A Thesis submitted in conformity with the requirements for the Degree of Master of Arts in the University of Toronto

© Robert E. Czerny 1971

Edited in hypertext by Andrew Chrucky. Reprinted with the permission of Robert Czerny.
Editor's Note: Due to the limitation of current hypertext, the following conventions have been used. Subscripts are placed either as simply concatenated suffixes with or without parentheses (depending on the context): thus, e.g., H(2)O is the chemical formula for water. Sellars' dot quotes are expressed by bold periods.

Logical connectors and quantifiers are all expressed by bold characters as follows:

--> = material implication
<--> = material equivalence
& = conjunction
~ = negation
v = disjunction
(x) = universal quantifier
(Ex) = particular ('existential') quantifier


A. Aims
B. Sense-Datum Theory
C. The 'Jones-Myth': Peripheral and Preliminary
D. The 'Jones-Myth': Meaning and Thought
E. Problems in Sellars' Doctrine
A. The Meaning 'Relation'
B. Ontological and Metaphysical Considerations
A. Introduction
B. "Language as Thought and as Communication"
C. Rules and the Model of Learning
D. Behaviourism, Meaning, and Language-Learning

An active cerebrum had Peter,
Which Skinner compared to a meter;
With sarcasm and passion,
All in his own fashion,
Peter said, "It's not true, but it's neater."


One of the gratifications of writing this M.A. thesis has been the friendship of those who helped me, both formally and informally. I would like to thank Christopher Olsen, Ron de Sousa, Theresa Haq, and Fred Elliston, my parents, for their sacrifice over a far greater length of time, in making things easier for me without making them less worthwhile; Michael; and, especially, Kathy.


"Being and Being Known", by Wilfrid Sellars
"Comment on Wilfrid Sellars' Paper", by Mikel Dufrenne (a reply to LTC)
"Empiricism and the Philosophy of Mind", by Wilfrid Sellars
"Intentionality and the Mental", by Roderick Chisholm and Wilfrid Sellars
"Language as Thought and as Communication", by Wilfrid Sellars
Science and Metaphysics. By Wilfrid Sellars


The overwhelming preoccupation of Anglo-American philosophy of mind and knowledge since Wittgenstein, has been to explicate the private, subjective world of the mental in terms of public, observable processes. This can involve a behaviouristic rejection of Cartesian dualism, which allows the world of the mental to be explicated autonomously and directly in its own terms.

Wilfrid Sellars, the American philosopher of science and language, can be seen as a moderating force in this rejection of dualistic philosophy of mind. He is concerned to 'rethink' behaviouristic and ordinary language philosophy of mind and knowledge in terms of the previous history of philosophy.

This programme makes Sellars' work extremely difficult to read, being full of brief historical allusions and references. At the same time, though, it makes a study of Sellars very rewarding. First, he points out and makes intelligible errors in the tradition of philosophy of mind. Second, he puts behaviourism and ordinary language philosophy on firm foundations by clarifying their methodologies, by tempering their denunciations of previous psychology and philosophy, and by combining their contributions with what is still worthwhile in the history of philosophy of mind. Third, he clarifies our everyday language for mental life by tracing the origins and development of some of our ordinary mentalistic concepts, and by explaining the logical relation between this ordinary mentalistic language and the specialized languages of philosophical and scientific explanation.

Sellars' philosophical method is the analysis of concepts primarily as they appear in ordinary language, and secondarily, as they appear in philosophy and science. As such, he is attempting a 'presuppositionless' philosophy: his analysis begins with that which, for him, is given in the most general and untheoretical manner, the public, observable, language-impregnated social world. He does not begin with metaphysical categories; these would appear only as a conclusion, after concepts have been analysed. In other words Sellars is dealing with problems of what is conceivable, not with the problem of what is. Applied to philosophy of mind, this means that he is dealing with concepts of the mental and their logical possibility.

In the works which I am considering, Sellars' questions regarding concepts of the mental fall under two broad descriptions. First, what is the relationship between the concepts and skills which I use to report my personal experience, and the concepts and skills which another person would use in describing my experience? To put it another way, this is the problem of the relationship between first person and third-person reports, between private personal experience and publically observable behaviour. Second, what is the relationship between those categories of intentionality (aboutness or reference to objects) which apply to the mental, and the semantical or metalinguistic categories which apply to language? In other words, what does it imply, that both our thoughts and our words are about something?

Before sketching Sellars' replies to these questions and my own criticisms of those replies, I wish to note that, since Wittgenstein, questions of thinking, knowledge, and mind have had to be dealt with in conjunction with questions of criteria, the logical possibility of teaching and learning, the social context of knowledge, and so on. Once all these concepts and concerns are in one 'family' together, it becomes very difficult to stipulate a necessary difference between the philosophy of mind and knowledge and the philosophy of education. I would consider the work of Wilfrid Sellars to be a prime example of the interpenetration of the concerns of 'philosophers of education' with those of other philosophers.

Sellars' essay "Empiricism and the Philosophy of Mind," the main topic of Chapter II of this thesis, gives us the framework and much of the substance of his replies to the two main questions facing him. In a 'myth' about the progression from behaviouristic language to a semantical and then mentalistic language, he tries to show that the language of introspection (the first-person reports which I now make about my present mental experience) depends logically on intersubjective observational language. Further, he argues that if we treat the relationship of ordinary mentalistic language to overt behaviour like that of scientific theoretical explanation to the data of observation, we can treat the aboutness of thoughts as a reflection of the meaningfulness of speech.

The concept of meaning is so central (via the myth) to both concerns, that the next chapter of the thesis is devoted to its elaboration. It is found that, ultimately, 'meaning' is to be understood in social terms and in terms of the logic of teaching and learning language.

Sellars' views on language-learning, in relation to his views on meaning and behaviourism, are considered in Chapter IV. The following three criticisms are presented, showing Sellars' views on these three topics to be inconsistent with one another. First, the awareness of meaningfulness is supposedly a language-dependent phenomenon; but, for epistemological reasons, we have to credit the pre-verbal child with a pre-verbal awareness of meaningfulness. Second, the understanding of action and rules, necessary for teaching, is akin to the semantical understanding of the meaningfulness of language. Thus, Sellars would have to claim that it is logically more sophisticated than observational language which does not contain semantical discourse; and yet, via his 'myth', Sellars claims that there could be an ongoing observation-language community -- so somewhere language is taught, without the action-understanding resources to do so. Third, it is impossible to pick out something in the world and call it a 'linguistic entity' without having some notion of 'meaningfulness' operating implicitly in this selection and classification. However, the methodological behaviourism with which Sellars purports to describe the early stages of language-learning and language use -- a description of 'linguistic entities' in relation to other things -- is not supposed to involve any concept of 'meaningfulness'.

Sellars' philosophy of mind, and these aspects of it in particular, are interesting to the philosopher concerned with education for at least two reasons.

First of all, a number of schools currently compete in psychologically explaining intellectual growth. Behaviourism is one of them; its strongest present rival is Jean Piaget's genetic-developmental epistemology. Their methodologies are radically different, embodying opposed philosophical assumptions about 'mind'. Obviously, the philosopher should distinguish his own work from that of the psychologist; but their interests overlap here, in methodology of psychology of intellectual growth. Sellars deals extensively with one of these methodologies -- that of behaviourism. This interest can be shared by any philosopher concerned with the proper description and explanation of intellectual growth.

Secondly, of even more direct interest to the 'philosopher of education' is the fact that Sellars deals directly with the learning situation itself. He avoids the oversimplification and distortion which comes from 'psychologizing' the learning situation. To explain it, Sellars uses philosophy of mind, knowledge, action, and language, in addition to methodological behaviourism. Any philosopher sensitive to the many dimensions of the learning situation can appreciate Sellars' careful contribution in this regard.

[Return to Table of Contents]


A. Aims

Before going into Sellars' essays, we can consult his own overview of problems in the philosophy of mind and philosophical psychology (Introduction to IM). Sellars defines the main issue as, how we are to construe the relationship between constructed scientific vocabulary of behaviouristic psychology, and common, everyday mentalistic vocabulary.

Such common mentalistic vocabulary can be found in the common man's report, "Jones believes that x". This report has a certain "aboutness or reference". Now, the psychologist says "S is in behavioural state y". If the latter statement does contain "aboutness or reference", then it is "because aboutness or reference is constructable out of the aseptic primitives to which he (the behaviourist) has restricted himself" (IM, 508). On the other hand, if the behaviourist's statement lacks such aboutness, he is 'leaving something out'. The behaviourist's counter-claim to this latter charge would be to deny the special intentional feature of common mentalistic language, saying that this is somehow fundamentally a mistake; 'mental language' would be just an "open ... informal" construction out of behavioural primitives. In other words, there is nothing 'really there' to leave out. And the dualistic counter-counter-claim is that behaviouristic psychology is doomed to deal just with the body (EM, 508).

Sellars' own position is as follows. He parts ways with the traditional behaviourist by refusing to deny the 'intentionality' of the common man's mentalistic statements; but on the other hand, he refuses to endorse the substantive explanatory claims which are classically used to shore up this intentionlity (e.g., immaterial existence of absolute natures, according to St. Thomas, cf. BBK). Rather, Sellars offers a revised account of the development of common mentalistic language. So too, he offers a personal interpretation of behaviouristic science and its 'primitives', and the meaning and development of the vocabulary of behaviourism. He wishes to allow statements within behaviouristic psychology to both be in a 'continuum' with common language as conceived of by behaviourists, and bear an 'aboutness' like that of common mentalistic language as conceived of by 'mentalists' (IM, 508 f.).

The setting of this exploration and debate is wider than my references to behaviourism might imply. As we can see from "Being and Being Known", more than from "Empiricism and the Philosophy of Mind" which was written four years earlier, Sellars' aim is to take current physiological sciences, especially neurophysiology, seriously, without thereby discounting any positive insights embodied in the history of philosophy of mind. To the extent, therefore, that the classical tradition in philosophy of mind stands against the possibility that neurophysiology has something to contribute to the study of 'mind', 'idea', 'sensation' and so forth, Sellars wants to both show how these points of incompatibility are errors within the classical tradition, and also make these errors comprehensible.

B. Critique of Sense-Datum Theory

Just as he points out errors in the classical tradition (the Thomists, Descartes, Russell) in the first half of the essay "Being and Being Known", so Sellars attacks sense-datum theory early in "Empiricism and the Philosophy of Mind". His attack on sense-datum theories is really a first step in a critique of the "framework of giveness" (EPM, 128) and its role in the general epistemological problem of the foundations of empirical knowledge (EPM, 164 ff.). Sellars wants to supplant epistemology which is based on the erroneous notion of 'givenness'. His alternative is a realism whose fundamental 'data' are all publically observable -- physical and linguistic items. The more particular consequence, in which I am interested, is that theories of language, meaning, and language-learning which depend on this same illusory 'givenness', are also supplanted by Sellars. The critique of sense-datum theories is a first step in this supplanting, because, by pointing out basic problems and paradoxes in these theories which depend on the 'givenness' of certain non-observables, Sellars makes plausible the rejection of the framework of givenness and the proposal of an alternative theory.

Since this thesis is more concerned with behaviourism and theories of language and meaning than with knowledge, the following brief summary is intended only as an orientation into the later discussion of meaning. Nevertheless, a few of Sellars' arguments will be presented in detail, as their pattern and content will be crucial for understanding later points.

The attempt by sense-datum theories to provide empirical knowledge with a foundation involves it in irreconcilable paradoxes. As something sensed, a sense content is a particular; as known, it is a fact. (Sellars uses the term "content" as neutral regarding the controversy whether or not that which is sensed 'exists' only when being sensed or 'given'.) If the sense content is a particular, then the sensing of it is non-epistemic and does not necessarily entail the existence of knowledge. If it is a fact, then the sensing of it is ipso facto a knowing (EPM, 128-9). But two factors count against the second interpretation. It seems to conflict with the basic position of 'givenness' in the sense-datum framework. Thereby, sensing is either "a unique and unanalysable act", thus containing no separable element which would necessarily entail knowledge; or it is analysable -- but the sense-datum tradition, when it analysed sensing, "did so in non-epistemic terms" (EPM, 130). The second factor is that, as fundamental in the sense of 'presuppositionless', simply sensing sense contents cannot presuppose acquired conceptual abilities. But for the empiricist tradition, 'unacquired conceptual ability' is anathema: 'knowing' is a species of 'classificatory consciousness', and thus involves learning and experience, etc. (EPM, 131). Thus the two demands -- that sensing sense contents be a 'knowing' and that it be 'presuppositionless' -- are contradictory.

In summary, the classical sense-datum theories are faced with reconciling the following "inconsistent triad" of notions. First, sensing sense contents entails having non-inferential knowledge of contents sensed; this is basic to the whole foundations-of-knowledge program, which is in search of veridical building blocks. Second, "the ability to sense sense contents is unacquired"; otherwise sense-datum talk is cut off from natural, spontaneous sensation talk, e.g., pain-and-tickle talk. Third, "the ability to know facts of the form x is φ is acquired", as empiricist theory of knowledge demands (EPM, 132).

The paradoxes of sense-datum theories in which data (or particulars) are confused with facts, are echoed in the analysis of 'looks'. Sellars agrees with the common-sense realism which wants to analyse the situation of the epistemic subject who looks and sees, in terms of the way things actually are. However, once we agree to analyse 'looks red' in terms of 'is red', we have to come up with a non-circular explanation of the seemingly necessary truth of 'x is red. <--> .x would look red to standard observers in standard conditions" (EPM, 142). The quick 'out' of denying that 'looks' is a relation is postponed, while Sellars distinguishes fact-stating and reporting uses of sensation sentences (EPM, 143). If one was inclined to say 'x is green' but then in different circumstances decided to say 'x looks green' ('blue' etc.), this is not a new kind of report based on a different 'minimal' objective (observer-independent) fact -- a 'looks-property' as opposed to 'is-property', a different sort of sense datum. Rather, "the experience of having something look green ; to one" is "very much like . . . seeing something to be green", taking both of these as experiences; the difference is that the latter is not just an experience, it is also a claim or assertion and especially an endorsement of that claim (EPM, 144).

Further, 'x looks r to me' and 'I see that x is r' are epistemic, not 'natural', facts, as opposed to 'x is r' simpliciter (EPM, 145-6). Thus, the move to analyse 'looks' in terms of 'is' is stymied; 'x is red' is necessarily connected to 'looks' not because of definition between them, but because the "standard conditions" (of "x looks red to standard observers in standard conditions") simply "means conditions in which things look what they are" (EPM, 147). In other words, a statement like "x is red. <--> .x would look red to standard observers in standard conditions" does not give a definition of the natural fact 'x is red' in terms of subjective and epistemic facts. Sellars rejects this sort of idealism. Nor does it tell us something which the natural fact necessarily implies. What it does say is that, given our rules and conventions for observation and expression, this natural fact entails certain epistemic facts and the appropriateness of certain linguistic expressions. The pattern of this argument is seen again in the distinction between definition, implication and entailment in semantical statements (e.g., IM, 536).

An interesting consequence of this position, says Sellars, is that the ability to recognize what colour a thing looks to be rests on the concept of being a colour, which in turn involves knowledge of colour-viewing circumstances. Sellars denies that this involves a paradox, of needing concept c before you can have concept c. Colour concepts do not come singly, but as part of a whole battery of concepts. While there is gradual acquisition of piecemeal habits, it is also true that one needs a basic few concepts concerning observables in space and time in order to have any one (EPM, 147-8).

Sellars shares the confidence of sense-datum theorists that 'x looks r to S' somehow necessitates x (or something else) being r and S being aware; but Sellars says that this can be accounted for without adopting the crucial core of sense-datum theories (EPM, 149). When asked if 'x looks r to S' involves some 'inner something' -- awareness, sensation, impression, immediate experience, or whatever -- there is a common-sense answer which rejects both these "dubious entities" and second-language reformulations. It favours, as an explanation, empirical generalizations in terms of the colour a thing has and the circumstances in which it is seen (EPM, 150).

The usual way to go past these generalizations is to speak of a descriptive content of 'seeing' and 'looks' experiences (EPM, 151-2). But specification of this content by the tradition in philosophy of mind always has taken sensation-talk away from ordinary sensation experience and talk and into the framework of givenness. Some (sense-datum theorists) would speak, for example, of 'two-dimensional particulars' as the basic given; as such, the commonsense physical-observational grammar of sensation talk cannot also be fundamental (EPM, 152-3). Others (the older mentalists) would liken awareness of sorts of items to thought and consider it an unproblematic given (EPM, 155-7).

Sellars thinks we can turn this empiricist tradition on its head by making the initial elements of experience not 'impressions of red' but red particulars, in which case the consciousness of sorts or repeatables would rest "on an association of words (e.g., 'red') with classes of resembling particulars". Of course, an associationism based on primordial awareness either that x is red or of given facts like x resembles y would just resurrect 'the myth of the given'. Our alternative to these mediations is psychological nominalism -- all awareness of abstract entities and even of particulars "is a linguistic affair". The "process of acquiring the use of a language" presupposes no primordial awareness of the types just mentioned (EPM, 160). This psychological nominalism has to be refined, all we have so far is a rejection of pure-awareness episodes in favour of "a mongrel categorical-hypothetical . . . a verbal episode as being the manifestation of associative connections of the word-object and word-word types". Also, now that 'sensation' and 'image' lack epistemic aboutness, word-world associations can be direct, not via private coloured (and other) particulars; there can be causal mediation by sensation, but without committment "to the mistaken idea that it is 'really' sensations of red, rather than red physical objects, which are the primary denotation of the word 'red' " (EPM, 161).

What happens, then, to our commonsense mentalistic vocabulary, like 'impression', 'idea', and so on? They are "linguistic affairs", says Sellars; in effect, they are part of a linguistic explanatory construction, like a scientific theory, with which we commonly operate; they are 'theoretical entities' (EPM, 160, 150).

Let us reword this tentative conclusion in terms of the essay "Being and Being Known". Sellars is concerned with the 'being' and the 'being known' of mental life. He claims that 'the mental' does not necessarily have to be what we conceive it to be. This common equation of the orders of being and knowing (an implicit premise in most philosophies of mind), involves the debatable theses of the self-presentation of mental states and the adequacy of introspection. But Sellars proposes that we explicitly deal with 'what could the mental be' and 'what could we conceive it to be' prior to making substantive claim concerning 'what is the mental'. If we want to contest Sellars' position, we have to attack this conversion of the history of philosophy of mind -- from a history of answers to 'what is the mental' to a history of 'what has the mental been conceived to be' -- on internal grounds, e.g., the upcoming accounts of meaning, language-learning, introspection. It does not help to go beyond this conversion to Sellars' tentative substantive proposal, that neurophysiology and cybernetics hold some sort of key to thought and sensation 'in themselves', for Sellars is not committed to these sciences. So long as we allow him his thesis about the theoretical status of our concepts regarding the mental, he is justified in telling us to quit philosophy of mind and take up philosophy of science, in the hope of aiding the scientific research (if not in neurophysiology and cybernetics, then elsewhere) which is the only thing which holds hope of going beyond our mere conceptions concerning the mental.

This ends the first half of the essay "Empiricism and the Philosophy of Mind" and the specific critique of sense-datum theories. It is important to ask how it and the second half (dominated by the"Jones-Myth") are related. Sellars makes his contribution to philosophy of mind in terms of the history of philosophy of mind. Therefore his myth (later in this essay) really has a dual purpose. On the one hand, concepts like 'introspection', and the language of thought and sensation in general, are given a position within a general theory of the emergence and meaning of thought and language. On the other hand, this general account serves as an explanation -- a psycho-analysis and pardoning, so to speak -- for the insights and errors, at a more sophisticated level, of philosophers within the classical tradition. So, the criteria which Sellars' 'myth' will have to satisfy are correspondingly two-fold: correct analysis of individual mentalistic concepts, and a plausible account of their emergence and social context.

C. The 'Jones-Myth': Peripheral and Preliminary

In the first half of "Empiricism and the Philosophy of Mind", Sellars proposed the destruction of one form of 'the Myth of the Given', namely, sense-datum theory. He showed that sense-datum theory has to assume as the fundamental level of cognition, an 'immediate awareness' which, paradoxically, should be both natural datum and epistemic episode or fact. In the second half of the same essay, Sellars offers an alternative framework in which the nature of 'immediate experience', and the logic of common observational and mentalistic language, are accounted for without making the assumption of givenness.

Our interest in language-learning and behaviourism gives us a twofold interest in this alternative construction. A major part of it is Sellars' own myth, the 'Jones-myth' -- a sort of logical chronology of language-learning which starts with a behaviouristic society and which attempts to show that the existence of observation-talk, meaning-talk and even thought talk does not necessitate the assumptions of 'givennes '. Second, the notions of meaning and thought which emerge here are necessary to under standing Sellars' later discussion of thought, language, and learning in "Language as Thought and as Communication".

In this sub-section I shall present the setting of the Jones-myth. More remotely, this involves certain conceptions of science and behaviourism (Sections 9, 13, and 14 of EPM). More proximately, Sellars presents the epistemological problem which his myth is meant to solve (EPM, Sections 8 and 10).

The Jones-myth begins with a society in which a primitive, ordinary language is spoken; it ends with a scientific theory explaining some 'ordinary' phenomena, and even replacing the ordinary language to become itself the usual idiom. For this reason, it is important to examine section "IX. Science and Ordinary Usage", which enunciates not only Sellars' general position on the relation of science to philosophy, but also the background to his main thesis about the status of the referents of mentalistic i discourse.

Scientific discourse is the flowering of a pre-existing, 'pre-scientific' dimension of common discourse, which dimension is 'writ large' in science. Failure to understand it in science leads to misunderstanding its role in 'ordinary usage' and thus "failure to understand the full logic of even the most fundamental, the 'simplest' empirical terms". Philosophy tends not to use 'results' from science. However, one needs to know the trends of scientific thought in order to appraise "the framework categories of the common sense picture of the world", for, as a continuation out of ordinary discourse, the scientific picture supersedes and replaces that of common sense (EPM, 172).

Over against this view of science is the widespread notion that "the categories of the common sense picture of the world" cannot be challenged; that common-sense statements are always ostensively linked to real entities and events; that theoretical statements are merely auxiliary and heuristic; and that the referents of theoretical statements exist in theory only (EPM, 173-4). Sellars rejects all this: the categories of science may well be superior to those of common sense (EPM, 172-3), and theoretical statements may point to real entities and events as well as common-sense statements can. In fact, he eventually claims that our common sense (non-theoretical) mentalistic vocabulary, of private, inner episodes, really cannot refer to anything at all. But if we take this same vocabulary as theoretical, as part of a scientific theory in continuity with an observational-behavioural non-theoretical vocabulary, then it can refer to real entities and processes.{1}

Because this thesis -- that scientific explanation grows out of and is continuous with ordinary usage -- is so central to the Jones-myth, it is important to note how shaky a premise it is. Sellars certainly argues from continuity, but where does he argue for it? It seems that we are supposed to suspend judgment until all the results are in:

basic concepts and distinctions . . . arc to be tested or 'proved' by the illumination they provide, and by the coherence of the story they make possible. (Science and Metaphysics, 1).

Sellars later repeats that theoretical explanation is continuous with the induction and deduction employed by common sense. He also tells us a bit about Jones as a theoretician. The languages of theory and observation are different. Classically, theory construction involves (i) postulating entities to behave in accordance with the theory, and (ii) correlating the theory with non-theoretical or observational entities. But this simple picture needs a corrective. One really starts not with a theory or formulae but with a model and its qualifying, delimiting commentaries on the relation (analogy) of the observed and the postulated (EPM, 181-2). Sellars will later speak of the errors in our common sense framework and in the history of philosophy of mind as failures to sufficiently restrict and specify the model proposed by Jones.

Sellars makes a point about the scicnce of psychology in general, and about his myth in particular, by claiming that scientific behaviourism need not be analytical. That is, it need not assert that "common sense psychological concepts are analyzable into concepts pertaining to overt behaviour". Behaviouristic psychology shouldn't have to follow such a closely delimited programme -- after all, physics and chemistry do not. Rather, "the behaviouristic requirement that all concepts be introduced in terms of a basic vocabulary pertaining to overt behaviour is compatible with the idea that some behaviouristic concepts are to be introduced as theoretical concepts" (EPM, 183-5). Jones will introduce theoretical concepts of thought and impression in terms of a basic behaviouristic vocabulary, without jeopardizing his own status as a methodological behaviourist. Overt linguistic behaviour will be his model. (When I speak in future of 'behaviourese', I mean a language with "a basic vocabulary pertaining to overt behaviour" and real-world obiects and events but nothing more. 'Rylese' means the same thing; I employ it to refer more directly to the Jones-myth itself, which starts with "our Rylean ancestors".)

The proximate setting of the Jones-myth is the epistemological problem of foundations of empirical knowledge and of private episodes. It has traditionally been felt that there must be a structure of ultimate, authoritative, and non-inferential episodes, reports of observational knowledge, which constitutes a foundation for empirical knowledge, an "ultimate court of appeal for all factual claims -- particular and general -- about the world" (EPM, 164-5). These episodes would bc instances of immediate awareness, unmediated by any concepts, verbal or quasi-verbal performances. That is (and this is the heart of the myth of the given), at this grounding or foundational level, knowledge is simply given.

Rather than review all of Sellars' reasons for rejecting this sort of givenness, let us merely remark with htm that the 'impressions' or immediate experiences which are supposedly given, are incapable of intrinsic description; i.e., predicates which we apply to them are primarily the descriptive predicates of real things. Thus, 'impression' talk becomes a mere secondary code for how we look at and see real things, rather than a 'foundation' (EPM, 175). Sellars turns this right around: "to have the ability to notice a sort of thing is already to have the concept of that sort of thing, and cannot account for it" (EPM, 176). For a particular judgment to express knowledge, a subject must already have concepts of general conditions relating sorts of observations to what they signify (EPM, 168).

Sellars thus proposes a new notion of the grounding of empirical knowledge. He does not deny that it has foundations, only that the term 'foundation' obscures the way in which observation reports rest on empirical propositions, in favour of the reverse dependence. This traditional 'givenness' view gives too static a picture: "empirical knowledge . . . is rational, not because it has a foundation but because it is a self-correcting enterprise which can put any claim in jeopardy, though not all at once" (EPM, 170). He also proposes a new analysis of 'inner episodes', which neither dismisses them as a category mistake nor considers their privacy an impediment to characterization in intersubjective discourse (EPM, 176). His Jones-myth provides us with the logic of such an analysis.

D. The 'Jones-Myth': Meaning and Thought

Sellars' Jones-myth is set in the very early history of human beings. The people here are limited to a Rylean language, one whose "fundamental descriptive vocabulary speaks of public properties of public objects". Though limited in resources, this language is highly 'expressive', using the "logical operations of conjunction, disjunction, negation, and quantification, but especially . . . the subjunctive conditional". It would also have a loose logic, a " 'vagueness' and 'open texture' " (EPM, 178).

Why, and in what sense, is this a Rylean language, a Rylean community? To see this, we have to turn to Sellars' preliminary remarks on meaning and thought. Meaning is not an associativity-based relation "between a word and a non-verbal entity". This would turn a meaning-statemcnt into a shorthand for a number of associations. Rather, the 'means-rubric' (". . ." means ---) states that "a mentioned word . . . plays the same role in a certain linguistic economy . . . as does the word . . . which is not mentioned but used -- used in a unique way, exhibited, so to speak -- and which occurs 'on the right-hand side' of the semantical statement". In this way, we do not need different rubrics to deal with the meanings of such divergent linguistic items as "red" and "and" (EPM, 163; there are no non-linguistic items with which to associate "and"). Thus, meaning is entirely within the linguistic realm, a correlation of linguistic items and roles. As a consequence -- since we are denied a simple association of 'this vocable' with 'that object' --, understanding the meaning of a word involves much "knowledge which classical empiricism would have held to have a purely contingent relationship with the possession of fundamental empirical concepts" (EPM, 164).

By referring to roles rather than non-verbal entities, Sellars wants to avoid injecting the 'myth of the given' into the language-learning situation. 'Roles' are subject to entirely public, observable exhibition (thus the name 'Rylean'). The alternative is to regard the language-learner as a 'pre-analytically', emergently aware neophyte in the completed, structured "logical space" of the philosopher. Then, it is assumed, the neophyte learns the language by being taught to discriminate elements of the given space (persons and physical objects in space and time) and to associate verbal symbols with these elements. This is not any better than the sense-datum theorist asking that the child discriminate his sense-data and associate words with them (EPM, 161-2).

Sellars agrees with Ryle in denying this sort of awareness. But he does not go all the way to rejecting 'thought-episodes', so his starting his myth with 'Ryleans' should not be taken to mean that he is elaborating a Rylean position. In fact, except for 'awareness' and immediate experience, Sellars much prefers the classical tradition, where thought "is a family of episodes, neither overt verbal behaviour nor verbal imagery", to which episodes these latter owe their meaningfulness as expressing thoughts. Thoughts are introspectable and self-presenting. However, this represents a mis-assimilation to the self-presentation of sensation and images. Sellars proposes that if we drop the requirement that thought episodes be immediate experiences, then we can rework the classical insight into "the idea that to each of us belongs a stream of episodes, not themselves immediate experiences, to which we have privileged, but by no means either invariable or infallible, access". Such episodes often and 'naturally' issue with overt verbal expression, but this is not necessary to them (EPM, 177-8).

With this further conceptual background, we can proceed with the Jones-myth. Sellars starts his myth in medias res with humans who have already mastered a Rylean language, "because the philosophical situation it is designed to clarify is one in which we are not puzzle by how people acquire a language for referring to public properties of public objects, but are very puzzled indeed about how we learn to speak of inner episodes and immediate experiences."

In going beyond the initial stage, of having a Rylean language, Sellars will show that some things have to be added to the Rylean language, in order that there be recognition of 'thought', 'feeling', etc., as these occur in ordinary language. What is to be added, "and 'How could the addition of these resources be construed as reasonable?' " (EPM, 179).

The first addition would be fundamental semantical discourse -- namely, ' ". . ." means ---' and ' ". . ." is true if and only if --- '. Sellars differs from Carnap in denying that these are logically contained within Rylean language, but does not argue the point as it is "not essential" (EPM, 179).

Having semantical terms is, plausibly, a step towards 'thought-talk', because both semantical discourse "about the meaning or reference of verbal expressions" and "mentalistic discourse concerning what thoughts are about" have a structure which deals with "intentionality, reference, or aboutness". A philosopher in the tradition of the Concept of Mind accounts for this similarity by regarding thought-talk as shorthand for other things in which language is involved; the intentionality of thoughts is reducible to semantical talk about verbal components. The classical position places an intentional-mental realm above the semantical-verbal level, and analyses the latter in terms of the former. Sellars wants to

reconcile the classical idea of thoughts as inner episodes which are neither overt behaviour nor verbal imagery and which are properly referred to in terms of the vocabulary of intentionality, with the idea that the categories of intentionality are, at bottom, semantical categories pertaining to overt verbal performances (EPM, 180).

The next step, after 'Rylese' is enriched by semantical discourse, is to enrich it further with theorerical discourse (EPM, 183). At this stage, our ancestors are ready for the birth of Jones, a genius ahead of his time as he is a methodological behaviourist.

The Rylean language spoken by our ancestors 'before Jones' was a language of non-theoretical behaviourism, says Sellars. In this milieu, Jones 'develops' a theory to account for 'intelligent' human behaviour which occurs in the absence of any overt verbal performances. His theory is that "overt utterances are but the culmination of a process which begins with ccrtain inner episodes". Overt verbal behaviour provides the model for these inner episodes. Jones has a unified theory: not only is 'inner speech' the "true cause of intelligent non-habitual behaviour", but also ". . . overt verbal behaviour is the culmination of a process which begins with 'inner speech' "(EPM, 186). ('Inner speech' does not involve verbal imagery, as Jones would lack "the concept of an image".)

The applicability of semantical categories to overt speech would be carried over to the inner episodes modelled on it; thus, inner episodes have meaning and aboutness. Furthermore, we may note a commentary which limits the model, i.e., 'there is no hidden tongue', 'inner speech is soundless' (EPM, 187).

Let us recapitulate. People first spoke a primitive language with a basic observational vocabulary; they could observe each other's behaviour (Rylese). They then were able to speak of the meaning of their words (the introduction of semantical discourse). Later, they aeguired the ability to construct crude theories about behaviour (the introduction of theoretical discourse). Now, Jones introduces a definite theory, that there are episodes inside language users, similar to overt verbal episodes except that they are silent, which explain the fact that some overt behaviour is intelligent even though the agent is silent. These inner, silent speech episodes come to be thought of as the source of intelligence of behaviour.

Jones' theory has introduced certain entities, 'thoughts', as theoretical. They are not observed, but they can still be thought to exist (EPM, 187).

Thoughts are not observed; however, an illusion of introspeetive self-observation can develop. You and I can both use the samc overt behavioural evidence to warrant our saying 'Bob thinks "p'' '; then, I can be trained, through the applause and frowns of others, "to give reasonably reliable self-descriptions, using the language of the theory, without having to observe (my) overt behaviour". Thus, "privileged access" is developed, somewhat as a skill. So "what began as a language with a purely theoretical use has gained a reporting role." What the 'myth' has helped to show is the primary and essential intersubiectivity of concepts pertaining to inner episodes, and the way 'privileged access' and the reporting role of these concepts depends on this intersubjectivity. The privacy of inner episodes is compatible with the intersubjectivity of language and language-learning; so privacy here is not an absolute privacy (EPM, 188-9).

Jones' second great theory, which concerns us less, is that there are inner entities called impressions which result from the impingement of outside forces on the body, and which explain our sense experience (EPM, 190-1). But, as with 'thought', it is not necessary to think that the entity existed, as described by the theory, before the theory ever arrived: "And notice that while our 'ancestors' came to notice inpressions, and the language of impressions embodies a 'discovery' that there are such things, the language of impressions was no more tailored to fit antcedent noticings of these entities than the language of molecules was tailored to fit antecedent noticings of molecules". Jones did not realize this -- he confused "his own creative enrichment of the framework of empirical knowledge, with an analysis of knowledge as it was". He comes to notice and report particulars, which he takes to have been given (194-6). All this applies to 'thought' as well as to 'impression'.

This is the point at which Sellars has been aiming. He had to go through a more general account of meaning, levels of language, and our concepts of the inner life, in order to support his thesis, that in propria persona impressions and thought could be particulars quite different from the descriptions given them by the classical theories; and that the 'descriptions' of inner episodes of classical theories are actually theoretical explanations, in content and structure, which however lend themselves to such frequent and easy practical use that their theoretical status is overlooked.

E. Problems in Sellars' Doctrine

Consistent with previous restrictions on my exegesis of Sellars, I shall limit myself to a few questions about the Jones-myth (section D above) which are not solved by a simple report.

(i) Sellars' non-relational conception of meaning entails that there is some knowledge which is actually necessary for understanding a word and/or for learning language, which has usually been thought of as merely incidental (EPM, 164). For example, if meaning-statements return our attention to the "roles" of words in language-impregnated human activity, then we need some incipient notion (not necessarily a concept) of "role" in order to understand words. Similar demands arise in another formulation (BBK, 56-7): "signification statements . . . say that two expressions, at least one of which is in our own vocabulary, have the same use"; ". . . a matter of certain items in the real order playing roles". So we need the notions of same, use and role as they apply both to words and to "items in the real order". So far, this demand is consistent with what Sellars has said; it is perhaps an elaboration of implicit points. However, what becomes of our picture of our language-learning neophyte totally lacking "awareness of logical space prior to, or independent of, the acquisition of language"? (EPM, 162). He may not yet be aware of persons or obiects, but he does seem to 'know' something of roles and uses. Obviously, he does have a logical space; but it is 'logical', and he is 'aware' of it, in broad senses of these terms, senses which do not commit us to ascribing explicit conceptual consciousness to the learner. (This suggestion is developed in Chapter IV, Section D.)

(ii) There is a problem with the early stages of the Jones-myth. Granted, we have started "in medias res"; we are puzzled by everything, from this point on, which culminates in the language of "inner episodes and immediate experiences" (EPM, 180). Granted, too, the first 'addition' to pure Rylese will be "the fundamental resources of semantical discourse" (EPM, 183). At this point, Sellars rather strangely says that it is not really relevant whether this is an addition, something logically unique; and he does not bother to indicate how the Ryle-speakers would have acquired these resources.

When he denies the necessity of tracing the emergence ab initio of our mentalistic vocabulary, Sellars is avoiding a fundamental issue. He wants to come to a certain conclusion about 'the given' which is given as epistemic (sense data, immediately-known thought, etc.), namely, that it is not 'given'. Otherwise, contemporary science could not legitimately encroach on 'mind'. So Sellars relegates the 'giving of the given' to analogical extension out of primitive language. But what distinction is there between treating the theoretical essentials of this argument concerning analogical extension, and postulating a sequence concerning how it would occur in rebus? Once we understand and acccpt Sellars' ideas concerning models, theories, and the relationship of theoretical entitites to ordinary language, there remains the question whether this sort of analogy could hold in this sort of situation. Thus, questions are necessary concerning both a full description of primitive levels of language use, and development from that point onwards. (I will return to this objection in Chapter III.)

(iii) What happens when Jones gets into the act? "Suppose, now", Sellars invites us, "that in the attempt to account for (etc.) . . . Jones develops a theory . . ." (EPM, 186). The criticism above regarding the introduction of semantical discourse applies here again -- how theoretical thinking comes into primitive stages of language use is not explained. Rather, we are told that theoretical explanation is well within the artillery of the common man, that science is continuous with common sense -- as Sellars will show. But here, it acts as a crucial premise. Furthermore, Sellars does not find it necessary to find concepts for the impulse on Jones's part to account for things. However, if his discovery is introduced by him as a theory, then to some extent, he would need concepts (not necessarily explicit) for the nature and purpose of theorizing.

(iv) Finally, Jones needs more than Sellars gives him to do this theory-building work: ". . . 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, 186). So Jones has noticed that there is something special -- intelligence -- to human action. How did he notice that? As Sellars has told us (EPM, 175-6), "to have the ability to notice a sort of thing is already to have the concept of that sort of thing, and cannot account for it". I don't see how the Rylean language could have contained a non-mentalistic concept of 'intelligence' (or a reasonably close equivalent). This idea would be covered in the category of overt verbal behaviour: all action which is or involves overt verbal behaviour is intelligent, and all intelligent behaviour is or involves overt verbal behaviour. Once a distinction, within this realm, is introduced between 'intelligent' and (e.g.) stupid or 'merely parroting' (c.f. IM, 525) overt verbal behaviour, then this 'intelligent' already connotes some extra something over and above overtness, verbal-ness, and behaving. I would be content to let Sellars say that the concept of intelligence is a special or specific development within and contemporaneous with the general development of semantical and mentalistic discourse. But what Sellars demands is that this concept (not necessarily explicit) already be there for Jones to use, before and in the course of developing his theory. (See further discussion in Chapter IV, Scction D.)


This ends my depiction of Sellars' Jones-myth, its context and aims. I shall now turn to Sellars' theory of meaning in greater detail, before finally relating his conception of meaning to his pronouncements on language-learning and his support of methodological behaviourism.

[Return to Table of Contents]


A. Meaning-Discourse

In this section I want to consider what Sellars has said so far about meaning, and to make initial links between meaning, behaviourism, and language-learning. I will explore the clarification of the concept of meaning in the 'Chisholm-Sellars Correspondence' (IM). And I will look for an introduction to Sellars' ontological commitments through the logic of Chisholm's refusal to agree with him on the analysis of intentionality.

While the ultimate, overall context of Sellars' discussion of meaning is the meeting between the history of philosophy of mind and modern science, the more immediate context, which I shall examine, is Sellars' reconstruction of behaviourism.

In large part, Sellars is trying to save behaviourism from itself. (For example, he proposes that it can adopt 'theoretical' entities without giving up its claim to be 'scientific': cf. EPM, 185). This wider interest in behaviourism is linked to the theses I am considering here in the following way: Sellars claims, contra Ryle, that ordinary mentalistic language is not just a careless and mistaken bastardization of (traditional) scientific 'behaviourese'.{2} And he goes beyond this (rather unexceptional) claim by suggesting that ordinary mentalistic language can be explained in the course of an account which is behaviouristic in methodology.

As we learned from the Jones-myth, and connected sections of "Empiricism and the Philosophy of Mind", Sellars proposes to go about his reconciliation of 'behaviourese' and 'mentalese' in this way. To begin, since behaviourism is not necessarily the same thing as physicalism, there is no prima facie opposition between behaviourism and 'mentalism', as there certainly is between physicalism and mentalism (EPM, 185-6, 193). On the other side (and from here on, we can follow the Jones-myth itself quite closely), by accepting the 'primordiality' of Rylese, Sellars accepts the familiar notion that the language which people usually learn initially is learnt as a sort of 'behaviourese'. However, if we go beyond this stage -- but not so far that a 'mentalistic' vocabulary of covert episodes like 'thoughts' is being suggested -- then we can imagine people developing semantical or metalinguistic discourse concerning their overt speech, speaking of their speech as 'meaningful'. The basis is thus provided for a later 'invention' of a 'theory of the mental', whose categories of intentionality are a transposition of the categories of semantics. Further, this theory would imply statements of the relations of thoughts to each other (as well as to other sorts of entities) analogous to those statements of the relations of speech episodes to each other (as well as to environment and non-verbal behaviour) which are implied by statements of the meanings of words. If all this is possible, then one can start with Rylean facts about the observable world (environment, verbal and non-verbal behaviour) plus a semantical (non-mentalistic) notion of meaning, and eventually account for the intentionality of ordinary mentalistic terms.

Such is Sellars' account of the origin and logic of mentalistic language. My problem is this: are only Rylean facts plus semantical meaning needed for an account of the intentionality of ordinary mentalistic language, an account which starts out from a Rylean language. That is, is the meaning-relation logically primitive in itself? Or, is it to be analysed in terms of things which for Sellars are logically (and perhaps historically) prior to it? Or may it have to be analysed in terms of concepts which, for Sellars, are logically (and perhaps historically) posterior to it?

Before examining these three possibilities, we must know what it means for something "to be analysed in terms of" something else. It is important to bear in mind that Sellars employs a rigorous conceptual, linguistic analysis. An analysis places a concept in its proper context, in its 'family' of necessarily related concepts. As such, it connotes neither causation nor a vertical hierarchy or derivation such as Chisholm seeks in one of his earlier letters. Chisholm sees the issue between him and Sellars as "whether (1) ' ". . ." means p' is to be analysed as (2)' ". . ." expresses t and t is about p ' " (IM, 524). He explains this as the problem whether living things and thoughts have a 'special something' which they 'pass on' to linguistic episodes, or vice versa. But Sellars claims that his denial that 'means' is to be analysed in terms of expression and thought, really comes down to saying that a person does not necessarily need the concepts of thought and expression in order to understand the concept 'means' (IM, 534). Sellars is wary of allowing a too quick assumption of what general or specific substances and properties may be involved; he conducts his analysis entirely at the level of the set of concepts necessary to understanding a certain chosen concept.

Sellars' notion of 'because' also comes down to 'outlining a family of concepts'. His 'because' is something which gives 'clues', without prejudice concerning hierarchy or causal progression between the terms involved. Normally, 'A because of B' conveys to us the idea that B is the origin of something (or everything) to do with A -- its existence, one of its properties, etc. But Sellars' use of 'because' is closer to the following example. Suppose a sonar-equipped ship is becalmed in total fog. The captain could say 'We may be moving, because the sonar indicates changes in depth." Here what follows "because" is a clue or reason, not a cause -- the sonar is not moving the ship. So, too, when Sellars admits that "statements mean states of affairs because they express thoughts which are about states of affairs", this 'because'-clause does not specify a causal source of the statements' meaningfulness (IM, 522). Rather, the aboutness of thinking is an excellent 'clue' that statements expressing thoughts possess meaningfulness (once the appropriate theory is known, see below, p. 34 ).

Given this sense of "analysis", the 'three possibilities' of the preceding paragraph but one are no longer mutually exclusive in any simple, straightforward manner. Taking 'analysis' in the way described above complicates our question, "Is the meaning-relation a 'primitive'?", by giving us two meanings for the term primitive. In its strictly logical meaning, primitive is opposed to derived or constructed. In an historical, social sense, it is opposed to sophisticated. Sellars respects both logical and ordinary language analysis; he refuses to give priority to either the logical or the historical sequence ( cf., IM, 536).

For Sellars, the meaning-relation and semantical discourse are logically primitive. They are not "constructed out of the resources of behaviouristics"; if we want to go beyond a behaviourese, " 'means' or 'designates' with its context ' ". . ." means --- ' " needs to "be taken as primitive" (IM, 523).

So the meaning-relation cannot be constructed out of or logically reduced to behaviourese. Does this mean that it cannot be analysed in terms of things which are logically (and perhaps historically) prior to it? Not entirely. Its logically primitive status implies that behaviourese is not a sufficient condition for semantical discourse; however, Sellars still considers it to be a necessary condition. Semantical discourses 'rises up' in the historical and logical context of a Rylean language. The meaning relation thus depends on the pre-semantical, pre-metalinguistic stage of overt speech (i.e., behaviourese) and its noting of overt dispositions and relations; it is also connected to a wider range of concepts and procedures, like 'role' and 'learning', which make up its context. The meaning- relation is in a 'family' with concepts re Rylean language and behaviouristic roles and language learning, all of which are involved in the analysis (or understanding) of the meaning-relation, without jeopardizing its status as logically primitive.

What of our third possibility -- whether the meaning-relation is to be analysed in terms of things which are logically (and perhaps historically) posterior to it? For Sellars, mentalistic language and the concept of thought are logically posterior to and dependent upon semantical language and the concept of meaningful speech. But taken as historically primitive, the meaning-rclation gradually becomes more sophisticated (cf., IM, 535 and below, the water-H(2)0 example) because it acquires new roles more things can be spelled out in terms of the means-rubric. Nevertheless, the meaning relation is not derived from or constructed out of this entire set of attributions (both posterior and prior). It is a logical primitive vis-à-vis these attributions, and historically it can be prior to (or set apart from) at least some of them.

In the following review of Sellars' thoughts so far on meaning, I will try to bring out these two aspects of the 'meaning-relation', its logically primitive status,and its inevitable connection to and its immersion in a context.

The context is set by the fact that our pre-Jonesian ancestors speak 'Rylese' a language stipulated by Sellars to have the logical character of language endorsed by anti-'theoretical entity' behavioural psychologists (EPM, 186). Its vocabulary would deal exclusively with "public properties of public objects"; and, as a behaviourese, we can assume that the learning of it would be summed up in traditional theories of S-P conditioning. None of this denies that it could be a complex, expressive language with an "open texture", says Sellars (EPM, 178). He takes all this to be nonproblematical. Rather, the problem for him is, how would people who speak this complex, open-textured, S-R conditioned language dealing with single and generalized public observables, come to have a mentalistic vocabulary as well? Sellars claims we can arrive at this new sort of language from Rylese via an intermediate step of adding semantical or mentalistic discourse, concerning the meaning and truth of statements, to the initial 'Rylese' (EPM, 179-80). He doesn't say where this 'comes from', although it certainly isn't something contrived entirely within the Rylean language.

What is this meaning-relation, which is contributed by the addition of semantical discourse? We have already seen, of course, that it is not a relation in the conventional senses of 'relation'. 'Meaning' does not associate some word with some non-verbal entity; nor is it a shorthand for a number of such associations. Therefore, if we speak of 'meaning-relation', this is merely an ellipsis for the concept of meaning and the rubric ' ". . ." means ---' (EPM, 162-3). (Although it may cause conceptual confusion, I will continue to use the words 'meaning-relation' because of their well-established familiarity.)

For the purposes of this examination of Sellars, I will grant that the 'meaning-relation' -- the concept of meaning and its context -- is logically an addition to Rylese which cannot be constructed out of the resources of Rylese (IM, 523). As such, the meaning-relation, or semantical discourse, has something unique (different, logically primitive) about it, which makes it impossible for Rylese to be sufficient (even if necessary) for its appearance.

What are the consequences of this unique feature? I am puzzled about why Sellars considers the issue of the uniqueness of semantical discourse "not essential to the argument" (EPM, 179). In the wider context of Sellars' aims, it is essential, both that a science of psychology with behaviouristic methods is both possible and adequate to the description of everything which ordinary mentalistic discourse describes; and that this new science be different from classical behaviouristic psychology and from the Rylean position on the category-mistake status of all mentalistic discourse. 'Rylese' is to 'Rylese' enriched by semantical discourse as classical behaviouristic psychology is to this new science proposed by Sellars. Thus the uniqueness of semantical discourse vis-à-vis Rylean discourse is an essential point, for it turns out to be the basis of the claim to uniqueness of Sellars' new behaviouristic science.

Furthermore, in the narrower context of the Jones-myth, Sellars on his own admission is interested not in how Rylese is acquired but in every stage past that point. Since the insertion of semantical discourse into the speech-world of the Ryleans is the first of these stages, I cannot see how it is inessential whether or not it is 'unique'. Either semantical discourse is surreptitiously contained already in Rylean discourse; or there is a plausible description of how semantical discourse is inserted into 'Rylese'. Sellars denies the former, and disregards the latter. (This is a continuation of a previous objection; see Ch. II, Sec. E.)

Therefore, I will be a little bit more concerned with construing the 'unique', 'primitive' aspects of the meaning-relation than perhaps Sellars appears to be. Let us analyse it. If meaning is not a relation, not an association or set of associations, then what is it? Sellars says that the means-rubric translates, or virtually equates, two words in virtue of their similar roles or uses or functions in their respective "linguistic economies" (EPM, 163). (I say "virtually equates" because the means-rubric provides a starting point; qualifications usually follow (IM, 528).) Two consequences follow from Sellars' hint that meaning is like translation. First, a single means-rubric, a single concept of meaning, is adequate to all words, since all words have roles or uses. Second, whereas meaning on the old model could often ignore roles in favour of a simple association of a word and an object, meaning as translation involves this wider set of (for others contingent) facts (EPM, 163-4).

We have to be careful here of the historical primitive sense of the means-rubric. If it is comparable to translation, then the linguistic economy would be 'ready for' the concept of meaning only when there is issue of more than one language, or more than one word for a use or role (synonymy) -- that is, only when translation is needed. And we could probably find empirical support for this: the time of Homer is earlier than the time of Thales; semantics and philosophy may have sprung up because an oral tradition of poetry, carrying on an older, unchanging language, presented the changing contemporary ordinary language with the problem of finding modern synonyms for the vocabulary of the ancient poets. However, if we construe the historical time of the means-rubric this way, we have also to imagine (i) the poet working without semantical discourse, and (ii) language being taught without semantical discourse. I think that (i) presents no problem; but (ii) will be investigated in light of Sellars' comparison of language-learners and language-users and teachers in "Language as Thought and as Communication".

Although translation may be the historical need which gave rise to meaning-discourse, we are still quite far from having an internal explanation of this phenomenon. And staying 'outside' just won't do. In another essay, Sellars says that we (human language-users) can speak of a robot's code-record of the world as meaningful by linking it via the means rubric to words and their roles in our language (BBK, 54, 57). Whether or not this anthropomorphizes the robot, it does presuppose the existence of human speakers who already have semantical discourse. Such a move cannot explain the coming-to-be of semantical discourse in the Jones-myth, since no humans already have it at the time it 'happens' in the Rylean community.

Sellars himself acknowledges this problem. He finds it easy enough to say what the meaning-relation is not -- for example, it is not a relation; but it is extremely difficult to say precisely what it is (IM, 531-2). What is it to 'invent meaning' (or even just to 'acquire' it), to begin to treat our own behaviour consciously as meaningful, rather than merely to have the power of meaningful behaviour? These questions are left open; we find ourselves 'in medias res' not only in the sense of bypassing the absolute beginnings of (Rylean) language, but also -- what Chisholm objects to -- in that we never learn what meaning is 'in itself', or where it 'came from'. We are left to rely on the overall plausibility of Sellars' myth and his philosophical system to convince us of their truth. To put it another way, our concepts of the mental are supposed to get worked out by analogy to concepts pertaining to meaningful speech (BBK, 43-4). But what constitutes the meaningfulness of speech? Our only hint is a coherent description of an imaginable, ongoing social group utilizing simple speech.

We are in a better position concerning the meaning and origin of mentalistic discourse. There is a difference between the introduction of 'mentalese' into Rylese enriched by semantical discourse, and the introduction of semantical discourse into the original Rylese. According to the myth, Jones invents mentalese, at first merely as suppressed inner speech on the model of overt speech, to explain the 'intelligence' of intelligent behaviour unaccompanied by overt speech (EPM, 186). In other words, Jones has a reason for inventing mentalese; mentalese (to put it in the language of evolutionary pragmatics) fulfills a need. But Sellars is unable to specify a reason for which semantical discourse would have been invented, a need which it fulfilled. Of course, the Ryleans are 'ready to receive' semantical discourse -- they have a language, and all their "routines" (IM, 526-7) will be entailed by meaning-statements. But how and why should this reflection of language on itself be invented in the first place? The previous suggestion, that the need to translate provides the historical occasion and need for meaning- discourse, does not answer the present logical question; since the concept of meaning is basic to the notion of translation, the concept of translation cannot be the basis of the concept of meaning.

To see why this query is crucial, we should recall that the introduction of the language of means is the first innovation of Sellars' myth. Not only is the language of mental events introduced in the course of development from this starting point; but also, introspection, the clarity and distinctness of immediate experience and its status as an infallible access to the substantial reality of the mental, is explained away in the course of the Jones-myth as a skill and an illusion. This introspection -- on the classical model, of course -- is the starting point for Sellars' opponents. Their counter- attack on Sellars could be something like this: they could say that the fact that introspection always seems to be mediated by language, is first of all untrue -- some sorts of (introspective) 'feels' experience seems to be 'beyond' or 'outside of' language (e.g., imagistic or musical experience). Second, mediation of introspection by language may be a contingent matter -- once a person learns a language, the power of language is such as helps it dominate all forms of thinking, some of which thinking, nevertheless, could and did take place pre-linguistically. Third, perhaps the domination of introspection by language can be overcome -- witness the via negativa of meditation among medieval mystical philosophers (negation of verbal-conceptual truths leads to intuition).

I am not interested in justifying these steps which a classicist would perhaps take against Sellars. I want, rather, to point out that introspection is a starting-point in the classical argument, and it would appear that Sellars undercuts the classical position by showing introspection to be a built-up skill, full of presuppositions, which gives the illusion of pure givenness. But if Sellars' own account, which leads up to this point, starts out with a premise which has no more justification than that of the classicists, then nothing more than a stalemate can be hoped for. Indeed, the classicists' premise has its own prima facie justification, in that many people have claimed to have had this unmediated introspection. Thus, if Sellars is to win this argument with the classical tradition in the philosophy of mind, he should find a reason or justification for the introduction of meaning talk into Rylese (this introduction being the starting point of his positive account of the meaning of 'mentalese' and at the same time of his revision of the classical tradition), and moreover, a reason or justification which does not land him in the camp of the classicists.

In his correspondence with Chisholm, Sellars faces persistent objections to the uniqueness of semantical discourse. He claims that 'means' "is not a logical term nor constructible out of such" (IM, 524); rather, it is the "core of a unique mode of discourse'' which is neither descriptive nor explanatory of empirical fact (IM, 527). Then Chisholm counters that a 'means' statement is either true or false, which would make it descriptive of certain empirical 'lexical facts' (IM, 529), Sellars' rejoinder is that a 'means' sentence implies but does not assert such facts as would make '' 'Hund' means dog" true (IM, 530). Now, what is it for facts to be implied, but not asserted, by a 'means' sentence?

We can find a clue to what a 'means' sentence is in a similar maneuver of Sellars' concerning colours. Sellars thinks 'that it just won't do to say that x is red is analysable in terms of x looks red to y". But he has to come to terms also with "x is red <--> x would look red to standard observers in standard conditions", which seems to be a "necessary truth" (EPM, 142). The solution is that it is a natural (light-wave plus public convention) fact that physical objects are certain colours; this is expressed on the left side of the equivalence. But the right side does not give a definition of the left; rather, it expresses what we mean by a thing's being a certain colour. On the right we have a social, epistemic and linguistic fact; on the left, a 'natural fact' which is not (necessarily) to be analysed in epistemic terms.

The application of this to meaning leads us to see a 'means' statement as embodying a 'natural fact' concerning the social, languaging world. It implies a set of Rylean facts about the relations between environment, verbal behaviour and non-verbal behaviour. But semantical discourse is unique, so what a 'means' statement asserts is something more -- maybe something about meaning itself, but Sellars does not say.

This implication-assertion structure is repeated in the case of mentalistic discourse. Once thought-talk is invented on the model of means-talk, the above relations are "the effective model for the roles played by these inner episodes" (IM, 530). A mentalistic relation of a thought to a material object is the analogue of a (Rylean) relation of a bit of verbal behaviour to a bit of environment. These relations are implied by statements concerning thoughts; but what is asserted is something more -- perhaps something substantive (and incorrect, says Sellars) about thought itself.

We have been shown a relation of the "unique'' means-rubric to its wider conceptual context of Rylean facts, facts which it implies. Another way in which Sellars makes this link is on an analogy to the history of science, which allows a true commonsense statement to be replaced by a more sophisticated statement once the appropriate theory has been proposed and understood. For example, now that molecular theory has been proposed, statements using the sign-design 'H(2)O' are taken to be implied by statements using the sign-design 'water', by those who have heard about and understand molecular theory; and the molecular statements are true if the water statements are true. However, until the molecular theory was promulgated, a 'water' statement could be true whereas an 'H(2)O' statement would have been just gibberish (IM, 535). Similarly, although we now say that statements concerning the meaning of overt speech episodes imply statements about the intentionality of thought, this can be understood as a historical sophistication of the means-rubric, which could have been used in making true statements long before such a theoretical accretion to it was proposed.

If our understanding of the meaning of words does not come ultimately from a notion of the mental, since 'the mental' (whether or not this constitutes a theory) is a subsequent sophistication based on meaning-discourse, then where does it come from? Sellars answers that people who can communicate in a language eo ipso have this understanding of

the syntactical or 'intra-linguistic' structure of the language, as well as such facts as that, ceteris paribus, a person who says "This is green" is in the presence of a green object, and a person who says "I shall do A" proceeds to do it. This understanding springs from the routines by which the language is learned and passed on from generation to generation by "social inheritance". (IM, 526-7)

Thus, the conceptual and historical context of the meaning-relation is again widened and specified: it includes (though it is more than) the understanding and activities which go into using as well as preserving language in general.

Sellars sums up the relation of his theories of meaning and of the mental "by the formula that just as semantical statements about linguistic episodes do not describe, but imply a description, of these episodes, so statements about the 'content' or 'intentional object' of thoughts do no describe thoughts, though they imply a description of them" (IM, 536). The direct description of the linguistic episode in its meaningfulness remains a puzzle concerning this primitive, the 'means-rubric', and this "unique mode of discourse", meaning-discourse. So too, the direct description of thought-episodes is still a mystery; but having shown that the classical theories do not directly describe thought episodes, Sellars feels free to speculate that neurophysiology may eventually give such a description (IM, 537).

What of Chisholm's position? Although he claims to agree with the Jones-myth, and he accepts various formulations (e.g., Sellars' use of 'analysis'), at the end he still finds things incomprehensibly paradoxical: "If the people of your myth were to give just a little bit of thought to the semantical statements they make, wouldn't they see that these statements entail statements about the thoughts of the people whose language is being discussed?" (IM 537). Yet Chisholm had admitted the possibility of people having a notion of language as meaningful before they had a notion of thought. And when he asks that people "give a little thought" to meaning-statements, he is not differentiating between thinking simpliciter -- which Sellars admits can be done even pre-linguistically -- and 'giving thought', which for Sellars involves having a theory of thought. Chisholm's metaphysics leads him away from employing this distinction in his arguments, as I will try to show below and at the beginning of the next section.

A summary of Chisholm's position has to start with this metaphysical dualism, as well as his equation of behaviourism with physicalism, which Sellars of course denies. When Chisholm divides the universe relevant to language into "marks and noises" on one side and thought and "living things" on the other, asking which side includes the "funny characteristic" of intentionality which is somehow caused to be extended to the other side (IM, 524), Sellars counters that this physical object, sign-design notion of language units is unnecessarily restrictive. "Marks and noises are, in a primary sense, linguistic expressions only as 'non-parrotingly' produced by a language-using animal" (IM, 525). Starting with two 'substances', the physical and the mental, Chisholm asks which is 'on top of the other' such that intentionality might 'trickle down' from its primary source to the other. With his 'conceptual-context' rather than causal sense of analysis, Sellars utterly rejects this 'vertical' image of the task of analysing intentionality. Rather, he says, we can legitimately start in medias res, with a common sense awareness of the meaningfulness of language as we and others use it. We do not recognize speech as speech, as opposed to random designs, by adding an 'Oh, this seems to evince a sort of mentality' to an 'Oh, what an interesting physical design'. If we agree to the possibility of this sort of awareness being primitively given (as Chisholm does at times agree), then, Sellars says, our first task is to stay with a sort of 'horizontal' analysis of the conceptual, social, and historical context of this meaningfulness-awareness.

Chisholm's own summation of his arguments illustrates this 'verticality' and this lack of interest in conceptual-historical sequence. A seven-step syllogism, it first asserts that the 'thought- family' and the 'linguistic-entity family' exhaust the universe of intentional things. Without linguistic entities, thought would still be intentional; but if thought did not exist, linguistic entities would not be intentional. Hence, among the things of the world, members of the thought family are the primary bearers and source of the characteristic 'intentionality' (IM, 533). Now, although Sellars can agree with Chisholm's construal of the universe of intentional things, Sellars could not begin an argument with the delimitation of these two classes; for his point is precisely that we are aware first of language, then of language as meaningful, and finally of both language and thought as meaningful or intentional. Chisholm disregards the thinker -- his psychology and history, and their epistemological bearing -- and treats 'thought' alone.

An aspect of Chisholm's disregard for the thinker is the lack of differentiation between thinking and thought. If there were no thinking in the world, then ipso facto there would be no 'humans playing language games in a society' in the world; then any random linguistic-entity seeming 'sign'-design in the world would not really be a linguistic entity, and would lack intentionality. With this, Sellars would agree. But to say that linguistic entities would lack meaning or reference if there were no thought in the world, is denied by the Jones-myth, which claims that the meaningfulness of language could be appreciated before a theory of thought was introduced, and thus, before there was consciousness of thought under mentalistic-concept theory.

The Chisholm position is reformulated by Sellars until it is quite close to his own. The essential difference between them is not whether people 'discover' thought and its intentionality; but if, as soon as they 'discover' thpught, they are simultaneously faced with the dependency of speech on thought for its meaningfulness -- even if the meaningfulness of speech had been noted prior to the discovery of thought (IM, 534). In other words, Chisholm thinks that this dependency is somehow 'given' in 'introspection' to people who "give just a little bit of thought to the semantical statements they make'' (IM, 537).

In opposition,Sellars complains against Chisholm's avoidance of a conceptual analysis of intentionality; instead, Chisholm resorts to "such philosophical wrenches as 'property', 'relation', 'attribute' ", characteristic, etc. (IM, 535). Sellars would prefer to come to such a vocabulary of dealing with things as they are 'in themselves' only after a proper ground-clearing vis-à-vis the development and interconnections of our concepts of things. Furthermore, he refuses to start with a predetermined ontology and metaphysics, of 'physical-mental' or whatever. If it is necessary, because of the progression of awareness, to speak of an ontological level of 'meaning in the world', Sellars wants to be free to do so.

B. Ontological and Metaphysical Considerations

In this section, I do not intend to defend definitions of 'ontology' and 'metaphysics'. Rather, I want to prepare for the discussion of the issue of language-learning by examining Sellars' commitments concerning the ultimate categories of 'what is'. This examination can best start with a contrast with Chisholm's commitments, based on the previous section's critical summary of his views expressed in the correspondence with Sellars.

We can distinguish three traits in Chisholm's ontology and his related mode of argument. He follows an overriding dualism, whereby everything is ultimately either mental or physical. If two things are ultimately different, one ('P') must be physicalistic and the other ('M') mentalistic. Chisholm proceeds to argue in the following manner: If 'P' and 'M' are two items which both have a characteristic 'C', then either 'P' is the source of 'C' in 'M', or 'M' is the source of 'C' in 'P' (cf. IM, 524). There can be no tertium quid as source of 'C' in both 'M' and 'P', since everything is either physical or mental. (Considering behaviourism and its language, therefore, Chisholm classes them ultimately as physicalistic, since they are not supposed to be reducible to mentalistic categories.) Such an argument has two interesting features: There is an idea of 'sequence' or 'hierarchy' involved, for one thing is logically prior to another in respect of a certain characteristic which each has. And, related to this and to the continuous reference to the physical-mental dualism, is the idea that such an argument is substantive rather than conceptual: it identifies a causal sequence rather than a conceptual interrelationship.

Filling in for 'C', 'M', and 'P' above, Chisholm's stand on the issue of the intentionality of language and thought is that 'Human mental life, especially thought, is the logically prior, causal source of the intentionality which characterizes some physical things, namely, (some) marks and noises'. Chisholm can debate only with another dualist who reverses the causal sequence: 'physical things, namely, certain marks and noises, are the logically prior, causal source of the intentionality which characterizes thought, which is part of human mental life'. Anyone outside this framework is being paradoxical, is trying to "have his cake and eat it":

to admit that "to analyze the kind of meaning that is involved in natural language" we need a distinctively semantical term ('means') which cannot be analyzed in physicalistic terms, but deny that the explication of this distinctively semantical term requires a reference to thoughts, has all the appearance of paradox. (LTC, 526)

This is the formulation of his own stand that Sellars gives on Chisholm's behalf.

But it is really Chisholm who puts himself in a paradoxical position, by granting Sellars the Jones-myth yet denying the consequent logical independence of means-talk from thought-talk. The paradox lies in his attempt to treat mental-talk as a 'special case' subject to fundamentally unique analysis. In granting the Jones-myth, as well as the preceding attacks on the 'myth of the given', Chisholm has granted that 'A looks r to S' is not a triadic relation between a subject, an object and a quality (EPM, 141). So too, he has acknowledged (IM, 524) that ' "x" means X to S' (where "x" is any linguistic episode) is not a triadic relation between a linguistic episode, a thought, and a subject (EPM, 163). But whenever these episodes are "m" (linguistic episodes employing mentalistic vocabulary), Cllisllolm seems to think that the denial of triadic relationship is no longer to the point. Rather, ' "m" means M to S' involves an epistemic experience and its referent which are both 'internal' to the subject (interpreting Cluisholm now according to the classical Cartesian position); thus, the ability to note the meaning of a mentalistic word is part and parcel of the subject's being in an immediate relation with the mentalistic referent.

The immediate relation just spoken of is, of course, the paradigm of 'introspection' or 'immediate experience'. And, as we already know, Sellars denies that whatever introspection there is must be on this classical model. Consequently, he rejects Chisholm's implicit distinction between mentalistic and other language, a distinction whose purpose is to allow that an 'internal', introspectionistic movement is legitimate for construing the meaning of mentalistic vocabulary.

I wish to argue that Sellars' theory is accomodative and ontologically agnostic (vis-à-vis these traditional categories). He can accomodate the insights of Cllisholm's classical position -- e.g., that there are such things as mental episodes, that meaning statements are intimately related to thought-and-cxpression statements -- while removing their ontological commitments without irreplaceable loss of meaning. He himself is agnostic as to the real status of thoughts; he leaves this question open. Thus, so long as we didn't use them to explain anything which his 'myth' had explained in a different fashion, we could as validly suggest that well-ordered armies of angels are thoughts in propria persona, as that thoughts are really the entities and processes investigated by neurophysiology. Sellars is not (logically) committed on this point. Rather, he has been trying to break down the modes of argument whereby the solutions to questions regarding the ontological status and substantive reality of 'the mental' were given in discussions of our concepts of thought and the mental, these discussions being dominated by the illusions of 'givenness' and 'immediate experience'.

This shows that Sellars is ontologically agnostic as to what thoughts really are, but that he is not without any ontological committments whatsoever. Although he is a common-sensc realist -- public objects are the referents of our object-language -- Sellars also asserts that the world of meaning or signification is fundamentally different from the world of Rylean facts (IM, 523). What are the ontological implications of this assertion of fundamental difference? Let me answer by continuing the contrast with Chisholm.

Earlier in this section I identified three traits of Chisholm's position -- physical-mental dualism, hierarchy or sequence in arguments, and substantive-causal rather than conceptual relationships. Were Sellars to accept these, he could not propose a tertium quid. However, he does oppose them. Against Chisholm's dualism, he claims that "logical behaviourism", even though it denies the 'givenness of the mental', is not to be identified with "logical physicalism" (LTC, 525). Against hierarchical argument and causal relations, he prefers delineating families of concepts; but he cautions that these families are more often constructed by the investigator than merely found. Therefore, he warns against metaphors which "suggest a sequential strategy" (LTC, 514 and footnote), as these can give an illusion of causal sequence. More positively, Sellars claims that behaviouristic methodology does not lead inevitably to substantive behaviourism; rather, it can also lead to naturalism concerning "the intentionality of conceptual acts" (LTC, 506).

Sellars' third course, then, between physicalism and mentalism, is some sort of 'behaviourism with a difference'. Sellars says that his methodologically behaviouristic concept 'thinking-out-loud' is the fundamental concept pertaining to thinking; the "intentionality or aboutness" which we regard as being part of this thinking-out-loud "is simply the appropriateness of classifying it in terms which relate to the linguistic behaviour of the group to which one belongs' (LTC, 527). The 'difference' then, is the fact that the categories of 'meaningfulness', legitimately applied to speech, are appropriate in virtue of a connection to 'society' rather than by virtue of the (introspected) internal character of speech as a (partly) mentalistic entity.

Why is this important? Sellars is showing that a realism which starts with what is ready-to-hand around us -- the world of physical objects, human society -- meets up with 'meaning'. It was previously thought that realism could be either introspectionistic or behaviouristic, and that 'meaning' would be a nonsense-word for the latter. Sellars is providing a middle course, a behaviourism in which 'meaning' is preserved. Of course, the initial difference between Sellars' behaviourism and traditional behaviourism is that Sellars accepts the possibility of the existence of 'inner episodes'. But it is the uniqueness of meaning discourse which leads to this acceptance; moreover, the meaning discourse gets, 'cashed out' in terms of 'ongoing social-linguistic groups' (as opposed to being cashed out in terms of thoughts).{3}

One way of situating 'meaning' more precisely is to see how it relates to non-intentional realms. Sellars says that the connection or relationship between the orders of signification and of reality is a matter of "items in the real order playing roles" (BBK, 57 footnote). For example, as it is applied to Socrates in a sentence 'Socrates is a man', the word 'man' is playing a role in the real order. But its way of playing a role is different from the way other things play roles in the real, non-intentional order -- for example, the cornerstone's role in supporting this archway. As a preliminary clarification, let us speak of symbolic "items in the real order playing roles". Being a symbol is a way of being real and being 'meaning-ful' at the same time.

How can this happen? At one point, Sellars says that it is in virtue of our treating them that way, that physical items can be considered meaningful, a language (BBK, 56-7). This sounds dangerously like Chisholm's attitude: in virtue of something which we humans uniquely have, we can infuse other items with certain characteristics -- that is, in virtue of the intentionality of our thoughts, 'vocables and printables' are infused with the intentionality or aboutness peculiar to linguistic episodes. Indeed, it seems to boil down to a physical-mental dualism once again -- as 'mental ', we infuse physicalistic 'picturing' with 'meaning'. However, Sellars claims, via his myth, that if we start in medias res with people speaking a primitive language -- rather than starting, as Chisholm does, with a listing of what is and what is not intentional (IM, 533) -- then, before we are ever faced with this sort of dualistic ontological choice, we come upon a sort of 'order of meaningfulness'. "Words as they occur in meaningful speech" are "related to the give and take of man's relation to himself and his environment" (BBK, 44, 57-8). It is this 'give and take' of social life with which the individual is primordially faced -- not with choices between 'physical' and 'mental'.

Therefore, it is inappropriate to conceive of a linguistic unit as a sum of physical and mental components. Sellars denies that the "linguisticist" seeks intentionality or aboutness in "marks and noises" conceived merely as sign designs: "Marks and noises are, in a primary sense, linguistic expressions only as 'nonparrotingly' produced by a language- using animal" (IM, 525). The response given by a human being to a stimulus after successful conditioning

is not as such the concept of the vocalizing as a linguistic response. For to classify an item as linguistic involves relating it to . . . a system . . . 'Word' goes not only with 'object' but with 'person', 'ought-to-be's', 'ought-to-do's , and much, much more. (LTC, 513)

Sellars is arguing that concepts of meaningful speech are part of a large and necessarily related family. It would be a denial of this necessary relation, to hold up one part or aspect and pronounce it to be the concept of (e.g.) linguistic expression. If anything, it is the procedures by which language and concepts (both semantical and mentalistic) are come upon and learned, which will have pride of place.

Another way to situate 'meaning' more precisely is to try, with Sellars, to achieve an isomorphism of the knower and the known (cf. BBK, 41 and passim.). In connection with this issue, Sellars tends to think of meaning in terms of translation. An isomorphism of word and world on the intellectual level is achieved if this translation-relation can link words, as they are (conceptually) for the knower, with something (conceptually) identical. Thus, since words are meaningful to the knower, the demand is that the world somehow possess a sort of 'meaningfulness'. But the world of sticks and stones, 'concrete reality', can involve 'roles' only in some very extended sense. Therefore, it would seem that this ontological level of 'meaningfulness in the world' could exist only in virtue of the existence of social groups. (Indeed, when humans are removed from all human social contact from before the normal age for language-learning onward, they seem incapable of ever appreciating or emulating the 'meaningfulness' of human linguistic behaviour; cf. Francois Truffault's film "Wild Child".) As Dufrenne comments, "once learned, it (language) constitutes a manner of being for the subject, the property of a nature informed by culture" (CSP, 531).

We have just seen, from two points of view, that according to Sellars there must be an 'order of meaningfulness', and that its existence is linked to that of society. First, words are both real and meaningful because they are linked to man's relationship to himself and society. This is philosophical 'bedrock', more primary than the question of an entity's physical-or-mental status. Next, to explain the phenomenon of know]edge, the world itself is to be seen under the guise of 'meaningfulness'; but this cannot be imagined possible without the existence of social groups. As Sellars says, "it is the linguistic community as a self-perpetuating whole which is the minimum unit in terms of which conceptual activity can be understood" (LTC, 512).

Let us look at this Wittgensteinian formulation carefully -- "self perpetuating" implies teaching and learning; "minimum unit" connotes the primitivity spoken of in the previous section concerning meaning; ''can be understood" implies that we are in a conceptual-context (not causal) investigations looking for a family of concepts. Noting all this, we seem to have here an adequate preliminary formulation of the site of the 'order' or 'ontological level' of meaningfulness. That is, given his rendering of meaning-discourse -- and, by implication, what it is to be a linguistic community -- Sellars can propose the ongoing language using community as the ultimate philosophical 'bedrock' for conceptual explanation of conceptual activity, provided that he can give an account of the teaching and learning of language. It is to this account that I now turn.

[Return to Table of Contents]


A. Introduction

Although it contains an explicit model or sketch of language-learning, Sellars' recent essay "Language as Thought and as Communication" cannot be understood simply as an explanation of the procedures of language-learning. This essay provides most of our material on Sellar's attitude to learning; but it sets learning in the context of the relationship between meaning, conceptual thought, expression, action, and communication. Let us explore this context briefly, in hopes of having a more accurate discussion of language-learning.

Our point of departure for setting the logical context of language-learning will be a remark of Sellars' concerning the context of knowledge:

empirical knowledge, like its sophisticated extension, science, is rational, not because it has a foundation but because it is a self-correcting enterprise which can put any claim in jeopardy, though not all at once. (EPM, 170)

Knowledge is not made up out of individual 'hits', each standing or falling on its own individual merit. Rather, characteristics of the enterprise as a whole constitute the rationality of knowledge. When speaking of individual concepts or performances, it is illegitimate to treat these as if they were isolated 'bits' possessing or lacking foundation. For example, having the individual concept of 'green-ness' is not the same thing as merely responding with the sentence 'This is green' to the proper object in appropriate conditions. To be said to have this concept, one also must know, for instance, that the conditions "are appropriate". Thus, this supposedly 'single' concept is a concept only to the extent that it is part of a "battery of concepts" (EPM, 147-8).

We are led back to the arena of actual learning by the fact that these concepts are in no way 'given' by basic 'awareness' of facts or things: "to have the ability to notice a sort of thing is already to have the concept of that sort of thing, and cannot account for it" : (EPM, 176). If 'noticing' is not to be our fundamentum, our primitive vis-à-vis concepts, what is? Sellars solves the problems of noticings and of concepts at the same time:

all consciousness of sorts or repeatables rests on an association of words (e.g. 'red') with classes of resembling particulars . . . not even the awareness of such sorts, resemblances and facts as pertain to so-called immediate experience is presupposed by the process of acquiring the use of a language (EPM, 160).

The associative learning of habits basic to language -- which does not necessarily take in any language learning, but does point at least to learning a basic observational language, a 'Rylese' -- is conceptually prior and basic to both the having of concepts and the ability to notice things (which itself depends on having concepts).

The preceding two paragraphs show that, in rejecting the 'foundations' logic for knowledge which the 'myth of the given' endorses, and in replacing it with a 'whole-enterprise' logic, Sellars ultimately must face the issue of learning as the logical ground of his new logic of knowledge. This is implicit in another remark. Sellars asserts that "the possession of fundamental empirical concepts" necessarily involves knowledge which the classical empirical tradition thought to be only contingently related to the possession of concepts (EPM, 164). What are the things, knowledge and otherwise, which contribute to the acquisition and possession of concepts, and which are in fact necessarily rather than contingently related to this? The answer comes after his "battery of concepts" remark, where Sellars acknowledges that concept acquisition involves "a long history of acquiring piecemeal habits of response" (EPM, 148). However, as we just saw, having the ability to respond accurately is merely a habit; it is not the possession of a concept unless the subject possesses all the concepts regarding the logic of observation and response.{3a} Although insufficient for having knowledge, 'habit' and 'habitual response' are a necessary part of the complex involved in being able to think conceptually. With subsequent questions about, for example, instinctual versus acquired habits, we are led by this conception of knowledge to the issue of learning as fundamental to epistemology.

Sellars himself introduces concepts to deal with the learning situation. For instance, he warns against thinking of the language-learner as minimally, then gradually more, aware of the "logical space" in which language-using adults operate. Rather, the learner is likely to be aware of and within his own logical space. When Sellars says that his own 'psychologically nominalistic' position denies ''that there is any awareness of logical space prior to, or independent of, the acquisition of a language", we must bear in mind that Sellars is attacking a particular interpretation of this awareness, that by which man is capable of immediate experience or introspection independently of mediation by language (EPM, 161-2). So it is still legitimate to talk of our language-learner, his logical space, and his awareness of it, provided we do not commit the logical anachronism of ascribing to him either the concept-employing 'noticing' typical of adult language-users, or the immediate experience/introspection which is possible only after language is mastered (EPM, 189). Thus, we need to explore the logic of terms like 'awareness', 'logical space', 'role', and 'notice', as they are when loosened from these two sorts of interpretations.

In order to examine the issue of language-learning, we are forced to consider both the learner and the teacher, together with the ongoing linguistic community. When Dufrenne lists the elements involved, on Sellars' view, in language learning, he is not referring to the position of the learner alone. Besides behavioural conditioning, says Dufrenne, Sellars employs notions of consciousness of objects, self-consciousness as agent, and a disposition to speak (which arises partly from the mastery of language) (CSP, 531). Let us look at the recent paper by Sellars, "Language as Thought and as Communication", to see how these and other notions relate to one another. We will also examine their relation to the 'all-at-once', rather than atomistic, notions of knowledge and learning which Sellars endorses in "Empiricism and the Philosophy of Mind".

B. "Language as Thought and as Communication"

Let me begin this summary by again noting a context-setting remark. Sellars thinks it is possible to give a naturalistic account of the intentionality of mental episodes, even though it would be arrived at by methodologically behaviouristic reasoning. He will try to accomplish this by showing, first, that not all rules pertaining to language are rules of action, and thus, a connection can be seen between conceptualization and verbalization; second, that there are at least three senses of the term 'express', and not all of them force one to assume either self-consciousness of an 'expressing' subject as an agent, or the existence of mental episodes as the things 'really' being expressed (LTC, 506).

For the purposes of technical philosophical discussion, "a rule is a general 'ought' statement" (LTC, 507). Besides 'hypothetical oughts', which deal with individual instances of practical reasoning, there are 'categorical oughts'. Two varieties of the categorical ought are considered by Sellars. The rule of action, or 'ought-to-do' (hence forth, 'a-rule'), demands for it to he properly exemplified, that the subject involved have the concepts necessary for the action and the concept of what it is to do something because of the rule and because of certain circumstances. The rule of being, or 'ought-to-be' (henceforth, 'b-rule'), demands only that the subject to which it applies conform to the rule, not that it do so consciously or under the guidance of concepts pertaining to the rule and to itself. But conscious subjects as agents are involved, since b-rules always imply a-rules. However, though the subject of the a-rule might be identical in type to the subject of the corresponding b-rule, i.e., a human conditioning another human, the subjects would not be identical tokens (LTC, 507-9).

While there may be a fashionable preoccupation with a-rules in language (due e.g. to Austin), a-rule subjects need concepts; and for Sellars, concepts involve b-rules. Therefore, the notion of b-rule is basic for understanding both the intersection of conceptual activity and language, and the rule-governed nature of language (LTC, 510). In other words, Sellars accepts (without argument in this essay) that the b-rule is the sort of rule involved in concepts; his innovation is to connect it to linguistic activity. The importance of the innovation is that once we have type-identity between rules for conceptual activity and rules for linguistic activity, a logical barrier has been broken down which could have prevented there being a close link (be that link what it may) between concept and word.

Lest the fact that b-rules imply a-rules return to a-rules the supposed logical priority which Sellars has just been trying to take from them, Sellars points out that both sorts of rules can be involved in any situation. For example, in a perception-situation, a subject could respond to a stimulus by merely uttering something; while another subject, employing the same words, could say something in light of certain conscious conceptual considerations. The latter, saying, is closer to what we think of as meaningful use of language. The first subject would be conforming to a b-rule; the second, obeying an a-rule. Both subjects can co-exist in the same situation, as the learner and teacher of language. Thus, because of the progression from learner to teacher, we need concepts to mediate between uttering and saying; and the notion of a 'self-perpetuating linguistic community' is conceptually necessary to deal with the context in which learner and teacher can co-exist and interact (LTC, 511-12). The full-fledged language user sees rules as rules and sees himself as an agent. As such, and as a teacher, a-rules are just as vital to his language-using situation as are the b-rules involved in conforming and in conceptualization (LTC, 513).

In the second and longer part of "Language as Thought and as Communication", Sellars claims that at least part of language which usually is taken to be communicative of a subject's thought, meaning, or intention, really ought to be understood as the subject's thinking in itself. Assimilating his analysis to that of expression/manifestation/actualization of dispositions in general, Sellars notes first that an episode of thinking is the 'primary manifestation' of a disposition to think. If we transfer the 'overtness' contained in the notion of manifestation to the thinking, we would say that expression is thinking -- a candid, spontaneous utterance by a language-knower is a 'thinking-out-loud'. These behaviouristic moves preclude our saying that here, a subject is consciously using the language; the thinking-out-loud is not to be seen as a consciously-used overt instrument or social act. In other words, not all meaningful linguistic activity is communication under the guise of the common-sense formula, 'We use language to communicate our thoughts'; some language is not communication but thought (cf. the title of the essay), is not talking or speech but just utterance, thinking-out-loud (LTC, 514-18).

Nevertheless, we shouldn't lose sight of this action or 'using' sense of language, which tends to be submerged by behaviouristic formulations like "actualization of dispositions". There are sayings which are consciously undertaken, and are not themselves complete thinkings-out-loud. This gives us a second sense of 'express' and we can add a third. The three senses of 'express' are:

(1) the 'causal', utterance-sense of express:
The subject's utterance of 'p' expresses his thought that-p;
(2) the 'action', saying sense of express:
The subject expresses his thought that-p by saying 'p' (this sense of express is open to many variations, taking into account the various intentions -- humorous, deceptive, etc. -- which the subject might have);
(3) the logical or semantical sense of express:
The subject's utterance of 'p' expresses the thought that-p. (LTC, 519-21)

Although Sellars does not mention this third sense again, a possible reason for its inclusion{4} is the following. Combining (1) and (3), we have a picture of the b-rule conforming language-learner. Combining (2) and (3), we have a picture of the a-rule obeying language-teacher{5} The element common to both their situations is (3), the semantical sense of express, the notion of meaning. This allows us to say that there is at least this one element common to the situations of both teacher and learner. I will consider later whether this can be the 'middle term' which Sellars is seeking between uttering and saying (LTC, 512).

Sellars next proposes a model of learning, such that a capacity to be in certain 'mental states', and the ability to employ corresponding linguistic expressions (in the logical or semantical sense), develop gradually and simultaneously. That is,

(a) A mental episode which is a thinking that-p is correlated, in a certain linguistic community, with a piece of linguistic behaviour which stands for (expresses in the logical or semantical sense) the thought (proposition) that-p;

(b) in the initial stages of the child's mastery of the language, whenever it has a thought that-p, this thought is manifested in a purely involuntary way by the corresponding verbal behaviour. (LTC, 521)

Sellars points out that (b) does not involve full language use; a response may be conditioned rather than instinctual without thereby being under voluntary control. But the fact that the child's "verbal behaviour . . . is the involuntary manifestation of (its) thinking" does not restrict the child's vocabulary; it can have "the language of intention and resolve as well as the language in which matters-of fact are stated", it could even "verbalize about verbal behaviour" (LTC, 521-22).

This analysis assimilates early language behaviour to the acquired, yet involuntary, 'cry', e.g., pain-behaviour. But we don't have to assume that the child cries out loud every time he is in pain; so too, we can allow that the child acquires the ability to be quiet without stopping thinking, that he would have voluntary control over whether to be silent or voluble. This does not change the logic of his situation -- the new control does not involve a-rules, and although the child can do thinking-out-loud sorts of actions (e.g., "deliberating out loud"), he does not do them as someone conscious of concepts and thoughts, and of himself as agent. We are still in the realm of utterance simpliciter, which is not a social, communication-context concept (LTC, 522-23).

It remains to see how Sellars' version of thought, expression and language-learning is different from others. The logical behaviourist would eliminate references to thought and mental episodes in the learning model; but he has difficulty with the action sense of express. The non-behaviourist would say that thinking-out-loud is really a combination of thinking and expressing, with the latter dependent on the former for meaningfulness (LTC, 524-25). As we can reconstruct it from other of his works, Sellars' alternative to these two is to retain mental episodes and the action sense of express, but not to treat the mental episodes as a source of meaningfulness.

Lest logical behaviourism be given too short shrift, Sellars warns us not to confuse it with logical physicalism -- and not to reduce it to mentalism either. He thinks that some of his present clarifications, especially about expression, would have helped in his correspondence with Chisholm. And he stresses the fundamentality of 'thinking-out-loud' among concepts pertaining to the mental, a fundamentality which he takes to be compatible "in the order of knowing" with the derivative status of concepts of mental acts or episodes (LTC, 525-27).

We can summarize this summary of the essay by listing some of those things characterized by b-rules opposite things characterized by a- rules:

learner subject conforms; uttering, parroting teacher subject is self-conscious as an agent and obeys; saying
chatterboxiness (which can be silenced without stopping thinking); i.e., 'thinking-out-loud', identity of conceptualwith overt linguistic activity use of language; consciousness of thinking as one thing and language (expression) as another

Where should the logical or semantical sense of express, the proposition, appear? (See above, p. 64 ). On both sides: both a conscious saying and a well-formed utterance will contain propositions. Thus, it is evident that the semantical or logical sense of express, which is common to the learner/uttering and teacher/doing situations, may be descriptively helpful in discerning a link between the left and right sides above. On the other hand, it is inadequate as an account of the transition between them. It defines their similarity; it does not define their difference-yet-connection. Rather, we need concepts for describing what it is to be at first a conditioned subject, what it is to be subsequently a self-conscious subject, and what is involved in the transition.

Before looking at other individual points, let us assess the difference this essay makes to the earlier correspondence with Chisholm. Does distinguishing "between the contexts 'person expresses' and 'utterance expresses' " (LTC, 526) solve some of their difficulties, as Sellars says it does?

Chisholm distilled his displeasure in his very last sentence:

If the people of your myth were to give just a little bit of thought to the semantical statements they make, wouldn't they then see that these semantical statements entail statements about the thoughts of the people whose language is being discussed? (IM, 537)

Our first task is to ascertain that Sellars' supposedly helpful distinctions do apply in this case. Some vocabulary just does not appear at the utterance stage, because it involves the "concept of saying 'p' without thinking that-p", e.g., deceptive telling (LTC, 522). However, Sellars seems to allow for semantical statements by the uttering child (LTC, 521-22: "verbalize about verbal behaviour"). Thus, Sellars' distinctions do apply here; we can conceive of the a people making and giving thought to semantical statements, in Chisholm's objection, as children who merely utter. We do not have to think of them only as adults who 'say'.

Suppose a child utters 's', some semantical-statement sentence. The child's uttering of it is an activity in which he is merely conforming to b-rules concerning vocabulary and meaning, and in which his uttering is an involuntary actualization of his disposition to think that-s (provided he is not now in a keeping-things-to-himself frame of mind) (LTC, 521-22). His uttering of 's' is itself his thinking-out-loud that-s (LTC, 522-23). Were the child to 'reflect' on this uttering -- i.e., were he to either volubly or silently verbalize about it -- he would not take it to be 'an expression of thought', since he does not yet have the concepts of expression and thought, especially not as separable items. What this 'reflection' is likely to remind the child of, is the context in which he learned the vocabulary for 's'. (This seems to follow from Sellars' principle, that b-rules imply a-rules; see below, p. 70 ff.) The child-learner, 'reflecting' on his speech, would think of his teacher. But here, too, Sellars would say that the child's 'reflection' would not necessarily appreciate the teacher's position as the teacher himself sees himself, that is, as a conceptually self-conscious agent. Rather, the teacher would be thought (-out-loud) by the child to be an item in the real world which remarks on certain patterns and connections in the world, and which is different from the child only in seeming to know more of these patterns and in having rewards and punishments to enforce attention.

In summary, if we specify children-people and utterance-statements, then "people" reflecting on their "semantical statements" will not necessarily be led to the idea of people's thoughts. And so, Sellars' distinctions between a-rule and b-rule, language-teacher and language-learner, do help us to resolve the perplexity which remained, in Chisholm's mind, at the end of the "Correspondence" (IM).

In this analysis of a child's 'reflection', Sellars is taken to be saying that the child's utterances express his thought in a causal sense akin to the way any instantiation i is, for a behaviourist, an expression of a disposition to do i. Thus, in the sense that his utterance is (an instance of) his thinking (out loud), "The child's verbal behaviour would express its thoughts, but the child could not express them" (LTC, 522).

On the other hand, once a person has a "conception of locutionary acts" (ibid.), he is a mature language-user who can 'express himself', can 'express his thoughts', and could certainly reflect on the thoughts of other people in the course of thinking about semantical statements. At this stage of language development, Chisholm is perfectly right (IM, 537); but he also is not in conflict with Sellars.

C. Rules and the Model of Learning

Talking about the child's 'reflection', I said it might follow the implication between b-rules and a-rules. What sort of implication is this? Or, more basically, what sort of rule is a b-rule?

Dufrenne criticizes this b-rule notion of rule or ought, whereby it is a rule that, for example, 'clocks ought to be on time', etc. Rather, Dufrenne proposes that we have here a defining condition, something which tells us what the essence of a thing is (CSP, 531). Secondly, he asks, what could be the "purpose of such a rule" of being (LTC, 508)?

I think that Sellars is speaking somewhat metaphorically, using 'rule' in a sense akin to pattern'. A child who is being conditioned and utters 'p' in the right place has allowed the 'purpose' of a 'rule of being' to be realized -- i.e., the pattern has been successfully achieved, and the utterance exemplified the rule.

Further, Sellars seems to be stacking his verbal deck. By speaking of 'rule' and 'purpose', which apply primarily to action contexts (CSP, 530), he is readying us for the formula 'b-rule implies a-rule'. It seems that Sellars wants our ready acceptance that one human action can cause another, and one a-rule implies another, to become an analogy for the idea that a non-action rule (b-rule) implies an action-rule (a-rule).

But surely, this latter situation is a more complicated one. Implication' is a logical term which can be used with varying degrees of strictness in various contexts. For example, one finds clearly defined and strict uses in mathematical logic. But in the present context, all Sellars says is that there is ' an essential connection" between b-rules and a-rules, "roughly, that ought-to-be's imply ought-to-do's" (LTC, 508). I cannot see anything intrinsic about an ought-to-be which would strictly imply an ought-to-do. In fact, patterns and conditions (b-rules) could exist in the world without any agents taking a-rule actions in relation to them. Further, it is unclear on what basis (if any) one could accuse such agents of 'sinning by omission'.

It is possible to make sense of this implication (of a-rules by b-rules) only by emphasizing, against Dufrenne and with Sellars, the 'ought' aspect of b-rules, at least as applied to the social context. Let us assume that Sellars is talking mainly about language here, and language as an artifact. Granted this -- that something is man-made and that an 'ought to be' does indeed apply to it in a certain way -- then human agency is implicated. This is easily seen in any instance of manufacture -- if cars 'ought to be' so and such, then some men have to tool up and run some machines. Thus, the implication between b-rules and a-rules is a 'social implication', at least where the b-rules have to do with artifacts. Taking it for granted that language is an artifact, it follows that implications hold between b-rules related to language, ruling word-word and word-world associations (LTC, 513), and certain a-rules.

At first glance, lack of further explanation by Sellars of this implication need not impede our understanding. B-rules and a-rules are continually involved in the analysis of the learning situation; but the implication between them does not seem very important to Sellars. The crucial point here is that we have to assimilate b-rules to social contexts in order to understand some of the things Sellars says about b-rules.

How does Sellars analyse the learning situation? At one point he claimed that there is no "awareness of logical space prior to, or independent of, the acquisition of a language" (EPM, 162). This 'co- emergence' of language along with cognitive thought ("awareness of logical space") is supported by his model of language-learning. Before being taught to speak, Sellars says that we can assume the learner to have the power of pre-verbal and pre-conceptual thinking "recognitional capacities" and so on (cf. the rat, LTC, 509). This would be in activation in the learning situation. For example, when the teacher tries to induce the vocalization 'Red', the learner is reacting pre-verbally and pre-conceptually to the red-coloured stimulus in that he 'recognizes' it, pays attention, etc. His subsequent verbalization provides not only a 'linguistic bit', but also the makings of a concept. The child's uttering 'Red' is both a linguistic and a conceptual event, conceptual because its speaking is a thinking-out-loud.

In this way, Sellars' learning model acknowledges factors which existed prior to the learning situation (like "recognitional capacities"), and the factors involved in that situation as a whole. It does not accept as its paradigm the simple pointing at a thing presumed to be appreciated correctly as it is by the learner, who by repeating what he is told and looking at what is pointed out, achieves learning by ostension and association.

When Sellars denies that meaning is a shorthand for talking about sets of associations (EPM, 163), we should take him to mean association as involved in learning by ostension. For if his own model of language learning is truly that, a model of language learning, then surely the words learned by the child 'have meaning'. And although Sellars does not say what the teacher actually does, we can surmise that what is involved is some sort of conditioning whereby the 'role' or 'use' of words is exhibited along with the vocables themselves, inviting associative learning by the child.

This associative learning situation is the proper locus for thinking-out-loud as "a form of meaningful speech which doesn't consist in talking to anyone at all, even oneself, and hence is not, in any ordinary sense, talking" (LTC, 518). We only have to think of how different the pedagogical question-and-answer is from questions and answers in ordinary communication. The teacher's question is not really a question, because he already knows the answer, lexically prescribed;and the student's response is not really an answer but the exhibition and repetition of a prescribed formula. Let us put this 'rote learning' into Sellars' framework: the teacher (i) provides a cue or stimulus, and (ii) demands that the learner relinquish any 'keeping-things-to-himself frame of mind'; the learner (iii) thinks-out-loud whatever response the stimulus has elicited. (Of course, this is the latter phase of a learning sequence; the former would be the 'exhibition' of the intended connections.) The crucial fact about this phase of learning is that it does not involve the learner's talking to anyone -- not to himself, not to the teacher. He is merely uttering.

Now, merely uttering is involuntary, habitual (LTC, 521), and early language learning is 'learning to utter'. So, early language learning could then be seen as merely a period of habit-formation. While this position is quite attractive, it is complicated by Sellars' response to a rival model of learning. Those who endorse learning by ostension would have us think that noticing things leads to our having concepts of them; but really, says Sellars, one needs the concepts in order to notice the things (EPM, 175-76). This response represents a complication because, as we saw, Sellars' model of learning does acknowledge factors existing before the learning situation. Some of these, like recognition, noticing, and so on, are normally analyzed (and analyzed by Sellars) as depending on having concepts. But if one has concepts as a result of learning, and recognition and noticing (etc.) depend on concepts, how can learning depend on recognition, noticing and so on?

The way out of this apparent circularity is to introduce 'liberalized' notions of innate or of developing pre-conceptual recognitional capacities -- that is, a notion of habit which does not itself imply the possession of concepts, but can lead up to a state in which the subject does have concepts. I think Sellars gives us an analogue of such a notion when he shows that behaviour can be learned but still be as involuntary as an instinct (LTC, 521).{6} He claims that there is a continuity between 'natural cries' and acquired habits of linguistic behaviour, acquired 'utterings'.

The parallel point for 'recognition', 'noticing', 'attention' , etc. is this. It is legitimate to speak of an unlearned, non-conceptual, involuntary precursor (the 'cry') of a learned, conceptual, involuntary behaviour (thinking-out-loud or uttering). The cry of pain and the pain-utterance have in common not only involuntariness, but also a certain relation to the normal characteristics, the 'nature', of the human subject. So too, we can speak of something like recognition as more than just conceptual or concept dependent activity. It is a capacity or power which exists preconceptually in a less complex form. That is, pre-conceptual recognition is to concept-dependent recognition as the natural cry is to the utterance.

As I interpret Sellars' learning model, he has to permit this 'liberalized' notion of recognition (noticing, etc.), which I have explicated on the model of the natural cry. But, as I hope to show in the next section, Sellars must withdraw the permission when asked to integrate his positions on meaning and behaviourism with this learning model.

D. Meaning, Behaviourism, and Language-Learning

I begin to diverge from Sellars' analysis of learning on the status of the learner as a social being. More generally, I shall be giving an anti-Sellarsian detailed characterization of the order or level of meaningfulness in the world, which exists in virtue of the existence of social groups (cf. Chapter III, Section B).

Sellars assigns the language learner a rather peculiar position in society: "the members of a linguistic community are first language learners and only potentially 'people' " (LTC, 512). He is giving the learner at least the status of a member who has his membership by virtue of being taught a language in a particular (spatio-temporal) community. But, for epistemological and language-learning reasons, this minimal, pedagogical-geographical analysis of the learner as social being, is really too minimal.

We are led to this realization by our having to reconcile the holism of conceptual thinking with the atomism of learning. I call conceptual thinking holistic because Sellars claims that one needs all of (a rudimentary) language before he really has any part of it, one needs all{7} the concepts in a given field (e.g., observation) before one can really have any one concept; and yet, I call learning atomistic because the child learns responses (words or phrases) individually, in individual learning situations (BBK, 49-50). This seems to imply that we need some sort of pre-conceptual, holistic 'glue' to hold together the learned linguistic bits until there are enough there for a complete rudimentary set, for a usable language. Other learning theories do not have this problem. For naive behaviourism, there is no difference between saying that a subject has learnt the correct response R by ostention and saying that he has the concept R. Thus, there is neither temporal nor logical gap between learning and 'concept-having'. Curiously, learning by abstraction, despite its different ontological commitments, also sees no temporal nor logical gap between learning (by abstraction or 'extraction') and 'concept-having'. As opposed to these two theories, which Sellars explicitly rejects (EPM, 179; BBK, 43), Sellars' learning-model demands that there be some 'glue' which preserves the sub-conceptual utterance-patterns through the temporal gap and logical shift to the stage of being part of a language and of a 'battery of concepts'.

We can call 'membership in the community' the 'glue' if we construe such membership as a child's acquiring roles and relations vis-à-vis his environment, including other human beings. To take an example: when the child is taught 'Red', why does he remember it? I think it is because he has already eaten tomatoes, yanked the red curtains, and torn the dust covers off books published by Routledge and Kegan Paul. (Of course, it is quite difficult to decide on a proper epistemological description, since we are talking about preverbal experience here, whereas epistemology's overwhelming preoccupation has been with conceptual and language-using thought.) Somehow, Sellars should account for the relationship of pre-verbal thinking (believing, desiring, etc., which exist before conceptual thought) to the acquisition of language (IM, 524, 525).

Perhaps a pictorial image would help. If the individual words and phrases learned are atomistic and unconnected, yet in virtue of the eventual structure they are 'appropriate' to one another, then we can think of them as different shaped stones which are being prepared for making a stone wall. The builder assumes that they are individual entities and unconnected to one another, yet 'appropriate'. However, there is also the assumption of a continuous piece of solid ground on which to build; the stones, once placed on the ground, will not change their position vis-à-vis each other, in virtue of this foundation. The ground is nothing at all like a wall; yet it is a sine qua non for the wall, and in its continuity and unchanging solidity it somehow anticipates the wall which will stand on it.

My point concerning language-learning is that we have to assume some sort of ground or foundation to this phenomenon of language learning, in virtue of which the atomistic, individually-learned bits of linguistic behaviour are not lost or dispersed despite their lacking that which (according to Sellars) eventually holds them together, the language as a useable whole. This ground or foundation would be the thinking, believing, desiring, and so on, of which children are capable before they learn to speak. Such thinking (etc.) would involve experience with interpersonal roles, rules and relations, and with the manipulability and other properties of ordinary physical things. In fact, these are the pre-conceptual counterparts of those Rylean facts which are implied but not asserted by the meaning- relation (IM, 536). And so, we can go one step further with the pictorial image, and say that this ground of thinking thereby anticipates the completed linguistic edifice.

Perhaps Sellars has himself hinted at such an analysis. He agrees that there could be thinking, of a primitive nature, before speaking (IM, 526). And he speaks of " 'recognitional capacity' " and the ability "to respond differentially" in connection with the conditioning of rats (LTC, 509). But both these latter concepts are extremely vague (not just one of them, as Sellars would have it). In training rats, the ability to respond differentially is defined as the disposition to make a statistically significant number of correct responses (turns in a maze, depressions of levers, etc.). Yet this doesn't say what the disposition is; for the disposition would be said to have ceased to exist as soon as the test scores altered significantly.{8} What is the disposition in itself? What, in itself, is the ability to respond differentially? If Sellars says that this is a neurophysiological rather than a conceptual issue, then he is referring ultimately, in the course of conceptual explanation, to the substantive category 'physicalistic.' Since this would be contrary to his programme as a whole, I take it that Sellars' alternative would have to be that there is a way of looking at thinking (believing, desiring, etc.), at the level of 'recognitional capacity and differential response', which reflects the eventual meaningfulness of full language activity, yet is neither entirely physicalistic nor mentalistic in the classical sense. I see no alternative but to propose that the child's thinking and other activity is, already at this stage, meaningful.

It is important to see that we reach this conclusion by following Sellars as far as possible, not by rejecting one of his premises or early arguments. We accept Sellars' 'battery of concepts', his notion of knowledge as a self-correcting enterprise rather than a collection of self-authenticating episodes. Accepting, further, that individual utterances are learned at different times, we need something to 'hold them together'. And this has to be something more than, simply, 'memory', because we also accept that these utterances are sub-conceptual rather than conceptual. The proposed conclusion is that the child's thinking is meaningful before he has learned to speak, although he lacks that awareness of it which depends on having semantical concepts -- that is, his thinking is not explicitly meaningful to him, but we can explicitly ascribe meaningfulness 'from outside'. And because this 'meaning-ground' is a factor which exists prior to the learning situation, we can say that the utterances of the early language-learner are also 'meaningful' (as does Sellars but for different reasons and with a different interpretation on 'meaning'). Again, explicit awareness is not involved, but rather, an ascribability of meaningfulness 'from outside' -- just as we could say that the language of the Jones-myth Ryleans is meaningful, without their being aware of it.

How does this thesis of the meaningfulness of 'pre-speech' thinking relate to Sellars' thoughts on meaning and language? Basically, it conflicts with his ontological thesis, that there is 'meaningfulness in the world' in virtue of the existence of language. It seems now that the paradigm for meaningful behaviour is not overt speech but something prior and more extensive: activity in the social life-world of the individual member in a linguistic community. Although, by this sort of definition, 'language' creeps back into that which is fundamental for meaningfulness (via the term "linguistic community"), I doubt that the actual possession of linguistic ability is as crucial as Sellars claims. This doubt is the basis for my second and third criticisms of Sellars. I have just claimed that the detailed development of Sellars' learning-model (in the context of other Sellarsian positions) puts a language-based concept of meaning in conflict with a needed 'pre-language' application. My next criticism is: teaching as analyzed by Sellars is based on a-rules; but a-rules seem to presuppose semantical discourse; and if so, there could not be language-teaching before there is semantical discourse. Finally, I shall try to show that Sellars' concepts of meaning and linguistic activity conflict with his outline of what a behaviouristic conception of language would be.

To decide how crucial the actual possession of language is to the existence of meaningfulness, let us look at the relationship of meaning and teaching.

If we follow Sellars closely, we learn that the questions of language teaching and of the existence of meaning-discourse (semantical discourse) are not connected at all. First of all, if we believe his 'myth' about the Ryleans, we can have an ongoing, self-perpetuating -- and ipso facto language-teaching -- linguistic community which, according to the myth, would lack semantical discourse. Therefore, not having semantical discourse cannot be an impediment to language teaching. My own historical reconstruction (Homer and Thales, page 37 above) agreed with this conclusion, as it implied that it must have been possible to teach language without having semantical discourse.

Thus, it must be that the existence of semantical discourse is unnecessary for there to be activity governed by a-rules -- for example, teaching (LTC, 512). (Of course, the existence of mentalistic discourse would also have to be unnecessary for teaching, since it is an outgrowth of semantical discourse.) It is this notion which I find rather difficult to accept.

The criticism I am about to make, by the way, need not conflict with accepting the Ryle-Jones myth; for we can take it to be only metaphorically a story about social-historical time. On the one - hand, taking the myth quite literally, we would expect there to be an 'era' of society or societies which possessed Rylean language exclusively; then an era of societies having Rylese enriched with semantical discourse; and finally, the era from Jones to the present, which possesses all this plus 'mentalese'. Interpreting the myth this way, we obviously cannot allow the language-teacher of any era to possess language or concepts whose logic belongs to a later era; this would be an anachronism. But on the other hand, we could say that the myth is literal only for the 'time', eras, or ages of the individual; that the myth as metaphor sets out as a chronological sequence what should be understood as a logical and conceptual sequence. For example, it could be said that chronologically, the second and third eras might appear simultaneously in the individual's development. Sellars' point is that even here, our logical and conceptual analysis would affirm that there is a logical, if not empirical, precedence of one 'era' over the other.

The notion which I find difficult to accept is that there could ever be any situation governed by a-rules (e.g., teaching) which did not also involve at least semantical discourse. If the notion of 'a-rules without semantical discourse' is conceptually respectable, then, taking the myth literally, we can postulate an ongoing, Rylese-speaking social community, because the Ryleans could teach their language even though not yet having semantical discourse. According to the non-literal interpretation of the myth, the thesis in question implies that we can accept a methodologically behaviouristic account of child behaviour as its first account (chronologically and especially logically), the addition to which of descriptive language employing semantical discourse is a unique, separate, and later (in both senses) matter. (I take this to be the import of relating the analysis of rules (LTC) to Sellars' myth (EPM): that Sellars is affirming the autonomy and fundamentality of a methodologically behaviouristic description of human behaviour.)

If I can make my objection stick, its force would be, in terms of the myth, that it just does not make sense to speak of an ongoing, Rylese-speaking community. The reason is that for there to be an ongoing community there must be language-teachers, and language teachers need a-rule concepts; yet (I hope to show) having a-rule concepts presupposes having the very semantical discourse which does not (yet) exist in Rylean communities. In terms which do not take the myth so literally, my objection implies that the logical-space awareness of the language-learner is not an awareness which can be described behaviouristically. Furthermore, it implies that the matter of description employing semantical (meaning) discourse is not really so separate and unique, even if possessing semantical theory and vocabulary is unique and separate from possessing other language. There is still much with which I agree -- I would like to grant that the 'myth of the given' is a myth; I would like to agree with Sellars on the illusions of 'immediate experience', 'epistemic objects' like sense data, 'introspection' on the classical model, and so on. Nevertheless, I wish to contest one of his steps which leads to these agreeable conclusions.

My objection is really rather simply put. Sellars specifies that the subject of an a-rule does action A because he perceives himself to be in circumstances C. Thus, besides having the concepts of being in C and doing A, he must have the concept of an action being called for by certain circumstances (and of course, he would have to know the rule for this particular situation, "If one is in C, one ought to do A"; LTC, 507-08). Now, Sellars does not go further into what it is to have the concept of an action being called for by certain circumstances. Staying within his framework, 'action understanding' is analogous to 'semantical understanding'. That is, to understand the relationship of actions to surrounding and antecedent circumstances is comparable to understanding the meaningfulness of words. Both deal with the role, use, or function of an item in a surrounding interpersonal economy, social and linguistic.

I already noted how the 'meaning-relation' is both unique or primitive, and related to antecedent and simultaneous contexts (above, Chapter III, Section A). Both these features are captured in the formulation 'A meaning statement does not itself describe, although it implies a description of, "Rylean facts about linguistic expressions"'(IM, 536). So too, an a-rule action statement is not itself a description of (e.g.) a typical motion and its spatio-temporal environment ('circumstances' considered behaviouristically), for it is somehow more than this. Nevertheless, it is essential to the logic of such a statement that it implies such descriptions, and can be analysed in these terms (the 'analysis' of concept-families and clue-giving). Otherwise, it could not be the rule for a particular action; it could not give instructions.

A careful continuation of this analysis shows that, following Sellars, there are two ways in which we must go beyond just assimilating 'action understanding' to 'semantical understanding'. In the first place, this thesis has been highlighting the notions of 'pre-verbal' thinking and, if concepts are "mental words", of 'pre-conceptual' understanding. But for Sellars, understanding and thinking are bound up with concepts and speech. If we speak of action understanding, then we are speaking implicitly of action-concepts and discourse as well. In other words, if we credit the agent with an 'action understanding' analogous to 'semantical understanding', then to follow Sellars, we expect him to possess action concepts and discourse analogous to semantical discourse. Sellars' own formulation emphasizes this: the agent must have the concept of action being called for by circumstances (I.TC, 508).

I specified the notion which I find difficult to accept as the possibility of there being any situation governed by a-rules which did not also involve at least semantical discourse. The second way of going beyond my initial assimilation of the action and meaning contexts justifies this 'at least'. Sellars also demands that if one is a full- fledged language user, a teacher, "one must . . . have the concept of oneself as an agent, as not only the subject-matter subject of ought-to-be's but the agent subject of ought-to-do's" (LTC, 513). I find it prima facie clear that in having a self-concept as agent, one has a concept which is mentalistic in structure. I am not arguing that a Thomistic or a Cartesian view of the self is necessarily correct; the eventual philosophical elaboration of a minimal self-concept as agent is not the issue here. Rather, I am accepting Sellars' claim that the core or structure of classical mentalistic theory was of a somewhat mysterious inner episode (thought) with the power of being a source of intelligence and meaningfulness (EPM, 177, 186). So too, the core or structure of agency, the minimal self-concept as agent, involves a somewhat mysterious power centre, a source of action as such. Because of this, action-statements cannot be fully analyzed into descriptions of extensional, Rylean facts of motions and of sequences of events.

Of course, Sellars agrees that eventually we have mentalistic concepts; but we have them only after having acquired semantical discourse. Yet if this is so, and if the self-concept as agent is a structurally mentalistic concept coming after semantical discourse, then there is a logical (or chronological) inconsistency in the myth. Sellars says that there is an ongoing Rylean-language (pre- semantical) community. But we cannot assume that language is taught by agents in a Rylean community, because teaching is post-semantical: it involves a mentalistic self-concept as agent, which (being mentalistic) is a post-semantical concept. Therefore, in this sense, there cannot be a Rylean community; or, the analysis of teaching is wrong.

The less literal interpretation of the myth conflicts, I feel, with the b-rule side of the teacher-learning situation. The import of this interpretation was said to be the possibility of a methodologically behaviouristic description as the logically and chronologically first account of human behaviour. Sellars' language-learning model (LTC, 521 and passim.) would imply that this first stage extends through the time when the child is an 'utterer'. But I will try to show that this concept, 'utterance', itself presupposes a description in terms of 'meaningfulness'; thus, the methodologically behaviouristic account will have to be dropped for one which includes some sort of 'meaning discourse'.{9}

What is an utterance? Sellars says that utterances are response uniformities, the result of training. As such they are mere 'parrotings'. Nevertheless, they are also linguistic entities, for it is "Linguistic ought-to-be's (which) are translated into uniformities by training" (LTC, 512). In their turn, linguistic entities are defined by Sellars as being 'more than' physical entities: "Marks and noises are, in a primary sense, linguistic expressions only as 'nonparrontingly' produced by a language-using animal"; "the primary mode of existence of a language is in meaningful verbal performances" (IM, 525). This would imply that one cannot undertake any description of entitites, as pertaining to language or the linguistic realm, without presupposing 'non parroting' and 'meaningful'.{10}

This, then, is the objection. First, in the interests of behaviouristic methodology, Sellars says that utterance is really just 'parroting'. Next, in the interests of linguistics, Sellars claims that utterance is really a linguistic entity and meaningful. But, third, Sellars opposes 'parroting' and 'meaningful'. Thus, if some person or methodology recognizes an entity as nonparrontingly produced, then he or it presupposes and employs some notion or criterion of meaningfulness. Recalling the definition of a methodologically behaviouristic description, this methodology can be behaviouristic only in a queer, modified, and very un-behaviouristic sense. I am not putting forward a positive account of the meaning and criteria of 'nonparroting'; I wish only to say that it is inter-defined with something (semantical or mentalistic) whose logic belongs to the post-Rylean eras of Sellars' myth. Therefore, it cannot be presupposed in the logic of behaviouristic description, if such behaviouristic is to adequately describe the Rylean era.

It seems, then, that a description which is behaviouristic in methodology cannot be adequate to accounting for even the primary level of linguistic entities; for the recognition of entities as linguistic involves a concept hidden in the methodology (or a capacity or power unreflectively employed by the behaviourist) which a behaviouristic methodology cannot employ.

I would like to end this examination of meaning, behaviourism and language-learning with one more criticism of Sellars' behaviourism. Sellars himself points out the need for something between utter and say (LTC, 512). Another way of putting it is to ask how the individual passes from the involuntary, chatterboxy 'thinking-out-loud' to the conscious use of language. We need to know not only 'what concepts are involved' (Sellars' preoccupation) but also 'where does this come from'. How do we get from 'conforming' to 'obeying', i.e., to consciousness of rules as rules?

One concept which Sellars offers us for this transition is that of 'acquiring' (LTC, 522) -- "We can suppose (the child) to acquire the ability to keep its thoughts to itself in the sense that it can effectively tell itself to keep quiet, without ceasing to think". Now if we were satisfied with vague answers like 'acquire the ability', we could solve all problems of knowledge, language and mind by saying that people acquire the abilities to do what they in fact do. This obviously is just obscuring the situation: 'acquire' is neither a mentalistic term, nor is it behaviouristic. It is a promissory note for a future analysis. So too, the concept 'thinking-out-loud': this is supposed to be the keystone of a behaviouristic analysis. If so, it should be arrived at behaviouristically. In that case, it could not contain the term thinking (with all the shortcuts concealed therein); nor should it even refer to out -- if a behaviouristic methodology does not believe in 'inner' processes, then there is no meaningful contrast which justifies the use of 'out'. Thus, the keystone concept of behaviouristic methodology vis-à-vis thinking, if arrived at in a purely behaviouristic fashion, cannot be 'thinking-out-loud' nor 'out- louding', but 'louding'.

[Return to Table of Contents]


Rather than re-establish the account of meaning, mind and language-learning based on a new behaviourism of 'louding' instead of 'thinking-out-loud' (see above, p. 88-89), I would prefer to anticipate a quite different account. First, though, I will recapitulate some of my objections to Sellars.

There is a great deal in Sellars' work with which I strongly agree. I support his arguments against the sense-datum theory and the 'myth of the given' (p. 7 ff.). These arguments help destroy the 'foundations'-conception of empirical knowledge -- the idea that rationality of knowledge is due to individual self-authenticating epistemic episodes, not verification procedures in the enterprise as a whole (p. 57 ff.) -- and again, I agree. Furthermore, I would endorse the wide set of arguments which points to (in my words) an 'ontological level of meaningfulness' (p. 42 ff.).

Thus, my objections to Sellars do not constitute a rejection of the overall programme of construing 'meaning' in social, real-world terms. Rather, I am concerned that some of Sellars' principles and assumptions prove embarrassing in the face of other demands. For example, problems arise because of Sellars' principle of 'no noticing without concepts' (EPM, 176). It is in direct conflict with his own 'Jones-myth'. In the 'Jones-myth', a transition is possible from observational language through semantical discourse to mentalistic language, only if the meaningfulness of overt speech and the intelligence of silent behaviour can be noticed, in some sense, before people have the concepts of meaning and intelligence (pp. 19-28).

This inconsistency implies an objection to Sellars' overall programme: while we may agree with his critique of traditional 'givenness' and Cartesian introspection, we cannot agree entirely with his positive thesis that the intentionality-categories of thought derive from the semantical categories of speech. The reason is that thinking must 'notice' intentional items (meaningfulness, intelligence) before it has the verbal-conceptual resources to do so. Thus, there is an aspect of the logic of the intentionality of thought which cannot be linked to concepts on a verbal model, and which, therefore, is not akin to the logic of the intentionality of speech. Without embracing dualism, we must allow intentionality of thinking which is not based on intentionality of discourse, because a subject can think about something before he has the related explicit, verbal-model conceptual means at his disposal.

The previous chapter ended with three main criticisms of Sellars. The first, concerning 'meaning' in an epistemological context (pp. 76-81), is a stronger version of the above objection. It states that we must credit the subject with 'awareness of meaning' before he has any verbal-conceptual resources.

I have already stated the principle which this violates. But I think that an underlying assumption is also being challenged. Although he implicitly agrees that 'thinking' can take place without verbal resources (IM, 527), Sellars tends to think of all thinking in terms of analogy to meaningful overt speech (BBK, 43-44). However, my objection shows, on the negative side, that neglecting the issues of awareness and meaning in pre-verbal behaviour in favor of starting with language behaviour, leads to inadequacies in the analysis of language behaviour.

Furthermore, the fact that certain behaviour is pre-verbal should not imply that it is beyond the pale of science. To say so would eliminate all science of animal behaviour. Moreover, to say so is peculiarly anti-behaviouristic: it implies that there cannot be scientific psychology without first-person verbal reports, i.e., without introspection.

On the contrary, my objection shows, more positively, that preverbal behaviour can be described by philosophy and science in terms of roles, relations, physical manipulations, etc. (p. 76 ff.) (In Piagetian terms, it can be described in terms of concrete thought rather than verbal, abstract, or logical thought.)

In the Introduction I argued that Sellars is attempting a 'presuppositionless' philosophy -- one which starts with what is most generally and untheoretically given, the language-impregnated social world, rather than with metaphysical categories or philosophical theories (p. 2). The assumption which my thesis challenges lies in the description of the given world. Certainly, to call it social in the sense of language-impregnated serves to distinguish the primordial, original world of humans from that of other species (some zoologists would disagree, but that does not affect my point). But Sellars fails to see that the distinguishing mark can be (and often is) narrower than the full definition. Starting with a Rylean-language community, Sellars analyzes thought and language in each others' terms. Against this, I would propose that the development of thought and language should be construed in terms of 'the world' and 'being-in-the-world' (cf. Heidegger's Dasein), not 'the language-using social world' and 'language-learning'. 'The world' includes Sellars' primordial 'given' but it is sufficiently broader to include pre-verbal and non-verbal thinking.

Similar points can be made on the basis of the second and third objections of the previous section (concerning action and language, pp. 81-88); but it would be better to end this concluding chapter with a clarification of the positive account (above) as it affects education. First, regarding the rivalry over psychological explanation of intellectual growth (pp. 4-5): this thesis represents a challenge to one of the rivals, behaviourism. The strength of the challenge is far more to Sellars' credit than to this author's. The thesis that mentalistic discourse is a theoretical discourse is not unusual in the behaviouristic and physicalistic literature. However, Sellars' version is unique in that he attempts to work out the details of development of language and thought in general, within which a theoretical mentalism arises disguised as immediate experience. Thus, even a small blow to Sellars is a large blow to behaviourism, and to the possibility of an adequate behaviouristic explication of intellectual growth and early language learning.

Second, this thesis shows that we can take many cues from Sellars in explaining the teaching-learning complex (p. 5). For example, we can follow him on distinguishing the self-awareness of the learner from that of the teacher; and we can use a difference in types of rules to do so. However, my thesis suggests that Sellars' ideas on action and linguistic entity conflict with his overall behaviouristic framework. Thus , for these further reasons, it is again implied that something broader than Sellars' methodological behaviourism is needed in order to deal with intellectual growth and the learning situation.

[Return to Table of Contents]


Chisholm, Roderick, and Sellars, Wilfrid,
"Intentionality and the Mental", in Minnesota Studies in the Philosophy of Science. Vol. II. H. Feigl, M. Scriven and G. Maxwell, ed. Minneapolis: University of Minnesota, 1958.

Dufrenne, Mikel,
"Comment on Wilfrid Sellars' Paper", Philosophy and Phenomonological Research. XXIX, 1969.

Sellars, Wilfrid,
"Being and Being Known", Science, Perception and Reality. London: Routledge and Kegan Paul, 1963.

Sellars, Wilfrid,
"Empiricism and the Philosophy of Mind", ibid. (Originally in Minnesota Studies in the Philosophy of Science. Vol. I, 1956.)

Sellars, Wilfrid,
"Language as Thought and as Communication", Philosophy and Phenomenological Research. XXIX, 1969.

Sellars, Wilfrid,
Science and Metaphysics. London: Routledge and Kegan Paul, 1968.

[Return to Main Page][Table of Contents]


{1} Thinking back to Descartes, we see why the ostension-illusion, linking commonsense mental talk with alleged entities, centres particularly on 'the mental'. In the realm of the mental, both the act of cognition and its object are 'inner', mental; thus, if our idea is 'clear and distinct', what possible reason could we have to doubt that our thought is in possession of the true reality of its object? The ultimate effect of the above distinction, and of Sellars' account of 'rule-following observation reports' (EPM, 166), is to allow a split between the content of our self-observation reports and the reality in propria persona of our mental life -- he accomplishes a removal of what we consciously assert (mental) things to be' from 'the way these things are in themselves'. [Back]

{2} 'Behaviourese' is a language whose basic vocabulary refers to observable properties of public objects, and all of whose vocabulary is constructed out of (and is reducible to) this basic vocabulary. 'Behaviourese' and 'Rylese' are synonyms. [Back]

{3} It is interesting that, in effect, the traditional behaviourist and the traditional mentalist are united against Sellars on this point. The mentalist proceeds from the character of 'the mental' to the legitimacy and character of 'meaning'. The behaviourist claims that 'the mental' is a superfluous and illusion-filled category, and proceeds from this to the superfluity of a 'meaning' category. Sellars is different from both in holding 'the mental' in abeyance and in claiming that the character of 'meaning' talk is a legitimate starting-point. [Back]

{3a} Whenever we speak of 'all the concepts' or 'battery of concepts', we understand a minimum logical range to be implied, not the maximum possible empirical range. For instance, to understand a colour word as part of one's language, one needs it plus concepts of standard conditions, etc. One does not, and could not, need to know all the colour names in order to understand any one of them. [Back]

{4} Another reason could be completeness, but incompleteness does not seem to worry Sellars. For example, he professes to have considered only some of the varieties of categorical ought statements, but he doesn't even mention those other varieties. [Back]

{5} The language-teacher is just one sort of mature, rule-obeying language-user, but the sort of greatest interest in this thesis. [Back]

{6} Sellars' idea is, roughly, that not all involuntary responses need be instinctual responses; some can be learned or conditioned without thereby necessarily being under 'voluntary control'. Instinctual involuntary responses and learned-involuntary responses differ only regarding their longstandingness and their origin. [Back]

{7} Whenever we speak of 'all the concepts' or 'battery of concepts', we understand a minimum logical range to be implied, not the maximum possible empirical range. For instance, to understand a colour word as part of one's language, one needs it plus concepts of standard conditions, etc. One does not, and could not, need to know all the colour names in order to understand any one of them. [Back]

{8} This means that the term 'disposition' is a dispensable shorthand for 'a certain pattern of scores'. Because it is redundant, it can be thrown out -- which leaves us with scores describing movements, but no interpretation, no theory. To maintain a science here, we must ask the interpretive, theoretical questions. [Back]

{9} A methodologically behaviouristic description involves no 'semantical understanding' and no concept of meaning. All concepts which it introduces are to "be introduced in terms of a basic vocabulary pertaining to overt behaviour" (EPM, 185). [Back]

{10} Perhaps it could be countered that 'nonparroting' and 'meaningful' are presupposed only by the main, or paradigmatic, i.e., 'primary' uses of the terms 'linguistic' and 'language'. On the contrary, Sellars affirms that one should not restrict 'meaningfulness' to full-fledged language-use alone -- it is unnecessary to say "that all meaningful linguistic expressions are actions in the conduct sense, and all linguistic rules, rules for doing" (LTC, 513). [Back]