1.1 Introduction

Wilfrid Sellars saw his account of categories and abstract entities as developing a long line of anti-platonist thinking about fundamental ontological categories such as substance, property, fact, universals, and particulars. Sellars traces the lineage of this outlook from Ockham and the medieval terminist logicians, through Hume’s nominalism, to Kant’s conception of the categories as “pure concepts” of the understanding; and then subsequently as transformed by the linguistic turn in the first half of the twentieth century. Crucial to the latter, for Sellars, were key “meta-conceptual” analyses in Wittgenstein’s Tractatus of “truths about our thoughts about objects in the world” (TTC VI §33),Footnote ¹ and Carnap’s analytic reconstruction of our references to abstract entities in terms of the metalinguistic resources of logico-linguistic frameworks. Sellars’s distinctive contribution to this longstanding (if currently on the defensive) metaconceptual approach to the nature of ontological categories was to interpret and reconstruct it in terms of his own “meaning as use” or norm-governed inferential role semantics. With these resources Sellars sought to preserve the genuine insights in the “realist” or broadly platonic traditions while simultaneously defending the idea that in the end, as he puts it, “a naturalistic ontology must be a nominalistic ontology” (NAO IV §129). His theory of the categories, as we shall see, turns out to be essential to some of Sellars’s most distinctive metaphysical and epistemological positions, including most notably his famous views on the myth of the given and on the ostensible “clash” between the manifest and scientific images of the world.

In what follows I begin by laying out some of the basic features of Sellars’s conceptual role account of abstract entities by focusing specifically on his account of the categories, beginning with his remarks on Ockham, Wittgenstein, and Kant. Assuming here the general viability of his general “functional classification” account of meaning and abstract entities in response to the usual realist objections,Footnote ² I proceed to focus on the complex role that ontological categories actually play in certain key aspects of Sellars’s metaphysical and epistemological views. Displaying the categories at work in this way raises fundamental questions both about their nature and about the role of categorial ontology in general, as well as casting a different light on the myth of the given and the relations between the manifest and scientific images of the world.

1.2 “Toward a Theory of the Categories”: Ockham, Kant, and Wittgenstein’s Tractatus

In his 1970 article, “Toward a Theory of the Categories” (TTC), Sellars presents a brief and highly reconstructive account according to which Ockham, Wittgenstein’s Tractatus, and Kant offer a series of insights for a possible alternative to the dominant varieties of realism about universals (e.g., qualities and relations) and the question of our “ontological commitment” to abstract entities, which Sellars traces back to Plato’s Sophist and Aristotle’s conception of the categories as “highest kinds” (summa genera) of entity. Plunging into the middle of things, Sellars illustrates the category of substance as an Aristotelian summum genus by means of the generalizing sequence, ‘Fido is a dog … Fido is an animal … Fido is a substance.’ The initially awkward parallel in the category of quality, he indicates, would suggest ‘x is a red … x is a color … x is a quality’, but characteristically this becomes the sequence, ‘Red(ness) is a color … Red(ness) is a perceptible quality … Red(ness) is a quality’, with the abstract singular term redness highlighting the ostensible ontological commitment to qualities as universals.

Sellars suggests that subsequently the “first major breakthrough in the theory of categories came, as one might expect, in the late Middle Ages, when logic, like knighthood, was in flower” (TTC V §17). In TTC Sellars presents the various views he discusses in terms of his own inferential role account of linguistic meaning and his analogous account of “mentalese” (i.e., of “inner” conceptual content). Roughly speaking, dot-quoted items (e.g., a •dog•) are contrived common nouns that serve to classify functionally any item in any linguistic or mental-representational system that has the same or a relevantly similar norm-governed functional role or “use” as that of the term illustrated between the dot-quotes, with which the language user is already familiar.Footnote ³ Sellars then illustrates Ockham’s strategy as one of explicating, for example, “Man is a species” as having the sense of “•Man• is a sortal mental term” (cf. TTC V §17), and similarly analyzing “Socrates is a substance (a primary individual)” as “•Socrates• is a basic mental singular term” (TTC V §21). Along the same lines, he indicates that “Yellow (yellowness) is a quality” would have the sense of “The •yellow• is a (one-place) predicate (in mentalese)” and the latter “would ‘reduce’ to” the statement, “•Yellows• are predicates.” On this view “to be a •yellow• is to be an item having a certain conceptual job, which would ultimately be explained in terms of the ‘word-word’ inferential and ‘word-world’ perceptual and agentive uniformities by virtue of which ‘yellow’s in one language and ‘gelb’s in German function as they do in basic matter-of-factual statements” (TTC V §29). Sellars sums up the general line of thought as follows:

On this general basis Sellars takes the nominalist to be in a position to grant, in one philosophical tone of voice, that of course there are such qualities as yellowness or triangularity, while nonetheless “denying that there really are such qualities” from a strictly ontological point of view. The latter is in light of the available paraphrase of such abstract singular terms (e.g., triangularity) into the normative-functional classification of the relevant rule-governed conceptual role players (the •triangular•). Thus “although there really are particular conceptual episodes of thinking that something is triangular, there really is no such entity as the quality of being triangular” (NAO IV §31) – though there do, of course, really exist the various individual triangular objects to which we regularly respond with that conceptual classification. Sellars considers the objection that this nominalist analysis merely saddles the theorist with the task of then explicating what the relevant conceptual roles are without thereby illicitly reintroducing “such exotic abstract entities as functions, roles, [or] rules” (NAO IV §137). But he responds by suggesting that the strategy simply applies harmlessly to itself as an instance, that is, that “abstract singular terms for these entities [i.e., functions, roles] are to be handled by the same strategy as is used to handle ‘triangularity’” (NAO Appendix, Correspondence with Loux, §23).Footnote ⁴

The overall upshot of Sellars’s account, as I understand his view, is that discourse about abstract entities and involving ontological categories is in reality a culturally evolved, metalinguistic, cross-language classificatory enterprise, the primary function of which in relation to our ground-level empirical thought and discourse is to track conceptually (at second-order) our ongoing attempts to conceptually represent (at first-order) the nature of empirical reality.Footnote ⁵ Sellars’s attitude toward the platonic and related realist traditions, so interpreted, is far from dismissive,Footnote ⁶ and we shall see that categorial ontology so conceived in fact plays a vital role within Sellars’s own philosophy.Footnote ⁷

In relation to Wittgenstein’s views in the Tractatus in this connection, Sellars in “Toward a Theory of the Categories” is primarily concerned, as elsewhere, to execute a complex two-pronged strategy, partly in criticism and partly in defense of his approach. On the one hand, Sellars employs his metalinguistic theory of abstract entities as sketched above to offer a deflationary, conceptual role-classificatory accounts of reference (e.g., the denotation of names) and exemplification (e.g., terms “standing for” properties), as well as of the idea of truth as a propositional “correspondence to facts.” All of these ostensibly world-relational semantic notions turn out, again, to be metaconceptual functional classifications of linguistic or conceptual roles, rather than being either basic empirical relations or primitive metaphysical relations between language or thought and reality. For example, in relation to the mathematical statement, “‘II + II = IV’ (in L) corresponds to the fact that 2 + 2 = 4,” Sellars holds that “it is implausible in the extreme to suppose that in this context ‘corresponds’ stands for a matter-of-factual relation between ‘II + II = IV’ (in L) and an extraconceptual entity” (TTC VII §48). (Of course, mathematical platonists will have various responses to make to this claim.) On Sellars’s analysis, “correspondence to facts” then “dissolves into” the functional role classification: “‘II + II = IV’s (in L) are •2 + 2 = 4•s,” and the factual character reflects “the ascription of [truth as] correct assertibility to •2 + 2 = 4•s,” with the criterion for mathematical truth for Sellars (with appropriate caveats) being a matter of provability within a formal system rather than of “word-thing connections” (TTC VII §48; cf. SM IV §62, V §55). Similarly at the subsentential level, denotative reference and property exemplification are analyzed in terms of the relevant name-role and predicate-role normative-functional classifications, rather than in terms of primitive semantic relations to things and properties.

On the other hand, Sellars argues that in the case of empirical or “matter-of-factual” discourse (as opposed to the cases of mathematics or ethics, for instance), the above sorts of intra-linguistic and interlinguistic conceptual role classifications reflect our generally systematic, socially norm-governed patterns of language use and of tokenings of thoughts in response to objects and in learned patterns of inference. The result is that in such empirical domains, specific sorts of spatiotemporal-causal relations and uniformities come to obtain (a) between language and the world (in norm-governed “language entry” perceptions and “language exit” volitions), and (b) in specific inferential patterns of linguistic tokenings, and these norm-produced and norm-governed natural uniformities are both entailed by and presupposed by our social-linguistic practices. It has not always been well understood in the literature that Sellars thereby defended a domain-specific, naturalistic yet norm-governed causal theory of empirical reference and of representational correspondence to the world (or “picturing”) in the case of basic matter-of-factual empirical domains, although of course his primary notions of “reference” and “correspondence” are the deflated conceptions sketched above. For the semantic notions themselves (of reference, exemplification, correspondence to facts, and truth) are understood throughout as various species of metalinguistic functional role classification that reflect those ground-level practices and resulting norm-governed uniformities (cf. O’Shea Reference O’Shea2007, 147–58).

Sellars concludes the above two-pronged analysis with the summary remark that Wittgenstein in the Tractatus – at least as Sellars wants to read him – was therefore “right in claiming that reference is not a matter-of-factual relation, although the fact that a term refers entails that it stands in certain matter-of-factual relations” (TTC VI §41). Our conceptual categories in general, then, have a similarly Janus-faced character on Sellars’s account: on the one hand metalinguistic and deflationary as far as “abstract entities” are concerned, but combined with a naturalistic ontological realism about empirical discourse and scientific theories in general. The categories themselves are metaconceptual, second-order functional classifications of the most fundamental types of first-order conceptual roles within a given norm-governed linguistic practice or conceptual framework. This is the anti-platonist, “psychological nominalist” aspect of Sellars’s account of abstract entities, in the spirit of Ockham and Kant. At the same time, as far as “matter-of-factual” discourse and empirical domains of inquiry are concerned, such first-order linguistic practices and resulting second-order conceptual categorizations are part of a naturalistic and realist account of how such practices enable us successfully to represent and track, or naturalistically “picture,” in Sellars’s sense, empirically mind-independent objects and events in the world.

The above sketch of Sellars’s theory of the categories in terms of his metalinguistic and nominalist treatment of abstract entities has been at a high level of generality. But what role do categories actually play within Sellars’s epistemological views and his metaphysics of nature? And what further light does that shed on his conception of the nature of the categories in general? The place to begin addressing these questions is with Sellars’s use of Kant in relation to what Sellars called the “manifest image” of the human being in the world.

1.3 Kantian Categories in the Manifest Image – And in the Scientific Image?

In my view Sellars was correct to interpret Kant’s categories, or “pure concepts” as metaconceptual or second-order rather than first-order concepts, in that they characterize the rule-governed functions that a given first-order concept must realize in our judgments and principles in order to be a concept of the given categorial kind under consideration (e.g., a “substance” concept, a “causal” concept, etc.). We need not endorse the specifics of Kant’s “metaphysical deduction” of twelve a priori categories by appeal to twelve corresponding a priori logical forms of judgment in order to take on board the general strategy of using logical form as the “clue” to the discovery of fundamental types of conceptual role in judgment. Kant’s primary concern is the role of the categories as classifying indispensable kinds of concepts as they function specifically in judgments and principles that, he argues, are necessary for the possibility of experience, that is, for the objective cognition of empirical objects in space and time. For example, the schematized categorial principle of substance as, roughly, a subject that persists permanently through all change in its “determinations” or accidents, classifies a functional role that will be realized by whatever empirical concept or concepts successfully play that necessary role in making our experience (in this case, of any duration over time) possible. Similarly, the second-order (“transcendental”) category and corresponding principle of causality are realized by whatever first-order empirical-causal concepts prescribe such a necessary connection between particular empirical kinds of alteration or event.

Sellars argued that Kant’s insights could be successfully reformulated and defended in ways appropriate to the subsequent revolutionary developments in logic, science, and the “linguistic turn” in twentieth-century philosophy:

In various places Sellars argued that any language or conceptual framework that makes possible empirical cognition must exhibit certain features that, if made explicit, take the form of various perceptual (or “language-entry”) reliability principles, material-inferential principles with counterfactual force, a skeletal spatiotemporal-indexical framework, and so on, where any such conceptual framework will necessarily have the Janus-faced normative and corresponding naturalistic dimensions discussed in Section 1.2 (e.g., SM IV §61, V §30; KTE §§36–45, 448–53; TTC §§50–6).

I have discussed this modified Kantian aspect of Sellars’s philosophy in detail elsewhere (O’Shea Reference O’Shea2007, Ch 5; Reference O’Shea, Stern and Gava2016b) and so I will not focus on it here. However, there is one difficult interpretive question that this account generates that is important for assessing Sellars’s own theory of the categories: Does Sellars take his “transcendental linguistics” to deliver certain second-order categories and categorial principles that govern all possible humanly knowable empirical conceptual frameworks, even across the radically alternative fundamental categorial ontologies that Sellars argues characterize the progress of scientific theorizing? This is not a simple question to answer.

The category of causality provides an important case in point. Sellars clearly holds that Kant’s account of the necessary applicability of the principle of causality in experience can be given a linguistic turn such that it becomes the claim that, for any empirically significant language able to serve as a “cognitive instrument” and “generate knowledge of a world to which it belongs” (KTE §40, quoted earlier), “it is only because the expressions in terms of which we describe objects … locate these objects in a space of implications, that they describe at all, rather than merely label” (CDCM §108, 306–7; cf. Brandom Reference Brandom2015, Chs 1 and 3, and O’Shea Reference O’Shea, Stern and Gava2016b). Brandom thus explicates and defends what he calls the modal “Kant-Sellars thesis,” according to which the practice of “deploying any ordinary empirical vocabulary,” however simple (e.g., “this is red”), already presupposes “counterfactually robust inferential practices-or-abilities – more specifically, the practical capacity to associate with materially good inferences ranges of counterfactual robustness” (Brandom Reference Brandom2015, 160, italics in original). There is no doubt that Sellars held that such nonformal “inference tickets,” as Ryle called them, are essential to any conceptually and empirically contentful language about the world.

However, the category of causality, on Sellars’s view, has undergone radical changes throughout the history of human thought, and in particular in the modern period as a result of scientific developments. Categorial ontology, for Sellars, is clearly a conceptual framework-relative notion, which complicates the question as to whether, on Sellars’s view, there are any categories or categorial principles that hold true across all such changes – never mind across all possible such changes, as envisioned earlier in “transcendental linguistics.” Sellars discusses the development of the category of causality during the course of his idealized account of humanity’s gradual development from (i) an “original image” or quasi-animistic conception of the world, in which nature’s own responsibilities for the rain, for death, and so on, are as it were modelled on the category of persons; to (ii) the more categorially “pruned” and empirically sophisticated “manifest image” of the perceptible world; and finally to (iii) the modern “scientific image” of the Kantian-phenomenal world’s ultimate basis in the postulated imperceptible particles, fields, and forces that are continually refined and often radically reconceived as a result of the ongoing explanatory successes of the natural sciences (PSIM II). In telling this story, Sellars offers some further clarification as to what he means by “categorial” analysis in general, as follows.

Categories for Sellars, as we saw earlier, are second-order role-classifications of the most basic types of first-order concepts as they function in implicitly norm-governed ways within a given linguistic community or a given scientific theory. Accordingly, these most fundamental categories – including those that in “matter-of-factual” empirical and scientific discourse seek to carve reality at its joints – will undergo change whenever our first-order conceptual practices and inquiries undergo significant conceptual change. Sellars in various places attempted to address the controversial question that then arises as to how or whether it is possible to distinguish in any principled way such fundamental changes in concept or meaning on one hand, as opposed merely to particular changes in belief or particular empirical refinements involving the use of the given term under consideration (see, e.g., SM V and NAO IV).

The most basic tool in Sellars’s metalinguistic account of the nature of meaning, abstract entities, and categories is a context-sensitive conception of judgments of functional role similarity and difference, relative to given background information and whatever the particular classificatory purposes at hand may be. (Compare functionally comparing the role of the bowler in cricket to that of the pitcher in baseball, for example.)

Conceptual roles in linguistic practices and in scientific theories are thus assessed in terms of what are judged to be the relevant first-order norm-governed inferential patterns and language “entry” and “exit” transitions involving the given term, patterns that change gradually in our social practices and in science often dramatically over time. With respect to the latter cases, Sellars compares, for instance, the revolutionary changes in meaning involved in the move from Newtonian •mass• to Einsteinian •mass•, though Sellars argued that assessments of functional role similarity enable rational comparisons of conceptual development and change over time. On Sellars’s view, then, “abstract entities, pace Plato, change” (SM V §42), and hence so do the fundamental categories in terms of which humanity has attempted to limn the structure of reality. The result on this view is that a “fundamental question with respect to any conceptual framework is ‘of what sort are the basic objects of the framework?’” (italics added), and what are their properties, relations, activities, and the groups to which they belong, where the “philosopher is interested in a classification which is abstract enough to provide a synoptic view of the contents of the framework but which falls short of simply referring to them as objects or entities” (PSIM 9; II §§18–19).

In relation to the example of the category of causality in particular, Sellars argues that “it is important not to confuse between an action’s being predictable and its being caused … In the ‘original’ image, one person causes another person to do something he otherwise would not have done,” so that a predictable or habitual action, for instance, would not normally be regarded as caused in this original sense (PSIM 13; II §37). While subsequent developments in the more sophisticated manifest image gradually led to a narrowing of the broader animistic category of “persons” in nature to apply primarily only to human beings, the common sense conception of causality largely retained its broader application, according to which something is caused to happen only when some relevant intervention brings about the resulting change. Sellars’s example is that a billiard ball’s continuing to roll in a straight line on a smooth table is predictable, but would not ordinarily or in the manifest image be said to be “caused.” Sellars is among those thinkers who hold that a “distinctive trait of the scientific revolution,” in contrast to the above “manifest” conception of causality, is “the conviction that all events are predictable from relevant information about the context in which they occur, not that they are all, in any ordinary sense, caused” (ibid.).

In fact, Sellars contends that the tendencies of modern theoretical science, combined with what he argues are the categorial changes that are and will be required to provide philosophical solutions to the deepest problems of mind and metaphysics, together suggest that an all-comprehensive alternative categorial framework to that of the manifest image is in fact what will be required in order to carve reality at its joints. Sellars’s envisioned alternative framework, roughly speaking, is a neo-Humean physicalist ontology consisting of patterns and uniformities of nondurational “pure processes” required by physics as well as, Sellars argues, by an adequate ontology of qualitative sensory consciousness. The result, as he puts it, would be “a truly heracleitean ontology” according to which there “are no objects” as the ontological category of object has usually been understood (including abstract objects), but rather all such categorial frameworks would be “replaceable by” or “eliminable in favor of” a nominalistic pure process ontology according to which the “world is an ongoing tissue of goings on.” As far as the category of causality in particular is concerned, in this replacement ontology there would be no causal powers or “potentialities in the basic objects” or pure processes that serve as the ultimately real posits or “logical atoms” of the framework (FMPP II §§95, 103–4).

Returning, finally, to what we saw is Sellars’s modified Kantian idea of a “transcendental linguistics” that “attempts to delineate the general features that would be common to the epistemic functioning of any language in any possible world” (KTE IX §41), what do Sellars’s views above concerning causality suggest about the question I raised earlier: namely, whether all such empirical-linguistic frameworks must include conceptual resources that, when made explicit, would include the category of causality? Kant’s own empirically realist conception of objective causal necessitation within the realm of (phenomenal) nature was evidently a modal realist one: the altering states of his “basic objects” as phenomenal substance within the spatiotemporal world – roughly, permanent matter constituted by universal forces of attraction and repulsion – are necessarily categorially conceived as subject to first-order empirical laws of objective physical necessitation. Is there a tension between the fact that in his writings on Kant, Sellars defends Kant’s arguments for this conclusion (i.e., an empirical realism about the causal modalities), yet the categorial framework of his own alternative heracleitean or pure process ontology purges objective empirical reality of the causal modalities, at least as realistically construed in terms of objective necessitation in nature?

In assessing this question we should note that Sellars’s various reconstructions of Kant’s arguments are framed in terms of his own meta-linguistic, material-inference license conception of causal inference, an account which Sellars always viewed as consistent, at the end of the day, with an underlying Humean or scientific ontology of the objective uniformities and patterns that are thereby – that is, as a result of these norm-governed material-inferential practices – naturalistically represented or tracked (i.e., “pictured”). For example, in his most substantial treatment of the causal modalities, Sellars argued “that the picture of the world in terms of molar things and their causal properties points beyond itself to a picture of the world as pure episode” (CDCM §51). Such tracking of uniformities and patterns of “pure episodes” or “pure processes” (CDCM §§51–2) within any humanly possible (but also in this case, scientifically ideal) representational system will, as I understand Sellars’s view, still require diachronic material inference tickets, as discussed earlier. In the case of this alternative pure process ontology, however, this would involve the explicit recognition that objectively speaking there are no necessitations “in the world” over and above the episodic uniformities that our scientifically refined inference tickets have enabled us to represent and track.

From the perspective of the manifest image ontology of the world, as well as from the perspective of many philosophers such as Brandom who are otherwise inspired by his views, Sellars’s ideal of a heracleitean and nominalist categorial ontology of pure processes in particular has appeared to be strikingly revisionary and implausibly reductive in various ways. This is currently a matter of lively debate, however, as the various contributions to the recent volume on Sellars and His Legacy amply testify (O’Shea, Reference O’Shea, Stern and Gava2016b). It is clear that the heart of Sellars’s philosophy – his naturalism with a normative turn, as I have characterized it (2007) – was devoted to attempting to show that there would be nothing even in such an all-comprehensive physicalist and nominalist ontology that would threaten our status as persons, and in particular as conceptual thinkers, experiencers, and rational agents who both institute and are governed by the intersubjective functional norms that make such personhood possible in the first place. Broadly speaking, this is because all of the conceptual capacities relevant to our personhood are normative-functional realities that Sellars attempted to conceive consistently with their full physical realization, however scientifically revisionary and metaphysically comprehensive the latter nominalist ontology might become.

What is particularly interesting in relation to the role of categories and categorial metaphysics in this overall story are the ways in which they figure in Sellars’s working out of the details as to how the idealized manifest and scientific images are thus to be stereoscopically fused with one another within one “synoptic vision” and one categorial ontology of the human being in the world. In Section 1.4, I will close by briefly examining one particular important issue in this regard, namely, the relatively neglected but essential role of the categories in Sellars’s famous rejection of the “myth of the given,” which was itself more central than is often realized to his own attempted reconciliation or synoptic fusion of the manifest and scientific images.

Before leaving the present topic, however, I should note that Robert Brandom’s recent sophisticated “pragmatic metalinguistic” and inferentialist analyses of Sellars’s views on both the categories and the causal modalities (e.g., Reference Brandom2015, Reference Brandom and O’Shea2016), which go into insightful detail in relation to their basis in Sellars’s semantics and his theory of abstract entities, also contain a detailed critique of Sellars’s attempt to draw the ontologically nominalist conclusions from those metalinguistic analyses that I have highlighted earlier. Brandom sums up his critique of this nominalistic aspect of Sellars’s views this way:

More specifically, the conclusion of Brandom’s argument is that Sellars’s metalinguistic analysis of abstract entities and categories “is at most compatible with” Sellars’ “ontological nominalism,” but provides no compelling argument for it (Reference Brandom and O’Shea2016, 43–5). Furthermore, Brandom argues that a realism of the sort that he himself defends about properties, propositions, facts, and the causal modalities is also consistent with Sellars’ metalinguistic analysis, but is philosophically preferable to ontological nominalism. Not even Sellars’s notoriously strong scientific naturalism yields an argument for the latter, according to Brandom, since “the language of science is just as much up for alternative interpretations, nominalistic and otherwise, as ordinary language” (ibid.).

I think this conclusion is not surprising. The burden of argument typically does fall on ontological nominalists to show that their more parsimonious resources are sufficient for the needs of semantics, mathematics, and science, consistently with the ostensibly realist surface grammar. The naturalistic motivations of most nominalists usually do derive from elsewhere, and in Sellars’s case he viewed his metalinguistic “psychological nominalism” about abstract entities and categories as essential to combatting “the persistent (if currently repressed) notion that relations between minds and abstract entities must be invoked by an adequate psychological theory of the ‘higher processes’” (EAE 436, II §12).Footnote ⁸ Sellars explicitly disagreed, on epistemological grounds, with Quine’s Duhemian-holistic reasons for being at naturalistic peace with such seemingly non-naturalistic (non-spatiotemporal, noncausal) positing of abstract entities realistically construed (cf. NAO I §§24–34), posits which continue to dominate current philosophy of mind, language, mathematics, and modality. However, if Sellars’s metalinguistic conception of the categories and of abstract entities really is fit for its purposes (I have noted that this is a question up for wider debate), then I would think that its consistency with ontological nominalism arguably ought to count strongly in its favor, at least with the many philosophers who, for a variety of reasons, continue to aim for some version of a non-eliminativist yet ontologically naturalistic philosophy of mind and agency.

1.4 Categorial Ontology at Work: Sellars’s Metaphysics and the Myth of the Given

Given its prominence both in Sellars’s thinking and in the literature, the problem of the so-called myth of the given and the related question of the ontology of the sensible qualities (such as color) is the most obvious case to consider in relation to exhibiting categories and categorial analysis at work in Sellars’s philosophy. In employing his theory of the categories as metaconceptual functional classifications of fundamental types of norm-governed conceptual roles, Sellars holds that with respect to any given conceptual framework (though in this passage referring specifically to the “manifest image”), “there is a correct and an incorrect way to describe this objective image which we have of the world in which we live”; and more generally, “there is truth and falsity with respect to” any philosophical attempt to analyze the categorial structure of any given framework. An important part of categorial ontology, for Sellars, is the attempt to diagnose and correct such mistaken analyses (PSIM III §§38–9). The recurrent philosophical idea of “the given,” in its various forms, “as a piece of professional – epistemological – shop talk” (EPM I §1), is a prime example of such a mistaken analysis, according to Sellars, but most importantly for present purposes, in this case his theory of the categories plays a distinctive and often neglected role in what is perhaps his most well-known philosophical contribution: his critique of the myth of the given.

It is not always realized the extent to which, for Sellars himself, in contrast to many of the more recent adaptations of his epistemological and semantic views on the myth of the given and the holistic “logical space of reasons,” the most consequential result of his critique of the myth is that, in his view, it first makes possible the only genuinely adequate, fully naturalistic synoptic fusion of the manifest and scientific images of the human being in the natural world.Footnote ⁹ Sellars diagnoses the myth of the given as coming in many different forms, including both empiricist and intellectualist or rationalist versions, and in neither case restricted to its most familiar and prevalent guise of epistemological foundationalism. Roughly, the latter is the idea that there is a stratum of noninferential knowledge or direct apprehension (or “acquaintance”), whether of particulars, properties, facts, or ways of experiencing or appearing, which is such that (a) it is not only noninferential – Sellars has no problem with noninferential cognition per se – but most importantly such items of knowledge allegedly (b) “presuppose no knowledge of other matter of fact, whether particular or general,” while nonetheless also (c) “constitut[ing] the ultimate court of appeals for all factual claims – particular and general – about the world” (EPM VIII §32). The most well-rehearsed Sellarsian arguments against any such alleged epistemic givens usually attempt to show, to put it in a nutshell, that either (i) the given does not have the right propositional or conceptualized form to provide any reason or justification for any empirical thought or claim within the “logical space of reasons” (EPM VIII §36), thus violating (c); or (ii) if the given is taken to have a sufficiently conceptualized form to be responsive to satisfy (c), it can in each case be shown not to be presuppositionless in the way intended by the philosopher appealing to the given, thus violating (b).

However, I think that what Sellars argues can be seen to underpin such foundationalist epistemologies are various forms of a more basic and typically unexamined assumption. This is the often implicit assumption that in some cases we are directly aware of some sorts of item – whatever these sorts may be, whether conscious states of ourselves, or properties (or property-instances), or external objects themselves, or directly evident propositions or intuited rational relations, for instance – and that we are aware of them as the sorts of item that they are, simply in virtue of immediately experiencing or directly apprehending them in this way (whatever that way may be), and hence independently of, and so not presupposing, any other capacities or abilities that we may possess or have acquired.

For example, Sellars argues that, despite their many other differences, “Locke, Berkeley, and Hume … all take for granted that the human mind has an innate ability to be aware of certain determinate sorts – indeed, that we are aware of them simply by virtue of having sensations and images” (EPM VI §28, italics in original), for instance, in having a sensation of red or of redness. Today this might be put by many philosophers of mind, whether they be foundationalists in epistemology or not, in terms of our being directly aware of or consciously experiencing qualia of various sorts, or perhaps being aware of the conscious character of other aspects of our own occurrent mental states, depending on the given view at hand. Other philosophers have recently defended more intellectualist or rationalist versions of the given in terms of the direct intuition of rational relations.

The arguments in this domain remain subject to lively controversy, of course, but what Sellars attempted to argue is that, at a deeper “categorial” level, as he characterizes it, even the most basic immediate experiences or direct apprehensions of various sorts – whether these are assumed to be sorts of experiencings themselves or to be sorts of experienced or apprehended objects, qualities, or relations – are themselves not as presuppositionless as they are implicitly assumed to be. This is the case whether or not the overall epistemology that scaffolds such basic awareness of sorts (or sorts of awareness) is taken to have a foundationalist structure. In his later writings, such as his 1981 Carus Lectures (FMPP; see also BLM VI), but developing essentially the same point he had made above in EPM VI in 1956, Sellars puts it this way: “perhaps the most basic form of what I have castigated as ‘The Myth of the Given’” may be stated in terms of the following “principle”: “If a person is directly aware of an item which [in point of fact, i.e. from the standpoint of the ‘best explanation’] has categorial status C, then the person is aware of it as having categorial status C” (FMPP I §44; the bracketed interpolation is from the principle as stated at BLM VI §122). I have elsewhere called this Sellars’s “myth of the categorial given” (2007 Ch 5: 115; examined in depth in O’Shea Reference O’Shea2021), arguing that this is typically the deeper basis of his familiar arguments against the myth of the epistemically foundationalist given.

Sellars’s reference to an item’s having “categorial status C,” and hence his own understanding of what is perhaps the most basic form of the myth of the given, is an application of his views on the categories as metaconceptual functional classifications. As we have seen, this brings in its wake the importance of conceptual change and hence of changes in categorial ontological frameworks in our ongoing attempts to reconceive and explain the nature of things. Categories, as second-order concepts that serve to classify functionally and normatively the most basic types of first-order concepts in terms of which we experience the world, are thus implicit in any conceptual framework put to use in describing and explaining the world, whether that framework be the “manifest image” of perceptible physical objects (colored, middle-sized dry goods) and persons as rational agents, or the ongoing scientific images (or the “regulative ideal” of “the” complete scientific image) of the human being and the world. The “categorial status” of any item is thus framework relative, and dependent upon a view of conceptual change according to which it is always an open and fallible question as to the best explanation or the ultimately correct ontological categorization of whatever it is that we conceptualize and experience within whatever conceptual framework we have inherited or are assessing, whether scientific or manifest.

The pivotal case that Sellars is discussing in the Carus Lectures when he brings up the myth of the categorial given concerns “The Place of Color in the Scheme of Things,” to borrow the title of Jay F. Rosenberg’s insightful “Roadmap to Sellars’ Carus Lectures” (2007, Ch 9). Sellars contends that our ordinary first-order awarenesses of such sensible qualities as the red or redness involved in seeing (or vividly seeming to see) a red apple embody an implicit ontological categorization that is as fallible and replaceable as any other implicit categorization of any “repeatable” item in experience. I have followed Sellars in using the ontologically neutral term items instead of, for example, qualia, states, objects, properties, qualities, contents, sensations, impressions, features, repeatables, ideas, sense-data, characters, tropes, et al. This is because what sort of items these items really are – for example, whether they are features of external physical objects, or rather states of perceivers themselves; whether they are “repeatables” or particulars (and in what sense); whether they are mental or physical, adverbial or substantive, and so on – is according to Sellars, not something that is directly “given” independently of implicit conceptual categorization (or in the case of infants and some animals, at least “proto-conceptual” categorization). How such items, whatever they may be, are (second-order) categorized will in each case be implicit in the material-inferential conceptual structure of our first-order characterizations of such items, which will itself depend upon the particular fallible conceptual frameworks in terms of which we confront reality in both ordinary life and in various stages and kinds of inquiry. In the case of our concepts of sensible qualities in particular, Sellars throughout his works provides detailed accounts of the complex ontological recategorizations, both scientific and metaphysical, that he thinks such concepts both have undergone and must further undergo. Any selective “highlight reel” sketch of his account of the sensible qualities would include the following general progression.

Sellars argues that the correct categorial analysis of our ordinary or “manifest” conception of the world – say in the case of the visible surface of a red apple I see in front of me – is that our manifest conceptions are directly realist in character, in the sense that what we take ourselves to see is a certain physical expanse of red skin as part of a persisting, physical apple. We also have the idea, both within the manifest image of common life and as refined empirically due to scientific developments (see Sellars’s “myth of Jones” account in EPM X–XVI), that in seeing the red apple we are caused to have corresponding visual sensations or “sense impressions” of a correspondingly appropriate sort, in this case, sensations of red (or more specifically, the sensation of a particular red expanse), the character of this sensing being also dependent upon the lighting, the state of one’s sense organs and brain, and so on. Sellars, throughout his works, then traces and develops a complex story about how our combined phenomenological, scientific, and metaphysical inquiries into the resulting problems raised by the nature of our sense experiences, both veridical and non-veridical experiences, have led (and continue to lead) to increasingly sophisticated accounts and disputes concerning the proper conceptualization and most adequate theoretical account of the place of color and other sensible qualities in the scheme of things.

Sellars himself ends up defending a rather radical view according to which, on the ultimately correct scientific and metaphysical categorial ontology of the world, as far as we can now coherently envision its structure according to our own best accounts, “the common-sense world of physical objects in Space and Time is unreal” (EPM IX §83). That is, the manifest categorial ontology of persisting and intrinsically colored physical objects will have been (and has already gradually begun to be) replaced by a competing and explanatorily more successful scientific categorial ontology. According to the latter, on Sellars’s reckoning, our experiences of the sensible qualities of things are ultimately to be fundamentally reconceived (and hence, recategorized) in such a way that – to put it very crudely – the given “expanse of red” ostensibly perceived “out there” on the surface of the apple finds its true ontological home, first, as an adverbial state of sensing in the of-a-red-expanse manner in the perceiver, and then ultimately as a “pure process” partly constituting what we would categorize (in macro-terms) as the perceiver’s central nervous system. (For brief classic statements, see PSIM 37, VI §107, and EPM XVI §61; for the fuller account, see FMPP passim.)

The key point arising from this account of the sensible qualities in relation to the myth of the categorial given, for Sellars, is that an experienced “expanse of red could be something actual” and yet turn out to be correctly categorized, in the end, in accordance with one of several different and incompatible ontological alternatives: for example (as Sellars puts it at one point), as being itself “either (1) a sense datum in visual space, (2) a manner of sensing, or (3) a spatial constituent of a physical object” (FMPP I, §88; numbering added). Sellars argues that his ‘givennist’ opponents, no matter how much they profess to be neutral with respect to such conceptual categorizations and recategorizations of the “immediately given” (in the case of the “sense-datum theorists,” for instance), or perhaps to be providing an entirely pre-theoretical phenomenology (Husserl, Merleau-Ponty) or linguistic analysis (Austin, Strawson) of the ordinary lifeworld, nonetheless inevitably presuppose an implicit categorization of such items that is incompatible with one or another of the possible ontological and scientific alternatives. For Sellars these implicit categorizations become particularly visible when seen by contrast with the possibility of his own radically revisionary scientific categorial ontology of sensory consciousness and of the physical world as a whole.

That is the real reason why the myth of the given was so important to Sellars: only a firm grip on the rejection of the myth of the categorial given can open up and keep alive the coherent possibility of the sort of radical scientific and metaphysical recategorization of both the nature of physical reality and of so-called immediate experience that Sellars took to be essential to an ultimately correct scientific categorial ontology of our own overall place in the scheme of things. Earlier I briefly alluded to the fact that, in addition to the key role of the categories in relation to the myth of the given and Sellars’s scientific naturalist ontology discussed in this section, Sellars also viewed his “psychological nominalist” metaconceptual theory of the categories and of abstract entities in general as equally indispensable for the possibility of a genuinely naturalistic ontology. Sellars rejects as insufficient the sort of merely methodological naturalism espoused by Quine, which is in principle consistent with the positing of any type of entity, whether it be Homeric gods or Cartesian souls or (as Quine argues science does require) abstract mathematical objects such as infinite hierarchies of sets. As Sellars wrote in a letter to Quine in 1978:

The question of whether or not the sort of comprehensive scientific-naturalist categorial ontology and thoroughgoing nominalism that Sellars envisioned is really achievable is beyond the scope of what I have attempted to show here. What I have argued is that Sellars’s metaconceptual theory of the categories exemplifies and extends a long line of nominalistic thinking about the nature of the categories from Ockham and Kant to the Tractatus and Carnap, and that this theory is far more central than has generally been realized to each of Sellars’s most famous and enduring philosophical conceptions: the myth of the given, the logical space of reasons, and resolving the ostensible clash between the manifest and scientific images of the human being in the world.