a man
who was dissatisfied
with a definition of instinct
as
a disposition
that he was concerned with a more general, more profound, more really metaphysical question than one who was satisfied with it.
Both used the same form of words,
'What is an instinct?'
but the one who is dissatisfied with the answer in terms of dispositions was really only using this form of words to ask a deeper, more metaphysical question,
'What is a disposition?'
And I want to claim that we should say the same of the man who was dissatisfied with Russell's definition.
I am well aware
that Russell's definition of numerical propositions
has important claims to
the title 'philosophical'
and even 'metaphysical'
which the definition of 'instinct' in terms of 'disposition' has not,
and that therefore the question to which Russell's definition is an answer has such special claims.
My point is that beyond Russell's question is another which has still better claims.
For there are and have been people dissatisfied, in the special way indicated, with Russell's definition, saying that it was not sufficiently profound.
What did they want?
¹ ...
² This is a slip. The latter expression means 'At least two people are interested in mathematical logic'. J. W. 1952.
[69]
Joseph,
in discussion some fifteen years ago,
said that Russell's definition was circular
in that his new definition involved the notion of plurality and was thus in essence still a numerical proposition.
It was said at that time that he had not grasped the point of Russell's definition and, as I have said, I do not wish to deny the logical, even the philosophical value of Russell's definition.
It removed the impression that numbers were a special kind of quality applicable to groups.
On the other hand those who derided what Joseph said were metaphysically blind, in that they could not feel the inclination to ask his wider, profounder question.
Let us look
at this profounder question carefully.
It differs from Russell's not merely in that some answers which would satisfy Russell would not satisfy the profound metaphysician, but in that
no answer or, rather, no answer in terms of a definition would satisfy the metaphysician who asks it.
For he asks that we should analyse numerical propositions into non-numerical propositions.
And yet it soon comes out that for him, if a proposition involves the notion of plurality, it is a numerical proposition,
and that a proposition involves the notion of plurality if it involves the notion other than.
Now no proposition involving the notion other than can be analysed into one which does not.
Further, if we take any numerical proposition p₁ it will, as everyone admits, involve plurality and thus the notion other than.
Hence if p₁ is analysed into p₂, p₃ will involve the notion other than and thus plurality;
and thus on the metaphysician's use of 'numerical proposition' p₃ will again be numerical.
Hence the class of propositions to which the metaphysician refers by 'numerical' cannot be analysed into propositions not in that class.
Which is what Joseph wished to show.
How natural it is
to proceed as the metaphysician does
may be brought out more plainly as follows:
The metaphysician asks that numerical propositions shall be defined in terms of non-numerical propositions.
At the same time he very naturally defines numerical propositions, not as those which are expressed by English sentences of the sort,
'Two S are P',
'Three S are P',
etc.,
nor as those which are expressed by these English sentences and their French, German, etc., translations,
but as those expressed by the English sentences and any sentences, whatever devices
[70]
they may involve, which stand for the same propositions, mean the same.
Nothing could be more natural, but the result is a self-contradictory request.
For to analyse numerical propositions is to translate sentences which stand for numerical propositions into others which do not.
But if a new sentence, S₂, is a translation of one, S₁, which stands for a numerical proposition, then S₂ will mean the same as S₁
and therefore, with the metaphysician's use of 'numerical proposition', the proposition it expresses will not be non-numerical.
Hence with this use of 'numerical proposition' numerical propositions are ultimate.
I prefer to put this by saying that the description
'analysed numerical propositions into others'
describes no process, or the description
'translated sentences expressing numerical propositions into sentences which do not'
describes no process.
The descriptions describe no process not merely in the sense in which
'completed the old course at St. Andrews in 25'
describes no process which has been or will be carried out,
but in the sense in which
'squared the circle',
'found a perfect movable pulley in which W did not balance ½W',
'found a leopard without spots and a purple Union Jack'
describe no process.
It might be thought
that there is a more drastic answer
which would help the metaphysician.
If a child were to ask us
'What does "Two men were murdered" mean?',
part of what we should do to explain would be to say
'Well, if Alf and Bill were murdered, or Charles and David, then two men would be murdered'.
And then, anxious to prevent the idea that only Alf and Bill, Charles and David, would count, we might add
'and so on'.
So we might suggest that 'Two men were murdered' means 'Either Alf and David were murdered or Charles and Bill were murdered, or . . . and so on'.
The metaphysician would say:
'When you say "and so on" is this short for a list you could put down?
Then your translation does not mean the same as the original.
But if it means "or any other man and another" then your translation is only another form of Russell's, and you have again given us a numerical proposition involving the notion other than'.
Or the metaphysician may object that the 'specimen' facts, e.g. 'Alf and Bill were murdered', themselves involve plurality.
Thus once more
it appears that any sentence
which means
[71]
the same as the original
will again be said to express
a numerical proposition.
No wonder that numerical propositions are ultimate.
3.21.
Formal puzzles:
(b)
Generality.
It is the same with general propositions such as
'Something moved'.
People complain that there is no finite list of propositions of the sort 'Smith moved' which disjunctively combined mean 'Something moved'.
We may say if we like that
there is an infinite list which does,
and this brings out an important feature of the relationship between 'Smith moved' and 'Something moved'.
This feature we can also bring out by saying
'The more specific propositions we put in our list, the nearer does a disjunction of them come to meaning the same as the general proposition'.
But when we say that an infinite disjunction would give the meaning,
we mean not only the fact just mentioned, but also the fact that
however long the list is it will not give the meaning unless we call a list with 'and so on' at the end a definition.
Or
again people complain
that the 'specimen' facts,
e.g. 'Smith was murdered',
again involve generality because 'Smith' is a disguised descriptive phrase.
The suggestion is that all would be well if we could analyse these into statements about our present sense data such as 'This is red'.
If these explanations
are not definitions
then there is no definition of general propositions into non-general.
It may now
be said
'The conclusion you have reached about numerical and general propositions,
and incidentally about fictional propositions,
is the perfectly familiar but utterly unhelpful one that they are ultimate'.
I am delighted to welcome this criticism.
This is precisely what I want to say myself,
that this answer
'Numerical propositions are ultimate'
is quite inadequate,
leaves the philosopher feeling that somehow there was something he wanted which he has been denied.
As I have explained, I want to go further and generalize and say that
whenever in answer to a metaphysician's question
'What are X propositions?'
the answer is given
'They are ultimate'
then that answer is
(1) correct, but
(2) inadequate.
3.22.
Category puzzles:
(a)
Time.
Let us now take an example from another class of metaphysical questions.
These arise, not from the 'queerness' of the form of the class of propositions which is felt to be queer, but because of the queerness of the
[72]
category of what they are about.
These are apt to present themselves in the form
'What is X?'
'What are X's?'
where 'X' is a name for the puzzling category, or a name for a species of that category.
We have for example,
'What are characteristics? What are numbers? What are abstract, necessary propositions about?'
and we have
'What are chairs and tables? What are material things?' 'What is a nation? Is it something over and above the individuals of that nation?'
'What is Time?'
Suppose we offer an answer to the last
on the lines suggested by Moore.
We say,
'Well, when we speak about Time, what are we talking about?
Such facts as this — that lunch is over, supper to come, that Smith's anger is past and so on.
Let us call such facts "temporal facts".
Then "Time is unreal" can be translated into the concrete (Moore's phrase) by "There are no temporal facts"'.
When we read this, we draw a breath of relief.
This is the stuff.
With this translation into the concrete, we get 'the cash value' (Broad), the predictive power, of the statement 'Time is unreal'.
What a contrast to the answer
'Time is an abstract entity, super-sensible, having a sort of existence all its own'.
For the latter answer only tells us that time is not brown or yellow, not big or little, not to be found in the bathroom, and like Space only different.
Such an answer only emphasizes what ordinary language suggests that, besides the facts that lunch is over and his anger past, there is the fact that Time is real.
True, Moore did not find a definition, but
he showed how it was a mere accident of language that we could not provide a definition
and thus remove an uneasy feeling about Time,
just as we did when we had the uneasy feeling that though the class of all men is not to be identified with its members, yet there were not in addition to the facts about men,
e.g. that men exist, that all are mortal, facts about the class of men,
e.g. that it exists, has members, has members which are mortal.
We may say here
that Moore meets a philosophical request,
even, if you like, a metaphysical request,
not indeed by finding but by creating
an analytic definition.
I do not wish to deny this
any more than I wish to deny
that Russell did the same with his definition of number.
They both translated sentences which trouble us into others which do not.
[73]
But of course there will be people
dissatisfied with this answer.
They fall into two very different classes.
In the first place, some will say
'What is meant by "Time is real" is different from what is meant by "Either lunch is over, or my supper yet to come, or his anger is past, or something of that sort (i.e. and so on)".
What I have in mind when I say "Time is unreal",
is very different from what I have in mind if I say 'Either lunch is over or etc."'
And yet these same people, some of them, will not be satisfied by any definition which does not put 'Time is unreal' in terms of individual events.
So they reject every definition either on the ground that it is incorrect or on the ground that it is not sufficiently profound.
The nature of this difficulty in metaphysics and the light it throws upon its nature will appear later.
In the second place,
there are people who will say
that even the definition in terms of individual events or temporal facts is not sufficiently profound;
they ask that the definition should be taken further.
And if we say
'By temporal facts we mean such facts as "Smith's anger is past" and so on'
they complain
(a) against the 'specimen' fact as again involving time,
or
(b) against the phrase 'and so on'.
It is soon clear that there is nothing to be done for them with regard to (a).
We may try saying 'His anger is past' means 'He was angry and is not'.
But it is a hopeless game.
The reply will be that 'He was angry' involves Time again.
And of course it involves it if any sentence using a verb with a tense involves Time.
And it is now apparent that this is how the metaphysician uses 'involves Time'.
He cannot translate sentences which work like the ones with tenses into sentences involving only a timeless 'is' such as 'Red is a colour'.
But this is what he wants.
Nothing short of it will do, nothing else will be a reduction of temporal facts to non-temporal.
No new language will help.
Suppose we invent a new language with no time-indicating words such as 'was', 'will be' and no time-indicating endings such as 'ed'.
Now the new sentences, if they are to provide translations, must do the work the old did;
so there must be differences between those which correspond to the old 'was'-sentences and those which correspond to the old 'will be'-sentences.
We might put them in different coloured inks
[74]
(compare Ramsey's writing negatives as positives upside down).
But then the new sentences would surely again express temporal propositions.
I am well aware that
there is nothing novel
in the conclusion
that temporal propositions are unanalysable, and that
this unanalysability is not a matter of our being unable to do or find the analysis
but of the nature of the facts.
Prof. Broad supports far more fully and carefully than I have done this very conclusion about temporal facts.
I am so sorry
that he advances his conclusion
tentatively.
He says
'And so, prima facie, the temporal copula has not been analysed away.
Of course it may be answered that this objection depends simply on defects in the language that we speak.
It may be so. But I am more inclined to think . . .'
What is he in doubt about?
He knows our own language well.
His own examination has shown that there are no sentences in English which we should say do not stand for temporal propositions which are nevertheless such that we can translate those which do into them.
We may bet anything that the same is true of French and German.
Is Broad asking whether, if we invented an entirely new and very different system of symbols into which we could translate temporal sentences, we should be satisfied —
should be willing to speak of having translated sentences which express temporal propositions into sentences which do not?
There may be people who would be satisfied by this.
Sometimes metaphysical difficulty is removed by inventing a word when it would not have been removed without such an invention.
We cannot say that Ramsey's writing negatives as positives upside down was useless.
On the contrary it distilled the problem of negation into such purity that it vanished;
but it did not achieve its result by analysing negative sentences into non-negative sentences.
And this is what signifying tense by colour would do.
After the removal of 'not' it could not be complained that the negative sentence was misleading in its shape.
We may doubt with regard to a particular person whether he would be satisfied if such a new language were introduced,
just as some might be satisfied by a definition of 'two' in the symbolism of
[75]
Principia Mathematica,
while they would be dissatisfied with any in English.
But we also know that there are many who would say that such people would be wrong to be satisfied.
By a study
of how these difficult people
use 'temporal sentence'
we come to learn that they mean what anyone reading this will now be inclined to mean by it, namely any sentence conveying temporal information.
By a study of how these metaphysically-minded people react to proffered definitions
we can see that no peculiarity in the way the sentence performs its work, the images it raises, the feelings it arouses, will prevent them calling it a temporal sentence
provided its cash value is the sort of cash value which
'He was angry'
has,
i.e. provided it is a translation of sentences with 'was' and 'will' and 'is now' in them, used as they are now used.
We know, then, that they so use 'temporal sentence' that
'a translation of temporal sentences into ones which are not temporal'
is a self-contradictory description.
Suppose a man says he cannot find a perfect movable pulley in which W does not balance ½W.
By prolonged search I find a golden pulley so smooth that the highest-powered microscopes detect no deficiency in its surface, on which, nevertheless, W does not balance ½W.
But he explains that by 'perfect' he wished to exclude so dazzling a material as gold.
From the fairies I obtain a gold which does not dazzle or from the angels a pulley in blue.
But now he complains of the microscope, says that a larger one would reveal defects even if the pulley has been sent down from heaven.
But even a heavenly microscope does not convince him.
God no doubt possesses a microscope of His own which would reveal what no other does, namely, the defect which must be there because if the pulley were perfect W would balance ½W.
Is it not apparent
that the well-known facts,
including
the fact that the smoother we make pulleys the more nearly does W balance ½W,
have induced him to so use 'pulley' or 'perfect pulley' that for him balancing W by ½W is part of the connotation of the expression, a necessary condition of a thing's being a perfect pulley?
Some people nowadays would not call a loaf 'bread' if it failed to react to the chemical tests for starch,
though others, whose usage of 'bread' has not been equally affected by science, would call it 'bread'
provided it was otherwise satisfactory.
[76]
Surely we have come to know
how this man uses 'perfect pulley';
and that in his usage
the request for a perfect pulley
on which W does not balance ½W
involves a description which describes nothing,
is self-contradictory?
There is no process of calculation by which we can show this to him,
as we can show him that when he asks for the proof that 2345 multiplied by 3 equals 6935 he is making a request which is self-contradictory.
But neither is there a process of calculation by which we can show him that a request for a proof that 1+1=1 is a self-contradictory request.
But how many of us are eating bread with no starch? And of those, how many do find it otherwise satisfactory?
The bread example is more obviously of relevance to 'Aesthetics', more so at least than any matter of calculation could be; and yes, it is here rendered as a case of some people reaching a different conclusion than some others. And yes, it does seem that to make such a divergence happen requires some forcing of the issue; e.g., in this case, one group has had their very taste irreparably affected by science. An apt example for our own times, this.
Now leave "science" aside. What else could play this role of catalyzing divergence? Spitballing: Americans of the first or second generation seem apt to get quite worked up over the authenticity of ethnic dishes served in restaurants, especially when it's a matter of 'how my grandmother made it.' Is this a "metaphysical" issue? It seems to become so only by virtue of circumstance, not because it is intrinsically important or even interesting. e.g. Appropriation of the dish by another group, or for profit by anyone at all, seems quite apt to raise hackles.
So, my deeply un-philosophical reaction here is to wonder if all "metaphysical" problems are not circumstantial; if metaphysics is indeed a game that we can get out of, or better yet, never get into?
In the bread example, the word people occurs for the 36th time out of 60 total occurrences . . .
difficult people
there are people who will say
of course there will be people dissatisfied with this answer
... again people complain that the 'specimen' facts ... involve generality ...
It is apt ... to make people reject such a statement ...
He's probably just talking about other philosophers, who presumably are in precisely the business of catalyzing their own divergence. As for all the rest of "those people", there's usually a simpler reason, no?
He appears to ask for something
because,
though his description
'a movable pulley in which W does not balance ½W'
may be used as he uses it,
there is, or we can easily imagine, a use in which such a description might describe something,
'Laws of nature broken; Scientist finds golden pulley, etc.'
There is, or we can easily imagine, a use of 'Union Jack' in which one could find one in purple, green and yellow;
you know what such a flag would be like — a Union Jack in purple, green and yellow.
When a philosopher asks for the unattainable, the fact that he is doing so is frequently obscured by the existence of a 'contingent copy' of what he wants — like those which we find for the pulley and the flag.
And the requests for analysis are no exception.
The philosopher complains, perhaps, that we say that his request for the analysis of X propositions into non-X propositions makes no sense.
'On the contrary', he says, 'it is quite easy to explain what I want.
You know what general sentences are, don't you?
And you know what non-general sentences are, don't you?
Well
what I want is that the former should be translatable into combinations of the latter without the use of such expressions as "and so on"'.
Now this describes a perfectly possible ideal.
What is there to prevent our altering our use of English so that this should be the case?
But this of course would not be what is wanted by the philosopher who is asking for the analysis of general propositions into non-general propositions.
It only looked as if it were what he wanted because we allowed him to translate his request into
'Are general sentences translatable into non-general sentences?'
and then forgot the reminder for the formal mode ...
Had we remembered this we should have realized that what he asks is that general sentences used as they are now used should
[77]
be translatable into non-general sentences used as they are now used
and without the use of 'and so on' used as it is now used.
And this is like asking that 20 shillings should have the value they now have and be equal in value to 20 half-crowns, meaning here again not just coins of the half-crown shape and size, but coins of half-crown value.
Here is
another factor
which prevents metaphysicians noticing how, as requests for definitions, their requests are self-contradictory.
Let us sum up
our conclusion so far
with regard to the question 'What is Time? or 'What are propositions about Time or involving Time?'
(1)
There are people
who ask these questions
and are satisfied by such a definition as Moore's
or by the definition
'"Time is real" just means "There are events such as race meetings, collisions, wars"'.
(2)
There are people
who are dissatisfied
with any definition of Time
in terms of
a class of temporal facts,
because such definition does not give us statements which mean the same as statements about Time — destroy its unity, etc.
(3)
There are people
who are dissatisfied
with any definition of Time
in terms of
'temporal facts',
because they are dissatisfied with a definition of 'temporal facts' as
'such facts as "Smith was angry" and so on'
on account of the specimen fact still involving Time and on account of the expression 'and so on'.
It remains
(a) to say something about the insatiableness of the request for an account of the notion 'and so on';
(b) to show that there is not some definition not in terms of temporal facts which would satisfy those hitherto unsatisfied.
But these points will reappear in other examples and may be ignored for the present.
3.22.
Category puzzles:
(b)
Nations.
Both Professor Broad and Dr. Ewing' express themselves as in doubt whether statements about nations can be analysed into statements about individuals.
But about what are they in doubt:
We know that there is no statement of the form
'Every Englishman . . .'
which means just the same as
'England has declared war'.
We know also that if asked by a child
'What does "England has declared war"
[78]
mean?'
we should say
'Well, a great many people in England are hurrying to make shells and tanks and guns and gas, and a message has been sent by the man who manages these things for Englishmen to the man who does the same for the people they are going to fight, to say that now they are going to fight and so on'.
And we should say that the child would after a while know what 'England has declared war' means.
Or
we might proceed thus:
'To say there is a nation in Arabia is to say that there are a number of people there nationally related'
and this means
'that there are a great many people who perhaps all have the same king or queen or have all decided that a certain two or three people shall rule over them, though of course this isn't necessary, they may have no governors but just be alike in . . .'
We are reminded of a doctor trying to explain what paranoia is.
We say 'he has no definition'.
And this is what many people say about our efforts on behalf of the child.
They say
" This is no definition;
you have not given an analysis of what one says in, for example, "There is a nation in Arabia which never remains long in one place".
You may have explained what you mean
but you haven't defined it'.
Now
some say our explanation
is no definition
because
as one might say
it is never completed
unless you call adding 'and so on' completing a definition.
Others might say
'Even if you could complete your infinite stories
you would not have provided a statement which means the same
as, for example, the simple statement that England is at war'.
Some might be dissatisfied with what we have done on the ground that we have not gone far enough and told them what individuals are, so that we have not explained the nature of our 'specimen' facts.
We may ignore these last people.
They will be satisfied no doubt when we reach the philosopher's paradise where a man speaks only of his own sensations and only at the time he is having them.
Here at last generality, uncertainty, and reduplication have disappeared,
for here are found the perfect proper names, the incorrigible knowledge and the ultimate categories that we have sought so long.
But we must concentrate our attention
(a) on those who say that statements about nations cannot be translated without loss of meaning into statements about individuals,
though the tests
[79]
for the truth of statements about nations may be the same as the tests for the truth of the long stories about individuals;
(b) on those who allow that nation-statements may mean the same as individual-statements ending 'and so on', but complain that this is not a definition because it is not complete —
that what they wanted has not been done because we have not completed a definition of what distinguishes those statements about individuals which are equivalent to statements about nations from those which are not.
It is the same when
if asked
'What is a chair?'
we reply
'A chair is a set of sensations chairishly related',
or
'To say there is a chair here is to make a certain sort of statement about sensations'.
People say
'What sort?'
It then appears that no statement about sensations ever entails a statement to the effect that a chair is present,
or, if you like, that
only at the point of infinity
does one reach
a statement in terms of sensations
which entails a statement about a chair.
And others object that even if one could complete the endless story (in other words apart from the objectionableness of 'and so on'), they would reject the definition.
They say:
'A statement that a chair is here is categorical,
and can never be equivalent to statements about what we should smell if we sniffed, what we should hear if we listened, etc.
And surely an arch-deceiver is logically possible'.
There is just no doubt that
no more extensive knowledge of English or more ingenious use of the knowledge we have
would ever enable us to produce something
in terms of English expressions at present in use,
which would be called by these people a definition of statements about chairs in terms of sensations.
And some will remain dissatisfied even if we invent expressions such as 'chairishly' and explain the analogy with Moore's invention of 'temporal fact.
It may be said
'You have not shown that those who allow that "There is a nation in Arabia" means the same as "There are individuals in Arabia who are nationally related" are insatiable.
For if you provided a definition of "nationally" they would be satisfied.
And the people who allow, as some would, that "There is a chair in the room" means "There are chairish sensations among our roomish sensations" would be satisfied if a definition of "chairish" could be provided'.
But this request for a definition
of what it is for individuals to
[80]
be nationally related
as opposed to non-nationally related
comes from another great class of unsatisfiable requests for analysis.
The only definition in terms of individual statements which will mean the same as 'nationally' will involve the expression 'and so on'.
The only possible definition of 'chairishly related' will also involve the expression 'and so on';
it will involve an infinite disjunction of conjunctions of statements about sensations.
And it will be complained that it is just this infinity which is mysterious.
It is true that
if a person is satisfied about the category of chairs (or nations)
then all that now puzzles him is the nature of the distinction between chairish groupings of sensations on the one hand, and non-chairish groupings on the other (between national groupings of individuals and non-national).
But about this he may well be insatiable.
3.23.
Differentia puzzles.
Let us look at other examples of this sort of request which arises
not from trouble about the form of a proposition
nor from trouble about the category of its subject
but from trouble about the differentia of its subject.
Googs:
... particularly in Aristotelian logic, differentia ... refers to the specific characteristic or attribute that distinguishes a particular species from other species within the same broader genus, ...
...
- Part of a Definition: A definition typically consists of a genus (the broader category, e.g., "animal") and a differentia (the distinguishing feature, e.g., "rational").
Sounds a lot like Rosch's superordinate semantic categories and basic level sematic categories.
Allowing that Henry is an individual, what makes him Henry and not Albert?
Allowing that a cow is an animal, what is the nature of the distinction between a cow and a buffalo?
It is no doubt, in the latter case,
the possession of a common property;
but what is this property:
People say,
'We cannot define what it is to be a cow'.
They often have at first the idea that what they are saying is that 'cow' is not definable in the way that 'brother' is.¹
Now
it happens that 'cow' is definable in this way
though one is apt to overlook it.
For cows are female kine.
But this definition is at once rejected by one who asks metaphysically
'What is a cow as opposed to a buffalo?'
He says
'Ah! yes of course that sort of definition. But
what is it to be kine?
What is it to be female:
Must animals if they are to be kine have horns?
Must they chew the cud?
Must they avoid canine ears?
But none of this is essential.
Is it then that they must either have the right horns, or chew the cud, or have the right ears?
But this disjunctive character hardly suffices to constitute an animal a cow.'
¹ Wittgenstein has given me the impression that he thinks that our trouble here consists in wrongly fancying that all words are definable like 'brother'.
But in the sense here involved 'brother' is also indefinable.
What is true is that the trouble arises from our fancying that the only reductive explanation is through
conjunctive definition of which the definition of 'brother' is a model.
[81]
At once we find ourselves
in the situation of a man
who wrote:
'We can say that horses are mammals, move on legs not wheels, are bred not built, and so on.
But are these qualities essential?
There is no selection of them which enables us to construct that "elusive quality of horsiness which we can detect only by acquaintance with the animal and watching it perform its functions"'.¹
This elusive quality becomes for Locke an unknown essence.
He says
'He that thinks he can distinguish sheep and goats by their real essences, that are unknown to him, may be pleased to try his skill in those species called cassiowary and querechinchio.'
We can easily imagine people saying that we know of its presence by the intellect or infer it from the qualities we detect by the senses.
Heaven knows how we justify the inference since
we have never observed the essence in conjunction with the signs of its presence.
Notice how a transcendental ontology is associated with scepticism, intuitionism and problematic inference of a very peculiar kind.
Gil-White wrote:
Surface characteristics are responsible for typicality effects but not category inclusion, for we treat the former as the typical manifestations or consequences of the essence rather than direct evidence of essence possession. What this implies is that the categories are not fuzzy even if our appearance-based guesses may sometimes be. For species categories, Keil (1989) has demonstrated that young children faced with gradual pictorial transformations of one species into another (e.g., lion into tiger) are very reluctant to designate an animal as simultaneously belonging to both. They will consider the animal to have been transformed, but this happens abruptly, for the children “saw the animals as fully changing in kind across the critical pair of adjacent pictures”. This suggests that even fuzzy transformations cannot overwhelm a strong intuition that the categories are not fuzzy.
(p. 522)
And earlier:
though people may make fuzzy guesses when assigning people to ethnic categories (if they have imperfect background information about them), this is fuzziness and ambiguity in identification procedures. It does not imply ambiguity in concept representation (category structure) if the fuzziness disappears when people are given the missing information. Claims by constructivists that ethnic categories are fuzzy (ambiguous, malleable, etc.) may result from confusing identification with representation ...
(p. 520)
Right on cue, our present-day constructivists also can be found kicking and screaming about the undue scepticism and problematic inference which seems (to them) to attach to any and all burblings-up of un-ambiguous category structure. Hopefully we can put all of that to bed by minding this crucial distinction between identification and representation. Then, we can remain duly skeptical about any suggestion that subjects must necessarily converge upon the same representations unless and until such people are given the missing information; and then, we can really wallow in our skepticism when we realize that "giving information" to, say, people of 80, 100 and 120 IQ, cannot be sufficient to engender agreement in any but the simplest conceptual matters; finally, we arrive at full-on despondency when it turns out that each one of these three "concepts", thusly formed, is crystal clear to the one who holds it, yet when we attempt to 'sum' them things get very fuzzy very fast.
Why does all of this make for skepticism? I think it is a due skepticism of the contingencies we face here and now; but also that these are contingencies, and they are largely the handiwork of our great Boomer "constructivists." (Please read this brilliant Substack, brilliantly entitled Boomerdämmerung.) Things indeed could (and should) be different; but things cannot be just any way at all. If the fruit of philosophical and scientific inquiry makes only for skepticism, then something else has gone wrong, and it must be something 'contingent; and so by then it should, on precisely this basis, be easy enough to say just what it is and what to do about it.
Anyway, floating somewhere in this eruption of blockquotes and vitriol is, I am certain, the balm for Prof. Wisdom's latest discomfort. I don't think I know any transcendental ontologists, but I do think I know how to conjure their specter. If that's not what we want, then we have to perform the opposite conjuring trick.
Now
what is it that is wanted
and is such that
the failure to find it
(1) is represented as ignorance
and
(2) leads to these entrancing tales about elusive essences?
What is the essence of electricity?
Do we not really know what it is for a wire to be electrically charged?
As usual,
it is possible to offer
a 'contingent copy' of what is wanted.
If 'cow' like 'skewbald' and 'spotted' could be defined as a conjunction or disjunction of sensory adjectives without the use of 'and so on', this would be what is wanted.
Now this is an understandable ideal provided it is not required that the words should function as they now do, i.e. stand for the qualities they now stand for.
But, of course, this is required.
And
as the use of the word 'cow' is not related to the use of the sensory adjectives on this simple conjunction-disjunction model
it is self-contradictory to ask whether, keeping the use of language unchanged, we can set out 'x is a cow' as a conjunctive-disjunctive function of a finite number of statements attributing sensory adjectives.
It does not follow that it stands for another character to be inferred
¹ Notice that uneasiness about the doctrine that acquaintance with the animal enables us to detect
a quality in addition to its colour, shape, etc.,
produces a half return to a positivistic doctrine
that horsiness consists in these and the functions of the animal.
If only this half return had been pursued. . . .
² ...
[82]
from the sense characters, as 'has an enlarged liver' is to be inferred from 'looks yellow etc.'
It is related to the use of them in its own way, a way with which we are perfectly familiar.
There are sense characters which are characteristic of cows though they are not necessary conditions of a thing's being a cow, and others which are not.
And we can go on explaining as long as we like which characters are characteristic and which are not.
Prof. Broad, many years ago, suggested a model which throws light on this.
Someone said,
'I know perfectly well what "horse" means
but I cannot analyse what it means'.
Broad said that it seemed to him that such words as 'horse', 'house', 'cow', are not analysable in the ordinary sense.
He said that in the case of such words
there are n characters such that if any m of these are present
we say that a thing is a horse, a cow, etc., as the case may be.
He said this with a view to helping those who said in a puzzled way,
'I know well what these familiar words mean. How then is it that I cannot give their analysis?'
Broad explained that
a thing's being properly called a cow, did not determine uniquely its having this or that character
but determined it with such and such a degree of freedom.
Of course
this again
is only a model.
There is no finite list of sensory statements related to 'That is a cow', even on these more complicated lines.
And it is just this infinity as usual which troubles people.
If a child asks us,
'What is a zebra?'
we explain that
it is rather like a donkey, has long ears and a mane which stands up, is rather less angular than a donkey and thus somewhat resembles a pony.
Must it have long ears?
Well no, but they usually have.
They have, of course, hoofs like a donkey and they are nearly always striped.
They eat grass, breed and so on, like donkeys.
Similarly
if a child asks us
what a tallboy is,
or a chair,
or what it is for people to form a nation.
In all these cases
we can
explain.
We finish always with the expression 'and so on' to cope with the infinite.
If it is said that this prevents our explanation being a definition, then we cannot find one
because there is nothing which the inquirer
[83]
would call a definition no matter how hard we were to work with the words we have or how many new ones we were to invent.
Here then
is another large class
of metaphysical requests for analysis
which like the others we have considered are distinguished from ordinary requests for analysis by the fact that they are requests that a certain class of statements should be exhibited as
finite conjunctive-disjunctive functions of another 'fundamental' class,
when this is impossible because the class of statements in question happens not to be
a finite conjunctive-disjunctive function of those which are taken as fundamental.
Some would allow that they are infinite functions of the fundamental statements;
others would insist that not even the infinite functions mean the same as the statements to be analysed.
3.24.
Do the transcendental theories
provide analyses?
It may now be objected
'You have reached the conclusion that if the philosopher is asking for an analysis then he is asking for the impossible, only because you have considered only the wrong analyses.
You have considered analyses of general propositions into specific propositions, analyses of super-individual things such as Time, Space and nations into individual things, and of individual things into sensations.
The right answer to the question "What are numerical propositions?" is that
they predicate a peculiar kind of trans-individual quality applicable only to groups,
or that they are analysable into general propositions together with the peculiar relation other than.
General propositions are about a variable entity, neither Smith nor Jones, neither this nor that.
Propositions about nations are about an entity which is not a person nor to be reduced to persons though it is in many ways like the persons which belong to it, and its nature is known to us only through them.
And Space and Time and a work of art, such as Hamlet, are again about entities which are "neither physical things nor minds nor sensa, neither particular existents nor abstract universals, neither substances nor mere qualities, relations, or states of substances, and therefore elude orthodox philosophical classifications.
These things cannot be said to exist in the sense in which particular things and persons can be said to do
and yet are essentially particular and not abstract universals"'.
[84]
Propositions about
numbers,
universals, or
propositions
are about what is objective, real and possible, but non-sensible,
neither subjective nor palpable and spatial.
These objects do not exist
but they subsist.
These theories are perfectly correct.
They could not be otherwise
since they proceed on the lines:
The peculiarity of facts about classes, numbers, propositions and characteristics is that they are about subsistent entities.
What are subsistent entities?
Well,
they are not chairs and tables.
Indeed
they are not known by the senses
but by the intellect.
Characteristics, for example, are subsistent entities, so are propositions, numbers and classes.
But the idea that there are questions
to which these theories and
the positivistic or reductive theories
are alternative answers
arises from a confusion.
The question
'What are X propositions?'
may mean
'What is the analysis of propositions with regard to propositions that they are X propositions,
e.g. "Something moved is a general proposition"';
and it may mean
'What is the analysis of propositions which are X propositions, e.g. "Something moved".'
Now these theories provide answers of a sort to the first question but not to the second.
If I ask
'What are propositions about instincts?'
and someone says
'They are propositions about dispositions which, etc.'
this is a definition which brings that class of propositions under a genus and mentions a differentia.
But no one supposes that this enables me to translate sentences expressing propositions about instincts into sentences which are not propositions about instincts.
Hence anyone who, wishing to do that, asks
'What is the analysis of propositions about instincts?'
will be entirely dissatisfied and will say
'But what is a disposition?'
Similarly, the people who ask
'What are numerical, fictional, general, national, material, propositions?'
in that sense of these questions which we have been considering, will be dissatisfied if offered one of these definitions in terms of a transcendental constituent.
He will at once re-express his questions in the forms,
'What is a variable entity?'
'What is a subsistent entity? and how do the numerical ones differ from others?'
'What is a particular universal?'
Just as when I asked a plain man,
'What is the proposition 2 + 2 = 4 about?'
and received the answer
'Numbers',
I asked at once
'What are
[85]
numbers'
and received the answer
'Oh, I don't know, I can't answer that, 2 and 3 and 7 are numbers'.
The transcendentalist likewise when asked
'What are these entities?'
replies
'They are ultimate. Such things as nations, institutions, poems, the Exchequer are particular universals. Such things as characteristics, classes, propositions are subsistent entities'.
That is, the transcendentalist agrees with the conclusion we have reached about our question, namely the conclusion that when the profound metaphysician asks,
'What is the analysis of X propositions?'
then his request is a request for the analysis of the unanalysable.
He agrees with us, too, that merely to say this is inadequate.
Something must be done to explain why we should so persistently return to our vomit, the positivistic theories;
and the explanation he offers is that the entities which the ultimate but 'difficult' propositions are about, are of a sort we have hardly recognized (Russell), because they are not detected by the five senses and consequently they 'elude orthodox philosophical classifications' (Ewing).
How this reminds us of the fictional entities upon which Russell poured so much scorn.
But at the moment we must emphasize our agreement with the negative aspect of the transcendentalist's conclusion,
namely, that the metaphysician's requests for the analysis of X propositions into non-X propositions is a request for what is impossible.
And now
the inadequacy of this conclusion
is still more apparent
because it is apparent that
the transcendentalist theories are common-sense answers dressed up.
The plain man, when asked what the proposition 2 + 2 = 4 is about, and how we know it, replies
'It is about numbers and we know it in a special way, and "Red is a colour" is about colours'.
The transcendentalist replies:
'Both propositions are about subsistent entities and we know them by intuition'.
The plain man, when asked
'What is "There is a chair in this room" about? and how do we know it?'
replies
'It is about a chair, of course, an article of furniture, and we know it by looking, by the evidence of our senses'.
The transcendentalist replies
'It is about a material thing which is not to be identified with our sensations and we know it by direct but
[86]
not immediate knowledge, or "in and through" our sensations,
but the knowing of it is not to be identified with sensing, it is something more than this,
and what is known by this process is something more than mere sensing reveals,
or perhaps we infer the chair from what we know by our senses'.
The plain man,
wishing to encourage us
to believe in the soul,
says
'You believe in the wind don't you, though you can't see it? Well, it is like that with the soul'.
What the phenomenalist resents is the suggestion that it is not by our eyes and ears that we know of the existence of the wind, when there is one.
This leads him to say
'The wind is nothing but slates off the roof and a smell from the gas-works.
You speak like the Irish coachman who whispers to children that goats can see the wind'.
But of course the wind isn't just that.
Or is it just slates off the roof and/or a smell from the gas-works and/or a rustle in the leaves, the hurrying of the clouds and so on?
Is it or isn't it?
Ordinary language
suggests analogies
which puzzle us.
From the rapidity of Smith's pulse I infer that the malaria germ is in his blood,
from the fall of the slates that Boreas is angry,
from the powerful smell that there is a cheese near which is going bad.
Transcendentalist language emphasizes and (usefully) collects these suggested analogies and thus increases these puzzles and distils them into more concentrated forms.
Positivism suggests a way of removing them.
The answer
'The classes of proposition which puzzle us are ultimate, unanalysable'
insists that the positivist's way out will not do.
But it leaves us where we were, except in the important respect that we can no longer imagine that definition will help us out.
We are left giving the answer
'"Smith is good" is about Smith and goodness, about an ego and a value predicate; it is not a statement about a body and how it behaves'.
A realist philosopher might say
'You ask
"Do we know the past and how, since it no longer exists to stand in any relation, cognitive or otherwise, to anything?"
the answer is that we do know it, by memory'.
But we do not need a philosopher to give us this answer.
Again, after reading Zeno's arguments to prove that we cannot reach the end of a racecourse we do not need a philosopher to show
[87]
us that we can and how.
A mere 'plater' can do both quicker.
But we know that there is another sense in which only a skilful philosopher can show us that we can and how.
He will show us what is wrong with Zeno's puzzling argument — give us what we want.
We have seen, however,
that in the case of
the profound metaphysical questions
we have looked at,
definition is not what is wanted.
They are of the form
'What are X propositions?'
or sometimes
'What is the analysis of X propositions?'
or
'Surely X propositions are not ultimate, what is their analysis?'
We have found that in these cases 'analysis of X propositions into non-X propositions' describes nothing,
so that to say that what we want here is definition of X propositions is to say nothing.
On the other hand to say that what we want is this conclusion that analysis is impossible, takes no account of our use of the phrase 'Surely X propositions are not ultimate' and leaves us still unsatisfied.
Nor is this dissatisfaction removed when it is explained how easy it is to pass from sensible requests for analysis to nonsensical ones because of ambiguities in 'X proposition' and 'mean'.
For we feel that even when these confusions have been removed and we are clear that we are asking with regard to an ultimate class of propositions, 'What are these propositions?'
still we are are asking for something.
But what?
4.
The metaphysician may be described as
seeking descriptions of a certain sort.
But of what sort?
and how do they help him?
The answer may be put in three ways:
(1) We might say
that we are asking for analyses
in a wider sense of 'analysis',
a sense which does not require that
if p₁, is analysable into p₂, then the sentence 'p₁' and the sentence 'p₂' mean the same.
This is Mr. Duncan-Jones's suggestion.
It has two important merits:
(i) It emphasizes the continuity of metaphysics with analysis.
(ii) It hints that our trouble lies in the fact that
we sometimes are more inclined to say that two sentences express the same proposition
(stand for the same fact)
than we are to say that they mean the same.
It will, however,
be objected
that this altering of the use of words
does not meet the case.
The question was, e.g.
'Can
[88]
general propositions be analysed into non-general propositions?'
and the trouble was that we were inclined both to say Yes and to say No.
Now it is no use saying 'In a new sense "yes", in the old sense "no"',
because the question was
'In the old usual sense of analysis, can general propositions be analysed into non-general propositions?'
We have shown that they cannot be unless a definition with 'and so on' in it is counted as a definition.
But is this a definition in the old sense or isn't it?
The answer
'It would be in a new sense'
is no more use here than it is when I ask
'Is a cassiowary a goat?'
or
'Did Smith exercise reasonable care or didn't he?'
You might as well say when asked whether a dog has gone round a cow who keeps her horns always towards him
'He has in a sense'.
(2) We might say:
'In a new sense of "meaning"
the metaphysician asks,
"Do general sentences mean the same as some collection of non-general sentences?
Does the statement 'There is a nation in Arabia' mean 'There are a number of individuals in Arabia who are nationally related'?"
He is confused by the old sense of meaning and so does not realize that the answer to his question is emphatically Yes'.
But again there will be the objection that the question asked was in terms of the old use of meaning.
We come on cases where it is difficult to tell whether two sentences mean the same, express the same proposition, stand for the same fact or not.
'Now', it will be said, 'there is no use in saying that there could easily be a sense in which the answer would certainly be "Yes" or a sense in which the answer would certainly be "No".
For what we want to know is whether in the usual sense of "mean the same" they do or do not mean the same.
And when people allow that
"There is a penny on the table"
means the same as
"If you look you will see a brownish circular or elliptical patch, if you . . . and so on"
but inquire whether this is a definition,
then it is no use saying
"There could easily be a sense of definition in which it would be a definition".
And this answer is also no use in dealing with disputes between philosophers as opposed to disputes within the breast of a single philosopher'.
We shall see later how far there is use in saying this and how far there is not.
(3) We might say:
'What the metaphysician asks for
when he asks
"What are X propositions?"
is not a definition but a
[89]
description,
and when he asks whether X propositions can be analysed into non-X propositions he is asking for a description of the relations between them.
He is apt to think that he is asking for a definition because in many cases to give a definition of X propositions has been the best way of giving a description.
When the philosopher says analytic propositions are a sort of verbal proposition, this must not be taken as a definition but as a description throwing light upon the nature of analytic propositions and their relations to verbal propositions.
By this means he describes a man asserting an analytic proposition just as one may describe, to a Red Indian, a man riding a motor cycle as a man riding a horse on wheels.
'And similarly
to say that a metaphysical statement
is a sort of analytic statement
but not quite
because it is a sort of description,
is to give a description of a metaphysical statement and of a metaphysician'.
Two objections may be made.
First, it will be said:
'The metaphysician asks
"Are general propositions to be identified with some combination of non-general propositions or not?"
"Are analytic propositions to be identified with verbal propositions or not?
Can the one class be defined in terms of the other, or not?"
You say
"They are not to be so identified with each other but the one can be described by reference to the other as rather like the other".
This is like the man who when asked whether a llama is a goat or not, replies,
"No, but a llama may be described as a sort of goat"'.
We can recognize in these objections the objection made to the first two accounts of what it is the metaphysician really wants.
There is a second complaint
which may be raised
against the first and second accounts
of what the metaphysician really wants,
namely,
that they do not give an analysis (new sense) of the new sense of 'analysis', or the meaning (new sense) of the new sense of 'meaning',
and that therefore they do not analyse (new sense) or give the meaning of (new sense) 'metaphysics'.
Similarly it may be objected that our description of a metaphysician is inadequate.
It may be said
'You say the metaphysician describes ultimate classes of proposition, or describes the use of ultimate classes of sentence.
But even if this is accepted you have not defined the metaphysician;
for not all descriptions
[90]
of ultimate classes of propositions or sentences are metaphysical.
Now which sort are?
Until you have told us this, you have not told us what metaphysics is'.
We are,
by now,
familiar with
this sort of complaint.
It is necessary to repeat that
the request
'What is a metaphysician?'
is not a request for definition.
We can explain what the felidæ are.
We do this
'internally' by explaining that leopards, lions, lynxes and cats are felidæ,
and
'externally' by explaining that dogs, monkeys and even Siamese cats are not.
We can explain what a pony is by giving a false definition, 'A horse under 14 hands high', and then explaining how this is not quite correct.
Likewise we can explain what a metaphysician is 'internally' by explaining that positivists and 'ultimatists' are metaphysicians, and 'externally' by explaining that grammarians, logicians and poets are not.
Here again we may in each case begin by a false definition
'Metaphysics is grammar',
'Metaphysics is logic',
'Metaphysics is poetry',
and then explain why this is not so, drawing attention to such statements as
'Life is a tale told by an idiot',
'The world was made by God',
'Thou canst not stir a flower without troubling of a star',
'All relations are internal'.
However,
it may still be complained
that the definition is
inadequate
in that the descriptions the metaphysicians give are connected in some way which has not been brought out by showing that metaphysics is not science, not logic, not grammar, not poetry.
We may say,
remember the 'internal' part
of our description.
The metaphysician is concerned with those descriptions of ultimate classes of fact which bear on the great groups of puzzles which we have considered.
It may now be said,
'Yes, but
how do the descriptions bear on the puzzles?
Bring out now the nature of the puzzles in such a way that it is clear how the descriptions bear upon them'.
It is now clear
that if we go on
to make our description more adequate
we shall be dealing also with the first complaint, namely
How does the answer
'General propositions may be described as a sort of infinite disjunction of specific propositions'
bear upon the questions,
'Can general propositions be identified with specific propositions?'
'Are the facts which correspond to them
[91]
identical?'
'Do the sentences which express them mean the same'?
The answer put briefly is this:
(a) The metaphysical questions,
'Are X propositions to be identified with Y propositions?'
'Do X sentences mean the same as Y sentences?'
'Do they stand for the same facts?'
arise and present difficulty not only because the expressions involved are ambiguous but also because they are 'vague',
especially the expressions
'mean the same',
'express the same propositions',
'state the same fact'.
It is because of this that the descriptions are relevant.
How the descriptions are relevant is obscure because the nature of the questions is misunderstood in a way in which the nature of questions which present difficulty because of the vagueness of their terms is constantly misunderstood.
(b)
The case is further complicated
by the fact that
the use of the expressions
'have the same thing in mind when I say S as when I say S’',
'mean the same by S as by S’',
'express the same proposition by S as by S’',
'state the same fact by S as by S’',
are connected in a way which forms a slide irresistible by any logician, and we are all somewhat afflicted that way.
For there are not cases in which one would unhesitatingly say one of these and deny another.
At the same time, there are cases where one hesitates more about one than about another.
5.
Let us look again
at 'primitive' metaphysical difficulties.
In order to see all this, let us look once more at metaphysical difficulties arising,
and let us look at them arising in people before they have been influenced by having the nature of their difficulties 'explained' to them.
I remember someone saying to me,
'You ask me to write an essay about Negation.
But what is the trouble about Negation?'
I may confess that I was somewhat at a loss.
After all, what could be simpler:
'It is Jones',
'It is not',
'Her coat was blue',
'It was not blue'.
What is the matter with the negatives?
Why prefer the positives?
Now
this case presents with simplicity
what we have seen so often.
First, an attempt is made to deal with the difficulty by definition in terms of expressions, new or old, which certainly mean the same as the usual sentences for expressing negative propositions.
Thus Ramsey suggests expressing negation by writing positive sentences upside down.
This throws a light,
[91]
but leaves a more refined question, leaves the shadow unreduced to the substance (W).
Then we try saying
'"This is not a tiger" means "This is a leopard or a lion or a giraffe or a donkey"'.
Then for accuracy we add 'and so on'.
Then it is said perhaps that this is just a negative sentence in disguise, because the whole infinite force of the negative lies in the 'and so on'.
Or perhaps it is said that this does not mean the same as the original negative sentence.
The failure of the positivistic answers
leads to the conclusion
that negative facts are ultimate,
that we are tempted to identify them with positive facts because the special sort of element they contain is not detectable by the senses.
But this again does not satisfy.
Now
what desire
produces these abortive efforts?
It is this.
No one minds admitting that there are negative sentences as well as positive ones,
but everyone feels uneasy when asked whether there are negative facts as well as positive ones.
Yet negative sentences have meaning, express propositions which are true.
If these propositions are not identical with propositions expressed by positive sentences, then surely what makes them true must be not identical with what makes true the propositions expressed by positive sentences.
In general:
The metaphysically-minded person
feels that
the actual world is made up solely of positive, specific, determinate, concrete, contingent, individual, sensory facts,
and that the appearance of a penumbra of fictional, negative, general, indeterminate, abstract, necessary, super-individual, physical facts is somehow only an appearance due to a lack of penetration upon our part.
And he feels that there are not, in addition to the ways of knowing the non-penumbral facts, additional ways of knowing employed for ascertaining the penumbral facts.
At the same time the penumbral do not seem to be identical with the non-penumbral and thus do seem to call for extra ways of knowing.
Now
this feeling
of
taking the same reality twice over (McTaggart),
this feeling of superfluous entities (Russell),.
[93]
this feeling of metaphysical double vision has been removed in certain cases by definition.
We can imagine someone saying
'What is the average man?
How do I know he exists?'
and then both troubles being dissolved by the definition
'"The average man is 5 ft. 4 ins. high" means "The sum of the heights of individual men divided by their number is ft. 4 ins."'
We can imagine someone saying
"All vegetarians are temperate" does not mean the same as "No vegetarians are intemperate"
and then having to say that we know that the one fact is always present when the other is because the propositions are connected by a necessary connection which we know by intuition.
We can even imagine someone saying
'"Geoffrey is George's brother" does not mean the same as "Geoffrey is a man and his parents are George's"'¹
and then again having to talk of two facts and two propositions connected by a necessary relation.
Or again we can imagine someone wondering how we know that when (1) there are two white goats and four black in a field then (2) there are six goats in the field or there exists a class goats-in-that-field which has six members.
Definition removes these troubles.
Take the last case.
When we saw that the two sentences meant the same or that the meaning of the one included the meaning of the other, then the appearance of plurality was explained by the plurality of sentences, while the assurance of identity was justified in the single meaning made true by a single fact.
And with the disappearance of the ontological puzzle
the epistemological puzzle vanished also.
No wonder the definition model fascinated.
But unfortunately,
as we have seen,
there are cases² where definitions cannot be found,
where no ingenuity reveals non-penumbral sentences which we can feel sure mean the same as the penumbrals.
And yet we cannot feel that the facts which make the penumbral true are anything but the positive, concrete, etc., non-penumbral facts which make up the actual world.
Indeed there are cases where we know that there are no non-penumbral sentences which mean the same in the ordinary use of 'mean the same' as the penumbral,
while yet some of us feel that there is no difference between the facts which make
¹ People sometimes say 'No two expressions really mean the same'.
² But for them, metaphysics would not exist.
[94]
those penumbral sentences true and those which make the non-penumbral sentences true.
Some people come down on one side of the fence, some on the other.
Thus Broad argues from the fact that the question
'Fido behaves in all respects intelligently, but is he intelligent?'
is not silly like
'Smith is rich, but is he wealthy?'
to the conclusion that the question is synthetic like
'I have given Smith two ounces of arsenic, but will it kill him?'
That is, he concludes that the sentences
'Fido behaves in all respects intelligently'
and
'Fido is intelligent'
do not mean the same, do not stand for the same proposition, do not stand for the same fact.
To do this is to represent the question
'Smith still breathes, and he nods, smiles, and talks as usual, but does he really think and feel?'
as like the question
'Smoke still comes from the chimneys, the lights go on in the evenings, but have the inhabitants fled?'
Has Smith's soul left his body but arranged with the nervous system that appearances shall be kept up in its absence?
Has his rha flown?
And yet we feel that the question is not like this — yet surely it must be, for Broad's premisses are true and there is no logical slip.
6.
Let us reformulate
the difficulties.
These difficulties arise from a misunderstanding of the nature of the questions,
'Are X facts ultimate?'
'Are X facts reducible to Y facts?'
Let us try reformulating them.
Instead of asking
'Are X facts reducible
to non-X facts
or to Y facts?'
let us ask
'Are X propositions reducible to non-X propositions or to Y propositions?'
And instead of asking this let us ask
'Do X sentences mean the same as any combination of non-X sentences or of Y sentences?'
And instead of putting this back into
'Do X sentences stand for the same proposition as any combination of non-X sentences or Y sentences?'
let us ask instead
'Are X sentences used in the same way as some combination of non-X sentences or of Y sentences?'
i.e.
'When we have an X sentence can we find a Y sentence which serves the same purpose?'
This reformulation
is permissible of course,
only
when we remember that it means
'When we have an X sentence used in
[95]
that way with which you are familiar, is there some combination of Y sentences used in that manner with which you are familiar, which serves the same purpose?'
We have then
the questions,
'When we have a general sentence can we find a combination of non-general sentences which serve the purpose it serves?'
and
'When we have a nation-sentence, can we find a combination of individual-sentences which serve the purpose it serves?'
and so on.
Compare these questions
with the questions
'Does the paper pound serve the purpose which the sovereign served before the war?'
'Could anything which is not a bridle serve the purpose which a bridle serves?
The Red Indian with his single strip of hide can stop and turn his pony with it, so the practical, debunking person will say 'Yes', may even insist that it is a bridle, which of course is quite untrue.
The instructor in equitation will say 'No' because he cannot obtain with it the balance and 'collection' he obtains with the bridle.
Others hesitate, wondering which is right.
But what sort of wondering is this, and what sort of right and wrong?
We see at once that we have only to put the question in the plural,
'Which of the purposes served by a bridle does the strip of hide serve?'
to see that the answer is a matter of describing these,
and if after this it is asked
'Well, does it serve the same purpose?'
the question is now obviously
'Shall we say
"It serves the same purpose"
or shall we not?'.
We can now see
how to deal with
the sovereign and the paper pound.
Those who say the value of a thing is its power in exchange will say the paper pound does serve the same purpose.
Others will say, 'No there was something about a sovereign which a dirty bit of paper can never do'.
And now
let us take
another case.
People ask,
'If when a dog attacks her, a cow keeps her horns always towards him, so that she rotates as fast as he revolves, does he go round her?'
We may imagine them offering reasons
'He does go round her, because he goes all round the place where she is standing, that is, he encircles her. Therefore he goes round her'.
'But', it may be protested, 'he never gets behind her, therefore he doesn't go round her'.
[96]
Suppose the disputants
now appeal to you
and ask
'Which is right?'
or
'Which do you think is right?'
There is at once an inclination to answer,
'There is a sense in which he does and a sense in which he does not go round the cow'.
But this is untrue.
There are not in English two uses of 'go round' in one of which the answer is 'Yes' while in the other it is 'No'.
Had there been, the question would hardly have produced difficulty.¹
But the answer is not useless because it brings out
how easily there might have been a use of language in which we should have had an answer ready,
and thus hints that the question is a matter of language.
It is, however, very necessary to explain what sort of a question about language it is.
This appears
if we set out
the right way to deal with the question.
One should say:
You speak as if you are asking a question about the dog and the cow.
But you know the facts about them.
And what is more
you know the answer to the question,
'What would ordinarily be said in such a case when the question is put "Did the dog go round the cow?"'
For you know that people would hesitate and
some insist, though with a certain bravado, that he did
and others that he did not.
In asking me this question
you are treating me like a judge of the High Court who is considering
a question of law
not of fact,
e.g.
'Was it in the case described reasonable and probable that someone would try to cash the cheque which Mr. Smith made out so carelessly?
Was it?
or was it not?'
Now
I can of course give a decision if that is what you want.
But you want more than that.
You want me to sum up and bring out the features of the case which incline one to say that the dog went round the cow, and those which disincline one to say this.
In this particular case, unlike other cases of the sort, this can be done fairly easily.
For the features which incline one to say that the dog went round the cow are summed up in the statement
'He circled round the place where she stood',
while those which disincline one to speak so can be summed up in the statement,
'He did not change his position with respect to the parts of the cow'.
Now you will notice that you who wished to say that the dog went round and your opponent who wished to say that it
¹ We might say
'Metaphysical difficulties have been so hard to remove because of the illusion that ambiguity is the only bar to logic'.
[97]
did not, had between you already described these features although at the time you regarded yourselves as giving reasons for its being right or wrong to say that the dog went round the cow.
So you see
you had already done what you wanted done
though,
because you mistook the nature of your question, you did this in a misleading way.
Had you put your question in the form,
'Which of the features which we expect to find in a case when told that A has gone round B do we find in the case of the dog and the cow?'
you would not have found yourself in difficulties.
Let us return to the question:
'Given such a sentence as "There is cheese here" does this mean the same as any string of sentences about what we should see if we looked, smell if we sniffed, etc?
Is there any sensation-sentence which means the same, expresses the same proposition as, stands for the same fact as, the cheese-sentence?'
Here again,
as we have seen,
there is an inclination to say
'There is a sense of "meaning" in which the answer is "Yes" and a sense in which the answer is "No"'.
But there are not two such uses of 'meaning' in English.
If there had been, the difficulty would have been removable in the way it is when someone says of a mare that she is a horse while someone else says she is not.
Here again
the proper answer
is:
'You are not asking anything about cheeses,
you know about them,
what they are made of, and so on.
But what is more,
you are not asking what would be said here nor what the language experts would say here.
No expert in heaven or earth could help you.
What you are asking for is
a decision and the reasons for it
in the sense in which reasons can be offered for a decision — by counsel for the plaintiff and counsel for the defendant.
But these as a matter of fact you have already set out —
the positivists have set out the "reasons" for deciding in favour of saying that the sentences mean the same, the "ultimatists" have set out the reasons against saying this'.
There are
cases of metaphysical dispute,
however,
where the answer
should not be quite as above.
Suppose A says that poetry which does not rhyme is not really poetry,
while B says it is.
Well, hello Art Problem, we've been expecting you.
Now
in so far as the question is one of logic
as to whether being poetry entails rhyme,
B is right,
because
language is in fact so
[98]
used as to make him right.
But A is of course well aware that this is how language is used and is not denying what he knows to be true.
But they think they are contradicting each other, and are thus wrong.
If now A asks,
'Don't you agree with me?'
one can only explain those features of one's attitude to their dispute which incline one to say, 'I agree with A' and those features of their dispute which incline one to say, 'I agree with B'.
(For neither the answer 'I agree with A', nor the answer 'I agree with B' is correct though neither is incorrect.)
Now this explanation will involve an explanation of the nature of the dispute, and this will involve describing
what leads A to say what he does
and what leads B to say what he does.
Similarly in metaphysics.
Broad shows how
as a matter of logic,
'Fido is intelligent'
does not mean the same as
'Fido behaves intelligently'.
But he has not grasped what those who deny this are doing.
And since they advance their thesis as the contradictory of his, it is not to be wondered at that he should not have grasped what they are doing.
On the other hand surely it is apparent that even when they have to allow that the two sentences do not mean the same, they are not going to be convinced that they were wrong.
No proof that taken as a statement in logic what they say is false convinces them.
The younger child who has been insisting against his pedantic elder brother that a fox has a tail may not be convinced that he was wrong even when an adult convinces him that experts and others do not say that a fox has a tail (W).
The situation is often complicated
by the fact that
the following steps seem impeccable:
'What I have in mind when I say S,
e.g. "Smith can do successfully all the tests for colour blindness and never fails in ordinary life either"
is not the same as what I have in mind when I say S’,
e.g. "Smith can see red and green and all the colours".
Hence S and S’ do not mean the same.
Hence they do not stand for the same proposition.
Hence they do not stand for the same fact'.
As we have seen, the argument may be reversed.
Now as a matter of fact, the features of the use of two sentences which incline us to use the expression
'stand for the same fact'
are not quite the same as those which incline us to use
'put the same ideas into our mind'.
Hencе it sometimes happens that we get from certain features of the
[99]
use of two sentences a strong inclination to say that they do not mean the same,
derived from a strong inclination to say that we have not the same things in mind when we utter the one as when we utter the other,
and at the same time a strong inclination to say that they do mean the same,
derived from a strong inclination to say that they do not stand for different facts.
And when this happens we may be induced to say that S and S' stand for different facts when but for faith in this logical ladder we should never have done so;
and we may be induced, by tipping the ladder the other way, to say that two sentences mean the same when we should never otherwise have done so.
It is this ladder or slide
which,
together with a wrong way of speaking about meaning and the assumption that the only purpose of indicative sentences is to state facts, that leads us
to look for fictional facts to fit the fictional propositions required by the significance of fictional sentences.
Here the assumption that
the sole function of indicative sentences is to state facts
reaches the limit of absurdity.
It will be noticed that
this assumption is the Verification Principle.
Having seen its falsehood in the case where the factual functions of an indicative are zero,
we are ready to look for non-factual functions in cases where the factual functions are not zero,
in ethics, in mathematics, in psychological statements, in metaphysics, and thus in the principle itself.
A flood of light is thrown on the numerous philosophical disputes arising from the fact that sentences may agree in factual function while they do not in other functions.
But it must be remembered that this method is far from foolproof because
the hardest problems are those in which there is an inclination both to say that two sentences stand for the samе fact and to deny this,
e.g. psychological and behaviour sentences (worst of all), analytic and verbal sentences, material thing and sensation sentences (less acute).
And such a conflict, as we have seen, can be met only by explaining its nature,
that is by explaining that the dispute is resolved by setting out
what has induced each disputant to say what he has said.
Well, . . .
what about the inextricability platitude?
"X cannot be considered apart from Y."
Most infamously,
"An artwork cannot be considered apart from its social and cultural context."
Suppose we said to these people,
'Why is it, then, that you yourselves cannot even refer to X and Y without "considering them apart" from each other? In declaring what you intend to do, you in fact do the opposite.'
To sum up:
The metaphysician
is concerned with
certain fundamental ontological and epistemological reduplication questions:
Are X facts to be identified with Y facts?
How do we get from knowledge of the latter to knowledge of the former?
[100]
Usually, even as questions of logic, there is no right or wrong answer to these questions.
I should be inclined to say this in every case where the question is in the form in which reduplication is most intolerable, namely,
'Are X facts nothing but Y facts or are they something over and above Y facts?'
Sometimes in the form in which reduplication is less intolerable, namely,
'Do X sentences mean the same as Y sentences'
(taken as a question of logic)
there is a correct answer 'No'.
But
in either case
the metaphysical dispute
is resolved by explaining what induces each disputant to say what he does.
This is done as follows:
First explain the nature of the question or request;
(a) Negatively — remove the wrong idea that it is a question of fact whether natural or logical;
(b) Positively — give the right idea by showing how, as in other disputes of this unanswerable sort, the questions are really requests for a description of
(1) those features of the use of the expressions involved in the questions which incline one to answer 'Yes',
and of
(2) those features of their use which incline one to answer 'No'.
In the case of ontological questions such as
'Are X facts to be identified with Y facts?'
'Do X sentences mean the same as Y sentences?'
'Does the sentence S stand for the same fact as the sentence S' '
the expressions involved are of course
(a) the expressions
'X facts',
'Y facts',
'X sentences',
'Y sentences',
'The sentence S'
and so on,
and
(b) the connectives
'stand for the same fact',
'mean the same'
and so on.
In the case of epistemological questions, expressions such as 'know', 'rational', take the place of the connectives.
Second
provide the descriptions
that are really wanted.
Fortunately
when the nature of the questions
has been explained,
then the nature of the 'answers', 'theories' and 'reasons' which they have been 'offering' and 'advancing' becomes clear to the disputants.
And then it becomes clear how much of the work of providing the descriptions has been already done, though under the disguise of a logical¹ dispute.
Thus the metaphysical paradoxes appear no longer as
crude falsehoods about how language is actually used,
but as penetrating suggestions as to
how it might be used so as to reveal what, by the actual use of language, is hidden.
And metaphysical platitudes appear as
¹ At one time under the disguise of a contingent or natural dispute.
[101]
timely reminders of
what is revealed by the actual use of language and would be hidden by the new.
To take an example which we have ourselves come upon:
Some have said
'Analytic propositions are verbal',
others have said
'They are not',
and, in supporting these 'views', they have between them done all that is primarily asked for by one who asks 'Are analytic propositions verbal or are they not?'
Thus
it appears
how it is that,
to give metaphysicians what they want,
we have to do little more than remove the spectacles through which they look at their own work.
Then they see how those hidden identities and diversities which lead to the 'insoluble' reduction questions about forms, categories and predicates, have already been revealed, though in a hidden way.
Logic and Language
(Second Series)
ed. Antony Flew
(1966)
[51]
chapter iii
UNIVERSALS
By D. F. Pears
'Do universals exist?'
This question was debated so long and vehemently because it was mistaken for
a factual question about some airy realm of being.
But why was this mistake made?
One diagnosis is that
general words were tacitly assimilated to proper names,
and that, when this practice is exposed, it becomes harmless but pointless.
But this is a description of what happened rather than an explanation; it gives something more like a symptom than a cause.
Could so many philosophers have been so silly in such a simple way?
...
'Universals exist'
has
a deceptive logic.
Realists offer it as the conclusion of many arguments:
but
unlike the premisses of these arguments,
it cannot be understood as a verifiable statement of fact.
Hmm. If the premisses can be verified but the conclusion cannot be, is that to say that those premises do not lead inexorably to that conclusion? (To any single conclusion?)
On the other hand,
if it is taken merely as
an esoteric way of stating those premisses over again,
the vehemence of the controversy becomes inexplicable.
Faced with this difficulty of interpretation, some modern philosophers suggest that it is no good puzzling about its literal meaning, just as it is no good puzzling about the literal meaning of dreams.
For traditional philosophy provided
a small set of possible conclusions
to arguments about the generality of thought and language, and tradition was strong.
Well, a small set is still . . . a set, rather than a unity or an identity; and rhetorical chaos grows exponentially, not arithmetically.
If a tribe educated its children to dream according to a tradition which restricted their manifest dream contents within narrow limits, it would be difficult to discover their much more varied latent dream contents.³
Similarly,
although realists are argumentative,
it is difficult to answer the question why they
[52]
maintain that universals exist.
Yes, the footnote (excised) cites Freud.
The point seems to be to anchor a metaphor, not to support an argument. Still, it's a curious example of a philosopher's appeal to empirical matters; doubly so, as the Freudian orthodoxy was to take quite a beating. What becomes of such appeals if the empirical applecart is later overturned?
Now, as for the metaphor: Why would narrow limits of manifest contents stymie interpretation? It's clear that this limits the interpreter's range, but why are we to think that this prevents him from getting at the truth? It may well be the only way to channel him towards it. The steep curve strikes again! Each additional meme "manifests" (n + 1)! - (n)! new memic axes for the interpreter to chew on, but it cannot manifest new truths. If it seems to do this, then the given practice of 'interpretation' is thereby given the lie.
The metaphor to argumentative realism is:
responsible
traditional philosophers
are
'responsible'
to
a mere
small set of possible conclusions
re: the big 'topics'.
But it seems deeply un-Wittgensteinian to suppose that rhetorical range and opacity of reasoning are so tightly bound up with each other.
Any answer must be based on a selection from among the many reasons which they themselves proffer:
and
a good selection will be diagnostic;
it will successfully explain the doctrine.
Well okay, the more there is to select from, the better can the diagnosis be fit to the problem. This, however, is more work, and it involves (probably) excluding a greater proportion of candidate reasons after first considering them. Small Set Pragmatism, then, is simply a way of saving the overwily diagnostician from himself, i.e., by giving him just slightly less rope than he can hang himself by.
There is no sharp boundary here between descriptions of the premisses of philosophical arguments and diagnoses of their conclusions:
because success in explaining, which is the criterion of a diagnosis, is a matter of degree, and because the reasons which philosophers themselves give for their doctrines sometimes completely explain why they held them.
Success in explaining diagnoses of conclusions
can never be complete,
but
reasons given
for doctrines
held
can
complete-ly explain
themselves.
Quine's remark,
that
realists find a universal for every property which can be existentially generalized,
is an extremely brief description.
The thesis of Berkeley and Mill was more than this:
it was a diagnosis, but an inadequate one.
I shall try to provide a less inadequate diagnosis.
And I shall hang on for dear life. It has already been a rough ride.
'Because universals exist'
is the answer to
at least two general questions:
'Why are things what they are'?²
and
'Why are we able to name things as we do'?
Though Plato and Aristotle sometimes distinguished these two questions, it was characteristic of Greek thought to confuse them.
Yet they can be clearly distinguished,
the first requiring a dynamic answer from scientists,
and the second a static answer from logicians.
Now philosophy has often
staked premature claims in the territory of science
by giving quick comprehensive answers to questions which really required laborious detailed answers.
And clearly this is what happened to the first of the two questions.
I'm not sure how this can be avoided, other than by science failing to exhaust its own mandate, i.e. by failing to explain absolutely everything.
If the march of science continues to move in steps halfway towards ultimacy without ever arriving there, then it's inevitable that philosophy will wander into its path without even realizing it, only to be squashed flat without science realizing or feeling anything under its foot.
When detailed causal answers were provided to it, the comprehensive answer 'Because universals exist' was no longer acceptable or necessary.
But what would detailed answers to the second question be like?
Presumably they would be explanations of the meanings of words.
✋
I did not presume this when I read the second question.
But philosophers are easily led to neglect such detailed progressive answers to the second question,
and to seek instead
a comprehensive and ultimate explanation of naming.
For, though comprehensive answers to the first question are clearly futile, there are no obvious penalties attached to answering the second question
¹ ...
² Aristotle criticized Plato's theory largely as an inadequate answer to this question.
³ Socrates in the Phaedo says that it is the only acceptable answer to the first question. But the advance of science has undermined this thesis more thoroughly than the advance of logic has undermined the thesis that it is an acceptable answer to the second question.
[53]
in a comprehensive way.
Yet, I shall argue — and this will be my first thesis — that
any comprehensive explanation of naming is necessarily circular:
and that philosophers think that, in spite of this disadvantage, such explanations have some point largely because
they wrongly assimilate naming to natural processes.
Yet surely naming cannot be utterly artificial?
My second thesis will be that the desire to understand naming leads to a hunt for a completely satisfactory analogy:
but that all other processes either
already contain the very feature of naming which was puzzling,
or else
are too natural or too artificial to be really analogous;
and that it was the inevitable oscillation between these three points which prolonged the controversy about universals.
It is unnecessary to produce evidence
that philosophers who proposed
the existence of universals
thought that they were explaining
the unity of classes
and hence the possibility of naming.
What is debatable is whether this was an important motive, and this can be decided only in the sequel.
My first thesis, which I must now try to establish, is that
realism is necessarily a circular explanation of naming.
Now the answer to the question
'Why are we able to name things as we do?'
is
'The reason varies'.
For
it is always possible
with more or less ingenuity, depending on the degree of atomicity of the name,
to give a detailed informative reason;
and this reason will vary with the name.
But ultimately
there must be some exit from the maze of words,
and,
wherever this exit is made,
it will be impossible to give an informative reason
except by pointing.
For the only other way of giving an informative reason is to give a new word, and this would prevent the exit from the maze of words from being made at this place.
Still
at the place where the exit is made
it is always possible to give a detailed reason
like
'We are able to call things red because they are red',
which is too obviously circular even to look informative.
Or alternatively it is possible to say
'We are able to call things φ because they are φ',
and this is a general reason which is almost as obviously circular and uninformative.
What philosophers who propose the existence of universals do is
[54]
to propose
a general reason
which looks informative
because it shifts to another level,
but unfortunately is not.
It merely marks time:
but marking time can look very like marching if only the movements of the performers are watched, and not the ground which they profess to be covering.
Yet this ground could not be covered.
For the reason could not be informative even if it were detailed;
since there could be a non-circular answer to the question
'What universal?'
only if the exit from the maze of words were made at some different point,
which would merely put off the moment of embarrassment from which in the end neither speech nor thought can be saved.
Thus realism fails to escape the limitations of all explanations of naming;
that
they can be informative only if they are not general but detailed,
and then only if they are not given at the point where an exit is made from the maze of words.
Uninformative answers have their point.
They are silencing.
What is wrong with realism is not this, but that it masquerades as an answer which advances knowledge one step further.
The analytic machine acquires a momentum which carries it beyond the point where it ought to stop.
And there is an inveterate philosophical habit which strengthens the tendency to go beyond this point, or rather to think that one has gone beyond it.
'A thing is called by a certain name because it instantiates a certain universal'
is obviously circular when particularized,
but it looks imposing when it is left in this general form.
And it looks imposing in this general form largely because of the inveterate philosophical habit of treating the shadows cast by words and sentences as if they were separately identifiable.
i.e. Words serve different sentences differently; Change one word, change the whole sentence, incl. (potentially) all the other words in it.
Universals, like facts and propositions, are such shadows;
and too often philosophers by appealing to them in general terms have produced in their readers a feeling of satisfaction which ought to have been produced only by specifying them.¹
But universals are specifiable only by referencе to words.
Similarly facts may be brute and propositions may be definite, but what exactly it is about them which is brute or
¹ The same trick is played by those who say that laws of nature exhibit connections between universals. This gives the impression that we could independently know the eternal framework in which temporal things move and change, rather as we independently know how a piston must move by looking at a cylinder: ...
[55]
definite can be specified only by reference to the sentences which were the unacknowledged starting-points.
In all these cases it is tacit re-duplication which makes philosophers think that they can enjoy the benefits of specifying without actually specifying.
Yet the explanation of naming is incomplete until a particular universal is specified, and, when it is specified, the explanation immediately fails through circularity.
Naming is hazardous, and any attempt to make it foolproof by basing it on an independent foundation must fail in this way.
It is impossible to cross the gap between language and things without really crossing it.
Incidentally, this is why 'doing' is superior to 'thinking.'
"I've had a hard time conveying to intellectuals the intellectual superiority of practice." (Taleb)
Since the failure of realism to perform this feat is inevitable,
its rivals fail too.
Oh?
Nominalism, conceptualism and imagism, in so far as they are rivals of realism, are
attempts to provide a unity which will explain naming.
Nominalism says that a name is merely connected with a multitude of things, sometimes adding that these things are similar.
Conceptualism says that the name is not directly connected with the things but only via a concept, thus changing the nodal point.
Imagism says that the nodal point is an image.
And realism says that there is really no nodal point, since a name, though it appears to be connected with a multitude of things is all the time connected with only one thing, a universal.
This is an over-simplification of what these theories say about the One and the Many;
but it is enough for my next purpose, which is to show that these rivals of realism cannot produce a non-circular explanation of naming at those points where an exit is made from the maze of words.
The two psychological theories say that one word can apply to many things only because of the mediation of a concept or of an image.
Locke's abstract general idea is 'the workmanship of the understanding, but has its foundation in the similitudes of things'.
And Berkeley replaces it by an idea which 'considered in itself is particular but becomes general by being made to represent or stand for all other particular ideas of the same sort'.
But what
[56]
similitudes, and what representation?
In the end both Locke's concept and Berkeley's image are completely identifiable only by their use.¹
Of course we can partly identify images by describing their features:
and in this way we may even almost completely identify them, since certain images most naturally stand for certain things.
And the same could be said of concepts, if they were not merely philosophers' reifications of mental processes.
But this will not completely identify either of them,
since
thought may not follow the most natural course;
nor is it always clear which is the most natural course.
It is not so much that thinking is speaking as that
thinking is like speaking in the only way that matters:
it uses one thing as a symbol to stand for many things.
And the only tool which could not be used differently is the use.
Even something which had its use written on it could be used differently.²
And, if the psychological tool, whether concept or image, can be completely identified only by the things on which it is used, it cannot explain naming without circularity.
Conveniently for Reybrouck and the Ecologists, the
"'circularity' of stimulus and reaction"
requires no naming.
There can therefore be
at least some
"meaning"
in absence of "naming":
This meaning is
"the perceived functional significance of an object, event or place for an individual".
But
naming is not (currently) absent from human cognition,
or at least
it is present in the vast majority of
"affordance"
scenarios.
If Pragmatism is what works, and if it works even in spite of some residual cognitive interference from naming-that-doesn't-work, this is a yet stronger indication in favor of doing over thinking; but it is hard to imagine that this interference really is overcome each and every time something works. As so often, it seems more likely that we're getting away with something which only rarely comes back to bite us.
For, unless we point, the use can be specified only by backward reference to the name.
Nor is this circularity surprising.
For psychological tools have no advantage over words:
they are like them in being symbols, and unlike them only in being shadowy symbols.
The type of nominalism which says that a name is applied to a number of things which are similar immediately falls into the same circularity.
For
'similar' is an incomplete predicate, anything being similar to anything in some way,
perhaps a negative way.³
And in the end the kind of similarity which is meant can
¹ This is due to Wittgenstein: cf. e.g. Tractatus, 3.326, 'In order to recognize the symbol in the sign we must consider the significant use'.
² W. T. Stace in 'Russell's Neutral Monism' in The Philosophy of Bertrand Russell, pp. 381-3,
complains that neither Berkeley's precise image nor Russell's vague image
...
succeeds in explaining the generality of thought.
But
no description of any item of mental furniture which included only its momentary properties and not its habitual use could possibly explain the generality of thought.
³ Hence the point of many riddles. Cf. Stuart Hampshire, 'Scepticism and Meaning' ... Also Plato, Protagoras 331 d.
The Platonic theory avoids the 'similarity' difficulty, but not of course the general difficulty of which this is only one form. Speusippus, who abandoned the Platonic theory, seems to have held that,
since every species is like every other species in some way, it is impossible to define one species without defining every other species.
...
Cf. Brand Blanshard on individuals (not species).
'One never gets what is fully particular until one has specified its relations of every kind with everything else in the universe',
The Nature of Thought (London, 1939), vol. 1, p. 639.
Curiously enough N. R. Campbell
[57]
be specified only by a backward reference to the name.
Equally the type of nominalism which merely says that a name is applied to a class of things cannot say which class without a backward reference to the name.
Here the circularity is so obvious and there is so little to cushion the shock of the realization that naming is naming that this type of nominalism seems hardly tenable.
For, however strongly nominalists react against realism, they can never quite escape its influence:
once somebody had said that universals exist it could never be quite the same again.
Surely, one wants to protest, there must be some way of giving the class besides reference to the name?
Well
there is,
of course,
enumeration.
But this answer seems to fail to allow for the possibility of ever using the name correctly in any synthetic sentence.
For, if the class is given by enumeration, surely every use of the name must be either incorrect or analytic?
Since,
if to call a thing 'φ'
is to include it in the class of things called 'φ',
then surely
either it is incorrect to call it 'φ'
or else the class cannot be given without reference to it?
It is the example of realism which encourages these protests.
But it is a bad example.
Such neatness is not to be had.
For,
first of all,
these classes cannot be given by enumeration
of all their members,
since,
except for words belonging to dead languages,
they are never complete.
Nor is it true even that each member must either contribute or not contribute towards giving a class;
since
a name may be applied to the same thing twice,
once analytically and once synthetically,
and even a single use of a name may be synthetic for the speaker and analytic for the hearer.
I have not quite absorbed this distinction by osmosis, so I broke down and asked Gemini,
in ordinary language philosophy, what is the meaning of "analytic use" and "synthetic use" of a word?
An
analytic use
of a word occurs when the truth of a statement depends entirely on the definitions of the words involved.
...
Example:
"A father is a male parent."
If you argue that a father is actually a "cloud," you aren't wrong about biology; you're simply
misusing the word.
...
A
synthetic use
occurs when a word is used to describe a fact about the world that isn't true by definition alone.
...
Example:
"That father is wearing a blue hat."
Nothing in the definition of "father" requires him to wear a hat. To know if this is true, you have to look at the specific person.
It gives analogies:
"checking the dictionary"
vs.
"checking the news".
So, "A father is a male parent" could actually be synthetic from the perspective of . . . a child-hearer, for whom the sentence (as of yet) "isn't true by definition alone", owing to (as yet) incomplete understanding of the implicated word-concepts?
I broke down again and asked about this too. It gives the example of a child learning (via experiment!) the molecular makeup of water, whereas for the (adult professional) chemist H₂0 just is the definition of water.
Close enough for now.
(23 Jan 2026)
In fact the disjunction 'Analytic or Synthetic' cannot be applied simply to the addition of a member to a class without further caveats.
But this in itself is not enough to remove the difficulty;
it only makes it reappear in a new form.
For if the addition of a member to a class can be synthetic for the speaker and analytic for a subsequent lexicographer, then to what class was the member added?
Well, if we permit enough time to pass, our entire understanding of the world will indeed disintegrate. But that's a cop out vis-a-vis the present discussion. Ego knows that neither he nor anyone else can account for this, hence his argument against alter cannot be that alter has not accounted for it.
Surely we now have two classes on our hands instead of one?
An analogy will help us to deal with this new form of the difficulty.
Naming is like electing the sort of member who makes
arrives independently at a similar conclusion about species, when he is discussing the definition of such substances as silver, mercury or lead (Physics. The Elements, Cambridge, 1920, p. 50).
All attempts to explain the unity of a species by similarity
— whether by similarity of the individuals to one another, or by similarities and differences between the species and other species —
suffer from the same incompleteness.
[58]
a difference to a club.
Strictly we cannot say without qualification to what club he was elected,
since
it was one club before he was elected
and another club after he was elected.
The club building might be pointed out, and of course there is no parallel move in the case of naming, although realism pretends that there is.
But, even if there were no building or anything else of that kind, the puzzle about the two clubs would not be very perplexing.
Similarly, when we reject the simple application of the dichotomy 'Analytic or Synthetic' the resulting puzzle about two classes is not very perplexing.
All that is necessary is to point out that
a class is incompletely given by a changing quorum.
This may be untidy, but why not?
There is something radically wrong with a request to be given a class which is not satisfied either
with a reference to the name
or
with progressive enumeration.
It is a request to be given something without being given it;
as if somewhere, if only philosophers searched long enough, there could be found something which possessed
all the advantages of a word
and none of its disadvantages,
an epistemological vehicle which carried all its destinations.
I now turn to my second thesis, that nothing is sufficiently like naming without being too like naming.
Defenders of realism, like defenders of the other theories of naming, might object that the criticism contained in my first thesis is obvious, superficial and directed against a man of straw.
For realism does not offer a non-circular detailed explanation of naming — how could it? — but simply
gives a general characterization of the sort of unity which makes naming possible.
But notice how very like a dream realism is.
Taken literally it seems to be of little importance.
But, if it is taken as the expression of a doctrine which, if per impossibile it were true, would give it great importance, the suggestion is immediately repudiated.
Yet it does express such a doctrine, even if its exponents intermittently deny that it does;
and it is to the devious expression of this doctrine that it owes most of its attractiveness.
Its manifest content is little more than a harmless caprice, but its latent content is a serious error.
But has realism no point when it is taken simply as a general characterization of the sort of unity which makes naming impossible?
One might answer that it has no point, and that it succeeds in appearing to have some point only by the device of inventing
[59]
a new comprehensive term:
and that this device is considered effective only in philosophy, since outside philosophy it is too obviously like making an impressive gesture in the direction of the interesting object,
opening one's mouth and saying absolutely nothing.
But such a denial would be tantamount to a denial that any general characterization of the sort of unity which makes naming possible could have a point.
And surely such a denial would be wrong, since something can be done towards explaining the general possibility of naming by finding analogous processes?
For instance, what makes naming possible is one thing which is in many things as an ingredient.
But does this analogy throw much light on naming?
Any feature of logical mixing which is at all interesting seems to distinguish it from all other sorts of mixing.
The values of an unrestricted variable are strange receptacles.
What prevents contrary ingredients from being put in together, or an implicans from appearing without its implicate, is never the causal consequences.
And anyway the whole notion of mixing ingredients which were not there before the mixing is peculiar.
Could there be a logical conjuring trick?
Here defenders of realism might object that a new misunderstanding had replaced the old one.
For,
if realism is to be understood,
not only must a general characterization of naming be allowed,
but also
the verification principle must not be applied too crudely.
And anyway, if mixing is not a good analogy, this only means that some better analogy must be sought.
This objection might lead to a tolerant examination of other analogies.²
But fortunately it also opens up a short cut to the heart of the matter, which I shall soon take.
Now it would be taking too short a cut to repeat the platitude that naming is sui generis.
For it is natural to seek an analogy even if the search can never be completely successful.
And anyway Butler's truism applies to everything.
What is needed in order to explain the peculiar persistence of the debate about universals is something slightly longer, a demonstration that no analogy can be
sufficiently close
to satisfy philosophers without being
too close.
It is most natural to seek a visible process as an analogy to
¹ ...
² Metaphors must not be dismissed just because they are metaphors, as, e.g. 'copying' and 'participation' are by Aristotle, ...
[60]
naming, particularly for the Greeks who began this controversy.
Now previously I insisted that it is impossible in the end to give a detailed non-circular description of what makes it possible to name anything.
Here, however, it would be unfair to object that, if naming in general is compared to a visible process, still that process itself must be named.
For this sort of circularity is the inevitable result of the philosopher's predicament.
However, it is dangerous to begin speaking at all where so little can be said.
For it is fatally easy to think that one has separate access to what makes a name applicable just because one has separate access to whatever stands for this in the analogy.
But, waiving this, let us now take the short cut and ask what sort of visible process could be analogous to naming.
Let us try a rough analogy and say that one word is connected with many objects in the same way that the estuary of a river is connected with its many sources.
But this analogy fails because this connection just happens naturally.
We might then try to mend the analogy by saying that water follows the easiest course.
But this could be called choice only anthropomorphically, in an extended and weak sense of 'choice'.
In order to introduce choice in a restricted, strong sense, it is necessary to alter the analogy and say that people by directing the streams choose which sources shall feed the river.
But, if the first process was too natural to be like naming, the second is too artificial, since, for the analogy to work, the sources ought to have something in common besides the fact that the river is fed from them.
And it is difficult to find an analogy which is neither too natural nor too artificial.
The characteristic of naming which is difficult to match is that
the objects have something in common besides being called by one name,
but nothing in common which counts except that in virtue of which they are called by one name.
And this characteristic can be matched only by allowing that something makes it
convenient
but not absolutely necessary
for people to canalize streams into the river in the way they do,
and that whatever it is which makes this choice convenient is the only thing common to the sources which counts.
But this compromise between the two extremes introduces into the analogy the very feature which it was intended to explain.
For just how something works in influencing usage was what was to be
[61]
explained.
Nor is there a fourth alternative.
So after all even general analogical characterizations of naming do fall into a circularity which is closely related to the type of circularity which my first thesis exposed.
Neither in detail nor in general is it possible to step outside language.
This short way with analogies looks too superficial.
For suppose that it is granted that one of the things that metaphysicians do is to seek the unattainable:
that they hunt for
definitions which would in no way involve their definienda,¹
and for
analogies which would in no way involve what they were intended to explain.
Yet even so metaphysics is a natural and inevitable pursuit,
since
the easiest way to discover how far one can go is to try to go one stage farther.
And anyway
there is a difference between complete failure and partial success;
It depends on the task.
In an interview (somewhere on YouTube), Searle counters Wittgenstein with the assertion that we'll never know if there is a workable 'theory' out there to be found if we don't try to find it. I confess that this sort of argument triggers my inner consequentialist like few others can. It's a very good argument
qua
argument, but it seems possible to really
believe
in this argument only by ignoring some basic observations about what has (so far) become of us and our theories. One of the most common (and insidious) ways to be more sanguine about our prospects here is to believe indiscriminately in
partial success.
Basically, this is to think that
some
good
must
come of our theory-building efforts, whether or not those efforts succeed in their own given task. Perhaps no one (celeb philosophers included) really believes this in quite this polyannaish form. But then, where fall the skeptical nodes along the rest of this continuum of belief? And to which do you ally yourself? Taleb has had an outsize influence on my thinking here. He leads me to think that the presence of bad information is an ultimate evil, whereas the absence of good information is something we've learned to live with by the time we're walking and talking, hence it cannot really be such a dire problem as pro-theorists often take it to be. The crucial task, and the one we are
actually capable of competently performing,
ends up being: to keep the bad info out;
whereas the task of generating truth is all but guaranteed to miscarry, yet its products arrive stamped with the truth-seeker's variois seals of approval.
since, so long as analogies do not reach the point of self-frustration they get better and better as they approach it.
Well, I really think it depends. Definitely don't let any journalists observe the approach. That was the whole point of Taleb bringing in the blurry pictures study: those who see more intermediate steps make more incorrect inferences.
These two qualifications are just but they only serve to strengthen my thesis that it was oscillation between the three points which prolonged the controversy about universals.
For unless the possible analogies are mapped out in this simple way, it seems always conceivable that some altogether better analogy might lurk in an unexplored corner.
And what more are the rival theories of naming doing than seeking a completely satisfactory analogy?
It is only jargon which makes them appear to be doing something more.
The type of nominalism which suggests that things which are called by one name have only their name in common
represents the extreme of artificiality.²
It suggests that there are never any ways of telling even approximately whether a word is used in one sense or two senses.
Hmm. Do we yet possess a true forensics of senses? Are we just taking 'forensics' for granted here? It seems to me that most divergence of "senses" must be contingent and circumstantial rather than cognitive. e.g. Whether you use 'disinterested' to mean (a) impartial, or (b) apathetic, probably is a contingent matter of your exposure (accidental and intentional alike) to one or the other "sense". Only from a bird's eye view on 'society' are there two senses of the word; I am taking that view right now, but I do not use the word in both senses. The reason why is highly 'contingent': in my last year of college I just so happened to attend a lecture which featured a 90 second professorial tirade against use (b). That makes me an outlier case vis-a-vis explicitness, but I'm not so sure it is an outlier vis-a-vis contingency. Divergence seems a contingent matter of scale and (dis)continuity. It seems like it would be easy to create con-vergence of usage by taking a small number of people and isolating them from all the other people; and at that point it becomes pointless to interpret their inevitable di-vergence from literally everyone else in the world as another window in on the same names. So, I am suggesting that what we call different "senses" of the same word are, 'really' and 'contingently', just different words altogether. But I don't know anything about this. Dad would know, maybe I'll ask him
At the other extreme stands
the type of realism which suggests that there is always one method of getting a precise answer to this question.
In between are all the other theories of naming, which allow that it is neither impossible for the lexicographer to succeed in answering this question nor impossible for him to fail.
None of these middle theories is really wrong, since of course we do bestow common names on certain chosen groups
¹ Cf. J. Wisdom, 'Metaphysics and Verification' ...
² There are traces of such an extreme form of nominalism in Hobbes. Cf. Leviathan, Pt. 1, chap. IV, p. 13 (Everyman edition).
[62]
of things which exhibit certain similarities (else why should we do it?) or instantiate certain universals (why else were they invented?).
But on the other hand none of them goes deep enough to satisfy the true metaphysician who is in all of us;
since though they take us to the bottom of naming, we were in a simpler way already there, and they do not succeed in showing us how naming is founded on something else which lies even deeper.
Hence each of these middle theories
(except imagism, which says something empirical which seems to be false)
On "imagism" I can find only references to the movement in poetry, among whose strictures is found:
"To use the language of common speech, but to employ the exact word, not the nearly-exact, nor the merely decorative word."
Maybe this notion of 'exact' usage is the empirically false something referred to here? Or maybe my queries have been confounded by the Lit Critters once again.
develops its own thesis with embarrassing success up to a point, and can discredit its rivals only by accusing them of not going beyond that point.
But, since naming cannot be explained by anything which really goes beyond a reasoned choice of usage, this is an unfair accusation.
And its unfairness is concealed from those who make it only because each tacitly and wrongly assumes that his own theory alone does go beyond this point.
Thus moderate nominalists maintain that similarity is a better explanation of the unity of a class than the presence of a universal.
(But why should people not just recognize the presence of universals?)
And moderate realists retort that this admits the existence of at least one universal, similarity.
(But why should the presence of a universal explain the recognition of similarity if it cannot explain the recognition of anything else?
Why should people not just recognize similarity?)
Really these are not two arguments but two bare assertions of superiority.
They are manoeuvres which are carried out in a way which suggests that they are difficult and that they must be advances:
but both these suggestions are false.
Yet these theories do seem to be striving towards something.
And they are.
Their goal is the unattainable completely satisfactory explanation of naming.
And,
as so often happens in metaphysics,
progress is measured by distance from the starting-point and not by proximity to the goal
whose unattainability each uses against its rivals without allowing it to deter itself.
Excellent.
So,
there is
NOT
a difference between
complete failure
and
partial success
after all?
(Or, not this time?)
Thus theories of naming,
which seem to flout the verification principle
without therefore saying nothing,
can be interpreted as disguised analogies.
And, though there is a common limit beyond which they cannot go, the success with which they stealthily approach this limit, camouflaged in the technical terms of epistemology, varies.
But if this almost mechanical oscillation is
[63]
avoided what else can be said about naming?
Certainly as the first part of this article showed, detailed answers to the question why we name things as we do will in the end be circular.
Only the trick of giving a general answer as if it were a detailed one cloaks their failure.
If a word is explained ostensively, then however difficult this process may be it really is explained ostensively.
It is no good trying to combine
the concreteness of ostensive definition
with the clarity of verbal definition.
Verbal definitions have such an easy task just because ostensive definitions have such a difficult task.
Surveyors find it easier to fix the positions of points which they can visit than to fix the positions of points which they cannot visit.
Similarly it is easy to fix the relative positions of words:
but the points in things to which words are related are in the end inaccessible to logicians.
Then what else can be said about naming?
How does the lexicographer tell when a word is used in two senses rather than in one sense?
Surely there must be something in common to all well constructed series of things?
Yes, just that they are well constructed.
For this question already contains the equivalent of any possible comprehensive answer which could be given to it.
And, though in one way it is hard to see what detailed answers could be given to it, in another way it is only too easy to see.
For
we never reach a point where an exit must be made from the maze of words.
Admittedly, if a verbal explanation is given at one point, it is only successful if at some other point a connection with things is already understood;
and at some points it is more natural not to offer more words.
But at no point is an exit obligatory.
I would like to hear much more about this, but the paper is almost over.
It's a relief to be off the hook for finding an
exit.
I'll take it!
But I wonder if this is (ironically) a poor description (or just a false one) of what the above-named
-ists
are trying to do?
Seems to me that 'security in words' is the more pervasive fantasy.
So, if detailed reasons why we call a thing what we do are required, it is easy to give them;
but never ultimately or in the end, since here ex vi termini it is impossible to give them.
But philosophers tend to ignore this kind of detailed answer and press on.
But where to?
Perhaps to experimental psychology, in order to discover how changes in the sense organs, in training and in interests alter the ways in which people group things.
But this sort of investigation only gives the varying tests of the good construction of a series, and not its essence.
But what could its essence be?
When general analogical characterizations of naming have been mentioned, and detailed reasons why we call particular things by particular names, and the psychological background
[64]
of all this, what is left?
The desire to go on explaining naming is to some extent the result of the way these three fields have been confused, and to some extent the result of a natural feeling that in such a vast territory there might be something which lies outside these three fields.
But above all it is the result of the Protean metaphysical urge to transcend language.
