Field Theory and Assemblage Theory: Toward a Constructive Dialogue

This paper engages with Manuel DeLanda’s Deleuze-inspired ‘assemblage theory’ from a perspective sympathetic to Pierre Bourdieu’s field theory. It first outlines DeLanda’s proposed new ‘philosophy of society’, focusing on his major works in this vein, and registers some scepticism as to its originality for sociology. It then introduces and responds to DeLanda’s critique of Bourdieu. Rather than simply reject assemblage theory outright, however, I draw on selected insights from DeLanda to push field theory in new directions. More specifically, I conceptualise the interplay of fields and assemblages and use notions of ‘exteriority’ and ‘possibility space’ to help conceive individual plurality of social positioning and its effects for subjectivity and practice.

For almost three decades, urban theorist and philosopher Manuel DeLanda has been trying to wrestle from Gilles Deleuze's giant opus a new 'philosophy of society'.Departing from the usual deployment of Deleuze in social theory as an extension on the Marxist-Foucauldian tradition of ideology/discourse critique to examine contemporary 'societies of control' (Clough et al., 2007), DeLanda endeavours instead to offer clear and consistent concepts that purportedly dissolve some of the age-old antagonisms of social theory while being of concrete use to practising social scientists interested in explaining all manner of historical and contemporary events.More specifically, he champions, elaborates and generalises one concept drawn from Deleuze's sprawling, enigmatic treatise co-authored with Felix Guattari on Capitalism and Schizophrenia (Deleuze andGuattari, 1984, 1988); a concept that is held by DeLanda to traverse physics, chemistry, biology, ecology, psychology, sociology and history, among others.This is the concept of assemblage.
The term 'assemblage' is also used by actor-network theorists (ANTs) and, like most ANTs, DeLanda subscribes to a 'flat ontology', eschewing recourse to deep structures or hidden mechanisms, and a 'generalized symmetry' in which social, material and symbolic entities are treated equivalently.Sometimes, therefore, he is cited alongside ANT and Deleuze himself as a force within the 'new materialist' or 'post-human' currents of contemporary social theory (see e.g.Fox and Alldred, 2017), or is otherwise seen as a crucial force behind the rising popularity of the concept of assemblage for non-conventional forms of conceptualisation and explanation across social, political and geographical theory (e.g.Anderson et al., 2012;Conway, 2022;Elliot and Urry, 2010;Marcus and Saka, 2006;Venn, 2006;Thrift, 2008).Like some ANTs, moreover, DeLanda has explicitly positioned his contribution against that of Pierre Bourdieu.As part of his mission to advocate 'assemblage theory', as he calls it, DeLanda has compared and contrasted his ideas with foundational contributions of major sociologists, from Weber and Goffman to Giddens and Tilly.Many of these are treated sympathetically and their ideas converted into the language of assemblage theory.Bourdieu, however, is given a noticeably rougher ride.While noting some points of apparent convergence or harmony, DeLanda ultimately rejects both the notion of habitus and the notion of field (or social space).These rejections have been noted, but not challenged, by other commentators (e.g.Adkins, 2013;Clough et al., 2007), and Fox and Alldred (2017) have even built on them in their synthesis of new materialist themes to discard Bourdieu's specific vision of social class.
The tasks of this paper are as follows.First, I aim to provide a clear summary of DeLanda's assemblage theory in abstracto and as it relates to human agency and social structures, registering my scepticism as I go regarding its originality or usefulness in toto.Second, after noting some affinities of language between the two thinkers, I will summarise and respond to DeLanda's critique of Bourdieu, essentially showing it to be highly questionable.Despite these problems, however, engaging with assemblage theory does prove fruitful for Bourdieusian social theory because specific insights from it allow us to push deeper on the interplay between fields, networks and the material world and the relational constitution of subjectivity than Bourdieu went, thus taking field theory in productive directions.
Some caveats before continuing.This paper does not try to compare the perspectives of Bourdieu and Deleuze himself, which would be a far larger and much more complicated task, nor am I interested in the degree to which DeLanda departs from or distorts Deleuze's own ideas, which others certainly take issue with.Whether another 'Deleuzian sociology' is possible and superior, therefore, I leave aside.There are others, moreover, who advocate versions of 'assemblage theory', including Jane Bennett (e.g.Bennett, 2010) and Ian Buchanan (e.g. Buchanan, 2020), offering similar ideas (without necessarily agreeing with one other), but not only do they refrain from targeting Bourdieu but they are much less interested than DeLanda in criticising, accommodating or otherwise comparing against large swathes of sociological or social theory.This is what makes DeLanda worthy of singling out, and why I will consider only the relationship between habitus/fields as defined and used by Bourdieu and assemblages as defined and used by DeLanda in what follows.

Assemblage Theory
DeLanda began contributing to social theory in the early 1990s, but assemblage theory per se only consolidated with the publication of A New Philosophy of Society (DeLanda, 2006a) and subsequent works (esp. DeLanda, 2005(esp. DeLanda, , 2006b(esp. DeLanda, , 2008(esp. DeLanda, , 2009(esp. DeLanda, , 2016a(esp. DeLanda, , 2016b)).In these he established that an assemblage is, at bottom, a processual fitting together of elements ('segments' or 'lines') into a whole.The notion is defined against 'strata' on the basis of how elements relate to one another.In a stratum, parts are defined by 'interiority', that is, exclusively in relation to one another, and exhibit complete homogeneity.Examples cited by DeLanda across works include geological strata (rock layers) but also social strata (classes) and family relations (DeLanda, 1997(DeLanda, : 59ff, 2016a: xx): xx).Assemblages, by contrast, are defined by 'exteriority': their component parts, while associated with one another, have relative autonomy from the assemblage -they are not exhaustively defined by it -and are heterogeneous.In DeLanda's eyes, this constitutes a refutation of the Hegelian idea, otherwise present in almost all social theory, that parts are always defined relative to their place in a whole, or totality.
The distinction between strata and assemblages is scalar rather than categorical: components can be bounded off and homogenised to different degrees depending on several factors, including 'coding', which is any process fixing the identity of an assemblage's component parts.Language is a key vehicle for coding.'Territorialisation', a term taken directly from Deleuze, is another factor, referring to the degree to which the boundaries of an assemblage are stabilised and made impermeable and its elements standardised.The difference from coding seems to be that territorialisation is not reducible to language.Instead, territorialisation can depend on either material or 'expressive' elements of an assemblage -a distinction which is, once again, scalar, according to DeLanda.Expressive elements do include linguistic segments -names, labels, etc. -but also symbols, iconography and so on, while material elements include specifically physical entities, such as walls or doors.Both coding and territorialisation, however, are fully changeable and reversible -they can be undone and progress or regress at varying speeds depending on the morphology of the assemblage.
Assemblages possess emergent properties and capacities irreducible to their component parts.They possess things and can do things that the elements on their own cannot, just as water can flow or has a freezing point, unlike its individual hydrogen and oxygen atoms.Thus assemblages constrain and enable their elements, providing limits and opportunities for activity (downward causation), but at the same time assemblages would not exist without their relatively autonomous elements (upwards causation).This is crucial, in DeLanda's mind, for avoiding the micro/macro division (which he also describes as the 'molecular/molar' division, after Deleuze) and the structure/agency division within social theory, because seeing assemblages as emergent entities avoids according them transcendent status over their constituents.Instead, it grants assemblages and segments equivalent ontological status -they are things equally.In the Deleuzian parlance, they exist on the 'plane of immanence'.At the same time, assemblages do not exist in categories with definitive essences but are always unique, singular individuals with their own trajectories (DeLanda, 2006a: ch. 2).
The properties and capacities of assemblages may be actualised or virtual, that is, exist in potentia without being actualised.To each assemblage, therefore, there corresponds a 'possibility space' or 'diagram' (sometimes referred to as the 'multiplicity'): a kind of n-dimensional map of possible developments for the assemblage and its components -their 'lines of flight' -including as they relate to (d)territorialisation and (de) coding (DeLanda, 2016a: ch. 5).When it comes to causal analysis, however, social scientists must eschew linear thinking (A causes B) in favour of thinking in terms of specific interactions and combinations of assemblages and components effecting the probabilities of specific lines of flight, including in the form of catalysis (DeLanda, 2006a: 20) or feedback loops (on this see esp.DeLanda, 1997). 1 They must also focus on the relevant emergent properties of assemblages rather than the activities of individual components if they are to avoid causal redundancy (DeLanda, 2006a).
Similar to ANT, then, assemblage theory refuses to bound off the social or human from the material or symbolic; it places all entities, whether apparently individual (human persons) or collective (groups, classes, etc.), on the same plane of existence; and it attributes agency -the capacity to affect possibilities -to humans and non-humans alike.Hence their conflation in many discussions.DeLanda is unapologetically realist in a more straightforward manner than most ANTs, however: assemblages, he says, exist independently of the conceptions our minds form of them (DeLanda, 2016a: ch. 6).In this he explicitly cleaves close to Bhaskar (1998Bhaskar ( , 2008)), though he distances himself from the apparent essentialism lurking in the latter's philosophy of science.DeLanda also heads off the common criticism levelled at ANT -rightly or wrongly -that it operates with an undeveloped view of human agents and their unignorable specificity relative to other actants or segments (Collins and Yearley, 1992;Pels, 1995;Vandenberghe, 2002).This is because DeLanda (2006aDeLanda ( : ch. 3, 2011DeLanda ( , 2016a: 26ff: 26ff) offers a vision of what makes up humans qua subjects or agents, i.e. on what basis they have the capacity to un/ intentionally affect possibilities, and perhaps it is no surprise that it boils down to seeing human beings as themselves assemblages.

The Model of the Subject
Inspired by Deleuze's reading of David Hume, DeLanda casts the human agent as an assemblage of experiences understood as sense impressions of varying intensity.These are bound together by relations of association and territorialised via habitual repetition, though they can also be deterritorialised by the acquisition of new skills, since these change the capacities and properties of the agent, or anything disrupting routine.Material elements of the assemblage include biological/neurological processes and expressive components include ideas and language, the latter of which can inform beliefs understood as 'attitudes toward propositions' (DeLanda, 2006a: 23, 52).However, DeLanda stresses the distinction of his Deleuzean-Humean model of human agency from a supposed neo-Kantianism holding that all sense impressions are read through existing classifications and, more specifically, a linguistic version in which human thought is always structured by language (he seems to be alluding to the Sapir-Whorf thesis here).This approach is falsified, he suggests, by the fact that experiential association facilitating object recognition and complex 'scene analysis' predated language acquisition historically (and also, we might add with a nod to Piaget, in childhood) (DeLanda, 2011).
In practice, however, DeLanda's model of human agency is neither original nor sound.It ultimately boils down to saying that reasons and motives should be factored into social explanation (DeLanda, 2006a: ch.1), echoing Weber, but only if they appear to make a difference to the operations of assemblages (if we can explain developments in terms of emergent properties, then we ignore them).These motives are based on beliefs and desires, which are the product of socialisation (DeLanda, 2006a: 24-5), and that socialisation is essentially a process of instrumental conditioning.Human beings 'match means to ends', and, to quote: the choice of ends depends entirely on the passions: on the habitual pursuit of those ends associated with pleasurable or positively valued passions [e.g.happiness or pride], and the habitual avoidance of those linked with painful or negatively valued ones [e.g.sadness or shame].(DeLanda, 2006a: 49;cf. DeLanda, 2011, where he explicitly embraces instrumental conditioning) There is no discussion of how else something might derive value other than the pleasure/ pain or happiness/sadness it brings, and DeLanda then cites Deleuze to the effect that this is reducible to 'the principle of utility'. 2All this sounds mightily like utilitarianism backed by behaviourist principles, à la George Homans (1961).DeLanda does factor assemblage norms and 'meta-norms' into the picture as constraints on action, but mainly in terms of punishments and perceptions of punishment attached to deviant actions -thus repeating James Coleman (1990).We started with Deleuze and Hume but, via Weber, end with rational choice theory -a thin, fictional conception of human subjectivity and how it engages with the surrounding world conjured from afar rather than based on sustained in situ observation.To be sure, DeLanda (2006a: 24) does express reservations about rationality as a causal principle, but only because it supposedly isolates mental calculation from the formation of passions (via conditioning) and physical elements of the means-goal relation, which hardly does justice to sociological rational choice theory.Elsewhere, moreover, DeLanda (2011: ch. 8) happily resorts to game theory, with all its assumptions, to conceptualise elementary social relations.

Relevant Assemblages for Sociology
The limited gains of assemblage theory carry over to its purported applications for sociology.DeLanda (2006aDeLanda ( , 2016a) ) works through a sequence of examples of social assemblages, including everyday interactions or conversations, social networks, social movements, organisations/governments and cities and nations.In each case he identifies material and expressive components, forces of (de)territorialisation and (de)coding and so on.Smaller-scale assemblages fit into or become segments in larger-scale assemblages, though DeLanda rejects any notion there are bounded 'societies', which includes discarding Deleuze and Guattari's own concept of the 'social field'.In all cases, however, DeLanda effectively accepts yet redescribes, with little obvious payoff, the work of other sociologists or historians on each of these domains: Goffman on interaction, Coleman on social networks, Charles Tilly on social movements, Weber plus Foucault on organisations and Braudel on cities and nations.Their own explanations, and even Weber's typology of legitimacy, are not questioned but simply considered compatible with assemblage theory.One is reminded of Giddens (1984) translating Paul Willis's (1977) research on educational inequality into the language of structuration theory with no obvious benefit to understanding the topic.
More generally, in fact, Giddens is brought to mind when reading DeLanda's efforts to overcome the micro/macro, individualism/holism or agency/structure divides by reference to constraints and enablements, unintended consequences and the virtuality of assemblages, which mirror the British theorist's language on social structure.DeLanda (2006a) rejects structuration theory, however, on the basis that its notion of the 'duality of structure' -the idea that neither agency nor structure exists separate from one another -posits a seamless (Hegelian) whole, thus excluding relations of exteriority from consideration.This in turn, like the focus on emergence more generally, is reminiscent of Archer's (1995) critique of Giddens' 'conflationism' and her alternative morphogenetic approach, or perhaps Bhaskar's (1998) transformational model of social life, both of which are based on a similar form of realism to DeLanda's own.However, one thing DeLanda does that most of those he accommodates or echoes do not do is to push the issue of materiality.A bureaucratic assemblage is not just a structure or network of power between humans, mediated by symbols, but a system of walls, rooms, corridors, doors, etc., constraining and facilitating what people or other segments can do, including in relation to those walls, doors, etc. (e.g.removing them, making them transparent or open, and so on).A village is not just a mesh of human beings joined by mutual awareness, and related symbols of identity (e.g. the village name or icons), but a complex of specific buildings, roads, paths and so on structuring possibilities for spatiotemporal and symbolic activity.Science qua assemblage is not just a body of practitioners but includes concepts and techniques and a collection of material entities -tools, equipment, laboratories, institutes -co-defining the space of possibilities or diagram.An army is not just a collection of soldiers but an assemblage of soldiers, specific technologies (weapons, tools, etc.), animals (horses, dogs, etc.), techniques (strategies, tactics, etc.), symbols (flags, emblems, etc.) and so on.
Others have brought materiality into sociology or social thought before, from Marx on the forces of production through to Foucault on carceral/surveillance techniques, Giddens on time-space paths and Urry on spatial affordances and mobility systems, and, indeed, DeLanda seems to more or less accept the insights of Foucault and Giddens for assemblage theory with only slight modification -the counterbalance with Weber in Foucault's case, and the counterbalance of deliberate choice (when relevant) against routine in Giddens' case.DeLanda differs, though, insofar as human, symbolic and material segments are treated with parity.Each possesses the same capacity for shaping the possibilities of others.One can look at the way in which humans shape the material world, but also the way in which material entities shape human activity -really, one must do both at the same time.The point is, an assemblage is not merely a 'social structure' but a distinct nexus between all kinds of things, and all those things are treated as capable of affecting its co-constituents in the same ways.Changing military technologies, for instance, such as the development of rifles or portable radio transmitters, have operated to deterritorialise armies and change their capabilities -making them looser, open to greater initiative and flexibility and so on -with profound effects on battles, wars and geopolitical relations.
Ascribing agency to material entities like this is not to 'fetishize' them -to obscure or mystify the relations of production or domination underlying them -as Vandenberghe (2002) worried of ANT.For sure, the posited effects of any material segment have been made possible by a chain of human interdependencies, struggles and interests.The agency of a type of rifle or a specific rifle, for example, implicates the interests of and struggles between weapons manufacturers and their nexus with suppliers, distributors and states.Yet those chains of interdependence can be identified as yet another assemblage, with the rifle's insertion into the different assemblages demonstrating its exteriority, and the rifle nonetheless bears capacities and properties in the military assemblage by virtue of its material form.The background human elements are not denied or suppressed but rather 'bracketed out', to use Giddens' (1984) term, for explicative parsimony.Even the interdependencies lying behind the capacities and properties of a material entity, though, cannot be reduced to human beings but implicate the capacities and properties of other material entities, including, in our example, the properties, capacities and distribution of the raw materials necessary for manufacturing rifles.

The Critique of Bourdieu
Having aired a degree of scepticism about assemblage theory while appreciating the focus on materiality, now we turn to DeLanda's (2006a) critique of Bourdieu.Bourdieu, to remind readers, conceived the social world as comprised of so many fields structured by the distribution of multiple capitals understood as properties and possessions misrecognised as inherently worthy or legitimate.Experiences associated with position and trajectory in a field shape actual and intuited possibilities and probabilities for activity within that field, specifically strategies for accumulating capital -striving for worth -or subverting the field, as well as a sense of what kinds of practices and activities are associated with what sorts of positions.Those with a similar sense of possibilities/probabilities or 'feel for the game' by virtue of similar locations and experiences are said to have the same habitus (Bourdieu, 1977(Bourdieu, , 1990a)).Science, law, religion, art and so on are examples of fields, but the class structure -or 'social space' -is the major field of a social order structured by economic, cultural and social capital (Bourdieu, 1984).Some of the language might seem similar to what we have seen so far: both Bourdieu and DeLanda talk about spaces of possibility, probabilistic causation, things being defined by their relation to others within a system, perceptual association and so on.Yet DeLanda rejects Bourdieu on two counts.

Habitus and Language
First, DeLanda rejects the concept of habitus for two reasons.One of these is the apparent automatism it implies: the lack of attention to conscious activity, intentional choice, articulated reasons and so on.This is an old charge that has been grappled with at length elsewhere, shown to be a partial reading of the concept and avoidable if its phenomenological heritage is emphasised (Atkinson, 2010).The other is Bourdieu's supposed commitment to the neo-Kantian theory of linguistically constituted experience, which is also said to lack a theory of codification -the coding of assemblages -of the kind seen in Tilly's work.This is based on a quotation from Bourdieu (1991: 105) in which he says that neo-Kantianism, which gives 'specifically symbolic efficacy' to language in 'the construction of reality', is 'perfectly justified'.
However, the charge of linguistic reductivism is misplaced. 3Bourdieu was certainly interested in the power of words to construct and shape social reality, as part of his theory of symbolic power, classification struggles and 'group making' (see Bourdieu, 2018).That does not mean that all perception is mediated by language, however.He folded the linguistic element of experience within a broader vision of the structuring of perception according to schemata generated by practical experience.How many times did Bourdieu emphasise that dispositions to act and evaluative schemes exist below the level of verbal articulation?What matters for Bourdieu -and neo-Kantianism generally -is not just language but symbols, as perceptual stimuli representing, alluding to or 'standing in' for something else.Words are merely a special type of symbol, which for the sake of clarity I will call -after Piaget -signs.Both symbols and signs are furnished with their specific meaning through practical experience in the world.Emphasising the phenomenological lineage again, we note the genesis of symbolic meaning -with Husserl (1991) -via active (conscious) or passive (non-conscious) synthesis and its operation via horizons of experience and protention, i.e. pre-predicative association and expectation. 4A phenomenological reading of Piaget furnishes evidence for the pre-linguistic genesis of associative symbolic (and motor/affective) schemes upon which linguistic associations are later layered (Atkinson, 2016).
Phenomenological synthesis is not the same as bare Humean associationism (or behaviourist conditioning), however, because Husserl, and later Schutz, emphasised the formation of mental or cognitive types via typification, i.e. the genesis of perceptual archetypes -which are not necessarily linguistic -for automatically classifying or categorising specific stimuli which may nonetheless vary from and inform the archetype in specific ways.Such is the basis of object recognition and scene analysis.In fact, DeLanda (2011) himself moves in this direction when he acknowledges that simulations of neural networks -models of brain activity -demonstrate the incremental formation of overlapping archetypes from which new stimuli are perceived to vary in differing ways and extents.Integrating this insight with the general relationalism drawn from structuralism (but also Cassirer), Bourdieu would emphasise that any typification or classification is inherently co-defined by that which it is not -by its oppositions and contrasts, in other words, which in Bourdieu's mind were structured by fields.Desire and affect also saturate typification because fields -including the family as a field -are the sources of human worth and purpose via the forms of (mis)recognition they bear (Bourdieu, 1990b(Bourdieu, , 2000a;;cf. Threadgold, 2020).
Finally, it bears stating that Bourdieu did have an explicit theory of codification (see esp.Bourdieu, 1990b: 76-86).It is not exactly the same as DeLanda's meaning, of course, but its purpose was to elucidate instances where specific 'rules of the game' and perceptual constructions of the world which are otherwise practically (non-linguistically) or even affectively/corporeally intuited are given formal, linguistic elaboration in the form of laws, official decrees, written records or formalised theories.This extends to the existence of groups or collectivities, as those invested with symbolic power declare the reality of specific 'groups', 'organisations', 'institutions', 'movements' and so on (or even 'communities', 'cities', etc.), fix their identities and boundaries in the minds of others -thus affecting their practice -and claim to represent, express or speak for them in some way (Bourdieu, 1991(Bourdieu, , 2018)).Incorporating the kind of 'delegation' of interest to Latour (2005) -scientists identifying and speaking in the name of 'species' or 'elements', for example -it becomes possible to extend the same logic to non-human entities too.

Assemblage or/and Social Space
DeLanda's second rejection is of Bourdieu's conception of social structure, particularly his vision of class in terms of social space.This is in keeping with his rejection of Deleuze and Guattari's notion of the social field, which seemingly covers similar ground.The substance of DeLanda's (2006a: 63) critique, moving away from his earlier crude suggestion that classes are strata, is that the distribution of resources -or capitals -does not exist in an abstract space but is only ever embedded in concrete networks of people or organisations.This critique is, in fact, very similar to that raised by social network theorists, such as Becker (2006) and Crossley (2011Crossley ( , 2022)), who prefer the notion of 'world' instead, though DeLanda emphasises the organisational and material aspects to a greater degree -i.e.resource distributions are not only dependent on connections or ties between people but between people, organisations and material entities.One can thus respond in a similar way (cf.Atkinson, 2020): put simply, networks/assemblages and fields/social spaces need not be mutually exclusive.
The utility of the concept of field or social space is to document the basic cartography of possibilities and probabilities in relation to a specific set of stakes: possession of a certain stock of capital shapes what is objectively possible, probable and desirable for people in their struggle for capital and the sense of the possible, probable and desirable, as well as the sense of what is possible or probable for others -grasped (via typification) at varying levels of anonymity and abstraction -in the field.These possibilities and intuitions unite individuals who are far removed from one another in terms of network linkages, generate similar practices and even inform the formation of networks (Lizardo, 2006).This has been empirically verified often enough.That does not mean networks do not exist or are inconsequential.Bourdieu (1996Bourdieu ( , 2000a) ) himself talked about 'legitimation chains' to convey the interpersonal chains of command and delegation through which symbolic power, as control over capital, operates, his major point, echoing Norbert Elias, being that the lengthier the chains become the more impersonal and stable becomes the symbolic power to the perceiver.More generally, though, the distribution of resources does indeed depend on specific networks and circuits of people and activity, and agents form their practical sense and discursive construction of the field in good part -but not exclusively 5 -through interaction with concrete others, meaning those similarly positioned in the field can have different interpretations of the field and their place within it with ramifications for their practice and even variant structures of possibility.As Crossley (2013;Bottero and Crossley, 2011;cf. Lahire, 2012) has noted in his more conciliatory moments, the relationship between fields and networks is best understood in terms of a dialectical interplay, though I would add: with the sense of the game (and one's place) and investment in the game, both oriented toward field structures, taking explanatory priority. 6Worlds can thus be reconceived as cross-field networks and circuits oriented around specific domains.The military world in a nation, for example, would encompass agents across the state (the bureaucratic field) and the military field strictu senso (army personnel) as well as arms and supplies manufacturers (in the economic field), each with their own interests relative to their fields.Some of these fields and circuits will be transnational, others national.There is even evidence to suggest Bourdieu himself sometimes worked with this notion of worlds without naming them as such (Atkinson, 2020).
The same logic can be expanded to assemblages incorporating material entities.Thus we flesh out the human interests and relations underpinning the production, distribution, accumulation, interpretation and uses of specific material entities within discrete assemblages, as Vandenberghe (2002) suggests, but we also acknowledge the effect of material entities within specific assemblages on field structures and possibilities.The military assemblage encompasses the military world just mentioned, but also highlights the (im) possibilities generated by the capacities and properties of associated technologies or nonhuman animals insofar as they are 're-structured' -as Bourdieu would have it (Bourdieu and Wacquant, 1992: 105) -into the logic and strategies of component fields (cf.Dominguez-Rubio and Silva, 2013).More broadly, the capitals defining a social space -especially economic capital and cultural capital -are actualised in material entities (coins, certificates, computer networks) embedded in assemblages of production and distribution, and their durability, size, weight, etc -made possible by the capacities and properties of their own material segments -make some strategies of accumulation or conversion possible and more or less viable both objectively and subjectively.Alchemy, for example, is impossible; forgery has only a chance of success depending on its method; paper money is easier to store and transfer than metal money, but not as easy to store and transfer as digital money (which also opens up possibilities for hacking); and so on.Still, the very drive to accumulate or convert the material manifestations of capital unites all agents within a social order and across numerous assemblages, and the sense of one's place is attuned to intuition, however fuzzy, of the topological distribution of capitals beyond the networks and assemblages in which one is directly embedded.The very production and distribution of materials perceived as capitals are also linked to the state's interest in accumulating or maintaining control over capital within a territory, documented by Bourdieu (2014), which is also, of course, affected by the properties and capacities of those materials.
Part of what is at stake here are what can be called -extending on legitimation chains -circuits of symbolic power, or the movements over time-space and between fields of specific people, symbols and material entities.A material entity -a singular artwork, a rifle or a standardised consumer commodity -may be produced via the struggles and strategies in one field, though also dependent on the properties and capacities of its constituent materials.It is then disseminated over time-space in accordance with the symbolic power of their producers and their connections within worlds, thence having a specific effect on what an agent or set of agents can and cannot do with consequences for their activity within a different field by virtue of its capacities and properties.This may take the form of catalysing specific symbolic revolutions or other major events within fields, or constitute feedback loops for the original field of production, and they may even operate to (de/re)territorialise specific assemblages or underpin codification or 'group-making' efforts.Such is the 'social life of things', to quote Appadurai (1988), but also the material life of the social.

Plurality, Exteriority and Relational Phenomenology
In responding to DeLanda's critique, then, I have discarded his associationist-utilitarianbehaviourist model of human agency in favour of the phenomenological habitus but also, rather than reject the notion of assemblage outright, considered ways in which fields and assemblages may co-exist and affect one another.There are other ways in which DeLanda's vocabulary may enrich field theory, however.One of these, implicit in the foregoing, is the concept of exteriority, which emphasises when the capacities and properties of an entity are irreducible to its insertion within an assemblage -and, we might now add, a field.In a social universe defined by a multitude of fields, the interpretations, uses and capacities of a single material entity vary substantially across fields.What a single rifle makes possible or impossible is very different in the economic field compared to the military field, the armed clash of state armies, and a familial field (where it may be a cherished heirloom, for instance).
The same applies to human beings.Bernard Lahire (e.g.Lahire, 2011Lahire, , 2012Lahire, , 2015) has long criticised Bourdieu for confining individuals to single fields, reducing them to 'beings-in-a-field' and nothing more, thus missing the full complexity and plurality of individual lives lived and dispositions forged across myriad contexts.There is some truth to that charge, or at least, as is now being increasingly recognised, a need to think about how to bring plurality into field theory -how, that is, individuals may be positioned within multiple fields (Atkinson, 2016;Schmitz et al., 2017).The notion of exteriority, then, serves to underscore, on the one hand, that an individual and their capacities, properties and activity are irreducible to a single field, and have some relative autonomy from it, by virtue of their being inserted within other fields too, as well as within networks and assemblages; but also, on the other hand, that what goes on in a field is not entirely reducible to the structure of the field, as 'outside forces', like an individual's family commitments or the material structures and limitations of their village life, play their part. 7 As DeLanda stresses, however, this is a scalar phenomenon: the degree to which an individual is defined by a singular field, and to which they are homogeneous in relation to others in the same field, can vary and surely has varied over history (cf.Lahire, 2011).There may be just one localised field in small-scale undifferentiated social orders, where members are relatively homogeneous, yet multiple field membership in others and great social dispersion within those fields.Bourdieu (e.g. 1990b: 73-4) himself recognised this, in fact, but never made anything of it.And then, of course, there is the isolated individual, like the Brazilian 'Man of the Hole', who appears to be an agent in no field but who still exists within assemblages of matter and whose material environment may still be shaped by circuits of symbolic power originating far away (e.g.state protection of his territory).Individuals and fields, then, are both objects equally amenable to social scientific construction; fields are comprised of individuals whose relations of mutual recognition generate emergent properties -structures, strategies, doxa, etc. -without being entirely reducible to those individuals or those individuals being reducible to it; and each human individual and each field are historically unique entities in the same way as assemblages are. 8Whether or not that makes for a 'flat ontology' is a moot point.However, DeLanda's use of exteriority to reject the Hegelian part-whole logic, which ultimately underpins both the structuralist version of relationalism adopted by Bourdieu and the interactionist relationalism of social network theory (and which I have tried to bring together in a 'unified relationalism'), is problematic.This is because it can always be retorted that the present possibilities of any singular entity are defined by its combination of assemblage insertions; it is, in other words, defined by the totality of its assemblage participations and the relations between those participations.
In the case of human individuals, for instance, they may be irreducible to single fields, but their singularity can still be understood in terms of their specific locations in multiple fields and circuits of symbolic power and the relations between those field/circuit locations in everyday life.Bourdieu (2000b), in fact, seemed to recognise this when he posited a term for grasping an individual's existence -their perceptions and dispositions -across multiple fields: their 'social surface'.This can be interpreted as a kind of 'metahabitus' and, using tools from phenomenology, it becomes possible to conceptualise the relations between fields and circuits within individual lifeworlds.Within any percept, for example, there may be loaded horizonal significance related to any one field in which the individual participates, but also significance related to how activity in one field/circuit bears on another -what becomes (im)possible or (un)desirable in one field by virtue of events in another.More generally, the individual's sense of self -of what matters to them and the hierarchy of those values -is constructed out of the fuzzy sense of plurality and circuit embeddedness, each intuited in relation to one another, whether that means they feel themselves fully integrated, neatly compartmentalised or 'conflicted' and 'internally torn'.And, of course, different fields will structure the time and space of lifeworlds, and streams of consciousness, to different extents, with some (e.g.'greedy fields' related to employment) demanding attention and structuring time-space paths more than others.
Here we can return to DeLanda (2006a), who, in offering a 'charitable' reading of habitus in his endnotes, suggests it might be reconstrued as an individual's 'internal possibility space' provided by their assemblage of beliefs, habits and skills.That is, in fact, a perfectly reasonable reading of the habitus, so long as we specify those beliefs and habits in phenomenological terms as expectations of the forthcoming, co-given with present perception, structured by pre-predicative typifications and soliciting associated activity.Elsewhere it has even been suggested that the habitus constitutes a sort of 'subjective field of possibles', i.e. a set of more or less doable, thinkable and conceivable practices and activities (beyond which lies the impossible, unthinkable, inconceivable or fantastical), attuned to the objective field of possibilities provided by position within a field or social space as it manifests in everyday experience (Atkinson, 2010).Now, however, we add in the structuration of perception by attunement to the possibilities and probabilities of multiple fields, awareness of their interrelation (how events and activity in one field will likely impact upon another, or are relevant to more than one field) -which we might call the 'lifeworld horizon' 9 -and the different strengths of desire for their respective stakes.
The result is a relational conceptualisation of the social surface: it is a (virtual) space of possibilities and probabilities of thought and action structured according to relations within fields but also the relations between fields as defined by spatiotemporal extension within the lifeworld, degrees of mutual interpenetration and strength of desire.It is in constant process, structured by time in the double sense of not just sedimentation of past experience and adjustment to the future but (bounded) flux in relation to the durée of everyday life and the 'situations' it encompasses.This is no metaphysical abstraction; the relational social surface or meta-habitus is an explanatory tool for making sense of observed events and practices -for comprehending why someone did something, or why they reacted the way they did to a situation or experience, whether its consequences are part of an ongoing feedback loop into myriad fields, a (de)coding or (de)territorialisation of an assemblage or 'group' or a catalyst for a thoroughgoing multi-field symbolic revolution.It can only ever be reconstructed by the analyst, but limning the relational structure of the internal possibility space, defined by spatiotemporal attention and intensity of desire directed towards fields, with all their solicitations, should facilitate a more refined sense of the conditions of possibility or likelihood of certain practices and affective states.

Conclusion
My intention was to put field theory and DeLanda's assemblage theory into dialogue.There is no disguising the fact that field theory was the starting point, and its foundational premises regarding human beings -the quest for value, misrecognition and symbol/sign manipulation -the ultimate anchors.An easy reaction to DeLanda's critique of Bourdieu would have been simply to dismiss it and assemblage theory altogether.And sure enough, there is much in DeLanda's theorising that seems to be flawed or unoriginal, especially the default to utilitarianism backed by behaviourism.Yet DeLanda's rendering of assemblage theory does bring fruitful insights to the table which should be taken seriously.It offers an opportunity to push field theory in novel directions, therefore, and develop new tools for the task of explaining social life.
The outcome of the dialogue can be summarised as follows.First, the causal power of 'the material' must be acknowledged; it has its own irreducible effect on fields; there exist distinct assemblages of material and non-material entities; these may even be usefully described as more or less (de)territorialised.Yet that does not preclude the existence of fields as trans-assemblage structures of capital structuring subjectivity and agency.Better to examine the interplay of fields and assemblages, and networks generally, via circuits of symbolic power.The desires and sense of possibilities related to fields may be the primary principle of practice, but they are co-conditioned and modified or individuated by the (sense of) im/possibilities furnished specifically by networks and assemblages.Second, the concept of exteriority gainfully underscores the plurality of positioning, and identity, of material objects and human individuals, all of which can exist in multiple fields, networks and assemblages while retaining relative autonomy from each, though DeLanda's posited escape from relational logic is doubtful.Finally, exteriority qua plurality has consequences for rethinking subjectivity as an internal possibility space, bringing a new layer of relationalism -beyond that provided by adaptation to a single field -to the phenomenology of experience and practice.
It might be supposed that the human/post-human or anthropocentric/ecocentric divides are new oppositions within social theory, opened up by Deleuze and the new materialism, to match the old ones of micro/macro and structure/agency.Perhaps the effort here could be considered a step toward reconciling, overcoming or at least questioning those oppositions insofar as I have tried to make field theory less 'human only' while recognising the specificity of human beings, including their tendency to form fields of mutual recognition and struggle.More important, though, is the possibility of providing new ways of conceiving the relation between material life and human life, and the structures of perception, that may prove useful for sociological explanation -for making sense of why such-and-such event or practice occurred, for example, when reference to the structure and forces of a single field or social space seems not to be enough.

Will Atkinson
https://orcid.org/0000-0002-1032-1241 Notes 1. Thus for each assemblage there may be any number of possibilities, or outcomes, that can feasibly happen, but observation of an assemblage, with all its interactions, determines that some outcomes are more or less likely than others, often statable numerically or statistically (e.g.20% probability, 60% probability, etc.).In this DeLanda explicitly makes common cause with Weber.2. The only other possible motivating force for human activity in DeLanda's Humean account of agency is habit and an associated quest to maintain the stability of personal identity -an idea echoing Giddens' (1984) appeal to 'ontological security' and leading to the same psychological reductivism (Atkinson, 2007).3.One also wonders whether DeLanda's own emphasis on propositional beliefs and reasons as explanations for human action does not also depend on a linguistic conception of experience.
Certainly it seems to unjustifiably exclude or downplay sensation and affect (Clough et al., 2007).4.This phenomenological account of experience is complemented by a neurological account of experience emphasising the formation, strengthening and activation of synaptic connections.5.It may also come, for example, from reading books, journals, magazines, news reports, etc. 6.This point can be pursued and demonstrated in empirical research.See, e.g., De Nooy (2003) and Basov (2022).7.This argument counters Fox and Alldred's (2017) explicit rejection of class as a concept by appeal to exteriority: it is possible to conceive of individual diversity and specificity via multiplicity but retain the notion of class as a practical means of describing relative similarity within one field (the social space), as indeed even they seem to do when referring to the 'class mix' of neighbourhoods (p.68).8.It might be argued that 'field' is a class of entity, thus violating DeLanda's postulate of singularity, but it is no more such than assemblages are.A field's emergent properties can be considered as 'parametrisable' (or scalar) as those of an assemblage; even the degree of struggle and heterogeneity is variable, the limit case of homogeneity being the 'apparatus' or 'corps' (Atkinson, 2020).9. Previously I have suggested Husserl's term 'world horizon' for this, but since that already has a useful meaning (pre-predicative awareness of there being a world beyond immediate perception), changing it to lifeworld horizon has the benefit of retaining that meaning while also being more precise in its coverage, i.e. awareness of the interplay of fields defining one's lifeworld.