1 Anti-naturalism

He approaches the critique of critical realism through the question of naturalism, in the sense of claims that the social world can be explained in similar terms to the natural world. This is where he starts to go astray. Roughly speaking the argument is, first, that we can’t produce ontologies unless we are creative and possessed of genuine agency, imagination and reflectivity; second, that there is no place for these in the subjects of the natural sciences; and hence third, the very existence of ontologies proves that the ontology of the social world and the methods appropriate for its study are different from those of the natural world (162–163). Thus, for example, after criticising critical realists for giving insufficient weight to agential creativity and reflection, he writes: ‘The ways in which we organise our coordination of communication, interaction and practices do not resemble the ways in which molecules are related, following a quasi-determinate compositional form. These projections of alien schemes onto the social world lead to theoretical absurdities’ (98). And he insists that there is no room for reconciliation with realists as long as they ‘invoke naturalist imageries of natural necessity, causality, composition and emergence’ (125).

He is right, of course, that critical realists believe that certain ontological principles apply equally well to the social as to the natural world, including causality, composition and emergence (I will return to natural necessity below). But those commitments are not dogmatic prejudices: they arise from examining when, how and how far these principles work and help us to understand social events, and they are backed up with careful consideration of the processes involved. I will give an example shortly, but first we need to open the black box of naturalism.

The first problem here is that Bouzanis is working with a crude binary version of the concept of naturalism and the alternative to it. In this version, naturalism means that the natural and social sciences are all alike in some crucial aspects of methodology or ontology, and the alternative is that the natural sciences are all like each other in these respects, while the social sciences are also all like each other, but different from the natural sciences. But, as Ted Benton has put it (in a paper that Bouzanis cites), in fact ‘the sciences display a “family resemblance”, of cross-cutting and overlapping differences and similarities of method’ (Benton 1985, 192). Not all the natural sciences can be experimental, for example, and not all the social sciences can employ interactive methods, while quantitative methods can be employed on both sides of the divide. As we shall see, critical realists argue that there are important differences between the two groups, but there can also be commonalities. It would be bizarre, for example, to believe that biological factors have no relevance to social behaviour or that the study of human biology has no relevance to the explanation of social behaviour. There can therefore be no a priori reason to believe that any given concept that applies in the natural sciences cannot apply in the social sciences.

Indeed, here’s a conundrum for critics who wish to pin naturalism on critical realism: we argue that ‘entities at different ontological levels have different sorts of properties that must be studied in different sorts of ways’ (Elder-Vass 2010, 197). That, of course, is an example of naturalism, in the sense that it is a general principle that applies to both the natural and social sciences, and also of anti-naturalism, since it leads to the view that social phenomena need different methods of study than natural ones. As Bouzanis recognises, for example, Roy Bhaskar specifies a series of reasons why social structures cannot be studied in the same ways as natural objects, of which the most important is that they are concept dependent. So, binaristic anti-naturalism is more or less incoherent, and critical realism is far from being purely naturalist.

2 Anti-realism

Bouzanis’s real concern, however, isn’t naturalism in general but quite a different argument that he seeks to undermine by labelling it as an application of naturalistic reasoning. Naturalism, in other words, is just a stalking horse here for his real target: social realism, or to be more specific the use of categories such as causality, composition and emergence to justify realist explanations of social structures and their influence on social events, which he takes to be incompatible with allowing creative agency to human individuals.

He returns here to the notion of natural necessity, employed by Bhaskar, and suggests that critical realists ‘project’ this naturalistic image from the natural sciences onto human–material relations (118). The implication, often repeated, is that critical realists think of human action as determined by external forces. This, however, rests on a misreading of Bhaskar’s use of the concept of natural necessity. He introduces the concept in A Realist Theory of Science, where he discusses three different versions of the concept: an empiricist, a Kantian and a critical realist version (Bhaskar 1975, 163–165). In the empiricist version, to which Bhaskar is implacably opposed, an event of type A always produces a further event of type B. In the critical realist version he advocates, there may be a law-like but not a deterministic connection between A and B, but only if there is a mechanism that tends to produce this sequence (see also Bhaskar 1998, 10). For critical realists, in other words, structures, whether natural or social, can have powers to tend to bring about certain effects, but the realisation of those effects always also depends on what other causal influences might interact with the first one in any given case (Elder-Vass 2010, Chapter 3). And as realists repeatedly stress, individual human beings have agentic powers: their decisions are not entirely determined, though they may be influenced, by external causal powers, and they themselves have a causal influence on outcomes (Elder-Vass 2010, Chapter 5). There is no external determination of human action here, though there is certainly influence, and it is difficult to see how anyone could deny such influences.

Similarly, Bouzanis suggests that the idea that social structural powers are emergent is incompatible with agency, despite the protests of critical realists to the contrary (6–7). To question this, let me introduce a case: my own account of the emergent power of organisations (Elder-Vass 2010, Chapter 7) and an illustrative hypothetical case, in which a salesperson sells a television on behalf of a retail business (Elder-Vass 2010, 170–176). I argue that organisations are composed (at least) of people and have emergent powers that the same set of people would not have, individually or collectively, if they were not formed into an organisation of this type. Those powers result from the particular ways in which the organisation’s parts, including the people, are organised (their roles) when they become members of the organisation. The addition of the relations between the people arising from their acceptance of roles in the organisation gives the group as a whole emergent powers, which are exercised through the parts of the organisation, primarily its human members. Thus, for example, a retail organisation has the power to sell and deliver a television. The salesperson who completes the sale does so on behalf of the organisation (at no point does the salesperson own the television!) and she does so because she is a member of the organisation. As a result of being such a member and accepting the role of salesperson, she knows that she is expected to sell televisions on its behalf, and in the hypothetical case she does so—an action she would not have taken if the organisation did not exist and/or she was not part of it. The existence and policies of the organisation have a clear causal influence on the actions of the salesperson. Yet they do not determine them. She can always choose to do otherwise, and she also has a degree of choice over how she chooses to sell the television. Her actions, like other social events but also like natural events, are multiply determined by many interacting causes, including in this case her own powers of agency.

There is thus no conflict between recognising that social structures such as organisations have emergent causal powers to influence human actions and also maintaining that the humans who are influenced continue to have the power to be reflective and creative. Reflectivity and creativity cannot operate in a vacuum: they necessarily interact with a context and causal influences from it. Social structure does not suppress agency but interacts with it and provides an element of its context—as indeed other critical realists such as Margaret Archer (1995) would recognise. And to return to the predicament Bouzanis poses, that’s quite enough agency for us to do ontology!

Bouzanis objects that I simply ‘project’ my generic analysis of emergence onto social examples (85). That’s a difficult criticism to understand when I have worked through some of the cases at this level of detail (and of course there’s more where that came from!). But it is also difficult to understand because I have examined the ontology carefully and made clear how it is affected by the unique characteristics of the social world. Most notably, I have argued that composition works differently in the social world, because the form of the relations between people that constitute them into social structures is different from the form of the relations between physical parts that constitute them into material structures (Elder-Vass 2010, 199–202). Rather than those relationships being spatial in form, as they are in material structures, our social relationships are intentional in form, in the sense that they depend on our cognitive beliefs and commitments rather than our spatial relationships to other people. That has several important consequences: it means that social structures can be spatially disarticulated in a way that material structures (except perhaps in the realm of quantum physics) cannot; it means that the same person can be part of multiple non-nested structures, unlike the parts of material wholes; and it means that people who are parts of social structures are not necessarily their parts continuously over time. This, in other words, is a radically non-naturalist account of composition, and Bouzanis’s claims that natural necessity, composition, causality and emergence make social structure incompatible with agency fall apart on closer inspection.

3 Imaginaries

In my experience, most critics of emergentist social realism want to locate all social causal influence at the level of individuals. Bouzanis, however, has a more substantive alternative of his own to propose, which comes in the form of ‘the social imaginary’, which he has adopted and adapted from Bradley, Boulding, Castoriadis and Taylor (Bouzanis 2023, 103–121). The social imaginary, he says, consists of ‘conflicting, mutually supportive, overlapping world-imageries which constitute the “object” of reflection for imaginative social agents participating in different communities’ and ‘a shared ideational background of interrelated imageries of the world(s) or sub-worlds that inform our categorical and perceptual engagement with events and objects’ (8). This, he argues, provides the basis for an anti-realist, and indeed an explicitly idealist, social ontology, which ‘renders the ideational dimension as the ultimate one, existentially prior to the material/structural dimension’ (180).

Particularly because he flirts with the earlier thinkers mentioned above, it is often quite unclear what form Bouzanis thinks the social imaginary takes. Idealists have sometimes thought of ideas as free-floating, as if they could exist or function entirely independently of minds; as existing or functioning in some sort of higher mind-analogue, like Hegel’s world spirit; or as belonging to a collective mind. Eventually, though, Bouzanis does decisively reject all the usual idealist formulations: ‘the loci of shared worldviews, normative orientations, roles and rules are individual “minds”. There is no transcendental, or non-mental, mysterious entity floating over and above our heads’ (120). This seems, so far, to be a thoroughly realist understanding of the form in which ideas exist: they are mental properties of human individuals.

A second dimension of his understanding of ideas is more questionable: he insists that their fundamental form is as images—and this is not just a metaphor since he repeatedly stresses that ideas take a pictorial form (e.g. 123, 155). His agenda here is to downgrade the significance of language in social ontology and to insist that ideas, including those that constitute ontology itself, do not take a linguistic form (62). The ideational level, he argues, is different from the linguistic level (26). There are familiar ways of justifying this—the separation of discourse from language in Foucault (2002, 116–117), for example, or the division between the signifier and the signified in Saussure (1986). We might think of this as the distinction between a concept (idea) and the various different words that can be used to represent it in different languages and different contexts. But here it seems to me that Bouzanis makes an error: he treats concepts themselves as essentially linguistic, rejecting, for example, Bhaskar’s insistence on the concept-dependence of social structure on the grounds that this makes it dependent on language. At one point, for example, Bouzanis says: ‘We can always use these concepts to name other entities’ (167)—but that’s not how concepts work, that’s how words work! Concepts are signifieds, not signifiers; they are the ideas to which language refers and can be shared between thinkers who employ different languages (language can influence but it does not determine how we understand concepts).

Having rejected concepts as the form of our ideas, Bouzanis seeks to replace them with images. But this is just as misguided. Perhaps there are people who think and store their cognitive commitments using, or in the form of, images. I don’t as far as I can tell. Perhaps Bouzanis himself thinks that way. I rather doubt it, though. My evidence? A book composed of two hundred pages of words with just one single image (consisting of three straight lines, an arrow and 30 words of text) (145). If people think about ontology in pictures, where are the pictures? No, pictures, like words, are just another way of representing our ideas in order to communicate them. I agree that we hold our ideas in a pre-linguistic form, but that form is also pre-pictorial.

The other important feature of Bouzanis’s social imaginaries is that they are shared, but again to say so is to beg a series of questions: in what sense, and how, are they shared, and in what way is this significant to social explanation? Starting with the last of these questions, he suggests that social imaginaries, or, sometimes, the social imaginary, provide ‘the ideational background which enables the common understanding “that makes possible common practices” ’ (115). Moving to the second, he tells us that ‘structurally positioned self-reflective agents … participate in the discursive elaboration of shared world-imageries and normative orientations which constitute the necessary ideational background’ (119). That still leaves us with the first question, which threatens to unravel the whole story. If individual agents elaborate imaginaries discursively, then they do so using language (and perhaps pictures!), but if what is being elaborated is ideational and not linguistic, then it is inescapable that there must be some process of encoding and decoding between the discursive and the ideational moments. Discursive elaboration, then, may lead to developments in the ideas held by individuals, but any consistency or similarity in those ideas is always subject to divergent interpretations by the participants. One of the wonders of language is that we do often end up with ideas that are sufficiently synchronised to enable us to coordinate our actions and to communicate, though always fallibly. Still, any sense that we ‘share’ ideas is always the tentative and uncertain outcome of a demanding process, rather than something that can be treated as unproblematic.

A second concern with concepts like “ ‘the social imaginary’ is that even to the extent it is achieved, the sharing of ideas and normative commitments is profusely heterogeneous and intersectional. Probably no two people in the world share the same set of ideas, and even quite restricted groups with common interests and experiences usually diverge in their understandings of even the concepts linked to their common interests. There is certainly divergence, for example, among critical realists in their understandings of the core concepts of critical realism. Of course, as I have already pointed out, Bouzanis sees imaginaries as ‘conflicting, mutually supportive, overlapping’ so I don’t think he would disagree with these points (8). But that leaves us with the question of what the social imaginary is good for—how does the concept contribute to the tasks of social ontology?

One indicator of what he expects from the imaginary is that he proposes to replace Bhaskar’s assertion of concept-dependence of social structure with ‘imagery … -dependence of social forms’ (165). Social structures, if I understand him correctly, don’t disappear entirely from his ontology; instead they become subsidiary to the imaginaries, the ideational background, which produces a kind of sense of social structures among social agents, much as Kenneth Boulding understands organisations as structures of roles that depend on public imageries (105–107). And this, it seems, is what he has in mind when he claims to be offering an idealist alternative to social realism. Critical realists, apparently, are prevented by their naturalistic commitments from recognising ‘the cognitional priority and existential pervasiveness of the ideational dimension, on which agents draw in order to re-construct institutions and transform social structures’ (147, emphasis in original).

‘Cognitional priority’ and ‘existential pervasiveness’, I’m afraid, are the sorts of phrases that theorists tend to produce as a substitute for substantive arguments. Without them, this argument reduces to the assertion that agents draw on ideas in order to transform social structures. The idea that critical realists are unable to grasp that agents draw on ideas when they interact with, and occasionally transform, social structures, however, is entirely at odds with Bhaskar’s concept-dependence thesis, and every other critical realist account of social structure of which I am aware, including my own and Archer’s. But to assert a kind of priority of the ideational is also to ignore the practicalities of our interactions with social structures. It misses the rootedness of our ideas about social structures in our day-to-day interactions as material persons with the responses of others influenced by social structures. When we observe and experience the enforcement of norms and laws, when we learn that organisations have powers to affect our lives, when we find that money confers massively different life opportunities, this shapes our understandings of structures. It is the experience of structures acting on us that lies behind many of our ideas about them. Yes, of course, those structures also themselves depend on the ideas that we collectively hold about them, but there is no priority here between the ideas and the structural reality, and the ideas themselves are real properties of the real people that participate in the social world, that tend to be homogenised not only by discourse but also by shared experiences.

4 Conclusion

Ultimately, then, Bouzanis is unable to sustain his challenge to critical realist ontology. We already have concept-dependence of social structure; we already have diversely intersectional culture that influences (but does not determine) structure; we already have ideas that need not take linguistic form but are firmly located in human minds/brains. None of that is inconsistent with the idea that social structures have emergent causal powers arising from interaction between the human individuals (and often other parts) that compose them. And none of that is inconsistent with recognising that those humans have imagination, reflectivity and agency at the level required for us to be able to create and debate ontology. Critical realism has already transcended the predicament that Bouzanis thinks philosophy and social theory face: the need for a social ontology that is consistent with an ontology of the physical things that underpin the social, and the need for a general ontology that allows for a social world in which it is possible for that ontology to be produced. He may well be right, however, that many other traditions would fail that test.