Foucault’s naturalism: The importance of scientific epistemology for the genealogical method

Abstract

This article offers a novel reconstruction of Foucault’s methodology that emphasises his respect for the natural sciences. Foucault’s work has long been suspected of reducing knowledge to power, and thus collapsing into unconstrained relativism and methodological incoherence. These concerns are predicated on a misunderstanding of Foucault’s overall approach, which takes the form of a historico-critical project rather than a normative epistemology. However, Foucault does sometimes make normative epistemological judgements, especially about the human sciences. Furthermore, there are outstanding questions about what secures the descriptive rigour of the genealogical method. To address these issues, I develop two claims that will significantly enrich our understanding of Foucault’s methodology. The first is that Foucault’s respect for the natural sciences is crucial in making sense of his normative epistemological judgements. The second is that the descriptive rigour of his genealogical method derives from the fact it is modelled on empirical inquiry.

Introduction

This article offers a novel reconstruction of Foucault’s epistemology that emphasises his respect for the natural sciences. Foucault’s work has long been suspected of reducing knowledge to power, and thus collapsing into unconstrained relativism and methodological incoherence. These concerns are predicated on a misunderstanding of Foucault’s work, which takes the form of a historico-critical project rather than a normative epistemology. Nonetheless, Foucault does sometimes make normative epistemological judgements, and it is unclear how these truth claims fit within this broader project. Furthermore, there are still outstanding questions about what secures the descriptive rigour of the genealogical method.

In response, I develop two claims that will significantly enrich our understanding of Foucault’s methodology. The first is that Foucault’s respect for the natural sciences is crucial in making sense of his normative epistemological judgements. While Foucault’s work is primarily oriented around a historical critique of knowledge rather than a normative epistemology, he does occasionally pass judgment on the epistemic credibility of different bodies of knowledge. I suggest that these judgements are predicated on a respect for scientific epistemology that must be situated within his broader historico-critical project. The second claim is that the descriptive rigour of his genealogical method derives from the fact it is modelled on empirical inquiry. On this basis, I argue that Foucault can appeal to evaluative norms and standards that have been formed under the pressure of power relations, but that are still epistemologically robust by virtue of their accountability to empirical evidence. I will support these claims by offering a close analysis of his engagement with scientific epistemology and contextualising his work within the French tradition of history and philosophy of the sciences, which was central to his intellectual development.

Motivating the discussion: Historical critique, normative epistemology, and the descriptive rigour of genealogy

In this section, I will demonstrate some of the interpretive difficulties that emerge from Foucault’s rejection of normative epistemology in favour of historical critique. Indeed, he does not offer a general theory that discriminates between the truth and falsity of statements. This lack of a normative epistemology led many of his early critics to worry that (a) he is an epistemic relativist who cannot reflexively account for the historicity of his own inquiry; and (b) the method of genealogy relies on a practice of interpretation that it is incapable of justifying. This line of criticism was most famously articulated by figures like Descombes, Habermas, Rorty, and Taylor (Descombes [1979] 1980; Habermas [1985] 1987; Rorty 1986; Taylor 1986; Bouveresse 2016). These concerns led such critics to claim that Foucault reduces truth to power despite his protestations to the contrary (FL 462; EW1 296, 298). If successful, this objection would leave Foucault vulnerable to accusations of epistemic nihilism and would appear to undermine the overall methodological coherence of his work.

However, more recent contributions to the scholarship have shown how this objection is predicated on a fundamental misunderstanding of Foucault’s aims. There has been an important interpretative trend that identifies the properly philosophical dimension of Foucault’s work as a historicisation of Kant’s critical project (Han [1998] 2002; A. Allen 2008, 24–29; Djaballah 2008, 198–199; Koopman 2013; Tiisala 2015; Han-Pile 2016). Indeed, Foucault understood his work as offering an inquiry into the historical conditions of possibility for knowledge and experience while eschewing the transcendental modalities of necessity, universality, and apriority (AK 203; PT 23–82; EW1 315–316). Accordingly, several critics have highlighted that his genealogical method does not pursue traditional epistemological questions but instead offers an inquiry into the historical depth conditions for knowledge, power, and subjectivity (Hacking 2002, 4; B. Allen 2010; Koopman 2013, 33). This line of argument has been highly influential and continues to yield powerful insights into Foucault’s work. Most recently, Lorenzini has offered an important defence of Foucault’s critical project that demonstrates how his historical problematisation of truth is based on a fundamental respect for the value of truth and truth-telling (Lorenzini 2023, 119–123). Lorenzini rightly argues that ‘Foucault’s rejection of “the Truth” as a timeless and suprahistorical concept does not amount to a rejection of truth altogether’ (Lorenzini 2023, 121). For Lorenzini, Foucault’s critical project is animated by an ethico-political commitment to truth-telling that recognises the immanence of truth to history and power, and thus avoids ‘reducing the question of truth to a purely logical or epistemological question’ (Lorenzini 2023, 7–9). Foucault’s respect for the truth is also supported by his claim that power does not inherently distort knowledge, but instead plays a productive and causal role in its constitution (DP 27; PK 131).

In a similar vein, Lorenzini and Tiisala have highlighted how the notion of ‘truth-telling’ plays two fundamentally different roles in Foucault’s conception of critique (Lorenzini and Tiisala 2023). The first is truth-telling as ‘avowal’, which refers to the confessional practices that often serve as the target of Foucault’s genealogies (Lorenzini and Tiisala 2023, 8). As Lorenzini and Tiisala argue, ‘Foucault’s genealogy of avowal aims to question the longstanding myth that where there is power, there cannot be truth, and that (telling the) truth will always set one free’ (Lorenzini and Tiisala 2023, 8). The second is truth-telling as ‘parrhesia’, which is an ideal of courageous speech that they argue serves as an important ‘method’ of critique in Foucault’s work (Lorenzini and Tiisala 2023, 10). For Lorenzini and Tiisala, this use of parrhesia helps to illustrate the second function of critique, which is ‘the possibility of questioning power about its discourses of truth’ (Lorenzini and Tiisala 2023, 10). In short, these scholarly contributions help to show how Foucault’s respect for the importance of truth is not only intelligible within the context of his historico-critical project, but also plays an important role in motivating it.

Despite the importance of these contributions, there are still outstanding questions concerning the way in which Foucault discriminates between competing historical conceptions of truth. Although Lorenzini is right to emphasise that Foucault avoids committing to any general theory of truth, I argue that Foucault does rely on historical standards of truth in formulating normative epistemological judgements (Lorenzini 2023, 7). Indeed, while Foucault’s approach to epistemology is mainly framed in terms of a historicist critique, he does sometimes pass judgment on the epistemic credibility of some bodies of knowledge, especially those he identifies with the human sciences. In other words, while he does not offer a normative epistemology, he does sometimes make normative epistemological judgements – that is to say, he makes claims about whether specific bodies of knowledge ought to be accepted as true. Furthermore, it is not immediately clear how these judgements fit within his broader historico-critical project. Although Lorenzini does broadly account for Foucault’s immanent reliance on regimes of truth, he overlooks Foucault’s particular respect for the natural sciences in guiding many of his epistemological judgements (Lorenzini 2023, 22–23).

This is one of the essential concerns motivating this article. On my view, Foucault appeals to epistemic standards and norms that have been derived from the natural sciences. More specifically, I argue that Foucault is attracted to the high degree of epistemological rigour that derives from their empirical accountability and self-correcting orientation. However, I will also emphasise that this appeal to scientific reason must be understood as part of his historical critique of reason more broadly. Therefore, a large part of this article will be dedicated to explaining how Foucault’s naturalism emerges within and complements his historico-critical project. It is worth noting that Foucault includes formal sciences like mathematics within his broader conception of science (FL 44; AK 188–190). However, my discussion will focus on his relationship to the natural sciences, as their empirical basis is more directly relevant both to his critique of the human sciences as well as his articulation of his own methodology.

There is also an outstanding question concerning the descriptive rigour of the genealogical method. While the recent scholarly focus on Foucault’s mode of historical critique has been highly fruitful, there is still more to be said about the basis of the validity claims advanced by genealogy. After all, while the genealogical method does not adjudicate these kinds of claims as such, it does attempt to offer valid descriptions of history. For this reason, one might wonder whether Foucault’s lack of a normative epistemology means that the genealogist cannot help but reduce truth to power. At the very least, Foucault’s acknowledgement that his mode of historical description involves some element of interpretation might lead us to wonder what epistemic constraints are imposed on this process. While Foucault does claim that ‘the genealogical mood will be that of a happy positivism’, it is not obvious how he can accommodate a positivistic approach within his broader critical project or how this fits with his sustained criticism of traditional positivism (OD 73).

This concern has been most effectively articulated by Descombes, who contends that Foucault’s epistemology is a ‘surprising mixture’ of ‘positivism and nihilism’ (Descombes [1979] 1980, 117; cf. Habermas [1985] 1987, 276–278). In this way, he claims that there are two contradictory strains in Foucault’s thought. On the one hand, he rightly points out that Foucault identifies his historical method as essentially positivistic, drawing a great deal of inspiration from the tradition of French historical epistemology as well as the broader movement of French positivism (Descombes [1979] 1980, 110–111). As Descombes correctly observes, the empirical basis of Foucault’s work is precisely what allows him to claim that he relies on a sophisticated form of positivism focused on empirical results rather than the construction of a grand metaphysical theory (Descombes [1979] 1980, 116). On the other hand, Descombes highlights that Foucault is also committed to the Nietzschean belief that knowledge is always constructed on the basis of situated interests, and that matters of fact can only be established on the basis of more fundamental interpretations of the world (Descombes [1979] 1980, 116).

According to Descombes, this leaves Foucault in the unenviable position of claiming that his method is rigorously positivistic and grounded in the materiality of the archive while also tacitly conceding that what we make of archival material is ultimately a matter of interpretation (Descombes [1979] 1980, 117). While Foucault tries to preserve something like the ‘the positivist notion of fact’, Descombes thinks he must logically be committed to the view that ‘facts’ are not determined by their correspondence to reality but are instead the product of a conflict between competing human interests (Descombes [1979] 1980, 116–117). In this way, Descombes thinks that Foucault fails to resolve the more fundamental issue of how we can discriminate between competing interpretations of the world, and thus claims that genealogy is susceptible to the very kind of unconstrained relativism that Foucault tries to avoid by grounding his insights in the materiality of the archive (Descombes [1979] 1980, 117; cf. Habermas [1985] 1987, 276–278). In the end, then, Descombes thinks that Foucault’s positivism is undermined by his supposed nihilism, resulting in the collapse of genealogy as a rigorous mode of historical description.

In response, my strategy will be to reinterpret Foucault’s positivism through his approach to historical critique. On my view, Foucault certainly does model his genealogical method on scientific inquiry insofar as he adopts an empirically accountable, fallibilist, historically reflexive, and self-correcting approach to analysing different forms of knowledge. This is crucial in explaining how he can safeguard his genealogies against the perennial threat of epistemic distortion by pernicious power relations. However, I maintain that Foucault’s genealogical positivism is predicated on a more fundamental mode of interpretation that breaks with the givenism of traditional positivism. In other words, the challenge is to show that Foucault’s positivism coherently fits within his broader approach to historical critique. This will help to demonstrate how he can secure the descriptive rigour of the genealogical method.

Foucault’s ‘Discontinuist Historical Ontology’: Clarifying the foundations of genealogical critique

In this section, I will clarify Foucault’s fundamental ontological and epistemological commitments to explain how his mode of historical critique can sustain normative epistemological judgements. As we have already seen, several critics have argued that Foucault reduces all normative epistemological standards to relations of power. While Foucault’s approach to historical critique is not addressed to questions of normative epistemology, there are moments in his work where he does make normative epistemological judgements – especially concerning the epistemological rigour of the human sciences. Therefore, it is necessary to contextualise these epistemological judgements within his historico-critical project. Accordingly, the main purpose of this section is to clarify his overall approach to questions of ontology and epistemology. While Foucault’s interest in ontology and epistemology is evident throughout his oeuvre, he does not systematically defend his views on these topics until the beginning of his genealogical period with texts like The Order of Discourse, Lectures on the Will to Know, ‘Nietzsche, Genealogy, History’ and ‘Truth and Juridical Forms’.

In earlier texts, Foucault sometimes appears to reach for the metaphysical concept of ‘reality’ to elucidate his views about the historical construction of knowledge. In his 1961 preface to The History of Madness, Foucault implies that there might be a more pure or authentic experience of madness that exists in the silence of modern discourse (HM xxvii, xxxii). Indeed, he speaks of a ‘madness’ that has been held ‘captive’ and ‘whose wild state can never be reconstituted’ in the absence of its ‘inaccessible primitive purity’ (HM xxxiii). As Hacking rightly observes in his foreword to the English translation of the book, Foucault would eventually come to repudiate this early view of madness as something that exists ‘in the wild, as something prediscursive, inaccessible, pure’ as well as the associated idea of an ‘archaeology of silence’, and instead comes to redefine archaeology as the study of material statements (Hacking 2006, xii; HM xxvii; AK 125). ¹ Similarly, in The Birth of the Clinic, Foucault claims that there is an ‘excess of the signifier over the signified’ (BC xviii). This is a view he again echoes in The Order of Discourse, when he claims that ‘we must conceive discourse as a violence which we do to things’ (OD 67; cf. FS 173).

These comments lead Kelly to suggest that Foucault embraces a ‘discontinuist epistemology’ that treats ‘the relationship of language and reality in terms of fundamental and irreparable discontinuity’ (Kelly 2019, 325). On Kelly’s view, this means that all validity claims and epistemes are underdetermined, and therefore exist in some degree of truth and error simultaneously. Kelly summarises this interpretation of Foucault’s view about the discontinuity between discourse and reality in the formula that ‘truth = reality ÷ episteme’ (Kelly 2009, 22). Although Kelly is certainly right to draw attention to the theme of discontinuity in Foucault’s work, there are several problems with his account. The first issue centres on the term ‘discontinuist epistemology’. Indeed, this ‘discontinuist epistemology’ is better described as a ‘discontinuist ontology’ insofar as it problematises the relationship between language and the idea of a non-linguistic reality, rather than describing Foucault’s views about the constitution of knowledge per se. The second and more substantive problem with Kelly’s account is that Foucault repudiates the idea that truth is an inadequate or partial representation of a non-linguistic reality. In The Archaeology of Knowledge, Foucault suggests that his approach in The Order of Things does not treat discourse as even a partial representation of actual ‘things’ in reality, but that it instead conceives of ‘things’ as discursive constructions that can never be compared against a mind-independent reality:

‘Words and things’ is the entirely serious title of a problem; it is the ironic title of a work that modifies its own form, displaces its own data, and reveals, at the end of the day, a quite different task. A task that consists of not – of no longer – treating discourses as groups of signs (signifying elements referring to contents or representations) but as practices that systematically form the objects of which they speak (AK 49). ²

For this reason, Foucault’s mature position should be described as a ‘discontinuist historical ontology’ insofar as it treats objects as purely discursive constructions and does away with any metaphysical notion of an objective reality. In his 1967 essay on Nietzsche that was included in the Cahiers de Royaumont, Foucault argues that interpretation is an infinite task precisely because we cannot posit anything like a non-linguistic reality against which our interpretations might be measured:

…if interpretation can never be completed, this is quite simply because there is nothing to interpret. There is nothing absolutely primary to interpret, for after all everything is already interpretation, each sign is in itself not the thing that offers itself to interpretation but an interpretation of other signs (EW2 275).

Similarly, in ‘Nietzsche, Genealogy, History’, Foucault directly opposes the genealogical method to the metaphysical concern with the nature of reality, and suggests that genealogy analyses the development of metaphysical concepts as discursive constructions that are entirely immanent to history (EW2 378–379).

This is not to say that Foucault was always clear or consistent about defining ‘things’ as discursive constructions when discussing his ontology. In The Order of Discourse, Foucault again emphasises the idea that there is a fundamental discontinuity between discourse and ‘reality’ (OD 67–69). However, his claim that ‘we must conceive discourse as a violence that we do to things’ is somewhat misleading insofar as it implies that there are really ‘things’ in ‘reality’ that discourse might violate (OD 67). Similarly, in the 1973 lecture ‘Truth and Juridical Forms’, Foucault claims ‘knowledge can only be a violation of the things to be known, and not a perception, a recognition, and identification of or with those things’ (EW3 9). Foucault repeats this ambiguity again in a 1983 lecture at Berkeley, when he characterises his problematisation of truth as the process of analysing ‘the specific relation between truth and reality’ (FS 173). Despite these comments, the logic of Foucault’s claims about discontinuity would rule out the possibility of appealing to any non-linguistic reality that might otherwise constrain our epistemic practices. After all, if there is an inseparable discontinuity between words and things, he cannot coherently posit ‘things’ that exist in some non-linguistic reality precisely because of this discontinuity. For this reason, he would have been well-advised to avoid talking about ‘things’ that exist beyond discourse altogether – or, more precisely, to avoid using the word in connection with the kinds of discursive constructions he deals with in his work. Indeed, these kinds of comments are most charitably interpreted not as a precise articulation of his view, but rather as rhetorical statements made for the sake of distinguishing his own view as a radical departure from conventional realist ontologies (OD 67; EW3 9). Therefore, the way he describes his position in The Archaeology of Knowledge is the more coherent and promising way of framing his ontology. In other words, the most plausible rendering of his position is to say that ‘things’ – or, more precisely, ‘things’ like madness and sexuality – should be understood as discursive constructions rather than metaphysical entities that exist in a reality beyond discourse.

At this point, it is crucial to emphasise how these ontological commitments fit into Foucault’s broader critical project. Djaballah and Tiisala have both convincingly argued that Foucault’s conception of discursive objects is deeply Kantian (Djaballah 2008, 239; Tiisala 2015, 665–666). However, Tiisala clarifies that Foucault is not concerned with the necessary conditions that make our knowledge of metaphysical objects possible, but instead with ‘the additional sufficient conditions’ that make discursive objects ‘thinkable’ (Tiisala 2015, 665). Indeed, Foucault presents his overall project as a critical inquiry into the historical conditions of possibility for knowledge rather than a metaphysical inquiry into ‘the universal structures of all knowledge’ (EW1 315; AK 203; PT 23–82; EW2 459–462). This further clarifies that his mode of critique is not addressed to traditional questions of epistemology but rather to questions of historical ontology. While this Kantian framing of Foucault’s work is persuasive, it leaves open an important question about how Foucault can accommodate normative epistemological judgements given he has eschewed transcendental modalities like universality, necessity, and apriority. While these kinds of judgements are not the primary target of his methodology, there are moments in his work where he does challenge the validity of particular conceptions of things like madness and sexuality.

Before I can explain how Foucault’s approach to historical critique can sustain normative epistemological judgements, it is also necessary to clarify the way he conceptualises knowledge as immanent to history and power. This is because his underlying epistemological commitments are important in understanding how his historical critique of reason can proceed. As Foucault moved into his early genealogical period, he reframed his discontinuist historical ontology through the Nietzschean theme of power. During this period, Foucault was more consistent in his rejection of the idea that discourse is even a partial reflection of reality and often warned against the temptation to recover the metaphysical origins of language. ³ Although Foucault certainly acknowledges his epistemological debt to Nietzsche in earlier texts, his 1971 essay ‘Nietzsche, Genealogy, History’ represents his first effort to elucidate his own ontology and epistemology through Nietzsche’s work. In this text, Foucault aims to radicalise the historicist dimension of Nietzsche’s thought by embracing the subtle distinction Nietzsche makes – but does not consistently observe – between the metaphysical ‘origin’ [Ursprung] of knowledge, and its ‘emergence’ [Enstehung] or ‘descent’ [Herkunft] within the field of history (EW2 370–376; cf. EW3 9). According to Foucault, Nietzsche is only truly a genealogist when he opposes the genealogical method to the pursuit of the origin – although he openly concedes that his reading of Nietzsche is selective insofar as it eschews any metaphysical impulse in his work (EW2 371). In this way, Foucault presents genealogy as a method that conceives of knowledge as something that should be understood as entirely immanent to history. ⁴

This leads Foucault to position genealogy as an alternative to a metaphysical account of knowledge insofar as it does not assume that the emergence of knowledge can be traced back to some originary perception, sensation, or disclosure in language that ensures it will correspond to reality (EW2 373; cf. OT 328–329). Instead, Foucault embraces the idea that knowledge is a human artefact that was invented for practical purposes, and is therefore liable to be shaped by strategic interests in ways that undermine the possibility of veridical perception. This leads Foucault to follow Nietzsche in claiming that our pursuit of knowledge is motivated by a violent impulse, going as far as to claim that ‘all knowledge rests upon injustice (that there is no right, not even in the act of knowing, to truth or a foundation for truth) and that the instinct for knowledge is malicious’ (EW2 387). On this basis, Foucault claims that knowledge ‘is not made for understanding’ but rather ‘for cutting’ – that is to say, that knowledge often begins with an interpretation motivated by practical concerns and self-interest rather than the desire to reach an accurate understanding of the thing in question (EW2 380; cf. BC xviii). For Foucault, then, knowledge is fundamentally a way of coping with the world rather than trying to represent it.

In the Lectures on the Will to Know, Foucault contrasts Nietzsche’s account of the ‘will to know’ with Aristotle’s account of what motivates the production of knowledge (LWK 5). Whereas Aristotle sees knowledge as motivated by a genuine desire to know the world as it really is, the Nietzschean approach developed by Foucault ‘claims that the Will to know composes illusions, fabricates lies, accumulates errors, and is deployed in a space of fiction where the truth itself is only an effect’ (LWK 197). In this way, Foucault understands Nietzsche as arguing that the will to know is really just a concealed effect of the will to power, and that knowledge does not originate from our veridical perception of reality, but rather that inquiry is at least initially motivated by situated interests. For Foucault, this insight should lead us to distinguish between two kinds of truth: ‘the truth that is error, lie, illusion: the truth that is not true’ and ‘the truth freed from this truth-lie: the truthful truth’ (LWK 219). While Foucault often speaks as if the will to know is reducible to the will to power, he sometimes implies that the will to know can manifest itself in ways that are more conducive to illuminating ‘the truthful truth’ (WK 55). Foucault explicitly draws this distinction once again in his written response at the end of ‘Truth and Power’ (EW3 126–133). In this piece, he makes it clear that his use of the word ‘truth’ does not refer to ‘the ensemble of truths which are to be discovered and accepted’, and that his work is primarily concerned with revealing ‘the ensemble of rules according to which the true and false are separated and specific effects of power are attached to the true’ (EW3 132).

This helps to explain how Foucault can treat knowledge as immanent to power without precluding the possibility of making normative epistemological judgements. In other words, it suggests that any critique or improvement of our epistemic standards will have to be immanent to the discursive practices it seeks to describe. Indeed, the ‘truthful truth’ does not have to be defined in terms of correspondence to reality but can instead be understood as the product of our most rigorous and successful epistemic norms. In this way, I suggest that we can rely on some accepted standards of rationality to call others into question. This is consistent with Foucault’s presentation of his work as an immanent critique of Enlightenment rationality (EW1 303–319). More specifically, my suggestion is that Foucault draws on standards and norms that were derived from the natural sciences within this historical critique of reason. This view will be outlined at length in the next section.

The force of scientific truth: Situating Foucault’s naturalism within his historico-critical project

Having offered a critical reconstruction of Foucault’s underlying ontology, I will now attend to the question of how he makes normative epistemological judgements within his historico-critical project. My strategy is to emphasise Foucault’s respect for the natural sciences, which demonstrates that he is willing to draw on rigorous epistemological standards that have developed within the field of history. I will argue that Foucault is impressed by the epistemic resources produced by the natural and formal sciences, and will demonstrate why he thinks the rigour of the natural sciences derives from their empirical accountability and self-correcting orientation. Although he does not always emphasise it, Foucault has a greater respect for scientific epistemology than was assumed during the initial reception of his work in the Anglophone world. As several scholars have since noted, Foucault’s relationship to the French tradition of history and philosophy of the sciences was overlooked during this reception, but was an important part of his intellectual formation (Dreyfus 1987, x–xi; Gutting 1989, 9–54; Dews 1992; Kelly 2009, 26–27; Webb 2013; Ross 2018). Indeed, these critics argue that charges of epistemic nihilism were likely fuelled by a lack of familiarity with its constitutive schools of historical epistemology and Comtean positivism.

Most prominently, Gutting has argued that Foucault’s approach to epistemology is unintelligible unless we situate his work within this tradition and especially in relation to Canguilhem (Gutting 1989, 52–54). More recently, Webb has traced the formative influence of this tradition on The Archaeology of Knowledge, paying particular attention to the role that figures like Bachelard, Cavaillès, and Serres played in the development of Foucault’s epistemological views (Webb 2013, 7–38). This tradition affords scientific epistemology a privileged epistemic status, despite also interrogating the ways in which scientific practice has been shaped by contingent historical factors and subtle forms of normativity. For this reason, it assumes that scientific knowledge is only ever provisional and is thus likely to remain in some degree of error (EW2 476–477). In other words, this approach holds that there is always likely to be some discrepancy between our accepted bodies of truth and what our most rigorous epistemological standards suggest is actually true.

It is much easier to understand Foucault’s appreciation for scientific epistemology following the contextualisation of his work as an immanent – rather than a totalising – critique of scientific reason. As Gutting rightly argues, Foucault clearly accepts that the natural and formal sciences are characterised by a core of methodological rigour, and there are good reasons to think that he might accept scientific normativity as a paradigm of rigorous inquiry even though he rarely defended this view in a sustained manner (Gutting 1989, 257–258). In ‘The Subject and Power’, Foucault makes it clear that while his approach is not based on ‘a dogmatic belief in the value of scientific knowledge’, neither does it amount to ‘a skeptical or relativistic refusal of all verified truth’ (EW3 330–331). Indeed, Foucault implies that his approach to knowledge is consistent with genuinely scientific standards insofar as it opposes bodies of knowledge that engender ‘secrecy, deformation, and mystifying representations imposed on people’ (EW3 329–331). In this way, it is broadly consistent with the self-correcting orientation of scientific inquiry. This also helps to explain why Foucault’s focus is on the epistemologically suspect ‘human sciences’ – such as psychology, sociology, and anthropology – rather than the highly rigorous natural and formal sciences (OT 344–387). In the final chapter of The Order of Things, Foucault even declares that the human sciences are ‘not sciences at all’, but have accrued their cultural authority due to their false claim to scientificity (OT 366). Similarly, in a 1968 interview with Esprit, Foucault’s respect for genuine scientificity leads him to claim that some bodies of knowledge – such as mathematics and physics – have stronger ‘epistemological structures’ than others (FL 44).

While Foucault’s views on scientificity are clearly indebted to Bachelard and Canguilhem, he has a more complex relationship with these thinkers than has sometimes been acknowledged. However, there is a recent wave of scholarship that is attentive to both the continuities and discontinuities within the ‘Bachelard–Canguilhem–Foucault lineage’ (Elden 2021, 26; Chimisso 2015; Ross 2018; Talcott 2019 ix–xii). Elden has rightly argued that while ‘there are important connections between the three thinkers’ there is also ‘a danger of reducing all three under the rubric of a “historical epistemology”’ (Elden 2021, 26; Chimisso 2003, 310–314; Chimisso 2015). With respect to Foucault’s place in this tradition, several scholars have argued that the epistemological focus of Bachelard and Canguilhem’s work contrasts with Foucault’s historico-critical appropriation of their ideas (Lecourt 1975, 119–120; Debru 2004, 79–80; Chimisso 2015, 65–70; Ross 2018, 145). For example, Chimisso has argued that while Foucault even came to influence Canguilhem’s conception of scientific ideology, the latter was still eager to emphasise his commitment to the epistemological distinction between science and non-science (Chimisso 2015, 65–70). Therefore, although Foucault’s respect for the natural and formal sciences is only intelligible in relation to this tradition, it is important to clarify the ways in which his treatment of scientificity differs from his intellectual ancestors.

In The Archaeology of Knowledge, Foucault offers his most sustained and systematic discussion of scientificity. He affirms that his focus is on the ‘dubious’ and ‘still imprecise disciplines that are perhaps doomed for ever to remain below the threshold of scientificity’ (AK 178). However, he also makes it clear that his method does not directly adjudicate their claims to scientificity, but rather attempts to track the historical emergence and development of their epistemological structures (AK 190–191). Nonetheless, he is clearly strategic in his choice of topics, and is clear about the fact that the human sciences fall below what he calls the ‘threshold of scientificity’ (AK 178, 186, 190). This term describes an important stage in the development of a discursive formation, which logically comes after the ‘threshold of epistemologization’ in which a discursive formation ‘claims to validate (even unsuccessfully) norms of verification and coherence’ (AK 186–187; Gutting 1989, 252–253). For Foucault, a discursive formation crosses the threshold of scientificity when it moves beyond these fundamental norms of verification and coherence towards more rigorous scientific rules (AK 187, 190).

This passage also serves as an occasion for Foucault to clarify his methodological differences with Bachelard and Canguilhem. According to Foucault, different types of historical analysis operate at different thresholds of discourse (AK 190). For example, he identifies Bachelard and Canguilhem’s work as focusing on disciplines at the threshold of scientificity (AK 190). In Foucault’s view, their histories show how different sciences were ‘established over and against a pre-scientific level’ and thus reveal ‘what the science has freed itself from, everything that it has had to leave behind in its progress towards the threshold of scientificity’ (AK 190). In other words, he thinks they are primarily interested in how pre-scientific concepts were ‘purified’ and ‘accorded the status and function of a scientific concept’ (AK 190). According to Foucault, their project is more traditionally epistemological insofar as it is designed to make normative epistemological judgements about when these forms of knowledge have crossed the threshold of scientificity:

Consequently, this description takes as its norm the fully constituted science; the history that it recounts is necessarily concerned with the opposition of truth and error, the rational and the irrational, the obstacle and fecundity, purity and impurity, the scientific and the non-scientific. It is an epistemological history of the sciences (AK 190).

By contrast, Foucault situates his own approach at the threshold of epistemologisation. This approach targets discursive formations ‘that are not necessarily all sciences (and which may never, in fact, succeed in becoming sciences)’ (AK 190). In Foucault’s view, this approach works at a level where ‘scientificity does not serve as a norm’ (AK 190). In other words, his method is not designed to offer normative epistemological judgements about claims to scientificity, but instead offers an ‘archaeological history’ that attends to historical conditions of possibility (AK 190; cf. OT xxii).

In this way, Foucault distinguishes his project from the work of Bachelard and Canguilhem by shifting away from their more traditionally epistemological history of the sciences and towards a critical inquiry into the historical conditions of possibility for knowledge. Furthermore, Foucault’s focus on the threshold of epistemologisation leads him away from the natural sciences, and towards the human sciences. ⁵ In the 1973 lecture ‘Truth and Juridical Forms’, he explicitly distinguishes between ‘two histories of truth’ that one might undertake (EW3 4). He describes the first as ‘a kind of internal history of the truth’ that ‘rectifies itself in terms of its own principles of regulation’ (EW3 4). Although he does not identify Bachelard or Canguilhem by name, he does characterise this approach as ‘the history of truth as it is constructed in or on the basis of the history of the sciences’ (EW3 4). In contrast, he identifies his own work with the second history of truth (EW3 4). This history does not take internal principles of regulation as the norm of inquiry into the natural sciences, but instead attends to ‘other places where truth is formed’ in our societies and where ‘a certain number of games are defined’ that allow different forms of subjectivity, object domains, and knowledge to ‘come into being’ (EW3 4). For this reason, he describes this second approach as an ‘external, exterior history of truth’ that does not rely on the internal norms and standards of the bodies of knowledge under investigation (EW3 4). In other words, this second approach concerns the conditions of possibility for the human sciences rather than adjudicating the claims to scientificity of the natural sciences (EW3 4–5).

However, the content of Foucault’s historical studies is often more varied and complex than these methodological reflections would suggest. After all, Foucault does not completely avoid more traditional epistemological concerns in these studies, even if they are never his primary focus. Indeed, I argue that he does sometimes make normative epistemological judgements when criticising the human sciences. In this way, Foucault’s respect for the epistemological strength of the natural sciences does inform his historical studies, specifically insofar as he privileges scientific normativity in making these sorts of judgements. Therefore, I suggest that Foucault does occasionally slip into a mode of analysis more characteristic of historical epistemology. This can be seen throughout several of his canonical historical studies. In The Will to Knowledge, Foucault chastises the discourse of modern sexuality for taking confessional practices as its paradigm while blatantly ignoring the insights of reproductive physiology (WK 54–55, 153–155; FL 44). Indeed, a central argument of The Will to Knowledge is that this discourse on sexuality was able to enlist the cultural authority of the natural sciences due to its fabricated proximity to biology (WK 52–73). According to Foucault, the nineteenth century saw the concept of ‘sex’ integrated ‘into two very distinct orders of knowledge’ in the West that developed simultaneously but without any ‘real exchange’ or ‘reciprocal structuration’ (WK 55). For Foucault, the pathologisation of sexuality in medical discourse was little more than a way in which ‘moral obstacles, economic or political options, and traditional fears could be recast in a scientific vocabulary’ (WK 55). Foucault goes as far as to claim that this discourse was exceptionally weak from an epistemological perspective:

When we compare these discourses on human sexuality with what was known at the time about the physiology of animal and plant reproduction, we are struck by the incongruity. Their feeble content from the standpoint of elementary rationality, not to mention scientificity, earns them a place apart in the history of knowledge (WK 54).

On the other hand, he claims that the epistemological strength of biology was secured by the fact it developed in accordance with a ‘general scientific normativity’ (WK 54). He even argues that while medical theories of sexuality were distorted by pernicious power relations, reproductive physiology was shaped by a genuine will to knowledge characteristic of rigorous and self-correcting scientific inquiry (WK 55).

Similarly, Foucault appears to accept that the medical conception of ‘disease’ is empirically rigorous, and thus understands The Birth of the Clinic as an investigation into how modern medicine developed under contingent historical conditions without questioning its basic epistemic credibility (EW3 111–112; BC xix). This lack of questioning is primarily because his study is a historico-critical inquiry into ‘the conditions of possibility of medical experience in modern times’ rather than a historical epistemology of medical science (BC xix). However, it is also due to his respect for the core of genuine scientificity that helps define modern Western medicine. In a 1976 interview with Fontana and Pasquino, he states that while ‘the epistemological profile of psychiatry is a low one’, he believes that ‘medicine has a much stronger armature than psychiatry’ even though ‘it too is profoundly enmeshed in social structures’ (EW3 111–112). Indeed, in The Birth of the Clinic, Foucault argues that the development of medical knowledge was motivated by forms of normativity that are deeply political and thus exceed purely epistemic considerations (BC 22–37). However, while Foucault thinks that power relations have shaped medical knowledge, he also thinks this interaction was productive and causal rather than distortive (EW3 117). ⁶ Therefore, he recognises that medical science is characterised by rigorous methodological standards that give it a high degree of protection against epistemic distortion.

On the other hand, he presents ‘delinquency’ and ‘sexuality’ as discursive objects that were produced by political interests in the absence of any rigorous epistemological structure, while the object of ‘madness’ appears to occupy a space between these two extremes (DP 277–292; WK 55; EW1 296). In Discipline and Punish, Foucault repeatedly emphasises that the figure of the ‘delinquent’ was not constructed through a genuine scientific normativity, but rather in accordance with highly dubious ideas about normality derived from human sciences like criminology and psychiatry (DP 252–256, 277–279, 293–296). More specifically, Foucault highlights how the emergence of the carceral system was supported by a judicial apparatus that drew its authority from the human sciences as well as its feigned connection to physiology (DP 293–296). Indeed, Foucault emphasises that the supposed legitimacy of disciplinary techniques was predicated on this false claim to scientificity. He argues that the ‘completion of the carceral system’ – as marked by the opening of the Mettray Penal Colony – coincided with the ‘birth of scientific psychology’ (DP 293–295). While he accepts that what happened at Mettray ‘was obviously of a quite different order’ to contemporaneous developments in the human sciences, he argues that their concurrence allowed the carceral system to draw on the cultural authority of scientific knowledge without actually meeting the appropriate standards of scientific rigour (DP 295–296). Of course, Foucault’s focus in this text is on the historical conditions of possibility for the carceral system and the discursive object of delinquency. Nonetheless, his decision to analyse the carceral system was a strategic choice, having explicitly stated that he was motivated to write the book due to his concerns about the ‘steep rise’ in ‘mechanisms of normalization’ and the effects of power engendered by ‘the proliferation of new disciplines’ (DP 306). With this in mind, his respect for the natural sciences plays a role in helping him to expose the lack of rigour that characterises the human sciences and the way they subtend the carceral system.

Foucault’s respect for the natural sciences also shapes his work on psychology and psychiatry. Although it is sometimes excluded from studies on Foucault’s work, it is worth attending to his analysis of the medicalisation of mental illness in Mental Illness and Psychology – the heavily revised version of his first book that he reluctantly approved for publication in 1962 (MIP). The fundamental argument of this text is that ‘mental pathology’ fails to meet the same standards of rigour offered by ‘organic pathology’ (MIP 1–13). According to Foucault, while psychiatry certainly claims to define ‘mental illness’ using the same conceptual methods as organic medicine, any suggestion of a ‘unitary pathology’ that captures both ‘the psychological and physiological domains’ is ‘purely mythical’ (MIP 7–10). In Foucault’s view, this is because psychology and psychiatry lack the degree of conceptual rigour characteristic of physiology and medical science, specifically insofar as psychology’s fundamental concept of ‘personality’ cannot be conceived as a totality of functional parts in the same way as the physiological concept of an ‘organism’ (MIP 10–13). In this way, we can see how this critique of psychiatry is predicated on a normative epistemological judgement about its lack of scientificity.

This concern for scientificity also underpins Foucault’s analysis in The History of Madness. In this text, Foucault once again emphasises that the supposed ‘objectivity’ of psychiatric practice was really ‘a reification of a magical type’ rooted in ‘a transparently clear moral framework which was slowly forgotten as positivism imposed the myth of scientific objectivity’ (HM 509). In this way, Foucault extends his previous argument that mental pathology fails to adhere to the scientific norms established by organic pathology (HM 186–198). He argues that it instead developed in accordance with a contingent moral normativity that made madness into ‘the stigma of a class that had abandoned the forms of bourgeois ethics’ (HM 378, 13–16, 355–380, 427–503). Similarly, in the sixth lecture of Psychiatric Power, Foucault highlights how nineteenth century psychiatric practice was able to establish its perceived epistemic authority by claiming fidelity to ‘psychiatric nosology’ and ‘anatomical-pathological research’, but without properly employing the knowledge of these disciplines or positively justifying its own claim to truth (PP 134–138). In other words, we can see that Foucault supplements his historical critique of the medicalisation of madness with an interrogation of its false claims to scientificity.

On the basis of this detailed textual analysis, I argue that Foucault’s respect for genuine scientific inquiry is crucial in understanding how he can accommodate normative epistemological judgements within his historical critique of reason. In contrast to the metaphysical ambition of capturing the true nature of reality, the dominant conception of the natural sciences is that they merely aim to establish evaluative criteria that are experimentally successful for the purposes of prediction and control. In this way, they can be understood to propose hypotheses that are empirically accountable insofar as they are exposed to rigorous experimental testing and can be continuously refined on the basis of new results. Therefore, there is no inherent contradiction between Foucault’s historico-critical orientation and his naturalism. However, this is not to say that Foucault takes an uncritical view of scientific practice itself. Despite this experimental success, Foucault makes it clear that it is essential to take a historical view of scientific practice to understand the ways in which even sciences with a strong epistemological structure have been shaped by power relations and subtle forms of normativity (EW2 473). In Foucault’s introduction to Canguilhem’s The Normal and the Pathological, he explicitly argues that:

…the history of the sciences opens up a domain of analysis that is indispensable if epistemology is to be anything else but the simple reproduction of the internal schemas of a science at a given time (EW2 473; Canguilhem [1943] 1991).

Therefore, Foucault’s own approach cannot be understood as an uncritical or naïve celebration of scientific practice that treats it as the straightforward accumulation of truth (EW3 330–331). Indeed, Foucault draws attention to the ‘discontinuous’ nature of scientific development as an important theme in Canguilhem’s work, and explicitly cites Canguilhem – along with Bachelard’s notion of the epistemological break – as an important influence on his own discontinuist account of the historical development of knowledge (EW2 470–471; AK 4–14; Bachelard [1938] 2002). With this in mind, Foucault thinks that the epistemological strength of science derives from the fact that it can be continuously critiqued and refined from within, and that this can be pursued through the history of the sciences as well as through scientific practice itself. In this way, Foucault draws on the epistemological norms established by scientific practice to expose the potential weak points of specific bodies of knowledge from within his historical critique of reason.

However, to embrace some norms and standards of scientific inquiry is not merely to hand over the game of epistemology to the natural sciences. The influence of historical epistemology also helps to distinguish Foucault’s naturalism from the less critical mode of naturalism offered by thinkers like Quine (Quine [1969] 2004; Roth 1999). Whereas Quine takes philosophy to be continuous with the natural sciences, Foucault follows Bachelard and Canguilhem in refusing any conflation of philosophy with scientific epistemology (Quine [1969] 2004; Elden 2021, 116–118; Vagelli 2019, 106–107). To be sure, Bachelard and Canguilhem both maintain that the history of the sciences must be accountable to current scientific practice (Canguilhem [1983] 2005, 205; Bachelard [1938] 2002, 22–27). However, they also emphasise that the history of the sciences must be supported by a distinctively philosophical approach to questions of scientificity if it is to maintain its normative force, and that this involves drawing on established scientific norms to conduct an immanent critique of the sciences (Canguilhem [1983] 2005, 200–206; Bachelard [1938] 2002, 22–27). In his 1983 essay ‘The Object of the History of Sciences’, Canguilhem makes it clear that ‘the history of sciences is not a science and its object is not a scientific object’ (Canguilhem [1983] 2005, 206). Instead, he argues that a ‘philosophical epistemology’ is required for the historian of sciences to pass judgment on the scientificity of knowledge without simply affirming our existing body of accepted scientific knowledge (Canguilhem [1983] 2005, 200–206). More precisely, he argues that the tight integration of history and epistemology allows us to establish the necessary distance from the sciences to make normative epistemological judgements about them (Canguilhem [1983] 2005, 200–206; Bachelard [1938] 2002, 22–27). In this way, as Vagelli usefully puts it, Bachelard and Canguilhem offer an ‘a posteriori normative approach’ to the history of the sciences that is accountable but not reducible to scientific practice (Vagelli 2019, 99–104). This distinction between scientific epistemology and historical epistemology is crucial in reconstructing Foucault’s naturalism and explaining how he can sustain his epistemological criticisms of the human sciences.

It is also worth saying something about how Foucault’s respect for genuine scientificity fits with his claims about power-knowledge. In short, Foucault’s view is that scientificity is always immanent to power, but that this does not give us any reason to think that scientific validity lacks rigorous epistemic constraints. In a 1984 interview with Concordia, Foucault claims that he is ‘absolutely not saying that games of truth are just concealed power relations’ and that such a description would be a ‘horrible exaggeration’ of his view (EW1 290; cf. FL 462). The complexity of Foucault’s view derives from the fact that he holds that power and knowledge are always co-implicated and mutually immanent (WK 98; DP 27; AK 184–186). For this reason, he believes that power can both play a positive role in the causal production of legitimate knowledge or have a distortive effect on what passes for knowledge. In this way, he recognises that political interests play an important part in motivating even the most rigorous forms of inquiry (PK 131–133; EW1 296). For all its methodological rigour, then, Foucault does not think scientific epistemology is in any way a pure representation of a timeless truth (AK 184–186). In fact, as Lorenzini rightly argues, he thinks it is ‘a regime of truth among others’ that is profoundly historical and stands in a circular relation with power (Lorenzini 2023, 23).

However, Foucault does not see all forms of knowledge as being at the same risk of being distorted by pernicious power relations. His respect for the natural and formal sciences leads him to suggest that domains of knowledge exist on a continuum from those with the ‘strongest epistemological structure’ – such as mathematics and physics – to weaker ones like psychology that do not meet the same standards of scientific rigour (FL 44). He also recognises the ‘intermediate’ sciences – ‘biology, physiology, political economy, linguistics, philology’ – that are more empirical than the highly formalised sciences and thus more vulnerable to subtle forms of normativity, but that are still much more robust than the human sciences (EW2 328; EW2 470; AK 187–190). In other words, while knowledge is always vulnerable to epistemic distortion by pernicious power relations, Foucault does not think this is what power inherently does to knowledge, and there is no reason to think that the mutual immanence of power-knowledge results in the permanent distortion of knowledge. Indeed, it is entirely possible to establish rigorous epistemic standards within the field of power. In this way, we can see that Foucault’s respect for the natural sciences is coherently immanent to his claims about power-knowledge as well as his historical critique of reason. ⁷

Empirical inquiry as a model for historical description: Genealogical positivism and the descriptive rigour of Foucault’s method

Foucault’s respect for the natural sciences is also crucial in making sense of the descriptive rigour of the genealogical method itself. Some critics have reasonably questioned what kind of epistemic constraints Foucault imposes on his method of historical interpretation. While Descombes recognises that Foucault aims to improve the rigour of our epistemic practices, he argues that Foucault’s commitment to positivistic description is undermined by his claim that we cannot identify any facts about a non-linguistic reality that is independent of discourse (Descombes [1979] 1980, 117; cf. Habermas [1985] 1987, 276–278). In this way, Descombes challenges the suggestion that genealogy can justify its validity claims and thus secure the descriptive rigour of its historical accounts. My response is to suggest that Foucault actively models the genealogical method on scientific inquiry, and therefore defends the descriptive rigour of his work by emphasising its empirical accountability to historical evidence. In other words, I will take seriously Foucault’s claim that his work is based on a form of historical positivism (AK 125; OD 73).

In The Order of Discourse, Foucault claims that ‘the genealogical mood will be that of a happy positivism’ (OD 73). By using the term ‘positivism’, Foucault clearly wants to suggest that his method resembles the empirical approaches of the natural and social sciences. ⁸ In the 1978 lecture ‘What is Critique?’, Foucault suggests that ‘positivist science’ is a central feature of Enlightenment rationality insofar as ‘it basically had confidence in itself, even when it remained carefully critical of each one of its results’ (PT 50). In this way, Foucault wants to suggest that the genealogical method adopts an empirically accountable, fallibilist, historically reflexive, and self-correcting approach to interrogating different forms of knowledge. Indeed, Foucault suggests that his approach to historical description is an empirical enterprise insofar as it is directly accountable to historico-discursive evidence that is accessible through archival research (AK 27). ⁹

For Foucault, then, genealogical description resembles scientific practice insofar as it both scrutinises existing theories and generates new hypotheses by drawing on the best available historico-discursive evidence. Although these theories are necessarily underdetermined, they can at least be refined, accepted, or rejected based on further historical research. In the encyclopaedia entry he wrote about himself under the pseudonym ‘Maurice Florence’, Foucault emphasises the idea that genealogy challenges metaphysical theories by putting them to the test of historical inquiry rather than rejecting them according to a priori philosophical criteria. In this way, he argues that genealogy adopts:

…a systematic skepticism toward all anthropological universals – which does not mean rejecting them all from the start, outright and once and for all, but that nothing of that order must be accepted that is not strictly indispensable. In regard to human nature or the categories that may be applied to the subject, everything in our knowledge which is suggested to us as being universally valid must be tested and analyzed (EW2 461–462).

Although Foucault models his approach to historical description on empirical inquiry, it should be clear that genealogy does not claim to adhere to exactly the same standards as the natural sciences. On this front, Foucault is again indebted to Bachelard and Canguilhem’s refusal to conflate historico-philosophical analysis with scientific epistemology. There are at least three ways in which genealogical positivism differs from scientific practice. The first is that genealogy deals with discursive events rather than natural ones. In The Archaeology of Knowledge, Foucault claims that his method does not directly deal with facts about the natural or non-linguistic world, but rather with ‘discursive facts’ that can be identified in the form of ‘statements’ (AK 25–27). Indeed, Foucault explicitly claims that ‘it should never be possible to assign, in the order of discourse, the irruption of a real event’ (AK 25). Second, genealogy lacks the rigorous experimental dimension that is central to the practice of the natural sciences. In his later work, Foucault does emphasise the importance of personal and collective experimentation in testing the historical limits of our subjectivity (EW1 315–316). However, such experiments are clearly quite different to the kind of controlled experiments conducted in biology or chemistry insofar as we cannot isolate the hyper-complex conditions of the social world. Third and relatedly, Foucault openly concedes that genealogy lacks predictive power due to the overwhelming complexity of history and society, the discontinuous nature of discursive change, and the unstable nature of power relations (AK 175–177; OD 67; WK 100–102).

With this in mind, Foucault does not offer a ‘positivism’ in the traditional meaning of the word, but only in a heavily qualified and restricted sense. In fact, Foucault explicitly opposes his work to the traditional positivism associated with eighteenth and nineteenth century scientism. In The History of Madness, Foucault explicitly criticises traditional positivism for its role in obscuring the profoundly normative foundations of the medicalisation of madness in the modern period:

Anyone who wants to pursue an analysis of the deep structures of objectivity in the knowledge and the psychiatric practice of the nineteenth century, from Pinel to Freud, needs to show that, from the very beginning, objectivity was a reification of a magical type, which could only be accomplished with the complicity of the patients themselves, starting out from a transparently clear moral framework which was slowly forgotten as positivism imposed the myth of scientific objectivity. The origins and the meaning of the practice were forgotten, but its use persisted and it was always present. What we call psychiatric practice is a certain moral tactic contemporaneous with the late eighteenth century, which is preserved in the rituals of life in asylums, covered over by the myths of positivism (HM 509).

Similarly, The Birth of the Clinic concludes with a discussion of how the formation of clinical medicine was made possible by certain forms of normativity, and that this historical process has also since been obscured by the myths of positivism (BC 198–199). In The Will to Knowledge, Foucault claims that the ‘scientia sexualis’ of the nineteenth century functioned ‘under the guise of its decent positivism’, having demonstrated earlier in the text that this discourse lacks any plausible claim to genuine scientificity (WK 70, 54). In the first lecture of Society Must Be Defended, he even states that his ‘genealogical project’ is ‘not a matter of some form or other of scientism’ and does not lead to a positivism ‘in the ordinary sense of the word’ (SMBD 8–9; cf. AK 206). It should be clear, then, that Foucault explicitly disavows any conception of positivism associated with scientism or the notion of scientific objectivity. In other words, he rejects the suggestion that discursive objects are given through scientific inquiry rather than constructed through value-laden interpretations rooted in specific historical conditions.

With this in mind, Foucault’s genealogical positivism must be understood through his broader approach to historical critique, especially insofar as it emphasises the historical conditions of knowledge that enable the positivities of different discourses (AK 126–131). Foucault hints in this direction with his critique of traditional positivism in The Order of Things. In the penultimate chapter, Foucault criticises positivism for trying to ground the philosophical notion of truth in empirical facts without accounting for the conditions of possibility that make these empirical truths intelligible to begin with (OT 319–320; Gutting 1989, 201–202). Therefore, he claims that traditional positivism is dogmatic in the Kantian sense, and thus collapses into a ‘pre-critical naïveté’ (OT 320). In The Archaeology of Knowledge, Foucault explicitly defines his own positivistic approach to discourse in terms of his historicisation of transcendental philosophy:

If, by substituting the analysis of rarity for the search for totalities, the description of relations of exteriority for the theme of the transcendental foundation, the analysis of accumulations for the quest of the origin, one is a positivist, then I am quite happy to be one (AK 125).

In other words, Foucault emphasises that his ‘positivism’ is of a heavily qualified type and must be situated within his broader critical project. He reaffirms this idea in The Order of Discourse, when he suggests that ‘the critical and genealogical descriptions must alternate, and complement each other’, before claiming that ‘if the critical style is that of studious casualness, the genealogical mood will be that of a happy positivism’ (OD 73).

While Foucault does emphasise the empirical accountability of his method of historical description, he is also reflexive about the element of interpretation that derives from his own situatedness in the field of history. Both these features are captured in the opening lines of ‘Nietzsche, Genealogy, History’, where he declares that historical description does not involve the straightforward accumulation of truth but instead involves a process of diligent and continuous revision (EW2 369). Therefore, when Foucault claims that his method is positivistic, he is trying to emphasise that it is empirically accountable, fallibilist, historically reflexive, and self-correcting. However, he also wants to disavow any claim to objectivity, as well as the scientism and dogmatism he associates with ‘militant positivism’ (HM 379). Instead, he maintains that his genealogical positivism is housed within a historical critique of knowledge that rejects the givenism associated with traditional positivism. Indeed, Foucault’s rejection of givenism is fundamentally grounded in his discontinuist historical ontology and Kantian conception of discursive objects, both of which were discussed in the second section of this article.

This clarification of genealogical positivism helps to address Descombes’s concern that Foucault’s ontology rules out any coherent notion of ‘fact’. While Descombes claims that Foucault does not believe in the positivist notion of fact, he incorrectly assumes that genealogical positivism must understand facts as claims that can be judged according to their correspondence to a non-linguistic reality (Descombes [1979] 1980, 116). In other words, this process of establishing facts does not necessarily rely on the metaphysical notion of correspondence, but could easily be understood through the fallibilist notion of empirical adequacy. Furthermore, as we have seen, Foucault is able to situate this epistemic standard within his broader critique of historical reason in a way that avoids claiming universal validity. In this way, his perspectivist rejection of metaphysical facts and hostility towards traditional positivism is entirely compatible with the robust epistemic standards derived from scientific inquiry.

Descombes also raises the concern that Foucault undermines the possibility of positivistic description by claiming that knowledge is always grounded in interpretation (Descombes [1979] 1980, 117; cf. Habermas [1985] 1987, 276–278). However, in ‘Nietzsche, Genealogy, History’, Foucault argues that genealogy does not directly interpret the world, but rather records different interpretations through the empirical work of historical description (EW2 378–379; EW2 275). Nonetheless, there is an important question about the degree to which this method of description involves an element of interpretation, especially given Foucault’s subordination of his positivism to his broader project of historical critique. Foucault’s view of the role of interpretation in his methodology appears to have shifted over the course of his career, making his position something of a moving target. In The Archaeology of Knowledge, he suggests that the archaeological method involves a rigorous description of discursive events, which allows us to define the rules that constitute discourses as coherent unities (AK 27). ¹⁰ Similarly, in his response to the Cercle d’épistémologie, Foucault emphasises that the empirical basis of archaeological description is what enables ‘the project of a pure description of the facts of discourse’ (EW2 306; RC 16). In this way, the Foucault of the late archaeological period appears to suggest that discursive formations can be identified and described without any major hermeneutical difficulties – on this view, one simply had to attend to the enunciative field and map out the relevant discursive relations between statements (AK 28). Indeed, the empirical basis of archaeology is precisely what he thought would allow the archaeologist to complete a description – rather than an interpretation – of discourse:

The analysis of statements, then, is a historical analysis, but one that avoids all interpretation: it does not question things said as to what they are hiding, what they were ‘really’ saying, in spite of themselves, the unspoken element that they contain, the proliferation of thoughts, images, or fantasises that inhabit them; but, on the contrary, it questions them as to their mode of existence, what it means to them to have come into existence… (AK 109).

However, the Foucault of the genealogical period found this account to be lacking insofar as the idea of ‘pure description’ is clearly in tension with his claims that all knowledge is grounded in interpretation and immanent to power (EW2 275–278; DP 28–29). ¹¹ In this way, Foucault’s genealogical expansion of archaeology seems to have been at least partly motivated by his desire to account for his own situatedness in the field of power-knowledge (Dreyfus and Rabinow [1982] 1983; Davidson 1986).

Dreyfus and Rabinow have offered what remains the most influential and convincing account of Foucault’s transition to genealogy. Their account is useful in explaining how Foucault can defend the descriptive rigour of the genealogical method despite his abandonment of ‘pure description’ as a methodological ideal. On their view, Foucault’s expansion into genealogy was designed to account for the ‘methodological failure’ of archaeology (Dreyfus and Rabinow [1982] 1983, 79–100). For Dreyfus and Rabinow, the Foucault of the late archaeological period found himself in the paradoxical situation of trying to ‘speak from a position of phenomenological detachment’ despite having shown the impossibility of occupying such a position (Dreyfus and Rabinow [1982] 1983, 100). On their view, then, Foucault came to realise that archaeological description itself had ‘to be accounted for and relativised’ before being assimilated into his expanded method of genealogy (Dreyfus and Rabinow [1982] 1983, 99). On this basis, Dreyfus and Rabinow argue that Foucault effectively abandoned the principle of ‘pure description’ and instead embraced an ‘interpretive analytics’ that sees genealogy as immanent rather than orthogonal to the discursive practices it seeks to understand (Dreyfus and Rabinow [1982] 1983, 89). On their view, then, Foucault’s expansion into genealogy allowed him to develop an immanent mode of critique that acknowledges the role of interpretation in making his rigorous form of historical description possible in the first place (Dreyfus and Rabinow [1982] 1983, 102–103).

Therefore, Descombes is right to suggest that Foucault sometimes struggled to explain how his positivistic approach to historical description could be reconciled with his own situatedness within the field of interpretation. However, Foucault’s expansion into genealogy gave him the resources to solve this issue in a relatively straightforward way. Building on Dreyfus and Rabinow’s account, my suggestion is that genealogy secures its descriptive rigour by drawing on epistemic standards that have been formed within the interpretative tradition of Enlightenment rationality, which is partly defined by the rigorous empiricism of modern scientific practice (PT 50). With this in mind, I follow Dreyfus and Rabinow in recognising that Foucault’s method of historical description relies on a more fundamental mode of interpretation. Although Foucault is sometimes cast as offering a totalising critique of Enlightenment rationality, he understood his own work as an immanent modification of this tradition that uses some of its existing conceptual resources to call others into question (EW1 312–316). In a 1984 interview with Concordia, Foucault outlines the logical structure of immanent critique as follows:

In a given game of truth, it is always possible to discover something different and to more or less modify this or that rule, and sometimes even the entire game of truth (EW1 297).

In other words, Foucault sees all knowledge as provisional and subject to revision, but does not deny the existence of epistemic standards that can help us discriminate between competing interpretations. As I have already shown, he seems to accept that some of our strongest epistemic standards are derived from scientific epistemology, and that these standards can serve as the basis for an immanent critique of knowledge. In this way, it makes sense to question the validity of some epistemic or hermeneutical resources while accepting others. Furthermore, the empirical basis of Foucault’s work provides it with a descriptive rigour that exceeds any crude notion of interpretation, even if it holds that our highest empirical standards were originally formed within a broader interpretative tradition.

Therefore, while Foucault’s work is clearly situated both within the field of power and within a particular interpretative tradition, its reliance on rigorous standards of description is precisely what protects genealogical insights from epistemic distortion. Of course, this is not to suggest that genealogy is immune from such distortion. Indeed, it should be clear that Foucault does not dismiss the possibility that his genealogies might be distorted by pernicious power relations and could even be exposed to genealogical scrutiny themselves. However, this vulnerability is common to all forms of knowledge, and this does not mean that any particular body of knowledge is permanently distorted by power relations simply because it is immanent to power. After all, power always serves as a causal condition on the production of our most rigorous forms of knowledge.

Conclusion

I have argued that Foucault’s respect for the natural sciences is crucial in making sense of his normative epistemological judgements. This has involved locating Foucault’s work within the French tradition of history and philosophy of the sciences and closely examining his views on scientific inquiry. More specifically, I have shown how he can appeal to rigorous epistemic norms and standards derived from the natural sciences within his broader project of historical critique. Furthermore, I have argued that he takes empirical inquiry as a model for historical description, and that this helps to secure the descriptive rigour of his genealogical method. In this way, Foucault can appeal to evaluative norms and standards that have been formed under the pressure of power relations, but that are still epistemologically robust by virtue of their self-correcting orientation and accountability to empirical evidence.