The Continued Relevance of Weber ’ s Philosophy of Social Science

Only a few writers have attempted to construct a comprehensive philosophy of social science, and of these Weber is the most relevant to the present. The structure of his conception places him in a close relationship to Donald Davidson. The basic reasoning of Davidson on action explanation, anomalous monism, and the impossibility of a ‘serious science’ of psychology is paralleled in Weber. There are apparent differences with respect to their treatment of the status of the model of rational action and the problem of other cultures, as well as the problem of the objectivity of values, but on examination, these turn out to be modest but, nevertheless, interesting. Weber’s use of the notion of ideal-types, though it is not paralleled directly in Davidson, allows him to make parallel conclusions about the relation of truth and interpretation: both make the problem of intelligibility rather than correspondence with some sort of external reality central, and each addresses, though in different ways, the dependence of considerations of intelligibility on normativity and the impossibility of a theory of meaning without idealization. Methodological and philosophical writings about the social sciences are plagued by the eternal recurrence, typically in slightly different guises, of the same issues: abstraction, rationality and its relevance to causal explanation, the problem of cultural and historically specific perspectives as a condition for constituting and defining topics and the problem of reconciling this with claims to universality of knowledge, and various other problems relating to the question of the a priori conditions of knowledge itself, which in the social sciences often involves conflicting choices between possible starting points. Dealing with any of these complex issues as they relate to the polymorphous and pluralized disciplinary and subject-matter worlds of the humanities and human sciences is difficult. But attaining some sort of reasonable perspective on the relations of these problems to one another and to the problem of the relations between the human sciences and such related enterprises as the natural sciences and the normative disciplines, for example law, is overwhelmingly difficult. The thinkers who have attempted to deal with these problems by reductive simplifications are numerous. The thinkers who have seriously attempted to deal with all * A different version of this paper was published in Etica e Politica/ Ethics and Politics, 2005. I thank the editors of this journal for the opportunity to republish the paper in this revised form. of them are few. One of them was John Stuart Mill, and the tremendous influence of his major work on the philosophy of social science, his System of Logic ([1843]1963-1991), testifies both to the difficulty of the task, with which he only partially succeeded, and the rarity with which it is attempted. Weber is perhaps the only other thinker who has left a coherent response to these issues which deals with the relations between the issues. Weber’s competitors in his own time, such as Durkheim, Pareto, and Karl Pearson (whose methodological ideas were transmitted and refined for social science by Franklin Giddings in the United States), each dealt with some of these issues but ignored, dismissed, or made arbitrary choices with respect to others. In what follows I will explain Weber’s views on each of these issues and the ways he put them together. In my conclusion, I will indicate some ways in which these issues remain live problems and also why some of the favored present day solutions to these problems in contemporary philosophy are inadequate or at least problematic. My stalking horse in this discussion will be, perhaps surprisingly, Donald Davidson, who is among the few contemporary philosophers to attempt to deal with what is roughly the same range of problems, especially problems of the nature of rationality as a model, the role of the normative in the constitution of problems, including the problem of rationality, and the dubiousness of such collective concepts as language. Davidson contrasts to Weber on these key issues, issues which are important to present day philosophy, and are especially important in the interface between contemporary philosophy and the philosophy of social science. But Davidson also, as we shall see, provides Weber a way out of the issue that has traditionally been regarded as a central defect of his thought considered as philosophy What may seen odd about this discussion is that Davidson is an ‘analytic’ philosopher, while Weber is and has usually been understood in terms of ‘continental’ philosophy. The great works on Weber as a philosopher, such as H. H. Bruun’s Science, Values and Politics in Max Weber’s Methodology (1971), Leo Strauss’s treatment of Weber in Natural Right and History (1953), or Alfred Schutz’s Phenomenology of the Social World (1932), have taken Weber as a part of this tradition. But today, especially in the light of Michael Friedman’s work on the origins of Logical Positivism and on the parting of the ways between what became ‘Continental’ and ‘analytic’ philosophy (Friedman 1999, 2000), which occurred for the most part after Weber wrote, these contrasts no longer have the categorical meaning they once did. And the rise of rationalchoice theorizing and its application to Weber has closed the gap further: Richard von Mises’s ‘Austrian Economics’ critique of Weber (Mises 1960), as we shall see, is as relevant today as ever– and the issues over the status of the model of rational choice that arise with Mises are also issues that are at the heart of the comparison to Davidson. 1. Reasons and Causes Weber understood at the outset that there was a fundamental conflict between the task of understanding action and the task of causal explanation, but he also understood that the conflict could not be resolved by collapsing one into the other. Reasons were not causes, at least in the sense of cause as governed by scientific laws, if these were understood by analogy with laws of chemistry, physics, or astronomy. The concepts we employ in the natural sciences, are, so to speak, developed to fit and explain the regularities that science discovers. In social science, in contrast, the terms in which we describe the things we wish to explain are given for us in what Weber characterizes as the language of life (Weber [1904] 1988: 209). There is no a priori reason that this is true, but it nevertheless is true: there are no nomic regularities that correspond to the facts that can be described in the language of life. Nevertheless, causal explanation to some extent is part and parcel of all of the human sciences, or as Weber typically called them, the historical sciences, even where these sciences are constituted and define themselves in such a way as to minimize the significance of causality. Weber resolved this conflict by arguing that there is a different relevant notion of causality, which he identified with a very specific legal theory of causal explanation developed by his Freiburg colleague, Johannes von Kries. This account defines ‘adequate causation’ as an increase in the probability of an event if we compare the probability of the event given the cause in question together with a selected set of background causes and the probability of the event given the background causes on their own. Von Kries provided him with a plausible and easily generalized notion of causality that does not depend on the existence of genuine nomic regularities and is entirely consistent with the formulation of both input and output causes and outcomes in ‘the language of life’. It is a method that allows us to speak of causes of the kind familiar to the historian, such as the causes of World War I, but also allows us to speak of the causal contribution of such things as the reasons that a given person had which contributed to an action, such as the elements of a decision. The notion of cause here is not an experimental notion but rather one based on expectations derived from the ordinary course of events which are then applied to particular actions in order to make sense of them both ‘causally’ and ‘rationally’. Weber’s solution in principle to the problem of reasons and causes, then, was to require that an explanation of an action be ‘adequate’ both at the level of understanding or interpretation, which is to say that it makes the action within its circumstances intelligible in terms of a type concept that is already understood, and adequate at the level of cause, which is to say, consistent with our expectations about the contribution of a particular reason as a cause in a set against the background of a similar set of circumstances or causal conditions. Is this a plausible conception of action in current philosophy? It compares to the writings of Davidson on anomalous monism ([1970]1980), though there are significant differences. Like Weber, Davidson argued that actions can be understood causally in terms of reasons together with what he called pro-attitudes, which are motivational sources of varying strength. He said also that ‘reasons explanations . . . are in some sense low-grade; they explain less than the best explanations in the hard sciences because of their reliance on propensities’ ([1987]2004: 109). Davidson also acknowledged that the genuinely nomic causes of human action cannot be articulated within the language of life ([1970]1980: 225). For Weber the typification of rationality with respect to action is one, though the most useful and powerful, of a myriad of intelligibility producing typifications, such as the typification of actions done out of emotions (in which, for example, the evidence of an angry face, together with sufficient motivating circumstances, justifies the application of the typification ‘done out of anger’). Thus, for Weber these various typifications are in effect on equal footing as intelligibility producers and potential causes, though obviously some will be more useful than others in particular contexts. Weber did not deny that such things as biological causes are causal contributors to human action even when they are not part of the language of life or a typification of the kind that renders an action intelligible. When he commented that the facts described by the concept of charisma seem to fade off from the intelligible into the unintelligible, where they can best be accounted for as biologically caused phenomena, he acknowledged the limitations of the understanding of actions as actions, that is to say as fully intelligible, and he freely conceded that large areas of even intelligible action are co-produced by reasons or causes construed in the language of life and biological causes. Davidson, in contrast, is committed to the indispensability of the paradigm of rational choice or decision theory, treating it as inseparable from the notion of intelligibility and reasons with respect to human actions, or to put it more plainly, that intentionality in the sense of the axioms of decision theory is coextensive and inseparable from action itself. Weber accepted what we might call piecemeal intelligibility, intelligibility in terms of a variety of distinct typifications, while Davidson believed that the wholesale conception of rationality is necessary to the understanding of action. This reasoning is not without its difficulties. It leaves Davidson, unlike Weber, with the problem of akrasia or weakness of will, a problem that arises from the ubiquitous gaps between the possession of reasons and actions in which its is rational to act but the agent does not act. Akrasia was not a problem for Weber, for two reasons. Because Weber considered typifications of actions to be ideal-typical in character and therefore holds that the extent to which an apparent action is in fact an action is a matter of degree and fact rather than a matter that can be settled a priori. Moreover, akrasia is a familiar and intelligible phenomenon, which, in Weber’s terms, we can typify. There is also a problem about the status of the model of rational action. As we will see, Davidson devoted considerable effort to this problem, which in the end comes down to a problem, as we shall see, of understanding an intellectual object that has both characteristics of a priori truth and of empirical truth. Whether one prefers to accept Weber’s difficulties or Davidson’s on this matter, it is relevant to note that the dispute over the status of what Weber thought of as the ideal type of total rationality was well developed in Weber’s own lifetime, and a contrast between the two interpretations of the status of decision theory was discussed by Alfred Schutz, who adhered to the position of Mises, who identified action by definition with rational action and thus, like Davidson, denied that actions which could not be made intelligible in terms of rationality were, properly speaking, actions at all. This difference remains: for Weber intelligibility in terms of typifications is not limited to the ‘typification’of rational action, and the characteristic of ‘being an action’ is itself a matter of degree, largely as a consequence of the fact that actions are co-produced by biological causes. Davidson in his own way acknowledged the co-production of action by biology and reasons. He argued that his own account of the logic of action explanation, in which reasons together with a ‘pro-attitude’ constitute the kind of ‘cause’ appropriate to human action, is connected to the brain, which processes information and desire in two different places. Thus like Weber he grants that having a reason is not enough to explain an action, and that the kind of cause appropriate to human action is not the same as that in the nomic sciences. But where Davidson takes the view that the two elements of action-reasons and pro-attitudes-have distinct biological sources, Weber took the view that in interpreting action, reasons and pro-attitudes are packaged together in typifications which make action intelligible. The probability of action outcomes is a separate issue for Weber, but the probabilities of outcomes are potentially influenced by co-producers, such as the biological causes that contribute to charisma, and not merely the intelligible features of action. Yet for both of them, these biological sources are unusable as action explanations because they are not in the language of life, and thus there is no possibility of reduction or ‘explanation’ in the nomic sense. Davidson’s and Weber’s views of the empirical status of the model of rational choice appear to be dissimilar, but the differences are limited, and closely connected to the notion of ideal-types. Davidson was involved in the continued discussion in the psychological field of decision science of the status of decision theory as a model of human decision-making, a literature faced with the problem that actual human decisions systematically vary from the ‘normative’ model, which nevertheless seems to be a priori valid. Davidson presented the issue in this way: It may seem that I want to insist that decision theory, like the simple postulate that people tend to do what they believe will promote their ends, is necessarily true, or perhaps analytic, or that it states part of what we mean by saying someone prefers one alternative to another. But in fact I want to say none of these things, if only because I understand none of them. My point is skeptical, and relative. I am skeptical that we have a clear idea of what would, or should, show that decision theory is false; and I think that compared to attributions of desires, preferences or beliefs, the axioms of decision theory lend little empirical force to explanations of action. (1976: 273) Weber expressed this same sense that ‘rationality’ was neither a priori nor empirical with his notion that rationality was an ideal-type, but the ideal type which represented the acme of intelligibility or Evidenz. To put the problem in Weber’s terms, the question is whether the ideal type ‘rationality’ is merely a typification or is something more, in spite of its lack of correspondence to actual human decision making. It is notorious that Weber’s uses of the term rationality were various, shifting, and irreducible to a single sense. But Davidson made a related point when he said ‘The issue is not whether we all agree exactly on what the norms of rationality are; the point is rather that we all have such norms, and that we cannot recognize as thought phenomena that are too far out of the line. Better say: what is too far out of line is not thought’ (Davidson, quoted in Engel 1999: 448). Weber would reach the same conclusion by way of typification: action we cannot typify and thus make intelligible as action is, for us, not action. And to be typified as action it must be to some slight extent instrumental, or otherwise to conform to one of our typifications of rationality. 2. Models and Ideal-types Weber’s use of the term ideal type and typification may appear to place him in the epistemic dark ages. Not only is this language no longer current, the neo-Kantian notions of concept and conceptualization from which it derives, such as Rickert’s idea of the hiatus irrationalis between concept and reality, is no longer current either. Nevertheless, the issues that Weber sought to address by the use of these terms are alive in contemporary philosophy. In the case of the problem of the status of rationality, as we have seen, the issue is phrased in terms of normativity. Decision theory is a ‘normative’ conception whose empirical significance and thus its explanatory power is problematic. The question is whether the normative conception is in some sense indispensable or necessary for the constitution of the subject matter, for example, for the interpretation of the babblings of other people as language ([1974]1984). For Davidson interpreting them requires us to impute rationality, at least in the sense of sharing many of our beliefs (though no belief in particular). Without this, he argued, we would be unable to make their utterances intelligible at all. The question of whether the normative conception of rationality contained in decision theory is indispensable to interpretation is not settled by these ‘constitutive’ considerations. But the fact that Davidson’s examples of the interpretation of apparently irrational beliefs involves the notion of error suggests that it might. Weber in his discussion of ideal types faced the problem of normativity in almost the same form, since he was compelled to distinguish the explanatory from the ‘dogmatic’ uses of the same concept, and to explain that the unique ideal of rationality was abstract and therefore ‘ideal’ in the sense of ‘ideational’ rather than ‘normatively preferable’. Weber gave as an example the problem of explaining the blunders made by a general in the battlefield, something that requires a model of correct decision making against which actual conduct could be compared in order to identify the blunder to be explained. He did not say that the normative model of rationality is indispensable, but its difficult to see how it would be dispensed with. This approach was the one taken up by Popper in The Poverty of Historicism ([1957]1961) under the label of ‘the logic of the situation’. Like Weber, and probably influenced by him, Popper argued that the normative conception can serve as a baseline for asking questions about the influences on action by allowing the analyst to see what produced deviations from ‘rationality’ ([1957]1961: 140-143). This is an approach totally consistent with psychological decision theory, which treats deviations from the normative as ‘biases’. But for Weber (unlike Popper), there was a rationale for treating the normative model (understood as an ideal-type), as a baseline: it is, as he argued, the most intelligible ideal-type, for us. Thus the justification for its use is purpose-relative rather than a priori, and the purpose is central to the aims of historical science as an ‘understanding’ discipline. It may be de facto indispensable in many contexts, however. In this instrumental use of normative concepts of rationality, the consideration of whether they are genuinely normative, a priori truths is irrelevant. Instead, the notion of ideal type compares closely to contemporary philosophy of science discussions of the instrumental use of models, which similarly are idealizations rather than law-like statements, and useful both to make simplifications of complex situations in order to make predictions and to use these simplifications to identify sources of interference that may themselves have important, even overwhelming predictive significance. As a construction (of such things as the patrimonial state, Cadi justice, or charismatic authority), an ideal-type implies nothing about their correspondence to reality, their actual predictive power, or their desirability as a state of affairs. The term ‘ideal’ here implied merely that the type could be fully articulated as an idea and thus be able to be used to contrast to a complex and messy reality for various explanatory purposes. If ‘rationality’ is taken as a kind of first approximation, it enables us to break down the problem of understanding into smaller parts-the parts remaining being the deviations, biases, and intelligible errors that account for the rest. The present relevance of Weber’s views on ideal types or models is, ironically, perhaps greater today than it has been for the rest of the intervening century, where social scientists, under the influence of the model of physical law and logical positivism, sought to construct ‘general theories’ of societies as systems, or when ordinary language philosophers obsessed about the differences between reasons and causes. The notion of models or ideal types in Weber’s hands cuts across these distinctions by treating reasons of particular types as typifications which are usually mini-models, and by treating rationality itself as a model. Weber was also interested in rational models for the production of aggregate patterns, such as the disappearance of a type of money in a two currency system, the case covered by Gresham’s law. This is a model rather than a law. It predicts what will happen if people reason about the value of currency in particular ways and acted on those reasons without interference from other reasons or causes. Here the explanatory part of the model is rationality, but rationality applied in a very specific setting, and reduced to a very small number of rational considerations. But Weber also constructed models that were not organized around ‘rationality’. Weber’s model of the ancient city as discussed by M. I. Finley (1986) is such a model. It is essentially a model of an ancient economy in terms of its inputs and outputs and the social and political structures that enable the economy to function. The uses of such a model are multiple. It predicts, enables interpretation, enables comparisons that point to explanations of differences, and so forth. But it is explicitly a simplification, or abstraction, and thus not ‘true’ but rather ‘applicable’ to rich and complex situations where the richness is an unusable distraction which needs to be reduced, but which cannot be usefully reduced to scientific (or for that matter economic) ‘laws’. It was formerly thought that the sciences which were in a position to reduce the richness or complexity of the subject matter by constructing artificial experiments had an advantage over the social sciences, because experiments in a very simplified set of circumstances could be constructed to match up with the issues in crucial experiments between theories or to produce direct confirmation of theories. But generalizing from experiment is a problem that requires the richness of cases to be taken into account, and the only means for doing this is the construction of an intermediate abstraction, a model, for this purpose. These models may be made up of largely of theoretical principles but they also (and often) involve interactions between variables whose properties are unpredicted by a theory but which can be estimated from the data itself. This puts social science and natural science on something closer to an equal epistemic footing, and at the same time rescues Weber’s usage from the antique language in which it is presented. 3. Cultural Categories Perhaps the most relevant of Weber’s core ideas, which I have already alluded to, concerns the cultural constitution of the objects of social science. Weber’s methodological enemies divide into roughly two groups. One group believes that the cultural constitution of the subject matter is irrelevant as long as we can identify data points and treat these as facts which can be related statistically or in some other quantitative way. Durkheim’s famous slogan, ‘treat social facts as things’, is an expression of this impulse, but it appears in an even more unmediated form in standard social science methodology, where any quantifiable material becomes, by virtue of its quantifiability, a potential subject of statistical analysis and ‘theory’. The second group of Weber’s enemies are what I will dub apriorist. The paradigm case of apriorism in the philosophy of social science is the distinction between reasons and causes itself, which grants a priori status of irreducibility, correctness, and descriptive validity to those descriptions of conduct or action which correspond to the reasons of the agent or of the agent’s linguistic community, represented especially by Peter Winch (1958). There are however, many other a priori starting points. As noted, Mises’ equation of rationality and action is one of them-for him all actions are by definition rational and nothing that is not rational is, properly speaking, an action. Here we see the a priori structure in a particularly naked form. But many other thinkers, such as George Herbert Mead, have constructed the problem of explaining human activity in terms of considerations that are also a priori. For Mead, the problem of priority is one in which particular features of human activity, such as symbol use, are taken for granted as valid descriptors and subjects to be explained, and the problem becomes one of accounting for the conditions for the intelligible use of these concepts in analysis. Thus Mead criticized the a priori notion of the self by showing that others were a condition for the construction of the self. This style of analysis, of taking some particular fact in the human realm as given, as Mead did, and inquiring into its conditions, is commonplace. Schutz, similarly, took human understanding of the other as the thing to be accounted for and inquired into the phenomenological conditions for such understanding. Weber’s response to all of this was to collapse the problem of the a priori into the problem of the arbitrariness of categories of, as he put it, selection from the meaningless chaos of existential judgments that would face us if we approached reality ‘without presuppositions’ (1949: 78). His argument was that the social sciences have already had the selection made for them, in the first instance, by the constitution of the subject matter in the language of life. For him, the problem of understanding is a matter of our interest. We choose to be concerned with human things described in human language. Terms like ‘action’ or ‘the city’ are, initially at least, simply terms of common usage which constitute our subject matter for us. We then make further conceptual refinements in our selections from reality based on our specific explanatory interests. For Weber, sociology itself is not a subject matter grounded in the nature of the universe but is rather grounded in the choices that we make to define the subject. His definition of sociology, as the study of meaningful social action, has no special status as a definition. It is not further grounded in an a priori, philosophically groundable ontology or epistemology which selects it as the only possible definition-something that many of his critics and interpreters, such as Schutz found to be a failing. Like the conceptual categories that Weber deployed, its only claim on us is the claim of utility, and utility is utility in reference to culturally generated and fundamentally ordinary kinds of purposes of understanding, such as the purpose of understanding the historical causes of present day capitalism, rather than universal epistemic goals, such as the final truth about the nature of society. This is an important move in relation to the second kind of criticism because it temporalizes, historicizes, and makes culturally relative all of the problematic philosophical distinctions, including the distinction between reasons and causes, which have been the subject of aprioristic philosophizing during the 20 century. The significance of doing this is that it undermines any sort of absolutization of particular concepts, which constitute the given on which this strategy depends. The notion of the self, for example, becomes available to us as a theoretical or conceptual term which we can use to define problems, but our use of it always comes with the implicit qualification that the term may be historically rooted in the language of life of a very specific historical community, namely ours, and that its applicability elsewhere cannot be taken for granted, and indeed that the applicability and historical origins and development of such concepts becomes fair game for sociological analysis itself. Thus the modern notion of the self is, as Weber implicitly treated it, the product of a particular theodicy that is part of the theology of reformation Protestantism, without which there would not be the concept of ‘the self’ in its present form. However, this strategy of historicization, applied reflexively, raises a question about Weber that was pointed out repeatedly by his critics in the twenties. Is Weber’s own conception of social science transhistorical, and if so, how can the idea of a transhistorical conception of social science be understood within the framework of his purpose-relative conception of social science in which fundamental conceptual schemes and categories are understood to be historical? His admirers, notably Karl Jaspers, argued that his conception of social science was itself purpose-relative, an instrument or a probe, which was itself subject to revision and which made no claim to finality or absoluteness. Understanding the arguments in this way brings Weber very close to the Heideggerian notion of every form of knowledge being a form of concealment, because Weber himself suggested that the content of social science changed as ‘the light of the great cultural concerns moves on’ (1949: 112). And this perfectly reasonable reconstruction of Weber places it in the general framework of Continental philosophy. Weber himself, I would suggest, would not have argued in quite this way, and his strategy in relation to the causal character of historical sciences provides a model that is useful for understanding how he might have responded differently to these issues. Weber argued that all historical sciences are concerned, at least in some minimal respect, with causal questions, and are thus compelled to deal with the issues that arise out of the problems of applying the notion of causality in the human realm, including the notion of the need for a selection of categories with which to define the causal questions, a problem that cannot be answered in terms of nature and cannot be separated from the language of life. In relation to the question of the grounding of his own methodological conception, I suspect that the most ‘Weberian’ response that could be constructed would be along the same lines: one which focused on the notion of intelligibility as the common element of the human sciences. The Weberian approach to the problem of collective concepts provides a useful illustration of this kind of reasoning. Weber famously proclaimed himself an enemy of collective concepts, but did not do so on metaphysical grounds, either on the grounds of an ontological individualism, or for that matter on the grounds that Popper would later construct as methodological individualism. Instead he argued that collective concepts were explanatorily dispensable for the particular purposes of sociology as he defined them, i.e in terms of ‘the interpretive understanding of social action and thereby with the causal explanation of its course and consequence’ ([1968]1978: 4), and that from this point of view they were fictions-useful and even, for the practical purposes of such sciences as legal science, necessary, but fictions for sociology. The thrust of his study ‘On the Categories of Interpretive Sociology’ was to show how one could understand these concepts without claiming they represent autonomous realities ([1913] 1988: 42774) . This is not, as in Popper, the result of a methodological decision to see how far one can go in explanation without appealing to collective concepts. But what does it establish? And how? What it seems to establish is the interest-relative character of such conceptstheir dependence on the specific purposes and thus presuppositions of special disciplines, such as the presupposition of the notion of legality for dogmatic legal science, or the presupposition of the notion of the state for political science. These are purpose-relative justifications of the use of collective concepts, rather than ontological or generic ‘methodological’ justifications. Weber’s distinctiveness is shown in the fact that theorists in these fields have typically attempted to avoid this kind of argument. Philosophers of law in the tradition of Kelsen and Hart, for example, have insisted that certain features of the law, such as the notion of obligation, are factual rather than fictional or ‘dogmatic’ (and that they cannot without circularity be understood as ‘merely legal’) and thus that the subject matter of the law cannot be explained without ‘changing the subject’ if such notions as legal obligation are not recognized as genuine in a non-purpose relative sense. The contemporary literature on collective intentionality is parallel in structure to this kind of philosophy of law. It advances a line of argument, which Weber did not address, that an individualist ontology has no more claim on our intelligence than a collectivist one, and that any preference for individual over collective ontologies is groundless. How could he have responded? Weber, like some contemporary philosophers, such as Susan Hurley (1989: 157-158), could have argued that the preference need not be ontological and need not be groundless. Our understanding of individual action and the beliefs of individuals as well as their intentions is for Weber (and Hurley) clearly superior to our capacity to understand ‘collective intentions’. Thus if a preference for greater degrees of intelligibility, which Weber could argue (parallel to his argument that all of the historical sciences are in some respect concerned with causal relations) is central to the shared purposes of disciplines concerned with human action, is a common ground, then preferring explanations in terms of the beliefs, intentions, and actions of individuals, to the extent that it is sufficient for the explanatory purposes in question, is justified by this more basic preference. This argument parallels the point Weber explicitly made about cause-that it is common to the historical sciences and in this limited sense more basic. When Weber wrote, the situation of the social sciences was such that it might reasonably have been supposed that the various approaches of the social sciences could be integrated, but that one or another of these approaches would prove overwhelmingly more fruitful than the others, and that the methodology of the social sciences could be reduced to the methodology of the most successful approach. It could reasonably have been supposed that some version of the model of natural science could successfully be applied to the social sciences raising the successful application to the status of science, thus freeing it from the limitations which Weber stressed as characteristic of the historical sciences. Much of social science writing and methodology in the 20 century reflected this idea of integration. But what transpired, despite the aggressive attempts of writers such as Talcott Parsons to prevent it, was a radical fragmentation of approaches, especially in sociology, but to some extent the other social sciences as well, in which different forms of problem constitution generated significant bodies of research scholarship, and constituted their problems in such as way that they could not easily be dismissed as nonscientific, irrelevant, and so forth. In short, methodological and theoretical pluralism established itself in practice, rather than merely in theory. Nevertheless these approaches were partially circular in just the way that Weber’s argument anticipated. The problems of an empirical nature that arose within them were problems that arose because they were constructed in particular conceptual terms, and there could be no argument of a convincing kind for the exclusive relevance of those particular terms. The social sciences thus became, in the language of the sixties and seventies, following Kuhn, multi-paradigm sciences in which diversity was irremediable and the dream of a single unified science of society or social life receded even in the writings of philosophers of social science themselves. 4. Normativity and Intelligibility The apparent, and partial, exception to this diversity of paradigms is economics, and the model of rational choice, as I have already noted. This model presents today, as it did in Weber’s time, a series of puzzles about the fundamental character and logical status of decision theory. And it is in the tangle of issues that arise in the characterization of rationality, with ‘problems of rationality’ as a recent volume of the collected essays of Davidson is entitled (2004), that Weber’s present relevance becomes especially apparent. The problems of rationality are inter-related, and, for Weber as for Davidson, it is the connectors which link the parts of their thought that are most puzzling and problematic. And it is in comparing the two in this respect that one can see most clearly the coherence of Weber’s position, and its distinctiveness. Begin with some common thoughts between Davidson and Weber: that there is an unbridgeable divide between the language of intentional action, which is to say the language of belief, and the language of scientific law. This means that there can be no ‘serious science of psychology,’ as Davidson put it, meaning a science that actually explains actions, or the rest of the ‘psychological’ domain, in anything like the language of intention and belief. Davidson put this point in terms of the propositional attitudes (believes that, knows that, fears that, and the like), and said ‘that there is an irreducible difference between psychological explanations that involve the propositional attitudes and explanations in sciences like physics and physiology (1999: 460). Weber put this contrast in terms of ‘subjective meaning’, which amounts to the same thing: for something to have subjective meaning for someone is to have a propositional attitude toward that thing. They agree as well that action explanation is nevertheless causal, that is despite not being transformable into the sorts of things that can be explained by laws, as discussed above. What is not shared is the way they characterized the difference between action explanations and nomic explanations. For Davidson, ‘the concepts we use to explain and describe thought, speech, and action, are irreducibly normative’ (1999: 460). Weber of course held the same view under the heading of ‘value-relatedness’. For Weber ‘values’ already enter into the descriptions one makes in the language of life. Moreover, Weber thought of ordinary language as the expression of a valuative world outlook, and thus distinctive to the epoch. Thus they agree on the normativity of descriptions of intentional action, but disagree on the reasons for it. Davidson, in ‘The Very Idea of a Conceptual Scheme’, dismissed the idea of distinct conceptual schemes on which Weber appears to rely, and indeed he might have listed Weber, along with Kuhn, as an example of the kind of thinker who, like Kuhn, ‘is brilliant at saying what things were like before the revolution using-what else-our post-revolutionary idiom’ ([1974]1984: 184). Against this Davidson argued that we could not understand another culture if we do not assume that many of the beliefs of the speakers in another culture are true and like our own, without which we would have no basis for translating and interpreting their beliefs, or even understanding them as beliefs and actions. This would seem to rule out genuinely distinctive Weltanschauungen based on different normativities or values, and indeed, Davidson, in a discussion of the problem of the objectivity of values, embraced the objectivity of values. But here the differences are not so great as they initially appear. Weber used the term Weltanschauung, but he also emphasized that it is an abstraction, an ideal-type construction of a more complex set of individual beliefs and ideas. And although I have loosely used the term ‘culture’ here to characterize his views, he did nothing to ontologize any of these collective mental concepts, or make them into fixed units, such as paradigms. He did not suppose that the thoughts (including the typifications and schemes of significance) of other people in other eras are always accessible to us, but did take the view that to the extent that they are, they must be accessible through our typifications, though of course these typifications may be refined and abstracted to be made more useful for the purpose of making sense of these other eras. For Weber we are condemned to the scheme of significance of our present, but we can extend it through abstraction. Where does this leave us with the normativity of rationality, and normativity generally? With respect to the normativity of decision theory, the two appear to have different views. Weber took ‘rationality’ to be an ideal-type among others, but primus inter pares with respect to its power to produce intelligibility. Davidson seems to be close to Mises, in making rationality inseparable from action. But there are some ambiguities in each of their formulations that parallel one another and bring them closer together. There is also an ambiguity in common, between rationality as something more or less fixed and something more or less local and ‘ours’. For Weber, ‘logic’, which might be taken by him to include calculation and decision-theory, was non-valuative, in contrast to the domain of described facts of the historical sciences, in which a ‘valuative’ element entered. What is valuative is, for Weber, what is ours: logic is everyone’s. But Weber nevertheless speaks, in a famous passage, of the ‘norms of our thought’ ([1904] 1949: 84) which in context is ambiguous between the local and the universal sense of norms. And there is the oddity of the claim that the laws of thought are normative. If they are, how could they be other than universal? The dilemma that produces this ambiguity is this. For Weber, to acknowledge that the typifications we use in interpretation are not necessarily valid transhistorically is to acknowledge an element of historical relativism. But to acknowledge that what counts as explanation or science for us is also not necessarily valid transhistorically seems to amount to making the methodological or second order claims also historically relative. And this is indeed what the critics of the fact value distinction, such as Strauss, did argue, treating Weber’s difficulties with this question as a stepping stone to Heidegger, and which is still argued today in the form of the claim that value-free science is itself a value choice (Williams, 2005). Davidson’s parallel problem is this: if psychology cannot be a ‘serious science’ like physics, is it not merely a restatement of (local) folk psychology with no autonomous claim to empirical validity? Davidson had a famous answer in general to such concerns which applies to folk psychology: we simply cannot understand the claim that our beliefs might be massively wrong. We can make sense of the claim that one or another of our beliefs is wrong. But we cannot as an interpreter ‘correctly interpret another speaker (such as one from another culture) in such a way that the speaker’s beliefs come out massively false in the interpreter’s opinion’. This has a dramatic implication. Folk psychology forms such a large part of our beliefs that it cannot be as a whole massively false, despite the fact that psychology cannot be a ‘serious science’. Why? Because if we meet up with a people whose language appears to lack terms corresponding to our fears that, wants that, and so forth, and we cannot translate what they do say to mean something that produces more or less similar results, we would simply be unable to translate them at all. And what holds for cultures here holds also for individuals. Davidson’s point here is similar to his point about rationality. We simply cannot make sense of the claim that the ordinary notion of rationality is false, for to reason about it we rely on it, as we do for example in understanding biases in decision-making by way of decision-theory. With the interpreter of another culture, we cannot even attribute intelligible error to another speaker without assuming that the other speaker is largely in conformity with our ordinary notion of rationality. Weber does not make this claim. As an interpreter, he treated ‘rationality’ as an ideal-type, as something that belongs to us, as part of our language of life and historical period, and for which there are no guarantees that it is transhistorically applicable or applicable to other cultures. This is not a profound difference, in practice. Both agree that understanding other cultures is a matter of degree, and that it involves making their utterances and actions intelligible in our terms. The difference arises over Weber’s categorical, rather than ‘difference in degree’, distinction between ‘logic’ and value. To understand this requires a little background. As a student and follower of Quine, Davidson began as a ‘holist,’ as a result of Quine’s basic insight (relevant to the distinction in Weber between universal matters of logic and matters of potentially different conceptual categories) that the distinction between analytic and synthetic statements should be abandoned. Quine was opposed to the earlier Logical Positivists, who argued that there were certain kinds of statements that were directly verifiable and which ought to form the empirical basis of theories. Quine’s point was that for any observation sentence, it was possible to make adjustments in other parts of the theory of which it was a part that allowed for it to be consistent with the theory and for the statement to be false. This is to say that when a theory is tested against the world it is tested as a whole. And when it goes wrong, which is to say it yields incorrect predictions, the adjustments needed to correct it might appear anywhere in the theory– including its most basic formal or logical elements. In the wake of quantum physics, which demanded revisions of our most basic notions of causation, this was a point that had more than theoretical interest. For Davidson, as a Quinean, there are no a priori truths-every part of our theory is open to revision. It is true that we have no clear idea of abandoning and replacing core elements of a theory, such as the principles of logic or decision-theory would be. But in theory at least, something can come along that could compel us to abandon them. One odd consequence of this adjustability of theory to evidence was the notion of underdetermination: more than one theory would turn out to be consistent with the same set of observation sentences, because there could always be adjustments to the theory that preserved the truth of the observation sentences but made the adjusted theory inconsistent (for example by introducing new theoretical entities) with the initial theory. Quine introduced a notion of translation that was congruent with this holism. If we think of a language as like a theory, that is to say as a set of true sentences with some logical connections between them, we can think of translation as the construction of sentences in one language that preserve truth and logical connections in the other language. But there is an oddity here as well: more than one translation would always prove to do the job. In his famous example, sentences using either ‘rabbit’ or ‘undetached rabbit part’ would have the same truth conditions, but ‘mean’ different things in the language of translation. Quine talked about this in connection with what he called radical translation, by which he meant translation from behavioral evidence alone. Davidson wrote an article in this vein called ‘Radical Interpretation,’ which asked a somewhat broader question, having to do with the propositional attitude of believing to be true-something that went beyond the behavioral, but which was essential to interpreting the meaning of an utterance. As Davidson put it, we can ascribe meaning only if we hold belief constant, or assume that the speaker believes what is uttered (1984: 135). To do other than to merely assume that someone is speaking truly we need to make this assumption into something more strongly based. And this requires us to connect one action of truth-telling to other actions and other utterances holistically, in a way that gives us grounds for thinking that a particular utterance is believed by the speaker to be true. These connections between action and belief are rational ones. So we must ascribe rationality. Without a conception of rationality, this connecting up can never get off the ground. Without a normative conception of rationality we have no conception of rationality, since the alternative models of reasoning, for example those provided by empirical psychology, depend on the normative conception. Thus interpretation is inherently normative. For Davidson, as for Quine, the problem of understanding other cultures, the analogue to Weber’s notion of understanding subjective meaning, is one of understanding their sentences. To understand is to construct a translation manual, something that is not possible unless many of the sentences that are true for them are true for us. This is a conception that does not require a ‘theory of meaning’ and indeed, as we shall see, precludes the kind of theory of meaning in which meanings are ‘out there’ at some sort of symbolic level, as Cassirer thought, or are part of a normatively structured language. Our access to the meaning of their sentences through translation is through determining what sentences they take to be true. Thus understanding, more generally, is a matter of triangulation, in which we check utterances against both the conditions we take to be the truth conditions of the sentence and also against the truth conditions that others take to be their truth conditions. This is meaning in a surface sense, and there is no more meaning to be had. This ‘triangulation’, in which holding belief and truth (and thus semantic meaning) constant, together with an assumption about rationality, enables the interpreter to make inferences about intention, and, similarly, holding any of the other two constant enables inferences about the third, became basic to Davidson’s later writings. Weber’s approach was different, as we have seen because it did not involve the same holism. Ideal-types applied more or less well to the facts in question, such as the expression on the face of the angry man, and made sense of them by identifying the supposed subjective meaning the actions had for the agent. But the types were particulars. Only in the case of the ideal-type of fully rational action would the logical relations between beliefs be relevant to the attribution of subjective meaning. In other cases, such as the angry face of a man, the facts to be found in the context determine the appropriate attribution of meaning. But even this distinction turns out to be less significant than it first appears. For Weber, the corresponding steps were made in terms of typifications. There is no direct comparison between some sort of independent ‘reality’ and our ‘typifications’, because the objects we inquire into are already partly conceptualized or typified. The elements of triangulation are implicit in the task, which he stressed, of making overt behavior meaningful to a particular historically located audience of intelligibility seeking hearers with their own armamentarium of typifications, who are practiced in applying them. There are checks on our applications in ordinary life: If we mis-typify in ordinary life, we are likely to discover that our expectations are wrong, and our actions in the rest of the web of typifications based on these wrong expectations will lead to difficulties with others. If we mis-typify an act as involving the intention of truth-telling, for example, we can revise our typification if new contextual evidence supports a different interpretation. This is the stuff of triangulation, in different dress. For Weber as for Davidson there is no getting beyond useful typifications to essences, or to a ‘reality’ that can serve as a standard. For both of them, the indispensable starting point is with us, and with our problem of making things intelligible, and here the standards of adequacy of interpretation are necessarily ours. We can refine them, but there is no God’s eye view which our refinements will reach. This leaves the problem of the relativity of values, which Davidson denied, and Weber affirmed, and the problem of the categorical character of the logic-value distinction. Just as Quine saw the problem of the analytic synthetic distinction as a matter of degree of entrenchment, Davidson saw the problem of values in terms of degrees of dispensability rather than categorical differences. He acknowledges that there are genuine moral conflicts ([1995]2004: 41), which of course are the centerpiece of Weber’s ‘relativistic’ discussion of values, and these seem to include Weber’s own cases. But he also notes that evaluative attitudes have as their objects propositions that have truth value ([1995]2004: 55), and argues that the issues of intelligibility that constrain our interpretations of other cultures arise here as well, so that ‘the more basic a norm is to our making sense of an agent, the less content we can give to the idea that we disagree with respect to that norm’ ([1995]2004: 50). This means that the norms of basic logic and the like don’t lend themselves to ‘disagreement’ claims, while the cases of value choice cited by Weber do. The difference is thus slight: both agree that some core elements of reasoning are immune to relativistic choice. Weber thought this was a result of a principled difference between these norms and values, but as his later critics pointed out tirelessly, he could not give a principled non-relativistic ground for this assumption. If Weber had Quine’s ‘Two Dogmas of Empiricism’, and extended it as Davidson did, Weber too could have treated the issue as a matter of degrees of dispensability, as he treated the ideal-type of rational action itself. This is an important result. It allows us to give a sense to Weber’s problematic phrase ‘the norms of our thought’ (1949: 84), which appears in the middle of his explication of the relation of (relative) evaluative starting points and results that are valid for all. The phrase can be treated as a reference to the validity-defining elements of our thought whose indispensability to understanding makes it difficult to give content to the idea that we disagree with respect to them. They are thus ‘ours’ but their universality follows not from some special logical feature but from their de facto indispensability to the understanding of others. This emendation is in the spirit of Weber’s sense that the application of such concepts as ‘rational action’ is a matter of degree rather than absolute-the feature of Weber’s thought that is usually seen as most antiquated. Yet by applying this reasoning to the core of the methodology, to the distinction between what is universal and what is relative, Weber could have avoided the main philosophical objections that have haunted him since the publication of the methodological essays a century ago. It depends on no problematic fact-value distinction, and no problematic metaphysics of ‘meaningless chaos,’ and no groundless distinction between ‘logic’ and ‘culture’.Yet to close the gap with Davidson in this way gives up nothing of Weber’s view of social science, or of knowledge generally. Nor, in a practical sense, does it give up anything about his approach to the problem of value choice that was a concern for him. The main objection to considering Weber as a significant philosopher arose from the fact that neoKantianism’s answers to these questions, which Weber’s position appeared to depend on, was groundless and thus ideological. In the emended form it is neither. Does the emendation do violence to Weber’s strong sense of the difference between matters of value and matters of logic? Perhaps not, and here we can find additional grounds for closing the gap with Davidson. In a rough sense, the fact-value and logic-value distinctions correspond to the differences in degree of entrenchment in relation to interpretation that are central to Davidson. ‘Values’ of the kind that were of interest to Weber, life-defining value-choices, obviously, as Weber insisted, do vary. Yet these choices, as Davidson would point out, are to some extent intelligible to us, and Weber was himself a master in making them intelligible in our terms. One is tempted to say that he made them into intelligible errors. But although Weber insisted on the diversity of ideals and the inability of reason to choose between them, in one respect he was a closer to universalism in ethics. When he spoke of evil, he did not regard it as a matter of value-choice. And this in an odd way anticipates Popper’s ‘negative utilitarianism’, which is built on the congenial premise that ideals are extremely diverse, but harms, though themselves often bound up with ideals that diverged, are less so. Davidson might say that without an understanding of what harms people we could not understand them. One suspects that Weber would concede this as well.