Understanding international practices from the internal point of view

Introduction

This article aims to provide a conceptual analysis of practices from what we term the internal point of view. Even though in what follows international practices within the field of International Relations (IR) will receive special attention, our principal aim is to elucidate the more basic, philosophical problem of practices. David Stern articulates this problem aptly:

What is a practice? No short answer will do here. At the very least, a practice is something people do, not just once, but on a regular basis. But it is more than just a disposition to behave in a certain way: the identity of a practice depends not only on what people do, but also on the significance of those actions and the surroundings in which they occur. This is only to begin to answer the question, how we are to understand ‘what people do’ when they are engaged in a practice, or just what a practice amounts to. (Stern, 2003: 186)

What complicates matters is that different IR scholars have drawn on different disciplines and vocabularies in developing their positions within the currently burgeoning debate on international practice theory also known as the ‘practice turn’ in IR (Adler and Pouliot, 2011a, 2011b; Brown, 2012; Bueger, 2012; Bueger and Gadinger, 2014; Jackson, 2008; Kessler and Guillaume, 2012; Kratochwil, 2006, 2011; Neumann, 2002; Pouliot, 2008, 2010; Ringmar, 2014). ¹ For example, Pouliot (2010) has based his international practices approach on Pierre Bourdieu’s (1977 [1972], 1990 [1980]) theory of practice, Kratochwil (2011) has connected his view of international practices to the tradition of American pragmatism (Dewey, 1981; James, 1995 [1907]; Peirce, 1932–1935) and Bueger and Gadinger (2014) have identified the importation into IR of five strands of practice theory originating in sociology – Bourdieu’s praxeology, actor-network theory, community of practice, narrativism and Luc Boltanski’s pragmatism. Such disparate articulations of practices leave us wondering what, if anything, makes a practice-based approach coherent.

To restore coherence, we propose a return to the internal point of view. Our central thesis is that the proper understanding of practices requires the internal point of view (internalism). To understand something is to recognise or make sense of it as a this or a that. Internalism, as presented here, is a fundamental standpoint for making sense of any social practices whatsoever, be they local or global, domestic or international, cooperative or conflictual. Practices, as a first approximation, are distinctive action domains inside which multiple individuals participate by being guided by common standards of action such as rules or norms. These rules and the attendant usages requisite for following the rules are ‘intelligibles’ that have meaning (Oakeshott, 1975). The premise is that we cannot qualify as participants in a practice unless we are able to understand what its constitutive rules and usages mean. We may seek to understand a practice for two principal reasons: either because we wish to understand what other agents, who are practice participants, do or have done in the past, or because we aspire to become participants ourselves. Either way, a proper understanding of any practice demands an internalist perspective.

To support our case, we differentiate between two basic standpoints for making sense of practices – the external and the internal one. Upon explication, they will be shown to be incongruent and not merely different. Their opposition resembles that between Verstehen (‘understanding’) and Erklären (‘explanation’), between hermeneutics and natural science, but is not identical to it. ² Our starting assumption is that the primary object to be understood or explained is a practice – a domain of intelligent activity that is already understood by its participants. Irrespective of whether we are engaged in the explanation of practices (practice externalism) or in the understanding of practices (practice internalism), such practices belong to the intelligent world of Verstehen.

The problem of how we are able to make sense of practices is akin to the problem of understanding and translating languages. The core distinction between practice internalism and practice externalism (‘internalism’ and ‘externalism’ for short), as we shall see, is closely linked to this conception of practices as language analogues. ³ Analytically, the distinction turns on the relation between the identity of the practice observed, on the one hand, and the language of observation used to describe it, on the other. ⁴ In the present context, ‘observation’ is a technical term broader than ‘scientific observation’ (Harman, 1977: 3–11). An act of observation has two aspects: (1) identity conditions specifying what type of object is being observed and (2) a language of observation (which may affect the identity of the object). In this discussion, we ask how an observer is able to make sense of a practice. To avoid confusion, we shall use the expression ‘language of action’ for the common understandings espoused by the participants in a practice, and ‘language of observation’ for the terms of the procedure employed by an observer who studies their practice. Our internalist thesis suggests that to be able to establish that a putative identity, a practice P, exists, the observer must use terms which the practice participants themselves use in understanding their own practice, P. Internalism, in brief, holds that the language of observation must match the language of action used inside the domain of a practice; externalism denies this.

It is not feasible to examine the whole spectrum of internalist or externalist positions available. Thus, we have picked views representative of each position. For practice internalism, we have chosen to focus on the ideas of Peter Winch (1958, 1972a [1959–1960], 1972b [1964], 1987 [1976]), a philosopher heavily influenced by the later Wittgenstein (1953, 1978). Similar internalist arguments have been developed by John Rawls (1955), HLA Hart (1961) and Michael Oakeshott (1975). ⁵ Winch characteristically departs from the orthodox view that social science is natural science applied to the study of society. In The Idea of a Social Science and Its Relation to Philosophy (Winch, 1958), he portrays social science as a practice in its own right (Gunnell, 2009), constituted by the commitment of its practitioners to the external point of view. As a prime form of externalism, social science relies on a uniform observational language which makes it unsuitable for capturing the puzzling diversity of ordinary human practices. ⁶ The task of philosophy is to recover the internal point of view and to reclaim the proper understanding of ordinary human practices that acknowledges their differentia. It is this Winchean insight that we wish to bring to bear on the discourse of practices in IR, as a sombre reminder that the status of IR as a social science might need re-evaluation. ⁷

Practice externalism is represented here by the writings of Adler and Pouliot (2011a, 2011b). For many, their ambitious and programmatic writings articulate the very ‘turn’ to international practices. The practice turn in IR has been examined by Chris Brown (2012) in the context of an Aristotelian international political theory and by Erik Ringmar (2014) via a critique of the cumulative growth of science. Employing a distinct set of premises that prioritise the externalism–internalism distinction, in the section ‘Externalism and international practices’, we set out to evaluate Adler and Pouliot’s externalist position. As we proceed with outlining the landscape of internalism and its difference from externalism, we raise basic questions about meaning, language and practice-based action. The section, ‘Understanding practices from the internal point of view’, is devoted to Winch’s internalist perspective on practices. Although Winch was preoccupied with reworking a notion of rule-following derived from Wittgenstein, the emphasis throughout is on Winch’s argument. ⁸ We conclude by canvassing a few core implications of the proposed internalist approach for the study of the global practices in IR.

Externalism and international practices

We now turn to Adler and Pouliot’s account of international practices. They define practices as ‘competent performances’ where the term performance is used interchangeably with ‘an action’ (or ‘meaningful doing’) (Adler and Pouliot, 2011a: 4–5, 2011b: 6–7). With respect to understanding practices, they favour the procedure of direct observation. For Adler and Pouliot, making sense of international practices demands looking at what top practitioners and decision-makers in IR do (2011a: 1–2, 28–30, 2011b: 3). For instance, to understand the practice of post-Cold War diplomacy between NATO and Russia, we must study the actions of NATO and Russian security practitioners in the post-1992 era (Pouliot, 2010). The premise is that what top IR practitioners do, the identity of their actions (practices), is intelligible as it stands. ‘Defining what counts as an international practice and what does not’, Adler and Pouliot write, ‘is best left with practitioners themselves in their actual performance of world politics’ (2011b: 7; see also 2011a: 6). In a positivist vein, it is assumed that the identity of action under investigation is transparent: it does not depend on a description used by the analyst.

Adler and Pouliot refine this preliminary picture by specifying five key characteristics of practices. They hold that a practice:

(1) ‘is a performance’

(2) ‘tends to be patterned, in that it generally exhibits certain regularities over time and space’

(3) ‘rests on background knowledge, which it embodies, enacts and reifies all at once’

(4) ‘weaves together the discursive and the material worlds’

(5) ‘is more or less competent in a socially meaningful and recognisable way’ (Adler and Pouliot, 2011a: 6–7, 2011b: 7–8, original emphasis)

This set of features is accompanied by an eclectic methodology that allows the researcher to pick and choose the object of inquiry at different levels of analysis (Adler and Pouliot, 2011a: 8, 2011b: 9). Thus, certain practices can be defined either widely or narrowly, as ‘general practices’ or as ‘specific practices’, and they can be seen as components of other, more complex practices or ‘constellation of practices’ (Adler and Pouliot, 2011b: 9, 20). Practices can be examined in terms of their ‘lifecycles’, as going through phases of generation, diffusion, institutionalisation and fading (Adler and Pouliot, 2011a: 24, 2011b: 17, 19, 23). What is more, practices can be recruited for various explanatory purposes. They can be an object of explanation (explanandum) as well as explanatory factors (explanans) (Adler and Pouliot, 2011b: 5, 18–19). One set of practices can be invoked to explain the change over time in another set of practices (Adler and Pouliot, 2011b: 19). Practices, in the sense of explanans, are ‘bundles’ of ideas, material and ideational structures and agency (Adler and Pouliot, 2011a: 14, 16, 2011b: 13, 15).

Apart from explanatory categories, practices operate as ontological ones. ‘Practices not only organise the world – they are also the raw materials that comprise it’ (Adler and Pouliot, 2011a: 15, 2011b: 13). Practices are the basic ontological entities that cut across the key theoretical paradigms of IR and allow for the emergence of a common language between them (Adler and Pouliot, 2011b: 10). Indeed, ‘the important reason why dialogue across paradigms via a focus on practice is a real possibility is the concept’s broader ontology which blends material and ideational factors, as well as structure and agency, into social doing’ (Adler and Pouliot, 2011b: 12; see also 2011a: 16). Those familiar with IR theorising would recognise in these statements an ambitious attempt to bridge two of its central meta-theoretical debates – ideas versus material factors and agents versus structures. ⁹ These debates pertain to matters of ontology or the ‘stuff’ composing the world of international relations.

The most perspicuous facet of Adler and Pouliot’s approach is that practices are taken to be the ‘gluons’ (2011b: 10) of the universe of international affairs. ¹⁰ Practices are atom-like units. Ontologically, each complex practice has a molecular structure: each can be aggregated or disaggregated depending on the analyst’s methodological preferences. The twin premises of methodological and ontological atomism enable a pick-and-mix approach to conducting social science. For instance, the social scientist might disassemble a practice and then focus exclusively on its material, or on its ideational, or on its structural components, and conversely, a set of individual practices can be combined into a ‘constellation of practices’ (Adler and Pouliot, 2011a: 8, 16, 2011b: 9, 17). Starting with a sociological account of the components of individual practices, the researcher can build up to a sociology of the practice of practices. The point is that a complex practice can be broken down and then put back together without loss of identity. This is permissible because the identity of a complex practice is defined in atomistic terms; it is an aggregate – a sum of elementary practices. As we shall see, this endorsement of atomism is typical for externalism.

Furthermore, Adler and Pouliot embrace an intersubjective conception of practices. Curiously, they cite Wittgenstein as one of its precursors (Adler and Pouliot, 2011a: 7, 2011b: 8n20, 12). Practices (according to their fifth thesis above) are performances ‘that are more or less competent in a socially recognisable way’. This invocation of Wittgenstein (1953) is insightful, for he is usually read as a defender of an intersubjective or ‘public’ language (1953: I, §§198–199, 202, 240–242). However, at this juncture, Adler and Pouliot’s attempt to ground their analysis on insights derived from Wittgenstein encounters a major problem. What is at stake is the question of internalism versus externalism or the relationship that can be assumed to hold between the object of observation and the language of observation used to describe it. Adler and Pouliot assume that one and the same procedure of observation can provide us with observational access to an array of different practices (2011b: 9–10). This is an externalist premise. It states that the language of observation (the investigator employs in the procedure) is neutral: it does not affect the identity of what is being observed – and the language that best fits the requirement for neutrality is a scientific one. In sharp contrast, Wittgenstein and Winch insist that any valid procedure of observation is bound to the concrete practice being observed: this is the message of internalism.

What Wittgenstein suggests, and is taken up by Winch, is that the observer must occupy a position inside a practice in order to understand what its own participants do. Internalism effectively collapses the distinction between observer and participant, between the language of observation and the language of action. The rationale obliging us to reason from within the practice is that practices are not reducible to one another. As Wittgenstein remarks in a well-known passage (1953: I, §66), no invariant set of properties is common to all practices, at most some practices share loose ‘family resemblances’ (1953: I, §67). It is therefore futile to search for an invariant observational language (science) that can reflect the foundation of practices. There is no such foundation. The avenue left open to an investigator is to study each practice on its own terms: as a concrete domain of action constituted by concrete rules and usages. This brings us to the master notion of rule-following which Wittgenstein (1953: I, §§145–155, 179–192, 195–219, 1978: VI) deploys to show that each practice is constituted, as a whole, by its own (domain-specific) standards of action (rules) that cannot be reduced – without loss of identity – to the standards of another practice.

Rule-following is an intricate notion that needs some elaboration. In paragraph 237 of his Philosophical Investigations, Wittgenstein (1953) writes,

Imagine someone using a line as a rule in the following way: he holds a pair of compasses, and carries one of its points along the line that is the ‘rule’, while the other one draws the line that follows the rule. And while he moves along the ruling line he alters the opening of the compasses, apparently with great precision, looking at the rule the whole time as if it determined what he did. And watching him we see no regularity in this opening and shutting of the compasses. We cannot learn his way of following the line from it. Here perhaps one really would say: ‘The original seems to intimate to him which way he is to go. But it is not a rule’.

What Wittgenstein means is that by the mere act of looking, we cannot tell whether the person who uses a pair of compasses, moving them along a line, follows a rule or not. Yet as it stands, Wittgenstein’s idea remains fuzzy. Winch’s major contribution is to have extended, refined and clarified this idea. To follow a rule, Winch says, is to know that there is a correct, as opposed to an incorrect, way of performing a given action: ‘If it is possible to say of someone that he is following a rule that means that one can ask whether he is doing what he does correctly or not’ (Winch, 1958: 32). Action, so construed, is based on rules that constitute criteria for its correct performance.

We must now consider the link between rule-following and social practice. The concept of rule-following does not by itself reveal what kind of audience is supposed to understand and accept the rule(s) in question. Wittgenstein pondered the scenario of a single person who is being taught to follow a rule (1953: I, §143). Imagine that this is a ‘+2’ rule of addition generating a series…0…2…4… .The rule offers a public constraint: there is a correct way of applying the rule, even though it serves to constrain the inner thoughts of a single thinker. A private rule-following has a public aspect absent from a private language (Wittgenstein, 1953: I, §256) – a language of inner sensations which its originator, and no one else, could understand. A rule, ex hypothesi, is a non-psychological entity (it does not refer to inner sensations): it is a public device. Suppose further that the solitary rule-follower continues the previous series by writing:…1000…1004…1008… . Is this a new rule (a ‘+4’ rule) or a misapplication of the old (‘+2’) rule? (Wittgenstein, 1953: I, §185). To be able to tell, we have to know what it means to succeed in following a rule. This success requires the agent to be able to extrapolate the same instruction and apply it to future instances of the same type. ‘The use of the word “rule” and the use of the word “same” are interwoven’, Wittgenstein (1953: I, §225) writes. Responding to this, Winch (1958) interjects: how do we know what counts as the same? To define the same we would need another rule (1958: 61, 83), and this would lead to a vicious regress. To eschew the regress, a practice, an entire domain of interconnected rules must be presupposed as a background. This insight marked a major change in Wittgenstein’s outlook that took place in the mid 1930s (Stern, 1991; see also Ben-Mehanem, 1998: 119). On this view, the rules and their applications are internally connected within a social practice. The practice is a social – or common – domain of activity because a community of participants jointly accepts its rules (this is the so-called ‘community view’; Bloor, 1997: 71; Canfield, 1996). The rules to be followed, then, are not ‘public’ in the sense of standards of reasoning external to the individual’s mind but ‘public’ in the sense of common standards of action internal to an entire practice. We say entire because a practice attains intelligibility at the level of the whole, as a ‘form of life’ (Wittgenstein, 1953: I, §241). Our conception of practices, ‘language-games’ in Wittgenstein’s phrase, ¹¹ refers to complex social wholes constituted by rules, a view known as practice holism (Stern, 1991, 2003: 185).

Examples of language-games include board games such as chess; sports games such as boxing and football; conferences; ceremonies; rituals; moral, political and juridical institutions; international practices such as war, diplomacy and balances of power. Each language-game is constituted by distinctive criteria that determine when an action falling inside its domain has been correctly performed or not. The upshot is that while practice internalists like Wittgenstein and Winch distinguish between an action (what one does) and a practice (a rule-constituted framework) within which the action is performed, practice externalists like Adler and Pouliot use these two terms – practice and action – interchangeably. This disagreement over the root conception of action has two important corollaries.

First, Adler and Pouliot adopt an externalist view of practices. To adopt such a view is to deny that action is inscribed in a specific framework of rules: a language-game. This amounts to a denial that there are criteria for correct or incorrect action performance; these criteria define the complex of rules constituting the practice (language-game) within which the action takes place. There is a second corollary: if there are no criteria which allow an observer to distinguish between mistaken, as opposed to correct ways, of performing a certain action – then the observer is free to impose a self-chosen description on the behaviour under observation. The analyst in effect imports his or her own criteria for identifying the actions observed. Since these criteria do not derive from the practice itself but are brought in by the analyst, from the outside, they represent external criteria. The investigation of practices from an externalist perspective may be very complex, subject to requirements such as consistency, reliability and parsimony. But the defining tenet of externalism is that it permits the researcher to use a language of observation that does not match the language of action used by the agents under observation. At the limit, in highly mathematical domains such as game theory, ¹² there is hardly any connection between these: the language of observation here is of a different logical order when compared to the language of action. The researcher cum externalist, therefore, is not constrained by an internalist requirement to grasp what the actors themselves would take to be competent performances, and research need not be conducted in a language intelligible to these actors. Whatever actions are thus identified as objects of observation by social scientists is the product of an external, scientific procedure. They are not a product of understandings achieved from the internal point of view where the language of observation matches the language of action.

Winch illustrates this problem by discussing the African Azande whose actions are being studied by an anthropologist. Where the internal point of view would enable us to identify the Azande as performing the action ‘consultation with an oracle’, which presupposes as its framework the practice of witchcraft and magic, the external point of view would take them to be performing what may be described as ‘irrational action’ (Winch, 1972b [1964]: 16–17, 22–23, 37–38). This irrationality might be explained by claiming that the Azande act out of motives of uncertainty about economic production in the future (bad crops, etc). ¹³ In IR, we might illustrate the same point with reference to Islamic State. We outsiders (i.e. we who are not members of this Jihadi movement) might describe the actions of members of the Islamic State in any number of ways referring to a holy war, or an insurgency, or clan based war, or a response to Western imperialism, or a war of Sunni against Shia or in some other way. But until such time as we have a thorough-going internal perspective, we shall not know what they are doing.

Why favour an internalist account of action over an externalist one? The answer is that internalism allows us a proper access to the inner space of meanings constituting a practice: it allows us not just to see it from some standpoint but from the standpoint which its own participants invoke in making sense of their actions, even – and this is key – when to us this sense might appear to border on nonsense. In the Azande example, to identify the action consulting the oracle, unambiguously, as an irrational attempt by the Azande to deal with uncertainty about future crops, is to commit what might be termed the externalist fallacy: to assume that all action must be understood in terms of universal criteria, rationality in this case. ¹⁴ But for the Azande (and for many believers in pagan religions), consulting an oracle is not an action properly understood in terms of instrumental reason. Until we comprehend the standing and expectations that the Azande themselves confer on oracles, we shall not know properly what they are doing in their consultative ceremonies. Just as someone not acquainted with the meaning attached to sounding the last post at a memorial ceremony in honour of those who fell in the First World War, would not know what the deed means in the absence of knowledge about the meaning of war in the practice of states and in the absence of knowledge about a complex set of beliefs regarding memorial ceremonies. It certainly would be wrong to see this as a bit of irrational proto-science directed towards resurrecting the dead.

The externalist argument for practices is vulnerable on another count. Recall that Adler and Pouliot claimed that all practices have five elementary traits – they are performatives, patterned, competent, draw on background knowledge and weave together the discursive and material worlds. To see these traits as invariant across different practices is to adopt the external point of view – its cardinal premise is that a uniform language of observation can be employed to analyse different practices. We may now ask to what extent these invariant traits advance our knowledge of practices that we do not yet understand from an internalist view. Consider the practice of beheadings carried out by Islamic State. To be informed that this practice, like all practices, will have the five features mentioned, does not seem to further our grasp of this practice of beheading very much. What we want to know is, is it a religious practice? Is it a practice of terrific drama? Is it a practice developed in retaliation for the killing of innocent civilians in response to the bombing raids of Western air forces? Many other possible accounts are available. We cannot decide which one of these is the appropriate one until the actions observed are understood in the context of the practice as the Jihadis themselves understand it. To express the point differently: the five components of practices that Adler and Pouliot enumerate are too abstract to be informative. Suggesting that a concrete practice one does not yet understand is patterned, material, ideational, and so on, does not facilitate the understanding of that practice any more than being told that ‘all languages contain words’ can help us understand Gaelic.

What stands in the way of abstract, social scientific analysis is the fact that practices are irreducibly different. The differences are ‘irreducible’ since practices do not have a common foundation, a base for reduction, as Wittgenstein (1953: I, §66) warned us. To differentiate between two or more practices, then, demands of us that we acknowledge that they present mutually distinctive meaningful wholes – or to put it another way – that each practice is a distinctive unity, a ‘form of life’. This sort of difference cannot be expressed on the model: a ‘base of invariant components’ (reflecting the invariance across practices) plus an ‘overlay of changing components’ (reflecting the difference between practices). As Winch notes, it would be pointless to ask a person familiar with both football and cricket which one would be the ‘right’ game to play: they are not variants of the same game, and one must simply choose between them (1987 [1976]: 201). But if each practice is a meaningful whole, its examination requires a holistic standpoint sensitive to meaning. The value of internalism is that it offers us such a standpoint.

Understanding practices from the internal point of view

The internalism we put forward rests on meaning holism of a particular sort: practice holism. On this view, the agent’s action has no meaning outside of the whole – the language-game or practice – which constitutes it. That a practice-dependent action is intelligible only within the context of a practice implies that without knowing what that practice is we would not know what action to observe. A practice, if you will, supplies the identity conditions for the actions that belong to its remit: the knowledge of these conditions enables an observer to pick out an object of observation (the action in question). More generally, whenever an action acquires its meaning as part of a larger language-game, we must first understand the meaning of the language-game, as a whole, or its defining rules, before we can consider the meaning of its parts, that is, the meaning of the actions falling within its domain: this is what internalism demands.

Once we subscribe to the holistic logic of practices, we cannot, on pain of inconsistency, disassemble a practice into discrete bits and pieces, any more than we can ask what the action ‘scoring a goal’ means by considering it apart from the game of football. If we started enquiring what the material bit of the action ‘scoring a goal’ is, and what the ideational bit is, and what the structural forces are which enable players to score goals, we would simply reveal that we had not understood the point of the game. For football is not an aggregate of ideas, materials and structures. It is a meaningful whole: a game constituted by rules and by participants who know how to follow the rules. In a like manner, it would be improper to decompose the action ‘negotiating an international treaty’ into ideational, material and structural bits.

From the holistic standpoint of internalism, it is misleading to present the international realm in atomistic terms, as consisting of discrete practices that can be piled together into a ‘constellation of practices’. Rather, the core practices we inhabit in the international realm have the character of macro-level social wholes – they are ‘global’ practices: notably, the practice of human rights and the practice of sovereign states. These count as ‘global’ practices not in the geographic sense, but in the normative sense of the term: no matter where on the globe we happen to live, we cannot exempt ourselves from their authority. ¹⁵ Insofar as we participate in the global public domain of moral and political life, we are subject to the authority of the language of rights and that of sovereign states.

Moreover, no categorical distinction divides what might be called ordinary practices, such as medical practices, local produce markets and legal practices, from the global social practices in international relations. By default, a practice is an action domain constituted by rules, ‘constitutive rules’, which specify performative conditions enabling participants to perform certain types of action (Rawls, 1955: 24; Oakeshott, 1975: 61, 79, 119–120,171). Constitutive rules define a type of action by stipulating that doing ‘x counts as y in a context C’ (Searle, 1969: 35). The rules of chess, for instance, define ‘check-mate’ as a type of action within the context of a game of chess. Knowing the constitutive rules of the relevant practice therefore is a condition for making sense of any practice-dependent action; it is not a mere desideratum. For example, someone not acquainted with the constitutive rules of diplomacy could not perform – or recognise that somebody else is performing – an action described as a demarche.

It follows that when we speak of practice competences, we work from the premise that most participants in a practice already know how to follow its defining rules. Practical competence, so construed, cannot simply be a matter of acquiring an individual skill in the absence of having the necessary knowledge of social rules. This is contra to what theorists such as Adler and Pouliot (2011a: 7, 2011b: 8) suggest. ¹⁶ By saying that competences are determined by antecedent rules, we imply that an agent is enabled and at the same time constrained to perform certain actions in virtue of injunctions (the rules in question) that have a ‘social’ or ‘public’ character. Within a practice, the activity of rule-following, as Wittgenstein and Winch pointed out previously, cannot be a solipsist business: it presupposes criteria of correct action that are publicly accessible to other participants in a practice and to those seeking to investigate it. ‘It is of course possible’, Winch writes, ‘… for an individual to adhere to a private rule of conduct. What Wittgenstein insists on, however, is … that it must be in principle possible for other people to grasp that rule and judge when it is being correctly followed’ (Winch, 1958: 32–33, original emphasis; see also 1958: 39). The agent who understands a practice from the internal point of view is not a Cartesian subject whose private domain of consciousness is sealed off from scrutiny by other observers. Internalism is an intersubjective point of view that is open to those who do not yet understand the practice but who are willing to learn how to understand or act within it.

At this point, we need to consider the notion of intersubjectivity more closely, for without further qualification it does not commit us to a particular standpoint (either external or internal), to a particular theory of meaning or to a particular conception of the units engaged in intersubjective communication. The brand of intersubjectivity that Winch and Wittgenstein recommend is tied to the standpoint of practice internalism. As such it must be distinguished from famous intersubjective positions associated with Weber’s (1978) doctrine of social action or Searle’s (1969, 1995) theory of institutional facts. For Weber, ‘action is “social” insofar as its subjective meaning takes into account the behaviour of others and is thereby oriented in its course’ (1978: 4). In Weber’s memorable example, when two motorists on a road collide inadvertently this is a natural fact. But when they try to avoid a collision by deliberately adjusting their behaviour, this is ‘social action’ attuned by intersubjectively shared meanings (Weber, 1978: 23). Notice that his account does not refer to following the rules of the road or to some social practice understood from the internal point of view. The focus of Weber’s analysis is on the individual agent:

Interpretative sociology considers the individual [Einzelindividuum] and his action as the basic unit, as its ‘atom’ … In this approach, the individual is also the upper limit and the sole carrier of meaningful conduct. (Weber, 1922: 415 quoted in Gerth and Mills, 1958: 55)

Action in the sense of subjectively understandable orientation of behaviour exists only as the behaviour of one or more individual human beings. (Weber, 1978: 13)

As a result, meaning is by default a subjective notion: it is a motive a subject has inside the head (Weber, 1978: 11, 18). Weber’s category of intersubjective meaning marks an important transition from inner thinking (motive) to outer acting (goal setting) and thereby allows the researcher to overcome the perils of psychologist- and empathy-based accounts of social action (1949 [1904]: 68ff; 1978: 5, 19). Despite this, Weber treats intersubjective meaning as a derivative notion – his argument that the individual formulates goals by ‘taking into account the behaviour of others’ appeals to subjective meaning. For Winch and Wittgenstein, in contrast, meaning ab initio is intersubjective, as the ‘community view’ that underwrites a social practice illustrated. From such a perspective, Weber’s standpoint which permits recourse to subjective meanings outside of a social world is an external one.

Searle’s position is curious not the least because it is an attempt to bridge the internal with the external points of view. Searle differentiates brute facts (e.g. James has brown hair) from institutional facts (e.g. James has five dollars) (Searle, 1969: 50–52, 1995: 27–29). Institutional facts exist by virtue of institutions such as money, the market and marriage – conventions constituted by rules. So far this might seem to be an internalist, practice-based account. However, for Searle, institutions are held together by intersubjectively shared beliefs. It is only because people in society S believe collectively in the institution of money that the individual act of Richard giving James a five dollar banknote counts as an institutional fact in S. This fact is external to the private beliefs of each individual. To ground institutional facts, Searle proposes a move from the ‘I think’ (subjective level) to the ‘we think’ (intersubjective level) of collective intentionality (1995: 23–26). But a residue of the Cartesian cogito is lurking behind the proposal since Searle’s unit of analysis is individual consciousness (as a source of private beliefs and intentionality). ¹⁷ Ultimately, Searle holds that human institutions (including language) can be explained by a scientific theory of consciousness rooted in physicalism (where consciousness is causally ‘realised’ as a macro-state of the brain) (1983: 265–266). ¹⁸ This endorsement of physicalism reveals that Searle’s approach is, at base, an externalist one.

The preceding remarks allow us to map out the demarcation line between internal and external practices. In our discussion, the line is demarcated relative to the baseline of an entire practice (‘internal’ means ‘internal to a practice’), and not relative to alternative baselines such as individual consciousness (Searle) or subjective meaning (Weber). Subsequently, the units that take part in an intersubjective agreement within a practice are not per se individuals or consciousnesses but participants. In light of these considerations, the significance of Winch’s earlier point about public rule-following can be appreciated – it is that in their public, intersubjective aspect of intelligibility practices present language analogues. The metaphor of ‘language-games’ discloses that the analogy is bi-directional. Not only is the speaking of a language a rule-governed activity (a practice or a game) but, correlatively, language, as a meaningful whole (comprising a lexicon plus rules for making well-formed expressions) provides a model for the notion of a practice.

For Winch, the limit of the external point of view is a direct consequence of the limit of the possibility of translation between different types of languages. In ‘Understanding a Primitive Society’ (Winch, 1972b [1964]), he argues that the practices of witchcraft and magic of the Azande tribes are not irrational, contrary to the contentions of proponents of modern science. ¹⁹ As he contends, what counts as rational belief and rational action is internal to the domain of a practice – the practice of Azande witchcraft and magic (Winch 1972b [1964]: 30–31, 34). In ‘Belief, Language and Relativism’ (Winch, 1987 [1976]), he considers two problems pertinent to translating one language into another. First, translation may be achieved relatively easily where different societies use different words for similar practices. It is not hard for an English speaker to translate the English word ‘bread’ into French as ‘le pain’ since in both societies bread making is presumably a common practice. Second, translation across different orders of language is more problematic. For example, the languages of science, of philosophy and of mathematics – are much more self-contained when compared to natural languages (e.g. French, German, English and Russian). To illustrate the difficulty, imagine that we have no prior knowledge of mathematics and that, having encountered a mathematics text for the first time, we attempt to translate its specialist terms into our everyday English. As Winch emphasises, in attempting to do this, what we would be seeking is not a new expression for an activity that we already understand but an understanding of a new activity (1987 [1976]: 198). In this case, we would need to establish what counts as mathematics to begin with, and doing this would require us to learn its grammar (in Wittgenstein’s sense): the internal system of meanings and rules, which, construed as a whole, constitutes the space of thinking and acting that we refer to as ‘doing mathematics’. ²⁰ But although grammar (in this sense) establishes what counts as proper representation, it is not itself a representation of external reality. ²¹

Winch concludes that the witchcraft and magic practices of the Azande form a ‘grammar’, and that it is within this grammar that we have to ask, and answer the question of what counts as rational belief, if we wish properly to understand the Azande’s worldview. To generalise, in order to make sense of actions located within a practice – or to understand action from an internal point of view, in our terminology – we should be able to first identify the background practice. The identity of a practice is contained in its grammar. Only its concrete grammar will supply us with the relevant descriptions of the actions that fit into the practice. To do something in a practice is to perform an action under some description of it. ²² Within the practice of witchcraft, we may find actions describable as ‘consulting an oracle’, ‘reading omens’, ‘finding witches’ and so on. Inside a common practice, a set of descriptions of actions must cohere with one another and be intelligible as a whole in the same way a set of idioms cohere within a common vernacular.

Winch’s internalist conclusions lead to a radical critique of social science as an aberrant, externalist practice. Science oversteps its legitimate boundaries when it seeks to present its own criteria (scientific rationality, universality, neutrality, etc.) as criteria that mirror reality, and on this basis, to legitimise such external criteria as a uniform language suitable for the analysis of different human practices. But science has no privileged access to the real. ‘For a form of the conception of reality must already be presupposed before we can make any sense of the expression “what science reveals to be the case”’ (Winch, 1972b [1964]: 27; see also 1958: 15). In more mundane terms, we may say that each practice constitutes its own domain of social reality: there is no such thing as reality tout court. If Winch is right in his assessment, as we think he is, then philosophers and social scientists err in supposing that a uniform procedure can be discovered – a set of rules of scientific investigation – which, when followed, would produce correct accounts of action across different domains of action. In this case, instead of relying on differentiated criteria that are domain-specific, social scientists recruit externalist, uniform criteria for action appraisal such as the formula ‘action equals behaviour plus meaning’. Here behaviour is the uniform component, factoring what the units under observation, even non-human ones, do in terms of physical movements (e.g. observing the behaviour of rats in reward/punishment experiments), and meaning is the add-on that accounts for the difference between units that can grasp meaning and units that cannot (e.g. comparing the responses of rats with human responses in reward/punishment models). This externalist formula is endorsed by Adler and Pouliot who argue that we can move from the plane of behaviour, to action (meaningful behaviour) and finally to practice (patterned action) (2011a: 5, 2011b: 6). But it is wrong to think of action as some sort of add-on to behaviour. As Wittgenstein quipped: ‘What is left over if I subtract the fact that my arm goes up from the fact that I raise my arm?’ (1953: I, § 621). My raising my arm exemplifies an action: it discloses the intentions, plans and decisions of an integrated actor, a choosing and thinking self and is a benchmark of an intelligent social world. ²³ Conversely, my arm going up exemplifies mere behaviour (since a robotic mechanism may be pulling my limbs upwards without any conscious intention on my part). This distinction between behaviour and action is not a pedantic joust. It points to the background texture of the world to which we belong, and by siding with, inter alia, Winch, Wittgenstein and Oakeshott in treating the concept of action as basic, we are choosing to understand the world from the standpoint of intelligent human conduct – from the internal point of view, not from a uniform, external point of view.

In bringing this discussion to a close, we may note that externalism is not coextensive with behaviouralism or the positivist methodology of science, even though they are perhaps its best known representatives. What we have said so far about the external point of view as a fundamental standpoint of making sense of what people do, which gravitates towards invariance, uniformity and commensurability, applies to game theory and rational choice theory, ²⁴ Althusser’s (2008) structural Marxism and the doctrine of ethical naturalism (Moore, 1903). Indeed, it applies to any approach to analysing action in which the language of observation is imputed from the outside: as when the actors observed are assumed to be acting on as-if conditions rather than on conditions that (if brought to their attention) would be intelligible to them. In sum, the externalist view cannot provide a proper or correct account of what we do, as participants in different practices, since it denies that there are different grammars of appropriateness that operate as orders of differentiation and meaning.

Conclusion

What are the key implications for the study of international relations and the role of the IR scholar as a social scientist that flow from our internalist analysis of social practices? The first thing to recognise is that we, as thinking and choosing human beings, are already competent participants in global practices: we already possess a great deal of understanding of the core standards constituting such practices. Of course not everybody has the same depth of understanding: mistakes, confusions abound, and there may even be deliberate foul play. From an internalist perspective, the social scientist can be seen as a participant in a practice whose task is to clarify, find coherence or detect incoherence, in the claims and counterclaims participants make within that practice. Winch’s internalism in effect treats the distinction between an analyst (scholar, scientist) and a practice participant as a matter of degree. The social scientist is not Plato’s philosopher who alone can bring the light of knowledge to the cave (Oakeshott, 1975: 27–31) but a participant in a practice who seeks better to understand what is already imperfectly understood. In an ethical sense, the scholar might assume the role of a social critic whose job is to expose hypocrisy and acts of injustice occurring within one’s own society (Walzer, 1987).

The major global practices in international relations, the practice of human rights and the practice of sovereign states, constitute us in a shared normative space. We cannot step outside their authority. We know for instance that a practice of sovereign states exists because its defining norms – peaceful settlement of disputes, non-intervention, observing limits in the conduct of war, keeping of treaties – are intelligible to us (Frost, 1996: 104–136). These norms are internal to this practice. We can understand the claims and criticisms made by other participants in the global practice of states, when they evaluate the actions of states (including our own state) in the international domain made by reference to these norms. To mention but a few of these: charges are made about unjustified intervention in the internal affairs of other states (Holzgrefe and Keohane, 2003; Nardin and Williams, 2006; Weiss, 2012), about failures to respond to gross human rights abuses taking place in specific areas (Alston and Goodman, 2012; Buchanan, 2010; Donnelly, 2013), about duties to build up underdeveloped states (Chandler, 2006; Paris and Sisk, 2009), about the rights and wrongs of acts of war (Falk, 2014; Rengger, 2013; Walzer, 2006). A rich and ongoing set of claims and counterclaims, descriptions and counter descriptions, are being made about what is happening within the global practice of states which can only be understood in terms of the internal point of view as we have described it. That these are not merely evaluations from some external, invariant point of view is evident from the way in which those criticised for their actions respond in terms taken from the same language in which the initial evaluations were made. The intelligibility of this common language to a group of agents reveals that they are participants bound by a common practice.

To reiterate, what participants in a shared practice accept are certain common conditions for action, which by setting standards of correct as opposed to incorrect action define the bounds of a practice. ²⁵ This acceptance has the character of normative recognition. It is reflected in Wittgenstein’s ‘agreement in form of life’ (1953: I, §241), in Oakeshott’s notion of citizens’ recognition of the rules of respublica (a system of law) (1975: esp. 149–151) and in Hart’s view that the rules defining a legal system must be accepted and internalised by its members rather than merely complied with on fear of external sanctions (1961: 55–56, 85–88, 197). But using the same language does not imply that interlocutors in a common practice necessarily agree on a set of goals or ideals about the good life. Moreover, because human agents participate in multiple practices at the same time, conflictual demands issuing from each practice are likely to provoke strain. At the global level, at the moment, there is a palpable tension between the two core practices – the practice of sovereign states and that of global rights (Beitz, 2009; Buchanan, 2010; Forsyth, 2012; Frost, 2002). This tension has become apparent in areas such as the infringement of human rights in the context of terrorist threats (Alston and Goodman, 2012: 383–488; Coady, 2004; Luban, 2002), in the debate over refugee rights (Barry and Goodin, 1992; Betts and Loescher, 2011; Cohen and Deng, 1998) and in the normative shift from the right of non-intervention enjoyed by states to a responsibility to protect doctrine which imposes a duty on states to protect citizens’ lives across borders (Bellamy, 2009; Evans, 2009; Pattison, 2010).

Finally, and to return to the heart of the matter, the internalist standpoint requires that anyone wishing to understand an action, any action, as a this rather than as a that, be it in chess, music, dance or international relations, has either to learn how to perform the action himself or herself, or to learn to recognise that it is being performed by somebody else within the practice concerned. To learn the moves, strategies and tactics of chess; the tunes and rhythms of Indian music; the movements and gestures of classical ballet; or the manoeuvres open to diplomats and others players in international relations, is to learn to make sense of and to perform actions, taking place inside the contexts of distinctive practices. There is no external, social scientific way of learning such practice-dependent performances. Despite all this, it is unlikely that the debate between practice internalists and practice externalists would come to a halt any time soon. In this discussion, we have supplied ample reasons to pause and reconsider the current trend among practice turn scholars in IR who treat the externalist route as the self-evident one. There is an alternative, internalist route leading to more profitable destinations.