Herbert Keuth

The Philosophy of Karl Popper

Herbert Keuth, The Philosophy of Karl Popper, Cambridge University Press, 2004, 384pp, $28.99 (pbk), ISBN 0521548306.

Reviewed by Robert Nola, , University of Auckland

This book is one of the best introductory accounts of Popper's philosophy and is to be recommended. It is wide ranging, covering, in its three parts, Popper's philosophy of science, his social philosophy, and his metaphysics. The summaries of Popper's positions are clear and succinct; relevant critical points raised by others as well as the author are injected appropriately into the discussion. The book reveals that Popper's philosophical concerns are broader than most other twentieth-century philosophers, whatever the critical response may be to his various doctrines in all these fields. In twentieth-century English philosophy perhaps only the concerns of Bertrand Russell surpass those of Popper in their scope.

Part I, which is about half the book, is devoted to Popper's methodology of science beginning with his anti-inductivism and then moving to his views on demarcation, methodological rules for science, the empirical basis of science, corroboration, truth and verisimilitude, the nature of theories, and finally his account of probability. Here some technicalities are unavoidable, but they are minimal and can be managed by most readers. Part II deals with Popper's views on method in the social sciences and a number of themes drawn from The Open Society and it Enemies. Here Keuth reviews Popper's views on Plato, Hegel, Marx, and his critique of the sociology of knowledge and his theory of democracy. There is a brief discussion of Popper's role in the "positivist dispute" and his interaction with other philosophers such as Horkheimer and Habermas. Part III considers Popper's metaphysical views concerning natural necessity, determinism, indeterminism, propensity, mind-body interaction, the doctrine of Worlds 1, 2 and 3, evolution, and the self.

The book does not get mired in the controversies between Popper and his contemporaries, from some of the logical positivists to his erstwhile colleagues or pupils. Nor does it cover some of the later Popper, especially those writings which have appeared posthumously in English but which were sometimes available earlier in German. For example, it does not delve into Popper's later work on the Presocratics. Keuth maintains that while Popper has many considerable achievements to his credit there are also some failures, especially in his late work. He makes a useful comparison with Kant whose claims about "transcendental knowledge" were known to be untenable even before Kant wrote on them. But just as Kant was an important influence in the Enlightenment, so Popper's critical rationalism is part of the twentieth-century continuation of that tradition. For German speaking Europeans, given their philosophical and political history since the time of Kant, the role of philosophers like Popper in the general intellectual culture is very important, a fact not often appreciated by those outside Germany and Austria.

Part I on Popper's philosophy of science opens with a discussion of Popper's first big manuscript Die beiden Grundprobleme der Erkenntnistheorie, only first published in German in 1979 but still not available in English. The two fundamental problems are what Popper calls 'Hume's problem', the problem of induction, and 'Kant's problem', the problem of demarcation. These are pervasive themes in Popper's subsequent work which Keuth sets in context. Only the most diehard Popperians now adopt his anti-inductivism in all its aspects. But what of demarcation? This too has fallen on hard times, but perhaps not with equally good reason. Though papers like Laudan's 'The Demise of the Demarcation Problem' are not mentioned, there are materials in Keuth's discussion of demarcation that show how the idea might at least be partially rehabilitated. Objections due to Kuhn, Kneale, and Grover Maxwell along with the "tacking paradox" (viz., if T is falsifiable them so is (T&X) where X is any arbitrary conjunct) are outlined, and ways around these objections are suggested. If no bright red demarcation line is to be drawn (perhaps never Popper's intention) then at least a distinction can be drawn on the grounds that scientific theories must at least be brought into relation with the stream of actual observations of the world, however minimal that relation may be; in contrast, for other intellectual endeavours (logic, mathematics or metaphysics), no such relationship need be required.

As well as the logico-epistemological demarcation criterion that Popper proposes as part of his epistemology of science, there are also a number of methodological rules that spell out more fully other aspects of his "definition" of science. This includes a rule which bids us to adopt his demarcation criterion as part of the goal of science. This in turn realises a further goal, that of the maximization of the openness of science to revision by test against observation. This is a distinctive feature of science which the young Popper recognised when it abandoned one of its best-established theories, Newtonian mechanics. But Popper's conception of science is not just the demarcation criterion; it is also to be understood as an activity bound by a further range of rules. A conception of pseudo-science then arises; these are systems of belief that do not evolve in accordance with the rules. Keuth lists at least twelve such rules in chapters 3 and 5. Oddly enough some of these rules are not given rule-like formulations by Popper (for example R1 and R5 cited in section 3.3); but these can be expressed as rules without too much difficulty.

Keuth comments that some of the rules are fairly trivial and can be readily accepted as definitive of science, but some other rules are problematic. For example, Popper's anti-ad hoc Rule (given as R8) attempts to expresses an initially plausible stricture against certain modifications of our theories in the face of counter-evidence. Often Popper treats such rules as rigid categorical imperatives rather than the defeasible principles they really are. However it is their expression in terms of increasing degree of falsifiability, or of testability, that raises problems. The requirement is that any modifications to a theory ought always to increase its degree of falsifiability. However, the problem here is that it is hard to find measures of degrees of falsifiability on the basis of which one can compare a theory before and after modification. Keuth also argues that there are some auxiliary statements that we might alter to save a theory but these need not always be those which are conjoined to a theory to yield testable consequences.

A commonly cited obstacle to Popperian falsification is said to be the Quine-Duhem thesis in one or other of its several forms. This is something which, as Keuth points out, Popper recognised in The Logic of Scientific Discovery (section 18) when he confessed that we falsify a whole system and that no single statement is upset by the falsification. Popper seems to pay little further attention to the problem of how falsification, or even corroboration for that matter, may arise by piercing through any surrounding accompanying statements, to target a hypothesis under test. However there is one out-of-the-way place where Popper does acknowledge Duhem's problem and suggests a solution, The Poverty of Historicism section 29, the end of footnote 2. Keuth does make some moves that are similar to Popper's suggestion; but it would have been useful to have had Keuth's commentary on Popper's explicit proposal, which has initial plausibility, viz., that we can construct a version of a "crucial experiment" and comparatively test two whole systems, one with and the other without some given hypothesis which is under test. (Glymour 1980 p. 34 has one negative response to this ploy.) Independently of Popper's problem with the Duhem-Quine thesis, that very thesis has come under criticism by Bayesians who argue that testing of target hypotheses is possible despite their being embedded in auxiliary hypotheses, thus freeing theories for Popperian falsification, but in an un-Popperian manner.

Keuth also considers Popper's falsificationism in relation to Quine's 'Two Dogmas of Empiricism' and the view, which flows from the rejection of the dogmas, that ''the boundary between speculative metaphysics and natural science is blurred" (p. 79) If correct, this would pose a challenge to Popper's demarcation criterion in that all our systems of belief form a seamless web without the possibility of even a somewhat dull red demarcation line being drawn. One dogma is the view that there is a difference between analytic and synthetic statements; the other that there is a reduction of each meaningful statement to a construction out of terms which refer to immediate experience. Popper would concur with the rejection of the second dogma but perhaps reject the inference from this to the claim that the unit of significance is the whole of science (as Quine claims). It turns out that Quine is not just a holist about meaning (a matter which Popper makes clear is not one of his concerns) but also a holist with respect to the way in which the whole of science gets positive or negative support. In part this takes us back to the credentials of the Quine-Duhem thesis and theories of confirmation and disconfirmation. Keuth focuses on the first dogma and argues that since we can establish logical truths and propose definitions, there are aspects of each scientific theory in which something akin to the analytic/synthetic distinction is not only possible but appropriate. Once this is spelled out in the context of a given theory, then there are grounds for distinguishing such logical truths and definitions from other aspects of a theory which, still taken as a whole, can then be open to test; and if this is so, something akin to demarcation is also appropriate.

In chapter 6 Keuth traces Popper's views on truth from his initial position, expressed in The Logic of Discovery, that it is possible to do without the concepts of truth and falsity, to his later position in which he adopts Tarski's theory of truth. But Popper's account of truth embraces more than the set of bare equivalences given by the various Tarskian (T) schemas 'X' is true iff p (where 'p' is a sentence and 'X' is a name of the sentence), which constitute an adequacy condition imposed on any theory of truth. Rather, Popper understands Tarski to have embraced a correspondence theory of truth in which the Correspondence schema (CT) holds: 'X' is true iff X corresponds to the facts. Keuth's treatment of Popper's account of truth is instructive and shows that Popper's version of his 'Tarskian turn' is in fact down a wrong, even blind, alley. His diagnosis is that Popper's commitment to a realist understanding of truth, especially the truth of the statements of our theories, has taken him in this direction; but, as Keuth points out, it is possible to have a realist understanding of the truth of such statements without the correspondence theory of truth.

In the following chapter 7 Keuth reviews the role of truth as a regulative principle in Popper's philosophy of science and his development of the idea of verisimilitude as a further articulation of that ideal. He also traces the development of Popper's more formal treatment of verisimilitude in terms of the comparison of the truth and falsity contents of statements. As is well known, these attempts ended in a spectacular failure, which Keuth sets out. Not explored, and appropriately so in a book just confined to Popper, is the ongoing research programme that this failure stimulated which, with the successes it has had, does keep alive aspects of Popper's important initial insight.

In Part I Keuth has further useful suggestions to make about Popper's rather decisionist and conventionalist view of the acceptance of basic statements which we use to criticise our theories -- a difficulty which many have raised about the rationality of Popper's whole falsificationist approach. He also discusses the development of Popper's ideas of probability. And he has a useful chapter on the development of Popper's idea of positive support, that is, corroboration, and whether or not corroboration can avoid any aspect of inductivism. Keuth finds fault with Popper's more formal attempt at a definition of corroboration. Even though corroboration is not obviously a probability, it is defined in terms of probabilities, including prior probabilities which, notoriously, are hard to define and so render any quantitative definition useless. However Keuth does defend the view, against many others, that corroboration does not need any associated principle of induction. When it comes to a choice of one of a set of theories as to the correctness, say, tomorrow, of their explanations, predictions, or technological applications, we choose that theory, T, which is the best corroborated up until today. Against the claim of many that there is an inductive step involved here, Keuth argues that no inductive principle is needed. All we need assume is that theory T is true; if T is subsequently falsified then our assumption was false. Induction is not needed, and we can get by without it. But the claim that it is not needed is not convincing. There does appear to be an element of induction in our keeping T on tomorrow, and the day after, when it does not succumb to falsification.

Most of this review has focused on Keuth's account of Popper's philosophy of science. Whatever shortcomings Popper's view may have, it, or a revised version of it, has been influential in providing a critique of theories of society. This critique Popper began in his The Poverty of Historicism and his The Open Society and its Enemies. Popper claims that many historicists such as Marx and Marxists, and also many sociologists of knowledge, have a faulty understanding of the nature of science and its methods. Once they have a richer account of those methods, the bottom falls out of many of their arguments about the separate nature of the social sciences. This, amongst others, is a theme that appears in Part II of Keuth's book; it can be extended to objections concerning an alleged misplaced scientism advocated in the writings of members of the Frankfurt School, such as Horkheimer and Habermas. Keuth's book is not only a useful account of Popper's views on science, along with promising revisions of these views; it is also a useful prophylactic when applied to some sciences, especially the social sciences. Keuth's treatment of these issues in Part II, and the metaphysical issues in Part III not mentioned here, continue the standard found in Part I. Taken together they provide a fully rounded and fair assessment of Popper's philosophy and its continuing interest.


Glymour, C (1980) Theory and Evidence, Princeton NJ, Princeton University Press.