Tuesday, March 18, 2014

Winch's 1990 preface

The second edition of The Idea of a Social Science has a preface in which Winch outlines what he did and did not mean to say, and gives some indication of how the book might be different if he were to re-write it. He acknowledges that "many things have changed both in philosophy and in the social sciences" (p. ix) between 1958 and 1990, but says also that, "It is not that I think the deep-seated errors and confusions which I tried to expose are no longer active" (p. ix). Helpfully, he also says that, "The central core of the argument is really stated in Chapter III, Sections 5 and 6." So, back to those sections.

What really seems crucial here is section 6, which begins with this paragraph:
On Mill's account, understanding a social institution consists in observing regularities in the behaviour of its participants and expressing these regularities in the form of generalizations. Now if the position of the sociological investigator (in a broad sense) can be regarded as comparable, in its main logical outlines, with that of the natural scientist, the following must be the case. The concepts and criteria according to which the sociologist judges that, in two situations, the same thing has happened, or the same action performed, must be understood in relation to the rules governing sociological investigation. But here we run against a difficulty; for whereas in the case of the natural scientist we have to deal with only one set of rules, namely those governing the scientist's investigation itself, here what the sociologist is studying, as well as his study of it, is a human activity and is therefore carried on according to rules. And it is these rules, rather than those which govern the sociologist's investigation, which specify what is to count as "doing the same kind of thing" in relation to that activity.
The last sentence need not be true. The sociologist can choose to count as the same thing any two things at all. Winch's point is that what actually interests us (presumably and for the most part) is activities as they are understood by those who engage in them. If we want to we can study mere hand-washing, but chances are if we are interested in "understanding human behavior" or "understanding a society" then what we mean is that we want to understand the rules that the people we study are following. And a Muslim who washes before prayer is not following the same rules as a surgeon who washes before getting to work. The criteria for performing the action correctly are (or at least may be) different. In a sense one must already understand what people are doing before one studies their behavior.

Is this a problem? I don't think it has to be. Understanding what something is need not answer all our questions about why it happens. For instance, we might want to know whether hooliganism increases or decreases when unemployment goes up. To answer a question like this we need to know what does and does not count as hooliganism, but then there is nothing to stop our counting instances and comparing them with unemployment rates. The main problem with an inquiry of this kind, I think, would be knowing whether any pattern discerned was a case of causation or mere correlation. Another problem would be knowing exactly what was meant by 'causation' in a case like this. Let's say unemployment tends to increase hooliganism. Still, not all hooligans are unemployed and not all unemployed people are hooligans. And unemployment is unlikely to cause hooliganism in the way that it might (be said to) cause attempts to get a job. A job applicant can explain that he is applying for a job because he is unemployed. A hooligan who sincerely blames his behavior on unemployment is (almost certainly) hypothesizing in a way that the job applicant is not.

I'm not sure what to say about this, but I think that correlations like this could be found and might be quite suggestive. The question that people like Winch and Read might ask is whether finding such patterns really counts as doing science. As long as we see the difference between this kind of case and what physicists do then it probably doesn't matter. But the difference is large. Big enough not only to leave room for doubt about the underlying or true cause, but big enough also for the related but not identical phenomenon of moral disagreement. The moral sciences are not the same as the natural sciences. If atoms or bodies in motion do this or that then there is no room for moral evaluation of their behaving in this way. If violent crime increases when income inequality increases (as I expect it does) this can be appealed to as an excuse or mitigating circumstance, but it does not remove the intelligibility, the possibility, of blaming the criminal. If there is causation here it is not of the same kind as the billiard ball kind. This itself is a fact about how we use the concepts relevant to making moral judgments, a fact about human behavior.

Although here things get complicated and/or confusingly meta. One person might insist that if there is enough evidence of a correlation between inequality and violence then we have no option but to accept that the former causes the latter, and that anyone who denies this is speaking nonsense, is simply unintelligible. Another person might disagree and say that there is room for reasonable people to take various views on the matter. In other words, simply describing uses of language without getting involved in the debate is not simple, and is perhaps not possible in every case. There is not always an undisputed fact about what makes sense and what does not. But often there is. And even when the space of possible meanings is disputed territory, diplomatic negotiations might still be possible. Not necessarily with a view to resolving the moral or political disagreement (although that might be possible too), but simply with a view to reaching agreement about who is saying something, however laughable or repugnant, and who is not. The possibility that attempts at description will get personal and/or evaluative does not rule out the possibility of (impersonal, neutral) description. This goes for both philosophy and social science. But it does mean that in an important sense we are not doing science in the way that physicists are doing science.


  1. Tell me if you think this is right: We have two ways of describing human behavior—internal and external. And if I understand that last sentence in your quotation from Winch, there is an ideal implicit in Winch of description of human behavior, namely that it all be internal. Do you agree that he has this ideal? Would you agree that he is making internality here a criterion for adequacy for description of behavior? And if so, would you say that this ideal is unrealistic or even a fantasy? – I know you hate using such words. I mean only this: Do you think we can or even should get rid completely of external forms of description of human behavior? Do you think for instance that we can or should get rid of words like “hooliganism” and “administration” and “working class,” which have their home in external modes of descriptions of human behavior?

    The fact that you say things like: “correlations like this could be found and might be quite suggestive” seem to indicate that you don’t think external forms of description are inherently problematic. Is that right? – If so, why do you think Winch thought they were (if you do)? – I can think of two (internally related) sorts of argument.

    (A) Moral argument: There is something morally wrong about external description because they disregard the humans behind the phenomena they describe. (Arendt makes a similar argument, I think, when she says that in the social sciences people are treated “functionally”: a person for the social scientist is never the full human being, but always only its function in a certain social structure.) This, it might be said, may even lead, as I think you hint, to the disappearance of morality altogether—of notions like responsibility and guilt.

    Even if we would want to reject this criticism, there are milder forms of it. For instance, the claim might be that the social scientists are typically unaware that something important is lost in external descriptions, and they are not in the habit of taking it into account, or of warning their audience, that the external angle from which they approach the phenomena might be morally skewed.

    (B) Logical argument: The external forms of description are unintelligible. Much of what you say goes against this conclusion. But might there be something in this still? What if Winch said something like: “Look, intelligibility is a function of what we need. And what we get from external, or functionalist, descriptions are is never ALL that we need. Such descriptions can never be fully satisfying, because (as you mention, and this is connected to the McManus criticism too) what motivates such research in the first place is an interest in humanity. So, for instance, then claim might be that hooliganism as such is not of interest to us, but only as it can be connected for us to the behavior of individuals—shed light on motivations and states of minds of real people.”

    Now, it is possible that this is false, and that this is not the real motivation for social scientific research. But if so, can Winch argue against the social scientist that they have never fully clarified to themselves what it is that they expect, and what it is that motivates their discussion, and that they hover undecidedly between a reductionist and a non-reductionist understanding of their own research? Would he be able to say that the social sciences often relapse into claiming they achieve much more than they really do, and perhaps that the only thing that keeps those research platforms alive is this illusion that they achieve more than they actually do?

    1. Thanks, Reshef.

      The internal/external distinction is helpful. I'm not sure what Winch would say, especially the 1990 Winch who doesn't say a lot. But I think that he does have an ideal of internality, yes, which relates to his ideas about what is interesting. I don't share that ideal, but I have some doubts about just how interesting suggestive findings of correlations might turn out to be.

      I think the answers to the questions you pose in your final paragraph are Yes and Yes. And this is very important.

  2. I have no serious views about what is and isn't science, but a sense that there are many sciences that have different primary subject matters, different methodologies, etc. At an abstract level, they all look at data, and try to understand something based upon looking at data.

    With that all sorted out.... I wonder what worries/concerns are driving your post. Is it something like this: that if social science (is a real thing and) studies causation in the social world, then the implicit determinism creates problems for making sense of personal accountability? That seems like a worry. But what immediately struck me in the hooliganism example is that one can simply say this about the correlation between rising unemployment and increased hooliganism: that rising unemployment creates social conditions in which more people will choose to engage in hooliganism. So there's more than a mere correlation, but the TREND doesn't force any particular (unemployed) person to start knocking over dustbins in Shaftsbury. (Old Bill Hicks joke about British hooligans vs. West Coast gangs.)

    Sorry if this isn't very helpful; I imagine I'm seeing only one tree in the forest of this post...

    1. Thanks, Matt.

      The main thing driving the post is just a desire to understand (and, I suppose, evaluate) what Winch says. I have a lot of sympathy for his view, but I also feel that a lot depends on what you call science, what you call a cause, and so on. (As I'm sure he would agree.)

      The main problem in the hooliganism case, it seems to me, is that once a correlation between unemployment and hooliganism is found (assuming that it is) then what justifies us in saying that it isn't merely correlation is something that looks very much like common sense, based on a kind of imaginative sympathy with the unemployed and with hooligans. And the role of that kind of sympathy makes this kind of study, as valuable as it may be, significantly different from chemistry or physics. Gathering data and analyzing it might be quite like other sciences (though also like history, for instance), but the role of human understanding is important and distinctive.

      If you or anyone else says, "So what?" then there might not be a lot more to be said. But not everyone has responded to Winch so nonchalantly.

    2. Oh, I don't mean to say "So what?" (for one, I haven't read this book, though it's been sitting on my shelf for quite some time...). Looking back at the paragraph you quote, however, I'm not sure that I understand Winch's point at the end. Is the point that one can only achieve an "insider's understanding" (e.g. of what counts as doing the same thing) if one judges based on the rules governing that activity or institution? And that, by comparison, there is no "insider's understanding" of the activities of electrons? Ok, I see that. But then I think I agree with you that, depending on what we are trying to do with our social science, we may not want to be beholden to the insider's rules for what counts as doing the same thing--especially, for example, if were were trying to make cultural comparisons or perhaps looking for structural similarities for different practices within a society that are, from within it, regarded as different or opposed. E.g. we might look at certain rituals or traditions performed by different sects within a larger culture and see these as different ways of doing the same thing, at a more generic level of description. Of course, such comparisons might be good or bad (strong or weak), but is that perhaps the sort of idea that Winch is resisting?

      You say: "In a sense one must already understand what people are doing before one studies their behavior." And that makes sense. But then couldn't one say that one must already understand certain things about electrons before one studies their behavior? (What I mean is, any hypothesis driven study is going to presuppose a background of information, e.g. gained through some purely observational study. Now, of course, I could be wrong about that, but it's a hunch.) That of course wouldn't even come close to suggesting that studying electrons is just like studying humans and human institutions, and perhaps your point about imaginative sympathy gets at an important difference (since it may help me to get into the heads, as it were, of other humans in a way that it won't help me get into the head of an electron--was that part of the point?).

    3. Oh, I don't mean to say "So what?"

      I didn't mean "so what?" in a bad way, just a kind of "OK, let's call it social studies rather than social science if you prefer" way. But anyway...

      we might look at certain rituals or traditions performed by different sects within a larger culture and see these as different ways of doing the same thing, at a more generic level of description. Of course, such comparisons might be good or bad (strong or weak), but is that perhaps the sort of idea that Winch is resisting?

      I don't know how much Winch would resist something like this. I suspect he would wonder what the point was, but I don't see a reason for him to object in principle.

      perhaps your point about imaginative sympathy gets at an important difference (since it may help me to get into the heads, as it were, of other humans in a way that it won't help me get into the head of an electron--was that part of the point?

      Yes, that's right. I think Winch sees the main goal of social studies as being to get into the heads of other people. At least this is a main goal.