Introduction

The inter-paradigm debate represents the theoretical core of international relations (IR), and its ability to confront post-positivist critiques has guaranteed a privileged epistemological status in the discipline. As a consequence of such theoretical triumph, neorealism and neoliberalism not only inform our understanding and perspectives about the international system but also shape how we approach international phenomena. Central concepts such as anarchy, agent rationality, war, cooperation, and complex interdependence, to name a few, have become an essential part of IR vocabulary.

The success of the neo-neo synthesis in generating explanations and predicting a variety of international phenomena has granted it a place of pride in the discipline, which in turn has come to suggest that these theories and the methods they entail are capable of producing quasi-universal claims about how international actors and the international system behave. Nevertheless, “universal methods and universal theories should be viewed with suspicion” (^{Cartwright 2007, 80}). Given that, in many situations, those phenomena that happen to be different play a key role in questioning our canonical interpretations of the world.

Perhaps due to its relative success in generating predictions about international phenomena, IR scholars have avoided the philosophical debate of neo-neo assumptions. Metatheoretical inquiries have been seen as a diversion by mainstream academia; however, some scholars have productively engaged with the philosophy of science and have made invaluable contributions to our understanding of what it means to explain international phenomena (^{Chernoff 2007, 2014}; ^{Lebow 2014, 2022}; ^{Kurki 2008}; ^{Kurki and Wight 2021}; ^{Suganami 1996}; ^{Wight 2006}). Based on this body of literature, this article explores the philosophical underpinnings of the law-like nature of neo-neo theories. We propose to confront Nancy Cartwright’s conception of nomological machines and capacities with neo-neo theories to unravel the mechanisms underlying their production of explanations. In so doing, we reframe the ontological terms of neo-neo theories’ hunt for explanations of international phenomena.

Drawing on Cartwright’s description of nomological machines and arguments about how they can alternatively operate to reveal capacities, we argue that the neo-neo synthesis can benefit from relaxing its rigid approach to central elements of the international arena-namely, anarchy, rational behaviour, and the role played by the structure of the international system. In this process, the strictness of the law-based theoretical discourse is loosened, favouring a more nuanced interpretation of international phenomena through the very same lenses of neorealism and neoliberalism.

The article is divided into three sections apart from the introduction and conclusions. In the first section, we present Cartwright’s arguments about nomological machines, central to our discussion of the neo-neo synthesis. Next, we briefly summarise the fundamental epistemological tenets of neorealism and neoliberalism, situating their theoretical logics under the framework of nomological machines. In the third section, we develop our understanding of how to fix neo-neo nomological machines and harness the full explanatory power of both theories.

1. Nomological Machines and Capacities

The various artifices IR scholars use to develop theories locate them in a long-standing tradition of positivist and neopositivist philosophical approaches to theory-building (^{Wight 2013}). In particular, Karl Popper’s philosophical reinterpretation of positivism reframed the social sciences in ways that a critical rationalism based on a hypothetico-deductive model of explanation has become pervasive as a metric for defining what (good) science is (^{Cartwright 2007, 179} ; ^{Chernoff 2014}; ^{Dowding 2016}; ^{Hawkesworth 2015}). To be sure, the hypothetico-deductive model resembles the archetypical explanatory model of natural sciences, where theories provide the conceptual framework from which scientists derive hypotheses to be tested against data extracted from the real, observable world, ultimately leading to the formulation or confirmation of overarching laws of nature. According to Hawkesworth:

Most of the research strategies developed within social science in the twentieth century draw upon either positivist or Popperian conceptions of the scientific method. The legacy of positivism is apparent in behavioralist definitions of the filed that emphasize data collection, hypothesis formulation and testing, and other formal aspects of systematic empirical enterprise, as well as in approaches that stress scientific, inductive methods, statistical models, and quantitative research designs. (...) It emerges in claims that social science must be modeled upon the methods of the natural sciences, for those alone are capable of generating valid knowledge. (^{Hawkesworth 2015, 33-34})

The quest for a social-cum-natural science of IR has produced a theoretical body that operates under the tenets of a nomological machine, a powerful device capable of matching the observable world to the laws of nature, as described by ^{Nancy Cartwright (1999)}. Cartwright departs from a Humean post-logical-positivist approach to define a law of nature as “a necessary regular association between properties antecedently regarded as OK” (Cartwright 1999, 49). Laws can be either deterministic-the association derives ipsis litteris from the properties entailed in the law-or probabilistic-associations between properties are governed by chance (^{Dowding 2016}). Every law describes real-world phenomena via properties, which can be measurable, tangible, or even abstract, and it is due to the operation of these properties through certain mechanisms inscribed into the law that generates the outcomes we observe in reality. Ultimately, a law holds for all cases that fall into the dicta of its properties.

Precisely here lies Cartwright’s understanding of laws vis-à-vis what she calls nomological machines. For a law to work, its properties must operate as it ascertains. In a world pervaded by natural and social phenomena, we observe regularities that we aim to explain and develop concepts that ascribe specific features to each element at work in generating those regularities. Examples of observed regularities and related concepts are planetary orbits and mechanical celestial forces; water flowing in a pipe, velocity, and laminar flow; collective decision-making leading to suboptimal outcomes; and majority-based electoral systems producing bi-party systems, to name a few. For each concept-regularity pair, rules of combination have to be assigned to each element for a law to be written down. When those elements are present, we can expect the law to say something about a specific phenomenon based on how its properties are inscribed into the concepts, and the regularities observed in the real world combine to generate that phenomenon. A nomological machine, thus, has been put into operation.

Yet, for that to be possible in a chaotic world of countless phenomena, laws are cemented on certain ceteris paribus clauses that confine the observed phenomena into the realm of the properties of laws. Cartwright calls it shielding, to the extent that nomological machines must shield themselves from exogenous phenomena that might interfere in their operation (similarly to how an interfering signal disturbs your radio programme). According to her, laws hold “only ceteris paribus-they hold only relative to the successful repeated operation of a nomological machine,” which, in turn, “is a fixed (enough) arrangement of components, or factors, with stable (enough) capacities that in the right sort of stable (enough) environment will, with repeated operation, give rise to the kind of regular behaviour that we represent in our scientific laws” (^{Cartwright 1999, 50}). This stable environment is ensured by shielding what we want to explain from any external interference not incorporated into the original model.

The following examples drawn from the natural and social sciences illustrate the operation of a nomological machine. In problems of convective heat transfer-when a solid medium exchanges heat with a fluid medium in its surroundings-physicists and engineers alike resort to Newton’s law of cooling, which is a basic formulation of the rate of heat transfer Q, which equals the product of heat transfer coefficient h, surface area A, and the difference between the temperature on the object’s surface T and the temperature of the environment T _∞ , i.e., Q = hA(T - T _∞ ) . Examining this equation allows us to see how a nomological machine is built on the mentioned description: it aims to explain a regularity, namely heat exchange between a solid object and a fluid in the environment. It establishes concepts that allow for explaining this regularity, such as the rate of heat transfer, heat transfer coefficient, and, more importantly, temperature. In defining these elements, it also sets the rules of combination for each of them-the equation itself. Nevertheless, this does not suffice to fully describe this nomological machine since shielding against the effects of the environment is still needed. A common effect would be neglecting radiation, for instance, or that the difference between the temperature of the object and the environment is relatively small. Shielding also occurs at the conceptual level. The heat transfer coefficient, for example, is estimated based on how the boundary layer of the fluid behaves relative to the surface (whether the flow is laminar or turbulent, for example) (^{Bergman et al. 2011}). In both cases, the nomological machine only begins to operate once the concepts are fully determined and confined to ceteris paribus clauses and the shielding from the environment is properly described.

In a similar vein, game-theoretical models are themselves nomological machines. Typical settings in game theory depart from the following elements: agents, assumptions about the agents’ behaviour, environmental sets of incentives and constraints, and, frequently, equilibria. Rationality, in whatever manner defined, is a classic concept in game theory that allows for modelling social phenomena where collective decisions need to be made. Agents are assumed to behave rationally and can assess information about the environment, ponder the sets of incentives and constraints imposed by other agents and the institutional architecture that informs the rules of aggregating individual decisions, and devise the optimal strategy to generate the best outcome. Once agents make their decisions, a scenario unfolds. Game theorists are interested in finding the equilibrium point resulting from the strategic interaction of individual agents (^{Gintis 2009}). Once again, concepts, regularities, and the rules of combination underlie the design of a game-theoretical nomological machine. Concepts such as agents, rationality, institutions, incentives, and constraints are fundamental to game theory, regularities stem from the various decision-making scenarios observed in the real world, and the rules of combination are frequently specified in terms of utility functions in game-theoretical models. Shielding is also in play at all levels-rationality is commonly assumed to be perfect or nearly perfect, uncertainty is usually domesticated to fit into probabilistic models, and the rules of the game, as well as the environmental incentives and constraints, are equally understood by agents. As we can easily see, these elements are also shielded via ceteris paribus clauses.

Ceteris paribus conditions instantiate the circumstances under which a nomological machine operates, allowing for the emergence of regularities of the kind we are looking for in the first place. This becomes evident when we scrutinise how scientists formulate equations to explain regularities. In the process, there comes the moment when they must specify the conditions where the equation can be applied (as in the mentioned case of Newton’s law of cooling equation). In so doing, the set of ceteris paribus conditions plays a special role in designing the explanatory mechanism entailed in the nomological machine: “They describe the structure of the machine that makes the laws true” (^{Cartwright 1999, 148}). Therefore, for a nomological machine to operate repeatedly and generate explanations and predictions across time and cases, it has to rely on the ceteris paribus structure that shields it from interference from the outside world that might bring instability and uncertainty to its operation. Indeed, this ceteris paribus claustrophobia is necessary for the proper functioning of the nomological machine; otherwise, the machine would break down should the shield fail, and other factors interfere with its internal mechanisms.

This approach to the scientific endeavour, namely to tailoring explanations, is the central matter in Cartwright’s critique of how laws are defined and valued in science. If laws depend on a nomological machine to generate them, and if nomological machines restrict the scope of explanation by establishing a single set of behaviours to be expected once the machine starts to operate, our understanding of the myriad factors capable of affecting real-world phenomena becomes profoundly limited, not to mention that our explanatory horizons become restricted to what the nomological machine says in its user guide. Instead of looking for potential behaviours a certain combination of factors might generate, we limit our attention to the components of the nomological machine and the ceteris paribus conditions that allow a specific set of regularities to emerge.

Against the law-like building of nomological machines, Cartwright argues that we should instead focus on capacities, which are ontologically more fundamental than laws themselves (^{Cartwright 1999}; ^{Hédoin 2014}). In her understanding-which, accordingly, draws from a long tradition of Aristotelian thought-capacities allow us to approach phenomena with an open-ended perspective, which acknowledges that social, economic, political, and natural phenomena might display a variety of manifestations. Capacities are to be understood as tendencies, abilities, and propensities to generate various kinds of behaviour that cannot be reduced to a single and unique outcome (^{Cartwright 1999, 59, 64}). She explains in one of her examples that we say in economics that taxes affect prices, but what happens to prices depends on the various possible ways of building economics (^{Cartwright 1999, 64}). Furthermore, she states:

The notion of capacity has three elements: (1) potentiality (capacities describe what a factor can do in the abstract, not what actually happens in full empirical reality); (2) causality (capacity claims are not claims about coassociation but about what results a factor can produce); and (3) stability (the ability to produce the effect in question must persist across some envisaged variation of circumstance). (^{Cartwright 1998, 45; emphasis in the original})

Capacities themselves do not presume any specific and predetermined behaviour; as they are scarcely observable in the real world, the nomological machine is needed to specify what exact tendencies we aim to observe in a given phenomenon. Think, for example, about the claim that a force can move a body, which is the underlying principle of Newton’s second law. By claiming the capacity of a force, we do not establish a predetermined motion, but rather the propensity to generate that motion. To provide a specific kind of motion, we need extra suppositions (assumptions, shielding clauses) that will set the nomological machine running.

But how do we identify capacities? According to Cartwright, by assessing the structures and qualities of natural and social phenomena-primarily via experimentation-we can identify the underlying capacities that lead to the manifestation of those phenomena, especially regarding the observable regularities that scientists attempt to explain. As Cartwright states: “We try to find out what capacities can be harnessed to produce predictable behaviours” (^{Cartwright 1999, 138}). As capacities are ontologically prior to laws, their discovery allows us to set the conditions and domains under which a certain law operates, acknowledging that the law itself is one of the possibilities entailed by the capacities-the possibilities prescribed by the ceteris paribus conditions stated in the law.

Indeed, harnessing capacities presumes the operation of nomological machines if one is willing to observe regularities in the real world. Once capacities are identified, the nomological machine runs according to the behaviours those capacities may elicit, this time in predictable ways. This poses a conundrum: If we acknowledge that the world is pervaded by nonlinear, unstable, and/or uncertain phenomena, how can we isolate capacities in a nomological machine to ascertain that they are responsible for generating the observable regularity of our interest? The answer lies in the fact that nomological machines are built upon principles that we as scientists dictate (^{Cartwright 1998}), and they are fundamental to generating the regularities that we aim to explain in the first place. In other words, our confidence in the design of the machine is what allows us to associate capacities with regularities (^{Cartwright 1999, 89}).

Up to this point, we have focused on Cartwright’s general ideas and arguments, eschewing a discussion of their impact on social sciences. In her interactions with other disciplines, she has primarily devoted attention to economics and how economists devise their socioeconomic nomological machines (^{Cartwright 1998; 1999}). The main challenges for economists reside in the fact that the discipline lacks “widely accepted general principles,” which requires economists to depart from thought experiments and assumptions that can be mathematically manipulated in a tractable model, and that they fail to isolate causal factors to explain how each works separately to generate a certain observable outcome (^{Hédoin 2014, 431-433}). As Hédoin puts this more directly: “If economics is about discovering capacities and if economic models are socioeconomic machines that fail to isolate capacities because of overconstraint, then economics fails to help us to learn how capacities work in the empirical world” (^{Hédoin 2014, 434; emphasis in the original}).

To a large extent, IR faces similar dilemmas: scholars in IR tailor explanations about regularities in the international system; the tenets of neorealism and neoliberalism are fashioned from a law-like, model-based approach to international phenomena, which ultimately leads to explaining observable regularities; these tenets are components of nomological machines that cannot apprehend the various factors we observe in the real world, which leads to their confining into claustrophobic ceteris paribus clauses that characterise the law-like machine. Once the machine is running and generates the regularities of our theories of interest, we claim that we have identified the explanatory link between the fundamental capacity entailed in the original design of the machine and the regularity itself. This is precisely the same plot followed by economists that ultimately confronts them-and IR scholars alike-with the problem of identifying capacities in the real world.

2. Neo-Neo Synthesis: Explaining or Shaping Reality?

To integrate Cartwright’s nomological machines with the theoretical enterprises of the neorealist-neoliberal debate in IR, it is worth discussing the foundational narratives of the discipline and the locus of the neo-neo synthesis in the paradigmatic development of IR theories.¹ The IR discipline has been narrated through great debates that unfolded throughout the twentieth century (^{Burchill and Linklater 2001}; ^{Lapid 1989}; ^{Smith 2021}; ^{Wæver 2021}). According to the traditional account of the history of the discipline, the debates fostered ontological, epistemological, and methodological pluralism within IR, distinctively characterising the discipline itself. As the story goes, at the early stages of the discipline, theoretical discussions opposed classical realism and idealism in what has been coined as the first debate.² It was then reedited in the 1970s and 1980s through a dialogue between neorealism and neoliberalism-the so-called neo-neo debate or neo-neo synthesis-also known as the third debate. More recently, from the 1980s on, post-positivists have profoundly criticised the ontological, epistemological, and methodological assumptions of the previous debates.

The second debate, which took place during the behavioural revolution in social sciences, systematically changed the epistemological and methodological landscape of IR. For one, the principles of behaviourism launched an understanding of social science that resembles those of the natural sciences, namely scientific rigour based on quantitative methods, observation, and measurement of regularities (particularly, behaviour), generalisation of findings, and conclusions (^{Farr 1995}; ^{Kirkpatrick 1962}). Neorealism and neoliberalism, hence, have followed these principles, establishing-in their terms-the scientific character of the IR discipline. In this sense, they can be seen not as rival theories but as theoretical approaches that share numerous similarities, especially ontological and methodological ones (^{Kaplan 1969}; ^{Kurki and Wight 2021}; ^{Lamy 2008}; ^{Nicholson 1992}). Thus, the neo-neo synthesis is often identified as an inter-paradigmatic dialogue, not as a clash of opposing paradigms.

The inter-paradigmatic debate differed in terms of actors, specific concepts, dynamics, dependent variables, and disciplinary boundaries but had significant internal and structural similarities (^{Banks 1985}). In general, the focus of this debate was on the criticism of the realist excess of state-centredness. Neorealism placed the state and force (military capability) at the centre of its theoretical analysis, whereas neoliberalism, though sharing these assumptions, sought to expand the types of actors in the international arena-incorporating non-state ones-and other expressions of state capability beyond the military sphere. Furthermore, ^{Holsti (1985)} argues that the similarities of these theories reinforce the inter-paradigmatic character of the neo-neo debate. Nevertheless, the inter-paradigmatic narrative, although compelling, is disputed in the discipline. For instance, ^{Smith (1996)} states that the conception of an inter-paradigmatic debate is misleading since “the proponents of each paradigm literally do not see the same world” (19), meaning that there is no competition between neorealism and neoliberalism. Wæver is also sceptical about the existence of an inter-paradigmatic debate:

Did it exist, the Inter-Paradigm Debate? Partly no, it was not actually an intense [...] debate occupying the minds of International Relationists, but an artificially constructed “debate,” mainly invented for specific presentational purposes, teaching and self-reflection of the discipline. (^{Wæver 1996, 161})

Despite these disputes about the history of the discipline, scholars understand that both theories emerged from the same ontological and methodological bases and are cemented on the same rationalist principles that characterised social sciences at that time. To a large extent, they resort to an ideal of science that is profoundly marked by the pursuit of explanations of observable regularities; in this process, the hypothetico-deductive model is extensively mobilised. We shall explore this next.

3. Fixing IR’s Nomological Machines: Capacities and Explanation

Neo-neo nomological machines are plenty and guide our understanding of international phenomena, namely what falls into the categories of conflict and cooperation. Taken in isolation, they provide us with rigid pictures of the international system, where specific regularities are worth observation and theorisation. This stems from neo-neo theories’ attempts to explain real-world phenomena not only by the force of their individual theoretical tenets but also by excluding each other’s interpretations and alternative views of the international system.

Indeed, any theory has to rely on a set of assumptions and concepts that, once combined, yield the theoretical discourse that examines and ultimately explains the world. Nevertheless, the specific law-like nature of neo-neo theories in IR, instead of enhancing their ability to tailor explanations about the international phenomena of their interest, creates a theoretical claustrophobia that excludes a significant number of factors that are useful to comprehend the intricate dynamics of the international system. More importantly, by eschewing a deeper ontological investigation of the basis of their explanations, neo-neo theories fall into the trap of reproducing a theoretical discourse limited to predicting what is going to happen by the operation of a nomological machine. In other words, they fail to explain why the phenomena of interest take place since they are confined to the law-like workings of the nomological machine. Relaxing the law-like element is paramount to illuminating these dynamics, and a capacities-oriented approach, as suggested by Cartwright, might be helpful in this process.

As Cartwright states, “Knowledge of capacities is more basic in that it is both more embracing and more widely useful than knowledge of regularities” (^{Cartwright 1999, 77}). Capacities are more embracing because, instead of talking about what things do in law-like terms, they tell what is in the nature of things to do given the exercise of their capacities under certain circumstances. For example, Newton’s law of gravity states that all matter, by possessing mass, attracts each other. This law holds for most objects and materials we encounter in our daily routine, which are subjected to the effects of our planet’s gravitational pull. But for subatomic particles, whose masses are negligible, gravity is not as important a factor as electric forces. Coulomb’s law comes into play, and gravity by itself cannot tell whether two particles will attract each other because, depending on their electric charges, we may observe repulsion between them. This does not mean that gravity plays no role. The particles do keep their capacity of gravitational attraction by virtue of possessing mass, but in comparison to the electric forces in play, the effects of gravity are negligible. In this example, the regular behaviour expected due to gravitational forces does not yield, for other forces assume prominence in this specific setting. Hence, forces are prior to the behaviour-they are manifestations of the capacities possessed by matter.

The key here is that capacities are primary, whereas regularities are derivative. As the example of the law of gravity demonstrates, an object possesses the capacity of attraction, but how this capacity will manifest itself depends on the analytical setting in the first place. The regular behaviour, therefore, is not immediately derived from the law because other capacities may elicit different types of behaviour and, therefore hinder the gravitational effects. Similarly, the underlying capacities of the international system generate various types of behaviour across countries; however, they do not have to be predetermined in a law-like fashion for the same reason that the analytical settings might influence how these capacities are exercised.

Anarchy is perhaps the most emblematic example of how a capacities-oriented approach to international phenomena is more productive than the original law-like perspective of neo-neo theories. Anarchy is a pervasive characteristic of the international system, considered by both theories in their accounts of state behaviour. In their readings, though, what results from an anarchic system is thought to be axiomatic-what we observe, according to each theory, is a set of regularities that fall into one of the categories of war or cooperation. Nevertheless, the traditional outcomes of war and cooperation are not directly derived from the components of anarchy and actor behaviour, but from a normative decision entailed in the theoretical discourse. If anything, these components have the capacity to produce either war or cooperation, but to do so, a specific nomological machine has to be assembled in the first place. There is no philosophical desideratum in anarchy or any other component by itself that leads to the immediate conclusions postulated by each theory individually. However, if anarchy were to be understood rather as a capacity in Cartwright’s terms,⁴ instead of rendering one specific type of outcome, it could be seen as capable of generating a variety of different behaviours.

It is worth exploring Waltz’s influential understanding of anarchy, given that it illustrates the usefulness of the capacities-oriented approach we have been discussing here. According to him, international anarchy is invisible. He describes anarchy not by its ontological elements but rather by the observable implications of its existence-state behaviour, the logic by which states function, the distribution of relative power (measured primarily by military capabilities), and the quest for survival, to name a few (^{Waltz 1979}). How the structure of the international system is organised necessitates anarchy; nevertheless, since anarchy is not a concrete, tangible element, Waltz can only talk about it in terms of what it elicits. As a consequence, anarchy constitutes a capacity of the international system that expresses itself in the mentioned various observable ways. We read anarchy by the force of its manifestations, which are far more diverse than a single set of behaviours pre-defined by a law of the international system operated via a nomological machine.

This is not to say that nomological machines must be jettisoned altogether. Instead, neo-neo nomological machines must be designed to unravel the capacities of the international system. Only by setting our nomological machines under the right and specific conditions entailed in theory can we reveal the underlying capacities of the international system and, then, the regular behaviour they may elicit. To be sure, “we get no regularities without a nomological machine to generate them,” and we can only be confident in our machine because of “our recognition that it is just the right kind of design to elicit the nature of the interaction in a systematic way” (^{Cartwright 1999, 89; emphasis added}).

But how can we achieve that with our current neo-neo arsenal of nomological machines? Shifting our attention to the ontology of capacities is the first step in the process of fixing neo-neo machines; however, more practically, we need to investigate at the ontological level what we mean by anarchy, rational behaviour, and the structural settings under which agents interact in the international system. Capacities lie beneath the international phenomena of our interest, and to unravel their operation to generate the outcomes we observe, we need first to unbind them from the strict conceptualization entailed in neo-neo theories. Take, for instance, rational behaviour. Typically, agent rationality is framed in terms of utility-maximising behaviour based on transitive preferences. Most game-theoretical applications in IR depart from the core assumptions entailed in rational choice theory (^{Gintis 2009}; ^{Hindmoor and Taylor 2015}), which is fundamental to solving for the game setting. In many cases, however, the predicted scenarios of a game’s solution do not yield, such as the tragedy of the commons and the inability of states to reach international agreements over various issues. If we understand rationality as a capacity instead, we may be able to fix our neo-neo machines that operate under its umbrella to encompass scenarios where alternative outcomes other than the equilibrium one are obtained (^{Lenine 2018}). In other words, by interpreting rationality as a capacity that operates in tandem with other capacities of the international system, we can extrapolate from the rigid equilibrium solutions imposed by solving archetypical game-theoretical settings.

The capacities-oriented approach relaxes the law-like explanations produced by neo-neo theories by freeing international phenomena from the shielding clauses entailed in the theoretical discourse. Instead of shaping the world to fit into neo-neo nomological machines and operate under the framework of laws of the international system, capacities allow for a variety of different but theoretically cemented behaviours to manifest themselves. It is by saying, for example, that anarchy might generate a cold war between superpowers in a bipolar setting or cooperation between Third World countries oblivious to the bipolar dispute that we can provide meaningful explanations about international phenomena-especially those that simultaneously occur at a given moment. The usual strategy of ignoring what neo-neo theories define as marginal phenomena is characteristic of the law-like nature of their nomological machines, and this is precisely what capacities aim to correct, for they allow both types of behaviour to manifest themselves and even coexist.

At the heart of this discussion is the issue of how we produce explanations about the international system and what we are explaining. If IR is to be a science capable of saying something meaningful about the world, it cannot rely on law-like nomological machines that confine phenomena to some specific, even ideal, conditions to explain the underlying patterns and/or forces of those phenomena. Instead, we should seek to understand the more fundamental elements that constitute the international system and are capable of producing a variety of different behaviours, ultimately as observable and predictable regularities. The predictions generated by neo-neo machines, hence, derive their explanatory power not from an idealised accuracy that depends on a set of rigid ceteris paribus clauses and conditions to emerge, but rather from their ability to represent the interaction of different capacities of the international system that operate to produce a certain effect. Therefore, by resetting our neo-neo machines to capacities instead of laws and by devoting serious efforts to investigating the ontological basis of the former, we are able to appreciate the various phenomena we see in the international arena without predetermining a single course of outcomes, especially in the face of real-world complexity.

Conclusions

The inter-paradigm debate has consistently informed how we see, analyse, and interpret international relations and, to a large extent, has become the landmark of IR as a discipline. Despite its theoretical and empirical achievements, various international phenomena do not fall into the predictions produced either by neorealism or neoliberalism-the end of the Cold War being the most widely cited example of such failure. While some have argued that neo-neo theories are deeply flawed at the level of their assumptions, others have diminished the importance of their empirical failures in the face of their success in explaining the mechanisms operating in the international system. However, as a result, the necessary philosophical debate about how neo-neo theories produce explanations has been avoided.

Neorealism and neoliberalism approach international phenomena via specially designed devices that operate under particular conditions to generate specific, observable outcomes. These devices are built upon law-like tenets entailed in the neo-neo theoretical discourse, and they are, by themselves, shielded from external interference of factors that might produce alternative outcomes than those predicted by these theories. Operating by repetition, they yield the well-known predictions of the neo-neo synthesis.

In other words, neorealism and neoliberalism design nomological machines to generate law-like explanations about international phenomena, where a law of the international system explains regularities observed by analysts in the real world. However compelling this approach might sound, it dangerously confines our explanations to shielding conditions entailed in neo-neo theoretical discourses. What they see is a world where behaviours are predetermined and shielded from external interference, whose complexity is subsumed to a set of ceteris paribus statements that produce the outcomes expected by the theory.

Neo-neo explanations, far from being a definitive representation of the world as it is, predicate what the world should be according to the theoretical discourse of the neo-neo synthesis. Consequently, neo-neo nomological machines are assembled to produce the world as these theories see it, connecting the outcomes with the theoretical discourse in a law-like fashion. This means that the predictions of neorealism and neoliberalism are always obtained, given that their nomological machines operate to produce the same predetermined behaviours.

Nevertheless, the explanation can only be meaningful if, and only if, it unravels the various capacities of the international system instead of confining international phenomena to a claustrophobic set of ceteris paribus conditions. Throughout the article, we have advanced the argument that resetting neo-neo nomological machines to unravel capacities allows alternative behaviours to emerge from the same constitutive elements of the international system. By redesigning neo-neo nomological machines to a capacities-oriented framework, which allows for the emergence of a variety of behaviours depending on how the capacities of the system interact, more meaningful explanations can be generated.