Brassier - Correlation, Speculation, and the Modal Kant-Sellars Thesis (Chapter 3 from The Legacy of Kant in Sellars and Miellassoux)

3 Correlation, Speculation, and the Modal Kant-Sellars Thesis Ray Brassier The motto of the age of science might well be: Natural philosophers have hitherto sought to understand ‘meanings’; the task is to change them. Wilfrid Sellars, “Counterfactuals, Dispositions, and the Causal Modalities”, 288 1. Introduction In After Finitude, Quentin Meillassoux proposes a speculative overcoming of “correlationism”—his term for all those variants of post-Hegelian philosophy rejecting the absolute.1 By ‘speculative’ Meillassoux means “every type of thinking that claims to be able to access some form of absolute” (Meillassoux 2008, 34). Because it prohibits any appeal to a necessary being, correlationism proscribes speculation. But although all metaphysics is speculative, “not all speculation is metaphysical and not every absolute is dogmatic” (Meillassoux 2008, 34). While metaphysical speculation identifies the absolute with a necessary being, Meillassoux’s speculative overcoming of correlationism prises apart necessity and being. The absolute, understood as what is necessary, is not what is but what could be. This is Meillassoux’s absolutization of contingency. Its startling consequences are easily overlooked until one remembers that Meillassoux has decoupled possibility from quiddity, so that what could be is entirely unconstrained by what is; indeed, for Meillassoux nothing is impossible, save something necessary. Meillassoux’s overcoming of correlationism reasserts reason’s a priori grip on being conceived as pure potentiality rather than as actual substance; the speculative rationalism ensuing from this overcoming jettisons the principle of sufficient reason as an empiricist prejudice. Thus Meillassoux’s rationalism embraces emergence ex nihilo: ontological discontinuity is not the exception but the rule. In this essay, I want to compare and contrast Meillassoux’s unfastening of reason from experience, and what I take to be its exorbitant consequences, with a version of rationality that, precisely because it retains something of empiricism, endows reason with the power to reshape experience. This is the version of rationality defended by Wilfrid Sellars. Sellars

68 Ray Brassier maintains a Kantian account of modality that situates it squarely within us rather than within things-in-themselves. Yet Sellars also upholds the reality of the in-itself by insisting on reason’s relation to sensibility, in contrast to Meillassoux, for whom reason is able to access the in-itself on its own. But, as we will see, Meillassoux unyokes reason from sensibility in one moment only to re-envelop thinking within sensing in another. Perhaps because he views Kant’s epistemic demarcation of the domains of thinking and sensing as part and parcel of the correlationism he wishes to repudiate, Meillassoux allows reason to transcend sensibility in order to access the absolute. It is by rejecting this transcendent conception of the in-itself as intelligible absolute that Sellars salvages the rational core of correlationism while avoiding its epistemic relativism, as well as the irrational consequences of Meillassoux’s speculative alternative. The first section of this essay will discuss some of the difficulties entailed by Meillassoux’s speculative thesis about the necessity of contingency. The second section will introduce the modal Kant-Sellars thesis and suggest that the distinction between semantic and pragmatic inference provides the germ for a powerful Sellarsian alternative to Meillassoux’s speculative exit from correlationism. The third and final section presents Sellars’ pragmatist rationalism as a way of reasoning about reason that exhibits greater responsiveness to contingency than Meillassoux’s speculative rationalism. 2. Meillassoux’s Speculative Overcoming of Correlationism After Finitude has been much criticized for disregarding the complexities of philosophy’s history since Hegel. But this disregard is deliberate. Like Descartes’ Meditations (upon which it is partially modeled), After Finitude eschews erudition in order to stage an intervention in contemporary philosophy. Meillassoux sets out a critical diagnosis of philosophy’s present by constructing a dialectical narrative leading from Kant’s ‘weak’ correlationism to Hegel’s speculative idealism, and from speculative idealism to ‘strong’ correlationism, exemplified by contemporary ‘post-metaphysical’ philosophy (whose patron saints are Heidegger and Wittgenstein). The decisive turning point in this dialectical narrative is Meillassoux’s master argument against strong correlationism, establishing the necessity of contingency and ushering in his own ‘speculative materialism.’ I have rehearsed the details of Meillassoux’s master argument elsewhere,2 so here I will focus on the three dialectical stages leading up to it. The three stages are the following: 1. Weak correlationism (Kant): we can know the for-us but we can only think the in-itself. 2. Speculative idealism (Hegel): we can know that what is for-us is also in-itself.

Correlation, Speculation, and the Modal Kant-Sellars Thesis 69 3. Strong correlationism (Meillassoux has in mind post-metaphysical philosophy broadly construed: not only Habermasian, but Nietzschean, phenomenological, deconstructionist, and pragmatist): the speculative identification of the for-us with the in-itself is only for-us. I will briefly elaborate on each of these, focusing solely on the characteristics singled out by Meillassoux. Weak Correlationism Kant carries out a Critical demarcation of dogmatic absolutism, here understood as metaphysical realism: all objectivity is correlated with transcendental subjectivity. The reality of objects (things, properties, relations) is relativized to the transcendental correlation. The in-itself exists as non-contradictory ground of appearances, but its structure is unknowable, i.e., it is non-objectifiable. Categories and forms of intuition apply only to appearances, not things in-themselves: transcendental structures are necessary for knowing appearances, but they are not features of things-in-themselves. What is significant for Meillassoux at this first stage is that transcendental reflection does not reveal why space and time are the only two forms of intuition or why there are twelve categories. Meillassoux glosses this feature by saying that there is no sufficient reason to account for such a fact. Thus the transcendental is characterized by a second-order contingency: not the contingency of empirical fact but the contingency of the condition for anything’s counting as an empirical fact: this is what Meillassoux refers to as the ‘facticity’ of the correlation. Speculative Idealism Meillassoux takes Hegel’s principal critical claim vis-à-vis Kant to be that what is constitutive of the for-us (the correlation) also manifests the in-itself (the absolute). Traversing the cognitive moments of immediacy, mediation, and mediated immediacy allows reason to think the processual identity of the for-us and the in-itself. Through this processual identification, reason grounds its own access to being: this is the absolutization of the correlation as necessary, rather than merely factical. Through this process, correlation is temporalized (historicized); but this temporalization has a rational structure: there is a reason (necessity) underlying the unfolding of appearances to and for consciousness (contingency). Speculative idealism relativizes the Critical absolute (the in-itself qua unknowable) to the correlation, reintegrating it as the in-itself of the correlate. Because the facticity of the correlation is in-itself for-us (we phenomenological subjects), it can be rendered knowable, and hence known to be necessary in and for itself through us.

70 Ray Brassier Strong Correlationism In Meillassoux’s narrative, just as speculative idealism supersedes weak correlationism, strong correlationism will supersede speculative idealism. It does this via two moves: first it relativizes every dogmatic absolute (every in-itself) to the correlation: the in-itself is always for-us. Then it blocks the speculative absolutization of the correlation by insisting on its contingency: the correlation could not have been or could have been otherwise: its necessity must remain unknowable. In other words, the correlation cannot become in and for itself through us. The question then is: what is it that prevents it? The answer given by strong correlationism is: facticity, or what Heidegger called finitude. Finitude is second-order contingency: the facticity of the condition for empirical fact. It is the response to the following aporia. Against metaphysical realism, strong correlationism must relativize every dogmatic absolute by insisting on the circle of correlation: every given absolute is posited as given. But against speculative idealism, strong correlationism must also de-absolutize the correlation by insisting on its facticity: the correlation could have been otherwise or could not have been. Thus the facticity of the correlation must be absolutized in order to de-absolutize the correlation: this is what prevents the correlation from becoming in and for itself. This is the point at which Meillassoux makes his speculative move. The non-existence of the correlation is not merely epistemically possible (for us), but alethically possible (in itself). This alethic possibility indexes a pure ontological potentiality uncoupled from the determinacy of substance: Meillassoux calls it “surcontingence” (supercontingency) or “absolute time.” Meillassoux’s speculative absolutization of facticity is the conversion of an epistemological thesis about the unavailability of reasons into an ontological thesis about the existence of a potentiality without reason. This pure, alogical potentiality is the root of possibility considered as real in itself, or as absolute (Meillassoux does not distinguish the latter two terms). Thus for Meillassoux: [. . .] ultimately the matter of philosophy is not being or becoming, representation or reality, but a very special possibility, which is not a formal possible, but a real and dense possible, which I call the ‘peut-être’—the ‘may-be.’ (Meillassoux 2014, 27) Because the absolute beyond the correlation is not a thing that is but the ‘may-being’ of every thing, Meillassoux describes his position as a “speculative materialism”: [W]hat we will discover outside the correlation is very different from the naïve concepts of things, properties and relations. It is a reality very

Correlation, Speculation, and the Modal Kant-Sellars Thesis 71 different from given reality. That’s why, ultimately, I prefer to describe my philosophy as a speculative materialism, rather than as a realism: because I remember the sentence of Foucault, who once said: “I am a materialist because I don’t believe in reality.” (Meillassoux 2014, 19) Materialism ascends from the dogmatic to the speculative register by relinquishing the metaphysical reification of materiality, whether as atom or force. It breaks with metaphysical realism by raising contingency, conceived as absolute potentiality, above chance, understood as the actual swerve of atoms or the actual play of forces. Supercontingency The crucial move in Meillassoux’s speculative deduction of absolute contingency is his conversion of epistemic contingency (what is contingent for us) into alethic contingency (what is contingent in itself). Pete Wolfendale suggests that the dialectical bridge that Meillassoux tries to construct between the for us and the in itself is vitiated by a failure to distinguish between the epistemic contingency of nomological necessity—our uncertainty about the necessity of the laws we know—and the logical necessity of nomological contingency—the logical impossibility of inferring universal necessity from finite regularities: [. . .] the absolutization of facticity hinges upon the overlap between epistemic modality and alethic modality when it comes to logical necessity, insofar as it converts the epistemological conditions of factual disagreement into the ontological structure of facts. In turn, this suggests that the inference from absolute facticity to absolute time hinges upon the gap between (atemporal) logical necessity and (temporal) nomological necessity, insofar as it converts the logical possibility that things could be otherwise into the nomological possibility that things could become otherwise [. . .] (Wolfendale 2014, 124) Wolfendale’s remark helpfully pinpoints what we will call the HumeMeillassoux dilemma. Hume’s skepticism entails the epistemic contingency of nomological necessity: lawlike necessity is rooted in contingent fact. Meillassoux’s rationalism entails the logical necessity of nomological contingency: the contingency of law is rooted in reason (Meillassoux’s ‘principle of factiality’). Meillassoux’s speculative resolution of the dilemma amounts to the ontologization of skepticism: where Hume’s empiricism took the ‘secret powers’ (sufficient reason) of things to be unknowable because the contingency of fact vitiates the necessity of law, Meillassoux’s rationalism takes them to be knowable because the contingency of fact is the only law.

72 Ray Brassier The absence of reason for us is converted into the presence of unreason in things. Since unreason is a positive feature of things-in-themselves, nature is necessarily discontinuous. The requirement of explanatory continuity between domains of reality (physical, biological, psychological, cosmological, etc.) becomes an empirical prejudice to be supplanted by the rational apprehension of necessary discontinuity: there are no explanatory gaps, only ontological emergences: As long as reason is identified with thinking the constancy of laws, it remains impossible to think rationally about the advent of life in matter, because it cannot be understood how the lifeless can produce a qualitative multiplicity of affects and perceptions from a certain ‘molecular geometry’ [. . .] This essential excess of life and thought beyond matter implies a scission that ruptures all continuity, leaving the divine and the soul free rein to fill the resulting chasm. Nevertheless, such ‘mysteries’ collapse once the qualitative component of life is identified with the advent of a Universe of cases that were in no way contained in the universe previously. Such a Universe gives us the advent of a pure novelty whose possibly regular concordance with material complexes does not obliterate the radical excess found in the affective qualities of suffering or jubilation and the travails of life or consciousness. From this we recognize that the qualities inherent in the affective and perceptive world of life are immediate signs that becoming makes its novelties emerge from nothing. A pain or pleasure does not pre-exist its effectuation in the living, because life itself does not pre-exist the material components that accompany its advent. (Meillassoux 2011, 180) What Meillassoux affirms here is the absolute excess of the effect over the cause: since empirical constancy is not subsumed by any law harboring the sufficient reason that would entail the existence of the effect from the existence of the cause, neither cause nor effect can be logically coordinated: their diremption is as reasonable as their conjunction. The absence of lawful continuity within domains of being—physical, biological, psychological, cosmological—is matched by the absence of lawful continuity across domains of being, from the physical to the biological, from the biological to the psychological, etc. Thus Meillassoux’s ‘Worlds’ (matter, life, thought, justice) are utterly discontinuous. The advent of the living is the anomalous rearrangement of lifeless matter required to support it; the advent of thought is the anomalous reorganization of thoughtless life required to enable it.3 Science is disunified because being is not One (not-all): there can be no single science (whether metaphysical or empirical) exhaustively describing and explaining everything that is. Together with the intelligible discontinuity of being, the contingency of laws entails that the empirical sciences are essentially descriptive, since

Correlation, Speculation, and the Modal Kant-Sellars Thesis 73 scientific laws and the explanations they enable remain beholden to empirical regularities: [. . .] the contingency of laws is manifested in their original caesuras, and that is what legitimates the essentially descriptive method of the empirical sciences. It is precisely in these descriptions of constancy that the epistemological notions of determinism and probabilism are legitimate. (Meillassoux 2011, 186) Since laws are facts, not grounds, descriptions of constancy circumscribe the domain of validity proper to the determination of one empirical phenomenon by another, as well as its probabilistic auxiliaries. Since the ambit of scientific explanation is circumscribed by the resources of empirical description, logical-discursive explanation articulated around distinctions between epistemic and alethic modality—between necessity and contingency in what we know, and necessity and contingency in reality—is superseded by the mathematical inscription of contingency, i.e., the claim that it is the ‘meaninglessness’ of the mathematical sign that best describes reality in itself because only it successfully inscribes its factial contingency.4 If explanatory change is propelled by changes in descriptive vocabulary, but not the other way round, theory change is caused but has no reason. Scientific explanation follows blindly in the wake of empirical description. This has one very important consequence for what will follow: the tissue of mandatory conceptual connections within which empirical-descriptive vocabulary is embedded is subject to abrupt and arbitrary rearrangements that cannot be conceptually adjudicated. Conceptual change is without why: the rearrangement of semantic webs cannot get an explanatory grip on empirical anomaly. The Factialization of Quiddity Advent ex nihilo entails another difficulty for both metaphysical and scientific explanation. It renders ontological classification anomalous: differences in ontological structure are no longer governed by logical differences: this is the factialization of quiddity. This means that differences in being, such as the difference between matter and life (between the animate and inanimate), cannot be deduced a priori. Factiality yields no principle capable of legislating differences between kinds of thing. How, then, do we establish the difference between those qualities characteristic of lifeless matter, and those that are characteristic of sensate life? Meillassoux’s insistence on the brutely sui generis character of quality compounds the difficulty: All quality as quality is without why, since none of its content refers to anything other than the advent ex nihilo of its being [. . .] A red is without why because no material underpinning can ever tell us how this

74 Ray Brassier red is red. A red is without why because there was nothing of this red in the world prior to its advent that would provide us with a reason by relating it to a pre-red where it had always been contained [. . .] Quality is a pure fact referring only to itself, and as such it displays the irrecuperable excess of a Universe of cases (namely, that of the living) on another (that of material configurations). It is given as a brute existence that essentially cannot be deduced, and which refers to its actuality alone [. . .] The remarkable thing is that the brute facticity of quality is where the inexistence of the Whole is immediately given. For the facticity of quality refers to its advent ex nihilo, which refers in turn to the absence of an originary Whole from which it could be inferred with complete necessity. (Meillassoux 2011, 181) If the ‘brute facticity’ of the sensation of redness is said to ‘give’ [donner] the inexistence of the Whole it is presumably by virtue of being directly apprehended in experience. But how could the sensation of redness, which is merely the empirical apprehension of a qualitative difference within a World, serve as the paradigm grounding the speculative (which for Meillassoux means rational) legitimacy of absolute qualitative differences between Worlds? Even if quality is a pure fact referring only to itself, there is still a question about the cognitive apprehension of such facts. Awareness of a quality is not awareness of it as the quality that it is. Failure to distinguish awareness of from awareness as short-circuits sensing and thinking. Meillassoux invokes the ‘brute existence,’ the sheer actuality, of qualitative difference as though it were self-presenting. By itself, the existence of such absolute qualitative discontinuities does not guarantee that they manifest themselves directly, whether in sensation or intellection—unless one assumes that the mere awareness of red suffices to guarantee awareness of it as “a pure fact referring only to itself.” But facts are thought, not sensed. Meillassoux seems to characterize the brute facticity of quality, its advent ex nihilo, as an ontological datum that simply gives itself in experience. Why should our experience of the difference between the sensate and the insensate be a reliable guide to ontological discontinuity? The point is not to query Meillassoux’s rejection of the hylozoist claim that matter is really alive, or its obverse, the eliminativist claim that life is really inanimate—he is surely right to reject peremptory metaphysical attempts to bridge the divide from one side or the other. The point is rather that the logic of his account calls for a criterion of discontinuity irreducible to the register of empirical description. By Meillassoux’s own lights, advent ex nihilo should be empirically undetectable relative to the constancy with which it breaks; if it is not, it is just another empirically describable fact, rather than a manifestation of ontological factiality, which is supposed to be the purely intelligible contingency of fact. Meillassoux elevates reason above sensation only to re-envelop thinking within intuition.

Correlation, Speculation, and the Modal Kant-Sellars Thesis 75 3. The Modal Kant-Sellars Thesis Like Kant, Sellars rejects modal realism, i.e., the metaphysical claim that necessity and contingency exist in things, independently of our representation of things. But Sellars subjects Kant’s transcendental account of modality to a pragmatist twist. Modal vocabulary allows us to regulate the explanatory frameworks within which our empirical descriptions are deployed, and in doing so it endows our theories with a rational responsiveness to the world’s unresponsiveness (i.e., to anomaly), enabling us to change our theories so as to maintain our cognitive (which also means practical) grip on the world. The upshot is that meanings are made not given: they are co-created by us in collaboration with the world, and this is precisely why they can be changed. Modal Expressivism The expression “modal Kant-Sellars thesis” is Brandom’s coinage.5 The thesis claims that the role played by modal vocabulary is expressive rather than descriptive: it does not describe properties of objects; rather, it allows us to say what we are doing when we make assertions about objects. Modal expressivism is the claim that modal vocabulary renders explicit features that are implicit in the use of all ordinary empirical descriptive vocabulary. Every empirical descriptive concept has modal consequences: these are the necessary conditions for its correct application that are rendered explicit using subjunctive conditionals. For example, the conditions for the correct application of the empirical descriptive concept ‘organism’ are rendered explicit in subjunctive conditionals such as: If this organism were to ingest a toxic substance, it would be harmed. But the inferences codified by these conditionals are non-monotonic, i.e., they are not robust under arbitrary addition of auxiliary premises. For example: If this organism were to ingest a toxic substance, and it had been given an antidote to that substance, it would not be harmed. All empirical descriptive properties are ‘modally involved,’ i.e., they locate objects in a space of implication. This space of implication is the inferential web within which assertions deploying descriptive terms commit the speaker to other assertions. For example: This is a cat  this is a mammal  this is a living thing. As Sellars puts it: It is only because the expressions in terms of which we describe objects, even such basic expressions as words for the perceptible characteristics

76 Ray Brassier of molar objects locate these objects in a space of implications, that they describe at all, rather than merely label. (Sellars 1958, 306–307) While description involves conceptual classification (characterizing X as Y), in which justification is implicit, labeling merely separates and groups: it differentiates according to a reliable responsive disposition, or what Brandom calls ‘RDRD’ (reliable differential responsive disposition). To describe X as Y is to commit oneself to justifying this description by giving a reason explaining why it is correct to classify Xs as Y. But it is an essential feature of the inferential relations in which descriptive concepts must stand that they can be appealed to in explanations and justifications of further descriptions. Thus the expressive role distinctive of alethic modal vocabulary is to make explicit these explanatory and justificatory relations. Alethic modal vocabulary allows us to say what we are doing when we use descriptive vocabulary. It explains why describing something as A entails describing it as B. Modal assertions provide ‘inference tickets’ from one empirical characterization to another. To say ‘As are necessarily B’ is to say that one is justified in moving from the claim ‘This is A’ to ‘This is B.’ Thus, to say ‘Dogs are necessarily mammals’ is to say that one is justified in moving from the claim ‘This is a dog’ to ‘This is a mammal.’ To appreciate this point is to realize that what we call ‘the causal principle’—every event has a cause, also known as the principle of sufficient reason—states in the material mode what is actually a formal feature of language use, viz., that the use of ‘necessarily’ or ‘possibly’ when conjoining descriptive predicates in empirical assertions either does or does not furnish us with an inference license from ‘This is A’ to ‘This is B.’ To describe correctly is also to explain (to give a reason for) what one is doing in describing. This is why Sellars insists that describing and explaining go hand in hand in conceptual practice: The causal principle [. . .] gives expression to the fact that although describing and explaining (predicting, retrodicting, understanding) are distinguishable, they are also, in an important sense, inseparable [. . .] The descriptive and explanatory resources of language advance hand in hand, and to abandon the search for explanation is to abandon the attempt to improve language, period. (Sellars 1958, 306) Once we understand that every description commits us to its justification, we realize that there is no descriptive vocabulary that does not implicitly harbor an explanation as to why things are described as being thus and so. To say ‘this is red’ is also to commit oneself to saying ‘this is colored,’ where ‘coloredness’ implies a theory explaining the difference between colored and colorless things. Thus our description of something as red is inseparable from our understanding of the nature of color (however imperfectly); an

Correlation, Speculation, and the Modal Kant-Sellars Thesis 77 understanding that is corrigible precisely because its conceptual character is of a piece with the obligation not only to justify, but to seek out better justifications for our claims. The interdependence between descriptive awareness and explanatory understanding obliges us to correct, revise, and refine our descriptions by developing better explanations for why things are thus and so. The descriptive dimension of language use operates in concert with evaluative and explanatory dimensions that require the constant rectification of use. To relinquish explanation would be to abjure description. Conceptual and Material Modality However, the claim that modal inferences are implicated in the meanings of ordinary empirical descriptive concepts seems to blur the line between conceptual necessity and possibility, on one hand, and material necessity and possibility, on the other. Conceptual necessity and possibility are knowable a priori; material necessity and possibility are only knowable a posteriori. Does modal expressivism commit us to a version of the synthetic a priori that makes real necessity conceptual? The dilemma can be formulated as follows: if conceptual classification is determined by reason independently of experience, how can it be answerable to experience? If it is determined by experience independently of reason, how can it be answerable to reason? Brandom points out that this gives rise to the following inconsistent triad: 1. Physical or causal necessity and possibility are a kind of conceptual necessity and possibility. 2. Physical or causal necessities and possibilities must be established empirically. 3. Conceptual necessities and possibilities can be established a priori. (Brandom 2015, 185) 1 and 2 are incompatible with 3; 2 and 3 are incompatible with 1; 1 and 3 are incompatible with 2. Sellars rejects 3: concepts are rules to which we bind ourselves, but we do not know in advance all the implications of the concepts/rules to which we commit ourselves. We discover what applications of empirical concepts are correct through the same process in with we discover which inferences connecting those concepts are correct. Thus, in Sellars’ words: [W]hile one does not inductively establish that A P[hysically]-entails B by armchair reflection on the antecedent ‘meanings’ of ‘A’ and ‘B’, to establish by induction that A P[hysically]-entails B is to enrich (and, perhaps, otherwise modify) the use of these terms in such wise that to ‘understand’ what one now ‘means’ by ‘A’ and ‘B’ is to know that A P[hysically]-entails B. (Sellars 1958, 287)

78 Ray Brassier Brandom makes the same point in the following way: “To find out what the contents of the concepts we apply in describing the world really are, we have to find out what the laws of nature are. And that is an empirical matter” (Brandom 2015, 186). Saying and Doing As we have seen, modal statements do not say that the relation of physical entailment between A and B holds; rather, they endorse a pattern of inference from claiming ‘x is A’ to claiming ‘x is B.’ The endorsement holds at the level of assertion, not of the objects of assertion. It is expressed in discursive practice without being explicitly stated in discourse. Thus modal vocabulary allows the practical ratification of patterns of inference.6 It stands to ordinary empirical descriptive vocabulary as a pragmatic (rather than semantic or syntactic) metavocabulary.7 This is just to say that modal vocabulary lets us say what we are doing when we describe something as thus and so. The distinction between saying and doing highlights the difference between semantic and pragmatic inference. For one can give a reason by doing something, as well as by saying something. The pragmatic ratification of inferential necessity between assertions is not the same as the conceptual ratification of necessary connections between objects. What is being done by saying is not the same as what is being said; what is implied by saying (pragmatic inference) differs from what is implied by what is said (semantic inference). Failure to observe this distinction can have metaphysical consequences: for instance, it can lead to the idealist fallacy. The idealist fallacy claims that since there must be thinkers in order for there to be thoughts about the world, every thought about the world entails the existence of a thinker. Sellars makes the following observation: We must here, as elsewhere, draw a distinction between what we are committed to concerning the world by virtue of the fact that we have reason to make a certain assertion, and the force, in a narrower sense, of the assertion itself. Idealism is notorious for the fallacy of concluding that because there must be minds in the world in order for us to have reason to make statements about the world, therefore there is no sense to the idea of a world which does not include minds; the idea, that is, that things might have been such that there were no minds [. . .] [But] just as it throws light on the status of mind in the universe to point out that it makes sense to speak of a universe which contains no minds; so it throws light on the concept of a law of nature to point out that it makes sense to speak of a universe in which there are uniformities which, although physically contingent, are without exception. (Sellars 1958, 301) Sellars here establishes an analogy between the mind’s relation to nature and the connection between physical law and necessity. That the concept of intelligibility is indissociable from the concept of mind does not entail

Correlation, Speculation, and the Modal Kant-Sellars Thesis 79 that a universe without minds is unintelligible. Similarly, that the concept of physical necessity (i.e., law) implies exceptionlessness does not entail that contingent (non law-like) exceptionlessness is unintelligible. ‘X is intelligible’ entails ‘X is thinkable,’ but it does not entail ‘Someone thinks X is intelligible.’ Claims about intelligibility do not entail claims about thinkings of intelligibility, even if it is always possible to think something that is intelligible. ‘X is a law of nature’ entails ‘X is exceptionlessness,’ but it does not entail ‘X is necessary.’ Assertions of uniformity do not entail assertions of necessity, even if they can in some instances be ratified as necessary. What Sellars provides here is another way out of the correlationist dilemma. Recall that, against metaphysical realism, strong correlationism must relativize every dogmatic absolute by insisting on the circle of correlation: every absolute is posited. But against speculative idealism, it must also de-absolutize the correlation by insisting on its facticity: the correlation could have been otherwise or could not have been. It is as a way out of this deadlock that Meillassoux introduces his speculative solution: the facticity of the correlation must be absolutized in order to de-absolutize the correlation; only this can prevent the correlation from becoming in and for itself. Sellars’ distinction between semantic and pragmatic inference provides a leverage point allowing us to prise apart correlating and correlated, thinking and thought. Claims about contingency and necessity at the level of thought-content must be separated from claims about contingency and necessity at the level of thinking-act. Correlationism conflates epistemic necessity at the level of thought-content with transcendental necessity as a relation between thinking-act and thought-content. It confuses the necessary link between intelligible-content and thinkable-content for a necessary link between thinkable-content and thinking-act. But the dependence of thought-content on thinking-acts is not a necessary connection. As Hegel saw, it is an indivision that is also a division, not a one-to-one correlation. Act and content are distinct facets of a single indivisible reality, which Hegel aptly described as the coincidence of scission and fusion. In one sense, the distinction between semantic and pragmatic inference is internal to the correlation, since what pragmatic inference does—the force of the assertion understood as what is implied by the doing that accompanies saying—is stateable as the content of another assertion. It is a representable happening in the world. In this sense it is the implicit underside of the explicitly asserted content. But this doing also has an underside and this is the non-intentional act effectuated by saying as what exceeds reinscription at the level of semantic content. Only part of what one is doing by saying is recapturable in another saying. The surplus or remainder is inferring as an act (rather than an intentional action), which is the non-semantic obverse of pragmatically inferable propositional content. The distinction between pragmatic and semantic inference cannot be aligned with the distinction between representing and represented; but the doing effectuated by pragmatic inference is rooted in an act that does not unfold within the space of reasons. Inferring occurs through a representing that is not part of the game of giving

80 Ray Brassier and asking for reasons. Thus Sellars’ claim that inferring is an act8 puts its finger on the point of transcendental exteriority internal to the correlation between mind and world: it is the difference between representing and represented, which he inscribes within nature, transcendentally understood as the domain of being-in-act.9 The epistemic primacy of the correlation between mind and world is maintained at the level of thinkable contents, but it is now grounded in the formal distinction between representing act and represented content, such that what is truly represented at the level of semantic content is the phenomenal aspect of representings-in-themselves. Because it does not transcend its instantiation in the discursive practices of language-using animals, epistemic correlation is contingent (language-using animals need not have evolved); but it is not necessarily contingent. The speculative redoubling is unwarranted. The contingency of the correlation is de-absolutized, without re-absolutizing correlation as a necessary being. Of course, Meillassoux could object that all we have done here is re-assert the Hegelian fork of the correlationist dilemma: what we just described as the ‘scission-in-unity’ of act and content is simply Hegel’s absolutized correlation in disguise. The act-content split which Sellars construes as the formal distinction between representing and represented implies the interdependence of inside and outside as aspects of a noumenal identity, and this, so the objection goes, amounts to a grounding relation. Interdependence implies necessity. Thus, what we tried to present as a Sellarsian alternative to speculative materialism—the fifth possibility, so to speak, after weak correlationism, speculative idealism, strong correlationism, and speculative materialism—would be nothing but speculative idealism in disguise. Such a rejoinder is envisageable, but it is vulnerable on two fronts. First, it continues to treat necessity and contingency as material mode meta-categories. To do so is to reify their expressive function and treat it as though it were a property of things—up to and including abstract entities like the correlation itself—rather than a way of saying what we are doing when we talk about things. The reification of modality perpetuates the metaphysical framework that speculative materialism claims to supersede. Second, the objection disregards how, according to the Sellarsian alternative sketched above, correlation is conditioned by a transcendental exteriority, but one that is immanent, not transcendent: what ensures the correlation of thought and thing at the level of representable content is a de-correlated real of nature acting in-itself, comprising both representings and non-representings whose actuality is independent of those aspects of the world that are currently represented as actual. This is Sellars’ transcendental naturalism. 4. Seizing the Means of Semantic Production All saying is a kind of doing, but not all doing is a kind of saying. Sellars’ distinction between conceptual and non-conceptual representings (which he calls “sense-impressions”) is accompanied by a commitment to the existence of

Correlation, Speculation, and the Modal Kant-Sellars Thesis 81 representings-in-themselves whose proper conceptual characterization cannot be stipulated a priori.10 Our everyday descriptions of things can be (partly) explained by postulating non-conceptual complexes of representations whose properties differ in kind from those of the things conceptually represented by them.11 Thus, for Sellars, the relation between representing act and represented content is not that between implicit and explicit: non-conceptual representings are doings that are not merely implicit in conceptual representings. Their properties are the postulated counterparts of the properties of represented objects, which serve as their models, but counterparts whose properties are modifications of their models: thus, for example, the sense-impression of a red rectangle can be described as a ‘red rectangular impression’ even though it is understood that strictly speaking the sensation itself is neither red nor rectangular. ‘Redness’ and ‘rectangularity’ are the models for the properties of the representing (the sensing) through which they are represented. What the postulated properties are cannot simply be deduced; it must be discovered by empirical science. Thus identifying the counterpart properties of representings is not just a matter of explicitation but of combining a priori postulation and empirical investigation. It falls to empirical science to tell us both what is being done by representings and how they do it. Sellars’ claim that “The descriptive and explanatory resources of language advance hand in hand” can be brought to bear on this account of the collaboration between postulation and investigation. It becomes possible to learn just how conceptual necessity—the practical ratification of patterns of inference—is conditioned by non-conceptual regularity; in other words, how patterns of activity both within and without the linguistic domain determine patterns of linguistic inference. As we learn how patterns of activity regulate patterns of inference, and which patterns in nature are responsible for which patterns in thinking, we acquire the resources to describe and explain the patterning of patterns. Describing and explaining the interaction between patterns of representing and patterns of inference allows us to understand the conditions under which linguistic change occurs and to begin to exercise control over it. The practical ratification of patterns of inference becomes subject to theoretical ratification. Changes in the meanings of terms become amenable to rational decision. Linguistic practice achieves semantic self-consciousness, allowing us to seize the means of semantic production. This is the ultimate import of the following lengthy quotation from Sellars: [W]hile one does not inductively establish that A P[hysically]-entails B by armchair reflection on the antecedent ‘meanings’ of ‘A’ and ‘B’, to establish by induction that A P[hysically]-entails B is to enrich (and, perhaps, otherwise modify) the use of these terms in such wise that to ‘understand’ what one now ‘means’ by ‘A’ and ‘B’ is to know that A P[hysically]-entails B. If to establish by induction that A causes B were to establish that (in all probability) (x) Ax  Bx, perhaps as a member of a set of induc-

82 Ray Brassier tive conclusions, there would be little reason to say that to establish by induction that A causes B is to decide on empirical grounds to give a new use to ‘A’ and ‘B.’ If, however, it is, as I am arguing, a matter of deciding to adopt a new principle of inference, then there is every reason to say that to establish by induction that A causes B is to modify the use of ‘A’ and ‘B’ and, indeed, to modify it in such a way that these terms can properly be said to have acquired a new ‘meaning’. Here two warnings are in order: First, the new ‘meanings’ do not involve a change in explicit definition. B has not, in this sense become ‘part of the meaning of ‘A’’. Yet the new role played by ‘B’ and ‘A’ does warrant the statement that the ‘meaning’ of ‘A’ involves the ‘meaning’ of ‘B’; for they are now ‘internally related’ in a way in which they were not before. Second, the relation between the new and the old ‘meanings’ of ‘A’ and ‘B’ is a logical rather than a purely historical one; as long, that is, as the espousal of this new inference ticket retains its character as a scientific decision [. . .] The motto of the age of science might well be: Natural philosophers have hitherto sought to understand ‘meanings’; the task is to change them. (Sellars 1958, 287–288, my emphasis) Semantic self-consciousness is a collective historical achievement. We discover which applications of empirical concepts are correct through the same process in which we discover which inferences connecting those concepts are correct. The local circuit linking perception to inference, inference to action, and action to perception at the ontogenetic level is enveloped within a global feedback loop making up our species’s cognitive ‘world story’ at the phylogenetic level. For Sellars as for Hegel, that story involves the development of resources for reasoning about reason, such that linguistic change becomes governed by reason, and no longer merely compelled by causes: Once the development of human language left the stage when linguistic changes had causes, but not reasons, and man acquired the ability to reason about his reasons, then, and this is a logical point about having the ability to reason about reasons, his language came to permit the formulation of certain propositions which, incapable of proof or disproof by empirical methods, draw, in the heart of language militant, a picture of language triumphant. (Sellars 1958, 307) Reasoning about reason is the unified process through which we correct prescriptive inferences and descriptive concepts; it proceeds on the basis of what Wittgenstein called ‘hinge propositions’: propositions that are not susceptible of proof or disproof because they provide the precondition for proof or disproof relative to a given subject matter.12 Such propositions express the regulative ideals that guide our inferential practices. But what picture of

Correlation, Speculation, and the Modal Kant-Sellars Thesis 83 ‘language triumphant’ do they draw? It is difficult to say. What can be said is that the ideals at issue are not supernatural phantasms but injunctions to act, such that it is constitutive of the ideal that it be actualized in the real. The necessity evoked here is that of a practical need, or what could be called a rational compulsion. Thinking is compelled towards what ought to be thought through the intermediary of what ought to be done. Notes 1. See Meillassoux 2008. 2. See Brassier 2007, 63–83. 3. “It could be that, like all radical novelty, the advent of life (the appearance of a hidden anatomical organization or cognitive activity) is accompanied by the simultaneous advent of material configurations that rupture with the physical laws in the midst of which they emerge. Indeed, nothing forbids us from thinking that the advent of the qualitative universe of vital contents should be one and the same as the advent of the material underpinning by which these contents are inscribed in the material Universe that precedes them. In that case, the appearance of the material organization of life would have no reason to obey the frequential constants of matter. The configurations of life would break the laws of chance, because they would not at all be the possible cases of matter, but rather the correlate within matter of the appearance ex nihilo of vital contents” (Meillassoux 2011, 183–184). 4. See Meillassoux 2016. 5. See Brandom 2015. 6. See Brandom 2015, 140–141. 7. See Brandom 2015, 49–55. 8. Sellars uses the word ‘doing’ rather than ‘act.’ He writes: “ ‘True’ is a sign that something is to be done for inferring is a doing” (Sellars 1991, 206). Sellars’ claim here can be interpreted as simply pointing out the connection between saying and doing, of which Brandom provides a systematic account. On this account, what is to be done by saying that something is ‘true’ is endorsing an assertion and taking responsibility for justifying it. These are moves in the space of reasons. What I am trying to suggest is that while Brandom’s expressivist account of the saying/doing doublet is very illuminating, it tends to disregard Sellars’ own naturalistic emphasis on the substructure of doing, i.e., the non-propositional substructure of representings-in-themselves. 9. This is a point I cannot elaborate on here, but I believe it is closely connected to Sellars’ otherwise perplexing commitment to a metaphysics of what he calls ‘pure processes.’ See Sellars 1981). 10. See Sellars 1968, 1–30; see also Brassier 2016). 11. For a fuller elaboration of this point, see Brassier 2016. 12. “That is to say, the questions that we raise and our doubts depend on the fact that some propositions are exempt from doubt, are as it were like hinges on which those turn” (Wittgenstein 1969, 44e). References Brandom, Robert, 2015. From Empiricism to Expressivism: Brandom Reads Sellars. Cambridge, MA, Harvard University Press. Brassier, Ray, 2007. Nihil Unbound: Enlightenment and Extinction. Basingstoke, Palgrave Macmillan.

84 Ray Brassier Brassier, Ray, 2016. “Transcendental Logic and True Representings,” Glass Bead Journal, Site 0: Castalia, The Game of Ends and Means. Retrieved on www.glassbead.org/article/transcendental-logic-and-true-representings/?lang=enview Meillassoux, Quentin, 2008. After Finitude: An Essay on the Necessity of Contingency, translated by Ray Brassier. London and New York, Continuum. Meillassoux, Quentin, 2011. “Excerpts from L’inexistence divine.” In Graham Harman ed., Quentin Meillassoux: Philosophy in the Making (Edinburgh, Edinburgh University Press), 175–238. Meillassoux, Quentin, 2014. Time without Becoming, edited by Anna Longo. Mimesis International. Meillassoux, Quentin, 2016. “Iteration, Reiteration, Repetition: A Speculative Analysis of the Sign Devoid of Meaning.” In A. Avanessian and S. Malik, eds., Genealogies of Speculation: Materialism and Subjectivity Since Structuralism (London and New York, Bloomsbury Academic), 117–197. Sellars, Wilfrid, 1958. “Counterfactuals, Dispositions, and the Causal Modalities.” In H. Feigl, M. Scriven, and M. Grover, eds., Minnesota Studies in the Philosophy of Science, vol. II: Concepts, Theories, and the Mind-Body Problem (Minneapolis, University of Minnesota Press), 225–308. Sellars, Wilfrid, 1968. Science and Metaphysics: Variations on Kantian Themes. London, Routledge and Kegan Paul. Sellars, Wilfrid, 1981. “Foundations for a Metaphysics of Pure Process.” The Monist 64: 3–90. Sellars, Wilfrid, 1991. “Truth and ‘Correspondence’.” In Science, Perception and Reality (Atascadero, Ridgeview Publishing Company), 197–224. Wittgenstein, Ludwig, 1969. On Certainty, edited by G.E.M. Anscombe and G.H. von Wright, and translated by D. Paul and G.E.M. Anscombe. Basil Blackwell, Oxford. Wolfendale, Pete, 2014. “The Necessity of Contingency.” In P. Gratton and P.J. Ennis, eds., The Meillassoux Dictionary (Edinburgh, Edinburgh University Press), 121–124.