1 Introduction: a compliment, a confession and a criticism

Let me start with a compliment and a confession before going on to a gripe and a comment (that is to a musing on ‘What This Book Made Me Think’). Though I intend to go on a bit about the gripe, I don’t want my complaint to take away from the compliment which should ring out loud and clear: The Analytic Tradition in Philosophy (henceforward ATP1) is an excellent successor to an excellent book. Essentially it is a much expanded rewrite of the first half of Scott Soames’ Philosophical Analysis in the Twentieth Century volume 1 (henceforward PATC1) minus the sections on Wittgenstein, Ross, Quine and the Logical Positivists but with two chapters on Frege, a new chapter on ‘Becoming G. E Moore’ and extra chapters and subsections on Russell. The chapters of Frege are good, though you get the distinct impression that Soames does not warm to Frege as much as he does to Moore and Russell. I think he would have done a little better to start with the Foundations of Arithmetic, which is not only a brilliant work of philosophy but also quite funny, rather than getting stuck into the logic straight away with a rational reconstruction of Frege’s system. But then, it is abundantly clear that, in comparison to Soames, I am a frivolous fellow in constant need of sugar to sweeten the pill of serious philosophical analysis, and those made of sterner stuff may find his approach refreshing. The chapters on Frege are mercifully free of the hosannahs that commonly glorify the name of Michael Dummett, who to my mind distorted Frege studies by reading Frege through post-Wittgensteinian spectacles. Instead Soames prefers Gregory Currie as a commentator (with a ratio of three references to one), emphasizing Frege’s epistemic concerns. Frege was, of course, a philosopher of language, but he was a philosopher of language because he was a philosopher of logic and he was a philosopher of logic because he was a philosopher of mathematics who wanted to reduce mathematics to logic, something that could not be done without inventing the appropriate logic and meditating a bit about sense and reference. Perhaps it would have been a good idea to have a chapter ‘On Becoming Gottlob Frege’, but whatever led Frege become Frege (I confess to horrible ignorance on this point), it is clear that it left him with the project of proving that arithmetic is analytic, which is surely an epistemic ambition. The book also contains chapters and subsections on the early Russell (including his relations with Sidgwick), on Russell’s meta-ethics (and his interactions with Moore), on The Principles of Mathematics, and on Russell on truth and judgment, including an analysis of the Russell/Moore polemics against the pragmatism of William James. This last is useful not only because it will remind the philosophical community of Russell’s excellent arguments against pragmatism (I would say ‘his devastating refutations’) but because it helps to answer a question that tends to puzzle students: ‘Analytic Philosophy as opposed to what?’ It is clear that part of the answer is ‘Analytic Philosophy as opposed to neo-Hegelianism’ but Analytic Philosophy also defined itself in opposition to both pragmatism and the philosophy of Bergson. (Bergson and Russell’s critique of Bergson are sadly missing from this book but then you can’t have everything!) Although one would not want to prescribe ATP1 by itself as an undergraduate teaching text (it would be way above the heads of any but the best students), there are parts of it that can be used in conjunction with PATC1 to illuminate not only the neo-Hegelian orthodoxy that Russell and Moore were reacting against but their debates with other anti-Hegelians. (I am in fact prescribing certain sections for the 200/300 level PATC1-based course on the History of Analytic Philosophy that I teach at Otago most years). The Russell-Moore revolt against Anglo-Hegelianism emerges as a much more protracted and messy process than a reader of PATC1 might have supposed, in which both Russell and Moore traversed a range of positions only slightly less weird and wonderful than the idealistic monism that they ostentatiously rejected. (Both Russell and Moore in autobiographical reminiscence, tended to telescope events and to represent their earlier selves as a lot more commonsensical than they really were. To some extent Soames went along with them in this in PATC1, an excusable defect that has now been corrected). It is common to praise books as ‘packed with argument’ (often with the conversational implicature that they are a bit hard going) but Soames packs in more argument per square inch of print than almost any other philosopher that I have read. Since ATP1 weighs in at 632 as opposed to the 405 pages of PATC1, with those 632 pages set in a much smaller type, it is a book which covers a lot less ground but in a lot more argumentative detail. This has two important consequences. Firstly, anything but a very extended commentary is bound to be superficial and selective. (That’s a kind of an excuse for the deficiencies of the present piece!) The book is so rich, and specifically so ‘packed with argument’ that there is plenty to keep us going for paper after paper. Secondly, although ATP1 is, in a way, the same sort of book as its predecessor, PATC1, it is addressed to a different and more sophisticated audience and should therefore be judged by different and more sophisticated standards. This means that what were pardonable, even necessary, omissions and simplifications in the first book may have metamorphosed into oversimplifications and sins of omission in the second. My principle gripe is with one such simplification (or omission) which has been allowed to persist from the first book to the second and has consequently undergone just such a metamorphosis.

That’s a first stab at the compliment and a hint at the gripe but what about the confession? Well, I might as well admit right now that I am one of the referees for Princeton that Soames so generously thanks in the Preface. Why is this a confession rather than a boast? Because Soames has accepted so many of my suggestions (and with such gracious thanks) that it seems a bit churlish to complain about the one suggestion that he did not accept (or accepted only to a very limited degree). But that, I am afraid, is what I am going to do. There is something wrong with APT1 that people need to be aware of, whether they are using it for purposes of private study or as a teaching text in graduate or (perhaps) advanced undergraduate classes. That way, they can correct for it in their capacities as teachers or work around it in their capacities as readers.

2 The compliment in context

Soames's earlier book on which this one is based, Philosophical Analysis in the Twentieth Century was an important contribution to philosophy itself, an important contribution to the history of philosophy and an important contribution to the teaching of philosophy. Even though I had read and pondered most of the philosophers that he talks about (and could even claim to be an expert on some of them), I learned an enormous amount from reading the twin volumes. To take a simple example, I had known about Grice for many years and had even used some of his ideas about conversational implicature in my published writings. But I totally failed to see the larger significance of his work until I read Soames’s chapter on Grice in the second volume, PATC2. The Oxford philosophers of the fifties (as well some of their American allies, such as Malcolm), endeavored to map the contours of philosophically interesting concepts by working out what a chap would be inclined to say under various circumstances. Thus because a chap would not ordinarily say that he knew certain basic facts (such as the fact that he had two hands and was a living breathing being) they thought that the concept of knowledge did not apply to such facts (and that it was somehow senseless to say that you knew them). But if what a chap would be inclined to say is determined not just by the concepts he employs but by certain conversational conventions (such as not stating the obvious), then the fact that a chap would not normally be inclined to say that he knew that he had two hands does not imply that he does not know that he has two hands (let alone that it would be senseless to say so). The fact that we have a conversational convention against stating the obvious does not imply that there are no obvious truths to state, including obvious truths about what we do and do not know. More generally, mapping the (ordinary) uses of a word is not a direct route to its meaning if those uses are partly determined by the conversational conventions that are in play. Thus Grice’s work represents an internal critique of the Oxford paradigm. Furthermore, this is not a point of merely historical interest, since much the same criticisms can be made of some of the methods of the currently fashionable X-Phi movement. The Oxford philosophers thought that they could map the contours of various problematic concepts by working out what a (rather well-educated) chap would be inclined to say in various circumstances. Not so, if what that chap would say is partly determined by something other than the concepts he employs. Some of the X-Phi crowd seem to think that they can map out the contours of various problematic concepts by asking the folk (or at least captive undergraduates) what they would be inclined to say in various circumstances. Not so, if what the folk would say is partially determined by factors other than the concepts that they employ, such as conversational conventions or even extraneous considerations such as the order in which the queries are posed on a questionnaire. Perhaps all this should have been obvious to me, but it was not obvious until Soames pointed it out. Afterwards, I could have kicked myself for two decades of stupidity, but it was Soames’s book that made me less stupid. It is some consolation to know that I am not the only one of whom this is true. Quite a number of famous philosophers have testified to the illuminating effects that Soames’s earlier books have had on them. If a veteran like the late Jack Smart, who lived through much of the history recounted in Soames book, could feel that there was something to be learned from the two volumes, it is not perhaps so surprising that they were enormously useful to a raw youth in his fifties such as myself.

The first book is also a tremendous resource for teachers. I use the first volume PATC1 to teach a course on early analytic philosophy to second and third year undergraduates, something I would never have dared to do without the aid of a really strong textbook enabling them to make sense of some rather testing material. So PATC and especially PATC1 were not only gifts to me as a professional philosopher, but gifts to me as a teacher of philosophy, enabling me to introduce a new generation to the ideas that led me to fall in love with the subject 38 years ago. (‘Why don’t you just introduce themes to the class using the original texts? After all, isn’t that how you learned about the subject?’ Well, this maybe all right at a highly selective university such as Oxford or Cambridge—which is where I studied as an undergraduate—but if the bulk of your undergraduates are merely good as opposed to academically tip-top, working from the original texts is a counsel of perfection that is likely to lead to a pedagogic disaster).

To repeat: PATC1, the earlier book on which this new book is based, is a very good book, a major contribution to the history of philosophy, to the teaching of philosophy and to philosophy itself. This new book, ATP1, is in many ways an improvement on PATC1 with a great deal of fascinating new material. It follows that ATP1 is a very good book indeed. Nonetheless, it suffers from one serious defect that readers and users need to be aware of.

3 The criticism in context

First to establish some common ground. ATP1, like the earlier book on which it is based, is intended is a contribution to the history of philosophy. But the history of philosophy is a borderline activity. Is it a contribution to philosophy or a contribution to history? The short answer is that it can be either or both. But there is a big difference in how we approach the subject, depending on whether our interests are primarily philosophical or primarily historical.

History is about human endeavour, about more-or-less rational agents operating in a problem situation, and about the unintended consequences of their actions and interactions. And the theoretical problems of philosophy are just as real (if not always as momentous) as the practical problems of politics or of military campaigns. Thus philosophy has a history which, to some people at least, is as interesting as the history of other kinds of endeavour. If we are doing the history of philosophy as a contribution to history, then byways and blunders can be quite important. The object of the exercise is to tell it like it was, warts and all, not to concentrate on those ideas that we think are going to be helpful in getting at the philosophic truth. But it is different if we are doing the history of philosophy as a contribution to philosophy. In that case the history of philosophy is transformed into a necromantic art. The idea is to revive the great dead (and sometimes the not-so-great dead) in order to get into an argument with them. And the reason is that we think that they are importantly right or interestingly wrong. (The trivially right and the boringly wrong are generally allowed to slumber in their graves unless they attract the attentions of a history-for-history’s sake scholar). The object of the exercise is to pick the brains of the deceased about matters of current concern (which is not to preclude the possibility that one of the benefits that we might derive from such a brain-picking exercise is the conclusion that some of the issues of current concern are, in some sense, the wrong issues!). This means that when we are doing this kind of history of philosophy, there is a lot we don’t need to know about the great dead. There are blind alleys that do not need to be explored and ideas that are so hopelessly flawed that nothing is to be learned from discussing them.

But though ATP1 and PATC1, are both intended primarily as contributions to philosophy rather than as contributions to history, the conversations that Soames is trying to stage in the two books are rather different. With PATC1, they are ‘getting to know you’ conversations, where it is OK if the acquaintance is going to be a relatively superficial affair. The book is aimed at undergraduates who might not be going on at all, and at graduate students who might well specialize in areas of philosophy other than those discussed by Russell, Moore and Wittgenstein (at least in their analytic incarnations) such as political philosophy or the philosophy of biology. Thus the conversations are the ones that you need to have with the giants of analytic philosophy if you aspire to philosophical competence, not the conversations that you need to have if you want to squeeze out every last drop that can be learned from talking to them. The book tells people what (in Soames’s opinion) they really need to know about the mighty dead of analytical philosophy in order to be philosophically literate.

With ATP1 it is otherwise. There the conversations are more in the nature of in-depth discussions. The book is aimed at people who are focused on the issues raised by the early analytic philosophers and are going to be worrying about this stuff for a large part of their professional lives.

With PATC1, simplifications—what a pedantic historian of Analytic Philosophy, anxious to protect his intellectual bailiwick, might regard as over-simplifications—are perfectly OK, and even de rigeur. There is no way you can teach (for example) the Ramified Theory of Types to a bunch of undergraduates, and no need to teach it to a bunch of graduate students, most of whom can expect to be working in (relatively) unrelated areas. But given the project of ATP1, if we over-simplify we risk missing out on what we might learn from an encounter with the mighty dead. If the readers don’t know about the Ramified Theory, there is a lot about Russell and a lot about the subsequent history that they won’t understand. (Why on earth is Russell rabbiting on about the Liar? What’s the big deal with Ramsey? Isn’t his theory just a minor variant of Russell’s? And why is the Liar an issue for Tarski? And what about this stuff about ‘the Axiom of Reducibility’ in Gödel’s incompleteness paper?) Furthermore, the readers won’t be equipped to address the question [once again a hot topic, because of the work of Priest (2002) and (2006)] of whether the semantic and the logical paradoxes ought to have a unified solution. If they don’t know (enough) about the Ramified Theory of Types, then readers won’t know whether its invention was importantly right or interestingly wrong (and in my view it was definitely one or the other!).

The same goes for sins of omission. There are things that it was right to leave out in PATC1 that it would be wrong to leave out in the expanded work. In a ‘getting to know you’ conversation, you don’t expect your conversational partner to tell you everything that you might want to learn from an in-depth discussion. It is different if you are really trying hard to pick someone’s brains. You want to have access to everything in those brains that is worth the picking (and when I talk about what is worth picking I mainly mean what is worth picking from a philosophical rather than a historical point of view).

4 The criticism in detail: the Ramified Theory of Types

This brings me to my gripe which I have already prefigured. In ATP1, there is an important over-simplification that should have been de-simplified and a sin of omission that ought to have been repaired. What the book should have had is a detailed account of the Ramified as well as the Simple Theory of Types together with an explanation of why Russell felt it necessary to deal with the Liar and the other ‘semantic’ paradoxes and to develop the ramifications. There are a few perfunctory pages, it is true (introduced, I suspect, so as to meet my earlier criticisms) but what is certainly lacking is detailed discussion of the theory and an account of how it is supposed to work. As it is, studious people will turn from Soames to the original sources (such as the Introduction to Principia or Russell’s 1908 paper) only to be totally flummoxed (a) by all this material about the Liar and the other semantic paradoxes and (b) by the baroque complexities of the Ramified Theory itself. From a historical point of view, it becomes utterly mysterious why Russell should have wrestled with the paradoxes for 5 years when he had already written up and published a version of the Simple Theory of Types in the Principles of Mathematics in 1903. [See Russell (2000) ch. 6 for the agonies that he went through, struggling with Liar.] Now of course, this is history of philosophy for philosophy’s sake not history of philosophy for history’s sake, but even so, there is something to be said against an analysis that makes total nonsense of the historical and biographical record. However, the ramified theory is important for philosophical reasons. These are hinted at by Soames on pages 534, 616 and 617 (the only pages on which the Liar rates a mention) but he does not say anywhere near enough. Russell was the first modern philosopher to address an issue which is nowadays very much a hot topic—that once you combine semantic notions such as Truth with classical logic, the threat of paradox rears its ugly head. Since the issue is discussed by Tarski (1983), Beall (2009), Field (2008), Priest (2002, 2006), Kripke (2011), and even gets an airing in Soames own book Understanding Truth (1999), I hardly need to belabour the point. But since Russell was the first major figure in modern times to address the issue and to appreciate its importance, and since the object of the exercise is to learn from the mighty dead, I think that much more should have been said about Russell’s efforts to deal with the problem. Furthermore Russell’s theory exhibits the besetting sin that plagues most attempts to solve (or dissolve) the Liar and the other semantic paradoxes. The Ramified Theory of Types tries to exclude the paradoxes by ruling out a range of potentially paradoxical expressions as meaningless. But among the expressions that it rules out as meaningless are those that are necessary to formulate the theory. (Again Soames alludes to the problem but he does not say anywhere near enough.) Thus if the theory is true, it is meaningless and hence not true. Since it is plainly not meaningless, it is false. Soames employs this criticism against the Tractatus in PATC1, but it is perhaps worth remarking (and indeed worth emphasizing) that the Ramified Theory of Types is the first modern theory to suffer from this catastrophic defect. (The Simple Theory of Types is also a theory that tries to deal with the threat of paradoxes by excluding certain expressions as meaningless but these do not include the expressions needed to formulate the theory. With the Simple Theory, the difficulty is that the allegedly meaningless expressions are obviously otherwise. Thus the theory is false but not self-refuting.) Furthermore, the Ramified Theory suffers from another defect reduplicated in many of its successors. If it is true, it irrational to believe it. For (a) it is not self-evident and thus requires an argument if it to be rationally believed, but (b) any attempt to argue for the theory would involve using expressions that are meaningless, and which, since they are meaningless, cannot be employed in a sound (or even a cogent) argument. Thus the theory is such that if it is true it is not rational to believe it (and if it is rational to believe it, it is not true), a problem that also afflicts the Tractatus.

A possible objection is that the Ramified Theory is just too difficult to explain. And it is certainly true that Russell’s star student Copi did a pretty poor job in his little book on the subject. (I remember struggling with it during the vacation when I was an undergraduate at Cambridge. I was thrown by some flat out errors in Copi’s exposition which I naively supposed were down to me). However I have been convinced to the contrary by Michael Potter in his book Wittgenstein’s Notes on Logic (2009), who manages to do a decent job explaining the Ramified Theory in a few pages as do Ongley and Carey in Russell: a Guide to the Perplexed (2013). So it is not a task beyond the reach of human ingenuity. Thus the moral is caveat emptor. There is one important aspect of the work of one of the ‘Founding Giants’ that is not adequately treated in this book. When it comes to Russell’s logic, those using APT1 as an advanced guide to the history of analytic philosophy, whether for personal instruction or as a teaching text, will need to supplement the book with other material.

5 Concluding reflection

Now for a brief Soames-inspired reflection before signing off. Reading APT1, it became clear to me that both Russell and Moore (and maybe even Frege) face a serious but unacknowledged problem. This is largely because of their addiction to Open Question-style arguments, which they deploy not only in ethics but also in their battles with the pragmatists and neo-Hegelians. They often want to argue that other peoples’ analyses of this or that are incorrect. If a sentence B is supposed to constitute analysis of another sentence A, it is sometimes possible argue that B is not synonymous with A because A could be true and B false or vice versa. Thus ‘It pays to believe X’ is not synonymous with ‘X is true’ since there are some falsehoods that it pays to believe and some truths that it doesn’t pay to believe. Hence the one cannot constitute an analysis of the other. But what about those cases where A and B are logically equivalent (that is true or false under the same circumstances) or even necessarily equivalent (as might be the case if both A and B were necessarily true)? How can you argue that in those cases the analyses are incorrect? What Moore and Russell tend to do is argue that A is not synonymous with B if a competent speaker, Sally, could believe A but doubt B or could believe A but without being constrained to accept B purely in virtue of her linguistic competence. That’s why they both think that ‘good’ does not mean ‘X’ for any naturalistic of metaphysical ‘X’ (though they also deploy similar arguments in other contexts). But now we have a problem. Moore thinks he can know material-object statements, for instance the proposition that he has two hands even though he does not know how ‘I have two hands’ is to be analysed. But if this is the case, and if we accept the Open Question Principle that no two sentences are synonymous unless this is obvious on reflection to competent speakers in virtue of their linguistic competence, then it surely follows that no purported analysis of ‘I have two hands’ can be correct. For no analysis can meet the constraint that no competent speaker could doubt it (since Moore himself is a competent speaker!). Unanalysablity threatens to run riot. But the problem is a lot worse for Russell. Consider the elaborate counterfactual analyses of material-object statements couched in the language of sense-data that are hinted at (though not of course fully developed) in and Our Knowledge the External World and The Philosophy of Logical Atomism. Could a competent speaker believe that ‘I have two hands’ whilst doubting Russell’s analysis of that statement (whatever the Hell it is)? The answer is obviously ‘Yes’. Hence Russell’s analysis (whatever it is) is incorrect. Much the same goes for Russell’s more precisely formulated analysis of the ‘occasionally useful’ proposition that 1 + 1 = 2. Linguistic competence alone is not going to take you from the belief 1 + 1 = 2 to a belief in Russell’s elaborate equivalent. Thus both Russell and Moore unwittingly subscribed to a criterion for synonymy, and hence for the correctness of conceptual analyses, which meant that most of their own analyses were false. Is it possible to devise a criterion for the correctness of conceptual analyses that does down the analyses they did not like without subverting the analyses that they preferred? It’s an open question. But what this shows is that despite the massive merits of their individual books and papers, the larger programme of the Founding Giants of Analytic Philosophy was in deep trouble almost from the world ‘Go’.

There is more, much more, to say about all this and about Soames’ stimulating book but I have already reached my word limit, so it is time to stop.