STRAWSON ON INDUCTION

GEORGE SCHLESINGER

One of the most significant results of ordinary language philos-

ophy is Strawson's dissolution of the problem of induction. 1 There
have been many attempts to solve the problem of induction by
people who held the view that it is a general principle that any
argument which may lead from true premises to false conclusions
is invalid and who were perplexed by the fact that inductive argu-
ments seem to violate this principle. Strawson's view differs sharply
from that held by these philosophers who in his opinion have been
laboring under the misapprehension that for inductive reasoning
to be justified it would have to be shown ultimately a form of de-
ductive reasoning. This he thinks is absurd since inductive reasoning
is precisely that: inductive and decidedly not deductive. The
question whether there are good reasons for believing in conclusions
arrived at inductively simply does not arise for by the very rules of
our language 'having good reasons to believe that p' means just 'p is
the conclusion of inductive reasoning. 2 . At most one can inquire
whether it is the case that by the rules of ordinary language if a
certain event has been established by induction to be going to occur
then it is called 'having a rational attitude' to expect that event to
happen, and the answer is, yes.
There is an important point here that has not always been appre-
ciated. One may be tempted to argue impatiently that the rules of
ordinary language are of little value in helping us to discover the
nature of reality and that they cannot create facts. Whether a certain
event E is going to happen or not is a factual question which cannot
be decided by linguistic legislation. But Strawson explains that the
linguistic rule according to which if it can be shown by inductive
reasoning that E is going to happen then it is reasonable to believe
that E is going to happen is entirely safe from empirical refutation.
For if it turns out that E does after all not happen it can still be
maintained that it was reasonable to expect that E was going to
happen s and that in the context of the circumstances prevailing at
the time it was correct to maintain that E was likely to happen.

199
G. SCHLESINGER

Strawson ridicules the question 'Is the universe such that inductive
procedures are rational?' since the nature of the universe has nothing
to do with the question what beliefs are rational to hold. It is purely
a matter o f convention which may be upheld in the face o f any
future experience that it is rational to assert that I have good reasons
now to believe E is going to happen if this belief is a conclusion o f
inductive reasoning.
While some philosophers have been antagonistic many others
have endorsed Strawson's view that the problem o f inductica i' a
pseudo-problem. Indeed the measure o f his influence may be gauged
from the fact that nowadays when someone is said to be engaged in
investigating the problem o f induction he is as a rule automatically
assumed to be engaged not in trying to justify induction but in
trying to describe in detail the rules of induction.

II

Before probing any deeper into the matter let me mention that
there are two points in Strawson's account which I am not going to
defend, which in my opinion he himself would withdraw were he
to give more thought to the matter and which are anyhow not
central to his thesis.
The first point is his drawing a parallel between induction and
deduction and going as far as claiming that in the case o f deduction
it is even more obvious than in the case o f induction that validity
is conferred by convention. He says that if someone asked what
grounds there were for regarding deduction in general as a valid
method o f argumant 'we should have to answer that his question
was without sense, for to say that an argument, or a form or method
of arjgument was valid or invalid would imply that it was deduc-
tive : It should however be obvious that the situation has not been
correctly characterized already from the fact that future experience
is fundamentally differently related to deductive than to inductive
reasoning. For suppose it has been observed that p, and q is said
deductively to follow from p, then if we now observe that q is false
then - unlike in the case of induction - we cannot continue to
insist that nevertheless we are correct in asserting that p logically
implied q. No, if p is known to be true then the fact that q is false
shows that the deductive rule we have used to arrive at q must have
been invalid. Thus when we are convinced that p logically implies q
then we can back up our conviction by something more than

200
STRAWSONON INDUCTION

linguistic convention since our conviction carries a definite claim

about the nature of the universe the truth of which cannot be
guaranteed by any rule of language. When we claim that p logically
implies q then at least part of our claim is that a situation can
never arise in the world in which p is found true while q is found
false.
The second point concerns Strawson's apparent attempt to go
beyond his original contention that methods other then induction
to form beliefs about the unobserved are inferior simply because it
has been agreed to call them irrational and say something stronger
in their disfavor. He considers a person who wishing to surmise what
the future is going to be, doses his eyes, asks himself the relevant
question and accepts the first answer that comes into his head.
Suppose we ask him whether he usually obtains by his method the
right answer. Strawson says:
He might answer: 'You've mentioned one of its draw-
backs; I never do get the right answer; but it is an extremely
easy method'. One might then be inclined to think that it
was not a method of finding things out at all.S
This remark is either irrelevant or unwarranted. When Strawson
says that this person's method 'was not a method of finding things
out he must have either meant that in the past it was found not to
be or that in the future it will not be found to be such a method. In
the former case we must ask: not assuming now that induction is
the only valid method or predicting the future - for surely such an
assumption would amount to begging the question - why should it
be thought relevant that the method was unsuccessful in the past?
What has past failure to do with future failure? In the latter case we
may ask: What is the justification for thinking so? Since we are not
assuming now the validity of induction what makes us inclined to
think that the method may not be successful in the future? there
does not seem to be any way in which Strawson's remark can be
presented as both relevant and justified.

II!

Now 1 propose to raise some questions which go more to the

heart of Strawson's thesis. He points out that it makes good sense to
ask about a particular belief concerning the future whether its
adoption is warranted for then we are asking whether by the
accepted standards of induction it follows that the belief should be

201
G. SCHLESINGER

adopted9 However, it makes no sense to go on and ask whether the

application of those very standards is well-grounded. The situation
according to him is very similar to that which obtains in the case
of questions concerning the legality of certain practices. One may
sensibly inquire whether a given action is legal to which the answer
will be given by an appeal to the existing legal system which sanc-
tions or prohibits that particular action9 However Strawson says:
9 it makes no sense to inquire in general whether the law
of the land, the legal system as a whole, is or is not legal9
For to what legal standards are we appealing. 6
Strawson's comparison is a very useful one and I shall use it in my
effort to'get a better picture of what his position amounts to.
( I ) Strawson is surely correct in saying that a person may well
ask for instance 'Is it legal to bring into the country merchandise
from abroad valued at $1000?' however upon being that it is not, he
is not to ask 'Is the law which prohibits such an act legal?' Suppose
however that we were told that in every country in the world
it was the law that the limit for bringing in undeclared goods
acquired in a foreign country was exactly $300. Also suppose
that we were told that this limit has been laid down by the approp-
riate authorities in each country without any prior consultation
with the authorities of any other country and without even the
knowledge of the law anywhere else. Surely we would regard this
quite incredible. How come that without any collusion all the
authorities arrived at exactly $300 as the limit for the value of what
may be legally be imported from abroad? Surely there is no distinct
feature of the universe which makes $300 as the natural sum for that
purpose.
Similarly then in the case of induction: what constitutes a valid
form of inductive reasoning happens to be agreed upon to the very
last detail in all the languages in the world9 How is it that all the
people without exception arrived precisely at the same convention
when there is nothing in the objective nature of reality which would
call for such a convention? It is for example a convention in the
English language that 'green' stands for the color green and we may
leave it at that. However if in every language in the world the color
green was denoted by 'green' then surely we would regard this as
most remarkable and investigate the reasons for it.
Strawson may well reply that he wished to claim only that we
cannot give logical reasons for adopting inductive practices but by
this he did not mean to imply that the rules which govern those prac-
tices are entirely arbitrary; we can and should give an explanation

202
STRAWSON ON INDUCTION

why there is a convention to employ these particular roles in empir-

ical enquiries. The explanation will be psychological. That is, while
Strawson will deny that inductive reasoning are objectively superior
to any other form of reasoning from the observed to the unobserved,
it does have the required features which make them psychologically
attractive. The explanation as to why everywhere in the world all
people subscribe to the same rules of induction because it is a uni-
versally shared human urge to anticipate the unobserved to turn out
the way they are predicted to turn out on the basis of these rules.
(2) While good sense precludes us from asking whether the law
prohibiting the import of unlimited amount of foreign goods is legal
it is perfectly in order to ask why such law has been enacted in the
first place. Such a question would not be left unanswered either;
the economic factors which necessitate the law could be stated in
detail in defending it. The same applies to every law: we do not ask
of course that they be legally justified but we do ask that they be
justified in a satisfactory manner, that is, they they be shown to
fulfill some legitimate purpose. It may safely be stated that if people
were to reach the conclusion that a particular law of country did not
serve any useful purpose and that it merely causes inconvenience
and its origin cannot be explained by nothing else but that it ful-
filled a psychological need of the legislators then they would fight
for the abolition of that law.
When it comes to liguistic conventions then in some cases we are
justifiably quite indifferent to the fact that there is no objective
basis for their adoption. For instance we are not worried why the
color green is denoted by the word 'green' and not 'red' but then it
makes no practical difference what word denotes green. We should
however not be indifferent to the question as to what linguistic
convention determines the circumstances under which we say of
event E that it is reasonable to expect it to happen. This convention
may have a great deal of practical implications. Suppose E is a highly
undesirable event which we should very much want to avert but
according to the rules of induction it follows that it requires
action of considerable complexity and strenuousness to prevent E
from happening. In a situation like this it seems we should want a
good explanation why these rules, which impose upon us the need
for such burdensome activities, had to be adopted. As in the case of
a law in connection of which we would be told that the sole reason
for its enactment was to satisfy some psychological quirk on the part
of the authorities, here too we ought to feel that the linguistic
legislation in question is an unwarranted imposition upon us. But in

203
G. SCHLESINGER

fact we never question the adequacy of the rules of inductive reason-

ing. Thus Strawson's claim that they are rules adopted merely by
convention and his comparison between these rules and the legal
rules which govern our behavior seem implausible.
(3) In the case of the law against importing foreign goods above
a certain value we can easily imagine that if the law were abolished
then we would act accordingly and cease giving another thought
about what we brought back with us upon returning from our visits
abroad. Would Strawson maintain that if by general consent of
writers, journalists and others who shape linguistic practices, the
convention concerning what is to be called a well-grounded belief
was altered, then all of us would cheerfully go along with this change
too? It would seem that we would not, but it might be contended
that the reason is something we have already mentioned: inductive
reasoning has strong psychological appeal. But if the convention
were changed we would be left with no rational grounds upon which
to defend our beliefs arrived at by the old rules of reasoning. Yet I
am sure there are many people like myself who feel that even if all
the philosophers who had the highest regard for universally pro-
claimed conventions tried their hardest they could not talk me into
accepting any new rules to determine the meaning 'a well-established
hypotheses'.

IV

In my 'Justification of Empirical Reasoning 7 I have advanced a

thesis which I shall briefly discuss here and through which Straw-
son's position may be vindicated. While nowhere in his writings
does he make any allusion to the thesis in question, it is by no
means necessary to assume that he would wish to reject it. The
essence of the thesis is that the inductive method as it is uni-
versally practiced is imposed upon us by a process of elimination:
that it is the one and only one generally applicable method of
forming hypotheses which will both account for the observed and
make a concrete conjecture about the unobserved, we are left with
after we have eliminated every other method as unsuitable.
Let me begin by considering for a moment the following rule for
the selection of our hypotheses to be adopted:
R: Always assume that the unobserved will be unlike the
observed.

204
STRAWSON ON INDUCTION

R has been called "counter4nduction". It is obvious that we

should reject the suggestion that we adopt R. Not so much because
it is not likely to lead to the correct hypothesis. But rather because
of something which is much more clearly evident, namely, because R
is simply not a useable rule. Given for example that in the past
Newton's law of gravity has been obeyed R tells us no more than
that in the future we should expect that some other law will be
obeyed. But there are infinitely many law-like statements which
differ from Newton's law. R does not lead us to the adoption of any
particular one among these. With the help of R we get nowhere and
can form no opinion what law will be observed to govern the inter-
action of masses.
But what about the rule always to assume that the unobserved
will belike the observed? This as it stands would not be an adequate
rule for the selection of hypotheses since it leaves us with the prob-
lem that there are infinitely many ways in which the observed may
be described each description giving rise to a different expectation of
what the unobserved will be like. This problem has been first dis-
covered by H. Joffreys and subsequently associated with the name
of N. Goodman. His version of the problem is as follows: suppose
we define
Grue = observed before the year 2000 A.D. and found green or
observed after the year 2000 A J). and found blue.
It is clear that all our past observations concerning the color of
emeralds permit us no loss to say that all emeralds have been grue
than all emeralds have been green. If however we describe our
experiences as having observed that all emeralds were grue and
assume this to be true in the future as well we expect emeralds to
look different after 2000 A.D., than if we subscribed to the hypo-
thesis that all emeralds are green.
A large number of solutions has been suggested. Here 1 shall
discuss only two. Goodman's own solution is that there is a method-
ological principle according to which, of all the possible predicates
we could apply to the subjects under investigation, we must apply
only the predicate which is most entrenched in our language as
'projectible'. Therefore since the predicate 'green' is well entrenched
in the language whereas 'grue' is not, the hypothesis 'All emeralds
are green' is projectible while 'All emeralds are grue' is not.
Goodman confines himself to stating it as a fact that everybody
regards it as natural to use the most entrenched predicates only;
he makes no attempt to defend this principle, I should like to point
out however that if indeed the question should arise, whether

205
G. SCHLESINGER

we should use (a) the most entrenched predicates in our projections

or (b) w e may (or should) use unentrenched predicates, then by
just once more by applying the very compelling principle that the
hypothesis-selection-rule we adopt must be at least useable in the
sense that it leads to the selection of a particular hypothesis, we
arrive at the conclusion that (a) must be adopted. For in case we
adopt (a), then in a situation where we have made a large number of
observations concerning the color of emeralds, the only entrenched
predicate we find applicable to our past experience is 'green' and we
are therefore led to the choice of the specific hypothesis 'All
emeralds are green'. If however we adopted (b) then we would not
epfid up with the choice of any specific hypothesis. There are after
all infinitely many unentrenched predicates such as 'grue'l which is
defined like 'grue' except that '2001 A.D.' replaces '2000 A.D.' and
'Grue'2 and so on. Then we have 'gred', 'grack', and the like when
'red' and 'black' replaces 'blue' in the definition. (b) provides us
with no instruction whether we should select 'All emeralds are
gruel ', 'All emeralds are grue2' 'All emeralds are gred' or what.
The other solution is due to S.F. Barker and P. Achinstein. s
They advance the rule that we must use non-positional predicates
only in our projections. Briefly, the difference between positional
and non-positional predicates is this: for all instances of application
of a non-positional predicate a single picture of representation can
be given. On the other hand at least two different representations
are needed to cover all instances of application of a positional
predicate. For example, a single patch of green paint in the present
can represent the colour of all green things irrespective of their
dates; but two present patches will be needed to represent all
grue things: a green one for cases up to 2000 A.D. and a blue one
for cases thereafter.
It is seen at once that in the context of the example in which
Goodman's problem has been raised the rule suggested by Barker
and Achinstein is adequate: it disqualifies all such predicates as
'grue'l, 'grue'2, 'gred', grack', etc. and permits the use of 'green'
only. On the other hand, the contrary rule which says that we
should use non-positional predicates would be useless for it would
leave us undecided whether we should adopt 'All emeralds are green'
or 'All emeralds are grue2' and so on.
I do not wish to claim that either of the two proposed solutions
have to be accepted. All I wanted to do is to illustrate that a solution
which may be advanced with any plausibility at all must be based on
the principle that we reject any rule that does not help us to select a

206
STRAWSON ON INDU(VFION

particular hypothesis. It is my contention that any other solution

that may look reasonable will presuppose this principle. It is also my
contention that this principle ultimately leads to the rejection of
all rules except the set of rules which constitute inductive reason-
ing as we know it.

I have not considered hypothesis-selection rules which rely on

empirical data other than those which the hypothesis to be selected
is to account for. Hypothesis-selection-rules which require empirical
data yet to be produced e.g. 'Choose the hypothesis that accounts
for the observed data and which comes first into the mind of person
P which requires that we obtain information about P's thoughts,or
'Choose the hypothesis that accounts for the observed data by inter-
preting the flight of a certain bird' are to be excluded for several
reasons. First of all, as 1 have already implied before, general appli-
cability is a basic desiratum in the method we should want to adopt
for selecting our hypotheses. Now a hypothesis-selection-rule which
requires no other empirical data than those the hypothesis to be
selected is to account for is equally available to everyone in the
possession of these data while with respect to extraneous empirical
information different people are differently situated; some may have
no access to it at all. Secondly, before we have adopted the set of
principles which we shall use in empirical reasoning we cannot
assume that a certain empirical method will indeed produce the
required results. For example we cannot expect that a given bird
will fly or even just that it will continue to exist the next moment
unless we have already adopted some method for the selection of
hypotheses which lead to the conclusion warranting such an ex-
pectation.
Rather than considering further examples let us assume that any
rule concocted by anyone and not belonging to the set of rules in
actual use in inductive reasoning can be shown to be inadequate for
whatever the nature of the universe, upon grasping the meaning of
the rule we at once realize that it is unuseable, and see how this
helps us in asnwering the question raised concerning Strawson's
thesis. To the question, what were the objective grounds upon which
the convention to regard a belief arrived at by inductive reasoning as
well-established was adopted in the first place the answer is: it is
the only feasible convention we could have since any alternative set

207
G. SCHLESINGER

o f rules for reasoning, is not adequate for adoption. To tile question

what would happen if the convention were changed, the a n s w e r is
that we are dealing here not with an arbitrary convention but one
whose reasonableness is inherent in the nature o f things. This fact no
one can change; it will always be true that tile convention sanction-
ing the present rules o f induction is the convention reason imposes
upon us.
What we may say then is that Strawson is quite right in claiming
that it is a universal convention to regard it 'reasonable to expect
that p ' in case p is the conclusion arrived at by the use o f the rules
governing inductive reasoning. However, it is not an arbitrary con-
vention nor one merely arising out o f our psychological needs. If we
are to agree on any rules at all then these are the only viable rules we
can have.

UNIVERSITY OF NORTH CAROLINA

CHAPEL HILL, NC 275 ! 4
USA

NOTES

I P.E. Strawson, Introduction to Logical Theory (London, 1952), pp. 248-

263.
a Ibid,m p. 249.
s Ibid.,p. 262.
4 Ibid., p. 249.
s Ibid.,p. 259.
Ibid., p. 257.
7 Philosophical Quarterly, July, 1979.
s 'On The New Riddle of Induction', The Philosophy of Science,
P.H. Nidditch Ed. (Oxford, 1968).

208