Naturalisms and Antinaturalisms
To appear in T.May & M.Williamson (eds), Knowing the Social World (Open University Press)
by William Outhwaite
Abstract
Perhaps the most pervasive question in relation to the systematic study of the social world is whether such study should be seen as roughly similar to the study of 'nature' or fundamentally different. More concretely, philosophers and social scientists have endlessly pondered whether the explanation (or understanding) of social processes is something radically different from the explanation of natural processes.
Many of these debates have been shown to rest on inadequate and misleading models of causal explanation, but the adoption of more adequate realist models framed in terms of tendencies, liabilities and mechanisms does not in itself resolve these dilemmas. Rom Harré and Roy Bhaskar, for example, are fairly close to one another in their accounts of natural science but diverge widely in their programmes for social science. This chapter will outline some of these dilemmas and illustrate them with examples from the history of, and philosophical reflection on, social science.
Introduction
If we assume that there is a domain which can be usefully discriminated from the (rest of the) material world under the description 'social' or 'cultural', the question immediately arises whether such study should be seen as roughly similar to the study of 'nature' or fundamentally different. The position generally known as naturalism denies that there is any such fundamental difference; the position which affirms a radical difference between the natural and the social sciences has often been called methodological dualism and I shall refer to it here as antinaturalism.
Most versions of antinaturalism rely on one or other, or some combination of the following claims:
- The social world is intrinsically different from the (rest of the) material world; thus our knowledge of it will be fundamentally different.
- our 'cognitive interest' in the study of the social world is radically different from our interest in the rest of the world, and thus our knowledge of it is radically different.
Although the first of these claims has an ontological flavour, and the second an epistemological or methodological one, in practice they tend to interpenetrate and reinforce one another.
1. History
The first move in this debate substantially predates the emergence of the 'social sciences' as we know them. Giambattista Vico (1688-1744), in his Scienza Nuova of 1725 and 1744, formulated the basic principle that our knowledge of what we ourselves have made (individually or collectively) is different from what we have not made. Thus the world of human society and culture is in some sense 'our' product; the rest of the universe is attributed to a residual maker known as God. 'We' know the human cultural world because 'we' have made it; only God, whose omniscience of course includes perfect memory, can know the rest of the world in this way.
This distinction is taken up in a rather more speculative form in Hegel's differentiation between reason (Vernunft) and the understanding (Verstand), though the maker here is the world-spirit coming to recognise its own productions (including, ultimately, the world itself) and its learning-processes as rational, in contrast to the essentially contingent states and relations found in nature and described by the mathematical and natural sciences. Something more like Vico's idea returns again in the second half of the nineteenth century, when Hegel's concept of objective mind is extracted from its surrounding developmental 'grand narrative' and treated more as a descriptive category, which is then used as the basis for a critique of the original (Comtean) version of positivism. (As the philosopher of history J.G.Droysen wrote in 1852, in terms which became familiar again just over a century later, 'Crass positivism unfortunately is finding great support...'(Spieler, 1970: 20 n.6.)).
To put matters very briefly, Droysen and his contemporary Wilhelm Dilthey (1833-1911) developed an antinaturalist programme based on the distinctiveness of human psychic expressions - the first of the two claims described above. In a more methodological vein, two other neo-Kantian thinkers, Wilhelm Windelband and Heinrich Rickert, focuses more on the second claim. The study of culture, they argued, is essentially interested in individual processes and in relating them to shared human values, whereas the natural sciences are concerned with general laws concerning objects which are essentially remote from questions of value. In other words, we are interested in the French revolution not just as a member of a class of revolutions exhibiting certain common features (this would be, for Rickert, a natural-scientific mode of approaching it), but as a unique event embodying, and perhaps violating, certain crucial human values.
This opposition between positivism and methodological dualism comes to structure the emergent human or social sciences, as the term culture increasingly gives way to society or, for those suspicious of the objectivistic overtones of the latter term, 'sociation' (Vergesellschaftung in German). There is a fairly strong direct line of influence from Rickert to Max Weber, and also through Georg Simmel, who discusses our knowledge of the social world in particular in his Problems of the Philosophy of History(1892) and in his Soziologie (1900), notably in a classic 'excursus' on the question 'How is Society Possible?' - an essay which can now be seen as the founding text of social constructionist theory.
For Weber, as he put it in an early essay, 'the course of human action and human expressions of every sort are open to an interpretation in terms of meaning (sinnvolle Deutung) which in the case of other objects would have an analogy only at the level of metaphysics' (Weber, 1975:217-8). He therefore later defines sociology, in the first sentence of his Economy and Society, as a science which aims at an interpretative understanding of action in order thereby to understand its course and its effects. Whether by this Weber means that explanatory understanding is itself a form of causal explanation, or merely complementary to it, the crucial point for him is that explanations of social phenomena must be both 'causally adequate' and 'meaningfully adequate'. As he put it in the earlier essay,
'Our need for a causal interpretation (unser kausales Bedürfnis) demands that where an "interpretation" (Deutung) is in principle possible, it be carried out; i.e. for the interpretation of human behaviour it is not sufficient for it to be related to a merely empirically observable regularity (Regel des Geschehens), however strict that law may be' (Weber, 1975: 128)
It is clear, I think, that for Weber our access to knowledge of the social world is importantly different from our knowledge of nature. It is not, however, in his view any less objective. He heroically attempts to hold together Rickert's principle that our perspectives on cultural phenomena, and our knowledge of them, are shaped by values (and for Weber, unlike Rickert, values are a matter of ultimately ungroundable existential choices), with the idea that the social sciences can attain a bedrock of solid and 'value-free' knowledge which would have to be accepted, as he sometimes curiously puts it, 'even by a Chinese'. But for all that one can distinguish between explicit value-judgements and the establishment of a relation between social phenomena and a potentially infinite set of possible value-standpoints - a position which one finds restated in Karl Mannheim's Ideology and Utopia of 1929 - the suspicion remains that in the end Weber is telling us how to steer a straight course in an ultimately arbitrary direction. Faute de mieux, of course, this is quite a good strategy to adopt if you are lost in a wood or at sea but, together with Weber's concept of the ideal-type, whose explanatory adequacy is again a matter of ungroundable judgement, it puts an uncomfortably heavy burden on the sensitivity and finesse of the individual researcher.
For this and other reasons, Weber's synthesis was pulled apart in both directions in the decades following his death in 1920. First, there was now a more stridently naturalistic and indeed reductionist variant of postivism: the logical empiricism of the Vienna Circle, in whose 'unified science' (Einheitswissenschaft) the statements of all sciences should be ultimately reducible to material-object language or to statements in physics, and verstehen was of no more importance, in Otto Neurath's vigorous formulation, than a good cup of coffee which sustains the social scientist. From this perspective, as Hempel and Oppenheim (1936) argued, the concept of ideal type was a mere preliminary to the serious business of constructing empirical indicators for theoretical concepts, and explanations in the social sciences and history could be shoe-horned into a covering-law model (c.f. Outhwaite, 1987b) . This formed what has been called the 'standard view' in Anglo-American philosophy of social science in the middle decades of the century.
From the other direction, Alfred Schütz initiated the tradition of social phenomenology with a book published in Vienna in 1932 with the title Der sinnhafte Aufbau der sozialen Welt (The meaningful constitution of the social world) a title chosen for the sake of the contrast with the empiricist philosopher Rudolf Carnap's Logische Aufbau der Welt (1928). Schutz felt, unlike Hempel and Oppenheim that the problem with Weber's ideal types was not that they were insufficiently scientific, but precisely the opposite: Weber was too quick to impose them on the phenomena he described, paying insufficient attention to their grounding in acts of typification performed by ordinary members of society. In other words, the social scientist is constructing second-order typifications based on those already carried out in the lifeworld. This theme was taken up by the phenomenologist Aron Gurwitsch in the United States, where Schutz had also settled, and by Peter Berger et Thomas Luckmann (1961), who finally put the term 'social construction' on the map, offering in the guise of a sociology of knowledge a paradigm more directly adapted to use in social research. By the time Schutz's book was republished, in Germany in 1970 and in the United States in 1967, the way had been prepared by Schutz's own later work (his Collected Papers appeared in 1962-6) and by the influential work of Berger and Luckmann.
Social phenomenology also of course had affinities with the well-established minority North American tradition of symbolic interactionism, which also experienced a certain resurgence in the 1960s with, for example, the republication of the work of G.H.Mead (1962, 1964) and that of Herbert Blumer (1969), and also the publication of various studies by Erving Goffman (1959-). But where the interactionists tended not to spend time on formal critiques of empiricism, Harold Garfinkel's ethnomethodology (1966) was more aggressive. A former student of the structural functionalist Talcott Parsons, Garfinkel was led by his study of the deliberations of a trial jury to emphasise, like Schutz, the importance of practical reasoning in everyday situations. He showed, in a kind of implicit hommage to the Weberian principle mentioned earlier, that our drive to interpret leads us to impute a meaning even to random processes in the social world, as in one of his experiments in which an interviewer playing the role of a counsellor replied to the clients' questions by a random 'yes' or 'no', leading them into more and more contorted interpretative efforts. The production of meaning is at the same time the production of social order - Parsons' major concern. Unlike his former teacher, however, Garfinkel insisted that social actors are not simply bearers of their social roles ('cultural dopes'), but active subjects obliged to practice social analysis in order to function in everyday society.
Social phenomenology found a somewhat unexpected ally in a convergent move within analytic philosophy. On the margins of the Vienna Circle, Ludwig Wittgenstein had drawn the main intellectual inspiration for his brilliant Tractatus Logico-Philosophicus (1921) from Frege, Russell and Whitehead. Coming to abandon the simple conception of a picturing relation (abbildende Beziehung) between propositions and the world, Wittgenstein was drawn into a more sensitive and holistic analysis of the practicalities of 'language-games' based on implicit rules and themselves embedded in what he enigmatically called 'forms of life'. Where Frege had insisted that it is propositions rather than individual words which have meaning, Wittgenstein went further to direct our attention to the use of words and expressions in more complex ensembles of what came to be called speech-acts and other forms of human practice. In 1958, an important book by the Wittgensteinian philosopher Peter Winch, The Idea of a Social Science and its Relation to Philosophy (Winch, 1958) drew the consequences for social theory, again using Max Weber as one of the foils for his argument. For Winch, knowing a society means learning the way it is conceptualised by its members. He thus revived the central principle of 19th century German historicism, according to which every age must be understood in its own terms. Winch explicitly cited Lessing in his frontispiece "...the same moral actions do not always have the same names, and it is unjust to give any action a different name from that which it used to bear in its own time and amongst its own people".
Winch went even further. He directly identified himself with the German idealist tradition by insisting that social relations are "like" logical relations between propositions (1958: 126). He also identified himself more concretely, in this book and in a later article , 'Understanding a Primitive Society' (1964), with an ethnographic field-work approach.
Hermeneutic theory also took a new turn with the philosophical hermeneutics of Hans-Georg Gadamer, whose Wahrheit und Methode, published in 1960 and translated into English in 1975, insists, in opposition to historicist hermeneutics, on the practical dimension of interpretation, conceived in the Heideggerian sense of an encounter (Begegnung) between the 'horizon' of the interpreter and that of the text itself.
Gadamer's radicalisation of hermeneutics was taken up in Jürgen Habermas' version of 'critical theory', which can be seen, along with certain realist positions and Anthony Giddens' structuration theory, as one of three particularly influential attempts in the final third of the twentieth century to reconcile, as Max Weber had done, the rival claims of explanation and understanding. At the outset of his career, Habermas participated in the famous 'positivism dispute' (Positivismusstreit) which began at the German Sociological Association conference in 1961 and ran on for some years. The initial exchange between Theodor Adorno and Karl Popper was continued by their respective supporters (Adorno et al, 1969). Habermas, with the same narrative skill with which he traced the history of positivism and its critics in Knowledge and Human Interests (1968), shaped the three thinkers discussed above, Schutz, Winch and Gadamer, into an argument which he presents in On the Logic of the Social Sciences (1967). The positivist thesis of unified science falls in Habermas' view, because of the intimate relationship between the social sciences and history, and the fact that they are based on "a situation-specific understanding of meaning that can be explicated only 'hermeneutically' (Habermas, 1967: 43). As he put it some ten years later:
"... the problem of 'understanding' (Verstehen) in the human and social sciences only achieved its great methodological importance because access to a symbolically prestructured reality cannot be gained by observation alone, and because the understanding (Verständnis) of a participant cannot be so easily subjected to methodological control as the perception of an observer" (Habermas, 1982: 549).
Habermas therefore reviews the three principal modern approaches to verstehende sociology: phenomenological (Schutz), linguistic (Winch) and hermeneutic (Gadamer). Each of these approaches goes beyond or transcends the previous one(s). In "the phenomenological approach [which] leads to an investigation of the constitution of everyday life-practice...language has not yet been understood as the web to whose threads the subject hang and through which they develop into subjects in the first place (Habermas, 1967: 117)." Similarly, Winch's linguistic approach which "concentrates on language games that at the same time transcendentally determine forms of life" neglects the Gadamerian requirement to mediate between alternative frameworks. "Winch seems to be contemplating a linguistic version of Dilthey. From his free-floating position the linguistic analyst can slip into the grammar of any language game without himself being bound by the dogmatism of his own language game, which would be obligatory for linguistic analysis as such" (1967:136). Finally, the hermeneutic approach opens up the issue of the practical fate and historical effectivity (Wirkungsgeschichte) of traditions, and thus points beyond hermeneutics to a sociological account of the power relations which shape and distort human communication. Habermas, along with his close collaborator Karl-Otto Apel, argued for a complementarity between an empirical-analytic approach oriented to the explanation, prediction and control of objectified processes and a hermeneutic approach concerned with the extension of understanding, in an emancipatory model of critical social science, instantiated by psychoanalysis and the marxist critique of ideology, which aims at the removal of causal blocks on understanding. Although Habermas has since come to abandon this model in its details, the basic idea remains in his more recent theories of reconstructive science and communicative action. Critical theory, it is worth noting here, is antinaturalist by inclination, but in an increasingly muted way.
In Oxford, at much the same time, the late 1960s and early 1970s, Rom Harré and Roy Bhaskar were developing their realist metatheory of science, drawing in particular on the work of Mary Hesse. Both Harré and Bhaskar, like Habermas, were substantially motivated by the desire to undermine positivistic theories and approaches in the social sciences. Harré and Secord (1972) developed a philosophy for social psychology based on the work of Wittgenstein and the analytic philosophy of language practiced at Oxford by J.L. Austin. Ordinary language, they argued, is better suited to the description of the mental processes of social actors than apparently more scientific artificial terminology, and they drew attention to models of research practice of this kind in the work of Goffman, Garfinkel and others.
Harré and Bhaskar were in any case interested in giving a more adequate account of science as a whole. For Bhaskar, accounts of the possibility of science had traditionally focused on facts about us: our experiences for empiricism, our cognitive structuring capacities (rationalism) or both (Kantianism). A more adequate account, he suggested, would have also to reflect on what had to be the case in the world for science to be possible - namely that the world be composed of relatively enduring structures and mechanisms, some of which could be isolated in scientific experimentation, given the entirely contingent emergence upon the earth of homo sapiens and homo scientificus. Thus a transcendental argument from the nature of science as an intelligible activity sustained the idea of a structured universe. The epistemic fallacy, for Bhaskar (1975), consisted in the swerve from direct questions about the nature of reality, most of them properly the concern of the sciences, to anthropocentric questions about the possibility of our knowledge of reality, leading inevitably, once you grow out of the search for protocol sentences, to the dead ends of scepticism and conventionalism.
An important aspect of the realist programme developed by Harré, Bhaskar and others was a conception of explanation as involving not an essentially semantic reduction of causal statements to covering-laws plus initial conditions but, in what Harré later called a 'referential realism', a reference to the causal powers of entities, structures and mechanisms. The constant conjunctions of events which, for empiricists following Hume, were all that we could know about causal relations, were in fact neither sufficient, given the problem of induction, nor necessary for the justification of causal statements. Such law-statements should properly be understood as statements about tendencies which might or might not be outweighed by countervailing tendencies. Two causal tendencies may neutralise one another, as do the centrifugal force of the earth's rotation and its gravitational attraction, with the convenient consequence that human beings and other animals are safely anchored to the earth's surface.
This and other features of realism meant that, as Russell Keat noted in a classic article in 1971, the whole issue of naturalism could be rethought. Human beings could be seen as having causal powers and liabilities, just like other entities; the fact that their relations rarely sustained any universal generalisations of an interesting kind, but rather sets of tendencies regular enough to be worth exploring. The fact that many of the entities accorded causal force in social scientific explanations were necessarily unobservable was not, as it was for empiricism, a problem of principle. Finally, in a move which I shall return to later, it would seem natural to include among the causes of human action the agents' reasons for acting.
The realist critique of traditional epistemology found an echo in social theory, notably in the work of Anthony Giddens, who had become similarly impatient with the residues of positivist social science as well as the more radical contentions of social constructionism. As Giddens put it in The Constitution of Society (1984), p.xx,