Hardly any reader would miss that there is some connection between Wilfrid Sellars’ Empiricism and the Philosophy of Mind (hereafter EPM) and Gilbert Ryle’s The Concept of Mind (hereafter CM).¹ Throughout the essay, Sellars occasionally mentions the Oxford philosopher by name, calls the protagonists of his main thought experiment “Ryleans,” and employs several of Ryle’s own terms, suggesting close familiarity with the work.² A fairly standard narrative about the history of the philosophy of mind holds that Sellars successfully refutes Ryle’s alleged philosophical behaviorism in favor of an early form of functionalism. However, closer examination of Sellars’ work shows that he retains a good amount from what he takes to be Ryle’s account of the mind,³ leaving the precise extent of his divergence from the Rylean picture hard to gauge. He explicitly rejects at least one aspect of it, namely the idea that the notion of “inner episodes” is a category mistake (EPM, sec. 45, 88); other passages that initially seem to suggest further discord reveal, upon careful inspection, a more ambiguous stance towards Ryle’s text.⁴

Despite the amount of attention that CM and EPM have received over the past half century, little work has examined their direct interpretive relation.⁵ This means (i) assessing the extent to which the Rylean account targeted by Sellars aligns with CM; (ii) evaluating what Ryle, on his own terms, would or should have learned from Sellars’ picture; (iii) considering what Sellars could have learned from a more accurate reading of Ryle, including his later essays. In what follows, my goal is to begin working through these questions. I shall argue, in Section 1, that Sellars mischaracterizes Ryle’s work as putting forward a form of what he calls philosophical behaviorism. In Section 2, I will try to show that, notwithstanding Sellars’ superficial reading, the myth of Jones in EPM suggests a way out of Ryle’s own dissatisfaction with his treatment of what Sellars calls “immediate experiences”—such as sense impressions and sensations. He does this by noticing that theoretical knowledge can transform our descriptive accounts of the world. In Sellars’ words: terms that are originally introduced as theoretical postulates, once properly integrated in our scientific and folk theories, can gradually come to play a role in observational sentences or reports. Sellars suggests that this might be the case of words such as “sense impression” and “sensation.” I argue that, to work through Ryle’s dissatisfactions about the contemporary use of these terms, we do not need to establish whether these have actually reached full-blown reporting status: it suffices for the transformative process that Sellars theorizes to be so much as possible. In Section 3, I will suggest that a more careful engagement with Ryle’s work on Sellars’ part would have dispelled some of the dilemmas that motivate the first half of the myth of Jones, which is devoted to thoughts as inner episodes. The reason is that Sellars, as well as the functionalist tradition after him, failed to grasp the full scope of Ryle’s rejection of the Cartesian dogma, as well as its implications for the notions of “overt behavior” and of “thinking.” This critique can be properly understood only if one takes “intelligent capacity” as an indispensable category within Ryle’s concept of mind. For Ryle, when we correctly say that someone is thinking we are not primarily reporting the occurrence of an overt or covert episode; rather, we are construing her activity as taking place within a broader intelligent practice—such as playing chess or doing mathematics. In other words, we are describing her actions as exercises of an intelligent (or rational) capacity. This paper is thus an invitation to learn from Ryle’s work while making room for Sellars’ insight about the possibility of enrichment and transformation of our “mental” vocabulary. The blind spot I detect in Sellars’ reading of Ryle and in his conception of thinking is one that, I believe, is also characteristic of most popular contemporary theories in the philosophy of mind.

1

I shall first introduce Sellars’ Ryle and say a few things about how he differs from the author of CM. As mentioned above, Sellars seldom attributes to Ryle the views he seeks to reject; indeed, he only does so once, in a passage from chapter X, which I shall shortly examine. Before doing so, it is useful to take a brief detour through chapter XIV of EPM, where Ryle is not mentioned, and where Sellars distinguishes between two varieties of behaviorism, which he dubs “methodological” and “philosophical.”

Methodological behaviorism, for Sellars, is not a thesis about the analysis of existing mental concepts, but a requirement concerning the construction of new concepts in psychology (EPM, sec. 53, 99). It is the demand that all concepts within a theory of mind be introduced in terms of vocabulary pertaining to overt behavior—even though they might be theoretical concepts (EPM, sec. 53, 100). It thus involves no metaphysical commitments about the nature of the mental. In particular, it does not deny the possibility of introspection or of privileged access to one’s own mental states; it merely claims that if we introspect, we do so in terms of common sense mentalistic concepts (EPM, sec. 53, 99). Sellars contrasts this version of behaviorism—which he favors—to what he calls “philosophical” behaviorism, the thesis that common-sense psychological concepts are fully analyzable into concepts pertaining to overt behavior (EPM, 99-100). This position is one that he finds excessively restrictive to the practice of psychology. In “Philosophy and the Scientific Image of Man” (hereafter PSIM), Sellars calls methodological behaviorism “simply good sense” and says that it “belongs to the manifest rather than the scientific image.” It is an example of “the continuing correlational sophistication of the manifest image” (SR, 390-91). At the same time, this form of behaviorism is compatible with the scientific image, as it does not deny any of the claims the latter makes about unobservable mental processes. By contrast, philosophical behaviorism precludes the scientific image from contributing to the study of human psychology (and vice versa), as it dismisses the possibility of interpreting the concepts of inner episodes as theoretical or postulated. For the philosophical behaviorist the notion of “thought,” for example, is only meaningful insofar as it can be explained in terms of overt behavior, and not in its use as a postulated mental episode with potential explanatory power in a broader scientific context. Thus, the observation that A is thinking about how to solve problem X is to be analyzed via categorical and dispositional statements such as “A looks concentrated” or “if asked to explain what X is, A would respond with such-and-such an answer.” The philosophical behaviorist sees no need to explain the statement in terms of A’s unobservable mental processes.

With this distinction in hand, we may now turn to chapter X, which contains the only explicit claim in EPM about Ryle’s commitments. It is worth quoting the passage in full:

… many philosophers who would not deny that there are short-term hypothetical and mongrel hypothetical-categorical facts about behavior which others can ascribe to us on behavioral evidence, but which only we can report, have found it to be logical nonsense to speak of non-behavioral episodes of which this is true. Thus, it has been claimed by Ryle that the very idea that there are such episodes is a category mistake, while others have argued that though there are such episodes, they cannot be characterized in intersubjective discourse …. It is my purpose to argue that both these contentions are quite mistaken, and that not only are inner episodes not category mistakes, they are quite “effable” in intersubjective discourse. And it is my purpose to show, positively, how this can be the case. I am particularly concerned to make this point in connection with such inner episodes as sensations and feelings, in short, with what has—unfortunately, I think—been called “immediate experience.” For such an account is necessary to round off this examination of the Myth of the Given. But before I can come to grips with these topics, the way must be prepared by a discussion of inner episodes of quite another kind, namely thoughts. (EPM, sec. 45, 88)

Here Sellars is claiming that, for Ryle, the notion that there are non-publicly-witnessable inner episodes—such as “immediate experiences” and thoughts—is a category mistake. This means that, for Sellars’ Ryle, the very notion of inner episode is a myth, and consequently that the terms supposedly designating particular inner episodes must be either rejected as meaningless or analyzed in terms of overt behavior. In other words, Sellars’ Ryle is a philosophical behaviorist.⁶ Sellars seems to think that this commitment underlies the whole project of CM, a book that he elsewhere describes as a “painstaking analysis of mentalistic terms”⁷ in terms of categorical, hypothetical, as well as mongrel categorical-hypothetical statements referring solely to overt behavior.⁸ This is, for Sellars, an overly reductive account of thoughts and immediate experiences.⁹ Let us now turn to why the author of CM was no philosophical behaviorist.¹⁰

For Sellars, Ryle’s alleged rejection of the notion of inner episodes opens up two ways of dealing with any given term that supposedly refers to an inner episode. One option is to reject the term as meaningless. This is indeed what the author of CM does with a handful of concepts, such as that of “volition” (CM, 62) and that of “sense-datum” (CM, 214), which he considers fantasies created by philosophers under the spell of the Cartesian dogma. However, in most cases, this is not feasible, since hardly anyone would be willing to say that the term “thinking,” for instance, is meaningless. Sellars’ Ryle concludes that these terms must be understood as shorthand for complexes analyzable in terms of publicly observable facts. If this were what Ryle actually thought, when turning to concrete examples of such analyses in CM, we would find no publicly unobservable remainder: Ryle’s discussion of any given mental term would exclusively contain categorical-hypothetical statements referring ultimately to overt behavior. However, this is not quite what we find. The book makes liberal use of words that prima facie refer to what Sellars would call “inner episodes,” without attempting to explain how these can be further analyzed in terms of overt behavior. In the chapter on emotions, for instance, Ryle analyses the ascription of a motive to someone by mentioning, among other things, the person’s past “fancies and feelings” (CM, 90) as well as their inclination to indulge in “certain sorts of daydreams” (CM IV(2), 92); yet, the activities of “fancying” or “daydreaming” are never subject to an analysis in terms of overt behavior. In the same chapter, Ryle explicates ascriptions of mood in terms of being “in the frame of mind to say, do and feel a wide variety of loosely affiliated things” (CM IV(4), 99, my emphasis). In chapter VIII, the notions of imagining and mental picturing are described as wholly legitimate, despite their proneness to being misunderstood as ghostly events.¹¹ Again, these are not notions that Ryle feels the need to fully reduce to complexes of overt behavior. Similarly, Ryle never denies that we engage in non-publicly-witnessable thoughtful activities such as silent soliloquy¹² or mental calculation.¹³ This example is particularly significant because the first half of Sellars’ myth of Jones is an attempt to account for the notion of ‘thought’ as an inner episode, which might be read as implying that Ryle rejected any such idea.^14,15

Even when he goes on to explicate mental concepts in terms of publicly witnessable events, Ryle is careful never to give the impression of providing a complete or reductive analysis of these concepts. He usually frames his discussion as providing “part of” or “a minimal component of” the whole story, thus implying that more could be said.¹⁶ He thereby markedly distances himself from firmly committing to any philosophical programme akin to what Sellars calls philosophical behaviorism.¹⁷ Crucially, Ryle does not make any explicit claim about the supposed analyzability of all mental terms in terms of overt behavior. On the contrary, he explicitly distances himself from the two variants of behaviorism he discusses in the last section of chapter X of CM. While it is true that he works through some of the puzzles stemming from a Cartesian interpretation of inner episodes—for instance by reframing them in terms of words that are less prone to a dualistic interpretation—this strategy need not be motivated by philosophical behaviorist goals. The point of focusing on overt behavior is, in my reading, that of loosening the grip of the myth of the Ghost in the Machine on our conceptual vocabulary.

The foregoing should suffice to show that Ryle was no philosophical behaviorist, meaning that he did not believe all mental vocabulary to be either the result of a category mistake or shorthand for descriptions involving solely publicly observable events. Should we then conclude that Sellars’ critique of Ryle’s analysis of the mind is merely the result of a superficial reading of CM? In the next section, I shall argue that this is not the whole story.

2

To bring out what I take to be a genuine point of divergence between the two thinkers, I shall focus on their discussions of the notion of sense impression and other “immediate experiences.” These notions raise issues that Ryle thought he was unable to address, both in CM and in his later writings. I argue that Sellars’ account of “immediate experiences” suggested in the notorious “myth of Jones” (chapters XII-XVI of EPM) illuminates a blind spot in Ryle’s account of mindedness that lies at the root of his dissatisfaction with his own account of sensation. In this respect, Sellars’ story enriches rather than rejects Ryle’s account.

In the chapter “Sensation and Observation” of CM, Ryle dismisses the notion of sense impression and kindred terms as the mere product of a mythical—Cartesian—picture of the mind. He concludes that “we have no employment for ‘object of sense’, ‘sensible object’, ‘sensum’, ‘sense datum’, ‘sense-content’, ‘sense-field’ and ‘sensibilia’; the epistemologist’s transitive verb ‘to sense’ and his intimidating ‘direct awareness’ and ‘acquaintance’ can be returned to store” (CM, 221). He extends this verdict to the more sophisticated uses of the nouns “sensation” and “impressions,” even though both of these bear meaningful ordinary senses (CM, 240, 242).¹⁸ In particular, “sensation” and “impression” are denied from playing any role in an account of perception or observation (CM, 210), pace the sense-data theorist (CM, 213). For Ryle, the issue with sense-data theories has to do with their confusing a causal, para-mechanical explanation for a philosophical account of the ordinary phenomenon of sense perception. When we talk of someone learning how to perceive, “we are not discovering any causal laws, such as those of physiology, optics or mechanics” (CM, 233-34). Even though we can study the causal phenomena—auditory or optical—that perception necessarily entails, Ryle takes the epistemologists’ notion of sensation or impression to be wholly unhelpful for the task, as well as for the philosophical one of elucidating the ordinary notion of perception. He accuses the epistemologist’s attempt to build a theoretical bridge between the two tasks of covertly relying on a para-mechanical conception of the mind.

However, Ryle himself remained dissatisfied with his account of sensation. In chapters VII and VIII, he admits to there being something “seriously amiss” with his discussion of sensation; he attributes this to a kind of disorder in our use of this notion—to “our lack of ‘neat’ sensation words” (CM VII(6), 240).¹⁹ On the one hand, we know how to use the terms “sensation” and “feelings” in their ordinary and unsophisticated sense, in which they refer to a kind of perception, such as when one says one has a disagreeable sensation in their eye or a tingling feeling up their arm (CM, 240-41). Ordinary cases of perception rarely involve sensations or feelings in this sense. On the other hand, “we use the word ‘sensation’ in a sophisticated tone of voice to show that we are conversant with modern physiological, neurological and psychological hypotheses” about the nature of perception (CM, 240). In this second sense, “sensation” is a technical term having little to do with the original use of the word. Ryle adds that people could speak sensibly of perception long before these theories were devised, and hence that sensation cannot be a component of the concept of perception. The main issue, however, is that people “are not ready, indeed they are not even linguistically equipped, to tell what impressions they are or have been having” (CM, 242). Nevertheless, “the theory was able to trade on,” becoming a piece of common knowledge and thus perverting our vocabulary related to feeling; Ryle condemns this fact as a “philosophical misfortune” (CM, 243).²⁰ Ryle seems to be troubled by how the sophisticated concept of sensation has trickled into ordinary use, thus making “sensation” a systematically ambiguous term. His uneasiness with this family of notions grows in his later writings, as he further realizes that he has stumbled upon a phenomenon for which he could offer no account. Ryle (1956) offers a discussion of the notion of sense-impression very similar to that in CM.²¹ In this article, however, he offers a more detailed diagnosis of the genesis of the philosophical misfortune, as well as a slightly more precise articulation of what he found troubling about his own account:

However, after all this has been said, I confess to a residual embarrassment. There is something in common between having an after image and seeing a misprint. Both are visual affairs. How ought we to describe their affinity with one another without falling back onto some account very much like a part of the orthodox theories of sense-impressions? To this I am stumped for an answer. (Ryle 1956, 362)

Here Ryle admits to having no solution to offer for the problem that chiefly motivated sense-datum theories, namely that of accounting for the similarity between veridical and non-veridical perception. We shall see that this is precisely the issue that motivates Sellars’ alternative account of “immediate experiences” in the second episode of the myth.

We can now turn to Sellars’ attempt to explain how we can make sense of the notion of inner episodes, such as sense impressions, without falling prey to a Cartesian picture of the mind. By writing that “not all the unobservable episodes we suppose to go on in the world are the offspring of category mistakes” (EPM, sec. 52, 97), Sellars is suggesting that he wants to give an account of inner episodes without giving up Ryle’s more fundamental commitments.²² As noted above, this is not a fair criticism of Ryle, since he never issued a blanket denial of the meaningfulness of all terms that denote unobservable episodes. However, as I have just shown, Ryle did think that the notion of sense impression was the offspring of a category mistake. We shall see how Sellars’ alternative picture invites us to conceive of impressions and other “immediate experiences” as having been introduced as theoretical terms. His account will help us understand what allows him, unlike Ryle, to speak at liberty of mental or inner episodes or processes in general. The explanation of inner episodes in EPM takes the form of the “myth of Jones,” divided into two episodes. The first introduces one kind of inner episode, namely thought; the second proposes a similar account of immediate experiences, such as sense impressions. For Sellars, even though both thoughts and immediate experiences are inner episodes, it is important that immediate experiences are not kinds of thoughts (even though they presuppose thoughts) and thought is not a kind of mental picturing (EPM, sec. 56-58, 103, 105-6, 115).

Sellars’ myth begins with an imaginary community—the “Ryleans”—whose language, originally, only allows for non-theoretical vocabulary describing overt behavior. Jones, an innovative member of the community, introduces the technical notions of “thought” and “immediate experience” as part of two broader theories. The first theory (introduced in chapter XV of EPM) accounts for non-habitual, non-overt intelligent behavior by postulating thoughts as theoretical entities that inherit some of the features of overt speech, such as its intentionality and semantic features. The second theory (chapter XVI) provides a causal explanation of sense perception and to account for part of the similarity between the case in which a perceiver sees a red object and that in which the object merely looks red—or that in which there merely seems to be a red object. Jones does this by postulating theoretical inner states called impressions.²³ Sellars observes that after Jones has developed the theory and taught his compatriots how to interpret and apply it, the Ryleans begin to use the language of thoughts and sense impression to make reports and self-descriptions: “What began as a language with a purely theoretical use has gained a reporting role” (EPM, sec. 59, 107). What Sellars has in mind here draws on the history of the physical sciences—think of concepts such as that of force or of atom. His innovation lies in the application of this familiar narrative to mental concepts, such as that of immediate experiences. The analogy with the physical sciences shows that the transition of these concepts from a theoretical to a reporting status can only take place over time and if accompanied by a proper integration of the theory of which it is part with our commonsense conception of the world. The resulting inner episodes, Sellars claims, are not ghostly entities, but genuine enrichments of the original Rylean language: they “are ‘in’ language-using animals as molecular impacts are ‘in’ gases, not as ‘ghosts’ are in ‘machines.’ They are ‘nonempirical’ in the simple sense that they are theoretical—not definable in observational terms” (EPM, sec. 58, 104).

Part of what makes Jones’ notion of sense impressions meaningful is that it serves a function in the neo-Ryleans’ language and in their explanatory picture of the world. Sellars distinguishes two roles of pure theoretical terms: (a) their role in explaining the selected phenomena of which the theory is the theory and (b) their role as candidates for integration in the “total picture … of man in the world” (EPM, sec. 55, 101). The “purer” a theoretical term, the less scientists are clear about (b), i.e. about how the term is to be integrated with other theories (EPM, sec. 55, 102). The initial roles of Jones’ notion of sense impression, following this schema, are (a) to contribute to a causal story of perception that explains the similarity between genuine and non-genuine perceivings, and (b) to play a potential role in connecting behavioral accounts of perception to neurophysiological theories. Sellars acknowledges that “until recently, at least, neurophysiological speculations in behavior theory have not been particularly fruitful” (EPM, sec. 55, 101), suggesting—as we move from the myth to the real world—that our notion of sense impression is still far from having fully graduated from its theoretical status.²⁴ However, he urges us against downplaying the significance of (a): what Jones has introduced among his compatriots is not a mere notational device or inconsequential language game; the Ryleans “have acquired another language, … one which, though it rests on a framework of discourse about public objects in Space and Time, has an autonomous logical structure, and contains an explanation of, not just a code for, such facts as that there looks to me to be a red and triangular physical object over there” (EPM, sec. 62, 116). He also notices that these explanations apply to future as well as, retroactively, to past instances of perception.

For present purposes, the most significant aspect of Jones’ revolution is that the introduction of the theories of thought and sense impressions is not a mere addition to the common-sense framework on which it is based, because it does not leave this underlying framework unchanged. Rather, it is a “creative enrichment of the framework of empirical knowledge” (EPM, sec. 62, 116). This enrichment thoroughly transforms common-sense notions, including those it seeks to explain. This is clear if we regard a concept’s content as partly constituted by its role within the language games it inhabits, and thus by its relationship with other concepts in those games. In the context of the second episode of the Jonesian myth, this allows us to see that the new notion of perception is one that essentially involves a certain kind of inner episode, namely sense impressions, and that it cannot be fully identified with the original, pre-Jonesian “perception.” Once fully assimilated in ordinary discourse and the total scientific picture of the world, the concept of sense impression thoroughly transforms the Ryleans’ underlying notion of perception. This is one of the instances motivating Sellars’ warning that failure to understand scientific discourse might lead to a failure to understand “its role in ‘ordinary usage,’ and, as a result, to a failure to understand the full logic of even the most fundamental, the ‘simplest’ empirical terms” (EPM, sec. 40, 81). Using terminology from PSIM, Sellars is suggesting that the “scientific image” of the world (SI) has the power to enrich and thereby transform the “manifest image of man-in-the-world” (MI), and thus refine it both empirically and categorically (SR, 375).²⁵ The account of perception put forth by the myth also buttresses Sellars’ claim that unpacking what is implicit in observational knowledge cannot be done simply by calling observation a form of “knowing how” (EPM, sec. 36, 75); rather, this can be done only by coming to appreciate how this form of know-how presupposes the knowledge(-that) of general facts, such as those about the causal role of sense impressions in perception.²⁶

We do not need to read Sellars’ myth as showing the untenability of Ryle’s philosophy. Indeed, I believe that the relevant insight of Sellars’ story complements, rather than rejects, Ryle’s account of perception.²⁷ For now, I shall focus on how Sellars’ story can be used to address Ryle’s misgivings about the ambiguous status of the notion of ‘sensation.’ While it is doubtful whether the notion of sense-impression has wholly graduated from theoretical to reporting status, it is much more plausible to suppose that it was introduced as a theoretical term—precisely to explain the similarity that Ryle worried about—on its way towards integration in our ordinary conception of perception. To appreciate what is innovative in Sellars’ story, it suffices to notice that it is merely possible for him to be on the right track, admitting that this might be a case in which mental concepts that are introduced as purely theoretical—devoid of observational content and mostly disconnected from ordinary meaningful discourse—can gradually acquire referential powers by being integrated in our scientific and ordinary discourse. Whether Sellars was correct about the case of sense-impressions depends on the (present or future) degree of integration of this notion into the layman’s vocabulary as well as on whether it does (or will) connect behavioristic accounts of mind with psychophysical or neurophysiological ones. If it does not, we might conclude, as Ryle (prematurely) did, that the notion was nothing more than a puffed-up variation on a category mistake, whose origin and development are nevertheless interesting to reconstruct and diagnose. Either way, the comparison between the two accounts of sense-impression brings out an aspect of his concept of mind that is wholly foreign to Ryle’s. The author of CM does not mention either the general possibility of an enrichment of ordinary concepts via theoretical postulations, nor, more specifically, ways in which the Cartesian doctrine might have positively enriched or transformed our understanding of the mind. His refusal to integrate much of the vocabulary employed by more technical accounts of perception can be read as a flat-out denial of the possibility of enriching our concept of mind through scientific discourse. It must be admitted that Ryle does mention, at the close of CM, one way in which the Cartesian picture of the mind can be productive. In his discussion of behaviorism (chapter X(2)), he states that “Myths are not always detrimental to the progress of theories. Indeed, in their youth they are often of inestimable value” (CM, 329). For instance, he maintains that behaviorism, as a methodological programme based on either a ‘Hobbesian’ (or ‘mechanist’) or a ‘Cartesian’ (‘para-mechanist’) picture of the mind, “has been of revolutionary importance to the programme of psychology” (CM, 328). Ryle even suggests that the Cartesian picture of the mind has actually been more productive than the mechanistic-Hobbesian one (CM, 329-30). However, he seems to interpret this contribution to philosophy as fruitful in a primarily negative sense: “[Behaviorism] has been one of the main sources of the philosophical suspicion that the two-world story is a myth” (CM, 329-30). The analogy used to illustrate this comparison in the last paragraphs of the book makes it clear that Ryle does not see the Cartesian myth as adding any positive content to our understanding of the human mind. This furthers the impression that Sellars’ theory of inner episodes contributes something that is genuinely absent from Ryle’s work. We can hold this independently of whether we think that Sellars was correct about the specific case of immediate experiences.

Sellars’ criticism might be fruitfully applied to other aspects of the human mind which some are inclined to describe as involving some internal (and potentially episodic) dimension, but for which no “neat” vocabulary has been fully established and integrated in what Sellars calls the total scientific image of man-in-the-world. The myth shows how we might come to develop the appropriate vocabulary to describe these internal phenomena, thereby bestowing upon them an increased degree of autonomy from intelligent overt behavior. We can imagine this applying, for instance, to feelings, to which Ryle seems to deny any internal dimension in chapter IV of his CM (see 83, 104-107). It might even be the case of thinking, the subject of the first episode of Jones’ myth and of Ryle’s later articles. Even though Ryle does not voice the same residual embarrassments concerning his own treatment of the subject, he does lament how the theme of thinking had turned out to be “at once still intractable and a progressively ramifying maze” (CE, xxi). It might not then be surprising, after all, if his intriguing and sophisticated account of thinking—to which we will return in section 3—admitted of further elaboration along the lines of Sellars’. While unpacking the details of this suggestion is too large a task for us now, we can begin to sketch how one might do so by focusing on how, in the first episode of the myth, the notion of thought, similarly to that of ‘immediate experiences,’ is first introduced as a theoretical term and then progressively acquires a reporting role. This invites a picture of human pensive life as evolving over time. Our capacity for reflection might have initially been wholly parasitic on non-pensive behavior—especially on overt speech—and later have progressively developed into an activity of its own, capable of being employed within an explanation of other intelligent actions.²⁸ Sellars’ account bears the special merit of explaining how the activity of reflection can both (1) arise from intelligent non-pensive behavior and (2) gain a degree of autonomy necessary to inform and even transform non-pensive intelligent behavior, as in the case where theoretical innovations influence our skillful practices or even our descriptive vocabulary. Ryle, instead, seems to struggle in the attempt to make room for (2) even in his latest writings:

what I aim to do … is to show that it is the notion of engaged thinking, like that of the tennis-player or the conversationalist, that is the basic notion, while that of disengaged thinking or reflecting, like that of le Penseur, is supervenient. The notions of being pensive and having thoughts do not explain, but need to be explained via the notion of intelligently X-ing, where ‘X’ is not a verb of thinking. (CE, 485)

Ryle here suggests that quiet reflection cannot explain what he calls “engaged” thinking—which I am equating, for the purpose of this section, with Sellars’ notion of overt intelligent behavior. He thereby denies the possibility of covert thinking alone transforming—rather than merely informing one’s future engaged thinking.²⁹ Sellars’ myth is not just an explanation of how disengaged thinking can arise as an independent activity; it also helps us understand how (2) is made possible through this very activity. The emergence of thoughts as autonomous inner episodes allows us to cite them in our elucidations of other phenomena—including those from which they arose.³⁰

This discussion sheds light on one further aspect of Ryle’s thought, namely the relatively marginal role he assigns to the scientific enterprise and its characteristic activities. Our capacity to engage in technically regimented, scientific thinking is a notable absence in CM; the reader gets the impression that Ryle considered it a dispensable activity of the human mind. I believe that this is what Sellars has in mind when, in PSIM, he includes among the philosophical currents that have been overly drawn to the pole of the manifest image—at the price of overly neglecting the scientific image of the world—“the trends of contemporary British and American philosophy which emphasize the analysis of ‘common sense’ and ‘ordinary usage’” (SR, 376). To see what Sellars meant, let us turn to one of Ryle’s most illuminating illustrations of his conception of the relationship between ordinary and scientific thinking. The second and the eighth lectures from Dilemmas liken the difference between everyday and technical concepts to that between civilians and army members in a country.³¹ The conduct of a civilian, like that of a soldier, is regulated by countless spoken and unspoken codes, habits, and etiquettes; however, that of the latter is subjected to stricter and more explicit regimentation. This mirrors the difference between the uses of ordinary and technical concepts. The analogy also serves to illuminate a form of continuity between the two dimensions of discourse. Civilians and soldiers are all citizens of the same country; civilians can be conscripted into the army; soldiers can be discharged. Similarly, ordinary terms can be given technical meaning, and scientific terms can be adopted by the layman. However—and this is not an aspect of the analogy that Ryle dwells on—everyone is born a civilian: no infant is born wearing a uniform. This seems to deny the possibility, suggested by Sellars’ myth, of terms born theoretical and raised to civilian status. Moreover, the analogy fails to capture the potential of technical discourse to transform ordinary notions, since the role of soldiers is not exactly that of transforming the lives of civilians. While Ryle did not mean for this analogy to illuminate every aspect of the relation between ordinary and technical notions, the fact that we can use the same analogy to describe Sellars’ criticism of the “philosophies of the manifest image” is rather telling. Sellars not only alludes, as Ryle does, to the continuity between ordinary and technical discourse (EPM, sec. 41-44, 81-82, 84; sec. 51, 97). He also warns his reader against thinking of “the authority of theoretical discourse as entirely derivative, that of a calculational auxiliary, an effective heuristic device” (EPM, sec. 19, 44; sec. 44, 84-85). For Sellars, familiarity with scientific discourse and progress is a prerequisite for fully understanding certain modern common-sense notions.³² This is evident in cases where scientific discourse replaces ordinary notions and thus informs the manifest image (EPM, sec. 41, 82).³³ Scientific discourse, for Sellars, is thus far from being a “peninsular offshoot from the mainland of ordinary discourse” (EPM, sec. 44, 84-85): it is part of the mainland itself. The capacity for scientific thought is a constitutive element of human mindedness as such. In this way, Sellars’ work sheds light on a subject that is central to Ryle’s concern with human mindedness, yet nearly unmentioned in his writings. In the next section, I will consider whether Ryle’s work can play a parallel role in understanding Sellars’ thought.

3

The goal of this last section is to show how a careful reading of Ryle’s account of thinking can illuminate a blind spot in Sellars’ own account. We can make better sense of Ryle’s account of thinking, I argue, if we take the notion of intelligent capacity, rather than that of overt behavior, as explanatorily primitive. I thus read Ryle as availing himself of categories that fall beyond what Sellars thought was available to a philosophical account of mindedness; this is precisely what I take to be Ryle’s fundamental insight. Ryle’s account allows us to question Sellars’ notions of “overt behavior” and “thinking,” thereby suggesting a way to work ourselves out of some of Sellars’ own concerns regarding the conflict that motivates the first episode of the myth of Jones. This difference reveals a deeper divergence between Ryle’s and Sellars’ conceptions of philosophy. For Sellars, the manifest image (henceforth MI) and the scientific image (henceforth SI) often provide irreconcilable accounts of one and the same phenomenon—leaving the philosopher with no choice other than choosing which one is the correct story. Ryle thought these conflicts were far from being mandatory; his philosophy aims at reconciling different accounts. My aim here is to show how this reconciliation unfolds in the case of thoughts.

Consider first the kind of tension that the myth of Jones addresses. For Sellars, the MI and the SI can provide genuinely incompatible accounts of one and the same phenomenon. This is why the scientific picture can occasionally replace or supersede the descriptive ontology of everyday life (EPM, sec. 41, 82).³⁴ An example mentioned in EPM is the issue as to whether physical objects are colored. Sellars does not want to say that the optical account of colors as the result of reflected light rays impinging on human retinas rejects the common-sense idea that physical objects have colors—at least not within the conceptual framework of the manifest image. Yet, he wants to make room for a sense in which optics can replace the manifest image as the correct conceptual framework to be used if we want to establish what, ultimately, is the case:

speaking as a philosopher, I am quite prepared to say that the common-sense world of physical objects in Space and Time is unreal – that is, that there are no such things. Or, to put it less paradoxically, that in the dimension of describing and explaining the world, science is the measure of all things, of what is that it is, and of what is not that it is not. (EPM, sec. 43, 83)

Sellars thus sees optics not only as enriching and transforming our ordinary concepts of colors, but also as a framework for understanding colors that genuinely conflicts with the manifest framework. A philosopher may accept the manifest framework and maintain that physical objects are colored—presumably trying to accommodate the theory of optics within the MI—or she can reject the manifest framework in favor of the scientific one and conclude that objects are not really colored, tertium non datur. While Sellars expresses his inclination to embrace the second alternative, the passage from PSIM I mentioned in section 2 might suggest that he sees ordinary language philosophers—a category of which Ryle is often heralded as a main exponent—as going for the first option, upholding the MI and thus rejecting part of the SI. What matters is that, for Sellars, every time the MI and the SI reach different conclusions, we are presented with a choice between two horns of a dilemma.

It is precisely this sort of irreconcilable conflict that motivates the first episode of Sellars’ myth, the one devoted to explaining thoughts. On the one hand, the very idea of inner episodes is unintelligible within the MI—championed by Sellars’ Ryle—on the grounds that for something to be an episode it needs to be publicly witnessable and reportable. Like the mythical Ryleans, Sellars’ Ryle can only speak of the items that populate the manifest image, such as overt behaviors. On the other hand, private episodes, such as thoughts, are not publicly witnessable. Hence Sellars’ Ryle must conjure a ploy to explain how we ordinarily speak of thoughts within the manifest image; he claims that “thought episodes … are not really episodes at all, but rather hypothetical and mongrel hypothetical-categorical facts about these and still other behaviors” (EPM, sec. 46, 89). This allows Sellars’ Ryle to speak of thoughts while preserving the claim that there is no such thing as inner episodes. Sellars is unsatisfied by this solution: he wants to defend a version of what he calls the “classical view” of thoughts, which claims that thoughts are inner episodes irreducible to overt verbal behavior or verbal imagery (EPM, sec. 47, 89-90; sec. 50, 94). Even though speech expresses thoughts—and overt discourse might indeed be the natural fruition of thoughts, as well as a prerequisite for developing the capacity to have thoughts—the latter, for Sellars, can retain a form of metaphysical independence from overt behavior. The idea that thoughts are sometimes inner episodes originates in something “like a theory”; it cannot be the product of the mere MI because it implies the existence of unobservable phenomena. The crucial point here is that the SI makes a claim incompatible with that of the MI, namely that there are inner episodes, and that when we speak of thoughts we are speaking of inner episodes. Sellars builds even more into the classical view when, while exposing Jones’ theory, he speaks of thoughts as the true causes of intelligent nonhabitual behavior (EPM, sec. 56, 103). I take him to imply that the MI and the SI, in this case, not only disagree on whether the word “thought” refers to an inner episode, but also disagree as to the true cause of intelligent behavior. This, in Sellars’ view, is the reason why Ryle cannot explain cases of intelligent behavior that is detached from any sort of (factual or hypothetical) overt behavior. These conflicts play an important part in motivating the first half of the myth of Jones, as we gather from Sellars’ articulation of his motives:

The philosophical situation [my myth] is designed to clarify is one in which … [we] are very puzzled indeed about how we learn to speak of inner episodes. (EPM, sec. 48, 91)

[Jones is trying to] account for the fact that his fellow men behave intelligently not only when their conduct is threaded on a string of overt verbal episodes … but also when no detectable verbal output is present. (EPM, sec. 56, 102)

This sets up a dispute: either thoughts are always analyzable in terms of overt behavior, or there are some cases in which thoughts are mere inner episodes that cause intelligent behavior. Sellars chooses the second horn. The point of the myth of Jones is to introduce a non-mythical account of thoughts that allows us to speak of them as inner episodes with causal powers. In what follows, I intend to show how Ryle’s work provides us with the tools to question the legitimacy of the assumptions underlying both horns of Sellars’ dilemmas.

Before turning to Ryle’s alternative conceptions of thinking and overt behavior, I want to first dispel the temptation to read CM as providing an account of the mind within Sellars’ “manifest” framework. To this end, it is helpful to recall how one of the central goals of CM is to reject the “two-worlds” picture underlying the Cartesian dogma. Not conceiving of intelligent behavior as comprising “a bit of theory” and “a bit of practice” (CM, 29) also mandates that we reject the ontological distinction between res cogitans and res extensa. The alternative “one world” that Ryle defends is not identical to either of the two worlds of the Cartesian. In particular, it is not the “physical” world—neither that of the MI nor that of the SI. One will not get Ryle’s world into view as long as one thinks of it as one among many worlds: “Indeed it makes no sense to speak as if there were two or eleven worlds. … Even the solemn phrase ‘the physical world’ is as philosophically pointless as would be the phrase ‘the numismatic world’, ‘the haberdashery world’, or ‘the botanical world.” (CM VIII(1), 199). This is not to say that the expression “physical world” is wholly pointless—it is used, for instance, to think of the world insofar as it is the subject of the interest of physical scientists. Rather, what is being suggested here is that the philosophical significance of the expression “the world of physics” is similar in kind to that of expressions such as ‘the world of poultry’ (Ryle [1954] 2015, 62). Ryle is thus suggesting that equating the notion of “world” with that of “physical world” is not as philosophically innocuous as Sellars assumes. The world within which Ryle locates the concept of mind is one that both the MI and SI presuppose; the “manifest” world and the “scientific” world are considered by restricting one’s scope. To appreciate Ryle’s account of thinking, we need to examine how the observable/unobservable and overt/covert distinctions figure in it, as we shall do shortly.

Sellars, in EPM, does not appear particularly troubled by these distinctions. In setting up the myth, he seems to assume that the notions of overt behavior and inner episode are relatively unambiguous and in need of no substantial elucidation. He leaves room for an interpretation of his myth in which the Ryleans’ notion of overt behavior, which does not make any reference to thought, shares a common core with our (or the neo-Ryleans’) notion of overt behavior. Sellars’ notion of overt behavior is thus intelligible independently of that of thinking. This leaves the door open for the possibility that two instances of overt behavior, one intelligent and one not—such as intentionally raising one’s arm to ask a question and one’s arm accidentally jumping up—share a highest common factor. The intelligent character of the former does not affect its overt character. What differentiates the former from the latter is presumably the addition of an extra factor, the occurrence of a thought. Similarly, Sellars assumes that an account of our ordinary notion of thinking must explain, before anything else, the possibility of thoughts as inner episodes; these need to be sufficiently independent from overt behavior to be able to cause it. Thus, a thought may or may not be involved in an overt action without changing in character. Similarly, an overt and a non-overt action could be the result of the same thought, whose character qua thought is wholly unaffected by the action it causes. Thus, Sellars’ explanation of the neo-Ryleans’ intelligent behavior is built up out of two kinds of constituent parts. The first—overt behaviors—are episodes that share a highest common factor with inert physical movements. The second—thoughts—are intelligent episodes, thus they are construed as somewhat analogous to physical episodes. This picture is reminiscent of the “two-world” para-mechanical account of mind that Ryle was after, in two ways. First, in identifying a highest common factor between intelligent and non-intelligent processes, it calls for the postulation of an intelligence-bestowing element to differentiate between the two kinds of processes. Second, it assumes that an explanation of intelligent behavior must be modelled after one of non-intelligent behavior, for instance by employing the same logical categories (that of “process” or “episode”). Ryle calls such an account “para-mechanical” because, even though it does not claim that mental concepts are reducible to observable physical events, it nevertheless assimilates the logical grammar of mental concepts to that of physical concepts.

That Ryle’s conception of overt behavior is less straightforward than Sellars’ is evidenced by his notorious suggestion that a description of someone’s qualities of mind can refer to their overt acts and utterances themselves (CM II(1)). I propose we do not interpret this as implying that, for Ryle, qualities of mind are reducible to mere physical acts and utterances; rather, the idea underlying this comment is that the more basic notion is that of intelligent behavior—more basic than those of overt and covert intelligent behavior. This distinction, for Ryle, can only be understood once intelligence is already in the picture. To be more exact, Ryle’s more fundamental category is that of intelligent capacities, which CM famously characterizes as “multi-track” dispositions, meaning that when we ascribe an intelligent capacity to someone we are licensed to “predict, retrodict, explain and modify actions, reactions and states” of said person (CM, 123) in an “indefinitely heterogeneous” way (CM, 44). When I credit someone with possessing a skill, such as that of being a good chess player, I mean that she could recite the rules of the game without having to look them up, but also that she can appreciate someone else’s mastery of the game while witnessing them play; however, it would be impossible to draw up an exhaustive list of the inferences I am licensed to draw from the ascription, or an exhaustive list of the exercises of said capacity. The difference between overt and covert exercises of a capacity is drawn within this list. In the case of the chess player, castling at the right time is an example of the former, pondering which move to make next is an example of the latter. At the same time, there are several exercises of the capacity that cannot be described as overt nor as covert—think of playing a game of chess, which might involve variable amounts of overt and covert actions depending on how the game is played.³⁵ This perspective sheds light on the fundamental difference between raising one’s arm to ask a question and one’s arm inadvertently jumping up. The very way I have described the two actions suffices to illustrate how they differ. While the first is an exercise of several intelligent capacities, such as that of asking a question, the latter is a mere physical reflex, one we do not impute to a person. Ryle would have deemed it misleading to even refer to it as overt behavior; he would have rather called it a “merely physical” process (CM, 32). The fact that the two events are physically indistinguishable does not mean that the intelligent action has to be construed as the result of adding some intelligence-bestowing ingredient to a mere physical process.

We can contrast Ryle’s picture with Sellars’ myth, in which the Ryleans do not seem to have any notion of intelligent capacity, even though they do have words for overt behaviors. If this interpretation is correct, we can read Sellars as assuming that the Ryleans’ meaning of the expression “overt behavior” shares a highest common factor with our notion of overt behavior. Ryle’s writings show us that this assumption is far from being mandatory. In his account, nothing needs to be added to overt behavior to make it intelligent. If, instead, Sellars did think the Ryleans had a conception of intelligent capacity, then Ryle has a story to tell us about how they might also develop a notion of covert intelligent behavior—which includes what Sellars calls nonhabitual, non-overt intelligent behavior, and Ryle calls reflecting or disengaged thinking. The details of this story are mostly found in his later writings. They allow us to see how Ryle questioned not only Sellars’ philosophically inflected notion of “overt behavior,” but also those of thinking and reflecting in Sellars’ “classical” view.

Much of Ryle’s writings from the 1950s and ‘60s examines what we do when we reflect, ponder, contemplate, or engage in other forms of covert thinking. This, for Ryle, is different from asking what “thinking” means in the general sense, as he insists that the strokes of a tennis player are exercises of intelligent capacities as much as the reflections of Rodin’s le Penseur. Ryle calls the former “engaged” and the latter “disengaged” thinking. Notwithstanding this distinction, Ryle believes that a proper elucidation of disengaged thinking must involve a discussion of the nature of thinking in general. The issue with most Cartesian accounts of thinking, he believes, is that they misguidedly attempt to identify some isolable and describable or elements of thinking—as if thinking were a chemical compound (CE, 309). Against this family of views, Ryle suggests that thinking need not be construed as an independent activity, but rather one that is “constitutionally inceptive,” meaning that it is parasitic on other intelligent activities that the thinker might be engaged in or preparing for.³⁶ This applies to both engaged and disengaged thinking. We can only describe someone as heeding the road if we already know that they are driving, biking, or walking along the road; we can only describe someone as doing mental math if they are trying to perform a computation; similarly, describing them as preparing their dinner speech involves an implicit reference to delivering a speech. These examples illustrate how a full (or, as he calls it, “thick”) description of engaged or disengaged thinking always involves a reference to some form of underlying intelligent capacity.³⁷ Usually, disengaged thinking is a form of preparation for an activity; Ryle likens it to training or experimenting (CE, 491). Running through a speech in one’s head prepares one for a dinner speech similarly to how practicing on one’s swing prepares one for a golf competition. Moreover, both activities involve a certain degree of detachment from the actual circumstances surrounding the full-blown exercises of the capacity they are a training for: in the case of golf, practicing one’s swing might not involve using a golf ball, and in the case of the speech, rehearsing it in one’s head involves detaching oneself from circumstances that constitute a full-blown speech, such as uttering words or having an audience.

Ryle’s mature writings on thinking and reflecting discuss many further aspects I cannot examine here. What I want to highlight is that Ryle presents us with a conception of thinking that elaborates on the account from CM and that does not assume, unlike Sellars’ updated version of the “classical view,” that the thoughts of le Penseur must be explained without making any reference to intelligent capacities. When we identify an action as thinking-laden, says Ryle, we are ipso facto connecting it to some further activity, an actualization of an intelligent capacity. In contrast, Sellars’ attempt to provide an account of the intelligence of non-overt non-habitual behavior, such as silent reflection, relies on the assumption that this intelligence must be explainable by some element concomitant to the behavior and intelligible independently of its broader context. While Sellars assumes that an explanation of intelligence can only take for granted the conceptual categories used to describe the merely physical world, Ryle suggests that intelligent activities and capacities fall under their own categories. In Ryle’s words, Sellars is trying to explain intelligent behavior by limiting himself to a “thin” description of the behavior, while understanding what makes an action intelligent usually requires focusing on its “thicker” descriptions – and there is no reason to assume that a description being thinner automatically makes it more basic or fundamental. Sellars’ defense of thoughts as inner episodes is the kind of attempt to reduce thinking to some irreducible element that Ryle condemns. Such attempts run the risk of missing the distinctive complexity of human intelligence. Moreover, they are far from being the only possible ones, as Ryle’s mature writings show that we do not need to identify the secret ingredient of thought to understand why silent reflection is a form of thinking.

We are now in the position to appreciate how Ryle can reject both horns of the dilemma that motivated the first episode of the myth of Jones. He agrees with Sellars that thoughts are not reducible to mere overt behavior, but not because they contain something over and above mere overt behavior; rather, because the very notion of human overt behavior presupposes that of intelligent capacities, and thus that of non-overt thinking. The fact that we speak of overt intelligent behavior implies that we can engage in covert intelligent behavior, such as silent contemplation. This does not imply, however, that the only way to make sense of covert intelligent behavior is to posit an unobservable inner process. In fact, we do not determine whether an action is an exercise of an intelligent capacity by abstracting it from its broader context. Rather, we do so by examining its implicit connections to broader intelligent practices, such as solving a mathematical problem or composing a poem. This is true for engaged thinking, which involves the performance of some overt activity, as well as for disengaged thinking, such as silent reflection, whose thicker description always involves a reference an activity other than motionless sitting. We can thus make sense of how disengaged thinking is a form of thinking without having to speak of unobservable inner processes. Importantly, making sense of Ryle’s alternative involves questioning the assumptions behind Sellars’ conceptions both of overt behavior and of thinking, rather than accepting them and choosing between one of the two horns of the dilemma.³⁸ I conjecture that many of the alleged conflicts between the SI and MI that Sellars was interested in admit of a similar Rylean dissolution. I hope to have at least successfully worked out one of these dilemmas.

Ryle’s crucial insight is that an explanation of covert intelligent behavior need not be an explanation of a certain sort of episode with causal powers, even though we might want to speak in those terms in the context of, say, a neuroscientific study of human beings. Instead, we can overcome our puzzlement about covert intelligent operations by locating them in the context of some intelligent practice, thus thinking of them as actualizations of some intelligent capacity, once we recognize that these dwell within their own logical space. This move is exemplary of Ryle’s own philosophical brand. Ryle does not see philosophy as a theory-building enterprise subject to continual revision, but as a more interpretive activity. He does not attempt to settle the disputes between science and common sense. This is because he sees the dilemmas arising from the scientific analyses of phenomena or the technical meanings of concepts as calls for further philosophical elucidation, and not as demands that we unconditionally reject our previous descriptions of the world. New theoretical concepts might be able to enrich and transform previous categories, but they cannot fully replace them. Speaking as a philosopher, Ryle is not as prepared as Sellars is to say that there is no such thing as the common-sense world of physical objects, or that, when it comes to describing the world, science is the measure of all things. This is not because Ryle thinks that common sense is the measure of all things. Rather, it is because he thinks that the task of philosophy is not the judiciary one of measuring theories against each other, but that of understanding how different accounts fit together.