Structure / Agency

October 4, 2010

This probably isn’t something I should bother people with – but I’m going to express passing frustration anyway at the ‘opposition’ between structure and agency sometimes discussed in the social theoretic literature. Presumably I’ll be able to articulate this point more rigorously at some later date; but since it seems pretty clearly true there doesn’t seem much harm in saying it this way:

Social structures only exist as persistent patterns of action by agents. There is no opposition between structure and agency. If you are talking about structure, you are already talking about agency.

Glad to get that off my chest. If there are any other perennial issues in social theory that I can resolve in a short blog post, please let me know.


7 Responses to “Structure / Agency”

  1. Mike Beggs Says:

    Hey Duncan – I would agree that it’s a false opposition involved in thinking there is a tension between structure and agency. But I don’t think you’re necessarily talking about agency when you talk about structure. I see structure as emergent out of human activity, but not reducible to it, and of course often not intended by agents – and it’s just as true that agency depends upon structure. They are at different ontological levels, and I don’t think you are necessarily talking about agency when you talk about structure, just as you’re not necessarily talking about cells when you talk about human bodies.

    Anyway, I’ve been binging on social theory in my attempt to tackle Poulantzas, so I’d be interested to discuss…

  2. duncan Says:

    Thanks Mike. I think this is probably mostly a terminological / framing thing (as indeed may be much of my beef with structure / agency talk). E.g. on this – I see structure as emergent out of human activity, but not reducible to it – I guess it depends what we take ‘reducible’ to mean. I tend to think that if a phenomenon is emergent out of other phenomena, it is in some sense reducible to those phenomena – i.e. it is ‘nothing but’ those phenomena, in their complicated aggregate action and combination (even if we don’t realise that). (This is what I take ‘emergent’ to mean, basically – I guess I’m operating with a concept of ‘weak’ emergence.) It’s not required that the agents intend the emergent structure, of course – I definitely agree with that. But you also don’t get something ‘extra’ when a structure emerges, other than a pattern in the simpler phenomena the structure’s emerging out of. The human body being composed of cells is a good example, I think: I take it that you are necessarily talking about cells when you talk about the human body, because the human body is composed of cells? Now it’s possible to talk about the human body without knowing you’re talking about cells, or without meaning to talk about cells – before the invention of the microscope and the development of cell theory, this would have been the case for everybody, for instance. But one was still referring to a group of complexly bound together and interacting cells, when one referred to a human body, even back then.

    I guess what I’m saying is that I don’t regard structure and the components of the structure (in the case of social systems, principally agents) to exist at different ontological levels, so much as at different epistemological levels of analysis of the same ontological plane (of which there is only one, really). This is probably a fairly meaningless terminological disagreement w/r/t how best to use the word ‘ontological’. But I think talking as if different levels of analysis are also ontological distinctions, while very useful shorthand, can also easily lead to the reification of the ‘structural’ phenomena, if not used carefully. (And I guess one of the big tasks of good social analysis is to show in some detail how the structures are made by their components.)

    (This latter is the sort of thing I take Marx to be talking about in the famous passage from the Preface to the First Edition of Capital:

    The value-form, whose fully developed shape is the money-form, is very simple and slight in content. Nevertheless, the human mind has sought in vain for more than 2,000 years to get to the bottom of it, while on the other hand there has been at least an approximation to a successful analysis of forms which are much richer in content and more complex. Why? Because the complete body is easier to study than its cells. Moreover, in the analysis of economic forms neither microscopes nor chemical reagents are of assistance. The power of abstraction must replace both. But for bourgeois society, the commodity-form of the product of labour, or the value-form of the commodity, is the economic cell-form. To the superficial observer, the analysis of these forms seems to turn upon minutiae. It does in fact deal with minutiae, but so similarly does microscopic anatomy.

    I take Marx here to be comparing the analysis conducted in Capital with the analysis of a body in terms of the cells that compose it, and suggesting that one of the things that’s novel and important about Capital’s analytic method is that its study of the ‘cell forms’ of social life can give a better analysis of how the ‘body’ as a whole functions, even though it was of course already possible to analyse the body as a whole, without adequately analysing or understanding the ‘microscopic’ components that are its substance. The body analogy can itself be misleading, of course, because it tends to suggest a much stronger organic unity to the capitalist social whole than I think Marx was meaning to imply. But still: when Marx says that the complete body is easier to study than its cells, I read him as saying that it can be easier to study large-scale social structures than to give a detailed account of the many social practices that these structures actually consist in – and that one of the virtues of Capital is that it bridges this analytic ‘gap’ (which is thereby revealed not be be a gap at all.))

    I definitely agree that agency depends on structure, also – I parse this as meaning that the agency of individuals and groups is formed by and out of the social structures of their (our) time, and that these structures can themselves in principle (if we want) be fully analysed in terms of the actions of lots of agents? So for instance class is a structural phenomenon, and so are the institutional, economic and ideological structures that reproduce class; but none of these structures have any existence beyond the actions of lots of people (and those people’s interactions with objects too I guess). Class is a structural phenomenon, that forms and also oppresses particular agents. But in any given person’s life, the reality of class is the reality of the way other people treat them. The structure just is those actions, taken all together – or rather the pattern of aggregate reproduced action those actions make. Which doesn’t mean of course that one can’t talk about classes as a whole, etc.

    This is a much longer comment than I meant it to be, sorry. Obviously I’m thinking about this sort of stuff at the moment – so thank you for giving me the chance to think out loud… On Poulantzas – I’m glad you mentioned – I was actually wondering if I should tag along on your reading project – I’d definitely be up for looking at his work, if I can find the time and headspace (which isn’t guaranteed, but…) Ed Miliband’s election to the leadership of the UK Labour Party made me think (among other things) that I really should get round to reading Ralph Miliband’s stuff. I was googling him the other day and I saw there was a debate between him and Poulantzas. Would that be a good way in to their work, do you think? (I’ve not read any of the Marxist state theorists, I’m really ignorant in this area.)


  3. Mike Beggs Says:

    Hey Duncan – thanks for that, really interesting. I’ll reply to the main part of the comment later on when I have some time… But briefly, on Poulantzas – it would be great if you wanted to join in on the reading. I’m starting with the early essays reprinted in the Verso collection that came out a couple of years ago – let me know if you want e-copies of the essays. It’s very heavy going – one reason why I’ve taken a long time to get started is that I’ve had to do some background reading on legal theory and late Sartre just to understand P’s terminology. (That lead to the great experience of reading Pashukanis, who is really interesting. I’m not planning on tackling the 1500 pages of the Critique of Dialectical Reason, but I read some summaries to get the language.)

    Miliband I highly recommend, and he’s much easier going. It’s common these days to say that the debate between M and P is misleading about both, that they were talking past each other, and that they both have something to offer. I’m definitely planning on dealing with this debate.

    I see it as analogous to EP Thompson’s polemic against Althusser (though Althusser never responded) – I’m sympathetic to good old Anglo common sense and plain language, and in both cases the Anglos are much more fun to read, but I also think there’s something to be said for abstract systematising and the critique of empiricism – because it’s really very hard to think through these social categories.

    As for other Marxian state theory – there really isn’t all that much. Poulantzas works by reviewing ‘the classics’, including Gramsci. Since the 1970s much of the best stuff came out of Poulantzas – Suzanne de Brunhoff is wonderful and I’ve written about her before – and Bob Jessop, whose intellectual biography of Poulantzas I’m using as my guide. Then there are the regulationists who are also interesting and came from the same milieu as Poulantzas. Leo Panitch was I think a student of Miliband and he is one of those urging the building of bridges between him and Poulantzas. Panitch’s very concrete stuff on the 1970s and 1980s in Britain (collected in ‘Working Class Politics in Crisis’) is great.

  4. duncan Says:

    Thanks, that’s a really helpful summary of the state theory space. I’ve heard other people speak highly of Pashukanis but he’d dropped out of mind, I should put him back on the to-read list. Lots to read anyhow. On Poulantzas: I grabbed Political Power & Social Classes from the library today (because it’s what they had – no Reader, oddly). Not sure I’ll be able to get into it for a while, but I’ll definitely try to absorb some when I can…


  5. mikebeggs Says:

    Hey Duncan – also thinking out loud, and ditto that this kind of discussion is useful to clarify what I actually think.

    Re: ‘emergent’. It’s a tricky word and maybe I shouldn’t have used it. I also agree with a ‘weak’, i.e. epistemological conception. I’m less certain about ontological emergence, and it is one of the great philosophical/scientific controversies of the age w/r/t consciousness. In any case, when I wrote about different ‘ontological levels’ I meant it in a practical sense in that different sciences have different ontologies (i.e. systems of what exists ¬_in their domain_ – objects and their relations) which they don’t know how to reduce to ‘more fundamental’ sciences. Even if biological processes, for example, don’t violate the laws of chemistry and physics, chemistry and physics as they are can’t entirely account for the biological level, and probably never will. This is because (1) biology is just too complex in physical terms and (2) a path dependent historical science, such that you can’t explain the physical makeup or activity of a cow, for example, without making reference to uncountable physical conjunctures over vast tracts of time. It’s not that there is a different kind of stuff, but different levels of organisation.

    So this is certainly epistemological, and perhaps it’s confusing to use the word ‘ontology’ in this context, but that’s what I was getting at with ‘ontological levels’. Biological organisms can have physical and chemical effects, but accounting for those effects without reference to biology is a practical impossibility, so that any explanation we give of interaction across domains involves causation across these levels and not reduction to a single physical level.

    So in terms of the social, I was trying to argue that structures are irreducible to individual action, even if they are created and sustained by individual action. But the more I think about it, the more I think it’s wrong to say that structures are ‘emergent’ out of individual action – because it is just as true to say that the individual level depends on social structure. Because while you can conceive of a physics without a biology, you can’t really conceive of individual people without social structures. Whatever human consciousness and agency are, they evolved within groups – social structure was always already there.

    I think rather that both social levels – individual action and psychology and social structure – depend on one another, without either being primary, or explained entirely in terms of the other. So I guess I disagree with you that structures could in principle be explained entirely in terms of individual action, because I don’t think you can give descriptions of actions without referring to structures. And finally, if you think social structures are reducible to actions, why stop at agents? Couldn’t you explain agents in terms of psychology and physiological powers, reduce psychology to physiology, physiology to chemistry and so on down the line?

    Re: Marx, commodities and cells – surely, though, Marx in the end doesn’t find the mystery of capitalist society in its cell-form – rather, he finds crystallised in commodities the effects of social structure – the division of labour etc – just as biological cells can only be understood with respect to their places in the structure of the organism, and the organism understood only within its ecosystem for that matter – a radically different kind of thing from a cell.

  6. duncan Says:

    Thanks Mike – I think we’re basically in agreement, but maybe with some differences of emphasis? I won’t have time for a few days to respond properly, but it’s definitely really useful to talk through this stuff. More soon…

  7. duncan Says:

    Sorry not to reply for so long Mike – very busy. Now that I’ve got time: I think there are two different things here.

    First thing: that we’re dealing with different epistemological frames or levels regarding the same basic stuff. I strongly agree with this. In no particular order:

    – Yes, it’s important that we talk about ‘weak’ rather than ‘strong’ emergence. It’s a while since I read any philosophy of mind, and I was never particularly keen on it, but my sense is that ‘strong’ conceptions of emergence are often used to smuggle tacitly non-naturalistic forms of explanation into discussions of neuroscience, etc. The concept of ‘weak’ emergence is entirely unproblematic, though, and I think very useful.

    – I think care needs to be taken when saying things like: any explanation we give of interaction across domains involves causation across these levels and not reduction to a single physical level. This can be meant and understood unproblematically – if when we talk about ‘causation across levels’ we are just talking about causal interactions different aspects of which we have chosen (often unavoidably) to discuss using different analytic frameworks. So if a scientist extracts DNA from a cell, for example, we have an interaction between, on the one hand ‘a scientist’ and on the other hand ‘a cell’. Clearly it would be ludicrous and impossible to describe and analyse the scientist in terms of cells, here. Our two units of analysis belong to different scales and different analytic idioms. But that doesn’t mean that the scientist isn’t made up of cells. We don’t really have ‘causation across levels’ here in any ontological sense – we just have causation, and the ‘levels’ belong to the different epistemological frames we have available for talking about the different aspects of the causal chain. So – yes – it would be ridiculous and impossible to describe the entire interaction between scientist and cell using only the idiom of cell biology. But the practical impossibility of translating one idiom into the other doesn’t mean that the objects of reference of one idiom are not structures composed of the objects of reference of another.

    – This is how to think about things like your cow example too. Evolutionary biology is a really useful example – on the one hand, we have the cow, which is an individual organism and can be analysed as such. Or we can analyse the species. Or the cell and its DNA. We know how natural selection works – we know that we can explain the cow’s organism in terms of the impact of its genes, sexual reproduction, selection from within species’ variety by environmental factors, etc. In the case of cattle, a major environmental factor is human society, so a socio-historical account of human farming practices would have to be part of the environmental narrative. Again, though, this doesn’t stop the cow being composed of cells, and all of the animals that are part of its genealogy being composed of cells.

    I think we basically agree about all this.

    You then move to a second thing, though, and here I think we disagree. You say, for instance: I think it’s wrong to say that structures are ‘emergent’ out of individual action – because it is just as true to say that the individual level depends on social structure. But I regard this as a false opposition.

    So: I definitely agree that the individual level depends on the social structure – I’m not in any way contesting that. The question, however, is what the social structure is. If you take this formulation (I mean the one I just quoted) to be in tension or disagreement with the claim of the original post, I think you must tacitly be seeing social structures as ontologically distinct from individual action, not just as epistemologically distinct. Which I think is wrong. Let me try to cash that out at greater length.

    There’s a common sleight of hand that’s used in a lot of methodologically individualist econ literature, though the same form of argument appears in other contexts. The argument runs something like:

    1) Large-scale social structures are just composed of lots of agents acting in aggregate.
    2) We can analyse agents individually.
    3) Therefore everything we need to know to understand agents’ behaviour can be understood by analysing agents individually (rather than the structures they form).

    #3 is nonsense, of course. There’s a move from the fact that social structures are composed of nothing but individual agents, to the idea that agents’ behaviour is something like spontaneously self-generating – as if individual behaviour could be fully understood in the complete absence of any environmental analysis. There’s absolutely no sense in which this follows – yet the ‘argument’ is apparently often seen as compelling.

    I think one thing that’s going on when arguments like this are made, is a mix-up over explanatory chains: a confusion of two different senses in which the individual agent can be taken to be explanatorily fundamental.

    Sense 1 is the sense in which individual agents are explanatorily fundamental because there’s nothing but individual agents in the structural phenomena we’re analysing. If structures are emergent effects of multiple agents, we have an epistemological-levels sense in which individual agents can be taken as more explanatorily fundamental than structures – just as cells can be taken as more explanatorily fundamental than cows. Cows are composed of cells, but cells are not composed of cows. This is true.

    Sense 2 is the sense in which individual agents are explanatorily fundamental because there’s (supposedly) no need to go beyond an analysis of an individual to explain the make-up of the individual. This is false. The parallel, here, would be claiming that there’s no need for any evolutionary story to give an account of the specific DNA in a cell (or, indeed, the type of cell one’s dealing with). Self-evidently, structural forms of explanation (for instance, in the biology example, evolutionary or organism-functionalist forms of explanation) are required in order to give an account of the forces that produce the individual unit. This can be true while at the same time there being nothing but combinations of multiple examples of this unit in the composition of the relevant larger structure (for example, a complex organism).

    It’s really important to keep these two senses of ‘explanatorily fundamental’ distinct. If you don’t, you end up with all kinds of bizarre arguments. In the case of social explanation, you end up with the ridiculous idea that individuals are entirely sui generis and environmental factors have no impact on the formation of self, based on the perfectly true and legitimate idea that societies are composed of individuals. It’s sort of weird that this confusion is so common, but it seems to be.

    I of course don’t think you’re advocating this kind of individualism; but I do think there’s a related slippage when you say you disagree with me about emergence, because structures impact on individuals. The point I’m trying to make is that it’s entirely compatible to say both that social structures are nothing but aggregate effects of many different agents and that agents are formed by social structures. This is what I was trying to get at in the original post when I was expressing frustration at the way structure and agency are often taken to be opposed. A parallel in biology would be saying both that organisms are composed of cells, and that the make-up of a cell depends on the kind of organism it belongs to, and the evolutionary history of that organism. These things have to both be true. If we think that emphasising one requires us to weaken our commitment to the other, there’s bad reasoning lurking in the background.

    So, in terms of social analysis, class (to continue with that example) is a pretty simple structural reality – a socially-reproduced pattern in the behaviour of large social groups. This structural reality is nothing but the pattern produced by the behaviour of all the relevant individuals. Nevertheless, class as a structural reality forms individuals. For example – if one is systematically rejected in job interviews because one does not have the right social capital to fit the well-paid roles one is applying for, one is experiencing a pattern in people’s behaviour. It’s a predictable pattern – it can be analysed as a pattern. Nevertheless, it’s nothing but individuals’ behaviour – it has no additional or ‘higher-level’ reality. I, as an individual, am going to be impacted by my failure to get a well-paid job; this is going to impact in turn on my own behaviour. All this can be analysed in detail – but such an analysis, if it’s going to be a good one, needs to be informed by the metatheoretical insight that structure just is agency in aggregate – it is a pattern visible in the behaviour of many agents. (And as you say, that pattern need not be intended by the agents; it very often isn’t.)

    The reason structural analysis still exists, even if structure ‘just is’ the action of the individual agents, is that 1) as you say, it’s often practically unfeasible to analyse the individual cases; but also, 2) in a real sense it’s the pattern, as well as the individual cases, that one’s interacting with: if I apply for 40 nice jobs and get rejected for all of them because I lack the required social capital (for example), it’s the pattern – the consistency – that matters, just as much as it is the specific motives of the specific interview panels (though the former can of course be explained in terms of the latter). The pattern might well still replicate, even if I’d encountered different interview panels – and thus the structural reality can be analysed with some degree of abstraction from individual cases, even though it has no reality beyond a regularity in multiple individual cases. And of course this type of analysis is absolutely imperative when one’s dealing with more ‘downstream’ patterns that don’t involve any specific judgement on the part of individual actors, but are, rather, complexly produced as unintended consequences of multiple divergent and seemingly unrelated forms of social behaviour (as is the case with many structural features of capitalist society). [I’ll add in my customary reference here to N. Pepperell’s influence on my thinking about all this.]

    Now my own sympathies, w/r/t social analysis, lie (as you know) with the world-systems end of the Marxist theoretical space – i.e. I’m interested in massively macro-social structural analysis (and indeed the world-systems folks are often accused of macro-reductionism). But I think it’s important to see that such macro analysis is not opposed to the analysis of agents; it is simply the analysis of agents at a particular scale. When doing world-systems analysis, one is still & already analysing agents.

    So I definitely agree with what you say at the end of your comment about how biological cells can only be understood with respect to their places in the structure of the organism, and the organism understood only within its ecosystem for that matter – a radically different kind of thing from a cell. But I take this to be more or less another way of formulating the point I was aiming at to start with.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: