Reading Guide to: Winch, P (1958) The Idea of a Social Science and its Relation to Philosophy, London: Routledge and Kegan Paul


Chapter 1 Philosophical Bearings

It is common in social science to distinguish between conceptual and empirical enquiries, but in (Wittgensteinian) philosophy, the two are intimately connected. Language constitutes the social world, and therefore conceptual questions are clearly the most important (and the only possible) ones.

Philosophy is far more than an underlabourer for social science, tidying up minor confusions in order to leave the path open for more important forms of inquiry. Philosophy is social science: all inquiries are based on the issue of intelligibility, and this must be the subject of conceptual inquiry. It is possible to see the individual sciences as researching areas governed by specific language games. There is no overall supergame, just as Wittgenstein argues, and so we cannot believe that the results of all these games will add up to some grand overall view of reality captured by social theory [a clear hint of relativism].

All social relations are expressions of ideas about reality. The notion of cause is not appropriate to human action: human beings follow rules, and we can understand them only by asking which rules they are following. Any attempt to ignore the central role of meaning, and to operate with structures or 'social facts', as Durkheim does, leads to error.

In terms of the relations between words and things, there is no easy identity to be discovered. Words have to be used in a consistent way to refer to things, but this is always sensitive to context [ another pointer towards relativism]. Again we need to consider rules to connect words and things, and there is not just a simple formula here -- we need to know how to apply these connections to future cases as well. The rules concerned may not be immediately intelligible, and they can vary to take into account the reactions of others -- that is, there is a social context, or social support for these rules. However, in principle all rules are discoverable. It is still possible to make mistakes, however, and to evaluate rule following, to establish a standard for it. Rules can never be merely a private matter, but must always have a social context. [Winch goes on to argue that the specialist languages of Social Sciences must be, and can only, based on these  shared understandings at the commonsense level].

Chapter 2 The Nature of Meaningful Behaviour

'Forms of life' should be the beginnings of our inquiries. These inquiries are epistemological matters (roughly , concerned with how we can know about the world). Language use is the main characteristic of human beings, although this is usually taken for granted, for example in many sociologies. This leads to some criticisms of explanations in conventional social science.

  1. Firstly, for Weber's sociology -- reasons and actions can be connected, as in his concept of 'subjectively intended action', if and only if a rule is being followed, and this cannot be ignored. Giving a reason for someone's behaviour makes sense only if the concepts in it are grasped by the actor himself -- we cannot say that someone has voted for a party because he felt there would be certain desirable consequences without assuming that the actor understands what it means to vote and what desirable consequences are .
  2. Secondly ,all actions follow rules, even when discussing the effects of the unconscious in Freud: infantile trauma affect us only if we can understand their effects, albeit not very clearly until we undergo analysis. The same points apply to any kind of 'expert' reasons (Weber again), which apparently do not require the understanding of the individuals concerned. Even these explanations 'must be in terms of concepts which are familiar to the agent as well as to the observer' (47), at least when referring to the terms that are apparently connected by the explanation. For example 'It would make no sense to say that... [a person's ] omission to post a letter... was an expression of... unconscious resentment against X for having been promoted over his head... if [the person] did not... understand what was meant by "obtaining promotion over somebody's head"' (48). Weber's category of traditional action also implies that a person must still be able to commit themselves to connect together tradition and a future act.

All these examples must involve rule following, and all refer heavily to a social context. It may not be possible for actors to be fully explicit about these rules, nor for observers to fully specify all the conditions at work. Actors do not need to formulate a rule, merely to be able to use it to decide correct and incorrect procedures: this is how children learn, by reacting to their mistakes, by gradually learning to apply the same criterion reflexively. This is not just a matter of habit, but of practice. Rules which subsequently undergo reflection and formalisation arise out of these practices.

Chapter 3 The Social Studies as Science

John Stuart Mill's notion of a social science is being criticised here, for arguing that we should develop a causal model rather than focusing on meaning, and attempting to see that human actions have causes rather than explaining irregularities in terms of following rules. [see file on Mill] However, the criticisms that follow apply particularly well to any kind of behaviourist explanation, any account their claims to be owed to describe behaviours without reference to meanings.

Social complexity meant that we could only develop probabilistic statements, but Mill argued that in principle the laws of social behaviour could be connected ultimately to the laws of nature, through a notion of laws of mental activity, discovered by empirical psychology. He found these laws of the mind in history -- an inverse deductive approach (and therefore rather dubious -- but actually rather common in sociology too).

Winch, by contrast, suggests that conceptual problems are more important than, and prior to, empirical ones. Social actions are complex because they follow a different rules in language games. Mill confuses empirical changes and conceptual ones. Winch illustrates the difference by considering two problems: (a) the relationship between temperature and the change from water to ice can be established experimentally; (b) what cannot be settled experimentally is a question such as 'How many grains of wheat does one have to add together before one has a heap?', 'because the criteria by which we distinguish a heap from a non-heap are vague [and not just a matter of degree]' (73).

Mill's categories are very commonsensical. It is a puzzle to explain why such simple mechanical models of human beings still exist, since they are clearly not adequate philosophically. [I think this is indicative of the limits of this kind of philosophical inquiry. Philosophers seem quite unable to explain the persistence of things that do not accord with their own standards of argumentation and rationality. Sociologists know only too well that vested interests take the place of philosophical rigour: behaviourism persists, despite its philosophical inadequacies, because to happens to express very well indeed particular views of human action. These views are obviously useful to those who want to control human action -- managers, teachers, behaviour-shapers, as in Foucault's account].

Even when we switch to motives instead of causes, we should realise that motives are more than just prior states of mind. They used in the justification of behaviour, in the construction of rules to explain irregularities, and they express acceptable standards of reasonable behaviour. Motives in this sense are again intelligible only in social contexts.

The same sort of argument applies to the specialist regularities of Social Sciences. They too reflect the consistent application of rules, validated by the social context in which social scientists find themselves, as much as the consistent relation to objects of study [an argument rather similar to Kuhn here]. There should be some connection between the rules of social sciences and the rules of actors themselves -- in practice, social science presupposes these commonsense understandings.

Prediction in social science is clearly inherently difficult because human activity is voluntary. Thus a failed prediction can still indicate a good understanding of current action. We must allow for indeterminacy and uncertain outcomes even when following rules. We thus have a clear difference between prediction in social science and prediction in natural sciences, and it is a category mistake to speak of them as if they were the same.

Chapter 4 The Mind and Society

It is misguided to think that the ideas of participants can ever be discounted, in favour of an expert's explanation, as in Durkheim or Pareto [just  take him as an example of attempts to define human action as primarily logical and rational].

When we consider the work of Pareto on logical action, we see that it is a matter of the correct connection between means and ends: much human action is not logical in this sense. However, the notion of a mistake itself implies the notion of a rule. All human action is rule governed, and so 'non - logical' action simply indicates that some other linguistic game is being played. Pareto really wants to privilege a particular kind of action, that which can be understood by [and imitates] science.

Philosophy should have no such pretensions. It should not prioritise but remain uncommitted in its search for explanations, and, indeed investigate its own explanations. Thus 'reality has no [master] key' (102): 'In Wittgenstein's words, "Philosophy leaves everything as it was"' (103).

Pareto also argues that people commonly do not act on the basis of actual ideas, as their influence is too variable to be an effective determinant -- they often rationalise their actions instead. Again, Winch detects a conceptual distinction here being made from the outside, and protests that such distinctions mean that most people cease to be fully social actors at all [ a bit of definitional privileging by Winch here too?].

This argument applies to Durkheim's famous attempt to explain suicide as a matter of social facts impinging on individuals, but also applies to Weber as well. Weber flirts with causal argument, when discussing the need for causal adequacy. This is simply not needed, and statistics cannot be used to check explanations of action. For example, counting the frequency of words which are used by a person will never help us to understand that person's speech.

Weber's causal analysis is suspect. Meanings do not act as causes, even in a logical sense. Rule following, not causes, produces predictability, and analysing causes implies some unreflected 'external' point of view again [ see below]. Causal analysis is acceptable as a way to assist understanding, to help us to step outside and question what is taken for granted, but it is quite wrong to accord it some priority as a better [ e.g. 'more scientific'] way to proceed.

Weber can at least be defended against charges of operating with mere intuitionism. He is not saying that explanations of the social action of others are based on some inner plausibility for ourselves [as in empathy, which Weber rejects]. He is better understood as beginning to see that we can use subjective understandings as a means to discover social rules. Weber should have pursued his attempts to understand, along the same lines as Wittgenstein does when explaining language use.

Chapter 5 Concepts and Actions

Social relations are best understood as embodiments of ideas and concepts. Social change can be understood as a development within an existing language game, or as the emergence of a completely new language game. As illustrations of the latter, when new concepts do arise, as in medicine, new social relations follow. Similarly, if we were to use numbers instead of names for individuals, we would develop new relations of impersonality which would transform our old concepts and ways of life [Winch uses the term 'debasement' to describe such a change, and refers to social relations being 'impoverished'. Realising that this might contradict the proud project of philosophy to leave everything as it was, he rapidly reverts to describing such events as mere changes -- page 123]. The meaning of concepts are determined by their use, and when we refer to use we are describing social relations.

Social relations must be 'internal' matters of following rules. Any external description cannot avoid a prior conceptual analysis -- they are already theory impregnated (thus to describe something as a response to a question is to imply that the actor has perceived the connection). The actual connections which actors perceive need not be rigorous, conscious, or abstract ones.

This is not to argue that language is some essentialist foundation for action [a possible objection from Popper is actually being countered here]: social relations are logical relations -- but only because logical relations themselves depend on social relations. 'It will seem less strange that social relations should be like logical relations between propositions once it is seen that logical relations between propositions themselves depend on social relations between men [sic]' (126)

Interaction is best understood as being like the exchange of ideas in a conversation, rather than an exchange of forces in a physical system. Of course non-linguistic conversations go on too -- involving dress, non-verbal communication, silences -- but these are closely connected to linguistic exchanges. The symbolic and non-symbolic can never be disentangled: all action follows conventions and rules.

This approach is openly idealist in grasping history as the recovery of subjective meanings. Despite the problems with this approach, such as that subjective meanings can never be recovered and relived easily, it is preferable to positivist history which assumes that we can explain events without recourse to internal relations of meaning. Theory is needed to establish these internal relations in the case of history, although this can never be a substitute for actual participation. In the same way, sociological 'laws' can be suggestive, and work well as analogies -- but they are never sufficient and can never replace the search for meaning.

Now try some criticisms? Try Gellner for a general one.