Jon Elster needs, as they say, no introduction to regular readers of the London Review, who will be familiar not only with his name but also with the cast of his mind and the breadth of his interests. What makes him distinctive is a combination of philosophical acuity and detailed attention to contemporary work in history, the social sciences and cognitive psychology. He sets out his own credo in the ‘General Introduction’ to Explaining Technical Change:
I am not out to tell economists or historians how to do their work. I believe, however, that philosophers of science can be of help in distinguishing true from spurious foci of disagreement within the empirical disciplines. Empirical work conducted in isolation from the philosophy of science may be no worse for that, whereas the philosophy of science atrophies if it is not in close and constant touch with the development of current thinking on empirical matters. Yet the asymmetry is not so radical as to make philosophy of science totally parasitic.
Perhaps this sounds pretty anodyne. But until recently philosophers who dabbled in history or social science usually started with the assumption that they already knew what constituted a valid form of explanation (drawn from some crude approximation to the practice of physics) and did not relate their ideas to the actual work of historians or social scientists. Either they made up simple examples that fitted the theory but bore little relation to what went on in the subject as practised, or some example, once introduced from the literature, was passed along from one philosopher to another, becoming increasingly stylised in the process. (‘If I hear about witchcraft among the bloody Azande one more time I shall scream,’ a sorely-tried Oxford colleague said to me some years ago.) Elster avoids these pitfalls. He has done his homework on the original sources and then thought a lot about what he has learned.
This is not to say that he is a paragon of all the authorial virtues. Both books are, I think, made harder going than they need be by Elster’s habit of laying down a dense barrage of distinctions and cross-classifications before doing anything much with any of them, and even then not doing anything with a great many of them. Explaining Technical Change consists of two parts of roughly equal length, one entitled ‘Modes of Scientific Explanation’ and the other ‘Theories of Technical Change’. The first part is a synoptic essay which can be read on its own and which brings together views about explanation that Elster has developed separately in a number of other places (including Sour Grapes). It is organised around the question of the subject areas (physical, biological and social) to which three modes of explanation (causal, functional and intentional) are appropriate, and contains in compressed form statements of all of Elster’s characteristic positions.
The most controversial of these (though not with me) is that functional explanation (roughly speaking, explaining the existence of something by pointing to its beneficial or survival-enhancing consequences) is, as it stands, no explanation at all. It can become one against an appropriate background, so that we can see how the consequences somehow feed back to produce the outcome. (The natural selection of genotypes on the basis of ‘inclusive fitness’ is, of course, the paradigm.) But the more the story is filled in, and the tighter the explanation gets, the more it becomes clear that a functional explanation, if successful as an explanation, is a kind of causal explanation. To the extent that a proposed functional explanation does not have the causal loop in place, it is simply a promissory note drawn on a causal account.
It might perhaps be thought that this dismissal of functional explanation (if we understand by that explanations that cite good consequences without supplying a feedback loop) smacks too much of the old-style philosophical police force, and Elster himself talks of ‘my argument – or should I say diatribe – against functional explanation of the more unreflective kind’. But I doubt whether anyone would ever have believed that anything could be explained by pointing to its consequences except as a result of having already swallowed some half-baked theory: either Hegelian teleology (or the aspect of Marx’s theory of history that unfortunately rests on Hegel) or the misinterpreted lessons of biology. So it seems legitimate for the philosophical cops to arrest the philosophical robbers who (as is the nature of functional explanation) seek to enjoy the advantages of theft over honest toil.
What shows Elster to his best advantage in this part of the book, however, is his frequent recourse to a method that might be called inductive methodology. By this I mean that he starts out with the idea that some writers are good at giving explanations and then tries to see how they do it. An excellent example of this comes in his discussion of social causation, where he draws from de Tocqueville’s Democracy in America four ‘basic principles’. The first is the ‘principle of holism’ – that what is true of exceptions may not be capable of generalisation to a whole society: ‘For example, even if love marriages tend to be unhappy in societies where they form the exception, one cannot conclude that they will also be unhappy in democratic societies where they form the rule. To marry for love in an aristocratic society is to court disaster, since going against the current tends to create hostility and in turn bitterness. Moreover, only opinionated persons will go against the current in the first place – and being opinionated is not a character feature conducive to a happy marriage.’ This is scarcely earth-shattering (and the same may be said of the others), but it is the kind of reminder that is genuinely useful and as far removed as could be from the notion, manifestly absurd to anyone who has ever seriously tried to explain a complex social phenomenon, that it’s all a matter of deducing it from some universal law.
The structural problem with the book is that, admirable as the first half is in its own right, it places a formidable roadblock in the path of anyone whose primary interest is in explaining technical change. For it is hard (or, speaking for myself, impossible) to keep it all in one’s head. Yet the reader is given no clue as to what points are going to be important later, and is liable therefore to engage in the self-defeating task of trying to keep a grip on all the twists and turns of the discussion.
In the end, Part II is not dependent on a lot of the detailed analysis offered in Part I. Elster does indeed sometimes refer back to it but I think that anyone who wants to read him on technical change would do best to plunge straight in and refer back to specific points in the earlier discussion when they are mentioned. Elster’s official story, as contained in his preface, runs as follows: ‘Part II is not to be read as an introduction to the theory of technical change. It is subordinated to the epistemological purpose of showing how the distinctions and propositions of Part I can be applied to a specific set of empirical problems.’ I think it is true to say that Elster’s interest in the theories he discusses is methodological rather than substantive. The kind of question he asks is: what is this theory trying to explain? Is it the right kind of theory to do it? Is it internally consistent? He does not, for example, bring to bear systematic evidence of his own to support or confute the theories.
But although the focus may truly be said to be methodological, the discussion is not in fact organised at all tightly around Part I. What we in fact get are expositions and fairly free-wheeling commentaries on four theories or types of theory: the ‘neo-classical theories’ of standard economics; Schumpeter’s theory, with its stress on the innovative role of the entrepreneur; a job lot of ‘evolutionary theories’; and ‘Marxist theories’ – mainly in fact Marx’s Capital. There is no concluding chapter, and it is hard to know what one would contain, since there are no conclusions. Elster’s individual discussions are, however, full of good things.
In Sour Grapes, the first chapter, entitled ‘Rationality’, is the stumbling-block. It covers a lot of the same ground as the chapter in Part I of Explaining Technical Change on ‘Intentional Behaviour’. But it packs in even more, and I think that anyone coming to it cold would be pretty bewildered. The best strategy would be to read it after the other three chapters. And even then it would be wise not to worry if some things go whizzing by too fast to catch hold of. The one point that is important for the rest of the book is this: it would be useful to have a concept of rationality that is stronger than consistency of belief or the adaptation of means to ends; but at the same time, we should be losing the distinctive critical force of the concept if we were to start from the other end and simply identify rational beliefs with true ones and rational actions with good ones. Elster’s ‘suggestion is that we should evaluate the broad rationality of beliefs and desires by looking at the way in which they are shaped.’ This is, however, a notoriously tricky business. We may start by saying that irrational beliefs and actions are those that arise from social or psychological compulsion and before we know where we are, finish up by concluding that, since everything has a causal antecedent, there can be no such thing as rationality. Elster is clear that this is a dead end. ‘All desires and beliefs have a (sufficient) causal origin, but some of them have the wrong sort of causal history and hence are irrational.’
The remaining three chapters, inasfar as they have a guiding thread, examine some forms that the ‘wrong sort of causal history’ can take. Elster is, I think, quite correct in taking the negative tack: it is easier to say what makes a belief or action irrational than it is to give a positive characterisation of what makes it rational. So, he says, ‘in the present work, autonomy’ – which he takes to be the quintessence of rationality in action – ‘will have to be understood as a mere residual, as what is left after we have eliminated the desires that have been shaped by one of the mechanisms on the short list for irrational preference-formation.’
The second chapter, on ‘States that are Essentially By-Products’, is concerned with the mistake of aiming directly at the achievement of some state of affairs which in the nature of the case can arise only as a byproduct of aiming at something else. Elster does not try to explain how this chapter fits the agenda, but it seems to me that attempts to do what is (logically or factually) impossible violate the minimal means-end kind of rationality rather than throwing much light on the richer substantive conception. Be that as it may, the idea itself is obviously sound enough. An example would be doing some altruistic act in order to get a glow of self-approval when such a glow arises only if the act is done for some other reason.
The interest and value of the chapter lie in the great variety of examples that Elster is able to find from all aspects of life. Although many of these are ingenious and illuminating, I am inclined to think that Elster over-presses his case. He commits himself to psychological assertions of necessity and impossibility that I see little reason for accepting. For example, he doubts whether anyone so ‘self-centred’ as to go deliberately to Calcutta to ‘understand emotionally and not only intellectually his duty to humanity’ could be affected by the experience. But it seems to me that that is simply a rather dramatic (and perhaps suspect only for that reason) version of a kind of self-management all of us go in for all the time when we either deliberately turn our attention away from ‘upsetting’ experiences or seek them out. The notion that our knowing what we are up to vitiates its effects is, I believe, just false. Here is another example. Elster quotes Tocqueville as saying that the important thing about jury duty is not its effect on the outcome but its educative effect on the jurors, and comments that ‘this effect would be spoilt if the jurors thought that the effect on their own civic spirit was the main purpose of the proceedings.’ What is necessarily true is that the effect on the jurors can be obtained only if they take seriously the obligation of trying to reach a correct verdict. Yet this does not seem to me inconsistent with their believing that the main reason for concentrating on the verdict is that only thereby can they reap the educative benefits. There is a difference between mental gymnastics and psychological impossibility which I think Elster frequently blurs in this chapter.
The chapter that gives the book its title, ‘Sour Grapes’, comes next. This is about ‘adaptive preferences’: the fox, having failed to reach the grapes, said they were sour, implying that he wouldn’t have liked them anyway. Again, the connection between the topic and the overall agenda put forward in the first chapter is not spelt out, but it seems fairly clear that the narrow conception of rationality has nothing to say about the irrationality of desires themselves, so if adaptive preferences (changes in preference in response to what is available) are irrational, it must be on the substantive conception of rationality. But are they?
The level of discussion in this chapter seems to me below par, because Elster tries to do too much in it. It was originally written to appear in Amartya Sen and Bernard Williams’s book on Utilitarianism, and in accordance with this, its accounced object is to argue that the phenomenon of adaptive preferences discredits Utilitarianism as a moral theory. But this means that there are really two discussions going on, one on the irrationality of adaptive preferences and one on what is wrong with a moral theory based on want-satisfaction. The two issues are of course related, but the effect of trying to discuss them at once is that both discussions are too superficial to carry conviction.
The discussion of Utilitarianism is undermined by Elster’s failure to specify adequately the criteria he has in mind. His set-piece example is the debate over the Industrial Revolution, which he encapsulates in the question whether things were better in 1850 than in 1750 or worse. He imagines two forms of Utilitarian principle to be used in settling this, one ordinal and one cardinal. The ordinal version is some sort of (unspecified) rule for getting us from a set of individual preferences (of the form ‘I prefer x to y’) to a social judgment. A simple example of such a ‘social welfare function’ might be: x is better than y if more people prefer x to y than prefer y to x. Now Elster points out that such a criterion might yield an indeterminate answer because a majority in 1750 might prefer living in 1750 (applying 1750 preferences to both) and a majority in 1850 might prefer living in 1850 (applying 1850 preferences to both). But this is to go much too fast. Since hardly any people were alive in both 1750 and 1850, what exactly is the comparison to be? Are we to match each person on one date with a counterpart on the other? (And how do we deal with the difficulty that we would have a lot of 1850 people left over?) Suppose this were done, and somebody in 1850 is asked: ‘Would you sooner be you as you are now or you in the situation of Joe Higgins circa 1750?’ Why should we imagine that to be a relevant question for deciding anything? A natural response might be: ‘I’d find Joe Higgins’s life pretty tough: but if I’d grown up as Joe Higgins it would probably have been all right.’
If we want to pose the question in terms of preferences, the relevant question would appear to be: ‘Would you rather be you now or Joe Higgins a hundred years ago?’ (But then what is the referent of ‘you’? These are deep waters, Watson.) The best way of making something out of that question would be to assimilate it to: ‘Do you think Joe Higgins was happier than you are?’ This is, in my view, a quite sensible question, but it immediately takes us outside the framework of ordinal Utilitarianism.
Cardinal Utilitarianism is understood by Elster as the criterion that the more wants are satisfied the better. He seems to assume that want-satisfaction should be defined as the absence of unsatisfied wants, and argues that it may be better to have more wants even if some are frustrated. But why should want-satisfaction not be defined as the presence of satisfied wants? Then somebody with more unsatisfied wants would score higher so long as he had more satisfied ones too. I do not pose this as the last word, but my point is that Elster does not push his analysis far enough.
The core of his objection to the criterion of want-satisfaction is, in any case, that, however we count want-satisfaction, we should give an independent weight to autonomy, for which frustrated wants are evidence. But autonomy, as we have seen, is a residual defined in terms of the absence of the ‘wrong sort of causality’. So we have to ask what is supposed to be wrong with the adaptation of preferences to what is available. As far as I can see, Elster’s case for saying that adaptive preferences are irrational is contained in the remark that ‘sour grapes’ – in contrast with purposive character-planning – ‘is a purely causal process of adaptation, taking place “behind the back” of the person concerned.’ But since, as Elster concedes, every desire has some causal antecedents, why should we regard the spontaneous adaptation of preferences as irrational (rather than non-rational but fortunate)? I suppose that most of us (Dante being an outstanding exception) have a tendency to acquire a taste for what we are able to get and lose a taste for what we are not, and why shouldn’t we? I do not think Elster tells us.
It is quite true that there are some instances of adaptation which we are tempted to call irrational, but since there are others that we are not, this shows that there must be some extra feature to them. Self-delusion is a good candidate. The fox in the fable who thought the grapes he could not reach would be sour (assuming he really thought it and didn’t just say it) is indeed a plausible example of irrationality. But the more typical case of adaptive preference-formation, exemplified in a greater taste for knicknacks and a lesser taste for poaching in 1850 than in 1750, does not on the face of it have anything of the same quality.
The final chapter, ‘Belief, Bias and Ideology’, is slighter than the preceding two. It contains valuable warnings against errors that are often made. Thus, there is no reason to suppose that an interest-induced belief (e.g. a piece of wishful thinking) serves the interest of the person holding it – if anything the opposite is more likely. And, in line with his antifunctionalist argument, Elster points out that ‘the mere fact that a belief furthers some interest in itself provides no explanation whatsoever.’ I think he is right in insisting that a causal mechanism has to be produced and that Marxists have a bad tendency to suppose that the presence of a belief is somehow explained simply by pointing to its conducing to the interests of the dominant group. However, Elster’s discussion of the ways in which the connection might be filled in is unconvincing.
A more pervasive problem with the chapter is the absence of any serious effort to tie the discussion to the issues about the definition of rational belief raised at the beginning of the book. Most of the phenomena that Elster deals with here involve rather straightforward weaknesses in information-processing: misplaced generalisation (the Tocqueville marriage example), failure to draw appropriate inferences, and so on. These are presumably irrational in the narrow sense of the term. That they are induced by interests seems neither here nor there in deciding us to call them irrational: they would be equally irrational however they came about. Whether or not anything illuminating can be said in general about the ‘wrong sort of causality’ in relation to beliefs remains open. I hope that Elster will return to this on another occasion. It is exactly the kind of question that is best approached by his characteristic method of accumulating and anyalsing a wide variety of examples.
With the word ‘example’ I sound my closing note. When Elster is at the top of his form the general ideas form the framework for the argument, which is itself actually carried on to a great extent through the use of examples. This is hard to convey in a review. But it is what gives Elster’s work its uniquely attractive quality. I would defy anyone to read, in particular, the second chapter of Sour Grapes without being charmed by the play of a subtle intellect on a remarkable range of materials.