Vous êtes sur la page 1sur 8

PHILOSOPHICAL

STUDIES
Edited by WILFRID SELLARS and HERBERT FEIGL with the advice and
assistance of PAULMEEHL,JOHNHOSPERS,MAYBRODBECK
VOLUME XVI

Contents

December

1965

NUMBER 6

A Possible Extension of Logical Theory? by Geothey Hunter, UNIVERSITY


OF LEEDSANDUNIVERSITYOF NORTHCAROLINA
Beliefs Which Are Grounds for Themselves by John Turk Saunders, SAN
FERNANDO

V A L L E Y STATE C O L L E G E

Necessity, Contingency, and Punishment by David BIumenfeld and


Gerald Dworkin, UNIVERSITY OF CALIFORNIA, BERKELEY

A Possible Extension of Logical Theory?


by GEOFFREY H U N T E R
UNIVERSITY OF LEEDS AND UNIVERSITY OF NORTH CAROLINA

"Even after the observation of the frequent or constant conjunction


of objects, we have no reason to draw any inference concerning any
object beyond those of which we have had experience."
Hume, Treatise, 1.3.12
THE ~EART of my paper consists of the following two propositions:
1. A man who knows of at least one case of an X being a Y, and who does
not know of any positive reason for thinking that an X might not be a Y,
has some reason for thinking that all X's are Y's.
2. The contradictory of proposition (1)--i.e., "A man who knows of at
least one case of an X being a Y, and who does not know of any positive
reason for thinking that an X might not be a Y, has no reason at all for
81

82

PHILOSOPHICAL STUDIES

thinking that all X's are Y's"--is absurd in virtue of the meaning of what is
said, absurd in virtue of the concepts employed being what they are. If by
"analytic proposition" you mean "true proposition that is true simply and
solely in virtue of the concepts that occur in it being what they are (or being
what they are made to be by the propounder of the proposition) and not in
virtue of any other facts," then proposition (1) is analytic. It is certainly not
merely contingently true.
An article by Lewy in Analysis for September 1939 provides a convenient
starting point for the explanation and defense of these two propositions. In
that article Lewy claimed that inductive conclusions logically follow from
inductive premises, and he gave the following example:
The premisses are:
(1) Whenever I have heard barking in the past there was always a dog
somewhere near; and
(2) I am hearing barking now;
and the conclusion is:
(3) I have good reason to believe that there is a dog somewhere near
nOW.

Lewy claimed that (1) and (2) together entail (3). Why?
Simply because it seems to me that the following proposition "Whenever
I have heard barking in the past there was always a dog somewhere near, I'm
hearing barking now, but I have no reason whatever to believe there is a
dog in the neighbourhood" is self-contradictory. I can not prove that this
is so, but I should ask you to reflect on how we use expressions like "I have
good reason to believe," "It's probable," "It's very likely" etc. If you reflect
on how these expressions are actually used, I'm sure you will see that the
proposition I've just stated is self-contradictory in the very same way in which
it is self-contradictory to say "All men are mortal, Smith is a man but Smith
is not mortal."
Now I believe that Lewy was very nearly right. But his example was a bad
one. It is easy to show that ( 1 ) and (2) together do not in fact entail (3).
Imagine a situation in which the propositions that follow are true: Whenever I have heard barking in the past there was always a dog somewhere near
(Lewy's premise (1)). I am familiar with radio and tape-recording, but I
happen never to have heard barking on any radio or tape-recorder. I am in a
small boat in the middle of the Atlantic, with my wife. It is a clear day, and
there is no land or ship or other boat in sight. I happen to know that there is
no dog on board. My wife says, "Did I ever play you the recording I made
of the sheep-dog trials I went to last year in Wharfedale?" I say "No"; she
plays me the recording; and I hear barking now (Lewy's premise (2)). In

A POSSIBLE EXTENSION OF LOGICAL THEORY?

83

that situation I do not have good reason to believe that there is a dog somewhere near. (This is the contradictory of Lewy's conclusion (3).)
Thus Lewy's premises (1) and (2) could be true while the conclusion (3)
was false. So (1) and (2) together do not entail (3).
What was wrong with Lewy's example? He failed to take into account the
requirement that the arguer must know of no positive reason for thinking
that an X might not be a Y. Lewy was, I believe, right in thinking that the
fact that all known X's are Y's affords a presumption that all X's are Y's. But
what he failed to see was that this presumption may be weakened, and even
canceled out altogether, by other considerations. 1
The first person I know of to state clearly the importance in inductive arguments of the clause nisi ratio positiva obstet ("unless there is a positive
reason against") was Roger Joseph Boscovich in his De Lege Continuitatis
(1754). He also makes use of the legal notion of "presumption." This is
what Boscovich says in Article 40 of his Theoria Philosophiae Naturalis (first
edition, 1758). A marginal note to Article 40 says that it deals with "ubi &
cur vim habeat induetio incompleta" ("where and why incomplete induction may have force") :
9 . . As regards the nature & validity of induction, & its use in Physics, I may
here quote part of Art. 134 & the whole of Art. 135 from my dissertation
De Lege Continuitatis. The passage runs thus: "Especially when we investigate the general laws of Nature, induction has very great power; & there is
scarcely any other method beside it for the discovery of these laws . . . .
Now, induction should take account of every single case that can possibly
happen, before it can have the force of demonstration; such induction as
this has no place in establishing the laws of Nature. But use is made of an
induction of a less rigorous type; in order that this kind of induction may be
employed, it must be of such a nature that in all those cases particularly,
which can be examined in a manner that is bound to lead to a definite conclusion as to whether or no the law in question is followed, in all of them the
same result is arrived at; & that these cases are not merely a few. Moreover,
in the other cases, if those which at first sight appeared to be contradictory,
on further & more accurate investigation, can all of them be made to agree
with the law; although, whether they can be made to agree in this way better than in any other whatever, it is impossible to know directly anyhow. If
such conditions obtain, then it must be considered that the induction is
adapted to establishing the law . . . (Art. 135). In addition, whatever absolute properties, for instance those that bear no relation to our senses, are
generally found to exist in sensible masses of bodies, we are bound to attribute these same properties also to all small parts whatsoever, no matter how
small they may be. That is to say, unless some positive reason prevents this
(nisi positiva aliqua ratio obstet) . . . When we consider absolute, not relative, properties, whatever we perceive to be common to those contained

84

PHILOSOPHICAL STUDIES

within the limits that are sensible to us, we should consider these things to
be still common to those beyond those limits. For . . . if there should be
any violation of the analogy, this would be far more likely to happen between
the limits sensible to us, which are more open, than beyond them, where
indeed they are so nearly nothing. Because then none did happen thus, it is
a sign that there is none. This sign is not evident, but belongs to the principles of investigation, which generally proves successful if it is carried out in
accordance with certain definite wisely chosen rules. Now, since the indication may possibly be fallacious, it may happen that an error may be made;
but there is presumption against such an error, as they call it in law, until
direct evidence to the contrary can be brought forward. Hence we should
add: unless some positive reason is against it. Thus, it would be offending
against these rules to say that large bodies indeed could not suffer compenetration, or enfolding, or be deficient in inertia, but yet very small parts of
them could suffer penetration, or enfolding, or be without inertia. On the
other hand, if a property is relative with respect to our senses, then, from a
result obtained from the larger masses we cannot infer that the same is to
be obtained in its smaller particles; for instance, that it is the same thing to
be sensible, as it is to be coloured, which is true in the case of large masses,
but not in the case of small particles; since a distinction of this kind, accidental with respect to matter, is not accidental with respect to the term
sensible or coloured . . .-2
That passage illustrates something of the wide range and variety of things,
short of knowledge of an actual refuting instance, that could in a given case
be "positive reasons against." Perhaps this is the place to say that the bare
logical possibility of an X not being a Y is not to be counted as a positive
reason against thinking that all X's are Y's: the bare logical possibility of
something's being the case is no reason at all for thinking that it is the case.
What I am maintaining is this: The premises (1) I know of at/east one
case of an X being a Y and (2) I know of no positive reason for thinking that
an X might not be a Y taken together stand in a logical relation to the conclusion (3) I have some reason for thinking that all X's are Y's, a relation
such that the conjunction (1) and (2) but not (3), though not self-contradictory, is absurd in virtue of the meaning of what is said.
If (1) and (2) are both true, then (3) cannot be false; and this is not iust
a physical impossibility. If someone asserted (1) and (2) and yet denied
(3), then I should not know what he was suggesting, and this would be because his words mean what they do.
There is a general point of logical theory involved here. Consider the possible Iogi~al relations between a conjunction of propositions, say p & q & r,
and another proposition, say s. What I am claiming is that there are not just
two possibilities, viz. either (i) p & q & r & ,~s involves a contradiction, or
(ii) p & q & r dr ,~s is perfectly all right so far as logic is concerned. There

A POSSIBLE EXTENSION OF LOGICAL THEORY?

85

is a third possibility, viz. (iii) p & q & r has a logical relation to s such that
p & q & r & ~s, though not involving a contradiction, is absurd in virtue of
the meaning of what is said. If someone asserted p & q & r & ,~s, then we
should not know what he was suggesting, and this in virtue of the meaning
of what was said: for p and q and r could not all be true while s was false, and
this impossibility is created by the meanings of the words used; it is in some
sense a logical and not a merely physical impossibility.
Neither (1) nor (2) by itself entails (3). It is obvious that I could know
of at least one case of an X being a Y (premise (1)) and yet have no reason
at all for thinking that all X's are Y's (contradictory of (3)). It is equally
obvious that (2) does not entail (3) : for it might be true that I knew of no
positive reason for thinking that an X might not be a Y, and yet that I had
no reason at all for thinking that all X's are Y's; this would be the case, for
instance, if I knew nothing at all about X's and Y's. It is only the conjunction
of (1) and (2) that stands in the logical relation I have described to the
conclusion (3).
The occurrence of even one case of an X being a Y affords the man who
knows of it a presumption that all X's are Y's, provided he knows of no positive reason against. Throughout when I speak of "some reason" I mean the
contradictory of "no reason at all."
So far I have considered only that form of argument in which one of the
premises is that the arguer knows of at least one case of an X being a Y, and
the conclusion is that the arguer has some reason for thinking that all X's
are Y's (where All X's are Y's is a merely de facto universal proposition). I
shall call arguments of this kind "simple inductive arguments."
There are conclusions which we should like to be assured of that cannot
be reached by simple inductive arguments alone; for instance, the conclusion that we have some reason for thinking that a moving body not acted on
by any external force would continue in uniform motion in a straight line.
There are no bodies not acted upon by any external force, so we cannot in
this ease use arguments that start from the premise " W e know of at least one
case of a body not acted on by any external force that . . ." etc.
In such cases the following form of argument seems to me to be a sound
form: (11 ) I have good reason for thinking that if p then q and r and s . . .
and (12) I have good reason for thinking that q and r and s . . . and (13)
I have no good reason for thinking that any definite suggestion that I know
of, other than the suggestion that p, is a true explanation of the facts that q
and r and s . . . Therefore: (14) I have some reason for thinking that p.
A premise of the form (11) may be known to be true even when the
proposition p contained in it is not known to be true, for the hypothetical
contained in such a premise may be analytic. For example, let "p" be "Ele-

86

PHILOSOPHICAL STUDIES

ments consist wholly of atoms, for each element its own kind of atom, and
every atom of a given element has the same weight as every other atom of
that element; and any given compound consists wholly of certain discrete
portions every one of which consists wholly of a whole number of atoms of
various kinds and has the same number of atoms of each kind as every other
discrete portion (in the sense explained above) of the compound." Then
though p itself could hardly be known to be true (unless it is taken as analytic), the hypothetical "If p is true, then the proportions by weight in which
the elements enter into a given compound are constant for that compound"
could be known to be true, for it is (so far as I can see) analytic.
Again, a premise of the form (12) may be known to be true, where q and
r and s are suitably chosen; they could, for instance, be propositions about
particular conjunctions of observable things.
The importance of premise (13) can be brought out in the following way.
Suppose we did not require a premise of the form (13), but allowed instead
that all arguments of the form "(11) and (12), therefore (14)" were sound
ones. Let "p" be "There exist imperceptible malicious demons with a spite
against the human race, who have no moral scruples and are not punished for
anything they do, and who have limited powers but are able to make cars
break down." Then we should have to allow the soundness of the argument
(11) I have good reason for thinking that if p were true, my car would occasionally break down in bad weather miles from anywhere; and (12) My
car does occasionally break down in bad weather miles from anywhere; and
therefore (14) I have some reason for thinking that there exist imperceptible
malicious demons with the specified qualities and powers--for both premises
are true, since the hypothetical contained in (11) is so nearly analytic that
I have good reason for thinking it true.
This is only the beginning of a possible extension of logical theory. For it
should be possible to go on to give, for other kinds of conclusions, sets of
forms of premises that stand in the logical relation that I have described to
their conclusions. For instance, if the conclusion I want a reason for is that
something or other is a law of nature, then one of the premises might be
"I know that properly controlled experiments designed to test the hypothesis
that it is a law of nature that . . . have been carried out and that they have
not falsified the hypothesis." (I am not saying that I can have no reason for
thinking that something is a law of nature unless a premise of that form is
true; only that a premise of that form, together with certain others that the
reader can work out for himself, and that could be known by the arguer,
would stand in the logical relation I have described to the conclusion "I
have some reason for thinking that it is a law of nature that . . .")
To finish, consider the quotation from Hume at the head of this article.

A POSSIBLE EXTENSION OF LOGICAL THEORY?

87

Is it true, understood in the sense in which it would most naturally be taken


by anyone who knew nothing of its original context (I do not want to discuss
the textual question of exactly what Hume meant by it)? It is not true. If
we have observed the constant conjunction of objects, then, provided we
know of no positive reason for thinking otherwise (and the mere logical
possibility of things being otherwise is not a positive reason for thinking
that they are otherwise), we have some reason for thinking that any unobserved objects of those kinds will also be conjoined; and this proposition
(i.e., the whole hypothetical from "If we have observed . . ." to " . . . will
also be conjoined") is in some sense logically true. Similarly, if we have
observed the frequent but not constant conjunction of objects, then, provided we know of no positive reason for thinking otherwise, we have some
reason for thinking that if there are any unobserved objects of those kinds
then some of them will be conjoined, and this proposition too (the whol."
hypothetical) is in some sense logically true. So if Hume intended to deny
either of those propositions, what he himself says must be logically absurd.
It might be objected that what Hume says cannot be logically absurd, because it is possible for a person to believe it and to act in accordance with his
belief, as Pyrrho is said to have done: "He led a life consistent with this doctrine, going out of his way for nothing, taking no precautions, but facing all
risks as they came, whether carts, precipices, dogs or what not, and, generally,
not letting himself be guided by his senses. But he was kept out of harm's
way by his friends who . . . used to follow close after him . . . He lived
to be nearly ninety.''a
There are two possible replies to this objection, either a simple one that
is probably right, or a more sophisticated one. The simple reply is to say that
the objection presupposes that it is impossible either to believe what is logically absurd or to act on such a belief, and that such episodes in history as,
say, Spinoza's claim that there must be an explanation for absolutely everything, and Hobbes's publications on squaring the circle, as well as the existence of several twentieth-century publications that the reader may think of,
are at least prima facie evidence that there is no such impossibility. The
other reply is this: "Pyrrhonian and Humean skeptics do not believe the
logically absurd things they say or write, for it is logically impossible to believe what is logically absurd (since there is nothing there to believe). All
that happens is that they speak or write or rehearse to themselves, or have
a tendency to speak or write or rehearse to themselves, various groups of
words. And the Pyrrhonian acts, not in accordance with a logically absurd
belief, but in accordance with some such belief as that expressed by the words
'I never have any reason at all for drawing any inference about objects I have
not experienced,' a belief that taken on its oven is not logically absurd. The

88

PHILOSOPHICAL STUDIES

logically absurd 'proposition' is 'Past experience gives m e no reason at


drawing any inferences about objects I have not experienced.'"
Received

a11 for

April 17, 1964

NOTES
An example of the way in which a very strong presumption afforded by a mass of facts
may be upset by the addition of one further piece of evidence occurs in Vance Packard's
The Status Seekers (London: Longman's, 1960), p. 47: "Sociologist Raymond W. Mack
told me of an interesting exercise he sometimes gives his students in stratification. He asks
them to 'place' in the class structure a man with these characteristics: "He is a graduate
of Indiana University and has a law degree from Ohio State. His father, a small businessman, was a high-school graduate. His mother had two years of college. He drives a 1958
Buick . . . he has his own law office . . . he is a Methodist . . . he has a $12,000
income . . . his two children are university students." At this point Mack asks the
students if they now have the man pretty well pegged as to status. Usually they nod that
they have. Then he adds: 'Oh, yes, and one other thing. He is a Negro.' "
Quoted from the 1763 edition, translated by J. M. Child (La Salle, IlL: Open Court
t'ublishing Co., 1922), pp. 57 and 59.
Diogenes Laertius ix. 62.

Beliefs

Which Are Grounds for Themselves


by J O H N T U R K S A U N D E R S
SAN FERNANDO

VALLEY

STATE COLLEGE

CAN BELIEFS ever be grounds for themselves? More specifically, can the
fact that I believe that P be one of m y grounds for believing that P? I wish to
show that it can, and thereby to refute the contrary dogma which is abroad
these days, for example: " . . . I cannot be said to rely on my beliefs at all.
It is absurd to speak of m y believing that P on the basis of the fact that I
believe that p.,,1 "My belief cannot be included a m o n g the grounds for itself. ''2 I use the term "dogma" advisedly, for I have seen no arguments offered by its proponents in behalf of this thesis.
It is possible that no defense has been provided because the thesis has been
viewed as self-evident. It may have been thought, that is, that its denial commits one to the legitimacy of circular reasoning. You ask me my reasons for
P and I cite N, O, and P. But this, of course, would be a mistake: the thesis
is not that P cannot be one of my grounds for believing that P, but that the
fact that I believe that P cannot be one of my grounds for believing that P.
And such a c o m m i t m e n t is entailed by the denial of the former, but not by
the denial of the latter.

Vous aimerez peut-être aussi