Causal Inquiry in Worldwide Relations
By Adam R. C. Humphreys and Hidemi Suganami
Oxford College Press, 2025
In Causal Inquiry in Worldwide Relations (2024), Adam Humphreys and Hidemi Suganami have made an impressive contribution to ongoing debates about causality in our subject and within the social sciences on the whole.[1] It is a guide that may change the best way a reader approaches the crucial process of asking and answering causal questions, which they outline as questions on whether or not and the way an occasion “contributed to the manufacturing, or bringing about, of” one other occasion (p. 14). Their central argument is that causal inquiry “hinges on the connection between concrete causal statements which do consult with particular occasions and summary statements which don’t” (p. 128), the place the latter statements contain propensities for one kind of occasion to supply one other beneath the precise circumstances. Alongside the best way, they critique main positions within the philosophy of causation and the apply of causal inference, together with Bhaskarian crucial realism and what they name “the tradition of generalization”: “the tendency to prioritise the manufacturing of ‘basic’ data over data of ‘specific’…details and occasions” (p. 4).
Humphreys and Suganami start their evaluation with an in depth studying of David Hume’s writings on causality. Hume, in lots of discussions of causality, is usually introduced as a defender of the “regularity view” of causation, in line with which causality is nothing however an thought fashioned within the thoughts after we see a daily conjunction of occasions — which collapses the excellence between causation and empirical regularity, and makes the remark of empirical regularity on this planet the required mark of a causal connection. The authors persuasively present that Hume doesn’t truly advance this view, and that although they themselves aren’t Humeans, the model of “Humean causation” that’s defended by neopositivists and criticised by scientific realists is a skinny caricature (p. 41). This, in flip, units up the authors’ dialogue of Roy Bhaskar’s argument that causality have to be one thing actual on this planet, and never only a projection of the thoughts onto observable empirical regularities. The authors draw on thinker Bas Van Fraassen’s place of “causal agnosticism” — “acknowledging that causal language is intimately sure up with how we, as human beings, relate to and make sense of the world and our place in it, however denying that it’s potential to know whether or not it latches on to part of the world present independently of human thought” (p. 104) — to recommend that it’s higher to depart metaphysics apart[2] and concentrate on what we truly do after we conduct causal inquiry.
If causality is neither a gloss on empirical regularity nor the manifestation of an unobservable-but-real dispositional property of occasions, what’s it? Humphreys and Suganami argue that we should always start by noticing that there are “two sorts of causal assertion: statements which relate to particular occasions, situated in area and time, and statements which summary from particular occasions and therefore possess a broader applicability” (p. 113). The place a concrete causal assertion, which they additional refine as an explanatory assertion — “paradigmatically ‘a prompted b’” (p. 148) — tells us that some particular occasion a caused some particular occasion b, an summary assertion as a substitute posits a causal relationship between occasions of sorts A and B “such that, if this relationship unfolds with out interference, an A-type occasion will produce a B-type occasion” (p. 126). That is the distinction between “my canine Zorri didn’t go on a stroll this morning, and this made her sad in the present day” and “not happening morning walks makes canines sad.”[3] Distinguishing between the 2 sorts of assertion is necessary as a result of it’s unattainable to offer direct empirical proof for or towards the notion that an summary causal assertion expresses; “not happening morning walks makes canines sad” isn’t a declare about observable frequency, however about propensity, and as such says not that most canines who don’t get morning walks are sad, however that canines that don’t get morning walks are sad except one thing else intervenes to forestall that from taking place. A propensity assertion can thus be true even when the proof is inconclusive, as a result of proof pertains to concrete explanatory statements, and Zorri not being sad regardless of not being walked on a given morning doesn’t imply that the summary assertion is fake.
How, then, will we arrive at summary propensity statements which can be in any sense legitimate? Humphreys and Suganami argue that the important thing right here is the realisation that explanatory statements suggest “corresponding summary statements.” If we all know that my canine Zorri’s unhappiness is attributable to her not happening a stroll this morning, that data will “not directly…assist a corresponding causal concept,” on this occasion, the causal concept that not happening morning walks brings about unhappiness in canines (p. 139). They additional argue that proof that persuasively helps an explanatory assertion should rule out “competing explanatory statements regarding particular occasions which have occurred, and that are situated in time and area” (p. 161). It’s not sufficient to merely rule out different explanatory statements, as these could be complementary to the explanatory assertion in query; as a substitute, proof have to be sought that distinguishes between rival explanatory statements that would not each be true on the identical time. On this occasion, we must search for proof that dominated out causes of Zorri’s unhappiness that have been incompatible with the shortage of a morning stroll being the trigger, e.g., {that a} change in her eating regimen produced her unhappiness — which might be dominated out by proof that her eating regimen didn’t change this morning, however that she didn’t get a stroll and was sad. The secret’s to search out proof per an explanatory assertion that invalidates rival explanatory statements.
The mechanics of discovering such proof may be tough, and the authors’ gold commonplace for doing so is a randomized managed trial (or RCT), though they’ve a particular understanding of such experimental analysis designs as ruling out competing explanatory statements by controlling for his or her posited causal components, as a substitute of the frequent understanding of RCTs as testing hypotheses derived from theories (pp. 162-163).[4] However growing legitimate summary causal statements doesn’t require the usage of RCTs; the hot button is growing the “confidence that the proof helps a selected explanatory assertion,” which in flip is determined by “our confidence that every one believable competing explanatory statements may be rejected” (p. 225), regardless of which strategies or strategies are employed in to doing. Discussions within the penultimate chapter of the guide interact each experimental and quasi-experimental designs together with “course of tracing” and historic reasoning based mostly in archives, inspecting how all of them depend on the sort of abductive inference that seeks to offer good causes “to simply accept a [concrete] proposition based mostly on how properly it explains sure noticed details” (p. 219).
So as a substitute of taking a look at many canines who did or didn’t did or didn’t get morning walks and ascertaining whether or not or not they have been glad, Humphreys and Suganami would have us take a look at concrete canines to see if their concrete unhappiness may be defined by the shortage of a morning stroll — a course of which includes eliminating rival explanations of their unhappiness in every case — and solely then advancing an summary causal assertion in regards to the relationship between the shortage of morning walks and unhappiness. This, in flip, means paying heightened consideration to context, and detailing the situations beneath which the shortage of a morning stroll brings about unhappiness.[5] The authors due to this fact don’t advocate eliminating or controlling for context in order to generate probably the most sturdy and basic estimates of common causal results, differing right here with nearly all the neopositivist and among the “qualitative strategies” methods for causal inference. As an alternative, Humphreys and Suganami insist on the gradual and painstaking technique of shifting from concrete explanations of causes and results in context to nuanced claims of any broader applicability, with out with the ability to take the shortcut supposedly enabled by empirical generalisation and its related, primarily quantitative, and formal strategies. In opposition to the “tradition of generalization” that advocates such ways, they advocate a extra grounded immersion in case-specific particulars as the muse of stable causal inquiry.
That is the place I diverge considerably from the authors’ evaluation of causality and causal rationalization. To my mind-set — elaborated most lately in Details and Explanations in Worldwide Research…and Past(Jackson 2025)—it’s removed from clear what it might even imply to contemplate an explanatory assertion to be purely “concrete” within the first place. The authors don’t argue {that a} causal relationship is instantly observable — what we will observe are inputs and outputs, and even the vaunted RCT strategy solely exhibits us that the proof is per some a bringing about b and inconsistent with components aside from a doing so, relatively than giving us causation “instantly” — and so they, like mainly everybody else engaged on causation who isn’t a neopositivist, distinguish sharply between noticed covariations on this planet and causation per se. However which means that within the very coronary heart of each “concrete” explanatory assertion, there’s something “summary”: the very notion that a prompted b, relatively than the notion that a solely preceded b or b solely succeeded a. With out this “summary” notion, all of the element on this planet about Zorri’s lack of a morning stroll can’t presumably be proof that this “prompted” something.
The issue appears to be that Humphreys and Suganami begin their evaluation from a hypothetical scenario of an investigator working with no causal theories by any means, and definitely with no causal theories that the investigator already takes to be legitimate. Whereas they do recommend that it’s potential to generate a causal rationalization “by making use of a well-supported causal concept,” this pathway takes a secondary place of their account, behind the strategy of “figuring out proof which instantly helps an explanatory assertion that’s to be provided as an evidence” (p. 168). However it is a extremely tendentious account of how we truly function in producing causal explanations. Whether or not we’re speaking about skilled scientific researchers or about individuals of their on a regular basis lives, John Dewey’s basic pragmatist evaluation in How We Suppose (1933) nonetheless rings true to me: individuals begin with habits which were established and are retained exactly as a result of they work properly sufficient in apply, and when these break down, inquiry begins. We by no means begin from zero; we enter a scenario with a set of pre-existing, pre-vetted notions which kind the premise of our investigation. We’re by no means restricted to re-stating or confirming these notions, however the level is that we begin from someplace and make modifications and enhancements accordingly. So, the scenario that the authors posit — with an investigator approaching a scenario with out preconceptions, needing to search out oblique proof for propensity statements that they solely generate after validating explanatory statements — by no means happens.
Utilized to causal rationalization, Dewey’s account provides us an image that’s in some methods distinct from the one which Humphreys and Suganami present. For the authors, causal rationalization is “formed, partially, by pragmatic issues” (p. 180 fn 11). In distinction, for Dewey, as for me, explanations together with causal explanations are solely pragmatic, responding to specific problem-situations and aiming to resolve specific perplexities. Definitely, causal explanations can’t accomplish that with out empirical proof to assist them, however they’re greatest understood as efforts to construct on previously-vetted propensity statements to construct a set of directions that may resolve the perplexity and remedy the issue. The act of explaining and the act of vetting an summary propensity assertion must be saved separate, exactly as a result of we all the time enter a scenario with a pre-existing inventory of such propensity statements from which we assemble each our novel explanations and potential rivals to them.[6] If we didn’t have such a pre-existing inventory, how would we even start to find out what rival explanations to hunt disconfirming proof of? And the way would we start to suspect that some issue prompted a selected end result, except we had prior cause to consider that it no less than might need?
I very a lot agree with Humphreys and Suganami that we should always preserve clear the excellence between causal explanations and the summary propensity statements that inform them. But when we don’t additionally preserve the method of vetting summary propensity statements separate from the method of explaining particular outcomes, we find yourself with both a thriller regarding how we ever cross from empirical observations to abstractions, or with a sleight of hand whereby an summary notion is smuggled from the start into what purports to be merely an empirical remark. It’s implausible to me that in suggesting that Zorri’s unhappiness is attributable to her lack of a morning stroll, I used to be not drawing on summary notions and propensity statements derived from prior expertise. Provided that we already thought that the shortage of a morning stroll has a propensity to supply an sad canine would we ever be capable of causally clarify Zorri’s unhappiness by seeing whether or not she had a morning stroll that day.
All of that mentioned, the sensible recommendation that Humphreys and Suganami provide about in search of empirical proof to tell apart between rival explanatory statements is stable, regardless of my disagreements with them in regards to the exact relationship between summary propensity statements and concrete acts of causal rationalization. I heartily concur with their argument that problems with causal inquiry in worldwide research aren’t going to be resolved by metaphysical hypothesis or debate. The methodological facets of causal inquiry deserve the sort of consideration that the authors give them, and the dialog that Humphreys and Suganami are advancing is crucial to the evolution of the sphere. They need to be recommended for a splendid assertion and acknowledged for the richness of the dialogue about their evaluation that may hopefully ensue.
References
Dewey, John. 1933. How We Suppose, Revised Version. Digital Version. The Later Works of John Dewey, 1925-1953, Quantity 8, 1933. Carbondale and Edwardsville: Southern Illinois College Press.
Humphreys, Adam R. C., and Hidemi Suganami. 2024. Causal Inquiry in Worldwide Relations. Oxford College Press. https://doi.org/10.1093/oso/9780198823582.001.0001.
Jackson, Patrick Thaddeus. 2025. Details and Explanations in Worldwide Research…and Past. London: Routledge.
Kern, Stephen. 2006. A Cultural Historical past of Causality. Princeton, N.J.: Princeton College Press. https://press.princeton.edu/books/paperback/9780691127682/a-cultural-history-of-causality.
Searle, John. 1995. The Development of Social Actuality. New York: Free Press.
[1] By the use of full disclosure, I’ll level out right here on the outset that I’ve been in dialog with the authors about these points for a few years, and that I participated within the workshop resulting in their 2017 co-edited particular difficulty of the Journal of Worldwide Relations and Growth on causation in world politics (through which I had an article printed). My very own current guide, Details and Explanations in Worldwide Research…and past (2025), can also be partially an end result of those ongoing conversations.
[2] That is what causal agnosticism does, even because it leaves room for the realist declare about causation being a part of the mind-independent world to be probably true. The agnostic merely doesn’t need to take a place on whether or not the realist declare is true or not (p. 109).
[3] On this instance, a is “Zorri didn’t go on a stroll this morning” and b is “Zorri is sad in the present day” whereas A is “not happening morning walks” and B is “sad.” “Canine”/”canines” is a scope and area situation for each claims. The authors particularly level out the significance of the usage of totally different tenses within the two sorts of causal assertion, such that explanatory statements in English are usually prior to now easy tense, whereas summary causal statements are within the current easy tense (pp. 128-130).
[4] Humphreys and Suganami’s understanding of RCTs is thus rooted in what some philosophers name IBE — inference to the perfect rationalization (p. 224) — or what a layperson would possibly consult with because the Sherlock Holmes investigatory strategy of eliminating impossibilities so as to arrive on the fact. On parallels and connections between detective fiction and the hunt for legitimate causal explanations, see (Kern 2006).
[5] Certainly, it might even be higher to say that the components for an summary causal assertion is “A causes B in context C,” which has the added benefit of paralleling John Searle’s (1995) components “X counts as Y in context C” for social and institutional details.
[6] Plenty of what we generally name “theorizing” in scholarly work is, I might say, a type of “vetting”: not the identical as conducting a laboratory experiment, however like such an experiment, meant to elucidate the propensity—the causal tendency or energy—of some component and render it believable to the reader.
Additional Studying on E-Worldwide Relations
