What is a deliberative system? A tale of two ontologies

Systems, boundaries, and evaluations

The systemic turn in deliberative democracy could be described as a set of theories with internal debates and disagreements. Equally, however, it is a self-conscious movement led by a number of prominent deliberative democrats, many of whom coauthored a manifesto expressing their shared commitments (Mansbridge et al., 2012). This manifesto defines a system as “a set of distinguishable, differentiated, but to some degree interdependent parts, often with distributed functions and a division of labor, connected in such as a way as to form a complex whole” (Mansbridge et al., 2012: 4). It then defines a deliberative system as a set of institutions, processes, and sites that “encompasses a talk-based approach to political conflict and problem-solving – through arguing, demonstrating, expressing, and persuading” (Mansbridge et al., 2012: 4–5). These definitions motivate the authors’ proposal to go “beyond the study of individual institutions and processes to examine their interaction in the system as a whole” (Mansbridge et al., 2012: 2). These definitions also motivate their argument that “the system should be judged as a whole in addition to the parts being judged independently” (Mansbridge et al., 2012: 5). The systemic turn thereby shifts our focus from the question of whether a particular site lives up to our ideals to questions about whether the overall system is sufficiently deliberative and democratic (Parkinson, 2018).

The systemic turn has tremendous potential. There are, however, important questions about what it means to understand and evaluate a series of practices as a deliberative system. One question is primarily empirical: it asks, how should we delineate the boundaries of a deliberative system, and how do we justify that delineation? Another question is primarily normative: it asks, how should we evaluate a deliberative system as a whole, and on what grounds?

A problem of boundaries arises once we move beyond particular sites and forums to look at a plurality of sites. We then have to decide how we are to sort out these disparate sites such that some belong as part of a deliberative system while others do not. Deliberative systems scholars need to provide criteria to delineate the system’s boundaries. Without such criteria, critics might reasonably complain that deliberative systems scholars manipulate their results by fudging these boundaries. Furthermore, we surely have to justify these criteria by reference to our concept of a deliberative system. We need to explain what features of the deliberative system make it such that the boundaries are drawn one way rather than another.

The manifesto provides a deceptively simple response to the problem of boundaries. It suggests that the boundaries of a deliberative system “define a decision-making arena that is at least loosely democratic,” and that this arena can in turn be defined “either institutionally or by reference to a particular issue” (Mansbridge et al., 2012: 7–8). This idea of a “decision-making arena” allows deliberative systems scholars to point towards a network of institutions or all the sites involved in a particular issue. They can thereby postulate systems that cut across sub-national, national, or supra-national divisions. Indeed, if there were a well-defined decision-making arena, the problem of boundaries might seem to disappear. We could simply identify the boundaries of a deliberative system with the boundaries of the relevant decision-making arena.

Unfortunately, the manifesto’s response only pushes the problem of boundaries back a step. The question now becomes, what counts as part of a “decision-making arena,” and how are we going to identify an arena? It is important to recognize that there are several different ways in which we might draw these boundaries. We could draw them empirically by reference to all and only those practices that pertain to making a particular decision. Alternatively, we could draw them functionally by including all and only those practices stipulated as serving the deliberative goals of the system. Finally, we could draw them normatively to include all and only those practices that at least loosely meet deliberative democratic ideals (Smith, 2016).

Should deliberative democrats worry about this ambiguity in their concept of a system? Dryzek (2016: 210–211) offers an interesting two-part argument suggesting they need not. He argues, first, that we could accept that scholars necessarily make an interpretive judgment whenever they pick out particular institutional arrangements, treat them as a deliberative system, and evaluate their actual performance in light of certain standards. He then argues, secondly, that it does not matter whether sites appear to be inside or outside of the system as long as we recognize their effects on each other. Although these arguments have much to recommend them, they pull us into two opposite directions at least if we are trying to understand what a deliberative system is. Dryzek’s second argument suggests that there are causal and functional relationships between different sites, that these relationships compose the system, and that these relationships exist independent of our interpretations. In contrast, his first argument suggests that researchers create the system through their interpretive judgments; they define the system to include some links but not others in accord with their purposes or beliefs. We thus return to the question of whether the boundaries of the system are a result of given causal and functional links or the purposes and beliefs of particular researchers.

Let us turn now to the problem of evaluation. This problem obviously overlaps with that of the boundaries of the system; it is, after all, difficult to evaluate a system if you cannot say what does and does not belong to it. More generally, the problem of evaluation is that of how and why to evaluate a deliberative system as a whole. The manifesto does not really address this question. It simply claims that we can evaluate the deliberative system as a whole through its epistemic, democratic, and ethical functions (Mansbridge et al., 2012: 10–12). It does not demonstrate how and why we can arrive at a judgment of these functions posited as the qualities of the system as a whole (Dryzek, 2017: 621).

One aspect of the problem of evaluation concerns the difficulty of evaluating a system rather than a single case. It is one thing to evaluate the performance of a single deliberative forum and quite another to evaluate a system composed of multiple deliberative and non-deliberative sites. Perhaps we can more or less decide if a single forum possesses certain deliberative qualities by looking at it. Perhaps, for example, we might decide that a forum is inclusive if it looks like it has procedures to ensure it is accessible to people from diverse backgrounds. If we want to evaluate a system as a whole, however, we might not even know where to look. What features of a deliberative system should we examine to see if it is inclusive? One proposed solution is to use specific empirical phenomena as heuristics by which to assess the performance of the system (Dryzek, 2017; Owen and Smith, 2015). For instance, if we assume that inclusive systems produce a civic culture in which people respect diverse opinions, we might use the civic culture as a heuristic to evaluate the inclusivity of the system as a whole. Another proposed solution is to look at how participants interpret one another and even the system (Bächtiger and Parkinson, 2019: 117–122; Boswell et al., 2016; Ercan et al., 2017: 201–203; Neblo and White, 2018). In this case, we would look to see whether participants believe they are included in the system, whether they think others should be included, and so on. Unfortunately, the existing literature does not explain why any of these heuristics would provide a sufficient basis for evaluating a deliberative system. More importantly, these two proposals again pull us in opposite directions when it comes to trying to say what a system is. The first proposal suggests the system consists of causal or functional links. The second proposal suggests it consists of the possibly diverse and even conflicting interpretations of participants.

Another aspect of the problem of evaluation concerns the relationship between part and whole. How should the evaluation of different parts of the system on their own relate to the evaluation of the system as a whole? Bächtiger and Parkinson (2019: 105–110) highlight two approaches to this question. The additive approach combines the evaluations of different parts to form a systemic level judgment. In other words, the more instances of deliberation the system has, the better is the system as a whole. However, Bächtiger and Parkinson worry that an additive approach undermines one of the key ideas of the systemic turn: it ignores the idea that non-deliberative practices could contribute positively to the deliberative system as a whole. The summative approach addresses this worry by locating the deliberative properties of a system not in its “constituent parts” but in the “configuration” of these parts (Bächtiger and Parkinson, 2019: 107), just as “no part of a car is itself a car” (Bächtiger and Parkinson, 2019: 107). However, critics worry that a summative approach has clearly counter-intuitive consequences: “within this conception of a deliberative system, no actual deliberation between citizens need take place as long as the process embodied in the sequenced institutions is functionally equivalent to a deliberative democratic exchange” (Owen and Smith, 2015: 219).

Bächtiger and Parkinson (2019: 110) try to find a happy balance between these two approaches. In brief, they appeal mainly to the summative approach while using the additive approach to require that the system meet a threshold of actual deliberation. Unfortunately, the existing literature, once again, does not explain why these approaches, or a balance between them, would provide a sufficient basis for evaluating a deliberative system. Surely if systems theorists stick firmly to their systemic commitments, they have no reason to care about the counter-intuitive possibility raised by their critics. Alternatively, if they do worry about that possibility, if they think there are good reasons to require actual deliberation to occur, surely they are doubting their own systemic commitments. Finally, if we face some sort of trade-off between, say, summative gain and additive loss, we lack criteria by which to decide how to proceed. Bächtiger and Parkinson (2019: 110) suggest that we need only to find the correct balance between these two approaches. However, this suggestion does not explain why a balance provides the right basis for evaluating systems, and it certainly does not resolve the apparently incompatible assumptions of the two approaches. Here the intuitions embedded in the two approaches point to completely different understandings of what a deliberative system is. The summative approach privileges the functional goals of the whole; it makes the parts deliberative in so far as they promote these goals, just as a motor counts as part of a car because it serves the functional goal of getting that car running. In contrast, the additive approach privileges the extent to which people see themselves and others as deliberating within the various parts of the system. The former ignores human interpretations. The latter makes them central.

We have argued that deliberative systems theory faces problems of boundaries and evaluation. We have also suggested that the existing, admittedly unsystematic, responses to these questions pull us in opposite directions. There is a lot of ontological messiness here because there is a confusion about what kind of system a deliberative system is. Sometimes deliberative systems seem to be about causal or functional links. At other times, they seem to be interpretive constructs. Perhaps readers think that ontology is unimportant, having no substantive implications for the content of deliberative systems theory. The remainder of this article will suggest otherwise. We have already shown how the neglect of ontology leads to theorists confusedly shifting back and forth between incompatible presuppositions. Later we will argue that ontological presuppositions define the field of normative possibilities and empirical agendas open to deliberative systems theory (Taylor, 1989: 160–162).

Functionalism

We have suggested that the existing responses to the problems of boundaries and evaluation pull deliberative systems theorists in opposite directions. They point to two different ontologies: functionalist and interpretive. Although these two ontologies are usually implicit, it is possible to reconstruct them as systematic approaches. This reconstruction builds on an emerging literature that asks what ontological commitments and empirical methods best fit with deliberative theories (Bächtiger and Parkinson, 2019; Bevir and Ansari, 2012; Boswell and Corbett, 2017). We will extend this literature by reconstructing two possible accounts of a deliberative system. We will also suggest that each account offers coherent responses to the problems of boundaries and evaluation.

The functionalist approach to deliberative systems draws on a broader sociological functionalism. This functionalism appears in the study of social systems by theorists such as Talcott Parsons (1977) and Niklas Luhmann (1995, 2013). In this tradition, a social system is understood to be analogous to a mechanical, biological, or ecological system. For a system to persist, it has to maintain its equilibrium despite constant changes in its parts and its external environment. Systems have to self-regulate and meet other functional needs. Functionalists here explain the operation of a system by pointing to its ability to meet its functional needs. Crucially, their idea of functional needs implies that a system possesses its own independent problem-solving logic. Any talk of subjective beliefs, motives, and goals is, therefore, at best a residual status of their functional explanations.

Social theorists often explicate functionalism using mechanical examples. Consider the case of a central heating system (Barnes, 1995: 38–39). Human engineers design a central heating system to keep a room within a temperature range—its equilibrium—amid changing conditions. Systems can maintain their equilibrium only if it meets certain functional needs. A central heating system, for example, must be sensitive to changes in the room temperature. It must include parts that meet this functional need, including temperature sensors to provide feedback information about the room temperature and to enable the output components of the system to adjust accordingly. Furthermore, because a sensor might misread the temperature, well-designed heating systems often have multiple sensors of different kinds thereby evening out the variations in reading and correcting for dysfunctional sensors. The extra sensors may look redundant, but this redundancy has a function: it enables the system to deal with internal disturbances, ensuring that no single part of the system is essential to its operation.

When functionalists turn from mechanical systems to social systems, they have to account for social change. Functionalists usually discuss social change in terms of processes of social evolution and systemic differentiation. In this view, as an environment presents new problems to a social system, so the system evolves by dividing itself into more differentiated parts and especially by dividing the functional labor between these parts. Parsons’ (1956) account of the family is a good example. Parsons argues that whereas traditional societies rely on kinship units to fulfill social, economic, and political functions, modern societies have become more differentiated in ways that strip the family of its functional role in maintaining social integration, economic production, and political authority. The nuclear family now serves only two functions: “first, the primary socialization of children so that they can truly become members of the society into which they have been born; second, the stabilization of the adult personalities of the population of the society” (Parsons, 1956: 16). Parsons’ account is functionalist in that, first, he presents the nuclear family as part of a social system and, second, he discusses it in terms of its functions within that system.

Functionalism can, but need not, be deterministic. Functionalists can allow that environmental contingencies, including the unexpected consequences of human action, might undermine one equilibrium and lead a system to another equilibrium (Clark, 2005). However, functionalism makes irrelevant discussion of human intentionality. Obviously, functionalists can refer to contingent beliefs, reasons, and actions. They can grant, for example, that people have various contingent reasons for getting married and having children. Equally, however, functionalists cannot allow these reasons to play an explanatory role in the functioning of the system. To continue with Parsons’ account, it does not matter why people get into marriages and have children. What matters is, rather, that they do so within nuclear families that fulfil the function of socialization for the system as a whole.

More generally, functionalist theories exclude human intentionality in two ways. First, because functionalists analyze a system as a whole, they examine the properties of the system as a whole rather than its parts. Functionalists thus ascribe to the social system a self-active problem-solving logic independent of human intentionality that allows it to maintain its equilibrium. Second, because functionalists explain social practices by their consequences for the system and its functioning, the reasons agents have for engaging in the practices are irrelevant. As long as people’s actions have the same effects on the system, it does not matter why they take those actions. ¹

The functionalist approach to deliberative systems largely resembles the functionalist approach to social systems more generally. Typically, however, a deliberative system appears less like an eco-system and more like a human-made machine than does a social system. For a start, whereas sociological functionalists limit the purpose of a social system to its own self-maintenance or persistence, deliberative democrats want their systems also to have or generate deliberative properties. The deliberative system, like a central heating system, should serve not only its own equilibrium or survival but also a human goal such as a desirable level of deliberation or a desirable temperature. In addition, whereas sociological functionalists treat evolution as independent of intentionality, deliberative democrats typically give policymakers and even themselves a role as engineers. They allow that policymakers and social scientists can tweak the configuration of deliberative systems just as one might mend a malfunctioning heating system. Nonetheless, they restrict human intentionality to this role. Once the central heating system is fixed, it runs by itself. The intentionality of the agents within the system does not have a significant effect on its outcomes.

Functionalism suggests clear responses to the problems of boundaries and evaluation. A functionalist approach to boundaries contains three distinct steps. First, functionalists postulate the functional goals of the deliberative system—the equilibrium in which the system needs to maintain itself—by drawing from the normative literature. Functionalists can take this first step because they treat a deliberative system as analogous to a human-made machine; they allow scholars and policymakers to decide the functional goals of the system. Second, functionalists derive the functional needs of the deliberative system from these functional goals. This second step follows logically from the first given the functionalist assumption that functional needs arise from functional goals. Third, functionalists include as parts of the deliberative system those practices and sites that currently effect, or at least have the potential to effect, these functional needs. This third step follows given the assumption that the deliberative system has, either actually or potentially, a self-active problem-solving logic independent of human intentionality. The capacity to realize functional needs is part of the system’s own configuration, not contingent human intentions. Any practice that has the potential to help the system to realize its functional needs or deal with disturbances is thus part of the deliberative system. Functionalists can hence proceed and investigate how this part has to be configured in itself or in its relationship with other parts of the system in order for it to realize its potentials. Crucially, for functionalists to proceed in this manner, the practice has to be defined by its configuration and function to the system and not by the human intentionality underlying the practice.

Let us now turn to the problem of evaluation. The analogy with a central heating system is useful here. Because engineers design the heating system to serve a particular function (i.e. keeping the room within a stable temperature range), they can evaluate its performance against that goal (i.e. by checking room temperature). Functionalists can thus define the goals for a deliberative system and then use empirical measures of outcomes to check whether it is delivering those goals. Furthermore, functionalists can fix the relationship between part and whole by distributing the functional needs of the system between its parts. Just as temperature sensors are responsible for informing the heating system about the room temperature instead of heating the room, so functionalists can say that the media is a part of the deliberative system in the sense that it generates and disseminates necessary information for deliberation, even if it is not itself deliberative. These functionalist positions again reflect their dismissal of human intentionality. They tacitly attribute a failure to realize deliberative goals to the configuration of the deliberative system itself rather than to contingent human intentionality. A failure to fulfill functional needs also would derive from the parts of the system or the relationships between the parts.

Interpretivism

Interpretivism, unlike functionalism, treats human meanings and intentionality as vital to social explanation. Interpretivists argue that humans act on languages, discourses, reasons, and beliefs. There are, of course, many differences between interpretivists. For our purposes, however, the key thing is that they typically treat meanings or beliefs as interconnected. They are reluctant to reduce or correlate a belief or attitude to or with an objective category in large part because they think the content of any belief or attitude depends on its place within a wider language or webs of belief. Interpretivists lean, in other words, towards a meaning holism that insists that beliefs gain meaning from their place within wider webs that sustain and inform them (Bevir and Blakely, 2018: 21). Whenever we try to explicate a belief or a reason, we enter what Hans-Georg Gadamer (2004) calls a hermeneutical circle. The explication must appeal not to objective correlates but to other beliefs and reasons. This hermeneutic circle does not become an infinite regress and undermine our daily communication with each other because we already share some social practices and ways of life. Our shared practices ground the possibility of mutual understanding. Ludwig Wittgenstein (1969: §110) explains: “As if giving grounds did not come to an end sometime. But the end is not an ungrounded presupposition: it is an ungrounded way of acting.”

This interpretivism has clear implications for the analysis of social concepts. For a start, interpretivists generally argue that social concepts should refer, at least implicitly, to human intentionality. Because people act on their conscious, subconscious, and unconscious reasons, and because these reasons make sense only in the context of their wider webs of belief, political scientists should explain actions by reference to reasons and webs of belief. Political scientists might adopt simplified analyses that avoid appeals to reasons or webs of belief; but when they do, interpretivists believe that they should recognize that their simplified analyses should be in principle ones that we can unpack in terms of postulated reasons and webs of belief. A full explanation of action should relate the action to the actors’ reasons and those reasons to the actors’ wider intentionality. Indeed, interpretivists often argue that even most descriptions of social life should be capable of being unpacked in terms of intentionality. Unlike the brute facts of natural sciences, social reality consists largely of institutions, norms, and practices—all of which are products of human actions embodying reasons within webs of belief. Social reality is constructed out of human activity, the meaning of which we can only understand through interpreting the intentionality of actors (Searle, 1995).

We can illustrate the role of intentionality in descriptive and explanatory concepts using the example of friendship. What is a friend? What does it mean to be a friend? To answer this question, interpretivists argue that we have to look beyond some atomistic definition of friendship and refer to a web of beliefs about friendship and the shared social practices that support these beliefs. These beliefs include beliefs about what a friend is for, the duties of a friend, and what counts as a true friend. They also include background beliefs about the sociability of humans and about norms such as reciprocity, hospitality, and sincerity. People acquire these beliefs through their immersion within a historical tradition, including guidance from adults, movies, games, novels, and so forth. Perhaps most importantly, they develop and adjust these beliefs, often subconsciously, as they interact with others and make friends. These beliefs vary from society to society and even between individuals within a society. Some societies, such as the ancient Greeks, seem to have valued friendship above even marriage and certainly not to have seen friendship as part of marriage. Other societies might value a good marriage above friendship and see friendship as an integral part of a good marriage. Interpretivists argue that therefore if they are fully to explain, or even just to describe, actions and practices based on friendship they should appeal to the relevant webs of beliefs. Interpretivists explain why people treat a friend in a particular way by reference to their beliefs and desires, and interpretivists explain those beliefs and desires by locating them in webs that derive in part from a tradition, whether it be ancient Greek, contemporary, or some other one.

Interpretivists generally argue, in addition, that social explanations deploy concepts that social scientists define pragmatically by reference to their purposes. Because intentional objects, such as beliefs, are constituted in part by their part in larger webs, they do not have essences. Friendship does not have a single correct definition in the way that, arguably, water or oxygen does. Friendship means one thing to one person in one cultural context, and another thing to another person in another cultural context. Interpretivists are thus anti-essentialists at least with respect to most of the social world. Anti-essentialists believe that we cannot find an essence—a fixed element or standard—by which to judge one use as uniquely correct. As Ludwig Wittgenstein (2009: §65) says, there is “no one thing in common which makes us use the same word for all – but that they are related to one another in many different ways.” This anti-essentialism implies that there is no right way to define aggregate political concepts, including liberalism or even legislature. Political scientists cannot justify their use and definition of these concepts by reference to some alleged essence. They can do so only by the role the concepts play in explaining or describing that which interests them. Political scientists can legitimately define liberalism one way if they are explaining or describing the ideas of the American Founders and another way if they are explaining or describing current trends in welfare policy. They justify their definition, in other words, not by saying that it captures fixed boundaries in nature but, rather, by saying that it best describes or explains the patterns of intentionality that interest them.

Crucially, interpretivism therefore points to very different responses to the problems of boundaries and evaluation from those associated with functionalism. An interpretivist response to the problem of boundaries should reflect the stress on anti-essentialist intentional concepts. Interpretivists should allow that there is a sense in which deliberative democrats can define the boundaries of a deliberative system pragmatically in accord with their research interests. Equally, of course, they will add that these boundaries must map on to patterns of intentionality. Deliberative democrats can choose which pattern they want to describe or explain, but there must be—or at least the social scientist must try to show that there is—a real pattern there. Because deliberative democrats can only interpret the meaning of practices through the intentionality of agents, whether these practices fall within a deliberative system, however defined, depends on the intentionality of the relevant agents. Indeed, because people can change and disagree over the meaning of their practices, the boundaries of a deliberative system or community are necessarily contingent and contestable; they are potentially subject to disjunctions and transformations (Ercan et al., 2017: 199–201).

An interpretivist response to the problem of evaluation similarly allows deliberative democrats to define concepts to fit the patterns of intentionality that interest them. On the one hand, a stress on intentionality definitely undermines a concern with functional relationships. For interpretivists, functional relationships exist only as contingent products of intentionality embedded in actions. Interpretivists are extremely unlikely, therefore, to have any interest in evaluating practices by reference to their alleged function in a deliberative system. On the other hand, a stress on intentionality does not require an interpretivist to evaluate agents in terms of their own beliefs and values. Interpretivists are free to choose whatever evaluative principles fit their normative agendas provided only that these principles are consistent with their ontology.

We can illustrate the role of intentionality in evaluating deliberative systems using the example of protests. Interpretivists will reject the functionalist suggestion that we evaluate protest in terms of a trade-off between its epistemic function of forcing “new reasons, facts, and information into public conversation” and its ethical dysfunction of eroding respectful mutual interaction (Mansbridge et al., 2012: 19). They will note that this suggestion appeals to a functional relationship between protest and a deliberative system, and that this relationship does not map on to a pattern of intentionality among the actors. For example, when activists rescue animals from animal testing laboratories, some of them might intend their act to be communicative, believing they are bringing attention to the issue, but other activists might simply want to rescue the animals and have no interest in communication (Humphrey and Stears, 2006). Interpretivists can only posit a communicative relationship between the animal rescuers and the parliament in the first case but even then not in a functionalistic sense. Interpretivists may then focus on different dimensions of this relationship and evaluate them according to different evaluating principles, depending on whether they are interested in deliberative democracy, direct action, animal ethics, or some other normative agendas.

Two different research agendas

Functionalism and interpretivism are rival ontologies offering competing analyses of deliberative systems. On the one hand, functionalists see any presumed deliberative system as literally a system with functional goals that might be distributed across its different parts. On the other hand, interpretivists see any presumed deliberative system as an abstraction that social scientists use to pick out a pattern of intentional human activity. Furthermore, these different analyses of deliberative systems inspire, in turn, competing forms of explanation. On the one hand, functionalists explain social reality as systemic outcomes that result from the configuration of the system. On the other hand, interpretivists explain social reality as the product of contingent and intentional human activity. When deliberative systems scholars try to combine these incompatible ontologies, the result is confusion. Once they distinguish them, they face a choice between two incompatible research agendas.

Most deliberative systems scholars have yet to recognize the choice they face between rival ontologies. Some lean unreflectively towards one of these ontologies. Others, even more problematically, unintentionally mix the two ontologies, creating tensions, ambiguities, and other problems for their research. It is, for example, because they fail adequately to address ontological issues that they have not satisfactorily addressed the problems of boundaries and evaluation. Consider, once again, their manifesto. At one point, the manifesto claims, “one can define the boundaries of a deliberative system either institutionally or by reference to a particular issue,” which suggests that researchers can make that decision in accord with their pragmatic interests in specific patterns of action (Mansbridge et al., 2012: 8). Yet, at another point, it claims that the different parts of the system should be assessed “according to how well they perform the functions necessary to promote the goals of the system” (Mansbridge et al., 2012: 11). Surely, however, if any given deliberative system is an abstraction postulated by social scientists, we have no reason to ascribe goals to it let alone to assess it in terms of such goals. Consider also the attempt by Bächtiger and Parkinson to reconcile the additive and summative approaches to the problem of evaluation. Their analogy for the summative approach—that “no part of a car is itself a car”—appeals to a functionalist ontology such that a deliberative system is an objective system with functional goals served by its parts. Yet, they then switch to an interpretivist ontology when they worry that this summative approach creates the possibility of a deliberative system that does not contain any act of deliberation. A consistent functionalist need not worry about that. Their worry arises only if one accepts the interpretivist view that the content of a deliberative system comes from the intentionality of the actors within it.

Ontological clarity allows us to clear up these confusions. Crucially, functionalism and interpretivism lead to two very different research agendas—empirically and normatively—that deliberative systems scholars might pursue. Let us begin by looking at their respective empirical agendas. Here a functionalist ontology supports a research agenda that asks if and how the configuration of a deliberative system enables it to achieve its functional goals. Functionalism establishes that a system can be configured in a way such that its outcomes remain constant despite contingencies. This ontology makes it plausible to assume that the functional successes and failures of a system can be explained by its configuration. This assumption implies further that either changing a part of the system or changing how the parts relate will bring corresponding changes to the system’s outcomes. This implication lends credibility to studies that focus on particular parts of a deliberative system (Almeida and Cunha, 2016) and studies that focus on the relationships between the different parts of a deliberative system (Cinalli and O’Flynn, 2014: 435–436), examining how their configuration influences its ability to realize certain functional needs. Furthermore, because functionalism suggests that a system can differentiate itself into parts and divide the functional labor between these parts, it promotes research that looks at the different ways a system differentiates and distributes different functional needs between its different parts (Bohman, 2012; Goodin, 2005). Indeed, all of these kinds of studies seem to rely, at least tacitly, on a functionalist ontology both to define what count as a part of a deliberative system and to postulate the system’s functional needs.

We can see how a functionalist ontology supports these research agendas by looking at the example of Almeida and Cunha’s (2016) analysis of the Brazilian Social Assistance Unified System (SUAS). Although Almeida and Cunha do not explicitly appeal to functionalism, they effectively posit political legitimacy as a functional goal of a deliberative system, argue that connectivity between different discursive spheres is a functional need that arises out of that goal, and focus their research on the question of which parts of a system are responsible for realizing that functional need. Their research question presupposes that we can reliably identify the parts of a deliberative system that explain its functional success, a presupposition that assumes a functionalist ontology such that the configuration of a system can fix its outcomes. Almeida and Cunha’s functionalist commitments require them to leave human intentionality out of the picture. Thus, although they study connectivity—that is, the ways in which participants in one sphere respond to participants in another sphere—they cannot appeal to the intentionality of these participants. They argue instead that the two parts of the system that govern connectivity are a properly configured institutional design and the circulation of participants. For Almeida and Cunha (2016: 299), the configurations “define” how the interactions happen; the former fix the latter.

An interpretive ontology supports a very different empirical agenda. For a start, interpretivism suggests that the concept of a deliberative system acts primarily as a heuristic with which to explore various aspects of social reality. This approach is credible because interpretivism implies that social scientists can define aggregate concepts pragmatically to reflect their interests. Equally, however, interpretivism precludes essentialist and functionalist explanations because it prompts social scientists to map their aggregate concepts, at least implicitly, on to patterns of intentionality. Interpretivism thus supports research that uses the idea of deliberative system as a heuristic to draw out the sequences in which ideas or claims pass through different spheres, but it does not allow us to presuppose that a sequential design fixes the transmission of ideas (Mansbridge, 1999; Parkinson, 2018: 439). Likewise, although interpretivism supports studies that look at how different groups and spheres are connected, it mitigates against any presupposition that these relationships are fixed by institutional design (Boswell et al., 2016; Ercan et al., 2017: 201–203; Hendriks, 2016). Additionally, although it supports studies that draw out family resemblances across various deliberative systems, the relevant comparisons cannot rest on essentialist definitions or rigid categorization (Boswell and Corbett, 2017). In addition, interpretivism typically implies that social scientists should pay attention to the contingency of the ways people reason against the background of diverse social traditions. It thus supports the use of deliberative systems as a heuristic to identify how participants from different spheres create and contest the boundaries of a system such that some discourses, spheres, and participants dominate over others (Ercan et al., 2017: 199–201). All these research agendas rely, at least tacitly, on an interpretive ontology to ground their pragmatic definitions of deliberative systems and to direct their research interests.

We can see how interpretive ontology supports these research agendas by looking at a comparative study of the deliberative systems surrounding honor killing debates in Britain and Germany (Boswell et al., 2016; Ercan, 2014). Although this study does not appeal explicitly to interpretivism, the researchers have endorsed an interpretive approach elsewhere (Ercan et al., 2017). The researchers use the concept of a deliberative system as a heuristic to draw out the diffused debates and meaning-making processes that contribute to defining the problem of honor killing and possible solutions to it. Their study suggests that because participants in Germany and Britain understand the issue differently, they end up treating different groups and perspectives as key to defining and resolving the problem. This argument relies on an interpretive ontology that both ties actions to the actors’ beliefs and unpacks these beliefs by referring to a wider web of background beliefs. Furthermore, the researchers seem to suggest that because human agency introduces contingency into the deliberative system, institutional design cannot fix the system’s outcomes. When they discuss institutional factors, they thus add the caveat that “it would be mistaken to reduce the differences between these two cases to the differences in the institutional settings of these countries” (Boswell et al., 2016: 269). If this caveat later seems a bit empty—if their study later appeals to institutional design to explain the outcomes without any reference to intentionality—a deeper understanding of interpretive ontology might help clarify what is at stake. From an interpretivist perspective, the problem is not reductionism; it is the assumption that institutional settings can explain action independent of intentionality.

Different ontologies arguably lead to incompatible normative as well as empirical research agendas. Functionalism and interpretivism certainly suit rather different defenses of deliberative democracy. A functionalist ontology is compatible, firstly, with an instrumental defense of deliberative democracy. This defense justifies deliberative democracy by linking it to outcomes we value. Many deliberative democrats argue, for example, that deliberative democracy produces better decisions. Sometimes they argue that the decisions are better because the act of deliberation changes people’s preferences, filtering out their pathological and irrational preferences (Goodin, 1986; Habermas, 2006) or aligning their preferences along a single coherent spectrum (Dryzek and List, 2003). Sometimes they argue that the decisions are better because of the epistemic diversity that comes from involving more and diverse people in the process (Landemore, 2012: 97–103). Crucially, these instrumental defenses of deliberative democracy work straightforwardly only if we assume that the outcomes are a product of the system, not the contingent intentionality and action of agents within it.

A functionalist ontology is compatible, secondly, with a distributive-proceduralist defense of deliberative democracy. This defense assumes that because different parts of a democracy serve different functions, they should be governed by different norms. Goodin’s (2005) sequenced model is a prime example. Goodin (2005: 189) argues that we can conceive representative democracy as consisting of four stages: the caucus room, parliamentary debate, election campaign, and post-election arguing and bargaining. He argues that the caucus room could exhibit the ideal of authenticity but only in secrecy, parliamentary debates could exhibit the ideal of rationality although nobody is expected to change their minds in the parliament, the election campaign is open but less reasoned, and post-election bargaining demonstrates respect for each other’s demands although not premised upon reasons. Goodin (2005: 193) argues that it is fine for these deliberative virtues to come sequentially as long as they are “coming in the right combination and the right order.” In this view, each part of a deliberative system needs to exhibit only those ideals that are relevant to its functional role. If each part plays its role, the deliberative system will necessarily fulfill its overall goals.

If functionalism fits well with instrumental and distributive-proceduralist defenses of deliberative democracy, it seems incompatible with intrinsic ones. Most intrinsic defenses offer reasons to encourage deliberative practices regardless of their outcome. Some claim that “citizens owe one another justifications for the laws they collectively impose on one another” (Gutmann and Thompson, 2004: 126). Others begin from the idea that democracy is “about making collective decisions and exercising power in ways that trace to the reasoning of the equals who are subject to the decisions” (Cohen, 2007: 220). These intrinsic defenses of deliberative democracy generally presuppose that we should respect people’s agency so deliberation and participation are crucial to legitimate decision-making. This emphasis on agency implies, in turn, that people are capable of forming their beliefs through their own contingent reasoning. Intrinsic defenses of deliberation presuppose, therefore, that the system does not determine the quality and content of people’s reasons (Benhabib, 1994: 33; Urbinati, 2014). The participants have an agency that means they decide how to make their arguments and whether to accept, reject, or modify others’ arguments. By writing individual agency out of its mode of explanation, the functionalist approach erases the intrinsic reason for ascribing normative significance to deliberation. If functionalists try to reintroduce contingent agency into their theory, they undermine their own attempts to explain outcomes by reference to the configuration of the system.

An interpretive ontology supports almost the exact opposite normative agenda from that associated with functionalism. Interpretivism suggests that the deliberative system and its functional goals cannot be an independent normative standard, but only a pragmatic stand-in for more intrinsic values. Assuming the citizens share those intrinsic democratic values, an interpretivist’s evaluation of a deliberative system in terms of those values is likely to map onto the self-understanding of those citizens. Even if the citizens do not value agency and participation, an interpretivist still might be interested in how well citizens’ beliefs, actions, and practices accord with those values. Crucially, in this latter case the interpretivist still will treat the deliberative system as composed of the contingent beliefs and actions of citizens, so their evaluation of the system will refer to the intentionality of the participants within it.

Interpretivism fits less well with instrumental defenses of deliberative democracy. One problem is that the interpretivists’ emphasis on creative agency implies that there is always some contingency about the outcomes of a deliberative system. Even if interpretivists value particular outcomes, they will resist attempts to suggest that deliberative systems, or certain types of deliberative system, are bound to generate those outcomes. As a result, they will be less ready to justify deliberative systems by tying them to particular outcomes. Interpretivists may emphasize that deliberative procedures can fail to “launder” people’s preferences if people mask their pathological interests with widely accepted social norms and cultural values (Young, 2001: 685). Interpretivists may add that deliberative procedures can fail to produce epistemically superior outcomes simply because human agents are prone to mistakes both individually and collectively: people can be led astray by false information and poor reasoning no matter how well a system is designed (Urbinati, 2014). Finally, interpretivists may argue that deliberative procedures can fail to make people align their preferences along a single spectrum because people can respond to the same set of arguments differently and take on varying directions of reflection (Cohen, 2007: 224–227). In short, once we are committed to understanding intentional agency as creative and contingent, instrumental defenses lose much of their force.

An interpretive ontology is also in tension with a distributive-proceduralist defense of deliberative democracy. For a start, the interpretivist response to the problem of boundaries counts against the attempt to assign different roles to different parts of the system. Because every part of the system is subject to reinterpretation and contestation, deliberative democrats cannot define and assign functions to the different parts prior to the task of interpretation. To put this more concretely, they cannot stipulate what a legislature is for or what activist groups are for until they have looked at the ways in which the relevant agents behave. In addition, because human agency creates and recreates the systemic outcomes, they cannot straightforwardly bypass individuals and assign responsibilities to the various parts of the system.

Conclusion

We have suggested that deliberative systems theory has much potential but it lacks ontological clarity. This lack of clarity prevents robust responses to the problems of boundary and evaluation, thereby hindering the development of the theory. We have also identified two distinct ontologies with clear philosophical commitments that suggest different responses to these problems and so different empirical and normative directions for deliberative systems theory.

Some deliberative democrats may worry that the demand for such clarity comes at a cost to the development of the field. Their worry stems from the belief that ontological ambiguities have enabled scholars from different disciplines and with different commitments to collaborate with each other. ² This worry should be largely alleviated once we differentiate between the work of individual deliberative democrats and the field of deliberative democracy. On one level, we would suggest that deliberative systems theory as a field might develop with these two ontologies debating and arguing with each other. Each might prompt the other to develop and improve its internal arguments. Each might provide insights that the other can adapt to fit its own perspective. On another level, however, individual deliberative democrats should avoid conceptual confusion, ambiguity, and inconsistency in their works by clarifying their views and debating with others who hold different commitments. Our analysis then presents deliberative democrats with a choice, perhaps even an uncomfortable one.

If deliberative democrats believe in the intrinsic value of things such as agency, reasoning, deliberation, and participation, they will struggle to reconcile their values with a functionalist ontology. If, however, they therefore adopt an interpretive ontology, they will have to drop the talk about deliberative functions. The problem they would then face is that the motivations for the systemic turn might seem to require just such talk of deliberative functions. If the point of the systemic turn is to restore faith in deliberative practices within mass democracies by showing how each practice feeds into a broader system, it seems to require talk of differentiated functions. More importantly, if the point of the systemic turn is to justify non-deliberative practices by showing how they can serve deliberative functions, they surely have to adopt a functionalist language that seems to undercut their focus on intrinsic values. Our general point is perhaps simply that one cannot “have it all.”