SELLARS ON THINKING AND THE MYTH OF THE GIVEN

CHARLES ECHELBARGER

SUNY at Oswego

Published in Philosophical Studies 25 (1974) 231-246.

      Many have thought that knowledge has a dimension which is 'basic' or 'given' in all of the following senses:

  1. knowledge at this level is non-inferential, not arrived at by reasoning from prior evidence;
  2. 'less basic' or 'higher' knowledge presupposes knowledge at the 'given' level, but the latter itself presupposes no prior knowledge as a condition of one's acquiring it;
  3. it is of facts (or entities of whatever kind) about which no mistake is conceivable.
Sellars believes that this idea is a 'myth', that nearly all philosophers have endorsed some version of it, and that "without flying in the face of reason, one can in general deny the entire framework of giveness". My aims in this paper are as follows:
  1. To show how Sellars' rejection of the Given leads him to the view that the concept of a thought-episode enters our conceptual scheme as a 'theoretical' concept.
  2. To argue that, if the concept of a thought-episode were introduced for theoretical purposes in Sellars' sense, the warranty for accepting and applying this concept would be disastrously small.
  3. I conclude that the concept of a thought-episode of which we do make use cannot be adequately grounded in purely theoretical style.
  4. I suggest some emendations which, though they incorporate a modified view of 'the Given', acknowledge the essential soundness of Sellars' critique of this notion.

     An example is necessary to illustrate the idea of Givenness. What is minimally necessary for one to know that there is such a quality as red? A typical partisan of the Given will say that if item a is red and S is a conscious subject presented with a, these conditions are epistemically sufficient for S to know that there is such a quality (and perhaps that something has it). Questions concerning S's ignorance about the lighting conditions and his own visual equipment may move the partisan of givenness to modify his claim to the effect that S can at least know that a looks red, and, therefore, that there is such a quality which a looks to have. Sellars' uncompromising reply is: even the knowledge that something looks red presupposes nothing less than knowing what it is for something to be red. A mind which does not possess the concept of something's actually being red could not frame any true judgement containing the term 'red'. "... being red", Sellars argues, "is logically prior, is a logically simpler notion than looking red." ([4], p. 142) "... the concept of looking green, the ability to recognize that something looks green, presupposes the concept of being green". ((4), p. 146). Sellars claims that, far from presupposing no prior concepts, one's taking some thing to have a property presupposes 'whole batteries' of conceptual abilities for identifying features of public objects. Knowing even as simple a fact as a is red "involves a long history of acquiring piecemeal habits of response to various objects in various circumstances..." ([4], p. 148). Perceptual takings are, as Quine might say, theory-laden.

I am able to 'see at a glance' that something is red only because I have a conceptual picture of myself as being in a situation consisting of such and such objects thusly located in Space and Time, a picture which I am constantly checking and revising, a picture any part of which and any principle of which can be put into jeopardy, but which cannot be put into jeopardy all at once. ([4], p. 339).

      According to Sellars, the basic logical confusion behind the Myth of the Given is that

... instead of coming to have a concept of something because we have noticed that sort of thing, 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. ([4], p. 176).
In his correspondence with Chisholm, Sellars asserts that there is a connection between his rejection of the Given and his (Sellars') philosophy of mind.1 I shall attempt to exhibit this connection. The partisan of the Given says that presentation of a red entity to a conscious subject is sufficient for S to acquire the concept red. Consider a similar approach to such a 'quality of mind' as thinking (that-p). A traditional cartesian will say that, if one is a thinking being, one can't help knowing that thinking occurs. One's own thinking cannot go on unnoticed. Call this the strong version of the givenness of thought. A weaker statement might be that the occurrence of one's own thoughts is sufficient for one to know that he thinks, but does not necessitate that one know this. Sellars denies both the strong and the weak versions of this thesis.2 His position on the empirical quality case seems to involve the idea that X is a knowable fact if and only if X is a public fact. If adhered to in the quality-of mind case, it implies that thinking, like redness, is necessarily a public quality of public entities (people); in other words, Logical Behaviorism. But Sellars does not accept this view. Thoughts, he says, "must have a determinate episode-ishness to bear the burden of... hypotheticals". ([3], p. 73). They are "private", though not "logically private", and we have "privileged access" to them ([4], pp. 176, 178). The question arising at this point is: if thoughts have the features just mentioned, how could we ever have come to know of their existence in a way that does not smack of givenness? ([4], pp. 176, 178). Sellars' answer is that this knowledge enters our conceptual scheme in the form of 'theoretical' concepts. To show how this process might have taken place, he invents a myth of his own, the "Myth of Our Ryleian Ancestors," a story which he initially characterizes as a piece of "anthropological science fiction". I will not repeat the story in full. I assume that its main outlines are familiar. The important features of the early Ryleian community's conceptual resources are that they have no concepts pertaining to entities that are not publicly observable, and therefore, that what 'goes bail' for the concept of thinking in their community is the concept 'thinking-out loud' - what we now call overt speech (O-thinking). Moreover the members of the early Ryleian community O-think their way through all their actions, step by step.3 What they at first do not suspect about themselves is that, even at this stage, they are all this while also "thinking things to themselves" in our present sense of the term ([3], p. 72). The myth's intellectual hero, Jones, one day notices the Ryleians beginning to do less O-thinking in the context of their otherwise normal behavior.4 A law-like relationship previously obtaining between O-thinking and other forms of behavior begins to break down. Undaunted, Jones formulates a theory to explain the anomaly. To wit: people have 'silent thoughts', inner states not publicly observable, which transpire during the course of their silent rational behavior. Jones' theory is revolutionary in that it goes on to postulate that O-thinking is not thinking in the proper sense but itself belongs in the class of behaviors to be explained in terms of his theoretical entities.5 O-thinking becomes 'overt speech', the normal culmination of processes which truly are thoughts. Jones intellectually radicalizes his comrades, first teaching each to interpret the other's behavior in terms of the theory, then teaching each to 'introspect' by approving utterances on their part of 'I am thinking that p' when their behavior supports this statement, disapproving them when their behavior does not support it. So trained, they can now report their thoughts, rather than inferring them from behavioral evidence. The fact that overt behavior is evidence for their thought-episodes is "built into the very logic of these concepts..." ([4], p. 189).

      In what sense is this story a 'myth'? I see three possible meanings. (I) It is armchair speculation at what our anthropological history must in some way have been like, i.e. a Historical Claim; (II) It is a conjecture that today's child acquires the concept of thinking in some such theoretical style, i.e. a Psychological claim; (III) It is an abstract claim, independent of the other two, that the concept of overt speech is epistemically, even logically, more primitive than the concept of a thought-episode in the following sense: one could have the concept of meaningful overt speech without having the concept of a thought-episode as something which overt speech expresses, but one could not have the concept of a thought-episode without having the concept of an overt speech-episode, and the concept of a thought-episode could be constructed by analogy with the concept of an overt speech-episode, i.e. it is a Logical Claim.6 I suspect that, if he could demonstrate the Logical claim, and conjoined to it his rejection of givenness, Sellars would maintain that (minimally) the Historical Claim follows. In the correspondence with Chisholm, Sellars remarks that his claim that the concept of a thought-episode must be 'constructed' is both a historical claim and a logical claim, nor, he says, can the two be "separated as with a knife." ([2], p. 536). More over, at the end of "Empiricism and the Philosophy of Mind", he muses that his myth has served to kill the Myth of the Given, and that, rather than being sheer myth, we can recognize in Jones Man himself in the middle of his journey from the cave to the drawing room ([4], p. 196). But, if we accept the Logical and Historical claims, reject the Given, and do not wish to assume that ideas are biologically inherited, it is difficult to resist the Psychological Claim as well. Though I believe my suspicions to be correct, I shall not raise any objections which are condemnations of a priori history and psychology. Indeed, for present purposes, I shall provisionally assume that the Historical and Psychological Claims are substantially correct, i.e. that our ancestors and present-day children did and do learn to use language effectively before learning, in theoretical style, that people have thought-episodes. I also grant the main implications of the Logical Claim, i.e. that a concept apparently similar to our concept of a thought-episode could be constructed from nothing but the resources potentially available to a Ryleian community (Behaviorese). Does nothing then remain to be discussed? Not necessarily, as we shall soon see.

      The route which I wish to explore begins with the familiar distinction between the internal consistency of a theoretical framework and the epistemic utility of accepting it. I shall assume that the force of this distinction should be attended to even in the abstract task of deciding whether the concept of thinking as a silent inner process analogous to speech is an acceptable explanans-concept where the explanandum is a community of rationally behaving persons who sometimes do and some times do not 'think-out-loud'. I hope to elucidate 'acceptable' and I will emphasize that it is only the weakest form of acceptability that I shall demand. Sellars refers to Jones as "an unsung forerunner of Behaviorism", a "Methodological Behaviorist" who ought not to be constrained by dogmatic proscriptions against postulation of unobservables, since this requirement is not considered binding on other scientists, e.g. physicists. Sellars very frequently compares Jones' theory to the Kinetic Theory of Gases, suggesting that they are similar in structure. Whether this is so and especially whether they have comparable epistemic utility is what I shall explore.

      Comparison with the Kinetic Theory of Gases does bear fruit in that there is no reason why a Behaviorese starting point should rule out such entities as molecules on the ground that they are unobservable. Postulation of unobservables to explain the observable is, per se, unobjectionable. However, I maintain that, to be acceptable, a theory must have a more than trivial explanatory power; that, when a theory does nothing more than imply the known phenomena, it suffers from fatal weakness. I distinguish between saying that a theory has large-scale epistemic utility and saying that it has the minimum epistemic utility which would warrant its acceptance. To illustrate the former, it is clear that the Kinetic Theory of Gases enables us to unify a large range of thermodynamic phenomena under the same set of mechanical laws that cover terrestrial and celestial phenomena. Moreover, it enables us to construct hypotheses which imply previously unsuspected facts, e.g. it predicted the unsuspected fact that the viscosity of a gas increases with its temperature. ([1], pp. 28,150-166). Moving to minimally desirable utilities, the following should not only be hoped for but demanded. A theory which purports to explain a range of observable phenomena by postulating a domain of unobservables must sufficiently define its theoretical entities by specifying enough of their properties as will enable one to non-trivially deduce the observable phenomena with reasonable closeness. Some of Sellars' own requirements will be added to these in due course. But consider the large scale utility of Jones' theory. As to unification, Kinetic Theory promotes this by finding a new application for an old set of laws, thus increasing the explanatory power of those laws, making them the fundamental principles of domains formerly described by distinct principles. Suppose that L is the set of laws connecting O-thinking and other behaviors in the early Ryleian community. Jones' theory is anything but an extension of L to a new domain. It is the rejection of L and the fundamental concepts of L. Thinking and speaking are one in L. In Jones' theory they are two things. Thus, he brings a new set of concepts into play. But the basic reason why Jones' theory does not unify is that there are not distinct theories to be unified. L describes normal, thinking out-loud behavior. The new phenomenon of silent rational behavior is covered by no laws at all. Before Kinetic Theory, of course, the macro-behavior of gases was covered by the classical laws of thermodynamics.

      The question of what testable hypotheses Jones' theory can be made to yield will depend on what general properties his theoretical entities are conceived to have. This means that we must first look to the matter of how well Jones' theory enables us to deduce the known phenomena, i.e. derive from its postulates the laws of L. Now assume for the moment that it is reasonably capable of this. What we are really interested in when we look for test hypotheses are implications of new phenomena or connections between phenomena that we had not previously suspected. As Sellars himself argues ([4], pp. 106-126 passim) we must not simply equate theoretical explanation with the derivation of empirical laws from theoretical postulates. To do so he says "... is to sever the vital tie between theoretical principles and particular matters of fact in the framework of observation" ([4], p. 120). To avoid this confusion between derivation and explanation is to see that

... theories about observable things do not explain empirical laws in the manner described; they explain empirical laws by explaining why observable things obey, to the extent they do, those empirical laws. ([4], p. 121).
The phrase 'to the extent that they do' points to another important explanatory role of theories.
(They) not only explain why observable things obey certain laws, they also explain why in certain respects, their behavior obeys no inductively confirmable generalization... ([4], p. 121).
Theories not only account for lawful behavior, but for behavior which is atypical and does not fit into any detectable pattern. Molecular chemical theory, for example, might explain the fact that gold dissolves in aqua regia at varying rates by supposing the gold used to actually be a mixture of two substances with different microstructures. For Sellars, this means that there is a requirement on any theory, viz. explanation of atypical or irregular cases must be by coherent extension of principles which explain the typical or normal cases.7

      Jones' theory fails to meet this requirement. He cannot show that his way of explaining the atypical case coheres with the way that his postulated entities explain the general conformance of people's behavior to the laws of L because he specifies no characteristic of the 'inner speech' episodes that occur in the normal cases which could profitably be extended to the irregular cases. Exactly the opposite is true of molecular chemical theory. The only thing Jones' theory postulates as the explanation of normal (O-thoughtful) behavior is also postulated without differentiating comment as the explanation of atypical cases.

      It might be thought that, one important test implication of Jones' theory would be that Ryleians can master the technique of reporting their thoughts introspectively when coached in Jones' fashion. Does not their acquisition of this skill tend to vindicate Jones' theory about the inner life of his comrades? It does not seem to. The test implication concerning an observable temperature-viscosity relation derived from Kinetic Theory arose from consideration of specific properties ascribed to the entities of that theory. Accordingly, experimental discovery of this relation tends to confirm the theory and weigh against any hypothesis which would deny the existence of these entities. But Jones does not ascribe any properties to thought-episodes which would allow us to nontrivially infer the introspection behavior of the Ryleians. The occurrence of this conditioned behavior in them is as compatible with the idea that Ryleians are thoughtless beings (proto-Ryleians) as it is with the idea that they think. There would appear to be no impossibility in supposing that a pliable proto-Ryleian creature of habit might respond equally well to Jones' coaching. The main reason why this is so is that Jones' version of 'introspecting' only requires that they be trained to preface the speeches they are normally inclined to make in certain circumstances with phrases like 'I think that...', 'I imagine that...', and so to respond whenever any one asks what is 'on their minds'. Introspecting, so conceived, is on a logical par with promising. Just as promising is not the sort of thing which must be a report or description of unobservable inner episodes (as Austin has shown), so with Jonesean introspecting. The ability in no way points to the existence of such processes.

      Jones' theory generally suffers from lack of empirical content. To illustrate this, consider a theory I shall call the Entelechy Theory of Moving Bodies (The E-theory). The E-theory stipulates that factors integral to Classical Mechanics such as mass, velocity, relative position, etc., are necessary but not sufficient to account for such phenomena as free fall, projectile motion, etc. It requires that we postulate that there are unobservable entities (entelechies) which inhabit all bodies and which behave in certain undefined manners specific to various kinds of motive phenomena. There is nothing logically inconsistent about the E-theory, and no phenomenon predicted by Classical Mechanics alone will tend to falsify it. Hence, even if Classical Mechanics is true but the set of theoretical postulates concerning entelechies is contingently false, the E-theory 'works' precisely because it is devoid of empirical content. Obviously, it is a parasite whose host is Classical Mechanics and its theoretical postulates should be dispensed with as epistemically worthless. Are Jones' theoretical postulates any better off? It would seem that they are not. By design, they predict only the phenomena they were introduced to explain. Unless we are told of some property other than that of being the processes which normally culminate in speech and other forms of intelligent behavior, we cannot non-trivially infer anything from his postulates, not even the laws of L.

      If these postulates are parasitical, their host is Behaviorese. To dispense with these postulates, we must show that Behaviorese has substantial explanatory power by itself, power at least as great as that of Jones' theory. On the face of it, there are many situations in which it is quite useful to believe that people are 'thinking', even when they are not expressing those thoughts. The question is not what do we mean by 'thinking' in these cases, but what citizens with none but purely Behaviorese resources could reasonably say in response to Jones' proposal about the meaning of 'thinks' in such cases. Supposedly, one of the points that Jones would make is that silent rational behavior is anomalous in comparison to traditional Ryleian behavior. But it is not clear why they should consider it anomalous. In the pre-Jonesean era, there must have been lapses into silence, however brief those lapses may have been. To say that rational behavior became anomalous when and not before silences appeared in it would entail that pre-Jonesean Ryleians were non-stop chatterboxes who could scarcely catch their breath. If we grant that Ryleians need not consider any and all silence during intelligent performance to be anomalous, why would it be wiser for them to adopt a novel framework of unobservables rather than expand their old set of principles which allowed for the non-problematic character of some silence?

      In this context, one can point to an important disanalogy between Jones' theory and Kinetic theory. Owing to the basic information contained in the postulates and laws of Kinetic Theory, we can always tell what is happening on the observable level by knowing what is happening on the theoretical level, and we can establish that there are certain kinds or classes of behaviors on the theoretical level by knowing what is happening on the observable level. With Jones' theory, it is impossible to establish precise and complete observation-theory connections of this sort. One must instead rest content with rough-hewn criteria which tell us when it is generally reasonable to believe that a certain sort of theoretical episode is taking place. The explanatory value of such a theory can be seriously questioned in the face of such difficulties, but the real point here is that precision is not available to Jones. How good an under standing of intelligently behaving persons is possible in purely Behaviorese terms? I think that Ryle and others have shown that it can be very good. If no more than reasonable, rather loose criteria are available for establishing observation-theory connections of the sort that Jones can propose, it would seem that at least this much is potentially available to a speaker of Behaviorese who accepts no theory committing one to unobservable inner episodes. Instead of such novel 'categorical' facts as these, the old-fashioned Ryleian can speak of the 'mongrel categorical hypothetical' fact that a silent agent could tell us the why of what he does, should an occasion arise in which it is appropriate for him to do so. Indeed, there are many complex forms of behavior (e.g. simultaneously smoking a pipe, listening to the radio, addressing an envelope, etc.) which we can quite reasonably say involve a large number of mongrel categorical-hypothetical facts, but which cannot very plausibly be explained in terms of a large number of simultaneously occurring theoretical 'inner speech' episodes. It need not be argued that purely behavioristic resources will never leave us in the dark about the explanation of what people do. But it does appear that a theory like Jones' has no particular explanatory advantage over the former.

      One may at this point begin to wonder whether Jones can be correctly described as constructing a theory in the scientific sense at all. In a number of places8 Sellars hints that this is the case, saying that what Jones constructs is not comparable in all respects to a scientific theory. On the other hand, there are enough places9 in which Sellars makes direct and strong comparisons between Jones' theory and scientific theories proper to justify a demand that we be told clearly why we should not expect Jones' 'theory' to have the same virtues as we expect of any scientific theory. I believe I have shown that it does not and perhaps cannot have such virtues. In view of this state of affairs, to be told that the story of Jones and the Ryleans is more like a 'social contract theory'10 or a platonic 'myth' is scarcely illuminating.

      Consider the following rebuttal to the thesis of the present paper:

Sellars does not present Jones' theory. Sellars simply characterizes the framework of inner thoughts as the classical framework, and hence as doing the explanatory job vis d-vis thinking-out-loud, propensities to think-out-loud, and their relations to behavior which Cartesians (or Aristotelians) attribute to them. The point of 'Empiricism and the Philosophy of Mind' is metatheoretical; to wit: the status of the Cartesian frame work, and its explanatory power can best be understood by comparing it to a theory. Sellars does not claim to describe, other than by reference to the classical account, how the theory deals with an interesting range of phenomena. Therefore, Sellars' account cannot be faulted on this score.
My reply is the following:

     (a) The strategy of my paper is to argue, first, that, if Jones is a sort of fictional character-cum-Weberian Ideal Type who constructs an Aristotelian/Cartesian theory of inner thoughts about his compatriots with no given as a support for it, and if its "explanatory power can best be understood by comparing it to a theory" (of the scientific postulational variety), then Jones' theory has no explanatory power for the same reasons that my fabricated entelechy theory is lacking in this respect.

      (b) If telling a pure bear story were the Ryleian Ancestor Myth's only reason for being, then its place in EPM is a mystery. I take it, therefore, to at least be a metaphorical way of referring (not necessarily to anthropological history or to the psychology of today's child, but) to what I call a Logical Claim about the epistemological priority of the concept of an overt speech episode's meaning that-p to the concept of a thought's being about (the fact that) p. At the very least (in Sellars' view) Jones and the Ryleians have being to the extent of populating logical space. As Rawls says concerning social contract theories, they are not so much intended to be history as pictures of hypothetical situations into which we can (for philosophical purposes) enter at any time. Since logical claims are atemporal in this sense, I take it that it must be connected to the sort of theory that Sellars does accept, for it is reasonably evident that he does endorse the sort of thesis that I call the Logical Claim. (See notes I and 6). Indeed, the latter would appear to follow from nothing more than his wholesale rejection of givenness together with a retail application of same to the cartesian view of the givenness of thoughts for thinkers thereof.

      (c) The culmination of my argument then is:

      (i) Any theory about thinking which is constructed within bounds that exclude givenness (in every sense) and which regards the thoughts that people have as unobservable inner episodes, must provide arguments for the acceptability of the idea that people 'have thoughts' in this sense (as opposed to the old-fashioned Ryleian sense). Any such theory of thinking must provide satisfying reasons why such unobservables are fecund enough in concept to explain the occurrence of interesting phenomena, particularly silent rational behavior. My argument is that, even if these unobservables are conceived by analogy with overt speech episodes (the concept of which is extremely rich), they lack the fecundity requisite to provide satisfactory explanations of the phenomena, even in a more or less commonsensical sense of explanation, let alone the technical, scientific style of explanation which fairly bristles with demands for epistemic utility.

      (ii) Finally, it seems that Sellars' account of thoughts is an instance of such a theory as is described in general above.11 Consider ([4] p. 178) especially Sellars' declaration of intent to "defend... a revised classical analysis of our commonsense conception of thoughts..." and his claim on the same page that the invention of the Ryleian myth will 'answer' the questions of how thoughts can be 'inner' (not merely conceived to be 'inner') and how they can be (not just conceived to be) 'linguistic'.

      I have criticized the idea that thought-episodes are private inner episodes precisely on the assumption that it has the form of a theoretical postulate valuable for explaining behavior, rather than the formulation of a fact which is 'given'. I must now emphasize (and Sellars would agree) that we certainly do know that there are such episodes, innumerably many of them, going on 'in' us. We also know that there is something 'more to' introspection than what would take place among suitably trained proto-Ryleians. But, then, what is the basis of this knowledge of ours? According to Sellars, it has the same sort of basis as our use of a concept like 'molecule'. I have argued that this is not the case. The original source of the difficulties I have uncovered is Sellars' radical rejection of givenness. Even if, as a matter of contingent fact, everyone acquired the concept of a thought-episode in theoretical style, the question would remain whether or not it could be acquired in some other way. For Sellars, we cannot reject the Given and still concede this possibility. I will argue that we can do this, provided the notion of Givenness be appropriately reassessed.

      Once more, the basic logical confusion which Sellars sees in the idea of givenness is that

... instead of coming to have a concept of something because we have noticed that sort of thing, 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 ([4], p. 176).
Consider the following:
Tl. For any and all properties K, one can notice that something is (exemplifies) K logically and/or temporally prior to having concept K
T2. There is at least one property K such that one can notice that something is K logically and/or temporally prior to having concept K
Pl. For any individual x, for any property K, one can notice that x is K if and only if one possesses concept K

I take Sellars' rejection of givenness to be equivalent to the denial of '(Tl or T2)' and an endorsement of Pl. To this extent, I agree with him. Moreover, like Sellars, I deny the strong version of givenness with respect to thoughts. It is not impossible to think without knowing that one thinks. But consider another possibility:

T3. There is at least one property K, such that being presented with a K thing is epistemically sufficient for one to acquire concept K

The denial of T3 does not follow from the denial of '(Tl or T2)'. Since the word 'prior' occurs in Tl and T2, T3 does not imply either of them. T3 would support the weak version of the givenness of thoughts, namely that their occurrence enables us, but does not force us, to know of their existence. To put flesh on these bones, suppose that a reasonably intelligent child who had learned the rudiments of our language were to be marooned on an uninhabited island by cruel parents who had never educated him to the idea of disengaged, silent thinking. Suppose, further, that through strength of will, resourcefulness and much luck, he survives well into his mature years. Might this person not one day realize some thing like the following?

Something like speech goes on within me even when I am silent and physically still.
I do not suggest that his realization would merely consist of noticing a stream of verbal imagery. I suggest that it may be the dim and crude beginning of the knowledge that there is a silent inner process which has in common with overt language such properties as truth, falsity, reference, etc. Since the limited givenness of such a fact would be logically harmless, I can see no reason for ruling it out a priori. It would be consistent with the idea that conceptual thinking is somehow analogous to language, and it would even support the idea that ability for thought presupposes ability for language. The hero in my brief legend is able to acquire the novel concept only because he already has a concept (speech) which is analogous to it. Thus, it would only be necessary to defend the idea of givenness with respect to some subset Kl . . . K,, of the set of all properties K such that for each Ki, one already has a concept K; where the property Kj is analogous to the property Kj. My extremely limited version of givenness as embodied in T3 is free from the sorts of difflculties that Sellars finds with givenness, in particular, T3 does not lead to the paradox that one must acquire concepts but have them first. Moreover, T3 may be shored up by means of indicating a very special property which, serving as the value for 'K', might very well render T3 true.

      If Chomsky12 is right we may have what might be called an innate idea of language (qua structure not qua overt speech signal). This is not the place to enter into an extended exploration of Chomsky's thesis, nor do I feel obliged to fight Chomsky's battles for him. I shall instead be content to speculate about likely possibilities, and then await further developments in psycholinguistics.

      I would be inclined to call the innate structure in question a proto-concept for two reasons: (a) the development of this innate structure is, according to Chomsky, maturational, rather than merely additive. (b) the 'conceptual' structure that a baby has (according to Chomsky) is much less specified in developed content than is an adult's.

      Let Kj = the formal structure of thought, and Ki = the deep structure of language L; then, Kj may be an instance of a true-making value for 'K' in T3.

      That we innately possess the proto-concept out of which the concept Ki develops may serve as a hypothesis which explains language acquisition and other aspects of performance. If one possesses any concept K innately (a la Descartes), then, obviously, no Sellars-like paradox of givenness can arise. But this need not arise even if one instead says that K develops maturationally from a proto-concept corresponding to K The hypothesis that we innately possess proto-Ki would, if confirmed, adequately ground a weak thesis of the givenness of thoughts (their occurrence enables us, but does not force us, to know of their existence.)

      Hailing my legend would mean that the myth of Jones and the Ryleians would lose its principle philosophical interest. Tutelage under a Jonesean master might be regarded as a convenience but not a necessity. Unless we preserve the possibility that each of us could in principle be his own master in this part of his education, the idea that there are private thought episodes is too weak to survive criticism by a Logical Behaviorist. By accepting a demythologized Given, this knowledge can be adequately grounded.

NOTES

1 "In your first letter you expressed agreement 'with much of what (I) have to say about the myth of the given'. Well, of a piece with my rejection of this myth is my contention that before these people could come to know non-inferentially (by 'introspection') that they have thoughts, they must first construct the concept of what it is to be a thought." ([2], p. 527). At this point, Sellars refers to the central thesis he maintains against the notion of givenness (to be found in [4], p. 176; quoted above on p. 232). From this point it is obvious that the connection is established in Sellars' own mind.

2 The quote from [2], p. 527 really supports this claim by itself. Other parts of Sellars' writings which indicate that he rejects both the strong and the weak version are: p. 177 of [4] section 47; second paragraph of section 45, p. 175 of [4]. Others could easily be adduced, but, in general, it is obvious that if Sellars did accept even the weak version of givenness, there would be no point whatever to the whole Ryleian Ancestor myth.

3 Section 56, p. 186 of [4]; A-9 on p. 522 of [2]; also p. 530 af [2]; [3], p. 75.

4 Section 69, p. 87 of [3]; [3] p. 97; [3] p. 151; [3] p. 159.

5 Pp. 186-7 of [4]

6 "I think that in my first letter I was so using the term 'analysis' that to say that X is to be analysed in terms of Y entails that it would be incorrect to say of anyone that he had to concept of X but lacked the concept of Y.... Thus when in A-12 I denied that "... means p" is to be analysed as "'...' expresses t and t is about p", I intended in effect to deny that the fact that a person lacked the concept of a statement's expressing the thought that p would be a conclusive reason against supposing him to have the concept of a statement's meaning that p." ([2], p. 534). This together with a denial of givenness constitues the essence of a Logical Claim.

7 P. 122 of [4]

8 [3], p 154 parag. 8; p. 155 parag. 10.

9 P. 155, parag. 11(b); p. 159; p. 165, parag. 36.

10 P. 71, parag. 25; p. 155, parag. 10; p. 165, parag. 37.

11 Also, see p. 32, 33 and 34 of [4], and p. 57, 58 and 59 of [4], especially passage "Our Direct Knowledge that the thought that-p has occurred to us involves the concept of an occurrent which is analogous to the statement that-p.", and p. 189 of [4], the passage beginning "My myth has shown that...", and especially the last sentence of same are particularly relevant.

12 See, for example, Language and Mind, passim. The idea that a conceptual structure is biologically emergent in the evolutionary scheme would seem to be consistent with a thoroughly scientific picture of persons. In a number of places, Sellars hints that he finds some merit in this idea. See, for example, Science, Perception and Reality, pp. 15-18, 90, 97 and 325. Also, p. 56 of 'Toward a Theory of The Categories'.

BIBLIOGRAPHY

[1] Brush, Stephen G., Kinetic Theory, Permagon Press, Oxford, 1965.

[2] Feigl, Herbert, Scriven, M., and Maxwell, G. (eds.) Minnesota Studies in the Philosophy of Science, vol. II, University of Minnesota Press, Minneapolis, 1958.

[3] Sellars, Wilfrid, Science and Metaphysics, Humanities Press, New York, 1968.

[4] Sellars, Wilfrid, Science, Perception, and Reality, Routledge and Kegan Paul Ltd., London, 1963.