Editor’s Note: This is a guest post by Stacie E. Goddard. It is the sixth installment in our “End of IR Theory” companion symposium for the special issue of the European Journal of International Relations. SAGE has temporarily ungated all of the articles in that issue. This post responds to Andrew Bennett’s article (PDF). His post appeared earlier today.
Other entries in the symposium–when available–may be reached via the “EJIR Special Issue Symposium” tag.
I am excited to blog about this EJIR special issue on theory and international relations, and am particularly pleased that I’ve been asked to discuss Andrew Bennett’s article, “The Mother of all isms: Causal mechanisms and structured pluralism in International Relations theory.” Bennett and Alexander George’s book was a touchstone for me as I wrote my dissertation and then book on indivisible territory. It was invaluable to have a book that was both pluralistic in scope and rigorous in its approach, and that explained in crystal clear language the benefits of a mechanism-based approach to international relations theory.
Not surprisingly, then, I agree with much of what Bennett has to say about international relations theory. Yet at the same time, Bennett’s forceful argument about mechanisms is disconcerting, in that it comes close to suggesting that the only way to destroy the paradigmatic debates is to build a new paradigm. In particular, Bennett argues that if IR scholars are to do away with the existing paradigms, we must embrace scientific realism: scholars must agree that there is a real (if ultimately unobservable) world out there; accept the fact that our knowledge is socially produced; but at the same time, agree that we can rationally adjudicate among theories. In other words, international relations scholars must embrace very specifically defined—and very much contested–ontological and epistemological foundations (I should note that I’m losing “paradigm,” rather loosely here, as I agree with Bennett, and with Jackson and Nexon, that the language of paradigms doesn’t fit the state of IR theory). Once we’ve accepted these paradigmatic foundations, we can then all proceed with the business of proposing and testing mechanisms.
And I have three substantive concerns about the paradigm Bennett proposes. The first is a discomfort with Bennett’s contention that scientific realism presents a more reliable path forward than other ontological or epistemological commitments. One key claim of scientific realism, for example, is that the entities described by a theory have ontic status, that is to say, they really exist “out there,” independent of the theory itself. So, to take one significant example, if a theory of international relations says that there are discrete agents and structures, then those agents and structures are real things that exist outside of the theory’s architecture.
But this is not the only way to approach theoretical concepts. To use the agent/structure example, Waltz most famously adopted an analytic epistemological position, where agents and structures—in Theory of International Politics, states and anarchy—were analytic constructs: they did not exist in the “real world” (for Waltz’s clearest statement on this, see his reply to Vasquez in the balance of power debate here; for a longer discussion, see my paper with Dan Nexon here). There are ways in which Waltz’s analyticism was frustrating. Yet one could argue that Waltz’s analyticism did a better job avoiding the reification of agents and structures into separate ontological entities than did Alexander Wendt’s work. Whereas Wendt’s scientific realism pushed scholars towards treating agents and structures as two discrete entities and, I would argue, reified dynamic processes of co-constitution as a result, Waltz made it clear that the separation of agents and structures was
A second concern I have has to do with the status of theory itself in a mechanistic approach. Where is the theory in Bennett’s argument? On the one hand, theory can generate lists of testable mechanisms: democracies cause peace is the starting point for identifying a host of mechanisms connecting regime type to peace. Perhaps more importantly, theory tells scholars when mechanisms are likely to operate. As Bennett argues, invariant causal models are problematic: it is not that one independent variable operates through a single specified mechanisms to produce a defined dependent variable. Rather, it is “combinations of mechanisms” that “interact in specified and often recurrent scope conditions or contexts to produce outcomes.” (p 12) In many ways, then, the most important role of theory is to establish scope conditions: we know that if we are in this theoretical world, then X set of mechanisms is likely to be operating, whereas if we are in this other theoretical world, then we should expect Y mechanisms to operate.
This, for me, begs the question of how interpret our evidence: in particular, in the face of an unexpected outcome, does this mean the mechanism is less powerful than we thought, or did the necessary scope conditions simply not apply? I suppose Bennett would argue that we can test scope conditions, and that this should be a simple matter of finding even more fine-grained evidence, and then engaging in Bayesian updating (PDF / PDF) in light of the results. If this is the case, Bennett has far more faith in Bayesian inference than I (and I think more faith than is warranted). Bayesians accept that we all approach the world with different theories but that, ultimately, we exist in the same world, we all have common knowledge about that world, and thus will interpret evidence in the same way and thus update our knowledge and theories rationally.
I’m a little more skeptical. While I agree that IR theories are not entirely incommensurable, there is an extent to which our ontological and epistemological commitments shape the way in which we interpret evidence. Social events rarely have a single interpretation, and thus all of our data—qualitative and quantitative—will be somewhat contested. This means that our adjudication of theories is more of a limited Bayesian inference: part rational updating, yes, but also part the privileging certain mechanisms over others, for no “scientific” reason other than the current status of an approach. We risk thus consistently turn to a handful of recognized mechanisms—commitment problems, signaling, etc.—not because they are objectively “out there” and our evidence tells us so, but because it is the language du jour of discussing social processes.
My third and final issue is with Bennett’s contention that a mechanistic approach is a better way to teach international relation theory to students than arranging our syllabi around the three “paradigms.” I guess my first question is, teaching international relations theory to whom? I may have a very specific perspective on this, because I teach international relations theory at a liberal arts college. Maybe I could see this approach working at the graduate level, where we are attempting to teach students a discipline, but at the undergraduate level Bennett’s argument just isn’t convincing.
It’s not simply that a mechanistic approach is somehow more complicated, or that I aim to teach my students more theory than “empirical” knowledge. There is a way to use the paradigms to teach IR theory that is less about hammering in the minutia of neorealism, neoliberalism, and constructivism, than it is about getting students to think about the big questions, not only of international relations theory, but of social theory in general. What is power? How do we think about power in relation to reason, and can the human gift of reason somehow transcend practices of brute power? What of agency and free will? To what extent can human beings alter and transform the world in which they live? These to me are the lifeblood of the paradigmatic debates. Stultifying as these arguments might have become in the field, they can still electrify a classroom of first and second year students.
And I would argue that asking these questions of our students create better policymakers as well. Bennett argues that moving towards mechanisms is a move towards more policy-relevant science. If we work hard enough at uncovering the mechanisms driving the democratic peace, for example, then we can give policymakers the concrete advice that they seek. Yet teaching students the paradigms is not to avoid these empirical questions. It is rather to argue that the most important skill policymakers can have is to critically question themselves and the state of their world, to realize how much their own theoretical commitments can shape their view of the world around them. It is to show how theoretical assumptions have the decisions of key leaders, how they have both enabled great transformations and blinded leaders to pitfalls obvious in retrospect. Paradigmatic thinking, in essence, can be the foundation of serious critical thought.
It has become quite fashionable to bash the paradigms, and as someone who sees herself as outside of the paradigmatic boundaries it is easy to be sympathetic. At the same time, the paradigms have forced IR scholars to be self-conscious in their theoretical commitments. I worry a world without them would make us more exclusive in our approach to the social world, and less critical in our thinking, than we were in the world of paradigmatic faults.
This reminds me of Kratochwil’s (2006) critique of Wendt and the dangers of trying to establish scientific realism as the basis for a new orthodoxy. You’re framing the problem from the perspective of an analyticist, but I think reflexivists would feel even more slighted. Bennett pays lip service to post-positivist claims (“observation is theory laden, knowledge claims are always part of mechanisms of power, meaning is always social”) only to completely disregard them a second later. He allows for interpretive methods and argues that interpretivists in IR have developed standards to judge the quality of the claims they make and that these standards are consistent with scientific realism. That might be true (for some people), but the reasoning behind these criteria varies from camp to camp. As Bennett notes, for scentific realism “use novelty” is akin to theory testing. For analyticists and post-positivists the heuristic power of a theory says nothing about the correspondence of our ideas and the world “out there”. Similar fundamental disagreements can be found in debates about predictive power and, crucially about what it means to offer a “convincing” account.
That said, I don’t quite agree that Bennett is intentionally or unintentionally proposing a new paradigm in the same sense that we call (however incorrectly) Realism and Liberalism paradigms. Adopting scientific realism as epistemology might condition broadly the ontological statements one can make about agents and structures for example (if we talk about them, it implies they exist), but is in principle silent about their existence (we can choose not to talk about them at all), their specific nature (the essence of things that exist) and how they relate to each other. It is also mostly silent on other ontological and substantive questions of interest to us as social scientists.
Hi Dani. Scientific realism an ‘epistemology’; in what way? In fact one of the most common criticisms of SR is that it pay too much attention to ontology and not enough to epistemology. Also, not sure I’m understanding your last para, but SR silent on the existence of agents and structures (I must have written a different book), or mostly silent on other ontological and substantive question of interest to SS. The must be somebody impersonating a lot of people. Me writing on terrorism, and institutional racism (surely issues of concern to SS), Heikki, on the global economy, democracy, Walt and Mearsheimer on Israel and the rise of China….I could go on. And of course, you can, and I know do, disagree with how the SR people do some ofthis stuff, and that’s fine; but silent?
Colin, I may be off base here, but I think of claims about the validity of knowledge and the kind of things that can be known as epistemological claims. So if scientific realism is, in part, a set of claims about the possibility of accessing “the world out there” and even knowing things that our senses can’t observe through abductive reasoning, then scientific realism is an epistemology.
On the second point, you misunderstand me (and I take full credit for that), and I think we’re more in agreement than you think. When I say scientific realism is silent about these things I only mean that adopting scientific realism doesn’t determine your views on any these things. Being a scientific realist doesn’t force you to adopt a structurationist ontology (right?) and doesn’t determine your views on democracy, terrorism or institutional racism or any of the other topics that you mentioned. It doesn’t carry with it assumptions about actors, their nature and motivations (you know, the things that defined our three classical paradigms). It can delimit the broad contours of the explanations you consider, methods you find valid etc., but even here it’s more permissive than deterministic.
Hi Dani, ok, probably my fault re the last para, so happy with all that. In terms of the former however, epistemology is that part of philosophy that inquires into the nature of (what is knowledge; an ontological question) and scope of what we call knowledge. Typically it raises questions about what knowledge is, how is it different (if at all) from belief, and what is the relationship between truth, knowledge and justification. Scientific realism is a philosophy, or theory of science. It doesn’t answer, and can’t, answer these epistemological questions, although answers to them will be implied (i.e that there is such a thing a truth, even if our access to it is only approximate, etc.). Scientific realism does not produce knowledge, or tell you how to tell whether any claim is at the level of doxa (belief) or episteme (knowledge). Scientific realism, and all such accounts of science, produce claims, but not all those claims are knowledge. The claims have to be held up to epistemological scrutiny; to refer to SR as an epistemology misses out that crucial step.
In answer to a typical epistemological question such as ‘how do you know X?; ‘because I’m a scientific realist’ is no answer; whereas because ‘I’ve observed it’ (empiricism), or ‘I’ve logically deduced it’ (rationalism), or because we ‘all believe it’ (conventionalism), or because ‘it works’ (pragmatisms), or because ‘my boss told me’ (epistemological authoritarianism) are. Confusing SR with an epistemology comes about (mainly) because of a residual unchallenged effect of positivism, which basically said that the only valid knowledge is knowledge produced according to the principles of positivism; everything else was meaningless. Hence positivism becomes synonymous with knowledge. But of course, not everything uttered by a positivist, or produces according to positivist principles (whatever they are) is knowledge. But insofar as we keep referring to theories of science as epistemologies then we are reproducing an aspect of positivism. You won’t find references to SR in books on epistemology, but you will see it in the Philosophy of Science.
Never thanked you for the terrorism exchange btw, so thanks, I’ve changed my view ‘ever so slightly, but in an important way’ as a result of that.
Colin, I take your point and am happy to call SR a philosophy of science, but I’d insist that it carries with it strong and specific epistemological claims (as well as scientific ontological claims, but those are essentially tied together it seems) that aren’t reducible to or conform exactly to this neat typology of epistemologies you present. In my (admittedly non-expert reading) SR makes different claims about logical reasoning than does logical positivism, different claims about observation than radical empiricism, different claims about heuristic validation than analyticism (hence Stacie’s criticism) and different claims about conventional knowledge than radical post-positivism.
I thank you as well, for taking the time to engage with my critique of your article on terrorism. Your response forced me to clarify my thinking on the topic and I’ve expanded on it significantly since then. I’m really looking forward to reading the book!
Hi Dani, but Logical positivism isn’t an epistemology (it’s another theory of science), and SR’s critique of observation is an ontological claim (there’s more to the world than what can be observed), heuristics is a methodological/axiological point (it’s not the end point of science), and that knowledge IS (ontological) not only the only is important) conventional (it’s also worth pointing out that SR is probably the dominant post-positivist theory of science). In all of these claims the epistemological issues are unresolved. So you can always ask me how I know LP is not an epistemology, how I know there’s more to the world than what we observe, how I know heuristics/analyticalism is not the end point of science, and how I know that knowledge claims are not all conventional. After making these SR claims, I still need epistemological support not provided by SR. SR doesn’t help me answer these questions, and because of its commitment to ontology then the relevant answer will differ with context. In this sense SR is epistemologically promiscuous; although given recent debates on this site I probably shouldn’t use that word. Put simply SR will not help you distinguish belief from knowledge, or give you any epistemological support any particular claim. Tbh, it’s not your fault it’s just the weird way IR talks about epistemology. And in fact, even the dominant view of positivism as committed to empiricism is incorrect; as they say, the clue is in the name ‘Logical Positivism.
Hi Dani, but Logical positivism isn’t an epistemology (it’s another theory of science), and SR’s critique of observation is an ontological claim (there’s more to the world than what can be observed), heuristics is a methodological/axiological point (it’s not the end point of science), and that knowledge IS (ontological) not only the only is important) conventional (it’s also worth pointing out that SR is probably the dominant post-positivist theory of science). In all of these claims the epistemological issues are unresolved. So you can always ask me how I know LP is not an epistemology, how I know there’s more to the world than what we observe, how I know heuristics/analyticalism is not the end point of science, and how I know that knowledge claims are not all conventional. After making these SR claims, I still need epistemological support not provided by SR. SR doesn’t help me answer these questions, and because of its commitment to ontology then the relevant answer will differ with context. In this sense SR is epistemologically promiscuous; although given recent debates on this site I probably shouldn’t use that word. Put simply SR will not help you distinguish belief from knowledge, or give you any epistemological support any particular claim. Tbh, it’s not your fault it’s just the weird way IR talks about epistemology. And in fact, even the dominant view of positivism as committed to empiricism is incorrect; as they say, the clue is in the name ‘Logical Positivism.
I admit to some confusion here, but I think you might also be defining epistemology too narrowly, in a sense… I can’t see how some of these claims you’re characterizing as ontological aren’t also epistemological (or at the very least necessitate positions on classic espitemological issues like internalism/externalism to be meaningful).
I didn’t mean to say that LP is an epistemology or that it’s committed to empiricism (in fact I think I wrote quite the opposite above). The point I was trying to make was that SR may draw on different epistemological propositions, but it doesn’t do so freely or indiscriminately and it seems to have different views on some of these than other philosophies of science that make similar claims such as LP or radical empiricism.
Hi Dani, I am defining narrowly, which in my opinion is far, far better than the scattergun approach used in most social science, that confuses it with ontological issues and most damagingly methodology . More to the point, I’m using it in a way that respects the history of the term, and how it’s used in philosophy. For sure IR is entitled to use any term any way it likes, but since it does use it to refer to issues surrounding ‘knowledge’ then it can’t just ignore that tradition that’s been going over that for centuries. We don’t need to become epistemologists to do SS. We use epistemologies and we should be more relaxed about them.
Anyway, give me some examples of where the positions I’m articulating do what you suggest and I’ll try and address them. But more to the point tell me how SR tells us what ‘knowledge’ is, tells us how to distinguish doxa, from belief, and so on. Internalism/externalism isn’t only an epistemological issue, or even primarily so. I believe there is an external world, and I think there are good arguments to support that belief (others think otherwise), but if you ask me why I’m happy to say that, with all due caveats, I think it’s ok to say that ‘we know’ there is an external world (rather than just believe there is one), then SR isn’t an answer, and neither is I’m a positivist. One thing I do with my students to get this point across is to get them to read an article and then ask themselves of every claim made by the author: ‘how does the author “know” this’, why should ‘I’ believe this ‘knowledge claim? Invariably you’ll find some claims grounded in empirical data, some in logic, some in ‘because as X has shown claims’, and some in ‘because it works’ claims. Because they are a SR, or a positivist, isn’t an answer; it’s a bit like saying because I’m English.
Likewise, can you explain why you won’t find SR or positivism discussed in philosophy books on epistemology? I didn’t mean to imply you said LP was empiricist, I said the dominant view. Finally, of course, SR has different views on LP they are competing views on science. Radical empiricism is a term sometimes applied to the Philosophy of science and sometimes just to a radical epistemological position. In the philosophy of science it describes the practice of science but then adds that ONLY empirical knowledge can count as scientific evidence. Nobody really takes it seriously today as it’s accepted that even the most stark facts need to be placed in framework of rational (logical) understanding and interpretation. In epistemology, however, it’s still a position some pursue, arguing that the the only thing that counts as knowledge is that derived from experience. You can be a radical empiricist in terms of epistemology, but not touch on the Philosophy of science at all. But in order to be a radical empiricist in the philosophy of science the ‘radical empiricism bit’ will only be a commitment to that epistemology, but that epistemology doesn’t exhaust the content of what ‘Radical Empiricism’ as a philosophy of science is; it’s part of it but not exhaustive of it. SR just says that, like Feyerabend, and Einstein, that the correct epistemological stance for a scientist is to be ‘opportunistic’. So SR will use the appropriate epistemology for the task at hand, but it’s not in itself an epistemology.
Don’t worry about sounding grumpy! The internet is a weird medium that always makes things sound way worse than intended (probably because we’ve grown accustomed to trolls), so I compensate and always try to read things in the best light possible.
I can see how it can be better to define it narrowly than too broadly, but I’m not sure it’s entirely consistent with how all epistemologists define it or at least treat it in practice when they write. Again, I say this as a non-epistemologist who reads this stuff not systematically, but I can think back to reading work “on epistemology” (memory is a valid source of knowledge right?) that explicitly discusses things you’re classifying exclusively as ontological (internalism/externalism, realism/anti-realism) or methodological (much discussion of appropriate methods of logical reasoning and inference, falsification/verification, etc.) and some of it discusses philosophies of science (SR much less than LP in my limited experience, but then again some of these you would probably and rightly say are philosophers of science writing about epistemology, like Philip Kitcher).
To give a very specific example of what I was trying to argue before, would you say that logical positivists and scientific realists have the same view on what constitutes valid observational evidence?
Hi Dani, memory can be a source of knowledge, but it’d fall under empiricism if the memory is of an actual experience, but if it’s a ‘false’ memory it’d be different (i.e. seems to have no valid grounding). Saying ‘I remember this’ doesn’t tell you whether that is a belief (doxa) or is known (episteme). Memory is a valid source of producing claims. But we are still left with the question of whether we can say of that claim that we know it, or just believe it. Just saying, ‘I remember this’ doesn’t make it knowledge; you still have work to do to epistemologically to justify that claim. We do this outside of science in relation to memory, but in science of course, we must do it, and claims can’t just be taken as knowledge on the fact of their assertion. Unless you are arguing that every claim made by anyone is knowledge, in which case we don’t need epistemology. Of course, the realism and the other issues come up in the discussion (so do ducks, optical illusions, bias etc.), but that’s because some epistemological positions imply realism, some don’t. And yes, there is a method of discussing epistemology and justifying epistemological positions, but that method it’s isn’t the epistemology. Re your point about LP and SR on observational evidence; well the epistemological point is because they both accept empirical evidence as a source of knowledge, so it’s a commitment to an empiricist epistemology at some level, but neither LP or SR are themselves the ‘epistemology’ that justifies that belief in observational evidence. But, it’s harder than that, because, there are all sorts of problems that have been well discussed in epistemology about the problems of just relying on the evidence of the senses, and how the senses deceive. Which is why SRer’s don’t just use empiricism. And of course, a belief in non-observables will require more than simple empiricism. And yes, you’d expect philosophers of science to discuss epistemology, and you can get epistemologists discussing the Phil of Science (after all for many it’s one of the best ways of producing knowledge we have). However, philosophies of science imply the ‘use’ of epistemologies hence can’t themselves ‘be’ an epistemology. Again, I think it’s quite straightforward to see this. Take any piece of research output produced by any tradition and ask your self of any claim in that research; how do they know this? Again, because there are a positivist isn’t an answer. but as long as we keep talking of these positions as epistemologies we aren’t going to bother doing the hard work that’s needed to really look into the knowledge claims. Research following SR produces claims, but it doesn’t tell you how to distinguish which of those claims deserve the status of ‘knowledge’ and which should remain at the level of ‘belief’. To be honest, I’ve pretty much given up on thinking that the social sciences really want to take this extra step and of course we don’t have to as long as we think of the likes of positivism as an epistemology.
It’s nice to wake up to such an incredible discussion. I’ll try do it justice. I’m going to reply to Dani’s points here, and then to everyone’s discussion on SR below
Dani–I’m embarrassed to say I have not read Kratochwil’s 2006 critique (but I will by the end of the week!). Much of my thinking in that first point about reification was drawn from my work with Dan on Theory of International Politics, including several background discussions that never made it into the piece. That’s one of the reasons I sound like such an analyticist here; I think you’re right that reflexivists will feel more slighted.
I think you are absolutely right that Bennett is not proposing a new paradigm. And by the way, that introductory paragraph is not simply flattery on my part. I’m a big fan of Bennett’s work and, having had the pleasure of talking to him in person I can say that he is incredibly thoughtful, careful, and generous as a scholar.
But I do think the article has a bit of a message of “if we just accept these very reasonable assumptions, we can go about our business.” And these assumptions contain some huge areas of contestation, not just in term of ontology (which seems to drive the discussion below), but also about the viability of Bayesian approaches. And if this is the case, then we are talking about paradigmatic thinking here (indeed, I would say it’s even more paradigmatic than realism or liberalism because of these assumptions about ontology and epistemology).
Stacie, if you’re right that that’s the message I don’t think that’s such a big problem, because that’s already the standard in American political science, with the difference that the “very reasonable assumptions” we’re expected to accept – KKV-style positivism – are even narrower and thus more exclusivist than the ones Bennett proposes. It’s not perfect, but it’s progress.
I just had to comment on this not as editor of the Special Issue, but as someone who knows (allegedly) something about scientific realism:
‘One key claim of scientific realism, for example, is that the entities described by a theory have ontic status, that is to say, they really exist “out there,” independent of the theory itself. So, to take one significant example, if a theory of international relations says that there are discrete agents and structures, then those agents and structures are real things that exist outside of the theory’s architecture.’
This is actually the polar opposite of SR; although it’s a common misperception (I blame PTJ and Fred Chernoff myself…:))
Whether the things posited in a theory exist or not, is what’s up for question. The theory suggests these things, and then these claims have to be tested. However, if the claim isn’t interpreted realistically, then testing makes no sense. I mean why would you try splitting an atom if you don’t ‘think’ its real? But in the testing we might find out the claims were wrong and we’d then be forced to drop or amend the theory (unfortunately the one about the atom seems to have been right). On your reading Stacie (Hi, I can’t remember if we’ve met or not; got a bad memory!) the articulation of the claims in the theory makes the things real, which is exactly the opposite of SR; if it were the case that all the existence of X depended on was the articulation of X in a theory, then SR’ers would idealists. In fact, the position you describe is pretty much instrumentalism, just they substitute the word ‘real’ with useful. But they carry on acting ‘as if’ the things were real if they are useful.
I think that you’ve interpolated into Stacie’s example the idea that any and all theoretical terms refer to real entities.
I thought that’s what she stated Dan, again, here:
‘So, to take one significant example, if a theory of international relations says that there are discrete agents and structures, then those agents and structures are real things that exist outside of the theory’s architecture.’
That’s simply not correct in terms of SR.
Perhaps she was trying to say that the theoretical entities posited by SR are possibly real, and their reality is what is to be established or disproved through testing? This is in contradistinction to the ‘analyticist’ view that theoretical entities are non-real, non-referring, or non-natural kinds – however you want to phrase it. If so, she might just have worded it poorly. Given the sophistication of her discussion, that not only seems like the more charitable interpretation but the most likely.
Yes it is the most charitable reading, but in a context where most people don’t know the nuances, it’s important to make the distinction that SR doesn’t move straight from positing a claim about entity X to entity X being real. Besides, the critique of Bennett on the issue of SR rests on just this claim, and if Stacie wants to be interpreted in the way you do, then that’s not something Andy would object to (I think). More to the point, once accepted then the idea that SR reifies the posits is also incorrect; or at least it is on any understanding of reification that I’m aware of. Don’t get me wrong, there are potentially all sorts of problems with SR that people can point to, but this isn’t one of them, because it’s not an SR position.
Actually, slight correction there; I’m allowing my marxist leanings to infect (in a bad way) my understanding of reification. On certain readings of reification it might be right, as in, interpret something abstract as a concrete thing, but that would depend on using ‘real’ and ‘concrete’ as synonyms; not something I’d do. But equally, to use ‘reification’ as a critique in that way, already implies that it is known that the abstract concept is not real (in which case it would be an error to treat it as real/concrete), but since the reality of the theoretical posit is exactly what’s at stake the claim of reification (abstract to concrete) is question begging. Abstraction, btw, is integral to SR and from what I know they’ve written more about it in social science than most other approaches. I also draw a distinction between ‘abstractive’ assumptions (which will be unrealistic) and ‘entity’ assumptions. So there is more than one way to assume something.
So the objection Stacie raises here is that SR ‘reifies’ by taking conceptual schemes and treating them as reflective of concrete reality, rather than convenient ways of ordering reality. If I understand correctly.
As far as I understand ‘reify’ – in broader-than-Marxian sense of the term – this is the right way to use it. The problem here seems more to be one of question-begging. A scientific realist would say that reality consists of many different structures, and that we can approximately describe those structures through our theories. So it is therefore only reification if someone goofs it up and treats a conceptual category as though it were referring when in fact it is non-referring.
Her critique of SR seems to go beyond this discussion of reification though. She mentions a tendency to fall back upon a delimited set of mechanisms that is defined by the familiarity of its members rather than by their superiority in describing the operations of Really Real Reality.
On an unrelated note, what do you see the biggest potential problems with SR as being? I am not an SR, but I find it to be one of the most sophisticated and powerful philosophies of social science out there. I have fewer problems with Elder-Vass compared to Bhaskar – E-V drops ‘strong’ emergence, for example – but I still have trouble seeing how agents can differ from structures and how transformation is possible outside of something like exogenous shock or chaos/complexity, neither of which to my knowledge have received much attention from SR. But since I’m a neophyte, I’d be keen on your own thoughts.
Hi Simon, sorry I didn’t see that point about a tendency to fall back on a limited set of mechanisms as a critique of SR as such, but a critique of any mechanisms approach. But anyway, I’d simply respond to that by saying that there’s no such tendency ‘necessary’ in SR, even if sometimes some people advocating SR might do so. SR, as Dani has pointed out in correcting me, has no pre-given account of what mechanisms there are, let alone what are the most important ones. In fact, whenever I get together with SR people we disagree about the mechs more than anything else.
And that kind of relates to your last question, which is far too big for me to get into here. But I’d say the most satisfying aspect (and something I struggle with) with SR in terms of SS is that it’s really arguing that in open systems with the possibility that anything and everything can be real and a causal mech (ideas, material things, intersubjective beliefs, subjectivity, processes, relations, events, and so on) then it makes life incredibly difficult. It’d be nice to be a constructivist and think it all comes down to ideas, or a classical realist, and think it’s all down to human nature, or some versions of poststructuralism and deal only with discourse, or language, or identity, or even claims that it’s all about relations, and so on. I’m sure life is much easier once you’ve got that kind of certainty about what’s important. I’m in a state of permanent confusion. :) In addition, there are still problems to be worked out about the use of SR in the social sciences, and it’s clear that the nature of intransitivity is very different (but there are ways to get around that). Then there’s the problem of finding a decent analogue in the SS for experiments, or the ability to isolate a particular mechanism.
As for the agent/structure, well it all depends what you mean ‘differ’. I think it’s an easy argument to see just by thinking of how agents move in/between/from structures to other structures. Mind you, I attend an annual interdisciplinary social ontology workshop run by Margaret Archer in Geneva, and she took me to task one year for suggesting that agents always take some of their structures with them. But if you don’t distinguish between them, then you’ll end with either a structural determinism or a methodological individualism. Anyway, I deal with all this in the book, so assuming you’ve read it, if I’ve not convinced you there nothing I say here is likely to persuade you otherwise. Re strong emergence, complexity needs it surely. Endogenous change is surely possible once you allow emergence within systems. But I know Dave’s work quite well so are you sure he drops strong emergence? If so, how do you make sense of his deployment of the causal power of structures?
You’re right, Elder-Vass doesn’t do away with strong emergence. I’ve got him mixed up with Daniel Little, who has a kind of epistemic view of emergence, because I just read a paper by the latter. I should re-read E-V’s book, then, because I remember really liking it for more than just being very clear, but also feeling like explanation of emergence was doing something sneaky. I also agree entirely with what you’ve said about ‘mechanism delimitation’ not being a good criticism of SR; I just wanted to point out that it is another part of Stacie’s argument.
I share your appreciation of SR’s capacity to model structures of anything as having causal powers, and the ultimate collapse of Cartesian divides implied by its physicalism. Particularly given physicalist theories of mind, which tell us that mental states are exhaustively caused by physical states, even if we’re not sure how and may never know. But wouldn’t you say that an essential aspect of SR is the possibility of (approximate) reference? That understanding of language seems a lot more contentious than to suggest experiences are caused by the interactions of Really Real objects, and from what I understand, the post-structuralist critique revolves around reference. Also, wouldn’t you say that ‘constructivism’ actually has a dominant SR streak in it, albeit one based on a rather naive reading of Searle with some old Wendt added in? Or are you speaking about idealism as an ontology? You know, I’ve never met an idealist. Maybe Rorty’s riff on how relativists only exist in our strawmen also applies?
I think your discussion of agency in your book is one of the best I’ve encountered, and yet it still feels to me as though you’ve simply ascribed a quasi-mystical capacity to structures of a certain type (agentic), and I still think that SR collapses into structural determinism. That makes it better philosophy of science than of social science, since my sense of the world is that at the macrophysical level it is indeed basically deterministic. But as I am quick to disclaim, I am new to the literature and I also don’t really have any tutors in it (few in my department care), so I’m probably missing something.
Hi Colin, that’s Bhaskar’s version, right? Someone like (the early, Realist) Putnam would say that “atoms” or “electrons” do exist (or rather, that we have very good reasons to believe that they do exist) even if human beings (and our attendant theories about the world) were to go extinct tomorrow.
The reason being: theories of electrons are very successful (under some definition of success) thereby giving us good reasons to believe that their terms do refer.
If so, then SR (or rather, one variant of Scientific Realism) really does “believe in the ontic status of theoretical entities (but only for successful theories)”
The slight parenthetical modification might then see such a position be un-applicable to the social scientific theories given how our theories have been very un-successful. But not too far from the actual position held by a (a significant-enough portion of) scientific realist philosophers of science, many of whom are concerned more with physics than social science.
Hi Colin! If we haven’t met, we certainly know each other by association right now. Either way, thank you for your comments.
I’m tempted to simply say “what Simon said,” because that is indeed what I meant to say: concepts are possibly real, and this is what we establish through testing. Certainly I don’t mean to imply that SR thinks concepts are magic, and simply articulating them instantiates reality!
I think there are two reasons my statements might be unclear. The first is that I am discussing Bennett from the position of an analyticist in this paragraph, someone who would suggest that concepts are simply a means for ordering reality (note: I’m playing the part of an analyticist for the purpose of debate. For the record, I don’t know what I am. I hope I figure that out by the end of my career. That’s why I like participating in exchanges like this).
Second, part of the conclusion might come from the fact that it’s a short post. In particular, I did not explicitly discuss Bennett’s chart of mechanisms at the end of his article. Basically, he suggests (I read as fairly exhaustive) categories of mechanisms, which include agent mechanisms, structural mechanisms, etc. It is this combination of SR with an exhaustive set of mechanisms that leads to the reification critique.
And finally, Simon, I cannot thank you enough for saying this so clearly (and I’ll quote it here): “She mentions a tendency to fall back upon a delimited set of mechanisms that is defined by the familiarity of its members rather than by their superiority in describing the operations of Really Real Reality.” Exactly. I worry that if we end up backgrounding our ontological and epistemological commitments, we’ll end up treating these incredibly important debates as settled, and privilege certain mechanisms over others for no apparent reason except hierarchies of knowledge within the field. Put another way, I have no problem with the commitment problem, but I fear that everything could become a commitment problem if we go down Bennett’s path.
Hi Stacie,
I’m glad I had the right impression of what you were trying to say. I am not sure I agree with you here, though. That is, I do not see Bennett’s ‘taxonomy’ of mechanisms, which seems to me to be deductively arrived at by considering the full realm of what is possible, as necessarily restricting the set of mechanisms SRists could posit as Really Real. I mean, maybe it does a little, in that it…reifies…existing disciplinary divides. But the bit about ‘agent>structure, structure>agent, etc’ seems pretty metaphysical. In fact, the very possibility of structure>structure is contentious, as some SRs associated with ‘analytical sociology’ argue that it is ontologically impossible because microfoundations. Daniel Little recently uploaded a paper on meso>meso causality, if you wanted to swing by his blog and read it. I think that well illustrates that Bennett’s typology seems more to be about organising theories than defining set-membership.
As Colin pointed out, there is nothing about SR, or any other mechanistic approach to causality (eg McAdam, Tarrow, Tilly 2001) that necessitates this delimitation of posited mechanisms. I’d agree with you that it seems to be a trend, but I think it’s a problem of intellectual laziness and social pressures related to peer review, etc, and not something intrinsic to the SR philosophy of science slash research tradition.
My view is that by establishing SR as THE ontology, with its implied epistemology (ie transcendental deduction), the cost is meta. That is, I think that competing philosophies such as those associated with pragmatist – what we’re calling ‘analyticist’ here – or interpretive social theories can provide us with critical answers to some of the questions that we’re asking. Of course, I am begging the question by framing my view like this, since I’m arguing from just such a pragmatist attitude. Another argument is simply that the ‘wagers’ of SR are still contentious amongst philosophers of science, and that there are noble traditions which contain some powerful arguments in favour of other wagers.
Simon, I’m confused as to what you think THE ontology is that SR is proposing? Also I don’t know that it’s got an implied epistemology; as I’ve said elsewhere SR basically advocates a very pragmatic (opportunistic) approach to epistemology. Also, SR doesn’t reject the insights of the interpretive tradition, but shows how they are compatible (with some modifications) with science. And there are certainly other accounts of science that are deserving of attention. I’d even say positivism gets some things right.
Right! I’ve read these marvelous attempts to subsume interpretive approaches within a SR framework. If I recall, you’ve got an old article on this, and Dave Elder-Vass also has one. I’m not sure if this manoeuvre works completely, but it does make SR more ‘broad church’ than neopositivism, from what I can tell.
I think THE ontology that SR is proposing is one in which structured arrangements of physical things acquire the emergent disposition to influence other physical things, in a layer-cake of structures of structures down to the quantum level. If I understand SR literature correctly, we’re advised to figure out what those structures are by examining phenomena and employing abductive reasoning skills to fallibly arrive at their causal conditions of possibility. I’m sure this is a gross simplification.
Not it’s not wrong per se, but I think it’s more a methodological point. i think I’d say that the ontology is incredibly broad; processes, things, states of affairs, relations, ideas, emergence, structure, chaos, complexity, change, continuity, agency, mechanisms, open systems, closed systems and so on. Tbh, I can’t see that a consistent realism would rule (the possibility of) anything out; how could it? If the world is external to thought what can be imagined can’t be allowed to limit what exists. Even God. I’m not a believer, but given my realism I have to accept I could be wrong about that. different object domains will prioritise certain things over others, but that’s a consequence of disciplinary boundaries not realism per se. Which is why I react in that way to the claim about THE ontology; it all depends which realist is articulating it.
Hi Colin, Simon, and Stacie, Thanks a lot for this discussion (& the post that prompted it, Stacie), it’s been illuminating. A quick point about SR’s ontology: although I’m nowhere near as well-versed as the rest of you, I haven’t seen such a broad ontological buffet advertised by SR as Colin offers. I thought SR ontological entities were at least unified by their intransitivity, durability and stability, and–as Simon put it–their multi-tiered arrangement. This is still quite broad, but less hospitable to chaos, fleeting changes, and some processes and emergences. I thought one of the payoffs of SR was to explain why things seemed chaotic and complex (e.g. because multiple causal mechanisms or entities collided or fired all at once), but an actually chaotic reality would seem to be off the menu.
Hi Stacie, yeah well you and me both in terms of trying to figure this stuff out. It’s a black hole, but one that’s a bit like a rollercoaster; terrifying but something you do for the terror. I absolutely agree that backgrounding the ontological and epistemological (although I have many issues with the way the discipline deals with this issue, but I’m writing on that) assumptions is problematic, which, in part, is why we did the special issue. As I’ve indicated above in response to Simon, I’m not committed to privileging any particular mechanisms, so that’s not an inevitable consequence of scientific realism. In fact, at a very basic level SR simply says, put your mechs on the table and let’s test them, but not if when the evidence seems to suggest they don’t exist you (not you literally) turn round and say; ‘ah that test doesn’t count because I’m not treating the mechs as real, only useful’. Of course, as a marxist leaning SR person, I have my own view on what mechs are more important, but that comes from my marxism, not SR. Look forward to meeting you non-virtually at some point.