Pragmatism, Inferentialism, and Modality in Sellars's Arguments against Empiricism
Pragmatism, Inferentialism, and Modality in Sellars's Arguments against Empiricism
Abstract and Keywords
This chapter places the arguments against empiricism presented in the first half of ‘Empiricism and the Philosophy of Mind’ in the larger context opened up by laying them alongside the further battery of arguments Sellars aimed at the empiricist treatment of modality. The focus is on Sellars's pragmatist insistence on looking at what one must be able to do in order to deploy empirical descriptive vocabulary, and his rationalist commitment to the necessary inferential articulation of the concepts expressed by the use of such vocabulary. His view is that anyone who has the practical ability to deploy ‘purely descriptive’ vocabulary already knows how to do everything he needs to know how to do to deploy modal vocabulary as well, since what one is doing in making a modal claim is endorsing a pattern of inference.
In this essay I want to place the arguments of “Empiricism and the Philosophy of Mind” into a slightly less local context, by tracing further, into neighboring works, some strands of argumentation that intersect and are woven together in his critique of empiricism in its two principal then‐extant forms: traditional, and twentieth‐century logical empiricism. Sellars always accepted that observation reports resulting non‐inferentially from the exercise of perceptual language‐entry capacities play both the privileged epistemological role of being the ultimate court of appeal for the justification of empirical knowledge‐claims and therefore (given his inferentialist semantics) an essential semantic role in determining the contents of the empirical concepts applied in such judgments. But in accord with his stated aspiration to “move analytic philosophy from its Humean into its Kantian phase,” he was severely and in principle critical of empiricist ambitions and programs in epistemology and (especially) semantics that go beyond this minimal, carefully circumscribed characterization of the cognitive significance of sense experience. Indeed, I think the lasting philosophical interest of (p.34) Sellars's thought lies primarily in the battery of original considerations and arguments he brings to bear against all weightier forms of empiricism. Some, but not all, of these are deployed in the opening critical portions of “Empiricism and the Philosophy of Mind,” where the ground is cleared and prepared for the constructive theorizing of the last half. But what is on offer there is only part of Sellars's overall critique of empiricism. We accordingly court misunderstanding of what is there if we do not appreciate the shape of the larger enterprise to which it contributes.
In an autobiographical sketch, Sellars dates his break with traditional empiricism to his Oxford days in the thirties. It was, he says, prompted by concern with understanding the sort of conceptual content that ought to be associated with “logical, causal, and deontological modalities.” Already at that point he says that he had the idea that
This telling passage introduces two of the master ideas that shape Sellars's critique of empiricism. The first is that a key criterion of adequacy with respect to which its semantics will be found wanting concerns its treatment of modal concepts. The second is that the remedy for this inadequacy lies in an alternative broadly functional approach to the semantics of these concepts that focuses on their inferential roles—as it were, looking downstream to their use, as well as upstream to the circumstances that elicit their application.
what was needed was a functional theory of concepts which would make their role in reasoning, rather than supposed origin in experience, their primary feature.
This second, inferential‐functionalist, semantic idea looms large in “Empiricism and the Philosophy of Mind.” In fact, it provides the raw materials that are assembled and articulated into Sellars's positive account of the semantics of the concepts applied in reporting thoughts and sense‐impressions. Concern with the significance (p.35) of modality in the critique of empiricism, however, is almost wholly absent from that work (even though it is evident in articles Sellars wrote even earlier). I do not think that is because it was not, even then, an essential element of the larger picture of empiricism's failings that Sellars was seeking to convey, but rather because it was the result of a hard‐won but successful divide‐and‐conquer expository strategy. That is, I conjecture that what made it possible for Sellars finally to write “Empiricism and the Philosophy of Mind” was figuring out a way to articulate the considerations he advances there without having also at the same time to explore the issues raised by empiricism's difficulties with modal concepts. Whether or not that conjecture about the intellectual‐biographical significance of finding a narrative path that makes possible the separation of these aspects of his project is correct, I want to claim that it is important to understand what goes on in EPM in the light of the fuller picture of the expressive impoverishment of empiricism that becomes visible when we consider what Sellars says when he does turn his attention to the semantics of modality.
There is a third strand to the rope with which Sellars first binds and then strangles the excessive ambitions of empiricism. That is his methodological strategy of considering semantic relations among the meanings expressed by different sorts of vocabulary that result from pragmatic dependencies relating the practices one must engage in or the abilities one must exercise in order to count as using those bits of vocabulary to express those meanings. I will call this the ‘pragmatist’ element in Sellars's multi‐front assault on empiricism. It makes a significant contribution to the early, critical portion of EPM, though Sellars does not overtly mark it, as he does the contribution of his inferential functionalism to the later, more constructive portion. The concern with what one must do in order to say various kinds of things remains implicit in what Sellars does, rather than explicit in what he says about what he does. As we will see, both the pragmatist and the inferentialist ideas are integral to his critique of empiricist approaches to modality and to his constructive suggestions for a more adequate treatment of modal vocabulary.
(p.36) 2. The Inferentialist and Pragmatist Critique of Empiricism in EPM
I think the classical project of analytic philosophy in the twentieth century was to explore how the meanings expressed by some target vocabularies can be exhibited as in some sense a logical elaboration of the meanings already expressed by some base vocabularies. The conception of the desired semantic relation between vocabularies (the sense of ‘analysis’) has varied significantly within this broadly defined semantic project, including definition, paraphrase, translation, reduction in various senses, supervenience, and truth‐making, to name just a few prominent candidates. I take it to be integral to the analytic philosophical project during this period that, however that semantic relation is conceived, logical vocabulary is taken to play a special role in elaborating the base vocabulary into the target vocabulary. The distinctively twentieth‐century form of empiricism should be understood as one of the core programs of this analytic project—not in the sense that every participant in the project endorsed some version of empiricism (Neurath, for instance, rejects empiricism where he sees it clashing with another core semantic program that was dearer to his heart, namely naturalism), but in the sense that even those who rejected it for some target vocabulary or other took the possibility of an empiricist analysis to be an important issue, to set a legitimate philosophical agenda.
Construed in these terms, twentieth‐century empiricism can be thought of as having proposed three broad kinds of empiricist base vocabularies. The most restrictive kind comprises phenomenalist vocabularies: those that specify how things subjectively appear as opposed to how they objectively are, or the not‐yet‐conceptualized perceptual experiences subjects have, or the so‐far‐uninterpreted sensory given (the data of sensation: sense data). A somewhat less restrictive genus of empiricist base vocabularies limits them to those that express secondary qualities, thought of as what is directly perceived in some less demanding sense. And a still more relaxed version of (p.37) empiricism restricts its base vocabulary to the observational vocabulary deployed in non‐inferentially elicited perceptual reports of observable states of affairs. Typical target vocabularies for the first, phenomenalist, class of empiricist base vocabularies include those expressing empirical claims about how things really or objectively are—that is, those expressing the applicability of any objective empirical concepts. Typical target vocabularies for secondary‐quality empiricism include any that specify primary qualities or the applicability of concepts that are not response‐dependent. And typical target vocabularies for observational vocabulary empiricism include theoretical vocabulary. All species of empiricism are concerned with the possibility of underwriting the semantics of the modal vocabulary used to express laws of nature, probabilistic vocabulary, normative vocabulary, and other sophisticated vocabularies of independent philosophical interest. The standard empiricist alternatives are either to show how a given target vocabulary can be semantically elaborated from the favored empiricist base vocabulary, on the one hand, or to show how to live with a local skepticism about its ultimate semantic intelligibility, on the other.
At the center of Sellars's critique of empiricism in EPM is an argument against the weakest, least committive, observational, version of empiricism (a critique that then carries over, mutatis mutandis, to the more demanding versions). That argument depends on both his inferential‐functionalist semantics and on his pragmatism. Its fundamental strategy is to show that the proposed empiricist base vocabulary is not pragmatically autonomous: that observational vocabulary is not a vocabulary one could use though one used no other. Non‐inferential reports of the results of observation do not form an autonomous stratum of language. In particular, when we look at what one must do to count as making a non‐inferential report, we see that that is not a practice one could engage in except in the context of inferential practices of using those observations as premises from which to draw inferential conclusions, as reasons for making judgments and undertaking commitments that (p.38) are not themselves observations. The contribution to this argument of Sellars's inferential‐functionalism about semantics lies in underwriting the claim that for any judgment, claim, or belief to be cognitively, conceptually, or epistemically significant, for it to be a potential bit of knowledge or evidence, to be a sapient state or status, it must be able to play a distinctive role in reasoning: it must be able to serve as a reason for further judgments, claims, or beliefs, hence as a premise from which they can be inferred. That role in reasoning, in particular, what those judgments, claims, or beliefs can serve as reasons or evidence for, is an essential, and not just an accidental component of their having the semantic content that they do. And that means that one cannot count as understanding, grasping, or applying concepts non‐inferentially in observation unless one can also deploy them at least as premises in inferences to conclusions that do not, for that very reason, count as non‐inferential applications of concepts. Nor, for the same reason, can any discursive practice consist entirely of non‐inferentially acquiring premises, without any corresponding practice of drawing conclusions. So non‐inferential, observational uses of concepts do not constitute an autonomous discursive practice: a language‐game one could play though one played no other. And this conclusion about the pragmatic dependence of observational uses of vocabulary on inferential ones holds no matter what the subject‐matter of those observations is: whether it is observable features of the external environment, how things merely appear to a subject, or the current contents of one's own mind.
Here the pragmatist concern with what one must do in order to be able to say (or think) something combines with semantic inferentialist‐functionalism about conceptual content to argue that the proposed empiricist base vocabulary is not pragmatically autonomous—since one must be able to make claims inferentially in order to count as making any non‐inferentially. If that is so, then potentially risky inferential moves cannot be seen as an in‐principle optional superstructure erected on a semantically autonomous base of things directly known through observation.
(p.39) Although this is his most general and most powerful argument, Sellars does not limit himself to it in arguing against the substantially more committive forms of empiricism that insist on phenomenalist base vocabularies. In addition, he develops a constructive account of the relations between (at least one principle species of) phenomenalist vocabulary and objective vocabulary that depends on pragmatic dependences between what one must do in order to deploy each kind, to argue once again that the proposed empiricist base vocabulary does not form a semantically autonomous stratum of the language. This is his account of the relation between ‘looks’‐talk and ‘is’‐talk.
It develops out of his positive account of what one must do in order to use vocabulary observationally. To apply the concept green non‐inferentially one must be able to do at least two sorts of things. First, one must be able reliably to respond differentially to the visible presence of green things. This is what blind and color‐blind language‐users lack, but non‐language‐using pigeons and parrots possess. Second, one must be able to exercise that capacity by reliably responding differentially to the visible presence of green things by applying the concept green. So one must possess, grasp, or understand that concept. “Grasp of a concept is mastery of the use of a word,” Sellars says, and his inferential‐functionalism dictates that this must include the inferential use of the word: knowing at least something about what follows from and is evidence for or against something's being green. This the blind or color‐blind language‐user has, and the pigeon and parrot do not. Only the performances of the former can have the pragmatic significance of taking up a stand in the space of reasons, of committing themselves to something that has a conceptual, that is, inferentially articulated, content.
The point of Sellars's parable of John in the tie shop is to persuade us that the home language‐game of the ‘looks’ or ‘seems’ vocabulary that expresses how things merely appear to us, without undertaking any commitment to how they actually are, is one that is pragmatically parasitic on the practice of making in‐principle (p.40) risky reports of how things objectively are. For what one must do in order to count as saying how things merely look, Sellars claims, is to evince the reliable differential disposition to respond to something by claiming that it is green, while withholding the endorsement of that claim (because of one's collateral beliefs about the situation and one's reliability in it). If that is what one is doing in making a ‘looks’‐claim, then one cannot be wrong about it in the same way one can about an ‘is’‐claim, because one has withheld the principal commitment rather than undertaking it. And it follows that phenomenalist ‘looks’‐talk, which expresses how things merely appear, without further commitment to how things actually are, is not an autonomous discursive practice—not a language‐game one could play though one played no other—but is in fact pragmatically parasitic on objective ‘is’‐talk.
My point in rehearsing this familiar argument is to emphasize the role played both by Sellars's pragmatist emphasis on what one must be able to do in order to count as saying various kinds of thing—using vocabulary so as to express certain kinds of meanings—and by his inferentialist‐functionalist insistence that the role some vocabulary plays in reasoning makes an essential contribution to its semantic content. Although Sellars does not go on to make this argument, the way these two lines of thought conspire to undermine the semantic autonomy of candidate empiricist base vocabularies provides a template for a parallel objection to secondary‐quality empiricism. For at least a necessary condition on anything's being a secondary‐quality concept is that it have an observational role that supports the introduction of corresponding ‘looks’‐talk, so that mastery of that ‘looks’‐talk can be taken to be essential to mastery of the concept—as ‘looks‐green’ arguably is for mastery of the concept green, but ‘looks’‐square is not for mastery of the concept square. What would be needed to fill in the argument against secondary‐quality empiricism via the non‐autonomy of its proposed base vocabulary would be an argument that nothing could count as mastering a vocabulary consisting entirely of expressions of this sort, apart from all inferential (p.41) connections to primary‐quality concepts that did not have this structure.
3. A Tension within Empiricism about Modality
Thus far I have confined myself to offering a general characterization of anti‐empiricist arguments that appear in “Empiricism and the Philosophy of Mind.” None of them involve empiricism's treatment of modality. Now I want to put those arguments in a somewhat different frame, by conjoining them with one that is presented elsewhere, and which does turn on the significance of modal concepts. The previous arguments concerned the suitability of some vocabulary to serve as the base vocabulary of an empiricist analysis—since plausible motivations for caring about such an analysis typically require that it be semantically autonomous. This one turns on the criteria of adequacy of the analysis itself. My remarks in this section concern Sellars's arguments in his essay “Phenomenalism,” which was written soon after EPM, and can be regarded as a kind of companion piece to it. (Later I will discuss another contemporary essay that I think should be thought of as yoked together with these two in a troika.) The first, modal, point is one that Sellars registers there, but does not linger on—his principal concern being rather with a second point, concerning another aspect of the vocabulary in which phenomenalist analyses would have to be couched. But given my purposes here, I want to make a bit more of the modal point than he does.
The basic idea of a phenomenalist–empiricist semantic analysis of ordinary objective vocabulary is that the expressive work done by talk of mind‐independent objects and their properties and relations can be done by talk of patterns in, regularities of, or generalizations concerning sense experiences characterized in a phenomenalist vocabulary. Saying that the curved red surface I am experiencing is an experience of an apple that has parts I am not experiencing—a similarly bulgy, red back and a white interior, for instance—is (p.42) properly understood as saying something about what I would experience if I turned it around or cut it open. That it continued to exist in the kitchen when I left the room is a matter of what I would have experienced had I returned. The first, obvious, observation is that an account of objective reality in terms of the powers of circumstances to produce, or my dispositions to have, sensations, experiences, beings‐appeared‐to and so on essentially involves modal concepts. The patterns, regularities, or generalizations in subjective appearances that are supposed to constitute objective realities are modally robust, counterfactual‐supporting patterns, regularities, or generalizations. Talk of what I actually do experience will not by itself underwrite claims about unexperienced spatial or temporal parts of empirical objects. Twentieth‐century logical empiricism promised to advance beyond traditional empiricism because it could call on the full expressive resources of logical vocabulary to use as the ‘glue’ sticking sensory experiences together so as to construct simulacra of external objects. But extensional logical vocabulary is not nearly expressively powerful enough for the phenomenalist version of the empiricist project. So the phenomenalist conditional ‘terminating judgments’ into an infinite set of which C. I. Lewis proposes (in his Analysis of Knowledge and Valuation) to translate the ‘non‐terminating judgments’ of ordinary objective empirical discourse, have to use his modal notion of strict or necessary implication.2 And similar points could be made about other phenomenalist reductionists such as Ayer. The consequence of this observation to which I want to draw attention is that one cannot use such a strategy in one's phenomenalist empiricist analysis, translation, or reduction of objective talk and at the same time be a Humean skeptic about what modal vocabulary expresses. Essential features of the only remotely plausible constructive strategy of phenomenalist empiricism are simply incompatible with the most prominent skeptical consequences characteristically drawn both by traditional and twentieth‐century logicist empiricism.
(p.43) This is a powerful argument. Sellars's principal concern in his essay “Phenomenalism,” however, is with a subsequent point. The conditionals that codify the patterns, regularities, or generalizations concerning sense experience corresponding to judgments about how things objectively are must not only be subjunctive, counterfactually robust conditionals, but in order to have any hope of being materially adequate (getting the truth‐conditions even approximately correct) their antecedents must themselves be expressed in objective vocabulary, not in phenomenalist vocabulary. What is true (enough) is that if I were actually to turn the apple around, cut it open, or return to its vicinity in the kitchen I would have certain sense experiences. It is not in general true that if I merely seem to do those things I am guaranteed to have the corresponding experiences. For, phrased in such phenomenalist terms, the antecedent is satisfied in cases of imagination, visual illusion, dreaming, hallucination and so on that are precisely those not bound by the supposedly object‐constituting rules and regularities. As Sellars summarizes the point:
It is a mistake because:
To claim that the relationship between the framework of sense contents and that of physical objects can be construed on the [phenomenalist] model is to commit oneself to the idea that there are inductively confirmable generalizations about sense contents which are ‘in principle’ capable of being formulated without the use of the language of physical things. . . . [T]his idea is a mistake.
(PHM, in SPR: 87)
This argument then makes evident
[T]he very selection of the complex patterns of actual sense contents in our past experiences which are to serve as the antecedents of the generalizations in question presuppose our common sense knowledge of ourselves as perceivers, of the specific physical environment in which we do our perceiving and of the general principles which correlate the occurrence of sensations with bodily and environmental conditions. We select those patterns which go with our being in a certain perceptual relation to a particular object of a certain quality, where we know that (p.44) being in this relation to an object of that quality normally eventuates in our having the sense content referred to in the consequent.
(PHM, in SPR: 84)
So the phenomenalist vocabulary is not autonomous. It is not a language‐game one can play though one plays no other. In particular, the uses of it that might plausibly fulfill many of the same pragmatic functions as ordinary objective empirical talk themselves presuppose the ability to deploy such objective vocabulary.
the logical dependence of the framework of private sense contents on the public, inter‐subjective, logical space of persons and physical things.
(PHM, in SPR: 84)
As Sellars points out, the lessons learned from pressing on the phenomenalist version of empiricism apply more generally. In particular, they apply to the more liberal version of empiricism whose base vocabulary is observational, including observations of enduring empirical objects, and whose target vocabulary is theoretical vocabulary. To begin with, if talk of theoretical entities is to be translated into, or replaced by talk of patterns in, regularities of, or generalizations about observable entities, they must be lawlike, counterfactual‐supporting regularities and generalizations. They must permit inferences to what one would observe if one were to find oneself in specified circumstances, or to prepare the apparatus in a certain way. For, once again, the patterns, regularities, or generalizations about observations the assertion of which an instrumentalist empiricist might with some initial plausibility take to have the same pragmatic effect as (to be doing the same thing one is doing in) deploying theoretical vocabulary must reach beyond the parochial, merely autobiographically significant contingencies of what subjects happen actually to observe. The theory is that electrical currents cause magnetic fields regardless of the presence of suitable measuring devices. And that can only be made out in terms of what is observable, that is, could be observed, not just what is observed. And that (p.45) is to say that the instrumentalist‐observational form of empiricism is also incompatible with Humean–Quinean skepticism about the intelligibility of what is expressed by alethic modal vocabulary.
And an analogue of the second argument against phenomenalist forms of empiricism also applies to instrumentalist forms. For, once again, the antecedents of the counterfactual conditionals specifying what could or would have been observed if certain conditions had obtained or certain operations were performed cannot themselves be formulated in purely observational terms. The meter‐needle would have been observably displaced if I had connected the terminals of a volt‐ohmeter to the wire, but that something is a VOM is not itself a fact restatable in purely observational terms. Even leaving apart the fact that it is a functional characterization not equivalent to any specification in purely physical terms, a description of the construction of some particular kind of VOM is still going to help itself to notions such as being made of copper, or being an electrical insulator (another bit of vocabulary that is both functional and theoretical). To satisfy the semantic ambitions of the instrumentalist it is not enough to associate each theoretical claim with a set of jointly pragmatically equivalent counterfactual‐supporting conditionals whose consequents are couched wholly in observational vocabulary. All the theoretical terms appearing in the antecedents of those conditionals must be similarly replaced. No instrumentalist reduction of any actual theoretical claim has ever been suggested that even attempts to satisfy this condition.
Though Sellars does not, and I will not, pursue the matter, one expects that corresponding arguments will go through, mutatis mutandis, also for the kind of empiricism that seeks to understand the use of primary‐quality vocabulary wholly in terms of the use of secondary‐quality vocabulary. What we mean by talk of primary qualities will have to be cashed out in terms of its powers to produce, or our dispositions to perceive, secondary qualities—that is, in terms of modally robust, counterfactual‐supporting generalizations. And (p.46) it will be a challenge to specify the antecedents of a materially adequate set of such conditionals wholly in the official secondary‐quality vocabulary.
4. A Direct Argument Against Empiricist Skepticism about Modality
The arguments I have considered so far set limits to the semantic ambitions of phenomenalist and instrumentalist forms of analytic empiricism, first by focusing on the pragmatic preconditions of the required semantic autonomy of the proposed empiricist base vocabularies, and second by looking in more detail at the specific sorts of inferential patterns in the base vocabulary in terms of which it is proposed to reconstruct the circumstances and consequences of application of items in the various target vocabularies. Here it was observed that the material adequacy of such reconstructions seems to require the ineliminable involvement of terms from the target vocabulary, not only on the right side, but also on the left side of any such reconstruction—in the definiens as well as in the definiendum. Modality plays a role in these arguments only because the material adequacy of the reconstruction also turns out to require appeal to counterfactually robust inferences in the base vocabulary. Insofar as that is so, the constructive semantic projects of the phenomenalist, instrumentalist, and secondary‐quality forms of empiricism are at odds with the local semantic skepticism about what is expressed by alethic modal vocabulary that has always been a characteristic cardinal critical consequence of empiricist approaches to semantics, as epitomized for its traditional phase by Hume and for its logicist phase by Quine.
In another massive, pathbreaking essay of this period, “Counterfactuals, Dispositions, and the Causal Modalities”3 (completed in February of 1957), Sellars argues directly against this empiricist (p.47) treatment of modality, completing what then becomes visible as a two‐pronged attack on the principle contentions and projects of empiricism, only the opening salvos of which were fired in “Empiricism and the Philosophy of Mind.”4 His principal target here is the “tendency to assimilate all discourse to describing,” which he takes to be primarily “responsible for the prevalence in the empiricist tradition of ‘nothing‐but‐ism’ in its various forms (emotivism, philosophical behaviorism, phenomenalism) . . . ” (CDCM §103: 303). The form Sellars addresses in this essay is the Humean one that one can find in statements of laws of nature, expressed in alethic modal vocabulary that lets us say what is necessary and what is and is not possible, “nothing but” expressions of matter‐of‐factual regularities or constant conjunctions (though he claims explicitly that considerations corresponding to those he raises for causal modalities are intended to apply to logical and deontological modalities as well) (CDCM §103: 302–3). His arguments are directed against the view that holds modal vocabulary semantically unintelligible, on grounds of inability to specify what it is saying about what the world is like, how it is describing things as being, insofar as by using it we are asserting something that goes beyond endorsing the existence of non‐modally characterizable universal generalizations.
Hume found that even his best understanding of actual observable empirical facts did not yield an understanding of rules relating or otherwise governing them. Those facts did not settle which of (p.48) the things that actually happened had to happen (given others), that is, were (at least conditionally) necessary, and which of the things that did not happen nonetheless were possible (not ruled out by laws concerning what did happen). The issue here concerns the justifiability and intelligibility of a certain kind of inference: modally robust, counterfactual‐supporting inferences, of the kind made explicit by the use of modal vocabulary. Hume (and, following him, Quine) took it that epistemologically and semantically fastidious philosophers face a stark choice: either show how to explain modal vocabulary—the circumstances of application that justify the distinctive counterfactual‐supporting inferential consequences of application—in non‐modal terms, or show how to live without it, to do what we need to do in science without making such arcane and occult supradescriptive commitments.
This demand was always the greatest source of tension between empiricism and naturalism, especially the scientific naturalism that Sellars epitomized in the slogan: “[S]cience is the measure of all things, of what is that it is, and of what is not that it is not” (EPM, §42, in SPR: 173; in KMG: 253; in B: 83). For modern mathematized natural science shorn of concern with laws, counterfactuals, and dispositions—in short of what is expressed by alethic modal vocabulary—is Hamlet without the prince, not just an impotent Samson but an inert, unrecognizable fragmentary remnant of a once‐vital enterprise. Sellars's general recommendation for resolving this painful tension (felt particularly acutely by, and one of the principal issues dividing, the members of the Vienna circle) is to relax the exclusivism and rigorism he traces to empiricism's semantic descriptivism:
(p.49) Sensitized as we now are by Sellars's diagnoses of semantic autonomy claims as essential to various empiricist constructive and reconstructive projects, both in EPM and in the “Phenomenalism” essay, and familiar as we now are with his criticisms of them based on the inferentially articulated doings required to use or deploy various candidate base vocabularies, it should come as no surprise that his objections to critical empiricist suspicions of and hostility towards modality follow the same pattern. For the Humean–Quinean empiricist semantic challenge to the legitimacy of modal vocabulary is predicated on the idea of an independently and antecedently intelligible stratum of empirical discourse that is purely descriptive and involves no modal commitments, as a semantically autonomous background and model with which the credentials of modal discourse can then be invidiously compared.
[O]nce the tautology ‘The world is described by descriptive concepts’ is freed from the idea that the business of all non‐logical concepts is to describe, the way is clear to an ungrudging recognition that many expressions which empiricists have relegated to second‐class citizenship in discourse are not inferior, just different.
(CDCM §79: 282)
In this case, as in the others, the argument turns both on the pragmatism that looks to what one is doing in deploying the candidate base vocabulary—here “purely descriptive” vocabulary—and on the nature of the inferential articulation of that vocabulary necessary for such uses to play the expressive role characteristic of that vocabulary. The argument in this case is subtler and more complex than the others however. For one thing, I take it that Sellars does not deny the intelligibility‐in‐principle of purely descriptive discourse that contains no explicitly modal vocabulary.5 For another, there (p.50) are special difficulties involved in, and corresponding delicacies required for, working out the general pragmatist‐inferentialist strategy so as to apply it to this case, by specifying the relation between the expressive role distinctive of modal vocabulary, on the one hand, and what one is doing (in particular, the inferential commitments one is undertaking) in using ordinary, non‐modal, descriptive vocabulary itself, on the other.
The pragmatic dependency relation that lies at the base of Sellars's argument is the fact that:
Descriptive uses of vocabulary presuppose an inferentially articulated “space of implications,” within which some descriptions show up as reasons for or explanations of others. Understanding those descriptions requires placing them in such a space. This pragmatist claim about what else one must be able to do—namely, infer, explain, treat one claim as a reason for another—in order for what one is doing to count as describing connects to the use of modal vocabulary via the principle that:
. . . although describing and explaining (predicting, retrodicting, understanding) are distinguishable, they are also, in an important sense, inseparable. It is only because the expressions in terms of which we describe objects, even such basic expressions as words for perceptible characteristics of molar objects, locate these objects in a space of implications, that they describe at all, rather than merely label. The descriptive and explanatory resources of language advance hand in hand. . . .
(CDCM §108: 306–7)
That is, what one is doing in using modal expressions is explaining, justifying, or endorsing an inference. So what one is doing in (p.51) saying that As are necessarily Bs is endorsing the inference from anything's being an A to its being a B.
To make first hand use of these [modal] expressions is to be about the business of explaining a state of affairs, or justifying an assertion.
(CDCM §80: 283)
The first sort of difficulty I alluded to above stems from the fact that there are other ways of endorsing such a pattern of inference besides saying that all As are necessarily Bs. One's endorsement may be implicit in other things one does, the reasoning one engages in and approves, rather than explicit in what one says. So from the fact (assuming, as I shall, that it is a fact) that the activity of describing is part of an indissoluble pragmatic package that includes endorsing inferences and the fact that what one is doing in making a modal claim is endorsing an inference, it does not at all follow that there can be no use of descriptive vocabulary apart from the use of modal vocabulary. The second difficulty stems from the fact that although Sellars may be right that what one is doing in making a modal claim is endorsing a pattern of inference, it is clear that one is not thereby saying that an inference is good. When I say “Pure copper necessarily conducts electricity,” and thereby unrestrictedly endorse inferences from anything's being pure copper to its conducting electricity, I have nevertheless said nothing about any inferences, explanations, justifications, or implications—indeed, have said something that could be true even if there had never been any inferences or inferrers to endorse them, hence no describers or discursive practitioners at all.6 These two observations set the principal criteria of adequacy both for Sellars's positive working‐out of the pragmatist‐inferentialist treatment of modal vocabulary, and for his argument that the purely descriptive base vocabulary invoked by the empiricist critic of the semantic credentials of modal vocabulary lacks the sort of discursive autonomy the empiricist criticism presupposes and requires.
(p.52) Sellars's central rhetorical strategy in this essay is to address the issue of what is expressed by modal claims about necessary connections by offering:
Officially, he is even‐handed in his treatment of the vices and virtues of the empiricist, who denies that the use of modal vocabulary can express any legitimate semantic content beyond that expressed by a descriptive, extensional universal generalization, and the rationalist, who understands that content in terms of entailments expressing rules of reasoning. In fact, however, as becomes clear when he launches into his own account, he is mainly concerned to develop a version of the rationalist account. As the second half of the essay develops, Sellars marks his abandonment of the disinterested pose by an uncharacteristically explicit expository shift:
. . . a sympathetic reconstruction of the controversy in the form of a debate between a Mr. C (for Constant Conjunction) and a Mr. E (for Entailment) who develop and qualify their views in such a way as to bring them to the growing edge of the problem.
(CDCM Introduction: 226)
Doing that requires careful investigation of the differences between and relations among four different sorts of item: (p.53)
It is now high time that I dropped the persona of Mr. E, and set about replying to the challenge with which Mr. C ended his first critique of the entailment theory.
(CDCM §85: 286)7
• Practical endorsement of the propriety of an inference from things being A to their being B;
• The explicit statement that one may infer the applicability of ‘B’ from the applicability of ‘A’;
• The statement that A physically entails B;
• The statement that As are necessarily Bs.
The third sort of statement expresses Mr. E's initial stab at an analysis of the fourth. It is the answer to the question: what sort of entailment is it that modal statements are supposed to express?:
The virtue of statements like “A physically entails B” is that they do plausibly codify the practical endorsement of an inference (p.54) that is implicit in what one does in the form of something one can explicitly say, without bringing in irrelevant commitments concerning particular expressions, the activity of inferring, or discursive practitioners. The remaining difficulty is that they seem plainly not to have the same content, not to say the same thing, as explicitly modal statements of objective necessity.
Mr. E has a ready answer. . . . it might . . . be called ‘natural’ or ‘physical’ entailment, for while any entailment is a logical relation, we can distinguish within the broad class of entailments between those which are, and those which are not, a function of the specific empirical contents between which they obtain. The latter are investigated by general or formal logic (and pure mathematics). Empirical science, on the other hand, to the extent that it is a search for laws, is the search for entailments of the former kind. (Putative) success in this search finds its expression in statements of the form ‘It is (inductively) probable that A physically entails B.’
(CDCM §56: 268)
Sellars's response to this problem is to acknowledge that modal statements do not say that some entailment holds, but to distinguish between what is said by using a bit of vocabulary and what is ‘contextually implied’ by doing so. Sellars says very little about this latter notion, even though it bears the full weight of his proposed emendation of the rationalist account. It is recognizably the same distinction he had appealed to earlier, in “Inference and Meaning,” as the distinction between what making a statement says and what it conveys. There his example is that in asserting “The weather is fine today,” I say that the weather is fine today, but convey that I believe that it is fine.9 That otherwise uninterpreted example suggests to me that what Sellars has in mind is the distinction between semantic and pragmatic inferences. That is the distinction between inferences underwritten by the contents of what is said or asserted, on the one hand, and inferences underwritten by what one is doing in saying them, on the other. The inference from “The weather is fine,” to “It is not raining,” is of the first sort; the inference from my asserting “The weather is fine,” to “Brandom believes the weather is fine,” is of the second sort. Inferences of these two kinds may generally be distinguished by the Frege–Geach embedding test: look to see whether those who make the inference in question also endorse the corresponding conditional. “If the weather is fine, then it is not raining,” is generally true, while “If the weather is fine, then Brandom believes it is fine,” is not generally true. (Compare the inference from my saying “That is an ugly tie you are wearing,” to “Bob is annoyed with me.”)
(p.55) If that is in fact the distinction Sellars is after, then it seems to me that the view he is expounding and defending can be put less paradoxically if we don't take a detour through entailment statements, but concern ourselves directly with the relation between the endorsement of patterns of inference and modal statements. The underlying rationalist insight is a pragmatist‐inferentialist one: what one is doing in making a modal claim is endorsing a pattern of inference. Modal vocabulary makes possible new kinds of sayings that have the pragmatic effect of endorsing inferences. To say that is not yet to say what they say, only what one is doing by saying them. But it does settle the pragmatic significance of such modal claims, in the sense of their appropriate circumstances and consequences of application.10 If one practically endorses the pattern of inference that treats classifying anything at all as an A as sufficient grounds (p.56) (“all on its own,” as Sellars says, in order to capture the way the pattern of inferences in question is counterfactually robust) for concluding that it is a B, then one is committed to the claim that all As are necessarily Bs. And commitment to that claim is commitment to practically ratify that pattern of inference. Assuming, as Sellars has claimed, that using ordinary, non‐modal, descriptive vocabulary requires practically endorsing such patterns of inference (“situating descriptions in a space of implications”), that means that anyone who has the practical ability to deploy “purely descriptive” vocabulary already knows how to do everything he needs to know how to do to deploy modal vocabulary as well. He need not actually do so, since practically undertaking those inferential commitments does not require that one have available a language with vocabulary permitting one to do that by saying something. But all a practitioner lacks in such a circumstance is the words to hook up to discriminative and responsive abilities he already possesses. In this precise sense, the ability to deploy modal vocabulary is practically implicit in the ability to deploy non‐modal descriptive vocabulary.
Sellars has claimed that the activity of describing is unintelligible except as part of a pragmatic package that includes also not just the making of inferences, but the making of counterfactually robust inferences: the sort of inferences involved in explanation, and licensed by explicitly modal statements of laws. Sellars summed up the claim admirably in the title of another one of his early papers: “Concepts as Involving Laws, and Inconceivable without Them.” Grasp of a concept is mastery of the use of a word, Sellars says. And that use includes not only sorting inferences (however fallibly and incompletely) into materially good and materially bad ones, but also, among the ones one takes to be materially good, to distinguish (however fallibly and incompletely) between counterfactual circumstances under which they do, and counterfactual circumstances under which they do not, remain good. Part of taking an inference to be materially good is knowing something about which possible additional collateral premises or auxiliary hypotheses would, and which would not, infirm it. Chestnut trees (p.57) produce chestnuts—unless they are immature, or blighted. Dry, well‐made matches strike—unless there is no oxygen. The hungry lioness would still chase the antelope if it were Tuesday or the beetle on the distant tree crawled slightly further up the branch, but not if the lioness's heart were to stop beating. The point is not that there is any particular set of such discriminations that one must be able to make in order to count as deploying the concepts involved. It is that if one can make no such practical assessments of the counterfactual robustness of material inferences involving those concepts, one could not count as having mastered them.
Against the background of this pragmatist‐inferentialist claim about what is involved in the ordinary descriptive use of concepts, Sellars's claim, as I am reading him, is that explicitly modal ‘lawlike’ statements are statements that one is committed or entitled to whenever one is committed or entitled to endorse such patterns of counterfactually robust inference, and commitment or entitlement to which in their turn commit or entitle one to the corresponding patterns of inference. Saying that about them settles what one needs to do to use such modal statements. It does not say how one is thereby describing the world as being when one does. It does not, in particular, describe a pattern of inference as good (though that saying does, in its own distinctive way, express endorsement of such a pattern). It does not do those things for the simple reason that the use of modal expressions is not in the first instance descriptive.11 It codifies explicitly, in the form of a statement, a feature of the use of descriptive expressions that is indissolubly bound up with, but not identical to, their descriptive use. Nonetheless, in knowing how to use vocabulary descriptively, one knows how to do everything one needs to know how to do in order to use modal vocabulary. And that is enough to show that one cannot actually be in the Humean predicament presupposed by the empiricist challenge to the intelligibility of modal vocabulary. For one cannot know how (p.58) to use vocabulary in matter‐of‐factual descriptions (“The cat is on the mat,”) and not have any grip on how to use modal, counterfactual, and dispositional vocabulary (“It is necessary for live cats to breathe,” “The cat could still be on the mat if the mat were a slightly different shade of blue, but not if it turned into soup,” “The cat would leave the mat if she saw a mouse,”). Although explicitly modal vocabulary is an in‐principle optional superstructure on practices of deploying descriptive vocabulary, what it expresses cannot be mysterious in principle to those who can engage in those base‐level practices.
In taking this line, Sellars quite properly sees himself as reviving a central idea of Kant's. The ability to use empirical descriptive terms such as ‘mass’, ‘rigid’, and ‘green’ already presupposes grasp of the kind of properties and relations made explicit by modal vocabulary. It is this insight that leads Kant to the idea of ‘pure’ concepts or ‘categories’, including the alethic modal concepts of necessity and possibility that articulate causal laws, which must be available a priori because and in the sense that the ability to deploy them is presupposed by the ability to deploy ordinary empirical descriptive concepts. The categories, including modality, are concepts that make explicit what is implicit in the empirical, descriptive use of any concepts at all. Though the details of which laws, the statements of which express counterfactually robust patterns of inference, actually obtain is an empirical one, that empirical descriptions are related by rules in the form of laws, which do support counterfactually robust inferences, is not itself an empirical matter, but a truth about the framework of empirical description. I want to call the underlying insight “the Kant–Sellars thesis about modality.” It is the claim that in being able to use non‐modal, empirical‐descriptive vocabulary, one already knows how to do everything one needs to know how to do in order to deploy modal vocabulary, which accordingly can be understood as making explicit structural features that are always already implicit in what one does in describing.
(p.59) 5. Conclusion
Articulating and justifying his version of the Kant–Sellars thesis about modality is Sellars's constructive response to the empiricist tradition's “nothing‐but‐ism” about modality: its demand that what is expressed by modal claims either be shown to be expressible in non‐modal terms, or be dispensed with entirely by semantically fastidious philosophers and scientists. This complements and completes his demonstration, in the “Phenomenalism” essay, that this critical consequence of an over‐ambitious empiricism is in any case incompatible with any constructive empiricist effort to reconstruct or replace the use of target vocabularies such as objective‐descriptive vocabulary, primary‐quality vocabulary, and theoretical vocabulary in terms of the favored empiricist base vocabularies, if that effort is subject to even the most minimal criteria of material adequacy. Together, these arguments show what Sellars eventually made of his early intuition that the soft underbelly of empiricism, in both its traditional and its twentieth‐century logistical form, is its treatment of modality.
My overall aim in this essay has been to place the arguments against empiricism presented in the first half of “Empiricism and the Philosophy of Mind” in the larger context opened up by laying them alongside the further battery of arguments aimed at the same target that derive from consideration of that tradition's views about modality. And I have been concerned to show that the methodological strategies that guide all of these discussions are Sellars's pragmatist insistence on looking at what one must be able to do in order to deploy empirical descriptive vocabulary, and his rationalist commitment to the necessary inferential articulation of the concepts expressed by the use of such vocabulary. I think that even fifty years on, there is still a lot of juice to be squeezed out of these ideas.
But I want to close with another, perhaps more frivolous suggestion. Every sufficiently engaged reading becomes a rewriting, and I have been offering here, inter alia, the outline of a different (p.60) narrative strategy that Sellars could have adopted in the late 1950s. Under some such title as The Limits of Empiricism, he could have re‐presented the material that in fact appeared first as roughly the first half of “Empiricism and the Philosophy of Mind,” and the second halves of each of “Phenomenalism” and “Counterfactuals, Dispositions, and Causal Modalities,” organized around and introduced in terms of the themes I have traced here. It is interesting to speculate about how his reception might have been different—and about where we would find ourselves today—had this been the shape of Sellars's first book.
Lewis, Clarence Irving, An Analysis of Knowledge and Valuation (La Salle, IL: Open Court, 1946).
Sellars, Wilfrid. “Concepts as Involving Laws, and Inconceivable without Them,” Philosophy of Science 15 (1948): 287–315.
Reprinted in J. Sicha (ed.) Pure Pragmatics and Possible Worlds: The Early Essays of Wilfrid Sellars (Reseda, CA: Ridgeview Publishing Company, 1980).
—— “Inference and Meaning” Mind 62 (1953): 313–38.
Reprinted in J. Sicha (ed.) Pure Pragmatics and Possible Worlds: The Early Essays of Wilfrid Sellars (Reseda, CA: Ridgeview Publishing Company, 1980). Cited as IM.
—— “Empiricism and the Philosophy of Mind,” (Presented at the University of London in Special Lectures in Philosophy for 1956 under the title “The Myth of the Given: Three Lectures on Empiricism and the Philosophy of Mind”), in Minnesota Studies in the Philosophy of Science, vol. I, eds. Herbert Feigl and Michael Scriven (Minneapolis: University of Minnesota Press, 1956): 253–329.
Reprinted in SPR with additional footnotes. Published separately as Empiricism and the Philosophy of Mind: with an Introduction by Richard Rorty and a Study Guide by Robert Brandom, ed. Robert Brandom (Cambridge, MA: Harvard University Press, 1997). (Cited as B.)
Also reprinted in W. deVries and T. Triplett, Knowledge, Mind, and the Given: A Reading of Sellars’ “Empiricism and (p.61) the Philosophy of Mind” (Cambridge, MA: Hackett Publishing, 2000). Cited as EPM, page references to SPR, KMG, and B editions.
—— “Counterfactuals, Dispositions, and the Causal Modalities,” in Minnesota Studies in The Philosophy of Science, vol. II, eds. Herbert Feigl, Michael Scriven, and Grover Maxwell (Minneapolis: University of Minnesota Press, 1957): 225–308. Cited as CDCM.
—— Science, Perception and Reality (London: Routledge & Kegan Paul, 1963). Reissued by Ridgeview Publishing Company in 1991. Cited as SPR.
—— “Phenomenalism,” in Science, Perception and Reality (London: Routledge & Kegan Paul, 1963): 60–105. Cited as PHM.
—— “Autobiographical Reflections,” in Action, Knowledge, and Reality, H. N. Castañeda (ed.) (Indianapolis: Bobbs‐Merrill, 1975): 277–93. Cited as AR.
—— J. Sicha (ed.) Pure Pragmatics and Possible Worlds: The Early Essays of Wilfrid Sellars (Reseda, CA: Ridgeview Publishing Company, 1980). Cited as PPPW. (p.62)
(3) “Counterfactuals, Dispositions, and the Causal Modalities,” in Minnesota Studies in The Philosophy of Science, vol. II, eds. Herbert Feigl, Michael Scriven, and Grover Maxwell (Minneapolis: University of Minnesota Press, 1957): 225–308. Henceforth CDCM.
(4) As in EPM (and even, though to a lesser extent, in “Phenomenalism”), in this essay Sellars describes himself not as denying empiricism, but rather as correcting it, protecting its core insights from the damage done by their over‐extension. But he also makes it clear that the result of such rectification is a Kantian view that gives equal weight to rationalist insights, when they are suitably reconstructed. So for instance he says:
And the final sentence of the essay invokes the “profound truth” of Kant's conception of reason, “which empiricism has tended to distort.”
It is my purpose to argue that the core truth of Hume's philosophy of causation is not only compatible with, but absurd without, ungrudging recognition of those features of causal discourse as a mode of rational discourse on which the “metaphysical rationalists” laid such stress, but also mis‐assimilated to describing.
(CDCM, §82: 285)
(5) This conclusion requires some interpretation, as Sellars is, frustratingly but characteristically, not explicit on the point. He says:
Sellars's view about this ideal is complex: there is sense in which it is intelligible, and a sense in which it is not. Such a discourse would be unreflective and unself‐conscious in a way ours is not. For reasons that will emerge, it would belong to what at the end of the essay he calls the stage of human language “when linguistic changes had causes, but not reasons, [before] man acquired the ability to reason about reasons” (CDCM §108: 307).
The idea that the world can, in principle, be so described that the description contains no modal expression is of a piece with the idea that the world can, in principle, be so described that the description contains no prescriptive expression. For what is being called to mind is the ideal of statement of ‘everything that is the case’ which, however, serves through and through only the purpose of stating what is the case. And it is a logical truth that such a description, however many modal expressions might properly be used in arriving at it or in justifying it, or in showing the relevance of one of its components to another, could contain no modal expression.
(CDCM §80: 283)
(6) Sellars connects this obvious fact with the observation that:
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.
(CDCM §101: 301)
(7) In fact, Sellars's ‘defense’ of Mr. C (see the passage from §82 quoted in note 3 above) consists of showing what concessions he needs to make to Mr. E. This proceeds first by Mr. C's qualification that “ ‘A causes B’ says that (x)[Ax→Bx] and implies that the latter is asserted on inductive grounds” (§62: 272), followed by the necessity of conceiving “of induction as establishing principles in accordance with which we reason, rather than as major premises from which we reason” (§83: 286). As will appear, the former concession, introducing the notion of what is contextual implied by contrast to what is explicitly said, is then dialectically made available to be pressed into service by Mr. E. This bit of dialectic is a pretty rhetorical flourish on Sellars's part, but I doubt that in the end it reflects any deep feature of the confrontation between the empiricist and rationalist approaches to modality.
(8) As Sellars says:
But one can know that Turks, for example, ought to withdraw ‘ . . . ’ when they commit themselves to ‘‐ ‐ ‐’ without knowing the language, whereas the statement that ‘p entails q’ contextually implies that the speaker not only knows the language to which ‘p’ and ‘q’ belong, but, in particular, knows how to use ‘p’ and ‘q’ themselves.
(CDCM §81: 284)
(9) Wilfrid Sellars, “Inference and Meaning” [IM] Mind 62 (1953): 332. Reprinted in J. Sicha (ed.) Pure Pragmatics and Possible Worlds: The Early Essays of Wilfrid Sellars (Reseda, CA: Ridgeview Publishing Company, 1980) [PPPW]: 280.
(10) It is the attempt to specify this peculiar and distinctive sort of pragmatically mediated relation between vocabularies that leads Sellars to say things like:
It is sometimes thought that modal statements do not describe states of affairs in the world, because they are really metalinguistic. This won't do at all if it is meant that instead of describing states of affairs in the world, they describe linguistic habits. It is more plausible if it is meant that statements involving modal terms have the force of prescriptive statements about the use of certain expressions in the object language. Yet there is more than one way to ‘have the force of’ a statement, and failure to distinguish between them may snowball into a serious confusion as wider implications are drawn.
(CDCM §81: 283)
Shall we say that modal expressions are metalinguistic? Neither a simple ‘yes’ nor a simple ‘no’ will do. As a matter of fact, once the above considerations are given their proper weight, it is possible to acknowledge that the idea that they are metalinguistic in character oversimplifies a fundamental insight. For our present purposes, it is sufficient to say that the claim that modal expressions are ‘in the metalanguage’ is not too misleading if the peculiar force of the expressions which occur alongside them (represented by the ‘p’ and the ‘q’ of our example) is recognized, in particular, that they have ‘straightforward’ translation into other languages, and if it is also recognized that they belong not only ‘in the metalanguage’, but in discourse about thoughts and concepts as well.
(CDCM §82: 284)
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.
(CDCM §101: 301)
(11) Sellars says: “[Mr. E.] conceives of induction as establishing principles in accordance with which we reason, rather than as major premises from which we reason” (CDCM §83: 286).