THE GIVEN AND THE A PRIORI: SOME iSSUES IN THE EPISTEMOLOGY OF C. L. LEWIS ' Lhesis fer the“ mega 6 cf Lb. D ALLCHLGALi 32' ATE ULLL VERSE ELY WLLLLML LLLSHARD CQNNOLLY 3973 ILLLLL ‘ Ll lmfi 3129 18162163.”- RY i. MlChigan State ~ University “'3 "WM This is to certify that the thesis entitled THE GIVEN AND THE A PRIORI: SOME ISSUES IN THE EPISTEMOLOGY OF C. I. LEWIS presented by William R. Connolly has been accepted towards fulfillment of the requirements for _Eh_.D__degree in Lin-@2917 Major professor Date_June 181. 1972 0-7639 - - IIIIDI m . " In“ 8 S": BOOK BINDLRY ENC. LIBRARY amocns gums-mm b. ‘7 APR ‘1 9 2000 ABSTRACT THE GIVEN AND THE A PRIORI: SOME ISSUES IN THE EPISTEMOLOGY OF C. I. LEWIS BY William Richard Connolly Jr. The dissertation deals with the roles played by the given and the a priori in the epistemology of C. I. Lewis. It is divided into two parts which deal respectively with these two notions. In Part I, I present a critical analysis of three criteria used by Lewis to distinguish analytically the given element in experience from its conceptual interpretation. In Chapter One I deal with Lewis' claim that the given ele- ment in experience is unalterable by interests and mental attitude. In Chapter Two I deal critically with Lewis' distinction between sense qualia and the properties of objects, a distinction which parallels the distinction between the given and its conceptual interpretation in as much as it is sense qualia which are given; knowledge of properties already marks the categorial interpretation of what is given. In Chapter Three I deal with a third criterion by means of which Lewis William Richard Connolly Jr. conceptually isolates the given from its conceptual inter- pretation. According to Lewis, the given element in any experience is the incorrigible datum element marked by the certainty of our awareness of it. Such awareness consitiutes the foundation of our knowledge of empirical objects and is to be distinguished from such knowledge which is always probable and less than theoretically certain. In Part II I deal with two aspects of Lewis' concep- tion of the a priori. According to Lewis, unlike the unal— terable given element in experience, the a priori element is just that element in knowledge and experience which always has conceivable alternative explications. In Chapter Four, after discussing the important motivating role of his earlier studies in logic, Lewis' claim that there are alternative systems of logic which we must choose among on pragmatic grounds is critically evaluated and found wanting. In the fifth and final chapter Lewis' conception of the role of the a priori in natural science is critically evaluated and a suggestion is made as to how, with some slight reinterpreta- tion of Lewis, a quite plausible conception of the role of the a priori in natural science emerges which constitutes an interesting and promising proposal concerning the analysis of the concept of scientific law. THE GIVEN AND THE A PRIORI: SOME ISSUES IN THE EPISTEMOLOGY OF C. I. LEWIS BY William Richard Connolly A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Philosophy 1973 Si Lt. W To Margaret for her patience and moral support ii ACKNOWLEDGMENTS .A special debt of gratitude is owed to Dr. Rhoda H. Kotzin, adviser and committee chairman. Her continuous encouragement and critical comments on earlier drafts were indispensable aids in the preparation of this dissertation. Appreciation is also owed to Dr. Charles McCracken, Dr. Herbert Garelick and Dr. Richard Hall for critical comments and suggestions on earlier drafts. iii TABLE OF CONTENTS Page PART I 0 THE GIVEN O I I O O O O O O O O O O O O O O O 1 Chapter I. THE TWO ELEMENTS IN EXPERIENCE: THE UNALTERA- BILITY OF THE GIVEN O O O O O O O O O I I O O 3 II. QUALIA AND PROPERTIES: THE SUBJECTIVITY AND INEFFABILITY OF THE GIVEN . . . . . . . . . . 26 I I I O SENSE CE RTAINTY O O O O O O O O O O O O O O O O 5 3 PART II. THE A PRIORI . . . . . . . . . . . . . . . . 114 IV. LOGIC MD THE A PRIORI O O O O O O O O O O O O 117 V. A FUNCTIONAL INTERPRETATION OF THE A PRIORI . . 182 NOTES AND REFERENCES . . . . . . . . . . . . . . . . . 215 SELECTED BIBLIOGRAPHY . . . . . . . . . . . . . . . . 229 iv PART I THE GIVEN Introduction The notion of the given element in experience is a crucial one in the epistemology of C. I. Lewis. In Part I Lewis' theory of the given will be critically evaluated. In particular, it will be argued that none of the three means by which Lewis tries to analytically isolate the given element in experience from its conceptual interpre- tation is adequate for the drawing of a clear and coherent distinction. The three theses defended by Lewis are: (1) that the given is that element in any particular experience which is unalterable by interest and mental attitude; (2) that sense qualia (the content of any given experience) are subjective and ineffable, while objective properties are neither; and (3) that our apprehension of the given, un- like our knowledge of objects, is certain, is not judgment and cannot possibly be mistaken. The following three chapters will deal critically with these theses. CHAPTER I THE TWO ELEMENTS IN EXPERIENCE: THE UNALTERABILITY OF THE GIVEN For Lewis it is requisite that any theory of know- ledge distinguish two elements in experience--the given and the pure concept. It is Lewis' contention that clas- sical theories of knowledge all accept this distinction in one way or another. The differences that do exist between classical theories can be attributed to difference in em- phasis. As Lewis puts it: There are, in our cognitive experience, two elements; the immediate data, such as those of sense, which are presented, and a form, construction or interpretation, which represents the activity of thought. Recognition of this fact is one of the oldest and most universal of philosophic insights.1 Furthermore, if there is no given, "then knowledge must be contentless and arbitrary; there is nothing which it must 2 be true to." And if there is no construction or interpre- tation by the mind, then the possibility of error would be inexplicable. If the significance of knowledge should lie in the data of sense alone, without interpretation, then this sig- nificance would be assured by the mere presence of such data to the mind, and every experience must be vera— cious.2 Stated in so general a fashion, Lewis' point is not likely to be disputed. Nor will the distinctions in such general terms be challenged here. If there is nothing that is not in some sense given to the mind (that is, something which is not created by the activity of thought), then there would be no means of telling which of two incompat- ible sets of belief are true. Failure to recognize that there is a given element in experience is inconsistent with the possibility of error. But so general a statement of the position does not get us very far. Granting that know- ledge presupposes both something independent of mind as well as some activity of mind (variously called by Lewis 'construction', 'inference' or 'interpretation'), how does Lewis make the distinction, and what does that distinction tell us about these two elements or aspects of experience? Lewis approaches these matters with several different dichotomies in mind which, he seems to assume, divide up experience in just the way he intends by the given/concept distinction. One may indicate these in several different ways. The distinction between the given and the conceptual elements in experience parallels the extent to which the mind is passive and the extent to which it is active with respect to the objects of knowledge and experience. Be- cause there could be no experience of objects without the conceptual activity of mind, there could be no objects of knowledge for the purely passive being.4 A second distinction which Lewis takes to parallel the distinction between the given and the conceptual elements in experience is the distinction between those aspects of experience which are unalterable by any act of my will and those aspects of my experience which are so alterable. For example, how I am to classify objects is not dictated by experience (al- though experience may make some systems of classifications more cumbersome or less useful than others). But whether there are any instances of a particular kind within a par- ticular classificatory system is something that is dictated to me by the content of experience. For Lewis, and this is the major thrust of his pragmatic theory of the a priori, the conceptual element in experience is just that element which always has alternatives.5 I do not take the above as in any way clarifying Lewis' position. All that I wish to point out is that, for a correct understanding of Lewis' distinction between the given and the conceptual elements in experience, it is necessary to keep in mind the way in which these other distinctions made by Lewis are used by him to explain why there is a need for the more fundamental distinction. With that in mind, we may now turn to the more particular argu- ments forwarded by Lewis on behalf of the distinction in the particular way in which he makes it. The first and in the end the most important charac- teristic of the given element in experience is its unalter— ability. In the final analysis unalterability is for Lewis the definitive criterion of givenness. . . . certain items or aspects of the content of ex- perience satisfy the criteria of givenness. These are, first, its specific sensuous or feeling character, and second, that the mode of thought can neither create nor alter it--that it remains unaffected by any change of mental attitude or interest. It is the second of these criteria which is definitive; the first alone is not sufficient, . . .5 There is, in all experience, that element which we are aware that we do not create by thinking and cannot, in general, displace or alter. As a first approximation, we may designate it as 'the sensuous'. It is not altogether clear what Lewis means in claiming that unalterability by interest and mental attitude is the (or a) definitive criterion of givenness. He does seem to be implying that it is a condition. However, is a definitive criterion in Lewis' sense a necessary condition, a sufficient condition, or both? It would appear that a definitive criterion cannot be simply a necessary condition, for in the passage quoted above he rejects sensuousness as a definitive criterion on the ground that sensuousness is not a sufficient condition of givenness. The content of a particular imaginary fancy, for example, might have a sensuous or feeling quality, but is surely alterable by interest and mental attitude. Thus by "definitive criterion" Lewis seems to mean at least sufficient condition. The ques- tion now is does he mean more than this? Does he mean by "definitive criterion" necessary and sufficient condition or just sufficient condition alone? If by "definitive cri- terion" Lewis means only sufficient condition, then there is no a priori reason to think that givenness has only one definitive criterion. Since Lewis later does offer other ways of analytically isolating the given element in any particular experience (e.g. that the given element is the incorrigible datum element, all the rest is a matter of interpretation), it would appear that givenness may have several definitive criteria. This would appear to suggest that by "definitive criterion" Lewis meant sufficient con- dition, since it is difficult to see how it would be pos- sible to have two (or more) conditions which are each necessary and sufficient, yet are independent of one another. But claiming that by a definitive criterion Lewis simply meant sufficient condition would precipitate prob- lems. If unalterability by interests and mental attitude and incorrigibility were each sufficient conditions for the given element in experience, then it might be possible to have the one (say unalterability) without having the other. This would appear to be an unhappy interpretation of Lewis, since it is unlikely that he would think that an element in a particular experience could be incorrigible without being unalterable (or vice versa). It would appear then that if these are in fact different sufficient condi- tions of givenness, at least they are not independent of one another. Indeed, it could be persuasively argued that the early emphasis on the unalterability of the given is to insure the incorrigibility of our awareness of it. But, if they are not independent, then it is at least possible that what we have are two different (but mutually dependent) necessary and sufficient conditions of givenness. Moreover, it would seem unlikely that Lewis would not include within a definitive criterion of given- ness necessary as well as sufficient conditions, since to do so would make it possible for an element in a particular experience to be given, but for it to be alterable by mental attitude and for our apprehension of it to be less than theoretically certain. I take it, then, that unalterability is being offered as a necessary and sufficient condition of givenness. From the manner in which Lewis defends the criterion of unalterability it would appear that with respect to there being an element of givenness in all experience there is no question of strict proof. This is typical of Lewis' discussion of the given. That there is such a thing as the given element in experience cannot be proved or argued for because it is just too fundamental for proof. To deny the given is not so much a mark of philosophical subtelty as a mark of insanity. As an unprovable fundamental epis— temological principle one can only try to remove any confu- sion or misunderstanding that may prevent the acknowledgment of this given element in experience. However, trying to be clear about the exact nature of this distinction carries with it a good bit of theoretical commitment, both ontolog- ical and epistemological. Moreover, what Lewis commits himself to in making the distinction is in no sense obvious, as I hope to show as we proceed. Once it is clear what Lewis has committed himself to in claiming that there is such a thing as the given, it just is not that obvious that there is a given in ttgt sense, and to deny its exist- ence in ttgt sense is not to fly in the face of sanity. However, since for Lewis this is not a matter for strict proof, it is not surprising to find him resorting to examples. Suppose at the moment that I have a pen in my hand. In describing this item of my experience I make use of terms whose meanings I have learned. In describing my ex- perience I perform two different but related mental acts. First of all, I abstract this item, which Lewis refers to as a presentation, from the present total field of my ex- perience. Secondly, in describing this presentation as a presentation of a pen, I relate this item of my experience to what is not just now present in my experience. That is, I relate this.present item of my experience to other actual and possible experience. This selecting and relating of a given presentation is significant in two respects. First of all,.it is related to further actual and possible 10 experience in ways that have been learned. Secondly, my taking of this presentation as a presentation of a pen re- flects my interest and purpose to write. Someone with different purposes and interests and who has significantly different past experience would attach a different meaning to this experience. For example, the savage, unaccustomed to writing, would take this ex- perience differently. He would selectively attend to different features of the experience. It might, for ex- ample, be taken by him as presentation of a plaything or a sharp weapon. What he selectively attends to would differ in his case as well as the relation of this presen— tation to further actual and possible experience conditional on actions of his own. Similarily, had my interests and purposes been different, I might have taken this experience as presenta- tion of a different kind of thing. For example, had I intended to illustrate a problem in solid geometry rather than to write, I would have taken the same experience as a presentation of, say, a cylinder. Having different pur- poses in.this case I would selectively attend to different features of the experience and would relate that experience to further actual and possible experience in a different manner. Just as in the case of the savage, the experience would have a different meaning. The same experience would be differently taken inthe different circumstances indica- ted. As Lewis puts it: 11 Something called given remains constant, but its char- acter as sign, its classification and its relation to other things.and to action are differently taken.8 The given is what is common to the three experiences indi- cated.above, that of the savage and the two experiences of the same mind with different interests. . . . what I refer to as the given in this experience is, in broad terms, qualitatively no different than it would be if I were an infant or an ignorant savage.9 However I describe some item of my experience, I can never describe it simply as given.10 In describing an item of a.present.experience as.a presentation of an object of a certain.kind, I supplement this given experience by a meaning having to do with its relations to further actual and possible experience and to my interests. All that.comes.under the broad term 'meaning' . . . is brought to this experience by the mind, as is evidenced by the fact that in this respect the experience is alterable by my interest and my will.11 Moreover, my taking some particular experience as a presentation of a pen or as a presentation of a cylinder is that element of experience which is alterable by the interests and purposes that I have. What is common to these two experiences, whether they be taken as the presentation of a pen or.of a cylinder, is what Lewis intends to refer to as the given element of this particular experience. The form of Lewis' argument is as follows. By appeal to ex- amples Lewis claims that there is something common to two qualitatively similar experiences, differently taken. 12 Secondly, by scrutinizing these examples, this common ele- ment of differently taken experiences is not presently alterable by any act of my will. While the same experience may be differently taken as presentation of a pen or as presentation of a cylinder, and in this respect is alter- able by my interests and purposes, nevertheless within the range of alternative takings there are limitations which reflect the unalterability of the given. The distinction between this element of interpretation and the given is emphasized by the fact that the latter is what remains unaltered, no matter what our interests, no matter how we think or conceive. I can apprehend this thing as pen or rubber or cylinder, but I cannot, by takingzthought, discover it as paper or soft or cubical. Thus, according to Lewis, we can isolate the given by its unalterability and its sensuous or feeling quality. However, we cannot describe the given as such, since in order to describe it we must subsume it under some category and such subsumption of the presently given relates it to what will or could be given in further actual and possible experience. Thus, applying any concept to the presently given, which is a necessary condition of any description, is significant of more than could be given in any "specious present". So that, in a sense, the given is ineffable always.l3 One should be careful not to attribute to Lewis the View that the fact that the given element in experience is an identifiable element in experience entails that the 13 given is ever experienced in isolation from conceptual ele- ments. Lewis is quite emphatic in maintaining that such isolation is improbable in fact and is unnecessary for epistemological purposes. In this connection it is impor- tant to distinguish between at least two different senses of the word "given". Most frequently, at least in philo- sophical circles, the word is used with something like the meaning which is here intended by Lewis. On other occasions, however, it is used to refer to the data with which philos- ophy in general begins. But what philosophy begins with is not, as Lewis puts it, the thin given of immediacy, but the thick experience of the world of things. We do not see patches of color, but trees and houses; we hear, not indescribable sounds, but voices and violins. What we most certainly know are objects and full-bodied facts about them which could be stated in propositions. Such initial data of object and fact set the problem in philosophy and are, in a measure, the criteria of its solution, since any philosophic theory will rightly be rejected as inaccurate or inade- quate if it does not measure up to, or account for, experience in this sense.14 While an identifiable element in experience, there is no question of the distinction between the given and the conceptual elements in experience reflecting any temporal sequence in the acquisition of knowledge. Distinguishing these two aspects or elements in experience does not commit one to holding that the given is first registered and then conceptually interpreted. "The given is $2! not before, 15 experience." The important question, for epistemological l4 purposes, is not whether the given can be experienced un- contaminated, so to speak, but whether this unalterable given element is an identifiable element in experience. The given element is never, presumably, to be discovered in isolation. If the content of perception is first given and then, in a later moment, interpreted, we have no consciousness of such a first state of intuition unqualified by thought, though we d9 observe alteration and extension of interpretation of given content as a psychological temporal process. A state of intuition utterly unqualified by thought is a figment of the metaphysical imagination, satisfactory only to those who are willing to substitute a dubious hypothesis for the analysis of knowledge as we find it. The given is admittedly an excised element or abstraction; all that is here claimed is that it is not an 'unreal' abstrac- tion, but an identifiable constituent in experience.16 A second confusion, partially resulting from Lewis' occasional use of the terms "given" and "data of sense" as roughly synonomous, should also be avoided. The term "data of sense" or alternatively "sense data" has connota- tions which are inappropriate for epistemological purposes, as Lewis points out.17 For example, "sense data" is often used as a psychological category, marked off from other mental phenomena by their correlation with events in the afferent nerves. But "sense data" with such connotations would be an inappropriate term to use to refer to what Lewis intends to refer to by the term "given". In the first place, the term "sense data" used with such connota- tions would entail that the given is, in some sense of the term, mental. But what Lewis intends to refer to by the term "the given" is without interpretation (that is, with- out reference to some particular category) neither external 15 reality nor explicit self. The division of experience into subject and object already marks the influence of concep- tualization; such division is only achieved when the given has been conceptually interpreted. "In immediacy, there is no separation of subject and object."18 Secondly, the basic categories of epistemological analysis are, in the nature of the case, prior to any find- ings of the special sciences. Interpreting the given or data of sense psychologically would render psychological findings prior to epistemological analysis.19 Thirdly, more qualities are given than what are strictly referred to as sense data. ". . . the pleasant- ness or fearfulness of a thing may be as un-get-overable as its brightness or loudness--thathuestion, at least, 20 must not be prejudged." The given "is the brute-fact element in perception, illusion and dream (without ante- cedent interpretation)."21 Moreover, one should distinguish between the given and the object which is given. When properly interpreted (that is, when referred to the proper category), the given is the presentation of something real. But what is given is never the whole object. In any presentation the object is given only in part. The real object as known involves the conceptual interpretation of some given element and 22 includes much more than could be given at any moment. Even the selective attending which sets off certain features 16 of a given field as presentation of an object already re- flects the activity of an interested mind. That is, such selective attention already marks the apprehension of a given experience as presentation of a particular kind of thing.23 There is one further point which should be noted. While the marking off within the total field of given ex- perience of presentations of things and events already marks the influence of an interested mind, "Experience, when it comes, contains within it just those disjunctions which, when they are made explicit by our attention, mark the boundaries of events, 'experiences' and things."24 That is, while the manner in which we break a given field of experience into the presentation of things and events reflects our interests and purposes, such selective atten— tion cannot mark disjunctions in an undifferentiated field. "The interruptions and differences which form the bound- aries of events and things are both given 32g constituted by interpretation."25 There are psychological considerations which appear to present difficulties for Lewis' claim that unalterability by interest and mental attitude is a definitive criterion of the given element in experience. After briefly discus- sing these psychological considerations, an attempt will be made to accommodate them within Lewis' general framework. However, the revised position will require that Lewis either l7 qualify or reject some of his other characteristic theses about the given (in particular, the thesis that it is the incorrigible datum element of empirical knowledge). There are two ways in which our interests and mental attitude influence our perceptual expectations. The first way does not constitute a challenge to Lewis' unalterability thesis. It is best illustrated by example; Lewis' pen ex- ample will suffice. Suppose that I have the purpose of writing a letter to an old friend. Further suppose that I am looking among articles on my desk for a writing implement. I am presented with a certain visual experience, for example of an elon— gated cylindrically shaped object on my desk. Having the intention of writing, I take this experience to be a pre- sentation of a pen. That is, I refer the presently given to the category of pen. Having taken the presently given as presentation of pen, I am led to have certain expecta- tions about the character of future experience, conditional on my acting in certain ways. For instance, I take the visual shape to be a sign of a particular tactual experience that would occur if I were to reach in a certain direction and make manipulative movements. In applying the concept pen to this given experience I have certain expectations about, make implicit predictions about, future actual and possible experience. Had my purpose been other than writing (for example, had I been looking for something with 18 which to open a letter), I might have taken the experience differently. For instance, I might have taken it as a presentation of a letter opener. Having different purposes than I in fact have, I interpret the given experience dif— ferently, and consequently have different expectations about, and make different implicit predictions about, my future actual and possible experience. Thus, in influencing the way I classify what is presented to me, the interests and mental attitudes that I have influence the expectations that I have of future experience. In the context of Lewis' theory of the given, this is a perfectly harmless sense in which interests and attitudes influence our perceptual ex- pectations. For Lewis, whether those expectations are ful- filled or not will depend, not on the expectations themselves, but on what my future experience brings, and that is precisely what is determined by the unalterable given element in my experience. The second way in which interest and mental attitude can influence our perceptual experience raises questions with regard to Lewis' criterion of unalterability. In the illustrations to be cited below, having sufficiently strong expectations may actually provide experiences which ful- fill these expectations. What is given to us will be seen to be, under certain conditions, alterable by mental atti- tude. The illustrations used are fairly commonplace; they are drawn from ordinary experience sufficiently familiar to all of us. 19 The first example is discussed by William James in his Talks to Teachers. When we go to a theater in which a play is being performed in a foreign language we may ex- perience much difficulty in determining what the actors in the play have said. Our difficulty may be due to two dif- ferent conditions. First of all, it might be that we have a poor seat and the acoustics of the theater are somewhat less than ideal. Secondly, our relative unfamiliarity with the language of the play may put us in the position of not knowing what words to expect in what order and in what con- text. Our difficulty in hearing the words of the players may be due to one or the other or both of these conditions. But had the play been performed in the same theater under the same acoustical conditions, but this time in one's native language, one might have little or no difficulty in discerning the various speeches of the players. In knowing that language to a greater extent, one knows what words to expect in what order and in what context. The result is that under the same acoustical conditions one now hears clearly what one previously heard as muddled or perhaps did not hear at all. In this second case, one's prior ex- perience with the language enables one to know what to expect, and in knowing what to expect, to fill in missing or muddled sounds. Another phenomenon quite similar to this one is the filling in of missing words in proofreading. When a 20 particular word is left out of a manuscript, but it is a word whose occurrence is expected on the part of the proof- reader, it often occurs that the reader, in some sense, "sees" the missing word in his proof reading. In such cases one's expectations actually tend to provide those visual experiences which will provide fulfillment of the expecta- tion.26 Another similar kind of situation is reported by D. C. McClelland and J. W. Atkinson in "The Projective Expression of Needs: The Effect of Different Intensities 27 After a preliminary of Hunger Drive in Perception." stage during which faint images were shown on a screen, no images were projected on the screen and the subjects of the experiment were asked to describe what they had seen. There was a significant increase in the subjects' reports of seeing food related images with increases in food deprivation.28 The suggestion is that these and similar cases constitute counter examples to Lewis' claim that the given element in experience is unalterable by our interests and mental attitude. It should prove of interest to consider how Lewis could handle these apparently disconfirming cases. Since experiences of this kind are not uncommon, it is prima facie unlikely that Lewis would have failed to take note of them. Lewis could argue that the perceptual qualia "filled in" are given. They are simply interpreted 21 incorrectly if they are interpreted as a sign of a certain kind of thing, for example if those in proof reading are taken as the presentation of a typed word. But the pre- sentation is nonetheless the presentation of some other kind of thing. For example, it may be correctly interpre- ted as a real hallucination or a real illusion. For Lewis, it will be recalled, the given is the brute-fact element in illusion, perception and dream, without antecedent inter- pretation. While this kind of reply might be initially persua- sive, it is not in the end a convincing way of maintaining unalterability as a definitive criterion of the given. The reply sketched in the last paragraph does not really address itself to the crucial question of whether and to what extent our interests, in influencing our perceptual expectations, influence what is given to us. While there are places in MWO which can be construed as claiming that such presentations are given, nevertheless this reply does not show that they are independent of, in the sense of unalterable by, our mental attitudes as reflected in our expectations. But if he has not shown ttgt, then he has failed to show how being unalterable by interests and mental attitude can be a definitive criterion of givenness. If such expectation fulfilling qualia are given, then the element of givenness in some experience is alterable by interest and mental attitude. 22 With this particular avenue of reply closed to him, Lewis could argue that since the aspects of the particular experiences in question are alterable by interest and mental attitude (in fact they are created by such attitude), then for that very reason they are not part of the givenness of those particular experiences. They reflect rather that aspect of experience which is due to our own nature as active, interpreting beings. Now, this avenue of reply might force us to conclude that less is given to us in experience than we may have previously supposed, that more of experience is alterable by our will. But, this would still enable Lewis to take account of the seemingly trouble- some psychological examples without having to abandon the main tenets of his theory of the given: (1) that there is a given element in experience; (2) that this given element is unalterable by interest and mental attitude; and (3) that this given element in experience has, roughly, a sen- suous or feeling quality. That he need not give up the claim that there is a given element in experience can be seen from the following consideration. In claiming that the psychological consider- ations noted above indicate that the element of givenness is at times or partially alterable by interest and mental attitude, such a critic may be merely regarded as presup- posing some distinction between the given element in 23 experience and what reflects interpretation. These psycho— logical considerations can at best be used to challenge the particular way in which Lewis makes the distinction between the given and the conceptual elements in any par— ticular experience. They cannot be used to attack the distinction itself. The critic may be regarded as merely proposing an alternative (empirical) criterion for isola- ting the givenness of any particular experience. On such a criterion the given might be that element in perception which results from the (causal) effects of external (phys- ical) objects on my sense organs. As an alternative to the criterion of unalterability, Lewis would take exception to it for the following reason. The critic seems to be suggesting that whether there is or is not an identifiable given element in experience and what criterion to employ in making the distinction are em- pirical issues for which empirical evidence would be ap- propriate. It seems clear, however, that for Lewis neither the claim that there is a given element in experience, nor the claim that it is unalterable by interest and mental attitude, is an empirical issue at all. If it were, one would have expected to see Lewis marshalling empirical evi- dence. That is not the way he proceeds, however. His arguments tend to be dialectical rather than empirical. Unless we recognize an unalterable given element in experi- ence, it would be impossible to distinguish reality from 24 a well constructed fairy-tale. But if Lewis' claim that there is a given element in experience and that this given element is unalterable by interest and mental attitude are not empirical claims at all, then it is difficult to see the point of these psychological considerations. Their only point would appear to be that much of our experience is alterable, and thus ought not to be included as part of the givenness of any particular experience. It should be recalled that Lewis did not say that illusion, hallucina- tions, etc. are given; it is only the brute-fact element in them which is given. While Lewis need not, therefore, give up his princi- pal claim regarding the existence and unalterability of the given, nevertheless some reconsideration of some of his other important theses about the given might be in order. All we can be assured of at this point is that there is an unalterable given element in experience. How- ever,the above psychological considerations make it clear that it is quite difficult to determine in any particular experience exactly where the given ends and interpretation begins. Now if all Lewis were interested in doing were to defend the claim that there is a given element in experi- ence, this would not present any difficulty. However, one of Lewis' main reasons for emphasizing the role of the given element in experience is to emphasize its crucial role as the certain (incorrigible) datum element underlying 25 empirical knowledge. But if one cannot be certain exactly where to draw the line in any particular experience between the given and the non-given, then it is not clear how such a given could perform this epistemic function. We shall consider the thesis that the given element in experience is certain in Chapter III. CHAPTER II QUALIA AND PROPERTIES: THE SUBJECTIVITY AND INEFFABILITY OF THE GIVEN As noted in Chapter I, the general distinction be- tween the given element in experience and the element of interpretation is reflected in several distinctions made by Lewis. In particular, the more general distinction is reflected in the manner in which Lewis distinguishes be- tween sense qualia and the properties of objects. Unlike the objective properties of objects, sense qualia are sub- jective and ineffable. After presenting an exposition of the way Lewis draws the distinction between qualia and properties, several internal and external difficulties will be pointed out. We may begin by considering Lewis' rejection of a position which may be called the two kinds of empirical knowledge view. Roughly, this view distinguishes two kinds of empirical knowledge. One kind is direct knowledge of objects (sometimes called knowledge by acquaintance). Such knowledge is gained by the presentation of objects in experience and does not stand in need of verification. The second kind of empirical knowledge is propositional 26 27 knowledge or generalization (knowledge about) "which con- cerns more than can be given at one time and requires some 29 Lewis mental synthesis of what is temporally disjoined." argues that there is no knowledge by acquaintance and that all empirical knowledge, whether of generalizations or of particular objects, always transcends the given. According to Lewis there are two sources that have given rise to the view that we have knowledge by direct acquaintance. First of all it is assumed that there are simple concepts which denote simple qualities, where by a simple quality is meant a quality which can be exhibited in a single experience. I take it that by a single ex- perience Lewis means something like experience confined to a single specious present. None of these Conceptions are particularly clear, however. The second source of the view that we have knowledge by direct acquaintance is that the word "knowledge" is sometimes used to denote that con- templative state when one is completely absorbed in the given as an aesthetic object.30 In understanding his point here it must be clearly understood that Lewis does not want to deny that there is direct apprehension of the immediately given. According to Lewis, there are qualitative characters of the given (Lewis calls these qualia). Furthermore, these qualia are a sort of universal in the sense that they may be repeated and recognized in different experiences and different 28 presentations. But, it is Lewis' contention that confusion results if we refer to this kind of recognition as know- ledge, and if we fail to distinguish qualia from the proper- ties of objects. As Lewis, puts it: There are no 'simple qualities' which are named by any name; there is no concept the denotation of which does not extend beyond the immediately given, and beyond what could be immediately given. And without concepts, there is no knowledge . . . . The quale is directly intuited, given, and is not the subject of any possible error because it is purely subjective. The property of an object is objective; the ascription of it is a judgment which may be mistaken; and what the predication of it asserts is something which transcends what could be given in any single experience.31 The second basic reason why Lewis denies that we have knowledge by direct acquaintance is that we cannot be mistaken about that of which we are immediately aware and knowledge is not possible unless there is some cor- relative possibility of error. Acquaintance, understood as a purely contemplative absorption in the given, is not susceptible of error. There is much difficulty with Lewis' arguments in this connection. We shall turn to them in Chapter III after having discussed Lewis' view of qualia and properties. In order to provide a general introduction to what Lewis takes to be involved in our knowledge of objects, we can do no better than to quote at length how Lewis illustrates his conception by means of an example. At the moment, a certain 'that' which I can only de- scribe (in terms of concepts) as a round, ruddy, tangy- smelling somewhat, means to me 'edible apple'. Now my 29 ultimate purpose toward it may be an enjoyment of an ineffable taste. But that taste not being given, I need a conceptual go-cart to get me over the interval between this round, ruddy presentation and the end pro- jected by my purpose . . . . It is the function of mind to bridge these by assigning an interpretation through which it becomes related to, or a sign of, a correlation between certain behavior of my own and the realization of my purpose. This interpretation has the character of a generalization which has been learned. I phrase it by saying 'That (denoting the presentation) is a sweet apple (connoting among other things the possible taste).' If I should be completely absorbed in the first given, as an infant might, then I should form no concept, it would have no meaning, and no action, unless a merely instinctive one, would be evoked. An object such as an apple is never given; between the real apple in all its complexity and this fragmentary presentation, lies that interval which only interpre- tation can bridge. The 'objectivity' of this experi- ence means the verifiability of a further possible experience which is attributed by this interpretation.32 On Lewis' account of the distinction between the given and its conceptual interpretation, even to describe the presentation as round or tangy-smelling is already to have placed an interpretation on the given. The denotation of "round", as Lewis would put it, already extends beyond what is immediately presented and is significant of future actual and possible experience. One reason that Lewis gives for the claim that qualia are distinct from properties is that the same quale may on different occasions be the sign of different proper- ties, and the same property may on different occasions be 33 signalized by different qualia. Consider, for example, the property of being round. 30 The real roundness of the real penny is seen as all degrees of elliptical appearance . . . . The judgment that the penny is round may be made because, it 'looks round' or it may be made because, under given condi- tions which are understood, it looks elliptical. But the givenness of the appearance is not the givenness of Objective roundness. Given the elliptical appear- ance the judgment 'That is round' may be in error. Indeed, given the round appearance, the judgment may still be in error, as measurement with precision in- struments might reveal . . . . A penny run over by a railroad train will look round when held at a certain angle, while one which looks elliptical may be round.34 That we confuse a quale with the property it signal- izes may be due to a confusion in the use of language. In general, the name of the property is also assigned to the appearance of it under certain optimum condi- tions . . . . This use of language has its obvious practical motives, but it would be an extra-ordinarily poor observer who should suppose that what the name means in ordinary parlance is the appearance as such and not the objective property.35 Moreover, that the same expression may be used to name a property as well as an appearance of it does not entail that the expression has the same meaning in these two occasions of its use. Thus, the fact that the same expres- sion is used to name a property on one occasion and to name one of its appearances on another provides no grounds for identifying the property with the quale which signalizes it. There are places in MWO where Lewis says that whether we say that there are concepts of qualia because they are recognized, or that immediate apprehension of qualia should be called knowledge because of its indispensable role in our knowledge of objects is simply a matter of how we are 31 to use words like "concept" and "knowledge". His recommen- dation that we not so use them is due to the fact that in Lewis' view using the same words in the case of our appre- hension of qualia and in the case of our knowledge of ob- ‘jects may cause us to fail to observe the distinction between "qualia and the immediate awareness of them on the one hand and the properties of objects and our knowledge of them on the other."36 A second reason Lewis gives for closely observing the distinction between qualia and properties is that Qualia are subjective; they have no names in ordinary discourse but are indicated by some circumlocution such as 'looks like'; they are ineffable, since they might be different in two minds with no possibility of discovering that fact and no necessary inconvenience for our knowledge of objects and their properties . . . if one such could be lifted out of the network of its relations, in the total experience of the individual, and replaced by another, no social interest or interest of action would be affected by such substitution.37 This is an exceedingly difficult passage to inter~ pret in such a way as to make it consistent with some other positions that Lewis takes. In the first place, the claim that qualia are subjective appears to be inconsistent with his earlier claim that, without interpretation, the given is neither subjective nor objective. Such distinctions are simply later distinctions.38 Perhaps, however, we are being too literal here. Perhaps where he says that qualia, unlike properties, are subjective, he means by "subjective" something other than 32 what he meant by it earlier when he claimed that, without interpretation, the given is neither "external reality nor explicit self". One way in which it might be determined what Lewis intends to mean in saying that qualia are sub- jective is to consider the claim in the context in which he makes it. That is, in saying that qualia are subjective he means that they have no names in ordinary discourse and that they are ineffable. It should be pointed out that when Lewis says that qualia have no names in ordinary discourse tag that they are ineffable, he is not repeating himself. He is claiming that qualia are subjective in these two different senses: they cannot be named and they are ineffable. In saying that qualia are ineffable Lewis means something quite different from their having no names in ordinary discourse. This can be gleaned from the remarks that immediately follow the assertion that they are ineffable in the passage quoted above. On Lewis' account to say that qualia are ineffable is to say this because "they might be different in two minds with no possibility of discovering that fact and no necessary inconvenience to our knowledge of objects and their properties . . . ." It appears then, that the fact, if it is a fact, that qualia have no names in ordinary discourse is not given as a reason for claiming that qualia are ineffable. And conversely, that qualia are ineffable 33 (in the sense that Lewis is using that term here) is not not given as a reason for claiming that qualia have no names in ordinary discourse. I take it that Lewis is here giving two reasons for claiming that qualia as distinct from the properties of objects are subjective. The two reasons Lewis gives are (1) that qualia have no names or sometimes cannot be named and (2) that qualia as such are ineffable or incommunicable. The second reason can be read as stating that the content of experience cannot be communicated, but only its struc- ture. Thus, qualia themselves cannot be communicated, but only the relations between qualia. In the first place, by way of internal criticism, Lewis himself uses words to name given qualia. At least he leaves it open that there be such names. For example, in his earlier discussion of some of the reasons why it has sometimes happened that qualia and properties have not been distinguished, Lewis pointed out that the same ex- pression has been used both to name a property and the quale that signalizes it "under optimum conditions". But he gave no reason to support the claim that if we observe that the expression is being put to a different use in being used to name a prOperty then the use to which it is put to name a quale which signalizes that property under optimum conditions, then the quale still has no name in ordinary discourse. If we observe the distinction between 34 the different uses of the same expression, then it is still open to us to name the quale by means of one use of the expression. Lewis might object, however, that such names are not names "in ordinary discourse." But then it must be made clear exactly what the force of Lewis' claim amounts to. Moreover, the fact that on Lewis' account qualia must be repeatable if there is to be any experience of objects implies that such names would count as general terms since they are applicable to more than a single oc- currence of a quale. The strongest claim the Lewis can make is that qualia, unlike properties, have no names in ordinary discourse, not that they have no names, or not that they could not be given names for special purposes. But even this weaker claim is not consistent with Lewis' own use of words from ordinary discourse to name qualia. For example, when Lewis pointed out earlier that by the term "the given" he meant to refer to more than what is usually meant by "data of sense," he gave the following as an example of this "more": ". . . other qualities than the strictly sensory may be as truly given; the pleasant- ness or fearfulness of a thing may be as un-get-overable as its brightness or loudness . . . ."39 But to say that pleasantness and fearfulness are given is to name qualia by the use of perfectly ordinary words. 35 It might be objected, however, that this criticism misses the point of Lewis' claim that given qualia have no names. That is, Lewis might be understood to be claim- ing that given qualia qua given cannot be named, not that given qualia when taken as presentation of an object or objective property cannot be named. That is, Lewis might be claiming that qualia can be named only in so far as a quale is taken as a sign of the presence of an objective property. For example, in naming a particular quale by the term "round" we are not naming the quale as such, but we are naming it only in so far as it is taken as a sign of the property of being round. Moreover, it seems quite plausible that Lewis might very well have made this objec- tion to the kind of criticism suggested here. What I shall appeal to in answer to this hypothetical reply on Lewis' part makes use of a distinction which Lewis himself made earlier in order to distinguish qualia from properties; namely, that the same quale may, on different occasions, be the sign of different properties, and conversely, that the same property may, on different occasions, be signalized by different qualia. On this assumption, it is quite pos- sible that I might use the word 'round' to name a quale with no intention whatever of taking that quale as a sign of any objective property at all, or at least without taking it as a sign of the objective property of being 36 round. It does not follow, then, that I cannot name the given quale as such. Furthermore, Lewis' claim is incon- sistent with what he actually says about qualia. Now it may well be the case that the names of qualia are, chronologically speaking, first used as the names of the properties they signalize, and are used to name the qualia that signalize these properties only at a later date, or for some purpose other than the purely cognitive; for example, if I had an aesthetic purpose in mind. It seems to me that Lewis would admit this. For cognitive purposes it is not usually important to describe the content of our "immediate experience." Indeed, being overly concerned with accurate description of the given may in fact conflict with our interests of action, and thus on Lewis' account actually interfere with the very purpose of knowledge. While this all may very well be true, it does not provide any reason for denying that given qualia, as such, can be named. What about the second part of the claim that qualia are subjective; that is, is it true that qualia are inef- fable in the special sense in which Lewis uses that term? There are serious difficulties with what Lewis says on this point, as has been pointed out by at least one commentator.40 The objections that will be raised here are far more crucial in regard to the kind of analysis of knowledge and experience that Lewis adopts than those raised against the 37 claim that qualia cannot be named. That is, one might admit that qualia can be named, but still maintain on other grounds the distinction between qualia and properties. The objections that will be presented against the claim that qualia are ineffable are such as to raise doubts about the desirability of making the distinction between qualia and properties in the manner done by Lewis. Consider the reason that Lewis gives for claiming that qualia are ineffable. . . . they are ineffable, since they might be different in two minds with no possibility of discovering that fact and no necessary inconvenience to our knowledge of objects or their properties. All that can be done to designate a quale is, so to speak, to locate it in experience, that is, to designate the conditions of its recurrence or other relations of it. Such location does not touch the quale itself: if one such could be lifted out of the network of its relatiSns,‘in the total experience of the individual, and replaced by another, no social interest or interest of action would be affected.by such substitution. What is essential for understanding-and.for.communication is not the quale as such but that4pattern of its stable.relations in experience which is what is implicitly predicated wheniit istaken.as.the.sign..of.anobjectiveproperty.41 The italicized portion of the above passage presents considerable difficulty of exposition. It should help us in understanding Lewis if we refer to one other passage in which he makes remarks similar to those quoted immediately above. The passage occurs in Chapter III of MWO, where Lewis is attempting to show that for shared knowledge and community of meaning to be possible, no "qualitative iden- tity of experience" at the level of givenness need be assumed. The intended force of Lewis' point here is that 38 even if we grant "qualitative difference" at the level of givenness, no dire consequences follow concerning the pos- sibility of shared knowledge and community of meaning. Even for large and crude distinctions, what assurance is there that our impressions coincide? Suppose it should be a fact that I get the sensation you signalize by saying 'red' whenever I look at what you call 'green and vice versa. Suppose that in the matter of immed- iate experience my whole spectrum should be exactly the reverse of yours. Suppose even that what are sensa- tions of pitch mediated by the ear, were identical with my feelings of color-quality, mediated by the eye. Since no one can look directly into another's mind, and the immediate feeling of red or of the middle C can never be conveyed, how should we find it out if such personal peculiarities should exist? We could never find out so long as they did not impair the power to discriminate and relate as others do.42 There are three different but related claims in this passage that Lewis is making. Before proceeding to explain and evaluate them, particularly with respect to the question of whether they support Lewis' manner of dis- tinguishing qualia from properties, a few preliminary remarks are in order. In the first place, Lewis makes one claim that I do not wish to deny. The claim is that In acuity of perception and power to discriminate, there is almost always some small difference between the senses of two individuals, and frequently these differences are marked. It is only in rough and ready terms that we can reasonably suppose that our direct intuitions and images are alike. 3 I take this to be a fact, but a fact which is dis- coverable by perfectly ordinary means. For example, if you are stronger than I, there will be upper weight ranges 39 whose differences you may detect, but I may fail to detect. This can, of course, be found out either by comparing our discriminations of weight with the weight discriminations others may make, or by weighing the objects on a scale. Similarly, your range of hearing may be greater than mine, in such a way that what are audible sounds for you are not so for me. This fact is also discoverable, for example, by measuring the associated sound waves, or by noting that others can make the same discriminations that you do. As Lewis himself points out, this is often what we mean by intelligence, the ability to overcome perceptual idiosyn- cracies by indirect means. Thus, Lewis seems here to be admitting that (1) it makes sense to speak about qualita- tive difference at the level of given qualia and (2) that there are perfectly ordinary ways of gathering (empirical) evidence of such quale difference. In the second place, it should be pointed out why Lewis holds that such idiosyncracies of sense can make more difficult the attainment of, but can never render im- possible, community of meaning. That is precisely the first point which needs to be emphasized; idiosyncracy of intuition need not make any difference, except in the aesthetic quality of the experience of the one as compared with that of the other. Let us take it for granted (it seems fairly obvious) that the sense data of one are seldom precisely those of the other when we address our selves to the same object. That, by itself, will in no way impede our common knowledge or the conveying of ideas. Why? Because we shall still agree that there are three feet to the yard; that yellow is lighter than blue; and that middle C means a vibration of 256 per second. In other 40 words, if we define any one of the unit ideas or con- cepts which enter into the expression of our thought, we shall define it in the same or equivalent fashion, and we shall apply the same substantives and adjectives to the same objects. So far as we fail in these two things, our knowledge really will be different and the attempt to convey our thought will break down. These, then, are the practical and applicable criteria of common knowledge; that we should share common definitions of the terms we use, and we should apply these terms identically to what is presented.44 We may now proceed to isolate the three claims that Lewis wishes us to consider. Lewis appears to be suggesting that it someone had certain idiosyncracies of sense (in particular if color and sound qualia were exchanged in the manner indicated), then such idiosyncracy could not be detected unless it affected one's ability to discriminate and relate as others do. That under these conditions quale-difference could not be detected is taken as marking a difference between qualia and objective properties. That qualia may be thus different in two minds is given as a reason for claiming that qualia are ineffable; this in turn is taken as part of what it means to say that qualia, in contrast to properties, are subjective. Thus, the supposition that "qualia might be different in two minds with no possibility of discovering that fact and no necessary inconvenience to our knowledge of objects or their properties"45 is a crucial assumption in two closely related of Lewis' theses: (1) that qualia and properties are distinct, and (2) what is really the 'more general thesis of which (1) is a particular application, 41 that one can make a clear distinction between what is given and what is constituted by interpretation. I should now like to indicate what I take to be the three hypothetical claims that Lewis made in the passage quoted above. The first claim amounts to the following. We are to imagine that two of our color qualia are to be inter- changed. This difference between one individual's color qualia and other's color qualia will not be discovered, because, for one reason, we shall all still call apples red and grass green, even though our color qualia will be diverse. The assumption is that such readjustment in our color qualia will simply be confined to the exchange of the two qualia, and thus will not affect our other color qualia nor the judgments we shall make about relations between colors. As Lewis puts it: . . . if one such could be lifted out of the network of its relations in the total experience of the in- dividual, and replaced with another, no social interest or interest of action would be affected by such substi- tution. What is essential for understanding and for communication is not thegquale as such but that pattern of its stable relations in experience which is what is Implicitly predicated when it is taken as the sign of an objective property.45 The second claim invites us to envisage the possi- bility that, in the matter of immediate experience of color qualia, one's spectrum would be exactly the reverse of another's. Once again, it is claimed that such a difference in immediate experience would not affect the relations 42 between color qualia in our experience, and thus no social interest or interest of action would be affected. There- fore, such difference in immediate experience would go undetected. The third claim resembles the first in that it ev- visages the simple interchange of two qualia. It differs from the first in that the two qualia that are to be ex- changed are qualia of different sensory modes; one is a color quale, the other is an auditory quale. It seems to me that this presents the most serious difficulties for Lewis. Now it seems to me that Lewis is correct in point- ing out that what is essential for common understanding is the pattern of a quale's relations in experience. In any case, what I should now like to argue is that the imagined difference in sense qualia that Lewis has imagined in fact will be detectable, and by perfectly ordinary means. Moreover, they will be detectable because a sense quale is not as independent of its relational context as Lewis seems to assume. Consider the first claim. That is, suppose that in my immediate experience of color qualia, red and green are interchanged. Lewis supposes that this difference will not be detected because we will fulfill the two criteria of common understanding. I will still apply the term "red" to the same things that everyone else applies the term "red" to, even though when I apply the term "red" I shall actually 43 be presented with the color quale you are presented with when you apply the term "green". Secondly, I will define "red" in the same way that others do by locating it in the 47 If these two criteria same place in the color spectrum. are fulfilled, then on Lewis' account we share a common color concept, even though our immediate experiences of color qualia are quite different. What I should like to point out in criticism of the first imagined situation is that the difference in our color qualia will be altered in discoverable ways and that a consequence of the exchange of two color qualia will not be limited to the simple fact that you will be presented with a qualitatively different sense quale whenever you and I see green. Now this exchange may happen in two dif- ferent ways. First of all it might happen that some time in the middle of my life as a result of some neurological disorder I suddenly see things as green that I formerly saw as red and conversely. This, of course, will be easily discoverable, for I shall, at least for a time, apply the color terms quite differently from you. The denotation of the two terms will be exactly the reverse of what they formerly were. Thus this kind of envisaged exchange of color qualia would be discoverable. The kind of situation Lewis imagines would be more like my congenitally being presented with qualitatively different color qualia from those presented to you when we are both in the presence of 44 ‘the same color property. Thus, I would learn to apply color words in a different way. Consequently, while actu- ally being presented with a green quale, I would report that I am seeing red and conversely. Moreover, I would locate my green in the color spectrum in exactly the same position that everyone else located red, and vice versa. I would like to suggest that even under these conditions, there will be ways of discovering that our presented color qualia are diverse. In the first place, Lewis has overlooked the role that color comparison might play in discovering quale di- versity in the first imagined case. If a color quale is thought of as a particular E2292 of color and if we are invited to imagine exchanging one £2292 of green for one EEESE of red, this will be discoverable by virtue of the relations of resemblance that these two qualia will bear to other shades of red and green. A consequence of the exchange of these two color qualia will be that my judgment concern- ing the resemblance between colors will, at one point in the color spectrum, be markedly different from yours and this fact will be taken as evidence of quale diversity. In this case, then, a color quale cannot be "lifted out" of the stable network of its relations to other qualia. Its being so "lifted out" is detectable in virtue of its effect on the relational context of the quale's occurrence in experience. 45 Lewis' second claim, the inverted spectrum case, is somewhat more plausible. Suppose, that is, that "in the matter of immediate sense qualia my whole spectrum should be exactly the reverse of yours." Imagine that A has this perceptual idiosyncracy; must it go undetected as Lewis supposed? I should like to make two remarks which have a bearing on this supposition. Such perceptual idiosyncracy as is envisaged in the inverted spectrum case will affect the relations between color qualia. Now, under these conditions Lewis might still wish to maintain that the difference in the relational as- pects of the given will and must go undetected. This may well be the case, but if it is, then Lewis will have to reconsider his reason for claiming that such exchange of color qualia will go undetected. That is, Lewis cannot give as a reason for claiming that qualia are ineffable the fact that the juxtaposition of color qualia will not affect the relational context of a quale's occurrence in experience. In the particular case at hand, the juxtaposition will af— fect the relational context of the quale, even though such an effect will not be discoverable. Thus, if the import of Lewis' claim is that qualia are ineffable if and only if their juxtaposition does not affect the relational con- text of their occurrence in experience, then Lewis will have to deny that the inverted spectrum case justifies the claim that qualia are ineffable, since in this case the 46 relational context of experience will have been altered. If it is the case that the relational context of experience will be altered by such juxtaposition, then it follows that qualia are not ineffable in Lewis' sense of ineffability. One way in which Lewis might avoid this objection would be for him to deny that not affecting the relational context of experience by their juxtaposition is a necessary and sufficient condition of a quale's being ineffable. He could not claim that it was a necessary condition since that would entail that it is not the case that qualia are ineffable, since the inverted spectrum case will alter the relational context of the experience of color qualia. Thus, he will have to claim that it is only a sufficient condi- tion of a quale's being ineffable. But even here Lewis is on weak grounds; his claim will then be that if a quale's juxtaposition will not affect the relational context of its occurrence in experience, then such a given quale will be ineffable. However, in both the first and the second case envisaged by Lewis' claim and in the third as I shall argue presently, such juxtaposition of qualia would affect the relational context of the quale's occurrence in experi- ence. Thus, none of the cases envisaged by Lewis' claims fail to affect the relational context of the quale's occur- rence in experience. Thus, the possibility of such exchange of qualia provides no grounds for Lewis' claim that qualia are subjective in the sense of being ineffable or incommuni- cable. 47 The third claim we are asked to entertain by Lewis seems to me fraught with the same general difficulty as the first; namely, that the kind of exchange of qualia and the re- sulting difference is our immediate experience which the sup- position envisages cannot be limited to the simple exchange of qualia and will consequently be detectable. Suppose that when you and I both report that we are seeing something red, I am presented with the sound quale you are presented with when you hear Middle C. And conversely, when you and I report that we are hearing Middle C, I am being presented with the color quale you are presented with when you see the color red. Must this quale diversity go undetected? No, and for much the same reason that the exchange of two color qualia will not go undetected. For example, what would I reply to you when both of us are looking at a brown table with a little red in it, or when both are looking at something red with a little brown in it? It seems to me that it is quite difficult for me to know what to report and that you would take this difficulty on my part to be evidence that my experience of sounds and colors is quite different from your own. You may not be able to tell exactly what the difference is, but you would surely know ttgt there was a difference. Thus, in the situation envis- aged by the third claim, that there are differences between the immediate experiences of you and me will be something detectable. Consequently, the possibility that this third 48 situation were to obtain does not establish Lewis' claim that qualia are ineffable. Once again, the difference in my experience will not be limited to the simple exchange of qualia, but such exchange, contrary to Lewis' claim, will always "touch the quale itself." What Lewis is suggesting about the nature of qualia presupposes a kind of atomism about qualia and presentations which is untenable. He is claiming that the only thing of significance for knowledge is the relational properties of sense qualia. So far as their intrinsic character is con- cerned, it makes no difference whether such intrinsic char- acter is presented as identical in two minds or not. So long as the relational properties remain the same, no cogni- tive interest or interest of action would be thwarted. Lewis' suggestion that it would be possible to exchange qualia without changing their intrinsic character, but only their relational properties, seems untenable. He is claim- ing a degree of logical independence of a quale's intrinsic character from its relational properties which will not bear scrutiny. One simply cannot change a sense quale's rela- tional properties without, as Lewis puts it, touching the quale itself. In this sense, the kinds of relations which, for example, color qualia bear to one another are internal relations in the sense that to change the relational proper- ties of the terms related is to change the intrinsic 49 character of the qualia themselves and to change the intrin- sic character of a quale is, necessarily, to change its relational properties as well. The manner of this related- ness emerges in several ways. It has already been suggested that the relations of resemblance that the qualia signaliz- ing various shades of the same (generic) color are internal to the qualitative character of the qualia themselves. To alter these relations of resemblance would not be just to alter the relational context of the quale's occurrence in experience, but would be to turn it into a different quale altogether. Such color qualia are not separable, not even conceptually, from their relations to other color-qualia. Furthermore, these relations are not limited to re- lations among qualia of the same sensory mode. Even if one could be aware of a certain color quale, one would have to be aware of a shape, involving spatial relations and color contrasts which set the presentation off from its background. This necessary connection between being colored and being extended is what marks the absurdity of envisaging an ex- change of a color-quale and a sound-quale. The quale which signalizes Middle C has no such necessary connection with the notion of shape. A comparison of Kant and Lewis on this point may be instructive. Lewis' distinction between qualia and the relations between qualia seems intended to mark a distinction similar to Kant's between respectively the material and 50 formal elements in perceptual intuition, the material ele- ment being contributed by the power of sensing, the formal element (or element of order) being contributed by the mind's a priori categories and the a priori forms of intui- tion, space and time. One crucial difference between Kant and Lewis on this point is that, unlike Kant, for Lewis the formal element in experience is all strictly speaking con- ceptual. In Chapter VII of MWO he specifically and explic- itly rejects the notion of a priori limitations on the form of our passive receptivity. The point of importance here is that what Lewis has in mind when he speaks of sense-qualia are such that they cannot be isolated from their membership in ordered rela- tions. That is, qualia cannot be located except in terms of their relation to other qualia. That Lewis was reluctant to admit the existence of such necessary relations may be due to his reluctance to admit that there are synthetic a priori truths of the sort these necessary relations imply. Lewis could be defended from this criticism in the following way. Far from being a criticism of Lewis, the above remarks serve only to highlight the main points he is trying to make. All of this simply goes to support Lewis' claim that for cognitive purposes it is only the relational context of the quale's occurrence in experience that is important. Its intrinsic character is irrelevent for empiri- cal knowledge. However, in order to explicate the distinction 51 between the given and its conceptual interpretation, Lewis is supposing that a certain conceptual distinction can be made between qualia without regard to their being ordered in certain relations and qualia regarded as members of such series. In the first respect they reflect the given element in experience; in the second they represent its conceptual interpretation. The suggestion being made here is that one cannot regard qualia (e.g. color-qualia) apart from the re- lations they bear to other qualia, since such relations are internal to the qualia themselves. It should not be forgotten that this whole discussion of the incommunicability of sense qualia is introduced in order to make a distinction between the given and its con- ceptual interpretation. The givenness of sense qualia is understood to be that content or intrinsic character of .qualia apart from their relations to other qualia. That is, Lewis still presupposes that this distinction can be made between the "content" of a given quale and its relational properties, if he wants to argue that its "content" is in- communicable, but its relational "structure" is communicable. The point of the above objections is that no such distinction can be made and that consequently Lewis cannot use such a distinction to explicate the more general distinction be- tween the given and its conceptual interpretation. It could be suggested that whatever effect this may have on Lewis' account of distinction between the given and its conceptual 52 interpretation (or between the material and formal elements in experience), the above arguments cannot be used to chal- lenge the distinction itself, since in fact they presuppose it. That we cannot simply apprehend or be aware of a sense quale apart from its relation to others shows only that nothing ever stands merely in the relation of givenness to the mind. It does not prove that nothing is given at all. Be that as it may, for Lewis one way of conceptually iso- lating the given element in experience from the element of interpretation is to distinguish sense qualia from the re- lational context of their occurrence in experience. How- ever, the result of such conceptual isolation does not provide us with a qualitative core of unrelated sense qualia. The claim that qualia are subjective in a way in which the properties of objects are not provides one way in which Lewis tries to make a distinction between what is given and what is constituted by interpretation. I have tried to show that Lewis cannot make out his claim that qualia are, in his sense, subjective. He therefore has failed to provide in the distinction between qualia and properties an explication of the more general distinction between-the~given and its conceptual interpretation. Lewis also offered an additional criterion in terms of certainty which was also intended to mark a distinction between the given and its conceptual interpretation to which we shall turn in the following chapter. CHAPTER III SENSE CERTAINTY Whether our awareness of the given is certain and whether perceptual judgments are ever more than probable are questions long associated with the epistemological work of Lewis. He seems to have held consistently that one either had to accept some such account as his of the certainty of our awareness of the given or else have recourse to some kind of coherence theory of truth and coherence theory of 48 And, moreover, it was Lewis' view that verification. both the coherence theory of truth and the coherence theory of verification face insurmountable difficulties. Before proceeding to evaluate Lewis' treatment of sense certainty, the particular difficulties which face both the interpretation and criticism of Lewis should be noted. These difficulties arise as a result of the fact that Lewis rarely treats in isolation the various epistemological prob- lems which he intended to solve with his theory of the given. For example, in his article "The Given Element in Experi- ence" Lewis lumps together three things that his account of sense certainty is to do: (1) to provide, in addition to that of unalterability, a criterion for isolating, 53 54 analytically, the given from the interpretation which, in the light of past experience, we put upon it; (2) to provide an account of the justification of our empirical judgments; and (3) to provide an account of the truth of the knowledge we gain by experience. Since I am mainly concerned with the first of these issues, in what follows I shall try, as far as possible, to keep these three issues separate. One further point should be noted. In MWO Lewis did not gay that certainty is a criterion of the given. His procedure is more that of noting, after isolating the given by the criterion of unalterability, that our awareness of the given is certain. To criticize a criterion of the given in terms of certainty would be to invite a charge of misin- terpretation. It could be replied on behalf of Lewis that he never intended certainty to be such a criterion. Happily, I need not base my case on this discussion alone, for in AKV and in "The Given Element in Experience" Lewis in fact proposes certainty as a criterion of the given.49 The discussion shall proceed in the following manner. First of all, I shall consider the various arguments Lewis proposes in MWO and elsewhere on behalf of the claim that our awareness of the given is in fact certain. I shall not consider a parallel argument (let us call it the logical argument) that Lewis presents to the effect that if any empirical knowledge is to be even probable, something 55 disclosed in experience must be certain. I shall neglect this argument for two reasons. First of all, the present discussion of the alleged certainty of our awareness of the given is concerned mainly with certainty as a criterion of the given and a discussion of the logical argument would take us too far afield. To present an adequate discussion of this argument would require a much more thorough treat- ment of the nature of evidence and probability than will be presented here. Secondly, if, as I shall argue, no sound argument consistent with other things in his account of empirical knowledge is forthcoming from Lewis on behalf of the claim that awareness of the given is in fact certain, then the logical argument will be in need of reconsideration. If it were true that no empirical knowledge claim could be justified as even probable unless something disclosed in experience were certain, then that our awareness of the given is less than certain would entail that no empirical knowledge claim is ever justified, since nothing in experience but the given could be certain. This kind of skepticism would surely be unacceptable to Lewis. Whether there is something wrong with the logical argument or whether the difficulty lies in some other aspect of Lewis' account of knowledge are ques- tions we shall not get into here. Secondly, in considering Lewis' claim that awareness of the given is certain in the context of his arguments in support of the claim that all (full-fledged) empirical 56 judgments, other than the terminating judgments of AKV, are at best probable, I shall present two different objections to Lewis' View. (1) I shall try to show that some peculiar consequences follow from the reasons Lewis gives for claim- ing that we do have certainty with respect to our awareness of qualia, but that we can never be certain about our know- ledge of objects and their properties. While not a compel- ling objection to Lewis' view of empirical knowledge, this may help to clarify what he meant by certainty. (2) I shall try to show that the argument which Lewis seems to have taken as the strongest in favor of the claim that our knowledge of objects is never more than probable is such that it entails that our awareness of the given is also never more than probable. Thus, given some of the other elements of Lewis' epistemology, certainty will not provide a criterion for distinguishing the given from its conceptual interpretation. This will bring to a close the sustained attack on Lewis' attempt to distinguish the two elements in experience. Before proceeding to a discussion of these matters in MWO, it will be helpful if we point out what Lewis refered to as "the different types or phases of knowledge." In MWO he claims that the following distinctions need to be made. . . . (l) the immediate awareness of the given, such as might be reported in statements like 'This looks round‘, IThis feels hard', 'This tastes sweet'; (2) knowledge of of presented objects, such as is expressed by, 'This is hard', 'This is a sweet apple', 'This penny is round'; (3) the a priori elaboration of wholly abstract concepts, like the formulations of pure mathematics, apart from 57 any question of possible applications; (4) the categori- cal knowledge of interpretive principles and criteria of reality, which is that form of a priori knowledge which arises when concepts have a fixed denotation and are applied to the given; (5) empirical generalizations, which are universal but not a priori. It would be inaccurate to say (and Lewis does not say) that each of these five represents a type of knowledge, for it is claimed by Lewis that what is represented by (l) is not a type of knowledge at all. Lewis took this view in both MWO 51 In this chapter we will be concerned only with and AKV. the alleged differences between (1) and (2). In AKV Lewis speaks of three types of apprehension rather than phrases or types of knowledge. These three types of apprehension are: . . . (l) of directly given data of sense (not excluding the illusory); (2) of what is not thus given but is em- pirically verifiable or confirmable; (3) of what is im- plicitly or explicitly contained in or entailed by meanings.52 The fact that Lewis here speaks of three types of apprehension rather than five phases of knowledge does not signify a change of position, so far as I can tell. Know- ledge of presented objects and of empirical generalizations are still to be distinguished, but both are recognized as that type of apprehension which is verifiable or confirmable. Thus, phases of knowledge (2) and (5) are cases of type (2) apprehension. Phases (3) and (4) are cases of type (3) ap- prehension, giving rise to a priori knowledge. Phase (1) of knowledge is a case of type (1) apprehension, but is not 58 itself a case of knowledge. Why Lewis denies that type (1) apprehension does not give rise to knowledge we shall see shortly. He gives several reasons for making this claim, some of which at least appear to be incompatible. Let us begin, then, by considering what Lewis has to say about these matters in MWO. One reason Lewis gives for claiming that awareness of the given is certain arises in his discussion of the alleged distinction between qualia and the properties of objects. He makes his point in the following way: The quale is directly intuited, given, and is not the subject of any possible error because it is purely sub— jective. The property of an object is objective; the ascription of it is a judgment which may be mistaken; and what the predication of it asserts is something which transcends what could be given in any single experi- ence.53 ' First of all, it is the apprehension or intuition of a quale which is in question, not the apprehension of a presentation. On Lewis' account the objects of such appre- hensions are ontologically of two different kinds. A quale is, as Lewis puts it, a sort of universal. "But although such qualia are universals, in the sense of being recogniz- able from one to another experience, they must be distin-~ 54 guished from the properties of objects." A presentation of a quale, then, is the apprehension of a sort of univer- sal.55 A presentation, on the other hand, is not a sort of universal, but is, as Lewis puts it, "an event and histori- "56 cally unique. Apprehension of a presentation, then, is 59 the apprehension of an event. Thus we may take Lewis as claiming that apprehension of a quale (a sort of universal) is certain in the sense that it "is not the subject of any possible error." Such apprehension of qualia is not subject to error because no claim is made which could be true or false, cor- rect or incorrect. Apprehension in the sense in which a quale can be apprehended is not propositional. It takes a thing, not a fact, state of affairs or proposition as its 57 Such apprehension could not be construed as making object. an assertion at all. Thus, the role of apprehension of qualia is that of a kind of primitive noticing which, while not itself a case of knowledge, is a necessary condition of knowledge. Indeed, a remark that shortly follows the passage here in question suggests just this. "Qualia," says Lewis, "are universals, and they are universals such that without the recognition of them by the individual nothing presented in experience could be named or understood at all."58 Lewis appears to be claiming that apprehension of the given (he explicitly refers to apprehension of qualia rather than of presentations) is not the subject of any possible error. Now consider the reason that Lewis gives for making the claim. Apprehension of a quale is not the subject of any possible error "because it is purely subjective." Assuming that the distinction between qualia and properties 60 can be made out, there are still some difficulties in claim- ing that apprehension of qualia is certain because subjective. In particular, why should Lewis have thought that the subjec- tivity of qualia entails the certainty of our awareness of them? His reason may have been the following. In order to see why Lewis took the subjectivity of qualia to entail the certainty of our apprehension of them, it must be made clear what Lewis understood subjectivity to consist in. Since the polar opposite of "subjective" is "objective," it should be helpful to look at Lewis' char- acterization of the term "objective." Consider the following characterization of objectivity given shortly before the pas- sage we are presently considering: "The objectivity of this experience (the presentation of a round, ruddy somewhat which is interpreted as the presentation of an edible apple) means the verifiability of a further possible experience 59 To say that which is attributed by this interpretation." something is subjective would then amount to saying that whatever it is that is being claimed to be subjective is not verifiable by a further experience, since there is no inter- pretation by means of which further experience would be relevant to determining its occurrence. (Recall that for Lewis the real object is the given conceptually interpreted.) There are difficulties here, but they will be treated in the context of the argument Lewis proposes in defense of the probable character of all empirical knowledge. 61 Lewis goes on to contrast the apprehension of qualia with the perception of (he actually says ascription of) properties. The property of an object is objective (i.e. its presence or absence is verifiable by the further experi- ence implicitly predicated in the making of some interpreta- tion). The ascription of a property is a judgment which can be mistaken. Its predication asserts something which "tran- scends what could be given in any single experience." As we shall see shortly, for Lewis apprehension of a quale is not a judgment, apprehension of it asserts nothing which tran- scends the given experience itself, which is to say the quale's presence, unlike that of the property's, is not veri- fiable (or Lewis sometimes says it stands in no need of veri- fication). This first passage which has been selected for discussion, then, seems to be making the following claim: our apprehension of qualia cannot be mistaken (or is cer- tain). The claim is justified by the following premises: (1) qualia, unlike properties, are subjective. In order to avoid committing Lewis to claiming that qualia are private, or mental, being subjective is here taken to mean "is not verifiable by any further experience." Moreover, apprehen- sion of qualia is not verifiable because such apprehension has no consequences for further experience which might be relevant to determining the truth of such apprehension.60 (2) Such apprehension of qualia is not verifiable because the mere apprehension of qualia, with no interpretation, 62 fails to relate the apprehended quale to further actual and possible experience. It is only when some concept with fixed meaning is applied to the given that such relations that the apprehended quale does have to further experience can be noted and can have any relevance to the ascription of a prop- erty in the knowledge of an object. And finally, (3) such apprehension is not a judgment which could be mistaken be- cause such apprehension is not a judgment at all. For Lewis, the perceptions of the objects of the thick experience of everyday life are themselves judgments. The apprehension of a merely given quale is not a judgment, since at the very least to judge is to have applied a concept to the given, which is to have already gone beyond what could be given in such apprehension. gama of the obscurities of Lewis' position here may be removed by considering some of the other contexts in which apprehension of the given is considered. Shortly after the passage quoted above, Lewis makes the following remark. Apprehension of the presented quale, being immediate, stands in no need of verification; it is impossible to be mistaken about it. Awareness of it is not judgment in any sense in which judgment may be verified; it is not knowledge in any sense in which 'knowledge' connotes the opposite of error.61 Here he is claiming not that apprehension of qualia is un- verifiable, but that such apprehension "stands in need of no verification." I take Lewis to be saying that it is for this reason that it is impossible to be mistaken about such 63 apprehension. Moreover, since such apprehension stands in no need of verification, it cannot be a judgment. Making a judgment, like drawing a conclusion, is something that one does. Moreover, it can be done correctly or incorrectly. As the drawing of a conclusion may be checked, so a judgment may be verified and such verification may reveal the judg- ment to have been false. But merely apprehending a presented quale is like simply registering something; no inference or interpretation, explicit or implicit, is made. The view seems to be that knowledge is a kind of structure, each new level of which is added by the making of a judgment,interpretation or inference. Moreover, any time such a judgment is made there is always a chance of mistake. But at the base of the structure there is the apprehension of qualia, where no mis- take is possible, since no judgment is made. It is assumed by Lewis that if no judgment is made, if no inference is made, or if no interpretation is put upon the given quale, then no mistake can be made. Mistake enters only when in- ferences, judgments or interpretations are made. This might well be questioned, for there are other things which, while not themselves pieces of knowledge nor cases of inference or judgment, nonetheless play a necessary role in knowledge and which furthermore may be done successfully or unsuccessfully. For example, one may fail to notice something and such fail- ure is surely a kind of mistake. Moreover, apprehension of given qualia is, in Lewis' account, a kind of pointing. 64 But this is still to have granted Lewis part of his case, namely that such apprehension of the presented quale is not knowledge. It is to have granted his claim that while we are acquainted with sense qualia, there is no knowledge by acquaintance. In addition, the mistake involved in failing to notice would itself be determined by-some mistaken inter- pretation or inference. In fact the failure to notice would be a direct result of mistaken interpretation through the erroneous expectations such interpretation would suggest. There is one passage in which Lewis is defending his view that qualia cannot be named that sheds some light on why Lewis held that apprehension of a quale is not verifiable. I shall quote only that portion of the passage which is of concern to the verifiability of the apprehension of qualia. If concepts are to be articulate and meaningful, then the application of them must be something verifiable; which means that what they denote must have a temporal spread. Not a momentary presented quale but an ordered relationship of such is the least that can be meaning- fully named.62 Lewis is here arguing that if anything is to be named or if a concept is to apply to it, it must have a temporal spread. It must exist "longer" than the specious present, since only then could the application of a concept be verifiable. But given qualia qua given exist only in the specious present. Thus concepts cannot be meaningfully applied to them. Since without concepts there is no knowledge, the mere apprehension of given qualia cannot be knowledge. 65 This seems to be the bulk of Lewis' case for the certainty of our apprehension of the given (really our ap— prehension of qualia).63 I shall merely cite in a footnote several other places in MWO where the issue is mentioned. For the most part they simply repeat what has already been discussed.64 We may now turn to what Lewis had to say about these matters subsequent to the writing of MWO. There is, I think, little change in Lewis' views on the issue of sense certainty. The value, for our purposes, of these later passages is that some of them come much closer than MWO does to using certainty as a criterion of the given element in experience. For example, the following two pas— sages, the first taken from AKV, the second from Lewis' 1952 paper "The Given Element in Experience," Suggest certainty as a criterion for isolating the given from its conceptual interpretation. The point is simply that there is such a thing as experi- ence, the content of which we do not invent and cannot have as we will but merely find. And that this given element is an element in perception but not the whole of perceptual cognition. Subtract, in what we see, or hear, or otherwise learn from direct experience, all that con- ceivable could be mistaken; the remainder is the given content of the experience inducing this belief.65 When I perceive a door, I may be deceived by a cleverly painted pattern on the wall, but the presentation which greets my eye is an indubitable fact of my experience. My perceptual belief in a real door, having another side, is not an explicit inference but a belief suggested by association; nevertheless, the validit of this inter- pretation is that and that only which could attach to it as an inductive inference from the given visual presenta- tion. The given element is this incorrigible presenta- tional element; the criticizable and dubitable element is the element of interpretation.66 66 The criterion which is suggested by these two passages is the following: x is the content of the given element in experience if and only if the apprehension of x is certain (or incorrigible, or indubitable or cannot be mistaken). The remainder of what Lewis has to say about sense certainty seems to be simply a repetition of views which he had defended earlier in MWO and which we have already pre- sented. I shall simply quote some of these passages in AKV which are most reminiscent of those passages in MWO which we have already discussed. Consider, for example the following passages: . . . the content of immediate awareness must be recog- nized as pertinent to knowledge, whether it be regarded as included in knowledge or not. Our direct experience of sense--and of dream and illusion as well--have a character of absolute specificity and of sheer datum. And without such data there could be no empirical know- ledge. But the presentation of what is thus immediately given cannot be called knowledge if that term be re- stricted to what is verifiable and stands in need of confirmation.67 When this direct awareness of the given is separated from any interpretation put upon it, it becomes evident that such apprehension neither has nor calls for any verifica- tion. The specific character ofgthe presentation content, and the givenness of it is, so to say, its own verifica- tion.68 Other pertinent passages include the following: Immediate apprehensions of sense possess certainty--if we are careful to restrict ourselves to just the directly given content and as it is given. But in that case, do they mean or point to anything not contained within the cognitive experience? And it is at least doubtful whether such apprehensions have a ground or reason. Perhaps we should say that they do not need one, being self-justifying or self-evident.69 67 Apprehension of the immediate--of the sensuously given-— is something which is essential to empirical knowledge; but it is not itself knowledge in the sense that there could be mistake about it. The contrary supposition could only arise through failure to mark off sharply what is genuinely given from interpretation which normally ac- companies it or inference which would usually (and in most cases validly) be drawn from it.70 Thus in both MWO and AKV Lewis holds that apprehension of given qualia is certain (or cannot be mistaken). He takes this view for the following reasons (not all of which are independent). ‘(1) The content of such awareness (i.e. qualia) is subjective; to say something is subjective means (in part) that without interpretation its existence is not verifiable by further experience or is not related by interpretation to any further experience. (2) Predication of a property is a judgment, and therefore can be mistaken, while the apprehension of a quale is not a judgment. (The assumption is that if an apprehension is not a judgment, then it cannot be mistaken.) (3) Predication of a property asserts something beyond or which transcends what could be given in any single experi— ence. Apprehension of a quale makes no such assertion. To say that something transcends a single experience must not be understood to mean that it transcends experience alto- gether. Lewis would never make such a claim. For Lewis, something may transcend some particular experience without transcending experience altogether. Something transcends a 68 particular experience if and only if it is related, by in- terpretation, to further particular experiences. (4) Apprehension of a quale is not verifiable, since such apprehension fails to relate the quale in question to further experience. (5) Apprehension of a quale stands in need of no verification, since it is self-justifying, self—evident or sometimes is its own verification. At least (1), (3), and (4) are not independent since they all have to do with whether or not the apprehension of a quale relates the quale apprehended to further experience. I hope to show that there are difficulties here after having considered Lewis' argument for the merely probable character of empirical knowledge. In MWO and in AKV Lewis purports to show that no empirical judgment can ever be certain. The argument by means of which Lewis attempts to show this will be called 71 Before proceeding to evaluate the verification argument. Lewis' arguments for the certainty of the apprehension of qualia, I should like to sketch Lewis' version of the veri- fication argument. As I shall subsequently point out, the verification argument and the defense of sense certainty are closely related. Moreover, we should learn indirectly what Lewis meant in claiming that the apprehension of qualia is certain or cannot be mistaken if we learn why, in Lewis' view, empirical judgments can never be certain and always 69 admit the possibility of error. Presumably, some of the differences between apprehension of given qualia and empiri- cal judgments will be definitive of what Lewis meant by certainty. Since the argument is so very well put by Lewis him- self, and it is very doubtful that I could improve upon it, I will quote at length the relevant passages from MWO and .AKV that constitute Lewis' statement of the verification argument. Before doing that, however, it is important that we note the kind of empirical judgment which Lewis claims is never certain, but at best only highly probable. For Lewis, besides our apprehension of qualia, there are two kinds of empirical judgments, which he refers to as terminating and non-terminating judgments respectively. Terminating judgments can be expressed as conditional state- ments, each constituent statement of which is expressed in what Lewis refers to as expressive language, the use of language to simply express our immediate experience of the data of sense. Terminating judgments do not themselves simply express the immediately given data of sense. They make implicit predictions about further experience condi-v tional on my acting in certain ways. The general form of a terminating judgment would be, then, "A certain quale or complex of qualia being given, if I seem to be doing x, then a certain quale or complex of qualia will result," or "S being given, if A, then E." The verification of such 7O judgments is complete and decisive, according to Lewis, once the hypothesis of this conditional is made true by my action, since the judgment asserts nothing beyond the content of that passage of experience. Thus, such judgments are said to be terminating.72 If, however, we take a different kind of empirical judgment (e.g. "A piece of white paper is now before me"), the circumstances are quite different. Such judgments are called by Lewis objective beliefs. It is these judgments which the verification argument is intended to show can never be certain. Such objective or non-terminating judgments are confirmable or verifiable because they are translatable into terminating judgments. But the verification argument intends to show that the verification of any finite number of the terminating judgments "implied" by the non-terminating judg- ment is not sufficient for the "complete" or "decisive" verification or certain assurance of that non-terminating judgment.73 Moreover, while the procedure of verifying term- inating judgments is completeable, the process of verifying non-terminating judgments is not. In AKV the verification argument is stated as follows: In the second example, as we considered it, what was judged was an objective fact: 'A piece of white paper is now before me.‘ This judgment will be false if the presentation is illusory; it will be false if what I see is not really paper; false if it is not really white but only looks white. This objective judgment also is one capable of corroboration. As in the other example, so here too, any test of the judgment would pretty surely involve some way of acting--making the test, as by 71 continuing to look, or turning my eyes, or grasping to tear, etc.--and would be determined by finding or fail- ing to find some expected result in experience. But in this example, if the result of any single test is as expected, it constitutes a partial verification of the judgment only; never one which is absolutely decisive and theoretically complete. This is so because, while the judgment, so far as it is significant, contains nothing which could not be tested, still it has a sig- nificance which outruns what any single test, or any limited number of tests, could exhaust. No matter how fully I may have investigated the objective fact, there will remain some theoretical possibility of mistake; there will be further consequences which must be so and so if the judgment is true, and not all of these will have been determined. The possibility that such further tests, if made, might have a negative result, cannot be altogether precluded; and this possibility marks the judgment as, at the time in question, not fully verified and less than absolutely certain. To quibble about such possible doubts will not, in most cases, be common sense. But we are not trying to weigh the theoretical dubiety which common sense practically should take account of, but to arrive at an accurate analysis of knowledge. This character of being further testable and less than theo- retically certain characterizes every judgment of objec- tive fact at'all times; eVery judgment that such and such a real thing exists or has a certain objeCtively factual property, or that a certain objective event actually occurs, or that any objective state of affairs actually is the case.74 The same argument is stated in MWO. I shall quote this version of the verification argument as well. While the argument remains unchanged, the passage in MWO brings out more explicitly the relation between its conclusion and the distinction between the given and its conceptual interpre— tation. It is because an interpretation (in terms of con- cepts) has been put upon the given that our knowledge of objects is probable only and less than theoretically certain. While the same view is implicit in the passage quoted from AKV, it is not made as explicit as in the following passage. 72 By how much does the interpretation which characterizes our knowledge of objects transcend what is given? What is involved in its complete verification? Obviously, in the statement ‘This penny is round" I assert implicitly evegything the failure of which would falsify the state- ment. The implicit prediction of all experience which is essential to its truth must be contained in the origir nal judgment. Otherwise such experience would be ir- relevant to it. All that further experience the failure of which would lead to the repudiation of the apprehen- sion as illusory or mistaken is predicted in the judgment made. Now suppose we ask: How long will it be possible to verify in some manner the fact that this penny is round? What totality of experience would verify it com- pletely, beyond the possibility of necessary reconsider- ation? I have here no theoretical axe to grind, but it seems to be a fact that no verification would be abso- lutely complete; that alI—verification is partial and a matter of degree. Even after the penny itself ceased to exist, a sufficiently important connection with other matters might still lead to a revival of the question whether it is really true that a round penny lay before me on my desk on the twenty-ninth of January, at four o'clock. And however difficult it might be at such later time to gain new and decisive evidence, theoretical tests of what increase or decrease the probability would be capable of formulation. Is it not the case that the sim- plest statement of objective particular fact implicitly asserts something about possible experience throughout all future time; that theoretically every objective fact is capable of some (partial) verification at any later date, and that no totality of such experience is abso- lutely and completely sufficient to put our knowledge of such particulars beyond the possibility of turning out to be in error? So far as this is true, all interpre- tation of particulars and all knowledge of objects is probably only however high the degree of its proEability. Every such judgment about the real external world remains forever at the mercy of future possible experience. Be- tween the immediate awareness, 'This looks round,‘ and the objective interpretation, 'This is round,‘ there lies all the difference between the present moment and all time; between an experience which is now complete and had, and a totality of possible experience which is un- I'fin'ited and inexhaustible. 75 It follows that all empirical judgments not merely empirical generalizations, are less than certain. The 73 argument applies both to judgments that predicate a property or a thing as well as to judgments of the existence of a thing, as Lewis points out.76 In MWO Lewis offers an additional argument in defense of the conclusion of the verification argument. The argument has the form of a reductio ad absurdum. Assuming that one could "completely" or "decisively" verify some empirical judgment (thus rendering the judgment certain), absurd conse- quences would follow, among which would be that no further evidence could be gathered for or against the judgment in question. Lewis recognizes that as a defense of the probable character of all empirical knowledge, the argument begs the question. Nevertheless, according to Lewis, it indicates the absurdities inherent in the alternatiVe.77 Its particu- lar importance for what we are about here is that it indicates what is involved in "complete" and "decisive" verification. It is the inability to achieve such verification which Lewis takes as his reaSon for denying certainty of any and all empirical judgments, with the exception of terminating judg- ments (which are at once verifiable and are capable of com- plete and decisive verification) and of the apprehension of qualia (which is not judgment anyway). The additional argu- ment is stated as follows: Perhaps this conclusion that the verification of empirical particulars is always partial, will be more evident if we investigate the alternative. Let us suppose that, for some interpretation of a particular experience, there is some finite and limited totality of later experience which, 74 together with the presentation itself, will absolutely and finally verify our interpretation as an empirical fact so that nothing further could possibly require the retraction of it. Let us suppose the presentation to be given and interpreted at time tO. And let us imagine the verification to be complete at time t . Now let us ask: will the total significance of this empirical fact be exhausted at time t , so that consideration of it will enable no furthertprediction of anything which would be discoverable at some still later time t2? We shall hardly be inclined to answer in the affirmative, since such an answer would mean that, after the date t1, the fact would have no further consequences and its existence or histori- cal verity would be no longer determinable. Because, if at time t2 there should be no consequences of this fact, which would be lacking if the event had not taken place, then there will be absolutely no means, at time t2, whereby it may be discovered whether there was any such fact or not. Hence, there can be no such time, t1, after which no consequences predictable from the interpretation of the presentation at to will accrue. The verifiable consequences of any fact last as long as time itself. Now suppose further that at some date, t2, we put our- selves in position to meet the consequences of this fact, which was completely established at t1. And suppose that these consequences fail to appear, or are not what the nature of the accepted fact requires? In that case, will there still be no.doubt about the accepted fact? Or will what was supposedly established at t1 be subject to doubt at t2? And in the latter case can we suppose that it was absolutely verified at time t1? Since no single experi- ence can be absolutely guaranteed to be veridical, no limited collection or succession of experiences can ab- solutely guarantee an empirical fact beyond the possi- bility of reconsideration.78 Before offering criticism of this argument, I should like to extract the verification argument as it is embedded in these three passages. With a clearer picture of the veri- fication argument before us, we should have a better idea of what Lewis may have meant by "complete and decisive verifi- cation" (which he thinks is necessary for any judgment to be certain) and consequently why he thought that apprehension of qualia is in fact certain.79 75 First of all it is important to note exactly what the verification argument is claiming. The argument purports to have shown that any time anyone says with respect to any empirical proposition (where "empirical proposition" is used so as to exclude what Lewis calls expressive statements and terminating judgments) that such a proposition is certain, that he has made absolutely certain that it is true, that the truth of the proposition has been determined beyond doubt, or something roughly similar, he is always mistaken. It does not prove simply (what is obvious) that we are often mistaken when we say we know something for certain, but that we are always wrong to make such a claim, unless we mean by "cer- tainty" what Lewis calls practical certainty, which is simply a very high degree of probability.80 Let S be an empirical proposition (or what Lewis calls a non-terminating judgment or statement of objective fact). The first step in the argument.states that S has consequences or that from S one may infer statements which are predictive of further experience. Two things should be noted here. First of all, such predictions are conditional on my acting in a certain way. For example, from "There is a piece of white paper before me" I can infer "I will seem to have a certain visual experience, if I seem to look in a certain di- rection." The state of affairs predicted is, then, a conse- quence of there being a piece of white paper before me. Secondly, the future consequence is not a consequence of S 76 alone, but of S conjoined with some other statements.81 For example, that I will have a certain visual experience if I look in a certain direction is not simply a consequence of there being a piece of white paper before me, but of that in conjunction with things like my organs of sight being in a certain condition, the conditions of observation being a cer- tain way, etc.82 Furthermore, when Lewis speaks of these predictions of future experience being implied by statements like S, he need not be taken to mean that these implied pre- dictions are entailed by S. I have some reservations about this as an interpretation of Lewis, since his analysis of the meaning of non-terminating judgments at least appears to com- mit him to claiming that the terminating judgments implied would confirm.some particular non-terminating judgment and are entailed by it. Some of what I will say about the verifi- cation argument and its bearing on the alleged certainty of our apprehension of qualia does hinge on interpreting Lewis in such a way as to deny that the various terminating judg— ments implied by and which constitute evidence for some non- terminating judgment are not entailed by that non-terminating judgment. Not all of my arguments will assume that interpre- tation, however. Moreover, it seems to me just false that what Lewis calls non-terminating judgments ever entail the 83 terminating judgments that are its consequences. Thus, if I were asked to criticize the verification argument construed as making that assumption, I would argue that the assumption 77 is false for reasons made clear by Chisholm in his article "The Problem of Empiricism". Moreover, it is unlikely that Lewis would have wanted to so closely connect the thesis of sense certainty with his pragmatic analysis of the meaning of non-terminating judgments. Thus, I have presented the verifi— cation argument in a way that does not assume the view that a non-terminating judgment entails the terminating judgments that are its consequences. Furthermore, this presentation of the verification argument assumes (what I think is true) that what are spoken of as consequences of a non-terminating judgment are consequences of it only in conjunction with other statements. Keeping these qualifications in mind, the first step in the verification argument is that S has consequences or that from S one may infer statements about further experi- ence. The second step in the argument states that the number of consequences of S or the number of predictions one can make from S is unlimited. An illustration may perhaps be of help here by way of clarification. We noted above that, with the above qualification, from some such statement as "There is a piece of white paper before me" one can predict that if I were to look in a certain direction, I should have a certain visual experience. (e.g. I should see a certain white expanse in my visual field). Now, other predictions can be made from S. For example, one may infer that if I were to look in a certain direction two seconds from now, I would have a certain 78 visual experience. One may also infer that if yaa were to look in a certain direction, you would have a certain visual experience. On the basis of S one can also make "predictions" about the past. For example, one may infer that if I had looked in a certain direction five minutes ago, I would have had a certain visual experience. Similarly, if you had looked in a certain direction five minutes ago, you would have had a certain visual experience is also a consequence of S. More- over, if any of these predictions should be borne out, this would constitute evidence for the truth of S. While each of these consequences of S depend on the piece of paper contin— Uing to exist (and to be where it is), there are other conse- quences of S which do not depend on such facts. Supposing that the paper were burned; then S would'have the consequence that if I were to look in a certain direction, I would have the experience of seeing ashes, and such an experience could be taken-as evidence for the claim that (at some previous time) there was a piece of paper before me. I think it can be seen from these illustrations that the second step in the verification argument, that the consequences of S are indef- inite in number and that each may be taken as evidence for the truth of S, is true. The third step states that one or any number of these consequences of S gay fail to occur. I will not discuss the correct analysis of this premise. It offers considerable difficulty, especially with regard to the interpretation of 79 the model expression "may fail to occur". The meaning of this expression is, however, crucial to the significance of the conclusion of the verification argument. For example, if we mean by "some of the consequences of S may fail to occur" simply that the consequences of S are not self- contradictory or are not analytic truths, then the argument would lose its significance. All that it would then tell us is the EELS of statement that S is and the kind of statement we might use to express the consequences of S. At least as intended by Lewis, the verification argument makes a stronger claim than that. The fourth step in the argument amounts to saying that if any of the consequences of S should fail to occur and the statements expressing them be false, then doubt would be cast upon the truth of S (i.e. S would then be subject to doubt.) For example, if I should look in front of me and fail to have the experience of seeing a piece of white paper, then, all other things being equal, I would then have reasonable grounds for doubting the truth of S. By ”All other things being equal" I mean to say that there is no reason to believe that the-other statements needed, in conjunction with S, to make the predictions in question are false. This qualification is required since without it we would have no grounds for doubt- ing specifically S. 80 The fifth step in the argument states that if at any time, say t1, there are grounds for doubting S, then at no previous time was 5 completely and decisively verified. Thus, S could not have been known with certainty at t As 1' stated, however, this fifth step is false. That it is false can be seen from the following reflections. Suppose there is a certain page of notes discovered by me in the philosophy office at Michigan State University. Suppose that these notes were written by Professor Arnold Isenberg. Since I never knew Professor Isenberg, am not sufficiently familiar with his handwriting to identify it readily and furthermore the notes, let us assume, deal with the relation between meaning and purpose ( a subject about which a colleague of Professor Isenberg's had a good deal to say), if I were at the time to consider the claim that the notes were written by Professor Isenberg,.I would have grounds for doubting it. More strongly, I would have good grounds for doubting it since I would have some grounds for believing that they were written by Professor Leonard. But it does not follow from the fact that 5 now have grounds for doubting that Professor Isenberg wrote those notes that at no previous time was anybody certain that Professor Isenberg wrote these notes. In particular, my present reason- able doubt is perfectly consistent with Professor Isenberg's certainty that ta_wrote the notes. There is no reason (short of begging the question) to claim that the following propo- sitions are inconsistent: "Connolly has reasonable grounds 81 grounds for doubting that Professor Isenberg wrote those notes" and "Professor Isenberg knew for certain that he wrote those notes”. What would be absurd would be for me to assert the following: "I doubt (with reason) that he wrote them, but Isenberg knew for certain that he did." The proposition asserted is not self-contradictory, but there would be some- thing self—defeating about my asserting that proposition. It may be that failing to note the crucial difference here between the self-contradictoriness of a proposition and the absurdity or self—defeat which may accompany one's asserting the same proposition is what led Lewis to claim something like step 5. That tayta should fail to heed this distinction is surprising, especially in view of his discussion of the notion of pragmatic contradiction as it occurs in his treat- ment of cognitive, moral and prudential imperatives.85 Can the argument be restated in such a manner as to avoid this difficulty? The original argument looks like this. 1. S has consequences. 2. The number of consequences of S are unlimited. 3. The consequences of S may fail to occur. 4. If some of the consequences of S were to fail to occur, then there would be a reasonable doubt that S is true. 5. If at any time there should be a reasonable doubt that S is true, then at no previous time did anyone make absolutely certain that S is true. 82 6. Therefore, no one did make absolutely certain that S is true. Following Malcolm, I have argued that as stated 5 is false. If one reflects on the general intent of the verifi- cation argument, I think that one can present an acceptable alternative to 5. What seems to be in Lewis' mind here is something like the following. S has a (potentially) infinite number of consequences. Each of these consequences of S constitutes evidence that S is true. The entirety of the consequences of S constitutes the whole of the evidence for S. Lewis wants to claim that unless all of this potential evidence is in, we cannot be absolutely certain that S is true. Since the consequences of S are infinite, and any one of them may fail to occur, no one can make absolutely certain that S is true since that would entail, what is surely im- possible, the carrying out of an infinite number of tests. To refer back to our counter example to step 5 as presently stated, we are being urged to imagine that even if I had all the-evidence available to Professor Isenberg, I would still not be absolutely certain that he wrote those notes. In fact, neither would Professor Isenberg. Harry Frankfurt has suggested an alternative to step 5 which is, I think, close to the spirit of the verification argument as defended by Lewis. He would replace the following for step 5. 5a. If at any time t2 there should be reasonable doubt that S is true, then at no previous time tldid 83 anyone make absolutely certain that S is true, provided that the evidence possessed at t2 includes the evidence possessed at t1. The argument, if sound, does not simply show that S is not certain, but, since for S we can substitute any empirical proposition and for t any moment in time, it shows that no empirical proposition is ever certain. As stated, however, the argument is not entirely clear. As Malcolm points out, step 3 must be understood in such a way that the modal expression means more than logical possi- bility. This is required since someone might perfectly well admit that in some sense of "may" or "possible" that it is possible that the consequences of S may fail to occur, but still maintain that it is nevertheless certain that they will occur. For example, one might understand "possible" to mean "not self-contradictory". So understood, admitting that it is possible that some of the consequences of 8 may fail to occur does not commit one to saying that therefore it is not certain. One further, but I think minor, qualification must be added if the argument we are considering is to be an ac- curate rendition of Lewis' version of this argument. Some of the consequences of S will be certain; for example, any of the consequences of S that I have in fact brought to the test will be certain. In the language of AKV, when any of the terminating judgments implied by a given non-terminating one are brought to the test, they are then determinable as 84 certainly true or certainly false. Step 3 must be read in such a way as to exclude reference to tests already made. (Since it plays no crucial role in the verification argument, I am ignoring the further complication that any tests carried out in the tast are less than theoretically certain, since the reliability of memory is presupposed.) Nevertheless, this does not weaken the argument since even omitting those consequences which have already been brought to the test, there are still infinitely many consequences whose truth has not been established. I am not quite finished with this argument, however. Both Malcolm and Frankfurt think that the argument is valid as stated. Malcolm attacks the argument not for being in- valid but because step 3 is either confusing or false. I am not quite able to see that the argument is valid, however. The conclusion one wants to draw is that S is not absolutely certain at t1. In order to obtain that conclusion from step 5a one must have as premise (or as an intermediate conclusion from onewor more previous premises) the proposition that at any arbitrary time there would be reasonable doubt that S is true. This is not obtainable from the argument as it stands. It is not stated as a premise, nor is it possible to infer it from any or all of the premises we now have available. How- ever, the intermediate conclusion we want is simply the con- sequent of the conditional we presently have as step 4. If we construe step 4 in the following way, we could derive the 85 intermediate conclusion. And that, in conjunction with step 5a would permit the inference to the proposition that no one did make absolutely certain that S is true. I suggest that step 4 be replaced by 4a. 4a. If some (any) of the consequences of S may fail to occur, then there would be reasonable doubt that S is true. With these revisions I think that the verification argument is valid. I shall not endeavor to evaluate the soundness of this argument. As Malcolm has pointed out one important question which must be-resolved in.determining its soundness is the question of how to interpret step 3. Moreover, there is a peculiar (Lewis would simply say strict) notion of certainty here. Some of this peculiarity will hopefully emerge in what follows. His view of certainty is such that one can be cer- tain of a proposition only if one has obtained all the pos- .sible evidence for it. Ordinarily one would say that there can be several ways of attaining certainty and that one need not have carried out all the tests involved. But this is simply to say (as Lewis would put it) that his notion of certainty (absolute or theoretical certainty) is much more stringent than the practical certainty that is embedded in ordinary usage, and appeal to such usage would not be a compelling objection to Lewis. Moreover, even if Lewis' notion.of certainty were shown to be esoteric and were shown 86 to depart radically from its usual employment, much of his case would remain intact. Whatever peculiar characteristic Lewis' notion of certainty might signify, it is a character- istic possessed by apprehensions of given qualia and termin- ating judgments, but which is not possessed by non-terminating judgments. If that much is granted, then the case that there is a fundamental difference between apprehensions of given qualia and knowledge of objects, which is all that Lewis needs to make the distinction between what is given and what ale ready reflects interpretation, will have been granted. This assumes, of course, that his notion of certainty, while peculiar, is at least not incoherent. My main concern is not with the verification argument itself, but with the rea- sons given for the (merely) probable character of all objec- tive beliefs. In particular, if the reasons given for the probable character of objective beliefs are justified, can it still be claimed that apprehensions of given qualia are certain? Are the reasons given for the certainty of our apprehension of qualia consistent with what Lewis commits himself to in the verification argument? As I hope to show, Lewis' notion of sense certainty is not coherent. There are three general types of criticisms which can be leveled against Lewis' theory of the given. Following Firth, these may be called (1) phenomenological, (2) onto- logical and (3) epistemological. We have already presented a phenomenological objection to Lewis in our treatment of 87 the criterion of unalterability. We have also raised an ontological objection to Lewis' account in a critical dis- cussion of his distinction between qualia and properties. Epistemological objections to Lewis' theory of the given may take several forms, all of which gather about the issue of sense certainty. First of all, one might argue that Lewis is mistaken in thinking that if any knowledge is to be even probable, something disclosed in experience must be certain. This type of criticism will not be discussed here since for adequate treatment it would require a more thorough analysis of the concept of evidence and probability than will be attempted here. Secondly, one might argue that while apprehensions of given qualia are certain, they are not uniquely certain since at least some objective beliefs may be certain as well. Thirdly, one might argue that, given the reasons Lewis has presented for claiming that non-terminating judgments can never be certain, Lewis should have concluded that apprehen- sion of qualia can never be certain either. And, fourthly, one might argue that Lewis fails to characterize a suffi- ciently coherent notion of certainty to make his case plau- Sible. The following discussion will concentrate on objections of the third and fourth types. One of the most frequent objections made against Lewis (at least the Lewis of MWO) is that if our apprehensions of given qualia are neither judgmental nor propositional 88 then it is quite unintelligible to say that such apprehension is certain or uncertain, since a necessary condition for the application of these terms is that they be applied only to what is in some way judgmental or propositional in nature.87 The assumption is that if something is to count as certain it must first of all be something capable of being true or false. Roderick Firth has suggested a way of avoiding this difficulty.88 He argues that if apprehension of given qualia is to provide a foundation for empirical truth, then such apprehension.must.be propositional of judgmental. If such apprehension is not so understood, then such apprehension can be neither certain nor uncertain. He suggests that this difficulty may be avoided by the following interpretative maneuver. The claim, frequently made by Lewis, that the apprehension of given qualia is not judgment should be under- stood in context. So understood, all that Lewis means to commit himself to in denying that apprehension of qualia is judgment is that (1) such apprehension does not refer beyond the presently given and (2) that there is, in contra- distinction to what we ordinarily think of as knowledge, no contrasting possibility of error. It is only in ttta sense that apprehension of given qualia is not judgment. This initially appears to be an implausible interpre- tation, since after all Lewis does say in MWO that such ap- prehension of given qualia is not judgment. However, the . _ 89 Lewis of 53! may be suggesting just this possibility by means of the use of the notion of expressive language. Unless apprehension is construed as apprehension taat, and not just as apprehension at it cannot play the justificatory role it is intended to by Lewis in his theory of knowledge. If such apprehension is to provide the ultimate and certain evidence for objective beliefs, then they must be in some sense judg- mental. Such judgments would, however, be restricted to being judgments about the given. In this way we can still distinguish between terminating and non-terminating judgments on the one hand and the mere apprehension of qualia on the other. However, while this suggestion avoids the initial ob- jection that non-propositional or non-judgmental apprehension can be neither certain nor uncertain, it nevertheless would raise other difficulties. It should be recalled that one of the reasons that Lewis gives for claiming that no non-terminating judgment can ever be certain is that the truth of such non-terminating judgments presupposes that the terms occurring in the judgment be correctly applied. For example, the truth of the propo- sition that this piece of paper is white presupposes that the term "paper" has been correctly applied. Similarly for the term "white". Keeping in mind that for Lewis the applica- bility of a term commits one to the indefinite number of terminating judgments included in its sense meaning, or I'f -. , I h _ L ‘ . its. 90 criteria of application, such application involves prediction and is consequently at the mercy of future experience and therefore (by the verification argument) is less than theo- retically certain. Thus, on Lewis' own account, since the assertion of any proposition presupposes that certain terms apply to a presented object, then the assertion of any propo- sition, even one descriptive only of the immediately given, is less than theoretically certain. Therefore, description of the immediately given cannot be theoretically certain, and consequently the criterion of certainty will not serve to distinguish the given from its conceptual interpretation. To this Lewis could reply as follows. The objection seems to make its case only by failing to take note of a fundamental distinction between terms in their objective use and terms in their expressive use. The argument applies only to the former, not to the latter. This suggests that the notion of sense meaning as criterion in mind not be applied to terms in their expressive use. While Lewis nowhere con- siders whether he would extend his fourfold analysis of mean- ing to expressive language, he might avoid the kind of objection presented here by refusing to so extend it. The result would be that terms in their expressive use have no criterion in the sense of sense meaning for their correct application. The applicability of such terms is not a matter of applying a criterion, but only an immediate recognition of phenomenal character. i . ‘ _i;h _- 91 However, this reply is not sufficient and forces Lewis into a dilemma. Either he must admit the theoretical uncertainty of expressive statements (really the judgments they express) or he must admit that expressive statements cannot serve their role as ultimate evidence in the analysis of empirical knowledge. The first horn of the dilemma en- tails the rejection of the criterion of certainty as a means of analytically isolating the given element in experience. The second horn of the dilemma deprives the given of its crucial epistemic role. The first horn of the dilemma follows if Lewis extends his fourfold analysis of meaning to terms in their expressive use. I should now like to show how the second horn of the dilemma follows from the rejection of such an analysis for expressive language. The result will be the appearance of a fundamental inconsistency in Lewis' analysis of empirical knowledge. The whole point of expressive statements and the .propositions they express is to provide ultimate experiential evidence for non-terminating judgments and the objective be- liefs they express. Thus the expressive statement "this presentation (denoting the presentation of a piece of paper) is white" or roughly "This looks white" is construed as (par- tial) evidence for the non-terminating judgment that this piece of paper really is white. But if Lewis does not extend his fourfold analysis of meaning to the term "white" in the expressive statement, then the expressive use of the term 92 "white" does not have the same meaning as the term "white" in its objective use. But then it is quite unclear how the expressive statement "This looks white" could be evidence for the non-terminating judgment that this piece of paper is white. ~That the same word is used should not necessarily lead us to believe that the same meaning is attached to it in each case. Thus, if the expressive use of "white" does not have the same meaning as its objective use (since it has no sense meaning as criterion in mind), then the expressive statement cannot serve the epistemic role it is intended to by Lewis.i The dilemma follows, since either we do or we do not extend the notion of sense meaning to terms in their expressive use. Moreover, if Lewis extends his notion of sense mean- ing to terms in their expressive use, a further difficulty arises. If the term "white" in both its expressive use and its objective use have the same meaning (i.e. if the color property which the piece of paper looks to have is the same as the color property which it actually has), then we are predicating a (color) property of two things that are sup- posed to be categoriallyfidiStinctr, namely a physical object and a phenomenal presentation. Since a necessary condition of something's having color properties is that it be a physi- cal object, it would follow that presentations are themselves physical and "white“ in its expressive use would signify a 93 physical property, a consequence which Lewis could hardly accept, so long as he wishes to distinguish qualia from properties.89 There is a further difficulty associated with the expressive use of language as descriptive of immediate ex- perience.Setting aside momentarily the objection presented above, I should like to argue that memory is presupposed in the application of an expressive term, and consequently, on Lewis' account of memory, we are committed to saying that the application of an expressive term to the momentarily given cannot be theoretically certain. If I am to correctly apply the expressive term "white" to a presented quale, I presup- pose that the immediately presented quale is the same or similar to a quale that has been presented in the past. Such qualia comparison presupposes the validity of a judgment of memory. Since on Lewis' account, a memory claim is never theoretically certain but only prima facie credible,90 it follows that the application of an expressive term is not theoretically certain, but only prima facie credible. The consequence of this is that one means by which Lewis proposes to distinguish awareness of the given from knowledge of objects will not serve to distinguish the given from its conceptual interpretation. If one reason for the merely probable character of non-terminating judgments is that such judgments presuppose the application of concepts, then expressive statements also presuppose such application. 94 The only way in which Lewis could avoid this objection (namely by not applying the notion of sense meaning to ex- pressive terms) would render expressive statements incapable of fulfilling their evidential role in empirical knowledge. At this point I should like to turn to some of the other arguments discussed above for the certainty of our awareness of the given. In this connection I should like to do two things. First of all, I should like to show that each of them presupposes that the reason for the certainty of our awareness of the given is that step one of the verification argument does not apply to our awareness of the given and the statements expressing it. Secondly, I should like to argue that on any reasonable interpretation of that part of the verification argument, it would follow that expressive statements are also less than theoretically certain. Lewis nowhere gives a definition of "certainty". Nevertheless, in the course of arguing for the certainty of our apprehension of given qualia and the less than certain character of non-terminating judgments he says a good deal about the differences between these two kinds of apprehension. Presumably, some of these alleged differences will be indic- ative of what Lewis meant by certainty. It is for this rea- son that a clear understanding of what, if sound, the verification argument shows about empirical statements is crucial. If apprehension of given qualia and their 5 ‘-‘ — i I ‘ I H .—..I| I A. . 95 corresponding expressive statements are certain, it must be that Lewis took one or several of the premises of the veri- fication argument not to hold of them. First of all, I refer the reader to the five reasons given, other than the verification argument, for the cer- tainty of the apprehension of qualia and the merely probable character of empirical judgments. Most of these reasons seem to me to depend on the verification argument. In fact several of them simply restate the conclusion of the verifi— cation argument; that is, they claim the theoretical uncer— tainty of empirical judgments. They are required for a correct understanding of Lewis' position since in addition to doing that, they indicate which premise or premises of the verification argument fails to apply to expressive state— ments and the apprehensions of qualia they express. Consider the first reason. The objects of immediate awareness are subjective. Several things need mentioning here. The term "subjective" is horribly illusive. One wishes that when Lewis had made this claim he had been more specific in indicating what it was that he meant by calling qualia subjective. We have already had some things to say about this in the section on qualia and properties. First of all, as pointed out above, he cannot mean, on pain of inconsistency, that qualia are subjective in the sense that they are mental states or mental entities. He cannot claim this if he wants to hold, as he apparently does, that without 96 interpretation, the given is neither external reality nor explicit self. The division within experience between self and not self already marks, for Lewis, the activity of an interested mind. That is, it already reflects the taking of a presentation as the presentation of a certain kind of thing. Lewis means in part by calling qualia subjective that with- out interpretation, a quale is not related to any further possible experience. In the context of the certainty of our awareness of qualia this entails that, without interpretation, the occurrence and character of a specific quale or complex of qualia is not verifiable by any further experience. Hav- ing at least some relation to further experience is a neces- sary condition of anything's existence or character being verifiable. Precautionary remarks are in order, however. Lewis cannot mean, nor do I think that he means, that, literally speaking, an uninterpreted quale bears no relation to further experience. Surely, it at least bears the relation of being the temporal predecessor, within the experience of an individ- ual, of some later presented quale. That a presented quale in fact bears such relations to further experience is, according to Lewis, presupposed by the fact that we can ex- perience objects. That is to say, if each quale were equally associated with every other, then we could not experience objects.91 For Lewis the possibility of applying concepts 97 is a necessary condition for the possibility of experience of objects. Furthermore, the application of concepts to the given presupposes that there is at least some discoverable order among presented qualia. For Lewis, we could not apply concepts to the given (and thus could not experience objects) unless we were justified in believing generalizations of the following sort: when I seem to be doing x, then with proba- bility a quale or complex of qualia of one kind will be fol— lowed by a quale or complex of qualia of another kind. But that we are justified in believing generalizations of this type presupposes at least some discoverable order among pre— sented qualia; and exhibition of such order presupposes that presented qualia bear certain relations to one another. In fairness to Lewis, we must take him to mean some- thing else. I would like to suggest that Lewis is not denying that a given quale itself is related to further experience (that would be tantamount to denying the existence of objects). What he is denying is that without interpretation we could ever be aware that a presented quale bears such relations to further experience. Thus, without interpretation, ya do not relate one quale to any other, even though qualia are so related. Thus, without interpretation (i.e. without applying a concept to a presented quale or complex of qualia, which is what Lewis means in MWO by giving a meaning to a presentation) there would still be objects (since qualia would bear the required relations to one another), but there would be no objects for us. 98 It is in this context that one can get some under- standing of Lewis' claim that there would be no reality for the purely passive being, a view he takes in both MWO and AKV. Whether a concept applies or fails to apply can be known only if one can come to know whether certain counter— factuals like the one above are true. Furthermore, these counterfactuals assert what would be the case, if a certain action were (or seemed to be) performed. A purely passive being, then, would not be able to apply concepts, since such counterfactuals would have no significance for it. But a being that is unable to apply concepts to experience could not have the experience of objects, and having experience of objects amounts to being able to distinguish real from unreal within a particular category. Since we now have some notion of what Lewis means by saying that qualia are subjective, why would he say that the subjectivity of qualia renders impossible the verification of their apprehension? For Lewis, we verify a belief in a state of affairs or a belief in the existence of an object, by finding out whether or not its consequences obtain in further experience. But such verification in terms of con- sequences requires that we have noted, by the application of a concept, that these further verifying experiences are related as consequences to the alleged state of affairs under consideration. If such relations failed to be noted, then further experience would not be relevant to the verification 4‘ 99 of the state of affairs in question. That is precisely what Lewis thinks is the case for the mere apprehension of a given quale. Since without interpretation, ya have not related the given quale to any further experience (keep in mind that this does not entail that the quale itself is not so related), we have no way of knowing what further experience is related, as consequence, to the presented quale. Since it is only in terms of its consequences that we can verify an apprehension or judgment that a certain object exists or has a certain property, we cannot verify its existence or character, since without interpretation we would have no idea as to what in further experience is related to the presented quale as .consequence. .It seems to me that this is what Lewis intended to say in claiming that the mere apprehension of a quale is not something.about which we can be mistaken. For Lewis, the mere apprehension of.a quale, out of relation to further experience,-cannot be mistaken because no implicit prediction has been made. ~Nor-can such apprehension be knowledge, since one can have knowledge only of that about which one could 92 There are, then, two requirements conceivably be mistaken. for empirical knowledge; (1) that it be verifiable, and (2) that it have a significant opposite, error. The apprehension of.a.quale fails to fulfill either of these requirements. It seems, then, that the first reason why Lewis claims that the apprehension of qualia is certain amounts to claim- ing-that the first premise of the verification argument 100 (namely "S has consequence") fails to hold of expressive statements. Of course, if the first premise fails to hold, then neither does the second. If expressive statements have no consequences, then a fortiori they do not have an infinite number of consequences. Simply showing that the verification argument does not apply to expressive statements does not suffice to show that expressive statements are certain, since having an infinite number of consequences is only a sufficient condi- tion for theoretical uncertainty, not a necessary condition. Why then should Lewis.think that having shown that expressive statements, on one count, cannot be mistaken, entitled him to claim that, therefore, on no count can expressive state- ments be mistaken? He made.this inference because he held implicitly to an assumption which, in conjunction with the verification argument, would permit the inference. No mis- taken (empirical) apprehension is possible unless in such .apprehension there is either explicit or implicit inference, that is unless such apprehension is judgmental. We have noted, however, that expressivestatements "have no conse- .quences” and, in matters empirical, it is only by their consequences that empirical statements can be verified or falsified.~ Since expressive statements have no consequences, such statements do not express judgments and their assertion makes no inference, implicit or explicit. But the making 101 of an inference, according to Lewis, is a necessary condition for a mistake's having been made. Therefore, no expressive statement can be mistaken and our apprehension of qualia is certain. I think that several of the other reasons that Lewis gives for the certainty of expressive statements and the apprehension of qualia amount to no more than what is involved in denying that they have any consequences or that they are inferential. And that, in conjunction with the assumption that the making of an inference is a necessary condition of a mistake's having been made, suffices to justify the cer- tainty of expressive statements. Take, for example, a second reason Lewis gives for the certainty of expressive statements and apprehension of qualia. Ascription of a property, ac- cording to Lewis, is a judgment and therefore can be mistaken, while the apprehension of a quale is not a judgment and thus cannot be mistaken. This reason seems to be based on the assumption that if an apprehension is not the result of an inference or not implicitly inferential, then it cannot be mistaken. In such immediacy there is no possibility of error.93 At other times Lewis says things to the effect that the predication of a property asserts something beyond or which transcends what could be given in any single experience, while apprehension of a quale makes no such assertion. This amounts to denying that expressive statements have any 102 consequences and in that respect differ from statements which express objective beliefs. At still other times Lewis says that the apprehension of a quale and the expressive state- ment which expresses it are not verifiable. This, too, is essentially a denial that expressive statements have any con- sequence, if it be kept in mind that, for Lewis, (empirical) apprehension is verifiable if and only if it has consequence. There is another reason Lewis gives for claiming that apprehension of qualia and expressive statements are certain. We have noted that expressive statements are not verifiable, since they have no consequences. Lewis also claims that ex- pressive statements and the apprehensions of qualia they express stand in no need of verification. This claim, unlike the previous one, does not reduce to pointing out that one of the premises of the verification argument fails to hold of expressive statements. It is probably the basis for the strongest argument for sense certainty at Lewis' disposal. Roughly speaking, it claims that since it is only the char- acter of a single presentation that is in question, the only evidence that could have any bearing on its truth or falsity is the-apprehended quale itself. Whatever evidence there could be is, so to speak, staring one in the face. Unlike statements-of objective-belief, one cannot imagine what evidence-there could be, since no verification is needed. Verification would have no meaning. 103 There are, however, difficulties with this position which relate to difficulties inherent in Lewis' notion of expressive statements. First of all, while it may be true that expressive statements are not the result of inference, such expressive statements and the apprehensions of qualia they express do presuppose other knowledge claims. fiIn.par- ticular, they presuppose memory claims concerning qualia com- parison. But then expressive statements are no more certain than the memory claims they presuppose., Since on Lewis' ac- count memory claims are never certain, but only prima facie credible, expressiVe statements cannot.be certain either. In short, since expressive statements presuppose other knowledge claims, they cannot be the foundation of empirical knowledge in the sense.intended by Lewis. On this account apprehension of given qualia, while a necessary condition for empirical knowledge,.itself presupposes the conceptual structure it is the partial foundation of. That is, apprehension can occur only within an already semi-articulate conceptual structure. Moreover, even if it were true that expressive statements "stand in no need of verification" it does not follow that expressive statements must necessarily be true, in which case they are still unable to function as the foundation of empirical knowledge. Even if the apprehension of qualia and the expressive statements by means of which such apprehension is expressed may be at the moment of such apprehension irrefutable, it does not follow that such 104 apprehension is certain in the sense of being certainly true. After all, we may be justified in believing something that is false. Now if Lewis were merely trying to offer an alter- native to the coherence theory of justification, the notion of sense apprehension as presently irrefutable might be suf- ficient. However, Lewis was also interested in offering an alternative to the coherence view of the nature of truth as well. It should be recalled that the notion of sense cer- tainty-was intended to do both jobs for Lewis. There is one final reason that Lewis gives which either.reduces to a claim that expressive statements, unlike statements of objective belief, are not affected by the verification argument or, if not, faces serious difficulties on other grounds. The argument I have in mind is implicit in the following passages which occur in the context of Lewis' attempt to make a distinction between qualia and the proper- ties of objects. . . . what constitutes the existence of an objective property and the applicability of a concept--even of the simplest sort-—is not a given quale alone but an ordered relation of different qualia, relative to different con- ditions of behavior. This pattern or order, which is what the.adjective names, will always be temporally ex- tended (which is the same as to say that the predication of the property is something varifiable), and always it will have relation to our own possible ways of acting toward the presented object . . . . . . If concepts are to be articulate and meaningful, then the application of them must be something verifi- able; which means what they denote must have a temporal spread.- Not a momentary presented quale, but an ordered relationship of such, is the least that can be meaning- fully named.94 105 This is not an easy passage to interpret. The argu- ment against the possibility of verifying the apprehension of qualia which is embedded in this passage is based on the assumption that a quale is momentary in its existence (i.e. it exists only for the duration of a single specious pres- ent), while a property, being an ordered relation of qualia, exists "longer" than a specious present. Lewis wants to argue further that this difference between qualia and proper- ties somehow entails that our apprehension of the latter is verifiable, while that of the former is not. Several ques- tions are-in order here. First of all, why should Lewis think that a quale's existence is momentary? If we grant the assumption that our apprehension of qualia is momentary, why should Lewis think that as a consequence of this, what is apprehended, namely a quale or complex of qualia, should itself be momentary? .One»possible interpretation of Lewis is that the predi- cation of a property and the empirical statement which ex- .presses;it.has.consequences for-future actual and possible -.experience,=while the apprehension of qualia and the expres- .sive statements which express them do not have such conse- quences.. If this is what is intended, then this justifica- tion for the certainty of our apprehension of qualia amounts to.claiming that the first step in the verification argument fails to hold of expressive statements. I am not at all 106 sure, however, that all that Lewis intended by claiming that qualia are momentary is that they have no consequences in the sense in which that is understood in the verification argument. But if he does not mean this, then the most plausible interpretation would be that the momentary existence of qualia is implied by the momentariness of their apprehension. But if this is what Lewis here intends, then he is contradicting an important feature of his theory of the given; namely, that without interpretation the given is neither external reality nor explicit self. I should now like to argue that on any reasonable interpretation of.the verification argument Lewis should have concluded that expressive statements also fail to have what he refers to-as-theoretical certainty. For Lewis, every statement of objective fact has further-consequenceS‘and“these'consequences-constitute.the evidence.that one may have or may seek for the statement of objective fact. No matter how many of these consequences may be brought to the test ' . there will be further consequences which must be thus and so if the judgment is true, and not all of these 'will have.been determined. The possibility that such further tests, if made, might have a negative result,” cannot be altogether precluded; and this possibility marks the judgment as, at the time in question, not fully‘verifiedandless'than"absoluteiy-certain.95" To obtain absolute certainty4with respect to a statement of objective fact would require, then, the carrying out of the 107 verification of the indefinite number of consequences implied by that statement. The crucial aspect of this argument is what precisely Lewis means in claiming that these further consequences maat be so if the statement is true. There are two ways in which this may be interpreted. On the first interpretation Lewis may be taken to understand this as a logical must. Each of the terminating judgments thus implied would, for Lewis, constitute part of the inten- sional meaning of the statement of objective fact. As Firth points out, however, this doctrine that the relation between non-terminating and terminating judgments is a matter of logi- cal entailment makes Lewis' argument relatively uninteresting, since one of its premises would be his controversial analysis of the meaning of empirical statements.96 Moreover, Lewis has in fact defended his view that statements of objective fact are never absolutely certain without making any specific use of his pragmatic analysis of meaning in his paper "The 97 It is reasonable Given Element in Empirical Knowledge". to conclude, therefore, that by "must" Lewis does not mean "logically implied". A second interpretation of Lewis' first step in the 'verification argument would claim that by future consequences Lewis did not mean logical consequences only, but any future Consequences which are evidentially relevant to the statement (”5 objective fact in question. Some of these consequences mi-<3ht be logical consequences, but others will be causal con- Se(Juences. On this interpretation of the verification argu- merlt, Lewis' point becomes much more plausible. 108 However, on this second interpretation of the verifi- cation argument Lewis is not able to claim that the first premise of the verification argument fails to hold of expres- sive statements. Thus he will not be able to claim that there is some sense of certain in which expressive statements are certain while statements of objective fact are not. An apprehension of sense is an event and historically unique. But as such it is an event in the natural order and conse- quently will be causally related to other events. But then, in accordance with the second interpretation of the verifi- cation argument, expressive statements will have future con— sequences which will be evidentially relevant to it. Since having such consequences is, for Lewis, a sufficient condi- tion for a.statement’s being less than theoretically certain, it follows that expressive statements are less than theoreti- cally certain. Thus, since the verification argument expli- cates what Lewis means by "less than theoretically certain", Lewis is not able to specify a sense in which expressive Istatements.are certain while statements of objective fact Eire not. If Lewis were still to wish to specify a sense in Which.presentations of sense play a crucial role in empirical knowledge, he might avail himself of the following suggestion (If R. M. Chisholm.98 Lewis could take a position with respect tC> statements expressing presentations of sense analogous 109 to the one he takes with respect to apparent memory claims in his discussion of the validity of memory. In this connec- tion Lewis claims that the possibility of knowing objects presupposes the following two claims about memory: (1) while no memory claim is absolutely immune to doubt or rejection, nevertheless a memory claim, qua memory claim, is prima facie credible because remembered; a memory claim is, so to speak, -innocent until proven guilty; (2) secondly, when the whole of our empirical beliefs are taken into account, those which are more-or less credible will be recognized as such by their mutual congruence.99 Extending these two notions of prima facie credibility and congruence-to sense perception would still enable Lewis to speak of empirical knowledge as having as aaa of its foundations data of sense. But the crucial and controversial role of sense certainty will have been eliminated. While departing from classical foundationalist epistemology, Lewis would not succumb to any coherence theory which would be too (ijectionable to Lewis' pragmatic approach. For Lewis, congruence, while not the same as mere lJogical consistency, is a certain relation between proposi- tions or a property of sets of propositions such that the antecedent probability of any one of them is increased if the rest.- are. assumed as premises.- But such a relation of con- gruiencevis not alone sufficient as a test of empirical cogency 110 The nature of this predicament (i.e. the lack of cer- tainty for isolated memory claims) suggests a coherence theory of truth as the solution of it. Specifically it suggests, as a possibility to be examined, that a body of empirical beliefs, each of which is less than certain and no one of which can be substantiated on empirically certain ground, may nevertheless be jusified as credible by their relation to one another. . . . But these pertinent facts are fundamentally dif— ferent, in the logical significance of them, from those theses, put forward by British post-Kantian idealism, which have the best right, historically, to the label 'coherence theory of truth'. In order to mark our de- parture from that historical conception, we shall speak of the congruence of statements, instead of coherence; and shaII.assign to this term 'congruence' a definite and limited meaning. A set of statements, or a set of supposed facts asserted, will be said to be congruent if and only if they are so related that the antecedent probability of any one of them will be increased if the remainder of the set can be assumed as given premises.100 However, this relation of congruence requires more than mere logical consistency, but less than that of logical deduci- bility. Still, neither consistency nor congruence can alone establish a belief as even in the least probable. Something more is required, since a set of propositions may be as con- gruent as you please, but congruence itself cannot provide a ground for truth. When dealing with memory claims and their Liustification Lewis is content with claiming that all that is frequired beyond-congruence is the prima facie credibility of E1 memory claim qua (apparently) remembered, not its certainty. The prima facie credibility is then defended by a transcen- dental argument, arguing that without the assumption of the Prima facie credibility of memory, we could have no conscious- ness, of. the past and thus no experience of objects. The sug- ge£3tion made here is that Lewis could treat perception in a 111 manner similar to the way he treats memory. Perceptual claims are prima facie credible because (apparently) perceived. Without that assumption, knowledge of (and thus experience of) objects would be impossible. In this way, the crucial role of perception in empirical knowledge could be assured, but the data of sense would not consist of certain acquaintance with given-qualia, but would consist of prima facie claims.101 Whatever the merits of this reinterpretation of Lewis, the following points deserve notice. First of all, while the justification of an isolated expressive statement will not be the result-of inference, nevertheless, as was argued above, such knowledge will not be ultimate in the sense of not presupposing any other knowledge of an empirical sort. -While acquaintance will still be a necessary condition of empirical knowledge, such acquaintance will presuppose other empirical knowledge. Thus much of the philosophical sting of Lewis' notion of the given will have been eliminated. And, more importantly for what we are about here, the criterion of certainty will not enable Lewis to distinguish the given from its conceptual interpretation. Concluding Remarks of Part I In Part I the following points have been argued. It was noted that a crucial aspect of Lewis' epistemology is the distinction between the two elements in experience--the given and its conceptual interpretation. It has been argued that 112 the various ways in which Lewis attempted to isolate analyti— cally the non-conceptual from the conceptual elements in experience are inadequate. In this connection we have defended the following claims: (1) Lewis' claim that the sensuous ele- ment in-experience is unalterable by interests and purposes does not provide a definitive criterion of the given element in experience. Since the a priori element in experience is defined as that element in experience which is alterable by interests and purposes, Lewis' account of the a priori should be re-evaluated accordingly. (2) Lewis' second attempt to distinguish the two elements in experience in terms of a dis- tinction-between qualia and properties has also been claimed to providelan inadequate criterion for making the distinction, and in addition was plagued by serious internal difficulties as well.. (3) Finally, it has been argued that Lewis fails to specify a sense of certainty in which our awareness of objec- tive empirical facts is less than theoretically certain, while awareness'of the given is theoretically certain. Thus, Lewis' criterion of certainty is also inadequate for distin- guishing the conceptual and the non-conceptual elements in experience. Since Lewis frequently characterizes the a priori ele- ment in experience negatively in terms of the given, it will be the purposeof Part II to take a fresh look at Lewis' prag- matic conception of the a priori in the light of the foregoing critical discussion. PART II THE A PRIORI 114 Introduction In Part I we were concerned with Lewis' attempt to isolate, analytically, the two elements in cognitive experi- ence, the given and its conceptual interpretation. In that discussion the great bulk of our consideration dealt with Lewis' notion of the given. In this part our concern will be with the conceptual element. It will deal with Lewis' explication and defense of his pragmatic conception of the a priori as that conception appears in MWO and two earlier papers which formed its foundation: "A Pragmatic Conception of the A Priori" and "The Pragmatic Element in Knowledge". After a discussion of the influence of developments in formal logic on Lewis' conception of the a priori, our concern in Chapter IV will be to explain the pragmatic theory and to consider the following features of Lewis' theory. It is a crucial feature of Lewis' conception of the a priori that, unlike the unalterable given element in experience, the ~a priori element is just that element that has alternative explications; the thesis of inevitability associated with necessary truth in traditional theories of a priori knowé ledge is a miStake due to the failure of those theOries to account for the senSe.in which and the extent to which the' a priori'is legislated.by mind, not diScovered.w As it' 115 was developments in formal logic which had originally sug- gested the pragmatic theory of the a priori to Lewis, in Chapter IV a question will be raised as to whether the claim that logical truths, the traditional paradigm of a priori knowledge, admit of adequate treatment in Lewis' theory. In precisely what sense, for example, can it be said that the law of non-contradiction is imposed on experience by the mind and admits of alternative explications? Whether Lewis' case can be successfully defended at this juncture is crucial, since logical truths are, and Lewis took them to be, paradigm cases of a priori knowledge and necessary truth. And, more- over, since it was his studies in logic which suggested the pragmatic theory to him, it is of the utmost importance for his theory that its case be made out in that domain.1 The question of whether Lewis is able to draw a sharp or at least clear distinction between a priori and a posteri- ori knowledge, analytic and synthetic propositions, will be the subject of more detailed discussion in Chapter V. There are places in MWO which suggest reasons for thinking that such a distinction cannot be made precisely, particularly when Lewis discusses the role of the a priori in natural science. In this connection some passages will be cited and interpreted-which suggest that what makes a proposition a priori is-our attitude toward it, in particular our refusal to consider the possibility of its being falsified. The re- sulting notion of the distinction between the a priori and 116 the posteriori, the analytic and the synthetic, while incon- sistent with a good deal of what Lewis says about these distinctions, seems consistent with much of his account and has a good deal to say for it.2 Before dealing specifically with Lewis' pragmatic conception of the a priori and the objections that may be raised against it, it will be useful to consider briefly the important motivating role of his earlier studies in logic.3 CHAPTER IV LOGIC AND THE A PRIORI (A) The Debate over Implication It was Josiah Royce, Lewis' teacher at Harvard, who first aroused his.interest in logic. CDuring the year 1910? 1911 Lewis assiSted Royce in two courses in logic and it was at that time that he first encOuntéréd formal logic developed in the logistic manner.4 What troubled Lewis at the time was the existence of some peculiar theorems of Russell and Whitehead's Principia Mathematica. The troublesome theorems were the now familiar "paradoxes" of material implication, that a false proposition implies any proposition and that a true proposition is implied by any proposition. His attempt to deal with these troublesome theorems lead to several arti- cles, notably "Implication and the Algebra of Logic" and "The Structure of Logic and its Relation to Other Systems", as well as his development of his system of strict implication in his Survey of Symbolic Logic.5 It was out of reflections on the criteria for choosing between some such system as Principia and the intensional system of strict implication that the pragmatic conception of the a priori developed. It 117 118 will be instructive to see how Lewis approached the problem, in particular how his studies in logic and the problems he found there inevitably led him out of the strictly logical realm to considerations of general epistemological issues. The difficulty with material implication is that these two theorems, as well as others which we need not go into here, seem to indicate the limitations of systems of material implication as formal codifications of the ordinary meaning of "implies". For clearly, that a false proposition implies any proposition and that a true proposition is im- plied by any proposition are incompatible with the ordinary meaning of "implies" or "is a logical consequence of". That mathematical logic should have initially moved in this direction is no accident, according to Lewis. Lo- gicians earliest interested in developing an exact calculus of logic turned first to the relations of concepts or classes, necessitating a choice between the logic of intension and the logic of extension. The extensional point of view required the special case that if there are no members of a class A, than "All A is B" will hold regardless of the connotation or intension of A or B. That is, the null class is contained in every class. When this algebra is applied to propositions and propositional functions the result is a sYstem contain- ing the paradoxes as theorems. If the algebra is to have this second application, then the properties of implication must be point for point analogous to the properties of 119 relations between classes understood from an extensional point of view. Thus if A=O--that is, if A is false-—than A must imply any proposition. If A-l--that is, if A is true-- then A is implied by any proposition whatever. (This second paradox of material implication is analogous to the corre- sponding theorem in the algebra of classes that every class is contained in the universal class.) Thus applying the algebra to propositions requires the two paradoxes of mateé rial implication. The relation thus designated by "implies" is such that a false proposition implies any proposition, but it does not indicate what it HEELS imply if it $253 true.6 It is questions of this latter sort which a system adequately representing the ordinary notion of implication must be able to answer. .Having become.aware of these initial difficulties involving material implication, there were two related sets of questions which Lewis began to consider. First of all, is there an exact logic, comparable to the extensional calculus of propositions, which adequately represents the intensional analogue of material implication? If so, is this intensional analogue the implication relation upon which the ordinary notion of deductive inference is based? Secondly, if such an exact logic could be developed, would it constitute an alternative to the system of material implication and in what sense is it an alternative? If they are genuine alternatives, 120 upon what criteria does one rely to choose any one of these alternative logics? Is this a question of truth or validity? How can criteria be appealed to without begging the ques~ tion?7 Before addressing himself to this second, broader, set of questions, Lewis proceeded to the logistic develop- ment of the logic of intension. The results of this work were summarized in Chapter V of his Survey of Sumbolic Logic, the resulting system became known as the system of strict implication. This system is both as exact as and more in- clusive than the extensional calculus of propositions, which it contained as.a.sub-system. Moreover, it was Lewis' view that the resulting system was more in accord with the ordi- nary notion of implication than previously strictly exten- .sional systems. Where it did diverge from ordinary logical intuitions, systems of material implication fared no better. That material and formal implication cannot be the basis of ordinary inference and that consequently only an intensional system can be adequate Lewis took to be evidenced by the following feature of material and formal implication. "A materially implies B" means precisely that it is not the case that A is true and B is false. Now while this is a necessary condition of ordinary deductive inference, it is not a sufficient condition. That it is not a sufficient condition.can be seen by noting the following special case of material implication. In particular, in case A is false, 121 there would ordinarily be no point in making the inference from A to B. In fact if A is false, then A materially im- plies B. But sometimes in ordinary deductive inference, we are concerned to make inferences from false assumptions. However, as canon of deductive inference, material implica- tion is here of no help whatever, since it would permit the inference of any and every proposition, a consequence fla- grantly inconsistent with our logical intuitions. That A materially implies B because A is false throws no light on the question of what A would imply if it were true. Simi- larly, A materially implies B if B happens to be true (no matter what A is). In cases such as this we would be un- likely, ordinarily, to make an inference, since our main con- cern in making inferences is to establish conclusions. If we already know that the conclusion is true quite independ- ently of any of the premises, there remains little point to any such inferences. Nevertheless, we are sometimes inter- ested in discovering what implies some known fact (as for example in the construction of scientific explanations), and in such cases the system of material implication is of no assistance as a critique of inference. In other words, "no use can be made of material implication in drawing valid inferences except in those cases in which the implication can be known to hold for some reason other than that the premise is false or that the conclusion is true. When we inquire how we can know that it is not the case that A is true and 122 B false, without knowing that A is false and without knowing that B is true, the only answer is: by knowing that if A were true, B must be true; by knowing that the truth of A is inconsistent with the falsity of B: by knowing that the situation in which A would be true and B false is an impos- sible situation."8 Thus, the only case in which an inference could be based on material implication is just that case in which it coincides with intensional or strict implication, where A Strictly implies B means that the truth of A is inconsistent with the falsity of B. Similar considerations convinced Lewis of the superi- ority of strict implication over formal implication as a codification of the ordinary meaning of "implies". In the formal implication "For every x, Fx materially implies Gx", if "For every x 'is understood to mean' for every actual x", then formal implication represents the ordinary relation of class inclusion and is just generalized material implica- tion. If, however, "For every x" means "For every possible x", then formal implication coincides with strict implication. However, there appeared to be consequences of Lewis' system of strict implication which Lewis had not anticipated and which seemed to threaten that system as the paradoxes of material implication had threatened systems based upon it as ,an.implication relation. These consequences have come to be called the paradoxes of strict implication. The two most 123 familiar examples are that a necessary proposition is implied by any proposition and that a self-contradictory proposition implies any proposition. I should like to indicate how these paradoxes arose and how Lewis attempted to explain away their paradoxicality. Consider first the paradox of strict implication that a contradictory proposition implies any and every proposition. Let "A and not A" be the contradiction in question. "A and not A" implies its former half "A". But by the rule of ad- dition "A" implies "either A or B" for any proposition B. But "A and not A" also implies its latter half "not A". But "not A" in conjunction with "A or B" implies "B". Recalling that "B" is any proposition whatever, it follows that a con- tradiction strictly implies any and every proposition. Consider now the claim that a necessary proposition is implied by any proposition. Assume some arbitary propo- sition "p". "p" implies "p and not q or p and q". (That is, if p is true, then either p is true and q false or p is true and q is true.) But this is equivalent to "p and q or not g", which implies "q or not q", which is a tautology and surely a necessary truth. Thus, from any arbitrary propo- sition we have derived a necessary truth. It follows that a necessary-proposition is implied by any proposition whatever. The consequence of all this is that while Lewis was able to develop an exact logic of intension comparable to the system of material implication (thus beginning contemporary 124 discussion of modal logic), it was not clear that the result- ing system of strict implication was an adequate formal codi- fication of the ordinary meaning of "implication". Lewis, however, did not think that the paradoxes of strict implica- tion implied that strict implication was not what is ordi- narily meant by logical implication. Unlike the counter- intuitive paradoxes of material implication, the "paradoxes" of strict implication were not so much counter-intuitive as r. non-intuitive. Moreover, the only way in which the paradoxes could be avoided would be by refusing to allow the legitimacy of intuitively valid rules of inferences, like addition, dis- junctive syllogism and simplification. Thus, while the para- doxes of strict implication were non-intuitive, they were not counter—intuitive since they followed from the acceptance of other intuitively acceptable logical principles.9 There are, nevertheless, a number of objections that can berraisedragainst the system of strict implication, which .will be discussed briefly here. First of all, it may be challenged-whether the logical principles used in drawing the paradoxes of strict implication are as intuitively plausible 10 This line of objection as Lewis seems to have thought. argues that the paradoxes of strict implication are as genu- inely paradoxical as the paradoxes of material implication. Thus, while strict implication may be a decided advance on material implication, it does not, as Lewis thought, ade- quately codify the ordinary notion of logical implication. 125 Consider for example the proof that an impossible proposition implies any proposition. It is necessary in constructing the proof that an impossible proposition implies any proposition that all such impossible propositions be expressible as ex- plicit contradictions of the form "p and not 'P. But doubt may be raised as to whether this is always possible. For example, the proposition "Whatever is red is extended" is a necessary proposition; or at least Lewis would have to admit its necessity given his explication of analyticity in AKV in terms of inclusion of criteria in mind. In Lewis' termin- ology, my criterion for applying the term "red" includes my criterion for applying the term "extended". The example Lewis himself uses is perhaps more congenial to his position. The proposition "All squares are rectangles" is necessarily true because in determining that the term "square" is cor- rectly applicable to a presented object, I will also have determined that the term "rectangle" is correctly applicable, and for that reason am entitled to assert that "All squares are rectangles" is necessarily true. But the contradictory of a necessary proposition is an impossible proposition. Thus, if "Whatever is red is extended" is a necessary propo- sition, then "Some red thing is not extended" is an impossible proposition. But if this impossible proposition is to be used as the first step in Lewis' argument for the first paradox of Strict implication, then some way must be found of construing it as an explicit contradiction. In the case of "Some square 126 is not.rectangular" this has some degree of plausibility. Thus, to say that whatever is square is rectangular might be construed as saying that "If anything is rectangular and equilateral, then it is rectangular". The contradictory of this would be "Something is rectangular and equilateral, but not rectangular", which is explicitly self-contradictory. Similarly, the contradictory of "Whatever is red is extended" would be "Something that is red and extended is not extended". Reconstruing these impossible propositions so as to render them explicitly self-contradictory provides the first step in the argument that an impossible proposition implies any and every proposition. But one may object to the explication of "Something is square and not rectangular" as "Something is rectangular and equilateral and not rectangular" on the grounds that this explication in some way begs the question. That is, we could not know that squares are equilateral rectangles without knowing independently that "x is square" implies "x is rec— tangular" or that "x is red" implies "x is spatially extended". Moreover, "red" simply does not mean "red and extended". Be that as it may, even if Lewis were to admit that not all im- possible propositions can be accurately expressed as explicit contradictions, we.are still left with infinitely many im- possible propositions which are explicitly self-contradictory and at least ttay could be used in Lewis type arguments for this paradox of strict implication. And, moreover, if one 127 objects to the View that impossible propositions imply any proposition, one is not likely to be satisfied that while this is not true of all impossible propositions, it is at least true of some, namely all those expressible as explicit contradictions. Moreover, Lewis himself might be perfectly content with this, since in AKV he defends the view that im- possible propositions of this puzzling sort are analyzable in terms of his theory of the modes of meaning of terms and propositions, not necessarily in terms of the argument in question here. Furthermore, Lewis' main concern here is with logic as a formal science, and therefore he is concerned only with arguments whose validity depends on their logical form. But the counter examples noted above are just the sorts of arguments whose validity depends on the special content of their constituent terms, not their logical form. The only other means of avoiding Lewis' defense of the paradox would be to question the principles of inference in terms of which the argument is conducted. This line of objection may take any or all of several forms. One may deny that from a conjunction one may infer one of its conjuncts (the rule of simplification), thus blocking the move from "A and not A" to "A" or to "not A". Alternatively, one may deny that one may infer a disjunction given one of its dis- juncts (the rule of addition), thus blocking the move from "A" to "A or B" where "B" is some arbitrary proposition. In addition, one may argue that the rule of addition is 128 justifiable if "either—or" is interpreted extensionally, but that the move from "A or B" and "not A" to "B" is justifiable only if "either-or" is interpreted intensionally. This line of objection criticizes Lewis' argument for the paradox on the ground that there is a covert equivocation on the expres- 11 More drastically, one might accept each sion "either-or". step in Lewis' argument, but deny the validity of the argu- ment as a whole on the grounds that either generally or in this particular case implication is not transitive. Each of these ways of avoiding Lewis' argument has been defended by 12 I should now like to indicate how one author or another. Lewis could avoid each of these objections. Consider first the avoidance of the paradox by deny- ing that one may deduce from a conjunction one of its con- juncts (i.e. the denial of the rule of simplification). One line of reply on Lewis' part will not do. His argument cannot be defended on the ground that each step in the argument can be justified by appeal to a previously proved principle in the system of strict implication, for ttat is the very ques- tion at issue.. The critic objects to the system of strict implication on the ground that it generates this paradox, among others. It is not satisfactory to reply to this line 0f objection that, in effect, the paradoxes are acceptable because they follow in accordance with the system of strict implication. Perhaps the most plausible reason for denying that one can deduce from a conjunction one of its conjuncts 129 is that in the context of ordinary reasoning such moves would have no point. Why atgaa that p is true when you already know that p and q are both true? One might similarly object to a proposition's being deduced from itself. Such an in- ference would have no point. If implication is taken to be a relation between propositions, then no proposition can imply itself, since then there are not two propositions that can stand in any relation to one another. In short, such moves as are represented by, for example, the rule of simplifica— tion, are not genuine moves, because they do not, so to speak, get us anywhere. In essence this line of objection amounts to saying that the move from a conjunction to one of its conjuncts is not a valid inference because it is so triVial, nothing "new" is obtained by making such an inference. But it is one thing to say that the validity of an inference is trivial and quite another to say that it is not a valid inference at all. More over, a good bit of the point of constructing proofs of argu- ments is to show that one proposition or set of propositions implies another by noting that the implication does hold in virtue of a series of trivial inferences that do not, indi- vidually, seem to get us anywhere at all. The rub is that while each step may be trivial, the result of the whole series may not be. Furthermore, while there may in certain cases be no point in making use of the rule of simplification, that will be determined not by whether one may infer from a 130 conjunction one of its conjuncts, but whether one will. And whether one will depends upon what consequences one wishes to focus attention upon as conclusion. There is another version of this objection to the rule of simplification which warrants mentioning. Sometimes the objection is raised not against the rule of simplification in general, but only against its particular application to explicit contradictions. Thus, it is argued not that any- thing follows from a contradiction, bdt that nothing does. Brand Blanshard picturesquely sums up this line of objection as follows: Take the man whose mind is intelligent, but as yet unen- lightened by symbolic logic; start with the statement that p is at once true and false, and try to exhibit what follows from it; how far will you go before he pro- tests that he does not know what you are about, that your very first statement, p is true and also false, seems senseless to him, and that the drawing of deduc- tions from the senseless seems equally senseless? This, one suspects, is the line he would take; and in this respect he stands for the rest of us. What we commonly mean by 'deducible' and what is here meant by 'strictly implied' are not the same.13 This line of objection will not bear scrutiny, how- ever. In the first place we do frequently draw inferences from contradictory propositions. For example, whenever we try to show that someone holds inconsistent beliefs, say about causal determination and the conditions of moral re- sponsibility, we do so by making inferences from some allegedly inconsistent set of premises. Putting restrictions on the making of inferences as suggested by this line of objection 131 would considerably restrict the scope of legitimate infer- ence. Bad as this may be, to do it in the name of what is commonly meant by "deduce" or "imply" is not plausible. The critic might reply as follows. When we try to draw conclusions from inconsistent premises we are not really making inferences from explicitly self-contradictory propo- sitions. Generally, each of the premises in such an argument is self-consistent: it is their simultaneous assertion that is inconsistent and impossible. In such cases we draw con- clusions, although we cannot from explicit contradictions. There are two peculiarities about this line of objection. First of all, the decision not to allow inferences from ex— plicit-contradictions, but to allow them in the case of inconsistent sets of premises, seems ad hoc. Its only justification appears to be that it avoids the paradox in question. Furthermore, to any argument with a logically inconsistent set of premises there corresponds an argument with only one premise that is explicitly self-contradictory and which is valid if and only if the original argument is valid.. If one.allows inferences in the one case, but not in the other,.one will be hard pressed to defend that dis- tinction.as any more than arbitrary stipulation. Such a proposal would make the validity of arguments depend not just 0n their.form or even their content, but on their manner of expression, a position that is hardly appealing and has little if any foundation in the common usage of "implication". 132 Similar replies may be made to the objection that one cannot infer a disjunction given one of its disjuncts as a premise. Thus, while such an inference may be trivial, its triviality and obviousness does not detract from its validity. Moreover, while in certain cases one may see no point in using the rule of addition, since a disjunction is always logically weaker than either of its disjuncts, one may be little concerned that the disjunction follows; never- theless, this in no way indicates that it does not follow.14 A further objection to Lewis' argument goes as fol- lows. The move from "A" to "A or B" is justifiable if "either-or" is understood truth-functionally. But, the move from "A or B" and "not A" to "B", the next move in Lewis' argument, is valid only if "either-or" is understood in some stronger, intensional sense. Therefore, there is no one unambiguous sense of "either-or" in terms of which the whole of Lewis' argument is valid. There is more to this objection than meets the eye. However, there is not sufficient space to do.more than provide the bare bones of it. In the hands of Anderson and Belnap the objection is serious since besides finding fault with the paradoxes of strict implication, they are among the few critics of Lewis who have taken the pains to propose alternative systems embodying their own views about implication. There are several points that need to be kept in mind in order to appreciate Anderson's and Belnap's objection 133 to Lewis. “First of all (and not without reason), they are extremely reluctant, on informal ground, to accept strict implication as the meaning of "entailment". Though not given by Anderson and Belnap, one might suggest the following as a reason for such reluctance. Not only does it seem paradoxi- cal to claim that a contradiction implies any proposition, but there are other results in the system of strict implica- tion which are even more shocking from an informal, pre- systematic point of view. According to the system of strict implication, a proposition which is not consistent with itself is not consistent with any proposition. Put symbolically, ’b (p o p) ‘3 ’b (p o q) . (See Lewis and Langford, smolic Eggtg, p. 167, Theorem 19.1) But according to the same sys- tem, a proposition which is self-contradictory or impossible strictly implies any proposition (i.e. NO pr; . p9 q). That a proposition should imply what it is not even consis- tent with seems not just counter-intuitive, but false to the point of self-contradiction. Secondly, the "fallacy" which Anderson and Belnap find in Lewis' "proof" of the paradox is really two-fold. The paradox as a whole commits what they term a fallacy of relevance. ". . . we regard a contradiction A- -A as in general irrelevant to an arbitrary proposition B, and we accordingly think of the principle "(A and not A) implies B" as embodying a fallacy of relevance." In their paper "The Pure Calculus of Entailment" they offered a formal theory 134 which was intended to capture this notion of relevance and which would be more in accord with our pre-systematic intui- tions about entailment. Nevertheless, even if this could be done, it would not itself alter the epistemological issue that Lewis is raising. Granted that the pure calculus of entailment adequately (or more adequately) represents the ordinary meaning of implication, what would justify us in adopting it rather than some alternative system? That it is more in accord with common meaning begs the question. Thus, even if Anderson and Belnap succeed, the epistemological issue will only have shifted ground, not changed in any essential feature. The second, more specific, fallacy which Anderson and Belnap.find in Lewis' argument is a fallacy of ambiguity. ". . . there is a sense in which the real flaw in Lewis' argument is not a fallacy of relevance but rather a fallacy of ambiguity." The ambiguity consists in the following. As did Lewis before them, Anderson and Belnap distinguish two senses of "either-or", an extensional and an intensional sense. In an extensional disjunction there is no connection between disjuncts other than their not both being false. In an intensional disjunction, the disjuncts are so related that the falsity of one implies the truth of the other. With this distinction Lewis especially would have no quarrel. However, Anderson and Belnap draw a consequence from this distinction with which Lewis would take exception. 135 According to Anderson and Belnap the fact that there are two distinct senses of disjunction implies that certain traditionally accepted rules of inference will be seen to apply to one sense of disjunction, but not to the other. In particular, the rule of addition applies only to the exten- sional sense of disjunction not to the intensional sense. With this Lewis would agree. However, according to Anderson and Belnap, if the relevant disjunction in Lewis' argument is extensional, then the use of the rule of disjunctive syl- logism in passing from "A or B" and "not A" to "B" is invalid since it commits a fallacy of relevance. More precisely, the principle of disjunctive syllogism does not, according to Anderson and.Belnap, validly apply to merely extensional or truth-functional disjunctions. Lewis' argument thus equivo- cates-on the work "or" and is therefore fallacious. Now there are some important facts behind the dis- tinction between extensional and intensional disjunction, and furthermore it is quite true that the rule of addition does not apply to intensional disjunction, but applies only when "either-or" is understood truth functionally. But it is an altogether different thing to say that the rule of disjunctive.syllogism does not apply when "either-or" is understood truth functionally. It is difficult to see why Anderson and Belnap take this position since the only reason they seem to offer is that common usage of the disjunctive syllogism rule is to 136 disjunctions when our grounds for accepting a disjunction are other than that we accept one or the other or both dis- juncts of the disjunction (i.e. where the disjunction is other than truth-functional). This does not seem sufficient, however. Suppose that we are dealing with a truth-functional disjunction "either A or B". Let us suppose that our reason for accepting such extensional disjunctions must be because we already accept one or the other or both of "A" and "B". That is "Either A or B" will be accepted if one of the fol- lowing holds: (a) we accept "A"; (b) we accept "B"; or (c) we accept "A and B". Now consider the following application of the dis- junctive syllogism rule: "A or B" and "not B" implies "A". Now, if "A or B" is accepted because "A" is accepted then what we have is a trivial inference. The same holds if "A or B" is accepted because "A and B" is accepted, since we then in effect have a case of simplification, which Anderson and Belnap admittedly accept. But, if "A or B" is accepted because "B" is accepted, then we have a contradiction. Thus in the end the rule of disjunctive syllogism is rejected because its.application to truth-functional disjunctions is trivial. But again, that an inference is trivial does not imply that it is therefore invalid. In addition, this argu- ment conflates logical and epistemic issues. What follows from a proposition does not depend on how one came to accept 137 it. This line of objection may be disposed of in the follow- ing manner. It is true that the move from "A" to "A or B" is valid only if "either-or" is treated truth-functionally, but the move from "A or B" and "not A" to "B" requires no more than a truth functional interpretation. One is simply inferring "B" from "not A" on the assumption that one or the other of "A" or "B" is true. No closer meaning connection or intensional connection between "A" and "B" is required in Lewis' argument. One final objection to Lewis' argument is that while each step in the argument is a valid implication, the argu- ment as a whole is not, because either in general or at least in this particular case implication is not transitive. Now to say that implication is not in general transitive would be absurd, especially if this line of objection is intended to justify the claim that strict implication does not repre- sent the ordinary meaning of "implies". And to say that implication is not transitive in this particular case is quite arbitrary unless some criterion for distinguishing transitive from non-transitive cases of implication is provided. And, moreover, the criterion cannot be based solely on its ability to avoid the paradoxes of strict implication, for then the question has been begged. Probably the underlying motivation for the reluctance to accept the paradoxes of strict implication is that both in the case of the derivation of any proposition from a 138 contradiction and of a necessary proposition from any propo- sition there arises the feeling that in order for an implica- tion to be a valid one, there must be some meaning connection between premises and conclusion. Similar motives may have been behind the apprehension which met material implication. Thus, one's reason.for denying that "McGovern will be elected President" is implied by "Today is Friday and it is not Fri- day" is that the two propositions have no significant meaning relation to one another. One would think that Lewis of all people would be sympathetic to this line of objection, since he persistently insisted that logical implication was a matter of intensional relations. But surely, there is no meaning connection between what day of the week it is and who will be elected President. To this Lewis would reply that while implication is ultimately based on meaning connection, such meaning connection justifies the paradoxes as valid implica- tions. Valid implication is based upon meaning in the fol- lowing sense: in a valid inference the meanings of the premises and conclusion are such that whenever the premises are true, the conclusion must be true as well; that it is impossible for the premises to be true and the conclusion false. But that is precisely what is intended by strict implication. Whatever other notion of meaning connection is required is unclear. One may, of course, continue to say that the paradoxes of strict implication are counter-intuitive, 139 but in the absence of better arguments, Lewis is entitled to claim that they are not so much counter-intuitive as non- intuitive consequences of the ordinary meaning of "implica- tion". Others, for example Quine, despair of the whole idea of intensional logic, arguing that the whole dispute over material implication and strict implication was the result of a confusion. The confusion resulted from failing to dis- tinguish implication, which is a semantic relation, from a statement connective like "if-then". "Implies" is a verb properly flanked, not by statements, but by the names of statements. Russell and Whitehead precipitated the contro- versy by their informally rendering the horseshoe as "impli- cation". Lewis, rightly objecting to this as the meaning of "implies", perpetuated the confusion. What Lewis wants to express by the use of the term "strict implication" can be adequately expressed at the metalinguistic level. To say that one statement implies a second is to say that the corresponding conditional is analytic. Thus, if "p" implies 15 Lewis, to be sure, "q", then "if p, then q" is analytic. would object that such a procedure in effect gives up the attempt to formalize the principles of logic. If it is those principles which justify inference that one is attempting to codify in a deductive system, then it is important that they be expressible in that system. Moreover, in this respect, the system of strict implication has a decided advantage over any purely extensional system. 140 Moreover, whatever its merits, Lewis persistently held the view that unless an adequate intensional logic can be developed, we will not be able to make sense of the idea of logical truth, and more particularly, that relying simply on extensional logic the distinction between the logical and the empirical, the necessary and the contingent, will have to be given up. Thus, shortly before his death, he made the following remarks concerning the importance of intensional logic. . . . those who carry forward this exclusively extension- alist conception of the logical, and refuse to recognize any significance in the logic of intension, are bound eventually to challenge the validity of the distinction between the analytic and the synthetic. And in result, barring some alternative notion which has yet been put forward, they are then left with no ground for distin- guishing a priori from a posteriori, and no clear con- ception of the.foundations of logic and mathematics as distinct from physics and psychology.15 Whatever views one may take as to the correct analy- sis of the issues involved, for our present purposes it is important to consider what significance Lewis attached to them, and in particular what significance they had toward his development of the pragmatic conception of the a priori. At this point Lewis had answered to his own satisfaction the first of the questions which motivated his logical investiga- tions. One could.develop exact systems of intensional logic and in particular could construct a system which codifies 17 the ordinary meaning of implication. But this only made his second set of questions more pressing and even more 141 difficult to answer. If there are different exact logics, how could one choose between these different logics without begging the question? Moreover, with the development of many-valued logics, the problem became even more difficult of solution. Ultimately, the question was one not of truth, but a question turning on what Lewis called pragmatic consid- erations. Moreover, even if the case can be successfully argued that strict implication is an accurate analysis of the ordinary notion of implication used in logical criticism, of what particular aid is that fact in justifying our use of that system rather than some other, say material implication or some many-valued system? To say that such choice is justi- fied simply because it is that notion of implication that we habitually use would be a flagrant begging of the question. It is to these questions that Lewis turned in his important paper of 1920, "The Structure of Logic and its Relation to Other Systems". At this point Lewis was forced by his own logical investigations to deal with an epistemological issue. (B) Alternative Logics.and themPragmatic.A Priori The first of two issues in Lewis' epistemology that I should now like to critically evaluate is his critical treatment of attempts to show that some particular logic is absolute and admits of no conceivable alternative. This issue is treated in his article on the structure of logic 142 referred to above. The same issues and arguments are dis- cussed again, with only minor alteration, in MWO Chapter VII. The second issue is his attempt to show in a later article that there are genuinely alternative, not simply equivalent, systems of logic, and that this constitutes strong evidence that a priori knowledge is precisely that element or aspect of knowledge which has alternative explications; that necessary truth does not connote inevitability and is in some sense made by mind. The article in question is his 1932 paper 18 After having explained "Alternative Systems of Logic." Lewis' arguments for these two theses it will be argued that only in a trivial sense are there alternative logics in the sense these are understood by Lewis and that the convention- alist interpretation of logic suggested by much of Lewis' position in MWO is not tenable, that in no clear sense is logical truth "made by mind" or admit of conceivable alterna— tives; in fact this position is quite inconsistent with Lewis' own criticism of conventionalist interpretations of analytic truth as these were developed later in AKV. The first series of arguments were presented in "The Structure of Logic and its Relation to Other Systems" to which we may now turn. Traditionally there have been three different views concerning the epistemic and ontological status of principles of logic. The first, or Aristotelian, position maintains that logic is formal and at the same time concerned with the actual modes of right thinking, that logic is both formal and 143 criteriological. The second view, the "modern logic" usually associated with idealism, treats logic as fundamentally con- cerned with modes of right thinking, that is as Criteriologi- cal, and for that reason denies that logic can be a purely formal discipline. The third view, the so-called new logic, associated with the tradition of Frege and Russell, treats logic as a purely formal discipline and for that reason re- jects any attempt to describe the nature of correct thinking as a psychological process. To attempt to do both of these is to fall into the trap of psychologism which Frege was de- termined to avoid. Thus conceived, logic is not a matter of depicting any process of thinking, but only the formal cri- teria of valid inference. It is this conception of logic which Lewis adopts. The question to be considered here, in the context of Lewis' epistemology, is whether or how funda- mental principles of logic can be justified. Traditionally, there have been three ways of justify-. ing principles of logic: (1) to present the principles of logic as demonstrable from self-evident axioms (e.g. the traditional laws of thought); (2) to argue that the principles of logic are a priori and necessary because they are presup- posed by science, or sometimes experience in general; and (3) to argue that principles of logic, like the law of non- contradiction, are necessarily true because to deny them is to affirm them, that to justify logical principles is to defend them by a particular method of proof traditionally 144 called reaffirmation through denial. Lewis argues that each of these views on the foundations of the principles of logic is fundamentally mistaken and that exposure of their mistakes will provide insight into the pragmatic nature of the a priori. More specifically, that the necessity of the a priori "is its character as legislative act. It represents a constraint imposed by the mind, not a constraint imposed upon mind by something else."19 Consider the first view, that to justify the principles of logic is to present them as demonstrable from self-evident axioms. This procedure is modeled, of course, on the informal axiomatic presentation of geometry associated with Euclid. However, much has transpired since Euclid and recent concep- tions of deductive systems lend no support to this view re— garding the necessity of logical principles. In the first place, to develop a system of knowledge in a deductive system, whether that be logic or some branch of empirical science, is not somehow to prove that system true. A deductive system is not a method of proving truth, but a method of presenting results; it is not a method used to prove conclusions, but a method of exhibiting relations.20 This illusion has been perpetuated because of the strong in- fluence of Euclidean geometry on historic conceptions of the a priori. However, advances in mathematics, particularly the development of non-Euclidean geometries, have dispelled this view of deductive systems in mathematics and requires a 145 re-evaluation of traditional conceptions of the a priori. Moreover, in more recent logistic developments of areas of mathematics, far from the axioms demonstrating the necessary truth of theorems by shedding their self-evidence upon them, the theorems are more likely to be considered self-evident than the axioms from which they are derivable. For example, in logistic developments of propositional logic, traditional self-evident principles of logic like the law of non- contradiction more frequently appear as theorems than as axioms. Moreover, the axioms for such systems are no more simple than its theorems. Accordingly, the selection of certain propositions as axioms is not based on any alleged self-evidence they may have, but simply on grounds of the deductive power and formal simplicity of the system thus developed. To say that such axioms are logically prior to theorems derivable from them is not to evidence their self- evident necessity, but simply their systematic relation to other propositions. Theorems of such deductive systems are no more proved true by their derivability from certain axioms than the axioms are proved true because certain theorems are derivable from them. Thus, the traditional term "axiom", long associated with self-evident necessity, has been replaced by such weaker terms as "postulate" or "primitive proposition". Therefore, the nature of deductive systems renders no support to the View that principles of logic are first principles in the sense of self-evident axioms from which other principles of logic are demonstrable. 146 The second view regarding the foundation of the neces- sity of principles of logic argues that such necessity is evi- denced by the fact that logic (though not logic alone) is presupposed by science or perhaps presupposed by objective experience in general. Since principles of logic are necessary presuppositions of science, as well as of mathematics, this is to somehow show the necessary truth of the principles of logic. To this Lewis argues in a twofold manner. First of all, the inotions of presupposition and logical priority associated with this view are the result of confusing logically necessary and logically sufficient conditions. And secondly, once this confusion is pointed out, this method of proving necessary truth-can be seen to be without merit.21 First of all, what is meant here by "presupposition" and what is it for one fact or proposition to presuppose another? -To say that B presupposes A, according to Lewis, is to say one or the other of the following two things: either (1) A is logically prior to B or (2) A is in some way a logi- cally necessary condition of B and is implied by it. Usually, proponents of this view of the necessity of principles of logic maintain both of these notions of presupposition simul- taneously. That is, they wish to maintain that if B presup- poses A, then it is true both that A is logically prior to B liaaa that A is, in some sense, implied by B. But it isn't clear that this view of presupposition can be consistently 147 maintained, and furthermore, even if it could be, it would provide no reason to accept the necessity or a priority of what is thus presupposed. Consider first the view that "B presupposes A" means that A is logically prior to B. But what could here be meant by "logical priority"? Ordinarily what is logically prior to a fact or proposition will imply that fact or proposition; thus, to say that B presupposes A is to say that A implies B, or is a logically sufficient condition of B. If so, then no support is given to the view that what is thus presupposed is either necessary a priori or indispensable. In the first place, "sufficient condition" should not be confused with "necessary condition". To say that B presupposes A in the sense under consideration is to say only that B is true if A is. That A is a sufficient condition of B (even if B is as- sumed to be true) is not even to imply the truth of A, much less its necessitya That we should fix on A as presupposi- tion may be due to nothing more than lack of imagination in considering alternatives. In this sense of presupposition that a principle or set of principles is presupposed by some fact or proposition shows neither the necessary truth nor the indispensability of what is thus presupposed. Sometimes when it is claimed that B presupposes A it is suggested that there is some logical connection between the truth of B and that of A. That in a special sense B implies A. But then to say that B presupposes A is to say that A is 148 a logically necessary condition of B. This interpretation of presupposition as necessary condition preserves the logical connection between what presupposes and what is presupposed, but would sacrifice the alleged logical priority of such presuppositions. So interpreted, the presuppositions of a fact or proposition would be as numerous as its logical consequences and, as Lewis puts it, the fine'glamOur of the word would be lost. ‘Furthermore,.showing that a given fact or proposition, B, presupposes another, A, does not thereby establish the necessity of A, unless B itself were already established as necessary. But this is of no help whatever if the facts or propositions with which we begin are contingent propositions. That a proposition is implied by a contingent proposition is not proof of its necessity. At least some of the logical consequences of a contingent propo— sition will themselves be contingent. Thus there is no clear sense of presupposition in terms of which it can be demon- strated that there are general principles of logic which are at once presupposed by science in general and established as a priori necessary in virtue of being thus presupposed. The third traditional method of demonstrating the necessary truth of the principles of logic is closely related to the method of presupposition. It has been argued that principles of logic are a priori and necessary because they are presuppositions of any (rational) thinking in the sense that to deny them is by implication to affirm them. While 149 Lewis admits that there are propositions which are genuinely implied by their own denial, he nevertheless argues that such propositions are not by that fact proved true, much less necessarily true.22 To show why Lewis claims this it is necessary to look more closely at what he says about the no— tion of reaffirmation through denial. Consider, for example, the attempt to prove the neces- sity of the law of non-contradiction by this method. The general point Lewis wishes to make may be summarily indicated as follows. Whoever asserts a self-contradictory proposition does not explicitly assert and deny the content of his asser- tion. He affirms it in fact, but denies it by implication only. Thus, if one denies a principle of logic, then he can either draw inferences according to the principle he denies, or he may avoid the principle in making inferences. If the former route is taken, then one will contradict oneself by affirming what one previously denied, but only because one has already assumed what is to be proved. If the latter route is taken, avoiding use of the rejected principle, no self- contradiction will be incurred whatever.23 Suppose we take the principle of non-contradiction in the following form: "It is false that p is true and p is false". Its denial would be "p is true and p is false". But, the argument goes, to affirm this is to deny it, as can be seen by the following argument. That p is true and p is false implies its latter half, that p is false. But that 150 implies that it is false that p is true. But if it is false that p is true, then it must be false that p is true and p is false. Thus, to deny the law of non-contradiction is to affirm it. But the argument looks persuasive only because, while we denied the principle of non-contradiction in the first premise of the argument, we introduced it in arguing from "p is false" to "it is false that p is true". But if we had, consistent with the original assumption, refused to use the law in question as a rule of inference, no contradiction need by incurred. All that the method of reaffirmation through denial indicates is that all demonstration in logic is circular. If you begin by assuming the truth of a prin— ciple of logic, then of course if you then go on to deny it you will have contradicted yourself. But you will have done so only because you were not sufficiently persistent in your denial of the principle in question. This is all that reaf- firmation through denial can show. It cannot be used to prove the truth of any particular system of logic, since it cannot distinguish "good" logics from "bad" ones. Thus, if one begins with a set of "bad" logical principles, but consist- ently reasons in accordance with them, to deny any one of them will be to reaffirm them. Thus, even a thoroughly false logic may possess the merely methodological consistency evi- denced by the method of reaffirmation through denial. Apparently those who set store by the 'reaffirmation through denial' have committed the fallacy of illicit conversion; they have reasoned: 'A.logic whose principles 151 are true will give their reaffirmation through denial. Therefore, whatever principles meet this test must be true.24 The consequence of all this is that there appears to be no way in which logical principles can be proved true. Any attempt along the lines suggested above turns out to be circular in that the disputed principles are already assumed in the demonstration of them, or they turn out not to have been demonstrated at all. If, as suggested by Lewis, there are alternative systems of logic and if no one of these sys- tems can be proved true, then we are faced with the necessity of choosing one of these systems (since as rational beings we cannot help making inferences and drawing conclusions), but in the strict sense there are no logical criteria for any choice we might happen to make. Whether such choice will be even rational will depend upon whether we choose to include what Lewis calls pragmatic criteria as rational. But are there genuinely alternative systems of logic and if so in what sense are they alternatives? To answer this question affirmatively, as Lewis does, is to assume a heavy burden of proof, for to speak of alternative logics goes implicitly against philosophical tradition and is prima facie incompatible with ordinary notions about logic. Lewis tried to defend his view in his 1932 paper "Alternative Systems of Logic". 152 Lewis there argues that, in the light of developments in mathematical logic, it is dubious whether there are laws of logic in the traditional sense that admit of no conceivable alternatives. More specifically, by means of brief illustra- tions of alternative systems of logic, Lewis attempted to show three things. (1) There are no 'laws of logic' which can be attributed to the universe or to human reason in the traditional fashion. What are ordinarily called 'laws of logic' are nothing but explicative or analytic statements of the meaning of certain concepts, such as truth and falsity, negation, 'either-or', implication, consistency, etc. which are taken as basic. A 'system of logic' is nothing more than a convenient collection of such concepts, together with the principles to which they give rise by analysis of their meaning. (2) There are an unlimited number of possible systems of logic, each such that every one if its laws are true and is applicable to deduction. These systems are alterna- tives in the sense that concepts and principles belonging to one cannot generally be introduced into another-- because of fundamental differences of category. . . . (3) Sufficiency for the guidance and testing of usual deductions, systematic simplicity and convenience, accord with our psychological limitations and our mental habits, and so on, operate as criteria in our conscious or uncon- scious choice of 'good logic'. Any current or accepted canon of inference must be pragmatically determined. That one.such.shou1d be thus accepted does not imply that alternative systems.are false: it does imply that they are--or would be thought to be--relatively poor instru- ments for the conduct and testing of our ordinary infer- ences.25 There are, according to Lewis, two fundamental diffi- culties encountered in the presentation of what he calls non- Aristotelian logics. First of all, we lack the appropriate intuitions for judging the correctness of any such non- Aristotelian system in its own terms. And secondly, the 153 development of any such system requires drawing inferences. How is it possible to make such inferences without falling back on Aristotelian principles in our manner of drawing them? Lewis attempted to solve both problems in terms of the matrix method. This method will provide a mechanical decision procedure for those cases for which we lack the appropriate intuitions. Sufficient practice with such tenchniques will enable us to draw inferences in a non-Aristotelian manner.26 By the matrix method Lewis means the familiar truth table method for specifying the semantics of logical connec- tives. He first illustrates this technique by introducing the truth-funcional connectives symbolized by the wedge and the tilde, showing how to assign values (in this case 1 or true and 0 or false) to formulas given the assignment of values to the formulas' constituent propositional variables. The matrix thus generated provides the meaning of the two logical connectives. The other truth-functional connectives can be defined in terms of the wedge and tilde. Thus, conjunction, material implication and material equivalence are introduced as defined logical connectives. To determine whether the formula is a law, one simply determines whether the formula always has the value 1 under every possible assignment of 1 or O to its constituent propositional variables. The logic thus introduced is non-Aristotelian in the sense that non- Aristotelian meanings are provided for the logical functions "p > g" and "p E qll.27 154 The same technique suffices, according to Lewis, to develop any number of different non-Aristotelian logics all of whose laws are true. [To show this Lewis introduces two other non-Aristotelian systems in addition to the one sketched above. The first is the three-valued system, generated by the following matrix: :C 1 1/2 0 N 1 1 1/28 0 0 1/2 1 1 1/2 1/2 0 1 1 1 1 "Np" represents "not p" or Jp is false". "qu" is what Lewis calls the implication relation of the system, although its meaning is not identical with the implication relation of the system of material implication sketched above. On the basis of these two functions, three other functiOns can be defined whose meaning is comparable to disjunction, conjunction and equivalence. The fundamental matrix together with the defin- ition of "O" (disjunction), "A" (conjunction) and "E" (equiv- alence) give the following truth tables for the functions of the system. bU w .n C9 0 I-Q ooo-oou-OL—II—H-“U o orao OFJO°OFHQ HP‘F‘ML‘F‘OHQFHU o-oI—I-u-oI—II-Jl-II-“U H'OO°\)t-'°\JO°\Jl-"'U OOOO'U‘OO'GH 155 In order to indicate the differences between the two systems thus generated it suffices to compare the laws of the two systems. Reading the functions informally, as suggested above, the following are laws in both the two-valued and the three—valued system.28 1. p implies p, every proposition implies itself p and q is equivalent to q and p p is equivalent to not not p '4. p implies p or q p and q implies p p implies q is equivalent to not q implies not p 7. If p implies q, then q implies r implies p implies r 8. If p is true, then q implies p (a true proposition is implied by any) 9. If p is false, then p implies q (a false proposition implies any) ‘10. If p is true, then not p implies p (every true propo- sition is implied by its own denial) Up to this point the two systems are completely analgous. Laws 1 through 7 are Aristotelian, but 8, 9 and 10 are not. However, the following four principles hold in the two-valued, but not in the three-valued system, and all of them are Aris- totelian. 11. If not p implies p, then p is true 12. If p is true, then it is false that p implies not p 13. If p implies not p, then p is false 156 14. Either p is true or p is false. (the law of excluded 7 4 middle) That 14 does not hold in the three valued system must not be misunderstood. The problem is one of interpretation. It ap- pears as if the three-valued system does not, unlike the two- valued system of material implication and Aristotelian systems, make the traditional law of excluded middle a lOgical law. This difference is generatedbecause the systems in question are categorially distinct; their basic categories(or values) differ. Moreover, since the systems differ as to their fund- amental categories, the symbol "0" has a different meanifig fromnthat-of the usual disjunction. In order to more clearly understand the three-valued system, Lewis proposes the fol- lowing interpretation of its fundamental categories (values). Let p=l be p is certainly true, let p=0 be p is certainly false and let p=? be p is doubtful. Thus interpreted, the function "p 0 q" is doubtful in case both p and q are doubt- ful, except in that case in which p and q are so related that if one is false the other must be true (what Lewis elsewhere refers to as intensional disjunction). In that case both p and q may be doubtful, but "p or q" is not doubtful. Thus, the fact that p 0 Np is not a law does not mean that for this logic the traditional law of excluded middle is false, but that that law is not expressible in terms of the categories "certain", "certainly false" and "doubtful". The 157 interpretation of "either-or" according to which "p or not p" is a law is not significant when propositions are classi- fied in terms of this trichotomy.29 In addition to this three-valued system Lewis also introduces another system, this time a four—valued one. As in the case of the two previous systems the laws of this system are not identical with those of the others. The funda- mental matrix for this system is as follows. v 2 2 3 4 - l l l l l 4 l 2 l 2 l 2 3 l 3 l l 3 3 2 1 4 1 2 3 4 l 4 Four other functions may be defined as follows: p q —(—pv-q) ("p and q are both true") p < q = - 0 (p -q) ("p implies q") qu = <>(p q) ("p and q are consistent") (p=q) = (p < q)(q < p) ("p is equivalent to q") The fundamental categories (values) of this system may be interpreted as follows: p=l, p is necessary; p=2, p is true but not necessary; p=3, p is false but not impossible; p=4, 30 The implication relation upon which this is impossible. system is based is again different from the other two, al- though it is closer to the Aristotelian notion. Interpreted in the manner suggested by Lewis, every law of each of the systems is a universally true formula. The question to be considered now is in what sense these 158 systems of logic are alternative and in just what sense can it be said that we are free to choose some one of these sys— tems over the others. Lewis was not satisfied with proposing alternative logics by the matrix method alone. Since it was the discovery of non-Euclidean geometries and the application of the logis- tic method to logic itself that first suggested the idea of alternative logics, it is not surprising to find Lewis de— veloping such a system to illustrate his general thesis re— garding alternative systems of logic.‘ In the course of discussing traditional views of the a priori as they relate to logic, Lewis offered the following postulate set as defin- ing one of those "queer" logics which, had our purposes been different, we might have chosen over the one we presently use. 31 A. -(-p) =P B. -(p< -p) c. (p < q) < (-q < -p) D. (p<(q p" is an independent axiom in some axiom set for truth-functional logic, then if we replace this axiom by its-denial, the result will be a formal system which is for- mally consistent. The result would be a logic which is an alternative to the traditional logic of truth functions, since in this system material implication would not be re- flexive. 178 This reply on Lewis' part will not do, however. In the first place, just as in the case of the derivation of theorems from axioms in both Euclidean and non-Euclidean geometry, we had necessary hypothetical propositions which have no alternatives, so here the proposition that given one set of axioms, certain consequences follow, and that given another set, different theorems follow, are both necessary, but are not alternatives in the sense that we must choose between them. The real lesson of this illustration will be discussed in Chapter V. As Lewis Carroll humorously indicated in "What the Tortoise said to Achilles", no inquiry of any sort can ever begin if everything is simultaneously up for review. In Chapter V a theory of the a priori will be developed along these.lines which is quite compatible with the general epis- temological view of Lewis. But perhaps this is to misconstrue what Lewis means by saying that various conceptual systems have alternatives. Lewis perfectly well admits that about any conceptual system there is an absolute truth which does not admit of possible alternatives. The element of choice involved is not a choice as-to the purely formal characteristics of Euclidean or non- Euclidean geometry. The question is, which of these two perfectly self—consistent geometries are we to choose to interpret in terms of physical or perceptual space. Whether 179 we are to choose to understand physical or perceptual space in Euclideanvor non-Euclidean terms given certain denotations assigned to their primitive terms is where there is an ele- ment of choice. The separation of the factors (in knowledge), however, reveals the fact that the pragmatic element in knowledge concerns the choice in application of conceptual modes -of interpretation. On the one side, we have the abstract concepts themselves, with their purely logical implica- tions. The truth about these is absolute. . . . . . . Such purely abstract a priori truth answers only to .the criteria of consistency and adequacy. It is absolute and eternal. On the other side, there is the absolute brute-fact of given experience. . . . It is between these two, in the.choice.of conceptual system for application and.in the assigning.of sensuous denotation to the ab- stract.concepts that there is a pragmatic element in truth and knowledge.41 Similarly, in his paper "The Pragmatic Element in Knowledge" Lewis argued that there are three distinguishable elements in knowledge, the purely given, the abstract concept and the act which interprets the one in terms of the other. The pragmatic element in knowledge is this act of interpretation.42 The question for consideration here is whether we can.treat logic in a similar fashion. I do not think that we can. In the first place, so long as we consider a formal system in the abstract, without any reference to possible interpretations, there is no question of truth or falsity at all. Consequently, there can be no question of necessary truth either. .While we can make statements about the formal system which can meaningfully be said to be true or false, the formal system itself does not, in the absence of some 180 suggested.interpretation, express true or false propositions. Thus, the fact that we can construct a different formal sys- tem by replacing an independent axiom in a system whose in- tended interpretation is truth-functional logic is no indication that in this new formal system we have an alterna- tive to the traditional logic of truth functions. That is a question of what possible interpretations can be made of this new formal system. Recall that the axioms of a formal system limit the range of possible interpretations of the primitive signs of the system. Thus the postulate "p > p" in some formal system whose intended interpretation is truth func- tional logic, limits the range of possible values of ">" to reflexive relations. In the new formal system generated by replacing this axiom with its denial, the limits of possible values.has thereby been altered. In this new system, there- fore,.">" cannot be interpreted as material implication since it is no longer reflexive. This new formal system does not, therefore, present us with a new logic which we must somehow adopt-or refuse to adopt in place of some other system. In conclusion, neither the existence of alternative geometries nor the elaboration of matrices for many valued logics implies that there are alternative logics with alter- native implication relations that we must choose among as alternative logics. What Lewis presents as alternative systems of logic turn out not to be alternative, and do not reject or ignore traditional laws of thought in an unequivo- cal sense. 181 Recall that for Lewis logic is a paradigm case of necessary truth and that it was his investigation of what he called alternative systems of logic that suggested Lewis' pragmatic conception of the a priori. We have argued that Lewis is unable to make out his case for alternative logics. Therefore, if logical truths are necessary truths, then Lewis is forced to admit that at least a subset of necessary truths do not admit of alternative explications. CHAPTER V A FUNCTIONAL INTERPRETATION OF THE A PRIORI In the remaining.chapter of this part I should like to raise and answer a question concerning the manner in which Lewis distinguishes the a priori from the a posteriori, the analytic from the synthetic, in MWO. In particular, it will be argued that there is a fundamental tension in Lewis' ex- plication.of.these distinctions. In MWO Lewis seems to vacillate-between two opposing views as to what makes a propo- Asition a priori.. The two views may be called respectively the functionalist.view and the absolutist view. According to the functionalist view, what makes a proposition a priori (and hence for Lewis analytic) is not the inherent nature of the proposition, but our attitude toward it. In other words, what makes a proposition a priori is our refusal to entertain the possibility of its being disconfirmed. According to the absolutist view, what makes a proposition a priori is the .proposition itself. According to the absolutist view, the functionalist interpretation appears plausible as an inter- pretation of Lewis only because of his occasional careless use of language. In particular, once we distinguish between a sentence's expressing an a priori or analytic proposition 182 183 and a.proposition's being analytic, it turns out that Lewis is not inconsistent and holds only to the absolutist view. It will be argued here that there is no totally convincing way of freeing Lewis from this inconsistency in MWO, but that the functionalist interpretation (1) is more in accord with the general flavor of Lewis' account of the a priori in MWO .and.(2) recommends itself on independent grounds as a plaus- ible account of the role of the a priori in natural science. In MWO Lewis develops his pragmatic conception of the a priori in.opposition to both rationalism and empiricism. The inconsistency I wish to point out is more clearly evident in his arguments against a kind of extreme empiricism which .denies.that there is any a priori knowledge at all. Lewis' arguments against this extreme view will be explained and particularly his argument that if any knowledge is to be pos- .sible,-some knowledge must be a priori. On the basis of this an attempt will be.made to indicate what Lewis meant by "a -priori" and how it can be persuasively argued that Lewis does .hold (at least at.some places in MWO) a functionalist view of the a priori-a posteriori distinction.43 First of all we must briefly review some of the criti- cisms Lewis makes of traditional conceptions of the a priori. This is important in understanding Lewis' theory, since it is by means of a dialectical interchange with these traditional conceptions that Lewis proposes criteria that any adequate theory of the a priori must satisfy. Some of Lewis' 184 criticisms.have already been treated in the above discussion of alternative.logics. The others bear mentioning, however, particularly as they relate to the role of the a priori in natural science. To this point we have seen that knowledge of objects involves the conceptual interpretation of the given. The epistemological problem of the a priori is the problem of how it can be known that conceptual order, which is of the mind, can be imposed upon an independently given content of .experience. There is no knowledge.without interpretation. If inter- pretation,.which represents an activity of the mind, is . always.subject to the check of future experience, how is knowledge possible at all? That.the interpretation re- flects the-character of past experience, will not save its validity- For what experience establishes, it may destroy; its.evidence is never complete. An argument from past.to future.at best is probable only. . . . For the validity of knowledge,.it is requisite that ex- perience in general shall be in some sense orderly--that the order implicit in conception may be imposed upon it. O O 0 Thus, if there is to be any knowledge at all, some knowledge must be-a.priori, there.must be some proposi- tions the truth.of which is necessary and is independent of the particular character of future experience.44 .A theory of the a priori which meets the apparently incompatible conditions of rendering the a priori necessarily true of reality and yet not refutable by the particular character of the content of experience is the kind of theory that Lewis wishes to defend. It is an a priori which does not compel the mind's acceptance, since it represents the activity of the mind itself. The necessity of the a priori is due to its nature as legislative act. "It represents a 185 constraint imposed by mind, not a constraint imposed upon mind by something else."45 The conception of the a priori which can fulfill the apparently inconsistent conditions of being independent of experience and yet applicable to it is a knowledge of the relations among our concepts used in classifying objects. It is at once a knowledge of the real (as specifying the criteria of reality within any particular category) as well as a knowledge which is independent of . given experience, since it merely explicates what is already i implicit in.our classifications. Thus, this particular con- L ception of the a priori takes account of two perennial prob- lems.associated with the a priori--the nature of necessary truth and the.part played by the mind itself in knowledge and.the.experience of objects. There are, according to Lewis, three general concep- tions of the a priori in.the rationalist tradition. (1) According to one conception a priori knowledge is distin- guished from empirical knowledge by some psychological cri- terion such as being seen by the natural light of reason or known by intellectual intuition. This conception of the a priori is usually associated with the rationalism of the early modern period. (2) According to a second conception, a priori knowledge is distinguished from empirical knowledge in that it is presupposed by science in general or even by the possibility of experience of objects. (3) On the third conception, which Lewis identifies with Kant's a priori 186 knowledge is possible only if the given element in experience is already limited by certain a priori forms of intuition. None of these conceptions of the a priori are acceptable, according to Lewis.46 Lewis in fact says very little about the first con- ception. He seems content to point out that it is too closely dependent on the notion of innate ideas, which Lewis took to be a dead issue. In any case, to say that we know necessary truths because we intuit them as being necessary constitutes no theory or explanation of such knowledge. To say that we intuit necessary truths is no more than to repeat the claim that we know them, which was the fact to be explained. That necessary truth-is what the mind finds compelling does not adistinguish necessary truths from contingent ones. And, moreover, as we pointed out earlier in our discussion of alternative logics, for Lewis there is a sense in which neces- sary truth does not compel acceptance, but is precisely that element in knowledge which always has alternative ex- plications. Conceptions of the second variety, which claim that what is.a priori can be seen as such by the fact that it is .,presupposed by knowledge in general or that it cannot be denied without inviting contradiction, are the result of mis- .taken notions about the nature of deductive systems. The notion of presupposition which underlies this conception of the a.priori suffers from two major defects. (1) It is a 187 confused amalgam of logically necessary and sufficient con- ditions. And (2) it conflates logical and psychological issues in mistaking a psychological inability to imagine alternatives for logical inevitability. The issues involved here have already been treated in our discussion of alterna- tive logics. It suffices to say that if the method of pre- supposition is unable to show the necessity and indispensa- bility of logical truths, then it would constitute a poor instrument for establishing necessary truths not included within the province of logic. While questions have been raised above against Lewis on this point, we will grant him his case for the purpose of argument. But if the a priori is thus conceived of as made by mind in its activity of classifying and Consequently is inde- pendent of the given content of experience, how can it be known that it applies to experience a priori? Kant's solu- tion, the third major conception of the a priori discussed and rejected by Lewis, is that we can be assured that an a priori, which is the mind's own creation, can be applied to the given content of experience only because the given is al- ready limited by certain a priori forms of intuition, space and time. These a priori forms represent "passive modes of the capacity of having representations." To be an object of experience is to be limited by these a priori forms of intui- tion. One consequence of this, of course, is that 188 is limited to possible experience and experience is limited by space.and time as a priori.forms of our capacity of re- ceiving sense impressions. Thus, the possibility of applying the a priori to experience is assured only by claiming that knowledge is not of the independently real, but only of phenomena (i.e. things as they appear to us already condi- tioned by the a priori forms of intuition.)47 According to Lewis this manner of meeting Hume's problem is not satisfactory. Since Lewis' treatment of Kant here-is extremely brief, his main objections to Kantian attempts to secure the possibility of the applicability of the a priori to experience may be indicated rather briefly. First of all, if in fact it were true that any object of our sense experience is already limited by conditions imposed upon them by our own manner of receiving impressions, we would have no way of determining whether these were limita- tions that pertained to us or to the independently real object itself. In the absence of any clear way of indicating that these limitations are imposed by mind, they would be taken as limitations on what has been given, not as part of the.machinery of mind, and its continuance in future experi- ence would have only the assurance of an empirical generali- zation.. Thus, Kant's proposed solution does not meet the skeptich challenge. 189 Now Lewis is a Kantian in some important respects. In particular, following Kant, Lewis emphasized the active function of mind in scientific knowledge and experience, as well as the prescriptive function of categories and a priori principles. But more importantly, and this marks a major differ- l ence between Kant and Lewis, no such limitation on our manner of receiving sense impressions is required to insure the applicability in general of the a priori to experience. That is, "mind may limit reality (in the only sense in which the e validity of the categories requires) without thereby limiting - experience."48 To see how this is possible it is important to be aware of a subtle ambiguity in Kant's notion of experience.49 That is, Kant tends to think of experience and the phenome- nally real as if they were the same. However, it is reality, not experience, that must be conceptually orderable. But we can be assured of that, in part, because of the system- atic ambuiguity of the notion of reality. Experience is not categorized as of the real or of the unreal, for then every- thing would be real and nothing unreal. To be real is to be a particular kind of real, to be real in some particular category, a real physical object, a real illusion, etc. Our categories are intended to cover all of experience only if the categories of illusion, dream and hallucination are in- cluded. An a priori principle of interpretation (i.e. the 190 criteria of reality or principles of classification within some particular category) is not required to bring all ex- perience within its compass. It is a priori applicable to reality precisely because it determines what it is to be real in any particular category. Thus, while a priori prin- ciples of interpretation do and are required to limit reality, they are not required to limit the content of experience.50 As criteria of Classification, a priori principles of inter- pretation are not refutable by any particular experience. That is a priori which we can maintain in the face of experience no matter what. In the case of an empirical law, a mere generalizatiCn from experience, if the par- ticular experience does not fit it, so much the worse for the 'lawL. But in the case of the categorial prin- Ciple, if the experience does not fit it, so much the worse for the experience.51 To the question how we can know in advance of any particular experience that if it does not conform to our categorial principle it will not be veridical (i.e. will not be expei- ence of the particular kind of reality determined by that category) the answer is that we can know this precisely be- cause-the-"categorial principle states the criteria of 52 reality of that categorial type." While it is not a priori determinable-that a particular experience is experience of some particular categorial type, one may still be inclined to raise the skeptical question of how we can know that some experience may not remain forever unintelligible and resist all.attempts at understanding. Before this skeptical question is taken too seriously, however, it should be pointed out that 191 understanding is a matter of degree and is always somewhat partial. Some particular experience may in fact remain for- ever unintelligible, but no particular experience must remain unintelligible. Even to classify something as unintelligible is to have obtained understanding of it, however partial. It is at least to understand that it does not exhibit certain types of categorial ordering. Moreover, in the matter of rendering experience understandable, we play, as Lewis puts it, a game of animal, vegetable and mineral with experience. We may thus be certain of at least the theoretical possibility of rendering experience intelligible in part because the claim that all experience can be understood (i.e. properly inter- preted) takes on the character of a regulative ideal of reason or a maxim for the conduct of inquiry.53 A priori knowledge (i.e. knowledge of the relations between criteria of classification) is thus independent of experience and is known prior to any particular application to experience because it simply expresses our own intentions in classifying. It arises simply from the analysis of con- cepts, freely adopted, which we employ in distinguishing, classifying and naming objects. Since the a priori is the explication of the concept as criteria of classification and application, "the a priori is not a material truth, delimit- ing the content of experience as such, but is definitive or analytic in its nature."54 192 This conception of the a priori satisfies those con- ditions specified by Lewis that any adequate theory of the a priori must meet. "Since it is truth about our own inter- pretative attitude, it imposes no limitation upon the future possibilities of experience; that is a priori which we can maintain in the face of experience, come what will."55 Moreover, we can recognize that the a priori is the product of the mind's own activity since it is simply the critical formulation of our own criteria of Classification. Since it is our own creation, it always has conceivable alterna- tives. That is, we can always conceive of different ways of categorizing, in the sense of classifying, our experience. Thus, Lewis' concept of the a priori meets the apparently inconsistent requirements of being known in advance of any particular experience to hold of all experience and yet al- 56 Thus, the alterna- ways having conceivable alternatives. tive to one mode of classifying is not its falsity, but simply its replacement by a different classificatory scheme. Moreover, since what is a priori can change (i.e. can be replaced by some alternative) it is not surprising that fundamental principles and categories should have Changed through history. This gives added support to Lewis' theory of the a priori, since this fact of historical alter- ability would be impossible on traditional accounts of the a priori. While the fundamental categories of common sense, like thing and property, law, cause, etc. and the categories 193 of logic (truth and falsehood, necessity and possibility) may remain relatively constant, the fundamental principles and categories of natural science do change throughout his- tory. Moreover, these fundamental principles cannot them— selves be discovered by generalization from experience, but must be provided by the scientist himself in advance of any particular experimental inquiry.57 Moreover, even that the fundamental principles of logic are of necessity permanent is not, according to Lewis, born out by twentieth century logical investigations. (Doubts were raised about this claim above.) Even in the case of such fundamental concepts as "thing" and "property", there is evidence of historical change, however infrequent such Change may be when compared to conceptual changes in science. Thus, the concept of matter is different as to its ancient Greek and contemporary inter- pretations. Similarly, the present distinction between mind and body, influenced as it is by Christianity and Cartesian dualism, is not identical with the distinction as found in the Greeks. While the names of such categories may be perm- anent, the concepts they express, their manner of classifi- cation, undergoes constant Change in the light of Changing 58 Alteration of our purposes and accumulated experience. categorial framework, which amounts to its rejection and our adoption of a new one, is never dictated by empirical find— ings alone. In the choice of categorial frameworks, 194 empirical findings, simplicity and conformity to fundamental human purposes all play their part, and frequently the most important role is played by the latter. We have already had enough to say conCerning the adequacy of this pragmatic theory of the a priori as an account of the necessity of principles of logic. I should F now like to turn to an evaluation of it as an account of the fundamental principles of natural science. It is an impor— L tant feature of Lewis' theory, as it was of Kant's, that QQ’HI‘I- 5—?— there is in natural science an a priori element. I“. Lewis expressed his belief in the fundamental inter- dependence of natural law, categorial frameworks and the partial determination of both by mind as follows: The a priori element in natural science goes much deeper than might be supposed. All order of sufficient impor- tance to be worthy of the name of law depends eventually upon some ordering by mind. Without initial principles by which we guide our attack upon the welter of experi- ence, it would remain forever chaotic and refractory. In every science there are fundamental laws which are a priori because they formulate just such definitive con— cepts or categorial tests by which alone investigation becomes possible.59 The crucial question for consideration here is how to interpret this passage in such a way as to render what Lewis says here about the a priori nature of "fundamental" natural laws consistent with the analytic (non-empirical) nature of a priori truth. There is some confusion here which I hope to indicate in what follows. 195 An initial difficulty with Lewis' view of the role of the a priori in natural science is brought about by his Claim that "All order of sufficient importance to be worthy of the name of law depends eventually upon some ordering by mind." Moreover, since such laws depend upon mind, this would show on Lewis' theory that they are a priori. It is difficult to render this consistent with Claims made earlier about the differences between empirical law and the a priori principles of categorial interpretation. In particular, it was claimed earlier by Lewis that the crucial differences between empirical law and categorial principle is that the former, but not the latter, is refutable by experience.60 Thus, we are led to believe that there is a difference be- tween a priori categorial principle and empirical law. The natural move then is to conclude that the kinds of funda- mental 1aw which Lewis has in mind in the above quoted pas- sage are not, strictly speaking, empirical laws at all. They are not empirical laws which result from empirical generalization (i.e. they are not, in Duhem's sense, experi- mental laws), but are in some sense the framework within which empirical laws are to be understood and which underlie our ways of classifying objects. Such laws might be thought of as categorial, and hence for Lewis, a priori. To obtain some idea of what these might be, it will be helpful to look at one example of these fundamental laws provided by Lewis. 196 Lewis will want to argue that these are both definitions and principles of procedure on the one hand and natural law on the other. The example of such a priori principles underlying the discovery of natural uniformities is taken from a popu- lar treatment of relativity theory by Einstein. The relevant 61 In this ex- passages are quoted by Lewis at some length. ample we are asked to consider criteria for determining the simultaneity of events at a distance. More particularly, how can we tell that lightning strikes a railroad track at two places, A and B, at the same time? The following sug- gestion provides a way of deciding this particular matter. Place an observer mid-way between A and B, supplying him with a complicated system of mirrors in such a way that the ob- server can literally observe the two flashes at the same time. If he in fact does observe them at the same time, then the two events occurring at A and B occurred simultaneously. A critic claims that as an answer to the question of whether these events occurred simultaneously, the suggestion is unsatisfactory because Circular. It assumes that it took light the same time to travel from point A to the observer as from point B to the observer. But we could know ttta only if we already had at our disposal the means of measuring time, and in particular, the means of determining simultar neity of events. The suggestion turns out to be mere sup- position or an hypothesis which ought not to be simply assumed. 197 The reply to this hypothetical critic is that the suggested means of determining simultaneity is neither a supposition nor an empirical hypothesis, but a stipulation, made in order to arrive at a definition of simultaneity. Without this a priori stipulation of definitive criteria, the questions with which natural science is concerned could not even be asked, much less answered. Such concepts are not verbal definitions nor classifica- tions merely; they are themselves laws which prescribe a certain behavior to whatever is thus named. Such definitive laws are a priori; only so can we enter upon the investigation by which further laws are sought.6 We are being asked to accept the following descrip- tion of the procedure of natural science. The job of the natural scientist is to discover empirical uniformities and to provide explanations of these uniformities in terms of theory. However, we can meaningfully test such experimental and theoretical uniformities only if we have definitive cri- teria for applying the terms that occur in the statement of empirical and theoretical uniformities. Thus, I cannot answer the question "Are all swans white?" unless I already have determined what criteria an object must meet in order to be a swan. Since such definitive criteria are what makes empirical investigation possible, they cannot themselves be determined by empirical investigation. They cannot be the result of generalization from experience. Such generaliza- tion is generalization from veridical experience. However, whether an experience is a veridical experience of a certain 198 kind of thing presupposes the determination of what counts as an object of that kind. But if such definitive criteria are not empirical then they must be a priori. They are the criteria of reality (of various types) which are the scien- tist's own free creation and which he must bring to his experimental investigation if that inveStigation is to answer any questions at all. Such a priori criteria of reality are thus presupposed by empirical generalizations, and conse- quently cannot be discovered by such generalization. However much the give and take between the purposes of science and discovered fact may contribute to alter the procedure by which those aims are sought, and may induce new principles and categories, still the naming, Clas- sifying, defining activity is at each step prior to the investigation. We cannot even interrogate experience without a network of categories and definitive concepts. Until our meanings are definite and our classifications fixed, experience cannot conceivably determine anything.63 We have here a most perplexing situation. Natural laws are "derived" from experience. But such experience must be veridical, i.e. must be experience of reality of some particular type. But what experience is veridical is itself determined by the criterion of law. Which comes first, then, the law or the veridical experience?64 The answer is that the law is first so long as and as far as we are prepared to maintain it as criterion of the real. As is consonant with Lewis' theory of the a priori, such a priori knowledge underlying empirical observation, admits of alternatives. Thus, historically, such basic prin- ciples have undergone substantial alteration. However, as is 4 .1-_.- ‘Siif‘ .-' 199 the case with any system of classification and the analytic principles it embodies, such alteration is not due to one's now believing or having determined that one such system or set of principles is false and another true. Both sets of principles are eternally true. We simply give up categor— izing experience in terms of one set of principles in favor of another. There is no question of truth or falsity here at all. The important issues at stake are of the type that Lewis calls pragmatic. Such change in fundamental principle that does occur may be due to strictly empirical discoveries, but such discoveries do not prove the rejected categorial framework and its a priori principles false. Concepts and principles of interpretation are subject to historical alteration and in terms of them there may be 'new truth'. But the situation in which this happens needs analysis. It does not mean the possibility of new truth in any sense in which new truth can genuinely con- tradict old truth. . . . New ranges of experience such as those due to the invention of the telescope and micro- scope have actually led to alteration of our categories in historic time. . . . But when this happens the truth remains unaltered and new truth and old truth do not contradict. Categories and concepts do not literally change; they are simply given up and replaced by new ones.55 Moreover, not only is this so, but the empirical discoveries which precipitate change of categorial framework do not compel any such alteration, except pragmatically. In the case of the Copernican revolution, for example, it was the invention of the telescope and the increasing accuracy of observation which mainly provided the impetus to reinterpretation. But these new data were decisive only in the pragmatic sense. Those who argued the issue supposed that they were discussing a question of empiri- cal fact.57 v. -emflirfl‘ )‘AW‘E‘ZR. 'J 200 In other words, the same data could be accommodated in either the framework of Copernicus or that of Ptolemy, the differences have to do with the complexity and difficulty of dealing with the relevant phenomena in the latter as com- pared to the former. The frameworks are competing, but not in the sense of being inconsistent with One another. The issue is not an empirical one, but a conceptual one of how to most simply represent the relevant facts. "But they would no more contradict each other than a measurement in pounds and feet contradicts one in grams and inches."68 Furthermore, that empirical findings do not dictate rejection of categorial principle is further evidenced by the logical point that the question of how long to maintain a categorial principle or of under what conditions to discard it admits of no generally applicable answer. A stubborn conservatism can be proved unreasonable only on the pragmatic ground that another method of categor- ial analysis more successfully reduces all experience of the type in question to order and law.69 The type of principle which Lewis claims to have discovered in natural science is not unique to it, but only more conspicuous there. Any employment of concepts and associated principles of classification, whether at the level of science or common sense, has embedded within it these a priori principles which are at once principles of Classifi- 70 cation and of natural law. The philosopher's job is the historical and critical one of perspicuously displaying 201 the categories and a priori principles present in our ordinary and scientific dealings with experience.71 To this point what Lewis has pointed out seems quite compatible with a clear distinction between the a priori and the a posteriori; what we have called the absolutist view of the a priori--a posteriori distinction. He has tried to re argue that natural science, and empirical knowledge generally, L would not be possible on radically empiricist principles. g Unless some knowledge is a priori, unless some order of the 5 nature of law is brought to experience by the mind, no know- L; ledge would be possible.~ However, if we look more closely at how it is determined what a priori principles are adopted at any particular time, something like the functionalist view emerges. What Lewis calls the a priori principles underlying scientific investigation turn out to be empirical proposi- tions which we happen to place great confidence in. The resulting distinction between a priori and a posteriori is both a matter of degree and context dependent. The absolutist position concerning the a priori--a posteriori distinction does not seem consistent with a number of passages in MWO. Thus, in attempting to answer the ques- tion of whether discovered law or the veridical experience from which it is derived is prior, Lewis claims that "the law is first so long as and so far as we are prepared to 72 maintain it as criterion of the real." This suggests that 202 an a priori principle is an empirical proposition functioning as a criterion of reality; a proposition which in a certain context of inquiry we are not prepared to reject. Rather than reject it, we elect to Change our categorial framework elsewhere. The following passage suggests the same interpreta- _. tion. . . . what is a priori criterion of reality in one con- nection may be merely empirical law in some other--for example, the law correlating photograph and thing photo- graphed, or the law of the behavior of solid bodies in translation which condemns the mouse that disappears without a hole, or the laws of perspective which exclude e; a landscape which recedes as we approach it. The suggestion here is that certain principles may be subject to empirical disconfirmation and confirmation in one situation, but not subject to such test in another. Thus, in the context of experiments in optics, laws correla- ting photograph and thing photographed may be up for empiri- cal test, but in some other context (as, for example, in determining whether spirits are physically real), the same law may function as a priori criterion of reality, condemning spirits as unreal because they cannot be photographed. In this respect, the a priori—-a posteriori distinction which Lewis has in mind is context dependent. It is not the nature of the proposition itself that determines whether it is a priori or a posteriori, but the attitude one takes toward the proposition. The a priori is what one tataa as cri- terion of reality. Thus, "All swans are white" is a priori 203 if having found that the term "white" does not apply to a presented object, we refuse to apply the term "swan". How— ever, this criterion does not distinguish strictly analytic propositions (i.e. propositions that are substitution in- stances of logical truths when non-logical terms are replaced with their definitions) from empirical propositions in which T1 we have such great confidence that we are prepared to reject as non-veridical any experience which fails to conform to it. Thus, "if definition is unsuccessful, as early scientific 3 . definitions have been, it is because the classification thus set up corresponds with no natural cleavage and does not correlate with sufficiently important uniformities of be- havior."74 However, on this interpretation, to reject an a priori criterion of reality is to reject it as false. Thus, if we should come upon birds, otherwise like swans except in coloration, we would reject "All swans are white" as false. It is not as if we have changed our habits of classification so much as we have found that color is not a reliable in- dicator of this particular species of bird. Moreover, since the degree of confidence we have in a generalization is notoriously variable, the a priori-—a posteriori distinction will not only be context dependent, but a matter of degree as well. Moreover, since the degree of confidence that we have in an empirical proposition may increase over time, what may have begun as an hypothesis subject to experiential 204 disconfirmation may, at a later point, he used as a criterion of reality, and thus become analytic. Such propositions function aa_tt they were analytic. Anything which fails to conform to them is condemned as non-veridical. Thus, the same proposition may at one time be a priori and at another be a posteriori (or analytic and synthetic, since for Lewis r these are coextensive). I A rather simple example of a proposition first being synthetic, then becoming analytic (in the functionalist sense) can be found in Chemistry. A chemical element, like gold, may be understood to have certain perceivable qualities like color or texture. Moreover, it is subsequently observed that gold thus Characterized has a certain atomic weight and atomic number. The correlation between something‘s being gold and its having a certain atomic number may become so well confirmed and so embedded in chemical theory that having that atomic number becomes a criterion for something's being gold. What was formerly a synthetic proposition, refutable by experience, has now becomes an analytic proposition in Lewis' sense. Anything which fails to have that atomic weight is not real gold.75 So on the functionalist interpretation of Lewis, the analytic-synthetic, a priori--a posteriori distinction be- comes a distinction which changes over time, is a matter of degree and is dependent on the context of inquiry. To say that a proposition is a priori or analytic is to give it a 205 certain privileged position. To this Lewis would heartily agree. But the passages noted above indicate that to say that a proposition is analytic is to say that relative to certain other propositions at some particular time and in some particular context, it functions as criterion of reality. However, Lewis' claim that if any knowledge is to be possible, some knowledge (in addition to the purely logical) must be a priori must now be reinterpreted. It may now be understood as claiming that unless some propositions are immune to refutation at certain times and in certain contexts, t no knowledge would be possible, since inquiry would be im- possible. This reflects the fact that, in empirical inquiry, each and every claim is subject to challenge, but in the nature of the case, they cannot all be challenged at once. In the functionalist context, the principles of logic would be unique in that they are subject to revision in no contexts and at no times, and may be said to possess the highest degree of analyticity. The resulting explication of analyt- icity shows some marked similarities to Dewey's attempt to explain the a priori and the a posteriori in terms of the functions propositions have in the context of inquiry. Thus it marks in the context of Lewis' own epistemology a move away from a formalist interpretation of the analytic-synthetic distinction toward a more pragmatic, contextualist one.76 206 Now it is true that this functionalist interpreta- tion of the analytic-synthetic distinction is not compatible with a good many passages in MWO (e.g. see passages quoted on p. 200 above). But then the same may be said of the absolutist interpretation as well. In concluding this chap- ter I should like to sketch a defense against this charge of inconsistency.77 ‘ To claim that Lewis is committed to a functionalist View of the a priori is to assume that because Lewis Claims that analytic propositions and a priori principles are con- IY~ ditioned by experience in the sense that strictly empirical discoveries may dictate a change in categorial framework, he must give up the absolute distinction between the a priori and the a posteriori, the analytic and the synthetic. But this is to fail to be even reasonably sympathetic in our interpretation; it is to fail to note what is and what is not pragmatically determined on Lewis' account. In AKV Lewis makes a crucial distinction between sense meaning as criterion in mind and linguistic meaning in terms of which the inconsistency noted above can be seen as apparent only. One may consider such criteria of application, as mean- ings entertained in advance, in terms of incipient be- havior or behavior attitudes if one chooses. . . . The real point is the distinction between meaning as indica- tive of and indicated by behavior on the one hand, and on the other, meaning as relation not to objects and objective occasions but to other linguistic expressions.78 207 When we have chosen to maintain a proposition in the face of all experience, then the signification of the predi- cate term (i.e. the properties necessary for the term's ap— plication) are implicit in the criterion for applying the subject term. Those properties are essential for the appli- cation of a term because we have chosen them as such. If P" we refuse to call anything a swan which is not white, or a ‘ substance gold which fails to have a certain atomic number, it is because we have made these properties part of the sense meaning, as criterion in mind, of the terms "swan" and ”gold". L That the term "gold" did not previously have that as part of its sense meaning does not show that the proposition that gold has an atomic number which was once synthetic has now become analytic. What it shows is that the linguistic ex- pression "gold" is now being used to express a different concept (since its sense meaning has changed). We have here separate and distinct meanings attached to the same linguis- tic expression. Thus when Lewis says that what is empirical 1aw in one context may be a priori principle in another he means to say that the same sentence may have once been used to express a synthetic proposition, which on another occasion or at a later time may be used to express an analytic propo- sition. Therefore, we do not have the same proposition being both analytic (in one context or at one time) and synthetic (in another context or at another time). The relation be- tween these two propositions is not a logical one, but an 208 historical or psychological one. Having made this distinc- tion between sense meaning and linguistic meaning, the pas- sages in MWO which suggested a functionalist interpretation of the analytic-synthetic distinction can be interpreted along the lines of the absolutist point of view. Moreover, it is just this failure to distinguish linguistic meaning and sense e meaning that, on Lewis' view, is responsible for the persis- tent belief in synthetic a priori propositions. In particu— lar, the position maintained by Kant that there are synthetic a priori propositions or principles presupposed by science T3"mvb v.37-‘-..‘D 'M""II}A? I. '4 is due to a failure to recognize this distinction between linguistic meaning and sense meaning.79 What we appear to have, then, is Lewis emphasizing two prima facie incompatible conceptions of the fundamental principles of natural science. On the one hand he emphasizes the analytic or definitive nature of them as criteria of reality. In this regard they are independent of experience and are eternally true. 'Such fundamental principles cannot and will not be rejected as false. Like systems of classi- fication, they are simply given up and replaced by new ones. On the other hand we find Lewis insisting that em- pirical laws may function as criteria of reality. If not strictly speaking refutable by experience, they are at least conditioned by it. Empirical laws and categorial principles seem to differ at best in degree and appear to usurp each other's function. 209 The key to the resolution of this apparent inconsis- tency is to recognize three things: (1) that empirical or inductive uniformities may, in certain contexts and under certain conditions, function as definitions, though defini- tions of an empirical sort, indirectly refutable by experi- ence; (2) that there are two different notions of analyticity F“ with which Lewis is working, formal analyticity (as in L logic) and functional analyticity (as in natural science); and (3) that for empirical scientific inquiry to be possible, -J analytic propositions of both the formal and functional h’ variety are necessary. If we can make use of these distinc- tions, then we can at once relieve the tension in Lewis' conception of the a priori (by recognizing its multi-faceted nature) and begin to see a contribution that Lewis has to make toward the epistemic analysis of natural law. To see the importance of the first two distinctions, consider the following way in which Lewis distinguishes the a priori and analytic from the a posteriori and synthetic. A proposition "All X are Y" is a priori if, finding that the predicate term Y does not apply to a presented object, we refuse to apply the subject term X. However, what Lewis seems to have overlooked is that it is not necessary for a proposition to be formally analytic to warrant such moves. All that is required is that we have sufficient confidence in the truth of the proposition. When a proposition func- tions in this way, we will speak of it as functionally 210 80 Thus, while formal analyticity is a sufficient analytic. condition of analytic functioning, it is not a necessary condition. Since functional analyticity is a matter of the confidence we place in a proposition, then functional analyt- icity would appear to admit of degrees. Again, from a functional point of view, propositions that are formally analytic (like logical truths) would represent the upper F limit of functional analyticity. From the functional point of view, logical truths, Lewis' categorial principles and empirical laws would be represented as different segments t; of a continuum. What needs to be shown now is that there are actual cases in which a proposition which may have begun as an empirical proposition, with which we were prepared to depart in the face of recalcitrant experience, may become function- ally analytic, but not thereby formally analytic. Such propositions would bear out what Lewis means by claiming that what is a priori in one context may be just empirical law in some other. We may begin by considering a simple example of a proposition beginning as an empirical generalization, but which, upon extensive confirmation, later functions analyti- cally. Definitions may be of at least two sorts: nominal and real. A nominal definition is simply a stipulation that .two linguistic ekpressions are to be used synonymously. In usuch'Cases the term defined usually has no meaning apart 211 from the stipulation. A real definition is a definition that is the result of empirical investigations and is conditioned by them.81 In such cases the term to be defined typically has a meaning independent of the definition. Such real def- initions are found when the empirical laws from which they are derived have received extensive confirmation. An example of such a transition from empirical gen- eralization to (real) definition is the historical develop- ment of the definition of whale in terms of mammalian Characteristics. Such a definition grew out of the empirical Hm]. fll’ F“ ‘n-lhflfl ‘eruy discovery that whales, contrary to first appearances, exhibit mammalian characteristics. The formation of such empirically grounded definitions changes what was previously an empirical generalization into a principle of classification, a "cri- terion of reality" in Lewis' sense. Thus what began as an empirical generalization now functions analytically. Such transitions to functional analyticity are not limited to classificatory sciences. Boyle's law, for example, after having received extensive confirmation, takes on the character of a criterion for what it is to be an ideal gas-- 82 Under such circum- it is a gas which obeys Boyle's law. stances, "all whales are mammals" or "ideal gases obey Boyle's law" function analytically and therefore cannot be refuted by experience in any straightforward sense. It is important, howeVer, to recognize that such propositions should be dis- tinguished from formally analytic propositions in being 212 empirically grounded. They can (contrary to Lewis) be rejected as false, though only by accumulated experience, not by isolated testing. Such protection from refutation by isolated experiment does not make them non-empirical or "eternally true". Should our adherence to any of them per— sistently fail to aid us in understanding some phenomena, P“ such principles of classification and criteria of reality 1 will be rejected as false and not just on pragmatic grounds. Another indication that laws may function analyti- ‘njr'smfr 9. - cally is the extreme reluctance with which we give up firmly entrenched laws. Noting, for example, that a planet is be— having abnormally, astronomers would be extremely reluctant to consider this a threat to the theory of gravitation. In fact, such observations would merely serve to motivate as- tronomers to search for the distrubing factors. That funda— mental 1aws in natural science are irrefutable by any par- ticular experience does not show them to be conventions merely. But by the same token, they are not empirical in a sense in which, for example, logical truths are not. Strictly speaking, to say they are a priori (and hence for Lewis analytic) would be misleading. Nevertheless, such laws may function analytically and thus become criteria of reality. In this way, Lewis has made several important points in his treatment of the role of the a priori in natural science. He has pointed out how natural law and systems of classifi— cation "grow up together". More importantly, he has pointed 213 out the absurdity in conceiving of natural laws as constant conjunctions or summaries of experience. They are in an important sense constitutive of experience since they serve to define reality within particular categories. Again, natural law is discovered by generalization from veridical experience. But veridical experience is itself determined FF by the criterion of law. Thus, unless law is presupposed, no inquiry into other laws is possible. As Lewis would put it, if any knowledge is to be possible, some knowledge must be a priori. ti But such law as is presupposed is not restricted simply to the formally analytic, but must also include those empirical laws which at any given time are functioning analytically. While any particular knowledge claim may be subject to revision, nevertheless if inquiry is to be pos- sible at all, they cannot all be subject to revision at any one time. Had Lewis paid closer attention to this funda- mentally Peircian notion of inquiry and taken note of the distinction implicit in his own work between the formally and the functionally analytic, then the confusion in his treatment of the role of the a priori in natural science could have been avoided. While the resulting conception is not consistent with some of Lewis (most notably when he speaks of the eternal truth of categorial principles), never- theless it is suggested if not actually implied by the gen- eral drift of his account of conceptual systems and conceptual 214 change. Indeed, the distinction between the formally anal- ytic and the functionally analytic is reminiscent of Lewis' own distinction between categorial systems considered with- out any reference to possible interpretation and those prin- ciples of interpretation through which they become prescrip- tive for reality. F9... 5. . S ConcluSion to Part II ; 5 After having raised objections to Lewis' account of Li -. the given element in experience in Part I, in Part II we L? have dealt critically with two fundamental aspects of his pragmatic conception of the a priori--as an account of logic and as an account of the fundamental principles of natural science. While our treatment of Lewis' conception as an account of logic was largely negative, we have found in Chapter V that there are present in Lewis' treatment of the a priori in natural science, the rudiments of a quite plaus- ible functionalist conception of the a priori. 215 Notes to Part I 1C. I. Lewis, Mind and the World_Order, Dover, Inc. N. Y. 1956; p. 38. Hereafter cited as MWO. 2mwo,p.39. 3MWO, p.39. 4MWO, pp. 140-142, An Analysis of Knowledge and r7 Valuation, La Salle, 111., Open Court, 1946, pp. 16-21. g Hereafter cited as AKV. L 5 MWO, p. 213. 6MWO, pp. 65-66. L 7MWO, pp. 48-49. by 8mm,p.so. 9mm,p.5m loMWO, p. 51. llMWO, p. 51. 12MWO, p. 52. 13MWO, p. 53. The claim that the given cannot be de- scribed is later rejected by Lewis in AKV, where he suggests that the expressive use of language is adequate for the de- scription of immediate experience. There are difficulties with this notion, however. It is not clear just how signifi— cant Lewis took this thesis to be for the analysis of know- ledge as well as for the practical importance of knowledge. See his paper "The Given Element in Empirical Knowledge" where he states that ". . . the near absence of any restric- ted vocabulary or syntax for expressive statements is an unimportant matter for empirical knowledge itself." Phil. gay., LXI (1952), p. 171. l4mwo, p. 54. 15MWO, p. 55. 16MWO, p. 66. l7MWO, p. 46. 216 18MWO, p. 46. 19The relations between epistemological, psychologi- cal and other accounts of knowledge is a much debated issue, but one we shall not get into here. 20That other qualities than the strictly sensory are given is a crucial assumption in Lewis' theory of value. On this point see Book III of AKV where Lewis defends natu- ralism and rejects subjectivism in the theory of value. Lewis' remarks, to be discussed later, concerning H the incommunicability of qualia seems to be at cross purposes 5“ with his attempted defense of naturalism in value theory. In discussing the difference between qualia and properties, Lewis argues that an individual's "perception" of qualia may be a quite different from that of others', but that so long as this L difference does not affect the relational context of the ~ qualels occurrence in experience, the purpose of knowledge will not be thwarted. That is, for purposes of knowledge 9 such differences in immediate experience will be relatively Li unimportant. However, the situation with regard to value qualia seems quite different. Moreover, in the case of value qualia, the immediate experience of them is at least as im- portant as the relational context of their occurrence in ex- perience. This does seem to indicate a difference between factual knowledge and value judgments to which Lewis fails to give sufficient attention. If the role of value qualia is thus different from other sense qualia, then that may indicate an important distinction between empirical knowledge and value judgment, and may.also.indicate a difficulty in combining, in the manner of Lewis, empiricism in epistemology and naturalism in value theory. As should be clear from my critical remarks on Lewis' notion of qualia, the most profitable way in which Lewis could avoid this inconsistency between his brand of em- piricism and his naturalistic analysis of value would be, not to accept a.radical distinction between factual and value judgments, but to reject the notion of incommunicable sense qualia which precipitates the inconsistency. Of course this is not all there is to defending naturalism in value theory. I suggest this merely as one way in which Lewis can avoid an inherent tension between his epistemology and his theory of value. 21MWO, p. 57. ZZMWO, p. 58. 23mwo, p. 58. 24mo, p. 59. 25mo, p. 59. 217 26For more on this, particularly with respect to representational art, see E. H. Gombrich, Art and Illusion, Pantheon Books, 1956. 27 Journal of Psychology, (1948), 25, pp. 205-222. 28For a discussion of similar cases from ordinary experience and psychological experiment, see I. Scheffler, Science and Subjegtivity, New York, Bobbs-Merrill Co., Inc., 1967, Chapter 2, "Observation and Objectivity". 29 MWO, p. 118. I...” s" I“ ‘ F 3°Mwo, p. 120. 31 l T. ‘ Jm‘ MWO, p. 121. 32Mwo, pp. 119-120. 33MWO, p. 122. ‘E 34MW0, pp. 121-122. 35MWO, pp. 122-123. By optimum conditions I take Lewis to mean those conditions under which it is most likely that one will make a correct prediction. This passage along with many others in MWO is inconsistent with Lewis' persis- tent claim that immediate experience cannot be described. To assign the name of a property to an appearance of it when one realizes that it is the appearance and not the property to which the name.is being applied is to assign a name to an appearance and hence to describe it. The continual difficulty that this position gets Lewis into may be one reason that he gave up this view in AKV, Claiming there that language can be put to what he calls expressive use. 36mwo, pp. 123-24. 37MWO, p. 124. 38MWO, p. 46. 39MWO, p. 57. 40For a brief, but illuminating, discussion of some of these difficulties, see Leonard Linsky, "The Incommunica- bility of Content", Journal of Philosophy, vol. 59, pp. 21-23. 41MWO, pp. 124-125. Italics added. 42MW0. pp. 74-75. 218 43MWO, p. 74. 44MWO, pp. 75-76. 4Smwo, p. 124. 46MWO, pp. 124-125. Italics added. 47It is not altogether clear why one should suppose that locating a color quale in the color spectrum constitutes a definition of the color term. This does not affect the arguments in question here, however. 48See the two excellent short papers by Firth and Brandt on these matters in the Commemorative Sumposium on C. I. Lewis published in Journal of Philosophy, 1964, pp. 545-559. See also the paper Lewis read in a symposium with Hans Reichenbach and Nelson Goodman, published in Philosoph- Fr ical Review, LXI, (1952) on the given. Lewis' contribution “L to that symposium is reprinted in The Collectad Papers of F5 C. I. Lewis, ed. by Goheen and Mothershead, Stanford, Stan- ford University Press, 1970. pp. 324-330. Hereafter cited as CP. 49See AKV pp. 182—183 and "The Given Element in Empirical Knowledge" in CP, p. 326. SOMWO, pp. 274-275. 51MWO, p. 275 and AKV, p. 30. With respect to valu- ation, which Lewis took to be a form of empirical knowledge, "direct findings of value-quality in what is presented" are not knowledge; or at least such findings are not "judgments", the assumption being that any kind of apprehension which is not judgment is not knowledge. See AKV, p. 365. It is, I think,.one of the attractive features of Lewis' pholosophy that the various different areas of philosophy (in this case epistemology and the theory of value) are so tightly woven together. Thus, just as in the case of ordinary empirical knowledge, one must distinguish between the immediate aware- ness of the given on the one hand and the knowledge of pre- sented objects on the other, so one must do in those forms of empirical.knowledge which are valuations. This systema- ticity has its problems, however. In particular, if the distinction is challenged, as it is here, then some revision might be called for in Lewis' theory of value. For example, Lewis Claims that one reason for the plausibility of non- cognitivist value theories resulted from the failure to distinguish these different phases of knowledge. Valuations, for Lewis, are forms of empirical knowledge, are true or false, and are subject to the same kinds of tests as other 219 forms of empirical knowledge. This cognitive character of valuations,.according to Lewis, "has often been obscured by failure to distinguish mere apprehensions of good or ill in experience from predictions of the possible realizations of these qualities in particular empirical contexts, from ap- praisals of the objective value-quality resident in existent things." (AKV, p. 365) Moreover, eliminating the distinc— tion between immediate awareness of the given and other forms of empirical apprehension would take from Lewis one of his more forceful arguments against non-cognitivism in value theory. What might.result in his ethics is another matter, since Lewis made a point of distinguishing valuations from ethical judgments. (On this distinction in Lewis, see a William Frankena, "C. I. Lewis on the Ground and Nature of the Right", Journal of Philosophy LXI, (1964), 489-496; "Lewis' Imperatives of Right", Philospphical Studies, 14, (1963), 25-28; "Review of The Ground and Nature of the ; Right", TheIPhilospphical Review, LXVI (1957), 3984402. See i also Roger Saydah, The Ethical Theory of Clarence Irving L Lewis, Athens, Ohio, Ohio University Press, 1969. $5 52 AKV, p. 30. 53mwo, p. 121. 54MWO, p. 121. 55Lewis' disclaimer to the contrary, he does seem to allow, if not simple ideas, at least simple qualities, namely-qualia. They have at least some of the qualities of the simples of the tradition, a fact which is revealed in Lewis' treatment of the incommunicability of content thesis criticized above. They are simple in the sense of having no parts. They cannot be named except in the sense of being merely labeled. For Lewis, though not for all in this tra- dition, qualia are what they are independently of their relations to-anything else, even their relations to other qualia. From the point of view of rationalistic idealism, this last claim would be.seen as claiming that all the relations-into which qualia may enter are external. Some of the.difficulties.of this view were noted above with special reference to color qualia. 56MWO, p. 60. 57Lewis does not take "fact", "state of affairs" and "proposition" to be equivalent, and I do not mean to imply here that they are. 58mwo, p. 123. 220 59MWO, p. 120. Of course, this is not all that Lewis had to say by way of characterizing the concepts of "objec- tivity" and "objective reality". For further remarks, see MWO, p. 138 ff. 60 t I o o The notion of consequences here is ambiguous and could mean either logical or causal consequences. 61de, p. 125. 62MWO, pp. 130-131. 63As noted above, I am not discussing what I have referred to as the logical argument for the certainty of our apprehension of the given. 64MW0, pp. 143-144, pp. 275-277, p. 292. 65 ‘M u! at! l “74'- rflm“ A us. AKV, pp. 182-183. 66CP, p. 326. For an astonishingly similar treatment of the given, see H. H. Price, Perception, London, Methuen, 1932, Chapter one. 67 AKV, pp. 25-26. 68AKV, p. 26. By presentation content I take Lewis to mean a quale or complex of qualia. 69AKV, p. 28. 7°AKv, p. 30. That Lewis still held to the soundness of the logical argument can be gleaned from AKV pp. 185-189, as well as from "The Given Element in Empirical Knowledge", 23, pp. 324-327. 71The name for the argument I have taken from Norman Malcolm's-article, "The Verification Argument", reprinted in his Knowledge and.Certainty, Englewood Cliffs, N. J., Pren- tice Hall, 1963, pp. 1-57. As Malcolm points out, it is not only Lewis who defended this argument. It was also defended by Carnap, Ayer and Popper. My debt to Malcolm's article is substantial. 72AKV, pp. 179-180. There is a complication here which will not be gone.into, but which I shall at least men- tion. It seems that.the "complete" and "decisive" nature of the.verification of terminating judgments is inconsistent with what is implied by other things in Lewis' account of empirical knowledge. I shall mention one of these. First of all, since, as seems obvious, this passage of experience 221 ’ will not be confined to a single specious present, the valid- ity of memory will be presupposed in testing the terminating judgment. But, according to Lewis, memory cannot guarantee the certainty of what is apparently remembered. See Lewis' discussion of the validity of memory in AKV Chapter XI, pp. 315-362. . 73AKv, pp. 180-181. For the difficulties involved in Lewis‘ notion of terminating and non-terminating judgments, especially those concerned with the sense in which the latter "imply" the former and are translatable into the former, see R. Chisholm,-"The Problem of Empiricism", Journal of Phil- oso h , Vol. 45 (1948) and Lewis' reply in the same volume, “Professor Chisholm and Empiricism". It is instructive to note that Lewis seems to have thought that if Chisholm's objections are decisive, our only means of justifying beliefs would be in terms of their coherence, a consequence which Lewis found devastating to Chisholm's case. "My own belief is that.if Professor Chisholm's point can be made good, then there would be nothing left for us but skepticism, because I am convinced that.any coherency—theory will have defects which are.fata1.“. The coherence theory of truth and the coherence theory of justification seem to have been the ene- mies from the start and the avoidance of any type of coherence theory was a prime motive for Lewis' account of the given. We shall.have more to say about this matter at the Close of this chapter. 74AKv, p. 180. '1 .Qm..m: lib. hi I ‘9.“: w ‘21.... 75mwo, pp. 279-281. 76mwo, p. 279. 77mwo, pp. 282-283. 78mwo, pp. 281-282. 79Whether Lewis adhered to the following proposition is not immediately obvious and will come in for discussion later.. The proposition in question is "x is certain if and only if x is completely and decisively verifiable." It’ appears, from the foregoing paragraphs that Lewis accepts this, or he at least.accepts the claim that complete and decisive verification.is a necessary condition of certainty, since.he.argues.that.non—terminating judgments are agt cer- tain because they are BEE completely and decisively verifi- able. This at least appears to contradict the claim that apprehension of a quale is unverifiable. If such apprehen- sion is unverifiable, then it cannot be completely and de- cisively verifiable, since it is not verifiable at all. But 222 if being completely and decisively verifiable is a neces- sary condition of certainty, then apprehension of a given quale cannot be certain. 80In..what follows I have relied so heavily on Mal- colm's exposition of the argument that detailed reference would be tedious. See his article, "The Verification Argu- ment" in Knowledge and Certainty, pp. 1-57. 81 On this see Chisholm, "The Problem of Empiricism". FA. 821 will ignore the further complication that these L conditionals which are predicted should, on Lewis' account, be couched in expressive language. I do not think that has a bearing on what I am about here. 83For a discussion of this question see Chisholm, "The Problem of Empiricism", Lewis' reply, "Professor Chis- holm and Empiricism", and Firth, "Radical Empiricism and Perceptual Relativity", Phil. Rev.,59 (1950), pp. 164-183 and pp. 319-331. 841 refer the reader to Malcolm's article for a dis- cussion of these issues. See also a reply to Malcolm by Harry G. Frankfurt, "Philosophical Certainty", Phil. Rev., LXXI (1962), pp. 303-327. , 85See the numerous discussions of this in a collec- tion of Lewis' ethical works, Values and Imperatives, ed. John Lange, Stanford, Stanford University Press, 1969. 86 .’ .—_EI. ETC—i. ‘— 1‘2'“ "- W . '2 .21” Frankfort, op. cit. p. 306. 87This kind of objection is made by Goodman in his contribution to a symposium with Lewis and Reichenbach in Phil. Rev., 1952 on the given. Goodman's contribution was entitled "Sense Certainty", Scheffler, following Goodman presents the same objection in his Science and Subjectivity, pp. 34-35. 88See R. Firth, "Lewis on the Given", in The Philos- opby of C. I. Lewis. ed. P. A. Schilpp, La Salle, IlIinois, Open Court, 1968, pp. 335-336. 89See Wilfrid Sellars, "Empiricism and the Philos- ophy of Mind", reprinted in his Science, Perception and Reality, London, Routledge and Kegan Paul, 1963, pp. 149- 150. 90For Lewis' discussion of memory, see AKV, pp. 315- 362. 223 918ee Mwo, p. 367-368. 92Onthe view that knowledge makes sense only if there is some correlative possibility of error, see MWO, p. 275 and AKV, p. 30. 93Recall that one of the needs for recognizing the e1ement.of.interpretation.in.experience is to allow for the possibility of error. See MWO, p. 39, p. 42. I will discuss later whether "impossibility or error" means "cannot fail to be true". 94mwo, pp. 130-131. 95AKV, p. 180. 96See Firth, "Lewis on the Given", in Lewis Schilpp ' L vilume, pp. 339-340. L 97 run -‘ TE : See Collected Papers, pp. 324-330. 98See Chisholm's article, "The Problem of Empiri- cism". 99See AKV, Chapter XI, pp. 315-362. 1OOAKv, p. 338. 101 For more on this notion of data as prima facie Claims, see N. Rescher, The Coherence Theory of Truth, Oxford, Clarendon Press,'1973, pp. 50-71. 224 Notes to Park II 1For a discussion of this line of objection to the— cries similar to Lewis' in certain respects, see the follow- ing: .W. Kneale, "Are Necessary Truths True by Convention?", Proceedipgs of_the Aristotelian Society, Suppl. vol. XXI, (1947), Brand Blanshard, The Nature of Thought, Vol. II New York, Humanities Press, 1939, pp. 385-398, 411-424; Brand.Blanshard, Reason and analysis, La Salle, Illinois, Ft Open Court, 1962, pp. 171-175. j 2 For a discussion of this line of approach see Arthur Pap, The A Priori in Physical Theory, New York, Russell and Russell, 1946 and Semantics and Necessary Truth, New Haven, Yale University Press, 1958, pp. 229ff, lle. 3For Lewis' own treatment of the connection between ti his work in logic and his work in epistemology: see the interesting autobiographical.essay, "Logic and Pragmatism", reprinted in_Collected Papers, pp. 3-19. 4 "Logic and Pragmatism:, p.4. .sThese two papers are reprinted in Collected Papgrs, pp. 351-359 and 371-382. 6See-"Logic and Pragmatism", pp. 4ff. 7Ibid., p. 6. 8Ibid., p. 8. At one point Lewis thought that the notion of strict implication would aid in the explication of truth conditions for contrary to fact conditionals. (cf. “Implication and the Algebra of Logic") However, he later.came to believe that there was an important class of contrary to fact conditionals whi h could not be represented either by material, formal or strict implication. See AKV, pp. 211-219. ‘ 9See "Logic and Pragmatism", p. 9. 10See Blanshard, The Nature of Thought, Vol. II, pp. 385-398, and W. T. Parry, "The LogiC of C. I. Lewis" in Lewis Schillp volume. 11On the distinction between intensional and exten- sional disjunction, see Lewis' "Implication and the Algebra of Logic" in Collected Papers, pp. 351-359. 225 leor a useful discussion of the arguments against strict implication as the correct analysis of the ordinary notion of implication, as well as a defense of Lewis' posi- tion, see Jonathan Bennett, "Entailment", Phil. Rev., 1969, pp. 198-237. 13 Blanshard, The Nature of Thought, Vol. II, p. 390. 14See A. Anderson and N. Belnap, "Tautological En- tailments", Philosophical Studies, 1962, pp. 9-24. 15For more on this analysis of the dispute over im- plication see Quine, Mathematical Logic, Cambridge, Harvard University Press, 1940, pp. 31—33 and H. S. Thayer, Meanin and Action, New York, Bobbs-Merrill Co., 1968, pp. 210-212. 16The philosophy of c. I. Lewis, ed. Schilpp, p. 659. 17Lewis is not without his contemporary defenders. See Bennett, op. cit. 18Reprinted in Collected Papars, pp. 400-419. 19mwo, pp. 197. 2ocp, pp. 371-372 and MWO pp. 202-205. 21cp, pp. 372-373 and see also MWO, pp. 200-202. 22cp, pp. 372-375. See also MWO, pp. 205-211. 23nwn, pp. 206-207. 24MWO, p. 209. 25See CP, pp. 401-402. 26CP, p. 402. 27CP, pp. 402-407. 28cp, pp. 407-409. 29cp, pp. 409-411. 3°Cp, pp. 411-413. 31See MWO, p. 209n. 32See CP, pp. 414-419. 226 33This section was greatly aided by the following: E. J. Nelson, "Deductive Systems and the Absoluteness of Logic", Mind, XLII, (1933), pp. 30-42; Wm. Kneale, "Are Necessary Truths True by Convention?", PASS, Vol. XXI, 1947; Brand Blanshard, Reason and Analypis pp. 271-283. 34 See Nelson, op. cit., p. 34. 35Onflthis, See N. Rescher, Many-Valued Logics, New York, MCGraw Hill Book Co., 1969, pp. l43ff. 36 I" MWO, p. 209n. 37See G. Massey, Understanding §ymbolic Logic, New York, Harper and Row, 1969, p. 207. 38See Kneale and Kneale, The Develppment of Logic, .4 Oxford, Clarendon Press, 1962, pp. 573-574. 3' . -J 39For a brief, but useful discussion of the discovery E of non-Euclidean geometries, see Howard DeLong, A Profile of Mathematical Logic, Reading, Mass., Addison-Wesley Publish- ing Co., 1970, pp. 38-39. 40See Wm. Kneale, "Are Necessary Truths True by Convention?" 41 MWO, p. 272. 42See CP, pp. 240-241. 43The question which forms the topic of this section was suggested by a paper by Arthur Child, "Toward a Functional Definition of the A Priori", Journal of Philosophy, 1944, pp. 155-160. See also A. Pap, The A Priori in Physical Theory, esp. Chap. 1. 44MWO, pp. 195-196. 45MWO, p. 197. 46 mwo, pp. 197-198. 47See MWO, pp. 212-219 for Lewis' discussion of Kant. 48MWO, p. 215. 49MWO, pp. 221—222. SOMWO, p. 222. 227 51MWO, p. 224. 52mwo, p. 225. 53See MWO, pp. 349, 353, 366. 54MWO, p. 231. 55MWO, p. 231. 56MWO, p. 232. 57MW0, p.233. 58$ee MWO, pp. 234ff. 59MWO, p. 254. 60See MWO, p. 225. 61See MWO, pp. 254-256. 62MWO, p. 256. 63Mwo, p. 259. 64mm, p. 262. 65MW0, p. 262. 66MWO, pp. 267-268. 67MWO, pp. 269-270. 68MWO, p. 270. 69MWO, p. 264. 7°MW0, p. 263. 7J‘See especially MWO, Chapter I. 72MWO, p. 262. 73MW0, p. 262. 74MWO, p. 257. 75For more on synthetic propositions becoming anal- ytic, see A. Pap, The A Priori In Ppysical Theory, Pt. 1, chapter 3 and his Semantics and Necessarerrut , pp. 229-236. ' .—.n. u 53.-“n a" I 228 76See H. S. Thayer, Meaning and Action, Appendix 6, pp. 508-521. 77For a defense of Lewis along the lines suggested here, see Sandra Rosenthal, "The Analytic, the Synthetic and C. I. Lewis“ in Tulane Studies in Philosophy, 1968, pp. 115-123. 78 AKV, p. 144, quoted in Rosenthal, p. 118. 79AKV, p. 163. 80See Arthur Pap, The A Priori in Physical Theory, pp. 4-5. 81 This distinction should not be confused with one that Lewis made later between symbolic conventions and ex- plicative definitions. See AKV, p. 10599. 82For- additional examples of this kind of transition see Pap, The A Priori in Physical Theory, Part I, chapters 3 and 4. Qy“.. Man-mu j N ; . I I,. ‘m. l SELECTED BIBLIOGRAPHY SELECTED BIBLIOGRAPHY Anderson, Alan and Belnap, Nuel. "Tautological Entailments”, Philosophical Studies. (1962), 9-24. Bennett, Jonathan. "Entailment", Philosophical Review, 78 (1969), 199-236. 1 Blanshard, Brand. The Nature of Thought, Vol. Two, New York, Humanities Press Inc., 1939. . Reaaon and Anatysis. La Salle, Illinois, Open Court Publishing Company, 1962. Brandt, Richard. "Epistemic Priority and Coherence", Journal of Philosophy (1964), 557-559. Child, Arthur. "Toward a Functional Definition of the A Priori", Journal of Philosophy (1944), 155-160. Chisholm, Roderick. "The Problem of Empiricism", Journal of Philosophy, 45 (1948). DeLong, Howard. A Profile of Mathematical Logic. Addison- Wesley Publishing Company, Reading, Mass., 1970. Firth, Roderick. "Coherence, Certainty and Epistemic Priority", Journal of Philosophy, LXI (1964), 545- 557. . "Radical Empiricism and Perceptual Relativity", Philosophical Review, 59 (1950), 164—183 and 319-331. Frankena, William. "C. I. Lewis on the Ground and Nature of the Right", Journal of Philosophy, LXI (1964), 489-496. . "Review of The Ground and Nature_pf tho Right", Philosophical Review, LXVI (1957), 398-402. . "Lewis' Imperatives of Right", Philosophical Studies, 14 (1963), 25-28. 229 230 Frankfurt, Harry. "Philosophical Certainty", Philosophical Gombrich, E. H. Art and Illusion. Pantheon Books, 2nd edition, 1956. Goodman, Nelson. "Sense and Certainty", Philosophical Review, LXI (1952). Harris, Errol E. Nature, Mind and Modern Science, New York, HumanitiesiPress Inc., 1954. if Kneale, William. "Are Necessary Truths True by Convention?", ‘ Proceedings of the Aristotelian Society, Supple- mentary Vol. XXI 31947). Kneale and Kneale. The Development of Logic, Oxford, . England, Clarendon Press, 1962. a Lewis, C. I. The Collected Papers of Clarence Irving Lewis, L ed. by_3ohn D. Goheen and John L. MotherShead Jr., Stanford, California, Stanford University Press, 1970 O . 5 Survey ofiaymbolic Lo ic, Berkeley, University of California Press, 1918. . Mind and the World Order, New York, Charles Scribners, 1929. and Langford, C. H. Symbolic Lo ic, New York, The Appleton-Century Company, 193%. . An Analysis of Knoyledge and Valuation, La Salle, Illinois, Open Court Publishing Co., 1946. . The Ground and Nature of the Right, New York, Columbia UniverSity Press, 1955. . Values and Imperatives, ed. by John Lange, Stan- ford» Stanford University Press, 1969. Linsky, Leonard. "The Incommunicability of Content", Journal of Philosophy, 59, 21-23. Malcolm, Norman. Knowledge and Certainty, Englewood Cliffs, N. J., Prentice-Hall, 1963. McClelland, D. C. and Atkinson, J. W. "The Projective Ex- pression of Needs: I. The Effect of Different Intensities of the Hunger Drive on Perception", Journal of Psychology (1948), 25, 205-222. 231 Nelson, E. J. "Deductive Systems and the Absoluteness of Logic", Mind, XLII (1933), 30-42. Pap, Arthur.. The A Priori in Physical Theory, New York, Russell and Russeli, 1946. . Semantics and Necessary Truth, New Haven, Yale University Press, 1958. Price, H. H. Perception, London, Methuen, 1932. Quine, W. V. 0. Mathematical Logic, Cambridge, Mass., Harvard UniverSity Press, 1940. Rescher, Nicholas. Many-Valued Logic, New York, MCGraw- Hill Book Company, 1969. ' . The Coherence Theory of Truth, Oxford, Clarendon Press, 1973. Rosenthal, Sandra. "The Analytic, The Synthetic and C. I. Lewis", Tulane Studies in Philosophy, 1968, 115-123. Saydah, J. Roger. The Ethical Theory of Clarence Irving Lewis, Athens, Ohio; Ohio UniversityiPress, 1969. Scheffler, Isreal. Science and Subjectivity, New York, The Bobbs-Merrill Company, Inc., I967. Schilpp, Paul A. (ed.) The Philosophy 9f C. I. Lewis, La Salle, Illinois; Open Court Publishing Company, 1968. Sellars, Wilfrid. Science, Perceptiohand Reality, London, Routledge and Kegan Paul, 1963. Thayer, H. S. haaning ang Action, New York, The Bobbs- Merrill Company, Inc., 1968. "‘lLILLLLLLLLLLLES