ANT at the Mirror: A Reflexive Analysis of the Theory and the Rise of a ‘New’ Action

Introduction

The heuristic principle at the basis of this analysis is well expressed by David Bloor: the requirement of reflexivity, like the one of symmetry, should be considered ‘an obvious requirement in principle, because otherwise sociology would be a standing refutation of its own theories’ (Bloor, 1976, p. 5). The gap between the theory proposed by the Scottish sociologist and the one of Bruno Latour ¹ is as wide as the difference between the two different types of constructivism they profess: social and realist, respectively. Nevertheless, with an adequate translation, the requirement of reflexivity (see also Woolgar, 1988) remains a valid ground for assessing consistency, even for ANT. Following Bloor’s assumption that ‘everything’ is socially constructed, the same theory used to make the claim that everything is socially constructed needs to be socially constructed too. Latour’s point of view is similar: given an ontology that is defined by and defines a theory, the existence of actors that represent an ontological exception could be a counterexample that is more or less significant. However, if the exception is the theory itself, considered as an actor in its own ontology, any difference between what is defined at a more theoretical level (the theory itself) and what happens to the theory as an actor in its own ontology would help clearly identify the limits of ANT. The analysis of ANT, considered face to face with itself, will reflexively characterise the research conducted here. ² In other words, ANT will be regarded as an existing actant with its own specifications in its own ontology. Since the analysis is limited to the above ‘own specifications in its own ontology’, this study may be considered of a reflexive character.

Introduction of a Turning Point in Latour’s Thought

Latour’s philosophy has evolved over the decades, which is why it needs to be broken down in different stages in order to allow a more detailed discussion. There are multiple ‘places’ where a boundary may be drawn, and they reflect the incidental requirements of the type of analysis to be conducted: for example, John Law (2009) chooses as a pivotal year ‘when the actor-network theory achieved recognizable form as a distinctive approach to social theory’. ³ For the purpose of this analysis, the essential requirement is being able to get an insight into ANT in a reflexive way. Since this is not practicable for the most part of modes of existence (hereinafter often referred to as MoE) ⁴ , it has been taken as an ideal turning point. Here the hybridisation of everything (Latour, 1984; 1991; 1999b; 2002; 2010) is accompanied by a ‘regionalisation’ of the ontology, which is more oriented to establishing boundaries rather than crossing them. As a matter of fact, with the introduction of this taxonomy of MoE, ANT loses its ontological status of a potential internal actor in many regions of existence, so that the essential traits that are required for inclusion in this analysis are missing. The only exception is the MoE [RES], a new way to refer to the old ‘network’, which is already investigated in the article.

‘Making Do’ vs ‘Letting Do’ and the ‘Occasionalist Description’

In this paragraph, we will focus on the status of the description and reports of ANT also from a semiotic perspective: ANT appears as an asymmetric actant compared to the other actants that form its objects of study. In addition, we will try to demonstrate how, if the ‘letting go’ (which will prove to be unjustifiable) was truly applicable, Latour’s reports would lose significance for the theoretical criteria of ANT itself.

The ‘making do’ has a semantic area compatible with the construction, ⁵ whereas ‘letting do’ refers to a neutral and inactive spectator. In the idea of ‘letting do’ a lack of interpolations by Latour is implicit, so that he could claim that his reports are able to ‘reflect’ as closely as possible what they describe. It seems to be the way in which he, as mentioned in (Latour & Woolgar, 1979), seeks to ‘come to terms’ with the problem of fallibility of reports and, more generally, of any forms of description ⁶ : making our presence evanescent through the ‘letting do’, so that the gap between report and studied object can be reduced to a minimum, following the extreme idea that there is a ‘correspondence’ between them.

However, before proceeding with a comparison between ‘making do’ and ‘letting do’, it is necessary to investigate the status achieved by the ‘description’ in an occasionalist context of pure immanence, like the one proposed by Latour (see also Mattozzi, 2006). Is it a concept that can enjoy full citizenship in such a philosophy? It might be useful to reflect on Madelaine Akrich’s assumption:

The description of text as a network of relationships, as proposed by semiotics […] produces an effect of symmetry the sociologists of science and technology are particularly fond of: misunderstanding and understanding are on the same level, which means they cannot be unilaterally attributed to a defective reader, nor an esoteric text. They are the product of a relationship established by the text between the author and the reader, which may or may not be actualised; what varies is the gap between the author constructed by the reader in the reading process and the ‘real’ author (or, likewise, the gap between the reader constructed by the author in his text and the ‘real’ reader). ⁷

Translating the above quotation in ontological terms, the fixity of a text is replaced by the kaleidoscopic approach offered by an occasionalist and relational ontology, which is based on the assumption that each actor is occasionally also ‘the author’ of the relationship with any other actor or actant. In other words, moving from semiotics to ontology, it is the distinction between the ‘narrative’ level, on one hand, and the ‘author’ and ‘reader’, on the other hand, to be undermined: it is implied by the very idea of network that ‘authors’ and ‘readers’ are actors to the same extent as the actors of the ‘narrative’. ⁸ In this sense, if from a semiotic point of view extracting a script (description) ⁹ of action is comparable to the transcript of a quote from one text ¹⁰ to another, under the ontological profile of a realist constructivism, mere quotations are not possible: each actor (which corresponds to the ‘quote’ in our semiotic analogy), once it has been related to a new co-text (the network), would no longer be the same; such ‘impermanence’ is achieved for a number of reasons that are exclusively ontological. In fact, moving from semiotics to ontology, both the author’s and reader’s levels are indistinguishable from each other as much as they are from the narrative level (which on the ontological level corresponds to a network of relationships) and all the actants of a network are authors-actors within a ‘narrative’. This shows that semiotics (whose heuristic importance in the formulation of sociology of translation is beyond doubt) and ontology are not completely overlapping in ANT, despite the ‘materialisation’ undergone by the former.

Based on this ontology, it becomes uncertain how the act of describing may imply a concern about correspondence. Can we imagine Latour, after he reassembled the social, wondering if it is actually the social that he has reassembled? In relational occasionalism there can be no social (and no other actor/actant) outside a specific set of relations (network) that determine its existence. Description is nothing other than a new ‘assemblage’ of something.

The occasions when Latour refers to his work as a simple ‘description’ lacking any interpretative grid, which simply follows the actors letting them present themselves for themselves, ¹¹ are quite frequent; here I want to mention a couple of them in particular, as they appear in a number of papers that the author himself, along with Michel Callon, defines in (Latour & Callon, 1992) the ‘ontological manifestos’ of ANT: (Callon, 1986) and (Latour, 1992). Regarding the absence of an ‘interpretative grid’, Latour and Callon refer to the interpretation in terms of the theoretical importance of the observer, who, in an immanent perspective, has no place. As regards ‘letting’ the actors act, Latour seems to confuse the ‘letting do’ with the ‘making do’, which is a key feature of ANT’s acting landscape. ¹² In his own words: ‘we are now interested in mediators making other mediators do things. “Making do” is not the same thing as “causing” or “doing”: there exists at the heart of it a duplication, a dislocation, a translation that modifies at once the whole argument’ (Latour, 2005, p. 217). Latour emerges like Janus, the two-faced Roman god of gates and doorways, who, with the ‘theoretical’ face, claims the ‘making do’ as an ontological feature, ¹³ whereas, with the ‘experimental’ face, he is more oriented to some sort of ‘letting do’.

In Latour’s approach, the ‘letting do’ takes on the features of a sort of arbitrary assumption—it is not a logical consequence of Latour’s theory (unlike the ‘making do’) or a necessity imposed by the experience—attributed to those who perform the action of describing. In addition, the introduction of a ‘letting do’ draws a distinct boundary between cognition (where a description is still a subset), equally ‘distributed’ as action amongst the actors of a network (Latour, 1986), and the description, which seems to be a prerogative of the actors who ‘let do’, as if there were still any facts to be faithfully captured in their neutrality, rather than ‘factishes’ to reassemble.

On the other hand, it is incorrect to draw such a sharp boundary between ‘action’ and ‘description’ in ANT, considered as a theory (regardless of reflexivity). In fact, Pasteur’s description of his experiment on fermentation belongs to a chain of ‘making do’ (Latour, 1999a, pp. 113–144), and this is also true for the Cartesian planes used to describe the motion of a point on a surface, for an equation describing a hyperbola, etc. as well as for the so-called laws of nature. Consider, for example, the first chapter of (Latour, 1999a), where the French author asserts: ‘Yes, scientists master the world, but only if the world comes to them in the form of two-dimensional, combinable inscriptions’ (Latour, 1999a, p. 29). The representation of two-dimensional inscription produced by the team of botanists, biologists and geologists that Latour follows all the way to Boa Vista can be found on page 57 of Latour (1999a, Fig. 2.15); it consists of a detailed description ¹⁴ of the soil samples investigated by the research team. It is precisely the ‘making do’ of all actors/actants involved that gives value to that description and to Latour’s activity, thanks to the transport of reference and the unfolding of an action that Latour can follow. In this sense, that description did not come up through a ‘letting do’ and this also applies to the photographs and maps of the same portion of soil that preceded it and served to prepare it. If all those inscriptions were guided by a ‘letting do’, they would not exert any translation and transcription action, suddenly losing value, meaning and reference.

The introduction of a ‘letting do’, motivated by the need to ensure a likelihood of reports with respect to what they describe, under the lens of a reflexive analysis, paradoxically leads to the opposite conclusion, in the sense that descriptions must be inextricably connected with a ‘making do’, if they claim to have a value as a result of their relation with an actual reference; removing the processes of ‘making do’ (turned into a ‘letting do’) involves blocking the course of reference and its translations, leaving Latour’s report without it.

Thinking about the principle of symmetry, Latour appears to be asymmetrical in respect to his objects of study.

Stuck in the Network

In this paragraph, we will show how ANT is not even affected by the ‘making do’ in the different networks it becomes part of. The asymmetry and singularity of ANT compared to its objects of study are even deeper.

‘“But, […] did ferments exist before Pasteur made them up?” There is no avoiding to answer: “No, they did not exist before he came along”—an answer that is obvious, natural and even commonsensical!’ (Latour, 1999a, p. 145). The ferment, in its relative existence determined by the set of relations between the actors of the collective that created it (the collective of Pasteur and his theory of fermentation) could not exist before the existence of that very collective (e.g., not as a phenomenon of vitalistic origin ¹⁵ ), and this is perfectly in line with Latour’s immanentism. Before then, however, ‘other’ ferments had their own existence, affected by the associations embedding them, for example with the processes of decomposition in one case, or with the spontaneous generation of Pouchet in another. The mutual gap between the two processes, as well as the ‘Pasteurian’ ferment, certainly does not lie in any difference inscribed in some mysterious essence, but in the actors and actants that are mutually articulated with them, the same that ¹⁶ ‘make do’ and are ‘made do’: ‘Latour gives us not just a metaphysics of actors, but of actors that come to birth only on the occasion of their associations; since these associations shift constantly in both tiny and revolutionary ways, we have actors that perpetually perish rather than endure’ (Harman, 2009, p. 80). Graham refers to the concept of ‘being-as-other’, which also in this article is considered a genuine metaphysical principle.

Considering this metaphysics, the term ontologisation rather than ontology should be preferred, since the former conveys that dynamic, active and changeable sense inherent to the actors of ANT that is missing in the latter. Ontologisation also conveys a sense of the kaleidoscopic immanent existence that characterises every actor and actant. ‘Each element is to be defined by its associations and is an event created at the occasion of each of those associations [italics added]’ (Latour, 1999a, p. 165). With the phrase ‘each element’, Latour refers to any actor and actant ¹⁷ that, continuing with the exemplification, can be referred to the new theory of the new Pasteur and to the new being, the ‘vitalistic ferment’. Since any change in the associations that make up a collective brought out a different collective, in the network deployed by Liebig, fermentation is a ‘residue’ from a chemical process, an actor that is clearly different from the vitalistic phenomenon created by the relationships in Pasteur’s network.

Reflexively, if the ferment consists of a different reality as an actant in multiple collectives, since it is associated with entities that differ from time to time—Annamarie Mol (Mol, 2002) talks about ‘different’ but ‘coexisting’ realities—the ANT actor, while assembling the numerous networks of associations where it belonged to, and performing the relevant reconstruction of ‘its’ case studies, should be at least as diverse as the networks where it appears. Better yet, it should not be the same actor, just like the ferments of the different laboratories of Pasteur, Liebig or Berzelius. Instead, in Pandora’s Hope and Reassembling the Social, just to name a few, Latour spends 300 pages to explain his theory, drawing on the multiple case studies to which he applied it. ¹⁸ In other words, from a huge number of collectives, and consequently from a kaleidoscope of relational diversity amongst actors, the result is a unique and irremovable actor ¹⁹ to the extent that the action exerted by the entities it investigates does not affect ANT itself. This is the counterpart of the ‘letting do’ mentioned in the previous paragraph: it appears that the ‘making do’ does affect ANT, neither actively nor passively. The asymmetry and the singularity of ANT are even deeper.

Immobility Persists: The Historicity of Things

In this paragraph we will see how the lack of action (in the form of ‘making do’ and ‘made do’) is not only ‘reflexively’ related to ANT.

Given the collective of a historical case ²⁰ (e.g., the one of Pasteur and the ferment), and assigning the letters ABCD to time₁ to indicate the relational chain between actors, when adding a further actor E, for example ANT (which, as a result of its analysis, produces a new narrative of the historical case), following Latour’s approach in Pandora’s Hope the above sequence would become ABCDE at time₂. Mutatis mutandis, another collective, for example GTRQ, would become GTRQE. But if a given actor is E before diving into new relationships and remains E even at a later stage, it turns, without any action, into a sort of primitive (Goodman, 1978) independent from them. It would enjoy a special existence, independent from the contingent relations in which it is inserted, and would constitute an ontological exception to that ontologisation inherent to each actor, which was mentioned in the third paragraph. Alternatively, since it does not act nor is acted, simply it would not exist in an immanent and relational collective.

As argued by Latour, if the addition or subtraction of an actor produces a difference, and ‘differences are all that we need’, the relationship between actors cannot be simply additive, even as regards the sequence named ABCD (or GTRQ) in the previous example, because the ‘addends’ themselves, thrown in new relationships, will no more be the same: there is a change in the relative existence of the actants entering different collectives. As remarked by Latour: ‘action is all that we require’, that is, the action that each actor exerts on those with which it comes into association and by which is equally acted. In terms of material semiotics, the actors that do not exert such an action are not even to be considered as existing, making existence the result of the relational influences deployed within a network: ‘there is no other way to define an actor but through its action, and there is no other way to define an action but by asking what other actors are modified, transformed, perturbed, or created by the character that is the focus of attention’ (Latour, 1999a, p. 122).

Here again, instead of action we find a lack of it. Harman seems to get straight to the point when he remarks: ‘for Latour, an actor is defined by its current alliances—but this does not mean that it has no problem entering new ones!’ (Harman, 2009, p. 114).

ANT and Relativity

In this paragraph, drawing a parallel between physical (Einstein, 1905; 1949) and sociological relativity, we may find a potential reason for the immobility of ANT discussed in the previous paragraphs: an action, so far gone unnoticed, that is connected with the metaphysics of ANT. We will also introduce a limit to the applicability of the sociological relativity suggested by Latour.

‘We are no more relativist than Einstein, and for the same reasons’ (Latour, 1988, p. 26). The underlying aims of Latour in (Latour, 1988) is to reconfigure Einstein’s argument (Einstein, 1920) through a semiotic analysis of the text, in such a way as to introduce in social sciences an equally clear distinction between relativism and relativity. In this sense, it is driven by the idea that, introducing a number of material networks, it is possible to give a unitary meaning to the observations in sociology that relativism would then brutalise. In the essay, a number of ‘superimpositions’ can be investigated (following the distinction drawn by Latour), where the movement (shifting) in and out of the narrator or the reference with respect to the author is compared with the changes in the reference system (i.e., point of view, hereinafter RS) from which any observations can be drawn. ‘Away from the work of inscriptions, subscriptions and transcriptions, no shifting in and out would be possible. We would be limited to a point of view (Latour, 1988, p. 31), that is, we would be limited to observations (or narrative perspectives for the part of analogy where he refers to the so-called ‘semiotic turn’) in which the hic et nunc where they occur would become an unsurpassable threshold. Basically, the equivalence of the physical observations made by heterogeneous RS(s), despite the differences in the measurements made within the relevant RS (e.g., contraction of spatial relations and dilatation of temporal relations), thanks to the invariance guaranteed by Lorentz transformations (Lorentz, 1886; 1892a; 1892b; 1899) and its processing through semiotic analogy, may find its equivalent in ANT. To the same extent that making commensurable the measurements made by different RS(s) may return a unitary meaning to the physical observations, so the ability ‘to switch’ from one sociological description to another without ‘losing’ the reference makes the observations of social scientist significant as well. For the latter, networks and ‘immutable mobiles’ ²¹ become the guarantors of the potential displacements without losing the reference:

It is to accommodate many examples of such a problem that I have proposed considering history of science as the history of centres which are growing through the management of traces that have three main characteristics: they are as mobile, as immutable and faithful, and as combinable as possible. The circulation back and forth of these ‘immutable mobiles’ traces networks—that is to say, two-way paths leading from the centre to the now-dominated frames. (Latour, 1988, p. 21)

This means that it is possible to collect and overlap traces of the actors thanks to the ability to move between ‘now dominated’ systems (now that sociology has become relativistic too); on the other hand, sociology is ‘dominating’ its frames thanks to traces that allow moving from one frame to another. Thanks to the circulation of reference along the chains of actors that translate it, the traces of this reference persist in the form of a relationship that remains stable while scrolling from one actor to another. Just like this immutability is preserved between a description or translation and its reference (the ‘frame’ and the ‘centre’ mentioned by Latour at the end of the previous quote), the same happens in the ‘displacement’ between different descriptions (or translations) of ‘the same’ reference (including the descriptions of the ‘same’ actor) that have overlapped in the flow of the history of science: this is the core element of sociological relativity. The shift from relativism to relativity is all about the rejection of the interpretive closure of a single description, primarily by unlocking interpretation from the core of a subjectivity that is never completely understandable and communicable, and secondly by ‘distributing’ it amongst all the actors of the network, thus making the different perspectives commensurable, because the reference is still traceable (actor by actor) against the plurality of the possible descriptions (or translations). In this sense, going back to the comparison with Einstein’s relativity, ANT does not coincide with a specific RS (i.e., a simple observation point), but it coincides with the theory of relativity tout court (with the relativistic world) on one hand, and with Lorentz transformations on the other hand (i.e., the means used by the theory for such ‘displacements’ from one frame to another). The metaphysics of ANT, just like any metaphysics, defines the world and its elements, ²² whereas its method provides the tools that are needed to move in and out and activate the chains of translation, as well as the immutable mobile. In other words, beyond its specific metaphysics there would be no acting reality (i.e., actors that are affected by the circulating action in a network) to be translated, and in the absence of its method, it would lack the tools to enact any translation and to relate with any actor.

Just like in physics one may consider equivalent the heterogeneous measurements coming from different RS, in the framework of a relativistic world and through the application of its own method (e.g., Lorentz transformations), so in sociology one can keep the reference unchanged against the multiple points of observation, in the framework of Latour’s world and through the application of his methods. ²³ Therefore, in order to enable the relativity of point of view in sociology (i.e., making possible to ‘travel from one frame of reference to the next, from one standpoint to the next’), ²⁴ it is necessary that these points of observation are integrated into Latour’s world and share the underlying metaphysics with it.

Just like Relativity would not be without a relativistic world and approach (Lakatos, 1978; Zahar, 1989), so sociological relativity would not be without the being-as-other of the ANT. The macroscopic difference lies in the fact that the former is not in the reflexive condition of appearing as an actant in its own ontology. In physical relativity, in view of the enormous freedom of movement of the observer, there is something that must always remain fixed (the relativistic world and Lorentz transformations, in the case of restricted relativity) in order to talk about relativity rather than relativism; in sociological relativity it is all about ANT and its metaphysics, so that when ANT circulates as an actant in its own ontology, it is subject to the ‘consolidating’ action of this internal instance.

Acknowledging the action exerted by the more general metaphysical level, it also becomes clear the reason of the paralysis incurred by ANT. A necessary condition in order for sociological relativity to be possible is that actors should be in function of one another (‘being-as-other’), hence the possibility of the ‘chains of translation’ and ‘circulating references’ (in the ‘old world’ of ‘beings in themselves’, this would be totally inconceivable). This means that it would be necessary to move internally in ANT’s metaphysics. Since we live in a ‘world’ (ANT) where everything is determined by the action exerted by related actors, this metaphysics must have to do with some kind of action: it is ANT that, after entering a network as an actor, makes metaphysics circulate. This circulation has a recursive form, that is, it goes through all the actors and actants of a given network and then returns to ANT, placing the entire network in the metaphysical conditions suitable for sociological relativity. If ANT was subject to the formatting action of the actors and actants of the network and was not able to make such action circulate, the new formatting would not ensure a metaphysical ground (which ontologisation and methods depend on) that is suitable for sociological relativity. Therefore, this type of action does not appear in MoE, because they are more connected with classification, rather than ontologisation. In the conclusions we will take a closer look at the flowing of this type of action.

If, as for the existence of actors, the most appropriate term to refer to this polymorphism is ‘ontologisation’, for the actor ‘ANT’, the term ‘ontological stillness’ would be more appropriate. Thankfully, the empirical-cognitive operations where it becomes necessary to find a connection of consistency between the theories as a whole and the theories considered individually as part of their own ontology are quite rare (in natural sciences, e.g., they are totally missing). Nevertheless, in cases where such a connection is required, as remarked by Bloor, complying with it becomes a compelling necessity for a theory not to be considered the constant refutation of itself. From these findings, ANT as an actor must be considered the refutation of what it describes as a theory in order to exist and in order for sociological relativity to exist. Then, should we just throw ANT away? In the conclusions of this article, we will see how that strangeness can be explained with a new kind of action.

Conclusions and Beginning of a ‘New’ Action

In paragraph one (after explaining the general structure of this research), a turning point in Latour’s thought has been defined and discussed based on the purpose of this article.

In the second paragraph we argued how the ‘letting do’ advocated by Latour is contradictory within the theory and it makes ANT an asymmetrical actant compared to the other actants. In relation to it, we argued that the idea of drawing pure descriptions in the context of ANT is little more than a fantasy, a kind of fantasy undermining the value of the whole theory by discrediting the function of the inscriptions and possible translations. In this sense, this research is oriented towards a reconstruction (not just a description) of ANT, where the flow of action—including the one of the writers—requires an internal circulation (due to the peculiar reflexive approach) within ANT itself that should not be interrupted.

In the third paragraph we argued how the ANT is immune to the circulation of formatting action in the networks where it has belonged. The lack of the ‘making do’ action exerted by ANT (Par. 2) has found its counterpart: the lack of the ‘making do’ action ANT should be subject to. We have also seen how ANT suffers this lack, not only from a reflexive point of view (Par. 4). Following these observations, there is an increasing need to identify some evaluation criteria for actors, which cross the network where they appear and are capable of justifying their permanence in the same form. Given the ontologisation promoted by ANT, would such a request make sense? On the base of the present analysis, it seems the answer is ‘Yes, as long as the permanence is caused by some traceable action’.

The fifth paragraph shows how the very idea of sociological relativity implies an internal layer within Latour’s metaphysics that undermines the hope of having turned sociology into a globally relativistic discipline. On the contrary, the intertranslatability between different perspectives is safeguarded internally to ANT. In the same paragraph we assumed that there is a type of action connected to the metaphysics of ANT, which can explain why the ‘making do’ exerted by the other actors on ANT does not produce different results across multiple networks.

In an immanent and relational philosophy, the changing of the actors when new relationships arise is the norm; in this sense, ANT as an actor contradicts what it states as a theory, leaving itself open to the consequence described by David Bloor in the formulation of the principle of reflexivity. On the other hand, in any theory, ‘anomalies’ are typically ignored until it is no longer possible to underestimate them. The fact that ANT looks like an anomaly to its own eyes is perhaps to be interpreted as reaching this limit. We suggested a type of action that has so far been ignored, which can explain the ‘stillness’ observed in ANT and can make sociological relativity practicable. For convenience, we will call this type of action metaphysical action. We have not returned to a world of interpretations that are ‘enclosed’ into the mind of subjectivities in front of which objects appear passively. Just because a decision in the mind of a subject is not enough in order for the metaphysical action to circulate, the actors and actants of a network should allow this action to circulate (which is also in line with Latour’s thought) and return (recursiveness) to where it started (ANT, in this case). Why are we introducing the concept of ‘allowing (the action) to circulate’ and ‘recursiveness’? The answer is simple: because the world of ANT is a kaleidoscope of actions. The recursive coming back of ANT’s metaphysical action, passing through the actors and actants that ‘allow the action to circulate’, creates (in the same sense intended by Latour) the unitary and democratic ground that is necessary for sociological relativity. As seen in paragraph four, in order for a relativity to be possible, there must be something which is not subject to change and can maintain unity across different RS(s). By circulating, metaphysical action constructs a network, but only after the real actors and actants have ‘allowed the action to circulate’: here we find again, though in a different form, constructivist realism. Whether they ‘allow (or do not allow) the action to circulate’ is a fundamental active component of every actor/actant. Amongst them, the ones that will make the action circulate will become part of that specific network, whereas the others will be excluded: a metaphysical action that is not capable of enrolling a suitable network is bound to fail. In this respect, the ‘stillness’ observed in ANT demonstrates the existence and the success of such action.

The metaphysical action not only responds to the need to explain some of the bizarre behaviours of ANT, but it is also more consistent with the concept of ontologisation that has been introduced as regards the ordinary way of understanding metaphysics (especially the one disseminated by neo-positivism). Latour in (2005, p. 50, 117) says: ‘to go from metaphysics to ontology is to raise the question of what the real world is really like’, since ‘metaphysics […] purports to define the basic structure of the world’ and ‘ontology is the same thing as metaphysics to which the question of true and unification have been added’. Using the concepts introduced here: ‘the metaphysical action purports to define the basic structure of the network as much as its unification, and the ontologisation adds the question of what really these actors and actants will become’. Now it is clear that metaphysics acts, and it is not something ‘enclosed’ in the head of an individual subject. As much as the ontologisation, it has to do with the pragmatic and immanent encounters between actors and actants.

The adjective ‘metaphysical’ is used to refer to this new type of action, not so much for its origin (the metaphysics of ANT) but rather for its function underlying the ontological level: to create a network where a contingent ontologisation of actors and actants can take place. In this sense, we can also find a sign of this action in the birth of ‘Pasteur’ ferment. In the words of Pasteur himself:

Dans tout le cours de ce mémoire j’ai raisonné dans l’hypothèse que la nouvelle levure est organisée, que c’est un être vivant et que son action chimique sur le sucre est corrélative de son développement et de son organisation. Si l’on venait me dire que dans ces conclusions je vais au delà des faits, je répondrais que cela est vrai, en ce sens que je me place franchement dans un ordre d’idées qui, pour parler rigoureusement, ne peuvent être irréfutablement démontrées. (Pasteur, 1858, 418, italics added)

Even though the terminology is the one of Modernity, we can translate it as follows: Pasteur network was assembled by the metaphysical action of vitalism, which the enrolled actors and actants allowed to circulate, primarily those linked with the experimentation. But in his Mémoire, Pasteur describes in detail the ferment that he observes under the microscope, including the moments when it feeds on the substances present at the time of fermentation. Therefore, why does he claim to go beyond facts that are not irrefutably provable? Because what allows the ontologisation of the living ferment cannot yet go irrefutably beyond the network where the actors actively allowed the metaphysical action of vitalism to circulate.

Based on the findings in these pages, I hope that the cataloguing of the possible epistemic modes of actors will not supplant future investigations on the types of action that are involved in the ontological and metaphysical determination of actors. I hope that this work will contribute to drive the research in this direction.