Inference and Scientific Progress in International Relations

This is the third in a series of posts by several of us at The Disorder Of Things on Patrick Thaddeus Jackson‘s The Conduct Of Inquiry in International Relations: Philosophy of Science and Its Implications for the Study of World Politics. Paul started things off with his post setting up Jackson’s methodology of politics in order to ask important questions about the politics of Jackson’s methodology. Joe continued with his post and a discussion of the relationship between the scientific and the normative, and their institutionalization within IR. Next week will see a final post, followed by a reply by Jackson himself.

Update (17 Feb): Meera’s post is now up.

Inference and Scientific Progress in International Relations

From a philosophy of science perspective, IR discussions on methodology and epistemology have always struck me as a bit bizarre. The outdated nature of most debates and the odd use of labels like ‘positivism’ have made IR philosophy of science too often seem like a muddled confusion, rather than an insightful debate. So it’s hard to overstate how fantastic it is to see a book like Patrick Thaddeus Jackson’s – precisely because it weaves skillfully through rigorous philosophy of science, and doesn’t remain bound by IR’s idiosyncratic frameworks of debate. I find myself highly sympathetic to a lot of what Jackson argues for in this book, and am a strong proponent of methodological pluralism. There are two major points I think Jackson’s book neglects though – one is more based upon my own philosophical position (an external critique), while the second is a problem more or less within Jackson’s position (an internal critique). In what follows I try to examine some missing elements of Jackson’s book, and suggest what might be an alternative approach. [1]

On Monism and Dualism

Jackson begins by setting out a 2×2 matrix of different fundamental philosophical orientations (‘wagers’). These are considered ideal types that help to clarify the vast field of philosophy of science. The first distinction is between mind-world dualism and mind-world monism. It is a distinction concerning the relationship between the researcher and his or her object. The second distinction is between what Jackson calls phenomenalism and transfactualism – or what might be also known as instrumentalism versus realism about scientific objects. The former sees empirical data as all that can be legitimately said to exist, whereas the latter argues we can deduce the existence of unobservable entities as well.

Phenomenalism Transfactualism
Dualism Neopositivism Critical Realism
Monism Analyticism Reflexivity

The 2×2 matrix: A scholar’s best friend.

As Jackson is clear about the ideal-type nature of this categorization, I don’t want to criticize that aspect. Rather, my point is that in his discussion of mind-world dualism and monism Jackson leaves aside one crucially important position (and the position undertaken by many in the so-called ‘speculative realist’ movement). [2] Whereas Jackson sets the empiricist, explanation-based, ‘scientific’ perspectives on the side of mind-world dualism, he sets the social constructivist, understanding-based perspectives on the side of mind-world monism. The former tries to bridge the gap between mind and world by creating accurate representations. The latter asserts that all of reality is intertwined with linguistic and conceptual baggage. (36) (This is precisely what Quentin Meillassoux will call the ‘correlationist’ position: the reduction of Being to the relation between mind and world. )

The problem is that Jackson’s discussion of the monist position neglects one crucial position – not a conceptual monism, but a materialist monism. [3] This is a monism that doesn’t accord any special ontological position to ideas or language or phenomenology. Such a monism upsets the framework that Jackson has set up, however. For one thing, it asserts a mind-independent reality (a dualist characteristic), but it also maintains the ontological sameness of mind and matter (a monist characteristic). It can still maintain a dualist separation between concepts and objects (see Ray Brassier, “Concepts and Objects”), but what is related is not made of different ‘stuff’ (a monist position). It does, however, definitively refuse phenomenalism as an anthropological conceit, so it can be firmly placed in the transfactualist category. If materialist monism doesn’t fit cleanly into Jackson’s categorization, this is predominantly because the epistemological relation must be reconfigured away from the standard story which sees scientific theories simply being mapped onto a pre-existing world via a principle of adequation.

With regards to the epistemological question there are at least two responses, both embodied nicely in Wilfrid Sellars’ work. On the one hand, there is the group now known as the eliminative materialists (EM), who emphasize the materialist side of Sellars and try to naturalize epistemology (Paul Churchland and Daniel Dennett, for instance). For Churchland, the idea here is that neural networks operate non-representationally, and adapt to their surroundings to produce usefully accurate conceptual maps. But these approximations of accuracy between the inner and the outer are themselves simply a product of a machine-like dynamic, and not any sort of logical, rational, or empiricist methodology. (See Paul Churchland, A Neurocomputational Perspective.)

A fancy-looking diagram that just says mental maps are created and shaped through neurological networks (in this case, of the visual system).

The second approach critiques (in my view, fundamentally) this first approach on the basis of its own self-justification. Namely, if EM is right, what are the inferential paths which would lead us to it? In other words, if EM demonstrates that all inferential paths are mere adaptations of neural networks, what can the status of EM itself be other than a simple accident? EM, from this second perspective, is self-denying. The result is a transcendental argument that there has to be a normative space of reasons that exists as well. Without this space, there is no rational justification for any position. (Pete Wolfendale’s work here has been crucial in my thinking on this.) It should be noted that the sense of normativity here is much wider than simply moral-type norms – it is the very norms of rational thought (such as objectivity, truth and falsity) which are in question. What behaviour must one follow in order to draw valid conclusions? As a consequence of these ideas, coming out of Wilfrid Sellars you get a normative (rather than a materialist) strand of thought – philosophers like Robert Brandom, most notably.

Three major results fall out of this transcendental argument for normativity. First, it negates any idea of a separation between scientific discourse and normative discourse. The former necessarily presupposes the latter as the proper conduct of thought. (Note that this doesn’t mean there are universal and pre-given norms of rational thought – Brandom will argue that there is a historical construction of rationality. Thus, in parallel with the development of empirical science we can find the development of transcendental norms.) Second, that there can indeed be a synthesis of the natural and the normative, along the lines of what James O’Shea has said about Sellars: Norms are conceptually irreducible, while being causally reducible. That is to say, we require them for consistent logical thinking, yet they are not situated above and beyond the material world and practice. That is to say, it’s not a dualism. This leads us to the third point – the conclusion that what is necessary for thinking through scientific thought is to proceed through the very notion of scientific inference.

Robert Brandom: Keeping beards in style since 1950.

On Inference

It is here perhaps that Jackson’s work falters most significantly – specifically, through the lack of any explicit means to compare and contrast the results of different methodologies. While he acknowledges that such interaction occurs, what is required is an explication of scientific inference that allows for commensurability between different methodologies. Without this, the placing of different theoretical orientations into different silos (so prevalent in contemporary IR) is inevitable. One of the more promising ways to potentially resolve this problem is through a development of inference – albeit in a much wider sense than King, Keohane and Verba give to the term. This would involve at the very least an understanding of the structure of inferential networks, how they provide meaning to terms and statements, and lastly, how their strength of connections is determined. In the work of Robert Brandom we can find important resources for the first two demands, while with Bruno Latour we can find an elaboration of the third demand.

An adequate picture of Brandom’s infererentialist position is impossible to give here (instead, there is his 740pp book that makes for a good afternoon read), but I’ll try to point out a few interesting aspects. First, inferentialism sets itself apart from the traditional truth-correspondence theories of meaning. Rather than statements matching up with some referent in the world, words receive their meaning by being situated within a network of inferential relationships with other concepts and – crucially – with objects. We can see here, then, that inferentialism already avoids the sort of mind-world dualism that leads to the logical problems that tend to surround correspondence theories of truth (what Jackson will call ‘the Cartesian anxiety’). Second, Brandom’s inferentialism is intrinsically embedded into social practices – in this sense it falls into what Jackson calls ‘analyticism’. In fact, for Brandom, the acts of giving reasons at the heart of the normative order can be precisely modeled by an elaborate game that he develops in his major work. The game of science, therefore, is a game of reason-giving that constructs the inferential networks that tie together various statements. The potential problem with Brandom’s work is that with all the focus on normativity and the space of reasons, it may fall into a (sophisticated) idealism.

The first way to avoid this is to argue that by justifying scientific practice and its results, Brandom implicitly at least lends support to the idea of science being progressive in the sense of producing more refined concepts, explanations, and predictions. If science can be considered progressive, however, then the ‘no-miracles’ argument can be employed to demonstrate that science must at least be grasping onto some real features of the world. These need not be absolute claims, incapable of being revised; instead they can simply be the best approximations we have to reality so far. But they are not simply ideal types grounded upon particular value-orientations, for instance.

As a result, the difference between Brandom’s inferentialism and Jackson’s analyticism is that whereas analyticism remains bound to the phenomenal level, the sort of inferentialism that Brandom develops can be wedded to a realist conception of science. It can produce transfactual knowledge about the mind-independent world, in other words. Thus while the inferential network may appear to be disconnected from the material world, it is in fact guided by norms of truth to produce knowledge claims that ever more closely (if asymptotically) approximate the features of real objects and their relations.

With Bruno Latour’s work, we can further tie the normative network of inferential relations even tighter to the material world. As we will see, central here is Latour’s rejection of the ‘modernist’ gap between mind and world, subject and object (or whatever one wants to call it). This will allow us to supplement Brandom’s lingustic-based conception of networked inferences, with a (paradoxically) non-lingustic type of inference. To see how this works, it needs to be understood that ultimately, for Latour, science is a matter of creating statements that are increasingly difficult to overthrow. Contra the naïve vision of science, this increased difficulty arises not simply from a theory being ‘true’ in the correspondence sense. Rather, for actor-network theory, this increased difficulty is tied up with two analytically distinct aspects. First, any particular statement must be made to fit in with things like the existing institutional structures, the various practices and standards of citation, and the personal networks between scientists. All these typically sociological aspects play a role in supporting a particular statement and making it increasingly difficult to criticize. This type of sociological influence, though, is not simply about power relations, economic interests, or cultural biases. It’s important to note that the types of sociological aspects that Latour is primarily interested in are what we would consider the everyday activities of an academic – taking note of existing theories, citing established evidence, looking for weaknesses in other’s research, etc. The sociological aspects of Latour are not straightforward distortions of the scientific project therefore.

Science? Yeah, it’s a bit crazy.

But for Latour, the stabilization of a statement also includes an important second aspect: the role of nonhuman actors as support. So the difficulty of overturning any particular scientific statement is simultaneously bound up with things like the instruments of science, the laboratory results, the products of field studies, and the outcomes of experiments. All these different aspects contribute to stabilizing a statement, and making it increasingly difficult to overturn. Similarly, the instruments used make a difference in how stable a statement is. For instance, the evidence provided by a mass spectrometer is worth more than the evidence provided by a chromatograph because the former embodies a longer line of research and theories within physics. Destabilizing the evidence it produces means destabilizing the numerous theories upon which it is built, which most scientists would be loath to do. Any particular statement, therefore, is built upon a large network of instruments and results from preceding experiments. Note that this is a non-linguistic series of inferences – put philosophically, it is about continuously transforming matter into form, and using the latter to derive conclusions (and is premised, again, on a monist ontology). (Latour’s excellent essay, “Circulating Reference” in Pandora’s Hope makes all of this clear with reference to an empirical case study.) In other words, it operates at an analytically (though not ontologically) distinct level from Brandom’s space of reasons. Crucially, there is no absolute ground here – any particular knowledge claim is contingently correct, and only supported through its support by materials, and through its role in inferential networks. Science, in other words, is always contingently revisable.

On the Progress of Science

The second oversight I see in Jackson’s position is the unarticulated notion of progress that is employed. As he rightly says, every science worth its name proclaims its progressive nature (194), yet what progress consists of for him is left aside. This is important for one crucial reason: without a notion of progress it becomes impossible to adjudicate between the competing methodologies. In fact, this is precisely Jackson’s position: the methodologies are incommensurable and each has incompatible means of evaluating progress. Yet it’s not clear then in what sense we can call this science rather than sciences in the plural. Can they really be said to be working on a common project? Furthermore, in practice the methodologies are often put in competition – an econometrics model may produce knowledge-claims about the relation between education and salary, whereas a reflexivist account may look at education as an ideological state apparatus that only relates education and salary as a result of the capitalist mode of production. How are we to say which is better? How are we to relate them to each other? Different methodologies produce different knowledge-claims, despite ostensibly speaking of the same object. Either they can speak to each other (in which case, how?) or they are divergent and we have multiple sciences. It is the notion of progress that can provide a unifying goal or measure for these different methodologies (with inference providing a unifying medium).

From what has been said above about inference and its expansion via Latour, I think it’s possible to derive a few aspects of a more general form of progress (i.e. one not limited to evaluating on the basis of a particular methodology). The first aspect of progress is the most intuitive sense: that of making conceptual, explanatory and predictive progress. Simply put, these become more accurate over time (with accuracy being measured by explanatory and predictive success, and not through correspondence). Importantly, this also allows for the overcoming of the supposedly mutually incompatible nature of different paradigms. As Philip Kitcher argues with direct reference to Kuhn, such incompatibilities are overstated when conceptual and explanatory progress are examined after the fact. For instance, the conceptual object of ‘phlogiston’ was picked out both by various experimental results as well as by a descriptive definition. While some of these characteristics were correct (insofar as they allowed for more accurate and novel predictions), others were later found to be fundamentally flawed. The downfall of the phlogiston theory of heat was not a complete rejection of a previous theory, but rather a conceptual refinement. Using this framework, conceptual progress can therefore be seen as the progressive fine-tuning of a term so that false modes of reference are dropped, and/or other modes of reference are narrowed down. A similar measure of progress exists as well for explanatory progress, where the web of dependencies between terms is given a progressive formulation on the basis of their abilities to improve explanatory accuracy and completeness.

The second major notion of progress we can derive is from Latour’s work – the strengthening of connections and the extension of supports for any particular claim. As any given knowledge-claim is only produced out of a long series of translations between matter and form, it follows that to stabilize and make stronger these connections, one can extend and expand the inferential connections supporting the claim. Thus progress in this sense can be measured via this strengthening over time.

The third type of progress stems from the embedded nature of mind into matter. This radicalizes the reflexivist position that Jackson sets out by undertaking not merely a critical reflection on the social structures that condition knowledge, but also by reflecting on the material structures as well. It is a progressive incorporation and refinement of the mind-world relation on both a scientific and philosophical level. In a long, but important passage, Gabriel Catren has formulated this position precisely in a recent work:

“Far from being an extrinsic philosophical operation capable of localizing the insurmountable limits of science, transcendental critique must allow the subject of science to identify and speculatively subsume the various transcendental conditions of scientific research. Among these conditions we can include: the (gravitational, thermodynamic, biologic, etc. conditions that make the emergence of localized and temporalized cognitive entities possible; the conditions defined by the anthropic principle; the physiological conditions of sensible intuition; the technological conditions of instrumental observability and experimental verifiability; the associated limits to the possibility of gaining empirical access to the different regions, strata, scales, and dimensions of the real; the ‘categories’ of human understanding, the available ‘imaginary’ schemata that allow us to connect these categories with sensible intuition, the formal and linguistic structures that convey theoretical reason, and the technical and conceptual operations of analysis, synthesis, abstraction, selection, coarse-graining, decoherence, and renormalization through which we can constitute finite objects and define what is relevant at a given stage of research. […] Indeed, the problem is not how to pierce a hole in the walls of the transcendental prison (built by philosophy itself), but rather to acknowledge that transcendental reflection is a necessary moment for absolving knowledge from the too human transcendental conditions of research. The infinite process of theoretical knowledge does not advance by attempting to grasp an ‘uncorrelated absolute’ through a philosophical ‘ruse’ capable of discontinuously leaping over the subject’s shadow, but instead through a continual deepening of scientific labour seeking to locally absolve it from its conjunctural transcendental limitations, expand its categorical, critical, and methodological tools, and progressively subsume its unreflected conditions and presuppositions. [4]

With Catren’s suggestion, we can begin to move away from the more or less incommensurable methodological positions that Jackson set out. Rather than accept the division between methodologies (and the inability to adjudicate between them), the scientific project can use the advances of science to reciprocally provide advances on their transcendental conditions. In other words, the progress of science is what makes possible the ability to sharpen our own philosophical (ontological, epistemological, methodological) assumptions.

While there are major difficulties remaining in the position briefly sketched out here, I hope that it can at least suggest a possible path forward that retains methodological pluralism, scientific progress, and materialist monism.


The likely response to the position put forth here is that it overlooks the massively contested nature of ‘science’ and its components within the philosophy of science field. As a result, the position here attempts to unify what is (for Jackson) intrinsically plural. This is true, to a degree. Reiterating some of the arguments made above, three points can perhaps be made against remaining at a pluralist level. First, that in practice scientists (in Jackson’s expanded sense of science) are always setting claims against each other, even when having different methodological assumptions.  There is an interaction going on in practice, yet this seems to go unaccounted for by Jackson’s work.

Second, the presumably unwanted outcome of Jackson’s pluralism is that we have four different and incommensurable fields of knowledge. To call all of these ‘science’ seems like a denial of the very different claims they produce, and the very different uses they aim at. This pluralism, in other words, seems like a deep pluralism (despite Jackson’s attempt to provide an all-encompassing definition of science).

Third, to repeat what was said before, there is no sense of general progress that could be derived from all four methodologies. Instead, at best, there would seem to be progress internal to each methodology, but not in any sort of overall sense. Indeed, Jackson uses this as an argument for why there is necessarily a plurality of methodologies – since there is no overarching framework for evaluating them. (208)

Yet it is hoped that the senses of progress and inference outlined above give some suggestion to how an overarching idea of progress might be constructed. What ultimately turns out to be a ‘better’ argument (i.e. methodology) will be partially contingent upon the situation it finds itself in (e.g. the dominance of neopositivism in IR is an example of a social influence on what is currently determined as a ‘better’ argument – but this is not a historical necessity). Yet there can be (and has been) philosophical and empirical progress that gives us more precise notions of the assumptions which divide methodologies (this is the importance of Catren’s position). If this is the case – if methodologies can be made to compete at the level of their assumptions (through empirical and philosophical inferences) – then it is not inevitable that they remain separate.

On International Relations

How does this all relate to the study of international relations? In many ways, I am in agreement with Jackson that the immediate consequence should be an opening up of the methodological space within IR. The great worth of this book is precisely to counter the misguided notions of what ‘science’ is within IR. One of the primary reasons I ended up at LSE to study IR was because of its methodological and theoretical openness.  And one of the potentially major roadblocks for more traditional IR is its inability to innovate, precisely because of its methodologically conservative nature. (My favourite, if depressing, examples are of quantitative researchers who produce statistical evidence for blatantly obvious truths.) This is not a request for less rigorous work, but instead a request to recognize the multiple ways in which conceptual objects are constructed, and the multiple ways in which they fit into inferential networks. It is a call for more clarity about the relationship between different knowledge claims, and how conceptual objects can be produced that are mutually overlapping even if not directly comparable. There is scientific progress, and on the basis of this we can say there is something like scientific truth, yet formal and quantitative methods do not hold a monopoly on this.


[1] I should stress as well, that as is so often in writing a piece, it’s only when one attempts to work out the connections of logic precisely on paper that gaps in the flow become apparent. There are a number of gaps I’m aware of in this piece, and I’m sure there are many more others can point out. So the positive project set out here should be considered a prolegomena to an introduction, at best!

[2] Speculative realism is a relatively new family of philosophical positions, united largely by their resistance to Kant and post-Kantian philosophy. Whereas Kant attempted to set limits on the ability to know the noumenal realm, and post-Kantians have maintained this in various ways, speculative realism argues that a relation to the noumenal can be sustained. For an overview of the entire emerging field, see (blatant self-promotion warning): The Speculative Turn: Continental Materialism and Realism. It should be made clear that what follows is my own position in the broad field of speculative realism. Others have different and varying approaches to studying the social and while I draw on many of them for inspiration, my own position diverges from any particular thinker.

[3] To be clear, this is a ‘materialist’ monism and not a ‘physical’ monism – the latter is a specific scientific ontology (to use Jackson’s terminology), whereas the former is the more fundamental philosophical ontology that underlies theories like physicalism. It argues that what we call minds are a small piece of a much larger world – they are objects alongside other objects, rather than being the frame through which all of reality is perceived. In this sense, therefore, materialism accords no special place to minds and their characteristics.

[4] Catren, Gabriel. “Outland Empire,” in Bryant, Levi, Nick Srnicek and Graham Harman (eds.) The Speculative Turn: Continental Materialism and Realism, Melbourne:, 2011, p. 342.


Brandom, Robert. “Inferentialism and Some of its Challenges,” Philosophy and Phenomenological Research, 74:3.

Brassier, Ray. “Concepts and Objects,” in Bryant, Levi, Nick Srnicek and Graham Harman (eds.) The Speculative Turn: Continental Materialism and Realism, Melbourne:, 2011.

Catren, Gabriel. “Outland Empire,” in Bryant, Levi, Nick Srnicek and Graham Harman (eds.) The Speculative Turn: Continental Materialism and Realism, Melbourne:, 2011.

Churchland, Paul. A Neurocomputational Perspective: The Nature of Mind and the Structure of Science, MIT Press, 1992.

Jackson, Patrick Thaddeus. The Conduct of Inquiry in International Relations, London: Routledge, 2011.

Kitcher, Philip. The Advancement of Science: Science without Legend, Objectivity without Illusions, New York: Oxford University Press, 1993.

Latour, Bruno. Laboratory Life: The Construction of Scientific Facts, 2nd Edition, Princeton: Princeton University Press, 1986.

Meillassoux, Quentin. After Finitude, New York: Continuum, 2008.

O’Shea, James. Wilfrid Sellars: Naturalism with a Normative Turn, Polity Press, 2007.


4 thoughts on “Inference and Scientific Progress in International Relations

  1. Pingback: Realist Social Science, and Other News « Speculative Heresy

  2. Pingback: Freeing the Pluralist Imagination, or on the wisdom of escaping Weber’s “Iron Cage” « The Disorder Of Things

  3. Pingback: Demarcation Problems: ‘The Conduct Of Inquiry’ Between Politics & Methodology « The Disorder Of Things

  4. Pingback: Of Consensus and Controversy: The Matrix Reloaded « The Disorder Of Things

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 )

Connecting to %s