Readers of Michel Foucault's diverse and multifaceted pronouncements on truth have often found it difficult to reconcile his tendency to revise the concept in his early work (culminating, perhaps in the famous 1976 interview on truth) with perfectly reasonable statements where he (like the rest of us) marks the uncontentious difference between what someone thinks is true and what is true, a perfectly clear and commonsensical distinction even Foucault had to rely on when investigating in his genealogies what we took to be true about madness, penal systems, sexual policies, "biopower", etc. My aim in this paper1 is to show that in exploring these (and other) issues, the early but extremely influential Foucault might have had in mind a cluster of distinctions and issues which make perfect sense: the distinction between reasons for believing something, and explanations of why we believe what we believe, the fact that the way we acquired certain contested concepts is not always fully transparent to us, and the key role of declarative speech acts in creating and maintaining institutional objects and social identities. But these insights can be fully articulated, explored and discussed with a minimalist conception of truth in mind. His observations anticipate – albeit in an unsystematic and often indirect way – current "rediscoveries" of those insights by analytic philosophers. A minimalist about truth holds that these insights do not require a revision of our ordinary concept of truth. The flip side of my argument is that Foucault and his followers should not have grounded his views in a substantial revision of the concept of truth and he should not have mounted an anti-realist (relativistic, idealistic or constructivist) conception of truth.2 Truth is and has always been "a thing of this world"; his idiosyncratic reconceptualizations of truth are not needed to defend social dimensions of belief systems, the way social facts emerge and the unintended social effects of speech acts, or so I will argue.

The plan of the paper is as follows. In Section 2 I explain my dialectical strategy. Section 3 presents a brief overview of a minimalist approach to truth, why such an approach shuns both relational and idealistic conceptions of truth and how it allows us to separate issues about truth simpliciter from issues about, say, the role of performative language, the value of true beliefs, the justification and explanation of beliefs, etc. In sections 4 to 6 I discuss various aspects of "criterial" and "perspectival" uses of truth in Foucault and show how the minimalist approach helps us understand how important Foucauldian insights can almost seamlessly be related to recent work in analytic philosophy, and why it was a strategic mistake of Foucault to have thought that his insights require a revision of the concept of truth.

According to C.G. Prado, Michel Foucault offered five interrelated "uses of truth":3 the criterial use, the constructivist use, the perspectivist use, the experiential use, and the tacit-realist use. The latter use is, as many of Foucault's early critics have argued, manifestly inconsistent with Foucault's account of truth as produced by power.4 Since all these "uses" (and there may be more of them) incorporate into the concept of truth issues that go beyond its minimal core, my dialectical strategy will be to isolate the core concept of truth, based on a family of minimalist conceptions of truth. I then defend that Foucault's proposals should best be interpreted as articulating insights about how social and cultural, non-epistemic factors shape our conceptual framework and explain why we hold the beliefs we have, how what people say has declarative powers which, when acknowledged by the intended or unintended audience, create overt and covert institutional facts.5

Foucault's "interrelated" but mostly anti-realist conceptions of truth have often been contrasted with and dismissed by analytic philosophers who ground their views in traditional relational and realist conception of truth. I contend that this dialectical move does not suffice to reject his insights, for even among mainstream analytic philosophers the substantial or realist conception of truth has become highly contentious, and it would beg the question to use those theories of truth as a foil against which Foucault's proposals should be evaluated (and, according to many, eventually disqualified). Versions of substantial correspondence theories which introduce "facts" or "states of affairs" to which sentences or statements "correspond" were defended by Bertrand Russell, J.L. Austin, and John Searle, among others,6 but these substantial accounts of truth should be contrasted with deflationary approaches, a family of views defended by Gottlob Frege, the later Wittgenstein, P.F. Strawson, W.V. Quine, and Paul Horwich.7 Although the deflationary view denies that the concept of truth can be analysed, it does maintain that truth is perfectly objective and that facts exist for the trivial reason that they are identical with true propositions (hence, for example, the use-equivalence of "that's true" and "that's a fact"). Facts, according to this family of theories, are mere shadows of true propositions.8 Minimalists can then plausibly deny – just as Foucault would have denied when deploying his "anti-realist" conception of truth – that "comparing" statements with non-linguistic facts makes sense. Philosophers who attack Foucault precisely because he would reject truth as correspondence with mind-independent facts therefore beg the question, not just against Foucault, but also against a prominent tradition that combines a non-relational account of truth with acceptance of the principles that we are fallible (we sometimes have false beliefs) and ignorant (there are unknown truths). Both are key features of our cognitive economy and capture our natural ontological attitude towards beliefs: there are more truths than the known truths, and what we believe is not always true. This much can be accepted by all parties.

While Foucault's cautious defenders worry too much over his alleged Nietzschean relativism, their anti-relativist critics, like Paul Boghossian for example, do not allow themselves to appreciate what makes his views on the truth/knowledge/power-complex and genealogies relevant; their insistence on an "objectivist" conception of truth and knowledge which is at least prima facie excluded by Foucault suffices to halt the discussion and dismiss his views as inconsistent or blatantly absurd.9 Other commentators hold that what Foucault says about truth often remains vague, elliptic or outright ideological, but it is not an option for them to justify Foucault's alethic revisionism by placing him in a tradition that goes back to Nietzsche or even Hegel, for such historical references merely explain why he himself held truth to be a revisable concept. Explanations of the origins of our beliefs should not be confused with justifications of those beliefs, a key distinction that lies at the very heart of Foucauldian genealogical approaches to concepts, ideas, an assumption shared by both debunking and vindicating genealogical approaches to concepts like knowledge and truth.10 And many have pointed out that the unrevised concept of truth figures prominently in Foucault's later work. Michael Peters and Tina Besley, for example, point out that Foucault provides an important analysis of parrhesia as discourse where "[…] the speaker has a specific relation to truth through frankness; a certain relationship to his own life through danger; and a specific relation to moral law through freedom and duty. In parrhesia, the speaker uses his freedom and chooses frankness instead of persuasion; truth instead of falsehood and silence; the risk of death instead of life and security; criticism instead of flattery; and, moral duty instead of self-interest and moral apathy".11 Foucault's early truth-revisionism is often seen as inconsistent with these and other remarks where truth is used with a realist attitude in mind, as when Foucault condemns political regimes that are "indifferent to truth".12 I will argue that this use of truth is entirely consistent with genealogical approaches to beliefs and "constructivist" insights in Foucault's work. To see why this is a perfectly acceptable (though of course post hoc) reconstruction of the stakes in the debate, I will briefly sketch the outlines of a minimalist account of truth. A minimalist account of truth allows us to appreciate Foucault's important insights without getting involved into distracting exegetical questions about his truth-revisionism or without dismissing them in toto based on some version of dogmatic realism.

A fruitful appreciation of Foucauldian insights with a minimalist conception of truth in mind assumes that there are good independent reasons to give truth neither a substantial "realist" or "relational" reading nor to relativize it to cultures, conceptual schemes or types of discourse. Minimalism is a family of views about truth. It rejects that truth has a deep metaphysical nature and that the concept of truth is an evaluative concept.13 Although the predicate that expresses it in English occurs in norms like "one should assert what is true" and in synoptic accounts of our epistemic practices and ambitions ("scientists aim at truth" or "truth is the goal of inquiry"), these claims are, according to the truth-minimalist, mere generalizations of particular norms such as "One should assert that grass is green only if grass is green" or "One should deny that grass is red only if grass is not red". These particular norms articulate propositions (truth-evaluable contents) and thereby make the reference to truth, and the use of the truth-predicate, redundant. The minimalist holds that you understand the truth predicate if you understand and accept all instance of the general scheme

for any proposition <p> that can be expressed in the language to which the truth predicate belongs. Frank Ramsey, one of the first do develop a minimalist conception of truth, held that the occurrence of "is true", as predicated of a belief or assertion the content of which is articulated, licences an inference to a content that recapitulates the original content, now conjoined with the original belief or assertion: if X' assertion that Michel Foucault was born in Poitiers is true, then X asserted that Foucault was born in Poitiers and, ("moreover", one might add), Foucault was born in Poitiers. This inference both illustrates what kind of inference assignment of truth to a belief under a description which articulates its content allows, and how the condition under which a belief or assertion is true, abstracts away from the believer, the act of believing, and the evidence in view of which X thought it was epistemically reasonable to hold that belief. The condition re-uses the content but kicks away reference to the believer, her state of believing, consensus with other believers, the evidence she has for her belief, the words or language she would use to express that belief, the contingent origins of her belief, the social forces that caused someone to believe it, etc. What is retained is a proposition that characterizes the world of the believer if her belief is judged to be true. Note that this approach leaves it entirely open what the metaphysical and ontological nature of the fact that p might consist in – it might be a fact of nature, a social construction, a state of affairs created by a performative use of language, or whatever. The simple point is that if one believes that p, and that belief is, or turns out to be true, then the actual world is such that p. Neither does the minimalist conception of truth come with an account of how we come to know the truths, which truths we should be interested in, or who is going to decide what we should and shouldn't know – the powers that control the distribution of true and false beliefs.14 The minimalist holds that an account of what is going to count as true, who decides what is collectively accepted as true, and the role of regimes of truth will involve careful investigation of historical, social, political and moral considerations that explain the origins of those beliefs, how they are sustained and which practices are served by having those beliefs. But such investigations are neither supported nor undermined by substantial accounts of the concept of truth.15 The philosophical excitement lies elsewhere and the battles over these issues must be fought elsewhere. This is not to deny that truth rhetoric isn't important or revelatory of what counts as true (is thought to be true, who decides what is true), but the rhetorics of truth tell us little or nothing about the nature of truth.

What is the relation between apercus such as "We aim at making true statements","Truth is the goal of inquiry","the truth will liberate us","Truth and beauty are one", and the value of truth? Minimalists or deflationists hold that little can be gained in support of these general claims and bits of worldly wisdom from the analysis of the concept of truth or from features that metaphysically constitute the property of being true.16 The platitudes and apercus (and the rhetorical moves and ploys they support) have their origin in cultural or religious doctrines, ideologies, personal and societal concerns values – the value of intellectual honesty, the delicate virtues of careful and accurate research, the superiority of scientific knowledge over folk ideologies and evidence-based practices over "local knowledge", traditional religious beliefs, etc. Qua synoptic statements they express or conversationally implicate bits of useful wisdom, but they are irrelevant in and for the philosophy of truth simpliciter – they cannot inform theories that are after an understanding the concept of truth.17 A truth-minimalist is free to argue that she doesn't adhere to some of these norms or values, or that she subscribes to them but (she will argue) precisely not on the basis of a correct understanding of the concept of truth or the metaphysical nature of the corresponding property. The alleged values and bits of wisdom they express or convey using the predicate "is (are) true" (in its capacity, qua predicate, to figure in generalizations over types or kinds of propositions or sentences) cannot be grounded in, derived from, nor rejected based on canonical use-patterns that ground the meaning of the predicate.18

Jane Heal has added to minimalism the thesis that we never seek truth for its own sake.19 When someone's epistemic actions – his or her attempts to find out something – can be described as trying to find out the truth, a more informative description – one which articulates contents and thereby renders the truth predicate redundant – is always possible, and the articulation will always refer to practices and projects. Which truths one needs to know, what it is that one wants to find out, are derived from projects which make inquiry intelligible and which explain the costs an epistemic agent wants to incur if and when trying to find out whether p. Jane Heal also points out that this view of the value of having certain true beliefs cannot be cashed out in terms of the value of a kind of "valuable" correspondence relation between our beliefs and the world required by a substantial correspondence theory of truth, for the fact that such a relation obtains is not as such the intended goal or object of inquiry. The real aim is to find out whether p – how the world is like – where the aim is guided and motivated by the inquirer's projects. The difference between a minimalist conception of truth and a more substantial relational one is that the latter misleadingly suggests that the object of an investigation is whether a certain harmonious relation ("correspondence", for example) holds between the inquirer and the world. Investigating the nature of such a relation may be the task of the epistemologist or cognitive psychologist, but it should not be confused with an investigation as to whether p. It is one thing to be interested in (the nature of) a harmonious relation between two items (one's belief, and the world), and having an interest in what's and what is not the case. Desiring the truth thus reflects an openness to the world, the desire to believe whatever the world offers to us qua believers. And this in turn, should not be confused with epistemic narcissism – the desire that what you (here and now) believe will turn out to be a true belief.20 That would amount to a self-centered desire, not one directed at the world.

Frank Ramsey held that if X' belief that p is said to be true, one asserts a conjunction – that X believes (asserts) that p – and p (cf. supra).21 (You can, by the same token, question the truth of a belief that p by asking: she believes that p, but p? – the question takes as its object the content of the belief, not the act or state of believing.) Ramsey's inference shows how closely truth and content are connected: if asserting that p amounts to or is judged to be a true assertion, you can infer that p, and if p, you can infer that any assertion to the effect that p (any assertion with the content that p), is going to be true. If you hold that someone's assertion was true, your attention is allowed to shift from the act of assertion to what is the case, and you can now legitimately ask further questions, like "Why p?" or "What would explain that p?". You do not allow yourself that kind of shift of attention if you hold that what someone asserted was false. In that case, attention shifts to the speaker herself and the etiology of her belief: What explains her mistake? Was she lying? What was her justification? Where did that belief come from? What were its origins? These are recognizable Foucauldian questions. The shift of attention from the mind of a thinker or speaker to the world is a move within the same conceptual sphere and not an inference from an inside (the conceptual sphere) to an outside that encloses it.22 That things are such and so if a belief or assertion that such and so is true specifies both the content of the thing said to be true and the world in which that very thing (a truth-bearer) is located. Within that same conceptual sphere, and now using modal language, it is possible to articulate such thoughts as: "even if p were true, we might never know it", and "although all of us believe that p, it still might not be the case that p." Ramsey's conception of truth ("a belief that p is true if and only if X believes that p and (moreover) p") also articulates the intuition that truth is a relation, but the relation is fully captured by a simple truth-functional operator – the conjunction.

The minimalist stance can tolerate that many statements (including moral pronouncements) are truth-apt because asserting that p is true just is asserting that p. It is the act of assertion, and its communicative role, that carries the burden of explanation, not the property of being true. "Why does she tell me that p?" "Why does she want me to know that?" "Why do the authorities proclaim that p?" are the relevant (and once again recognizable Foucauldian) questions. This makes room for the observation that substantial evaluations of the speaker's communicative act made pertinent by the moral or epistemic profile of the conversation are much more revealing of what they effectively want from others when producing discourse: that something was the right or proper thing to say, that a dismissal was premature, or that what she said was accurate enough for our purposes. This reflects the often-neglected observation that it is not (or not only) the truth of what I say, but what exactly I want you to know, how I want you to see things, what kind of commitment I want from you, that reflects and explains how I approach you when asserting something. It is often said that we should care about the truth in moral affairs, but we care even more about the effects of our moral communicative acts, and then truth may be less important than effectuating an intended shift in view, the change of a commitment or the adoption of different values. A moral pronouncement consists in more than just saying something I intend to be true because (or because I think) others ought to know how I judge the situation. What I want from you when I pronounce my moral views is that you to adopt my way of morally approaching a subject matter, that you coordinate your decisions and actions with my moral outlook, that you follow me. And if you agree with me, you can publicly signal that by saying that is true. The result is common knowledge that we share a value or set of values. This key point about moral discourse captures, as Lorenzini extensively documents, the nature of the performative dimension of the parrhesiastic utterance that Foucault explored in his later work.23

Minimalists hold that, if for some reason or other you find it very important that p (rather than q), nothing is added by finding it important that it is true that p (rather than that it is true that q). We value truths (true beliefs, true utterances) for reasons external to truth as such. If it is true that it is going to rain, the relevant source of value of that truth is how upcoming rain and our knowledge of it contributes to our projects. Without knowing that it is raining, you may miss an opportunity to realize your project, but then the added value resides in believing, or knowing that it rains, which is different from (and irreducible to) the value you attach to the rain itself.2425 The value attached to the fact that p cannot be reduced to the value you attach to knowing (or believing truly) that p. The goal of inquiry is not maximizing true beliefs, but to acquire the right beliefs – the ones one is, or ought to be interested in, things one needs to know in order to realize one's projects. There is no such thing as commitment to truth as such; we are committed, via our projects and practical concerns, to obtain the beliefs we need to know.26 It is then (once again) a recognizable Foucauldian question how our projects are shaped by the powers that be and how societal forces can prevent us from what we have a right to know.

Unlike the concept itself, philosophical theories and ideologies about truth are eminently revisable and volatile. The distinction between central cognitive habits in which the concept of truth figures essentially, and philosophical theories about truth will help us avoid a potentially dangerous equivocation at work in Foucault's early work. When Foucault famously said that "[w]hat I owe to Nietzsche derives mostly from his texts around 1880, where the question of truth, the history of truth and the will to truth were central to his work",27 he (and perhaps also Nietzsche) can be read as making pronouncements about (i) the history of conceptions and (substantial) theories about truth, (ii) the possibility of the concept of truth itself having a significant history, or (iii) the history of (true or false) beliefs about a particular subject matter (madness, sexuality, in the case of Foucault). The second option should be rejected.28 Our concept of truth is the same as (say) Aristotle's (that's why we understand him so well when he talks about truth), but countless philosophical theories of truth (the truth, the Truth, the use of the predicate "is true") have been developed, often in the service of philosophical views or Weltanschauungen. One has every right to seek revisions and refinements of historically situated philosophical theories of truth, but it doesn't follow that the concept of truth itself needs revision, for that would amount to an unsustainable revision of central cognitive habits. I hold that Foucault must have had option (iii) in mind, and I will argue that this option doesn't require a revision of our concept of truth.

Avoiding the confusion between philosophical theories of truth and the ahistorical character of our ordinary concept – basically the distinction between options (i) and (ii) – lies at the heart of the redescription of Foucault's project proposed in this paper. Apparent inconsistencies could have been avoided had Foucault been clearer about one (but not the only) feature of his own project: criticizing theories of truth and knowledge rather than proposing revisions of our concept of truth. A minimal or deflationary account of truth is consistent with the emerging insight that many concepts and beliefs are shaped by complex social factors the workings of which may not be fully transparent to us.

I will now focus on how Michel Foucault handles the interaction between the epistemic and the practical sphere of belief, reflected in what C.G. Prado labels the criterial use of truth in Foucault.29 Its use figures in well-known pronouncements in Foucault (1980, Power/Knowledge):

Prado understandably adds that "(i)t is [a] misperception of Foucault's criterial use as exhaustive of his conception of truth that prompts analytic philosophers to summarily dismiss Foucault's views and contentions as instancing fashionable but ill-conceived irrealist postmodern relativism."31 But the real problem with Foucault's sweeping claims was never relativism (whatever that doctrine means). It was, firstly, a potential conflation of what is true with what is held true, and, secondly, a failure to distinguish between merely asserting or pointing out what's true (or false) and issuing overt or covert declarative speech acts which create new social roles and objects and thereby make it possible to form true or false beliefs about the institutional objects and properties thus created. Foucault explicitly refers to procedures for the production and distribution of statements—the first conjunct in the minimalist conjunction (cf. supra)—and it is of those statements that we can ask such questions as: "Who pronounces them?", "Where do they come from?" and "Which practices do they explicitly or tacitly legitimize?"

For anti-revisionists about truth, the first problem with Foucault's "criterial use" of truth is that his pronouncements are ambiguous between what passes for truth – what is taken to be true, what is held true, what is believed within a community – and what is true (the second conjunct in the minimalist conjunction): believing or holding true that p does make it true, and the truth of p does not entail that someone believes that p. This distinction allows us to hold that (for example) those in charge of saying what counts as true (for them) need not have stated any truths at all; and we sometimes want to say this not because it supports a particular philosophical view of the concept of truth, but because it assumes a central cognitive habit – the distinction between what is true, and what is held or thought or said to be true, which is a datum any theory of truth should accept.32 This reading has the advantage that "what passes for truth" can be applied to any type of belief; genealogy asks the further question where what passes for truth has its origins.

Foucault's idea that truths can be produced can also be taken as a pertinent observation about how the existence of institutional objects and properties depend on shared beliefs about their existence.33 One way of creating the kind of common knowledge within a community that is constitutive for the existence of an institutional fact is by issuing declarative speech acts intended to publicly classify people. Talk of "production of truths" by "those who are charged with saying what counts as true" makes eminent sense because speech acts can create institutional facts with which various obligations and permissions are associated. Performative utterances, legal pronouncements in courts, certain ways of "officially" classifying people and their activities create institutional objects and properties about which further true or false statements can be made – statements that are true or false in virtue of the entities created by those public pronouncements. That someone comes to count as mad or a hysteric or a fundamentalist can depend on declarative utterances produced by (self)-appointed authorities and accepted by others.34 Not all of us have the privilege of being in a position to issue such declaratives. Hence the relevance of power in explanation of how pronouncements and regimes of truth can have immense societal effects.

What is even more important in this area is the fascinating and still none too well understood phenomenon that people's beliefs may unintentionally introduce social or institutional objects.35 That "discourse" creates the very phenomenon it described may itself be a covert phenomenon and it is a major achievement when an unintended institutional phenomenon, often presented as and taken to be a "brute" or "natural" fact, is publicly exposed as being created by accepting declarative pronouncements, a phenomenon which may result in sometimes quite radical conversions within a community or group of believers. One of Foucault's key insights was that it is not always transparent to them what people do with words and how far the effects of words and discourse can reach: are they merely describing objects, events or persons when they are declaring something to be the case, or propagating a certain classification? Descriptive speech acts can be disguised as declaratives, and declaratives can be played down as mere observations. Foucault's "power-produced truths" must therefore not be interpreted in the strong sense that all truths are constructed or that truth can be "monetarized"; it should be understood in the very real and almost mundane sense that institutional facts and the practices in which they figure are intentionally or unintentionally constructed, that we can form true or false beliefs about those institutional facts, that we can discover how, where and when they were created, who was responsible for maintaining them, and under which specific historical circumstances their institutional and constructed nature became fully transparent to those involved. (Unlike Foucault, I think it is an open, empirical question whether the phenomena he himself thought were constructed – psychiatric conditions, for example – were indeed "created").36

The key insight is that Foucault's observations do not require a substantial revision of the concept of truth in the direction of relativism or idealism. My minimalist reading preserves the important Foucauldian insight that "discourse" sometimes does create new facts, but it doesn't follow from the fact that someone was authorized to make certain declarations, or that accepting certain pronouncements was the right thing to do, that such acts and classifications were justified. As Rae Langton aptly observes, we can have knowledge about things which should not have been brought into existence as institutional facts via declarative statements. Her example is pornography: pornographic discourse (in the broad sense of speech, graphic and sexually explicit material) "can help create knowledge" because "the beliefs of the powerful become (proven), in part because the world actually arranges itself to affirm what the powerful want to see".37 These and other "looping effects" are now well-known and intensively studied in, for example, game-theoretical approaches to institutions and the proliferation of labels.38

The constructivist use of truth in Foucault explains how a proposition can come to be a candidate for truth in a community. Here is a key (and famous) pronouncement:

As Prado is keen to point out, "Foucault's constructivist use [of truth] does not allow a distinction to be drawn between underlying truth and apparent truth: the constructivist use deals with the production of all truth", and "there is (according to this conception of truth, FB) no truth other than discursive truth".40 This conflicts with a central cognitive habit explored earlier, viz. that what someone believes to be true need not be true: if a power makes a proposition or belief or statement true (truth as the effect of power), the distinction between truth and what is believed to be true collapses.

But there is also a reading of Foucault that reveals an important insight that is fully accessible for a minimalist. Notice, first, that many have pointed out that making the facile distinction between truth and ideology ("power produces ideology-laden beliefs") was not Foucault's intention, and Foucault leaves no doubt about what he means: "[Ideology] always stands in […] opposition to something else which is supposed to count as true. Now I believe that the problem does not consist in drawing the line between that in a discourse which falls under the category of scientificity or truth, and that which comes under some other category, but in seeing historically how effects of truth are produced within discourses that are neither true nor false."41 Foucault denies that ideological distortion of underlying ahistorical truth is the object of his analysis. And there are further problems with the superficial reading. Does power or discourse also produce falsehoods? How should we distinguish between powers (discourses) that produce truths and powers (discourses) that produce falsehoods? What if two conflicting discourses are at work in the same society, about the same subject matter? Foucault seems to have excluded this possibility, as Merquior rightly points out – but on what grounds?42

The real insight behind Foucault's genealogical approaches to beliefs, conceptual frameworks and regimes of truth can be clarified by a distinction that lies at the heart of genealogical methods in general: the distinction between epistemic reasons (reasons for believing that belief that p is true, that is, according to the minimalist, reasons for adding to that belief the conjunct "and, moreover, p") and non-epistemic explanations of why and how we came to believe and know what we believe and know, a distinction that doesn't require a revision of the concept of truth and doesn't appeal to substantial, non-minimal accounts of truth. Following Peter Railton and Robert White, we distinguish epistemic reasons for believing that p and causal social or institutional explanations of why an agent came to believe or know that p or became a p-believer (a certain type of believer).43 That distinction applies to almost all beliefs. Take Nina's knowledge of the simple mathematical truth that 7 + 5 makes 12. Being asked for a justification, we would say that she knows rules of addition and can competently apply them to 7 and 5. Asking for an explanation, we would say that Nina's single mother managed to gain enough funds to send her oldest daughter to a primary school. The distinction between epistemic reasons for/against X' beliefs and (in this case, social) explanations of why X believes or knows what she believes is not always fully appreciated because the question "Why does X believe that p?" is ambiguous between the epistemic and the explanatory reading.44 Questions like "Which factors explain why X knows that p and Y doesn't know that p?", "Why X never even heard about the possibility of p's being true?", "Why was X deliberately deprived of coming to knowledge that p?" or "Why X was forced to know that p?", give rise to social explanations of (for example) why people came to entertain beliefs we now recognize as obviously false, racist, or sexist. One key question Foucault never lost sight of had precisely to do with social explanations of attitudes– explanations of why and under which social or political circumstances people became p-believers and (to complicate matters) under which circumstances their acceptance of (un)intended declaratives (cf. supra) turned them into p-believers with true and false beliefs about the institutional facts created by those declaratives.

The distinction between epistemic justifications and social explanations of beliefs need not always be transparent to us (witness the ambiguity in "How does she know that p?"), but once the distinction is firmly in view we can once again see Foucault's questions about "the production, regulation, distribution circulation of statements" (my italics, cf. supra) as articulating concerns that do not require revision of the concept of truth. His point was simply that our justifications for the truth of beliefs say little or nothing about what explains why (a group of) persons came to be believers of those truths, and, conversely, that a surprising social explanation of how beliefs and knowledge "spread" within a community and who is in charge of controlling where and when it spreads need not preclude the possibility that those beliefs can constitute knowledge – justified true beliefs.45 Having a belief and sharing it with others is subject to social explanations, and for some beliefs – beliefs we have about sexuality, madness, politics, the legal order, criminal behaviour, gender, etc., for example – it is important not just to find out how agents justify their beliefs but also to track the origins of those beliefs. Social explanations explain what we tend to find important to know, what we identify with, what we deem useful to know, what we are unjustly denied to know, and they point out that we are not always aware of these inclusionary and exclusionary non-epistemic constraints under which we epistemically operate.46 While reasons may be transparent to believers, the social explanation of why X or Y has a specific belief or cluster of beliefs it is a matter of empirical research and not a priori accessible to the believer herself.47

There is thus no reason to confuse epistemic reasons for believing that p with historically situated practices, traditions, and institutional realities ("regimes of truth") that explain why individuals or segments of society hold the beliefs they have.48 Surprising social explanations of why we know what we know are consistent with excellent (and, of course, not so excellent!) epistemic justifications for what we know. But, as Foucault would certainly have recognized, making this distinction at a reflective level is one thing, making it transparent to those who have beliefs and making them sensitive to this distinction is not as easy as we might wish it to be. That is, I contend, why the truth/ideology distinction was important to Foucault. Ideologies precisely tend hide the justification/social explanation distinction from those who are in thrall of certain beliefs.

This deflationary reading also explains why, as Barker suggests, the Foucauldian recommends that we should shift our attention away from the question "Is this true and what does it mean?" to the question "What are its effects?".49 I take this to recommend looking at what kinds of effects having knowledge has, not in the sense in which theoretical reason complements and underpins practical reason (justified true beliefs tends to produce successful actions), but in the non-epistemic sense that acquiring beliefs has a price: what is the effect of distributing or withholding knowledge? What price are people willing to pay for knowledge? Which extra-epistemic (social, political) factors set our epistemic agendas? Who determines which truths we ought to have access to?

The "perspectivist use of truth" is the most delicate aspect of Foucault's alleged anti-realism, and it is often based on a superficial reading of many of Foucault's early pronouncements. Foucault's perspectivism denies "that we can meaningfully assert that things are a certain way independently of how we take them to be", a position which derives from Nietzsche, who famously asserted that "there are no facts, only interpretations".50 According to Prado, Foucault's perspectivist use of truth is "limited to [a] […] denial of the possibility of a global or holistic description within which diverse perspectives could be 'reconciled' or 'rationalized' as so many true but incomplete points of view",51 and it "[…] has less to do with the issue of an extralinguistic or extradiscursive reality. The perspectivism of Foucault's perspectivist use is the denial of the possibility of descriptive completeness".52

Be that as it may, a minimalist has a disarmingly simple reply to the allegation of perspectivism. When asserting that p, one inevitably presents oneself to one's interlocutor as believing that p.53 It is thus a conceptual truth that when a speaker meaningfully asserts that p – that things are a certain way – she publicly presents herself to others as taking things to be a certain way. A more contentious reading would hold that Foucault's perspectivism is better characterized by a rejection of the possibility of referring to and discerning a determinate state of being beyond our interpretations, which situates him in a long idealistic tradition, starting with Berkeley and running via Kant and Nietzsche to postmodernism, which accepts one or another version of Kant's classic argument:

But this style of argument has now long been shown to be fallacious, as David Stove, Alan Musgrave and James Franklin have forcefully argued.55 It deduces a contingent proposition from a necessary truth, and it presents a necessary condition for thinking or cognitive activity ("when we think about the world, we must inevitably rely on our concepts") as preventing us from having genuine knowledge about the world. The obvious reply is that while we inevitably think with concepts – ("our concepts", but that is trivial addition) and form true or false beliefs about the world using those concepts,56 it doesn't follow that those concepts somehow prevent us from acquiring knowledge, just as the fact that we see things with "our" eyes somehow prevents us from perceiving items in our surroundings or that we dig holes with a spade doesn't prevent us from digging holes in the ground. It is not a condition on knowledge that we be in "unmediated touch" with something "as it is in itself", and it is not even clear what the expression "things in themselves" means. As Peter van Inwagen forcefully put it, "there are all sorts of adverbs and adverbial phrases that can be meaningfully used to qualify 'has', for example when the object has a property: 'apparently', 'essentially' and 'according to what this audience believes', for example, but 'in itself' is not one of them […]. [The addition] has no significant connection with the words that surround it."57

This makes room for a final important insight behind Foucault's perspectivism. A reading which doesn't invite commitment to the idealistic fallacy can connect the trivial observation that I think with my concepts (exactly in the sense in which I always speak my language or look at the world with my eyes) with the deflationary readings of Foucault's criterial and constructive uses of truth proposed earlier. It is precisely because we think with concepts, and because those concepts are constituents of our beliefs, that social explanations of why we believe what we believe are possible and explanations of how societal forces, institutions and authorities have a grip on how we conceptualize things, make sense. As Sally Haslanger put it in the context of a critical discussion of the spreading of racist and sexist concepts, "what concepts and so what ideas we have is the result of social-historical events; who is in the business of denying that? It would seem to be a matter of common sense that concepts are taught to us by our parents, our teachers, the newspapers, 'those who are in charge', that different cultures have (to a certain extent) different concepts, and that concepts evolve over time because of historical changes, science, technological advances, and so on. Even those who believe that we have concepts that cut nature at its joints, will agree that a group may have those concepts through socio-historical processes."58 These platitudinous facts do not entail or support a revision of our concept of truth, but they make it immensely urgent to further explore what explains that one can become, in Railton's apt description, a p-believer (cf. supra): our beliefs and the concepts that characterize their contents may have quite unexpected social or institutional explanations and Foucault's work was, among other things, devoted to exploring and documenting these phenomena.

The correct but insufficiently appreciated reply to the charge of relativism and constructivism about truth is that Foucault's insight do not presuppose or require that truth itself be made relative to a culture, a conceptual scheme or a perspective. A minimalist conception suffices to isolate the idea that we have beliefs, think those beliefs are true, and that having those beliefs – especially when they are contested, socially relevant or simply contentious – make genealogical explanations relevant. That we have culturally acquired certain concepts and beliefs and that many concepts that refer to social or cultural classifications have their origin in intended or unintended declarative speech acts, that the institutional facts they intentionally and unintentionally create have a contingent existence and, perhaps most importantly, that it is not always fully transparent to us that the facts so created are institutional facts, or that they should have been brought into existence, were Foucault's key insights in his early work. It contained the germs of important proposals for framing the sociology and epidemiology of beliefs and the role of declarative speech acts in introducing and maintaining social and institutional facts. These insights do not require a contentious revisionist reconceptualization of our familiar concept of truth that minimalists have recovered. Truth should never have been the issue.