In his new book, Miller returns to his central philosophical interest – to critical rationalism. Readers who are familiar with his previous book Critical Rationalism. A Restatement and Defence (Open Court 1994) know that Miller there reaffirms and further develops Popper's falsificationism and considers it not just a methodological issue relevant to science but a philosophical thesis about rationality. In what new directions does Out of Error take us, given the fact that Critical Rationalism presents a pretty comprehensive account of the most important problems of Popper's methodology, including a systematic enumeration of objections voiced by his critics over the years, followed by their elimination? In this review, I will argue that readers will not be disappointed; Miller both provides new insights to the problems he dealt with before and addresses new problems, especially problems concerning applied science, the demarcation criterion, the use of Popper's rationalism against the fashionable postmodern currents, and the employment of paraconsistent logic in falsificationism.

The book can be divided into three main parts: chapters 1 and 14 were written in memoriam, in the second part (chapters 2–7) Miller carries out a philosophical investigation of critical rationalism, while the third part (chapters 8–13) is more technical and deals with various logical aspects of critical rationalism. I will focus on and discuss mainly the problems of the second part of Out of Error.

The first chapter, Karl Popper: A Scientific Memoir, and the last chapter, In Memoriam, were written after Popper's death and Miller looks at Popper's intellectual development (1) and links to his life and personality (14). Since the (pre)-war period is well documented by Hacohen, readers will probably find more interesting Miller's description of the time that Popper spent in England. Miller, a graduate student at LSE, one of Popper's assistants, his colleague and life-long friend, guides us through Popper's academic and personal life in great detail and with moving sensitivity.

Chapter 2 defines three basic components of critical rationalism that, as Miller stresses, must all be accepted if we want to become truly critically rationalist. The Three Stages of Critical Rationalism are fallibilism (denial of certainty), negativism (exclusively negative method of testing) and scepticism (utter impossibility of ever justifying anything). There is not much trouble with the first (epistemological) stage, but it takes some courage to accept all the implications of the second (methodological) and the third (epistemological), especially in Miller's version, which is even more radical than Popper's. Miller makes it clear that the negative method and the absolute scepticism are non-negotiable and must not and need not be tainted with any inductivist or justificationist flavour; at the same time, radical falsificationism should not be confused with irrationalism and relativism. This is the leading idea throughout Miller's book, only briefly touched upon in this chapter and developed in various directions in the following chapters. I appreciate the emphasis that Miller puts on the negative method as constituting reason; reason is thus defined by its negative action, not by provisions of (good) reasons. “Rationality … is wholly a matter of method” (p. 50), he says, and the “disappearance of the epistemological dimension of rationality is what is least understood in Popper's philosophy” (ibid) and what Miller set out to rehabilitate.

In this chapter, Miller also makes a sharp distinction between falsificationism and justificationism (positivism) in their respective philosophical focus. Critical rationalists are in their empirical investigations concerned primarily about our relation to the world, justificationists are concerned about our relation to theories (about what can do to the theory). Critical rationalism proposes that “theories are unchanged by their interaction with experience” (p. 54) and even falsification of a theory is a decision that “goes beyond the empirical investigation itself” (ibid). I take it that Miller wants to show a change of emphasis in critical rationalism from knowledge to learning, learning itself facilitated by critical investigation of the conclusions of the theory. It is out of error that new problems are born. This message, however, does not come out entirely clearly; perhaps, Miller could have given more space to explaining these issues instead of including in this chapter yet another portion of personal reminiscences.

Miller makes it up to us in Chapter 3 in which he elaborates on the negative, destructive methodology of critical rationalism. The problem in the title of this chapter, What Do Arguments Achieve?, is a serious one and Miller outlines it in the opening paragraph (p. 63): “Arguments are either conclusive or they are inconclusive. Conclusive arguments are deductive arguments, and are circular; inconclusive arguments are non-deductive and are invalid. Circularity and invalidity are bad things. Why, then, bother with arguments? What do arguments achieve?” His reply follows two main lines: 1) arguments achieve nothing if we expect from them some kind of justification of the investigated proposition, and 2) arguments do not matter; “… the total repudiation of all schemes of partial justification, the total repudiation of any attempt to use arguments to support the hypothesis that we adhere to … amounts to saying that it is not the argument that matters, but only the conclusion” (p. 78) and, of course, our determination to show their falsity. Critical arguments criticize and stop there; this is enough to promote the search for truth - not (partially) justified truth.

Miller disqualifies three types of answers, namely persuasion (dogmatic), discovery and justification (cannot be done, for logical reasons sketched in the opening). Miller then explores the “unnourishing pastures of inconclusive justification” (p. 75), (Lipton, Bunge, Musgrave, Pargetter & Bigelow and others) and concludes that the considerations of probability or ‘degrees of rational belief’ or reasonableness make “little impact on the problem of justifying inductive inferences” (ibid). He rejects the criticism blaming critical rationalism for destroying any secure (rational) foundations for the acceptance or the elimination of hypotheses (after all, theories cannot be confirmed or disconfirmed by experience); “… if we are lucky we can sometimes be right (though we never know when)”, says Miller, and wonders why this should surprise justificationists who, too, do not think that the methods of justification are persistently successful.

Also worth noting is Miller's stern warning against linking his denial of the relevance of arguments (that may sound relativistic) to postmodernism, the proponents of which happily revel in relativism and “dismiss objective truth as an uninteresting delusion” (p. 79). This warning taken up in Chapter 6 entitled Back to the Frying Pan. Miller is infuriated by the argument offered by Sokal & Bricmont in Intellectual Impostures (1998) that Popper – in his extremely negative falsificationism—contributes to the rise of irrationalism. Of course, Miller is sympathetic to Sokal and Bricmont's undertaking to mock the fashionable postmodern trends and to expose their pretentious postmodern jargon as gibberish. But they commit, says Miller, exactly the mistake he himself warned against, and proceeds to list the ‘vices’ frequently attributed to Popper (and also cited by Sokal and Bricmont); these are the destructive and too complicated nature of falsification, the absence of good reason, the lack of grounds for reliable predictions. Miller then argues that the alleged vices are in fact virtues, enabling us to formulate “a smarter, leaner, tougher, more wholesome, and less dissolute rationalism than most rationalists are accustomed to” (p. 133).

As much as I agree with Miller in promoting critical (here “radical”) rationalism, I am not impressed with his depiction of postmodernism. He throws all postmodernists (naming just a few criticized by Sokal & Bricmont, and omitting the most representative figures of postmodernism) into one huge sack, without giving any serious consideration either to their ideas or to their differences. Despite not having discussed any of their philosophies he does not hesitate to call them names: fad, ruffians characterized by insulting infantilism or meretricious desipience (pp. 64, 133). Such scorn for postmodernism does not help the case of critical rationalism. It would be more effective to consider that Popper shares with postmodernists a relevant concern about the dominance of the authoritarian rationalism in the Western philosophical tradition; in this tradition, philosophy is seen as a discipline having a privileged access to some indubitable foundations (metaphysical, epistemological, textual) that can legitimate certain discourses or, in Popper's terminology, justify certain knowledge – and thus enable us to make meaning of signs or the truth manifest. Whether we call these philosophical models classical rationalism, logocentrism or the metaphysics of presence does not change the basic fact that they all are united by the need for grounding, this being the primary target of criticism for both postmodernism and Popper. Taking postmodernism more seriously places Popper right in the middle of the contemporary philosophical disputes and strengthens the impact of Popper's solution as an alternative to both the discredited, hollow authoritarian principles endorsed by the old oligarchs of ‘rationalism’ (p. 145) and postmodern relativism.

In Falsifiability: More than a Convention? (Chapter 4), Miller “treads over well-trodden ground” (p. 81) of the problem of demarcation between science and non-science (or, as he prefers it, between what has/does not have empirical content). He takes the view that falsifiability is more than a mere convention but an “irresistible proposal” enabling us to make “sense [of science] as an objective enterprise” (p. 81). He then ties the problem of demarcation to the problem of how we learn from experience and criticizes the inductive models in which “what you know is what you learn (WYKIWYL)” (p. 85). By contrast to the view that learning produces knowledge, Popper's epistemology begins with another question, namely “what we can know through experience”? (p. 83). Miller's answer is, as we might expect, that we learn from experience that our hypotheses are false. Therefore, knowledge (however poor) must come first and a hypothesis must be accepted before it is rejected (or tested). This is the crucial point of this chapter and Miller goes well beyond Popper who sometimes wrote about acceptance in an inductivist sense; for Miller, “testing cancels acceptance, but it does not create it” (p. 125). In other words, “there is no such activity as accepting as true a hypothesis that has already been accepted” (p. 91). If the hypothesis is not falsified nothing happens to it, “all unrefuted hypotheses are epistemologically equivalent” (p. 86). Miller also deals with some objections against this deductive epistemological conception. He rejects the methods of so-called data mining (popular in the sciences), questions the objections typical for naturalism and naturalized epistemology, namely that the history of science does not conform to the conjectures-refutations pattern, discusses the relevance for critical rationalism of Duhem's thesis, and also deals with the problem of existential statements in science. To me, the most interesting issue is Miller's dispute with naturalism, which, as he shows, sets aside justification-centered epistemology but at the cost of resigning on its normative role; a price that we should not and need not pay.

I can fully understand that Miller's firm stand on the priority of acceptance to testing is necessary in order to remove any ‘whiff’ of induction from corroboration. But why is it necessary, in order to maintain the hypothetico-deductivism in methodology (and rationality), to deny that induction might play a role in learning? I am probably missing something important here but it is not clear to me why we could not simply declare that knowledge is logically antecedent to testing regardless of how it is produced; from the rational perspective, all knowledge is purely conjectural and only refutation-attempts are permitted in evaluating it. The imperative of falsification can be maintained as a methodological norm and, simultaneously, as a standard of reason. Once we rid reason of its justificationist role we have a logically consistent theory of rationality (ratio negativa) defining reason by its critical activity; the procedures of knowledge acquisition (or belief formation) should be investigated empirically, not forced to comply to a logical pattern, as Popper suggests in his principle of transference (“what holds in logic … holds in psychology”, Objective Knowledge, Oxford 1972, p. 24).

In Chapter 5, Induction: A Problem Solved, Miller defends the purist deductive position against some objections that blame critical rationalism for not providing a rational criterion of acceptance of theories, for its lack of predictive import and for the implicit inductive assumption of the uniformity of nature. Popper's notorious remarks – that it is rational to prefer the best-tested theory as a basis of action (since it has greater verisimilitude) – are often quoted and several objections are raised: Popper's advice assumes that the future will be like the past (inductivism) and if it does not then the rationality of our preference between theories and their use in the applied science are undermined; science ends up in irrationalism. Miller again rejects this ‘clumsy insistence’ on the need of grounding for rationality, as it was first declared by Lakatos and than further supported in various forms by Worrall, Zahar, Musgrave, O'Hear, Watkins, Shearmur, Salmon and others.

A critical rationalist, says Miller, should not give any advice on what theory we should prefer: “What must not be admitted is the suggestion that a proposal that has been subjected to critical scrutiny, and has survived it, thereby qualifies as a better proposal than one that has not been subjected to critical scrutiny. That would convict deductivism not only of inductivism but of clairvoyance …” (p. 124). Miller recommends a more ascetic approach than Popper did and replaces Popper's positive rule of choice by a negative one: “Refrain from any practical proposal that does not survive critical scrutiny as well as others do” (ibid). This approach involves a radical change of perspective; in critical rationalism one accepts that the process of the growth of knowledge, as well as the process of living, is an open-ended and risky yet a rational business. Its rational core, however, lies purely in its critical method, not in justifying its own activities, achievements or decisions. In other words, the fact that “neither beliefs nor acts of belief, nor decisions, nor even preferences, are reasonable (p. 111) … does not imply that people cannot think or make decisions reasonably or rationally” (p. 112). If an agent accepts spatio-temporally universal generalizations he also accepts that in some respect the future will resemble the past. But “no metaphysical principle of induction is needed to generate such predictions” (p. 115) since accepting a proposition involves accepting as true all its logical consequences – “universal laws do pass judgment on their own future performance, and if we classify them as true, so do we” (p. 116). Yet, Miller continues, “it seems not to be reason, but another decision, that leads us to adopt the best-tested theory as the basis for action” (p. 118). No need to add that this decision contains no warranty for the future: “if [the agent] desires to avoid ill success tomorrow, he can cross his fingers or he can make the effort to be rational today” (p. 119). Miller then discusses the problem of applied science and makes a sharp distinction between science and technology; the latter being concerned with generic plans and practical proposals how to ‘tame’ nature according to our designs. He provocatively says that “for most of their histories science and technology proceed independently of each other” (p. 122) and that “science has very little to contribute to the advance of technology except for criticism” (p. 123). This startling statement certainly needs a more thorough explanation than Miller provides here.

Being an Absolute Sceptic (Chapter 7) was originally published in Science and considers the relevance of critical rationalism to scientists. Miller lists the following ways in which critical rationalism could beneficially influence science: scientists should stop exaggerating the power of scientific rationality and likewise, the public should moderate their expectations of what can be accomplished in science. Further, scientists should be less harsh on their colleagues who make interesting but false conjectures, and be more ready to admit ignorance. If scientists (as well as philosophers of science) cease “to attribute to well-tested hypotheses a security or reliability that they do not posses”– but stop short of a slide into relativism – it is possible to “rescue science as a … rational enterprise par excellence” (p. 150).

The more technical part of Out of Error will now be briefly surveyed. Chapter 8 can be characterized as an investigation within the inductivist tradition, an attempt to see whether good sense can be made of the idea that induction ‘presupposes’ the uniformity of nature, especially in the context of the practice known as ‘assumption spotting’ (p. 69). Chapter 9 deals with the problem that statements are language constructs that may not correspond to facts; it is an external relation. Miller imposes two requirements, namely Tarski's requirement that the language in which truth is defined (the metalanguage) assumes more than the language for which it is defined (the object language)—this is why acceptance of a statement as true goes beyond ‘the statement itself’- and the dualism of facts and decisions – this is why acceptance of a statement as true goes ‘beyond its truth’. In Chapter 10 Miller subjects to rigorous criticism the somewhat unsophisticated proposal of McAllister, endorsed by Kuipers, that theoretical beauty is somehow indicative of increased truthlikeness. The main thesis of Chapter 11 is that the many attempts to defend language-dependent theories of verisimilitude all fail. Miller admits that we must reject all such theories and all theories that tie verisimilitude to syntactic structure, despite the fact that there is not much to put in their place. In Chapter 12 Miller discusses the suggestion that certain Central African peoples (the Azande and the Nuer) adhere to inconsistent systems of belief. He argues that it is not desirable to ascribe paraconsistent thinking to any groups who appear to hold contradictory beliefs, or do not endorse all the consequences of what that they believe, be it the Azande, or the Nuer, or white Westerners. In Chapter 13, Miller develops critical rationalism further beyond the Popperian framework and comes up with the suggestion that we should introduce into falsificationism some kind of paraconsistent logic, specifically a dual-intuitionistic system. To those who think that he lost his grip on philosophical reality by doing exactly what was dismissed by Popper in his criticism of dialectics, Miller replies that we can make fruitful use of paraconsistent logic in the task of understanding the permanent existence of problems (contradictions). He calls this alliance “a perfect marriage of convenience” (p. 250) and explains how the dual-intuitionistic logic arises naturally within a classical falsificationist context.

Out of Error is a thought-provoking book written with clarity and punch. Inevitably, certain key arguments from Critical Rationalism are repeated – and they should be. The principles of radical falsificationism need to re-affirmed for several reasons – and here I speak as Miller's sympathizer. First of all, they are frequently misunderstood and various alliances of critical rationalism with some elements of justificationism are sought and deemed necessary for rationality. Secondly, Popper's ban on justification represents a radical denial of almost all traditional epistemology (perhaps except Socrates) and is hard to digest. As Miller remarks in Critical Rationalism, the majority of intellectuals are still justification junkies to whom “cold turkey is recommended” (p. 49). Thirdly, Popper's ambition was to propose more than just a methodology but a non-foundationist theory of rationality. This philosophical challenge of critical rationalism has not been adequately appreciated in the anglophone world. On all these counts, Out of Error is a valuable contribution to contemporary debates on rationalism, a contribution that might help to revive Popper's philosophical legacy.