> CEB@ #jbjb (RlHHHHHHH8p,0444444xzzzzzz,+ K^H44444DHH44,DDD4H4H4xD\66HHHH4xD DdHHd dOkddddDdDBOOK REVIEW
[Review of David Stoves The Rationality of Induction, Canadian Philosophical Reviews, vol. 7 (1987), no. 11, pp. 464-466.]
D. C. Stove.
The Rationality of Induction.
Oxford: Clarendon Press, 1986.
Pp. iii + 231.
ISBN 0-19-824789-3.
On the back of a recent paperback edition of Popper's The Logic of Scientific Discovery one finds the following two evaluations of the book, both of which originally appeared in Erkenntnis. The first, expressing outright condemnation, is from Reichenbach: 'The results of this book appear to me completely untenable I cannot understand how Popper could possibly believe that with respect to the problem of induction his investigations mean even the slightest advance.' In contrast, the second is a ringing endorsement. According to Carnap, the book is 'One of the most important contemporary works in the field of the logic of science.' Today, some fifty years after the first appearance of Popper's LSD, Stove's The Rationality of Induction will no doubt evoke this same intensity of vigorous and diverse opinion, in equal measure scandalizing and enraging some, delighting and exhilarating others.
The book takes as its main task a defense of the rationality of inductive inference in light of the now prevalent presumption in favour of inductive scepticism. It was Hume's thesis in the Treatise that all inferences from premisses about the observed to conclusions about the unobserved are unfounded, and it is exactly this thesis against which Stove argues. Of course, Stove does not mean to argue against the thesis that inductive inference is fallible. Rather, his only aim is to show that, given minimal and plausible logical and probabilistic assumptions, such inferences often can and do raise their conclusions' probability in such a way that it becomes irrational not to accept them. To the sceptic who wishes to disallow discussion of logical probability altogether, Stove points out that the sceptical thesis itself makes use of such probabilities since the sceptical thesis is in fact a statement about how probable the conclusion of an inductive inference is in relation to its premisses (43). After all, in its strongest form, the sceptical thesis is just the denial of the proposition that a conclusion's probability is ever altered as a result of observational evidence (40, 42, 44). It is in this sense that the premisses of an inductive inference are often thought to be 'irrelevant' to their conclusion. In its minimal form the thesis is just that 'such conclusions never have, in relation to their premisses, as high probability as non-sceptics suppose' (44). Against both versions of the thesis Stove produces convincing counter-examples.
Central to Stove's purpose is his claim that inductive inferences, whenever they are in fact rationally justified, are justified necessarily. That is, whenever e is an inductive reason in favour of h, it is so necessarily. After all, even the sceptical thesis is simply a particular kind of assessment of the probability that one proposition has in relation to another, and such assessments are never contingent.
Stove gives two main arguments in support of this claim. The first is that, in the case of inductive inferences, such assessments of probability are compatible with every contingent proposition. Since a proposition is itself contingent if and only if there is some contingent proposition with which it is inconsistent, it follows that such assessments must themselves be necessary (4f). Secondly, if the sceptical thesis were in fact contingent, as well as universal, Hume himself would have been forced to conclude on empirical grounds that there was no good reason for accepting the thesis, just as he was forced to conclude that there was no good reason for accepting any other contingent, universal proposition (46f). Hume could no more consistently advocate the sceptical thesis than he could the proposition 'All ravens are black.'
It follows from the necessity of such assessments that there could never be some contingent presupposition, such as the uniformity of nature (23), the long run success rate of inductive inferences (22, 45), or Mill's law of universal causation (45) upon which rested the justification of any particular inductive inference. It also follows on this account that many inductive inferences would be justified, and could be known to be justified, even if no proof of this fact was itself known. After all, although one can often prove a necessary truth to be true, the proposition so proved may still be known independently of the proof. Still less does the proposition itself owe its truth to such a proof (26). Despite all this, Stove believes that in the case of induction the requisite proofs are forthcoming. In fact, Chapters V & VI have as their aim the presentation of three such proofs.
Many of the arguments which Stove presents in this book have been seen before, either in his Probability and Hume's Inductive Scepticism (1973) or in his numerous journal articles. Even so, it is valuable to have them polished and improved and bound together in one volume. Also new is an improved version of D. C. Williams's attempt in The Ground of Induction (1947) at using the law of large numbers to prove induction justified. Most of the second half of the book is a spirited defence of the theory of logical probability upon which the proof rests.
Chapter IX, entitled 'The Myth of Formal Logic,' can and should be read separately from the rest of the book. In this chapter Stove abandons the thesis that either deductive or inductive logic is purely formal. In the latter case, this denial is used as the basis of a solution to Goodman's so-called new riddle of induction. As Stove points out, once the idea that inductive inferences gain their strength from any purely formal characteristics is rejected, the force of Goodman's argument is diminished greatly, and perhaps entirely (78, 132). However, one need not subscribe to Stove's corresponding views concerning deductive logic in order to appreciate his argument with respect to grue or, for that matter, the rest of the book. And, in contrast to the rest of the chapter (which, by Stove's own admission (127), can only charitably be described as radical), Section (ii) is a much needed discussion of the question of whether there can ever exist purely formal judgements of invalidity it should be required reading for every logic instructor who enters the classroom. The following chapter, which addresses the question of whether deductive logic is empirical, is almost as good.
Although the book is tightly argued, Stove still manages to marshal a distinctively large dose of the same type of outrageous, polemical flourish which was characteristic of his recent Popper and After (1982). His clear, crisp and rigorous writing style makes the book a pleasure to read and serves to emphasize the adage that, next to truth, the most important virtue in philosophical writing is clarity. It is an uncommon book that allows one always to be able to point to the exact step in an argument where disagreement occurs. It is an enviable characteristic of this book that nowhere is one left with that vague, annoying feeling that somewhere, somehow, we have unaccountably been led astray on so slippery a topic as induction.
Given the current state of play with respect to induction, this book deserves to become mandatory reading for both the contemporary inductive sceptic and the traditional dogmatist. Friends and foes alike will not be disappointed. The book is vintage Stove.
A. D. Irvine
University of Toronto
&BDb2T AJ>?df]Qh V"j"####;OJQJ6OJQJOJQJ#
;+"###$da$ $da$dd&d(dPR$da$
ST { |
;+"###########hvw0#########$a$$a$&
00P/ =!"#$%
i8@8Normal1$CJOJQJkH'mH F@F Heading 1$$d@&a$5>*OJQJ<A@<Default Paragraph Font,@,Header
!, ,Footer
!R z z z z z3
jT#%##&(#'#AH5
=
;D~DK<FG7]k:::::::
Andrew Irvine:Macintosh HD:Andrew:Publications:Reviews:Stove.87/10 [ROI]
Andrew IRVINE5Aristotle HD:Users:irvine:Desktop:Stove Induction.doc@웲|@STimes New RomanTimes5РtSymbol3ArialSTimesTimes New RomanY New YorkTimes New Roman"Ah3֔F
X4!xx20ms@
Andrew Irvine
Andrew IRVINE
Oh+'0`
(
4@HPX'ososAndrew IrvineondrNormalIAndrew IRVINEo7drMicrosoft Word 10.0@e@NCB@ XcX
՜.+,0hp|
'UBC4m$
Title
!"#$%&'()+,-./013456789;<=>?@ADRoot Entry F(ncF1Table*WordDocument(RSummaryInformation(2DocumentSummaryInformation8:CompObjX FMicrosoft Word DocumentNB6WWord.Document.8