James W. Cornman, "Materialism and Some Myths about Some Givens," The Monist Vol. 56, No. 2 (April 1972): 215-233.


James W. Cornman
University of Pennsylvania

Not long ago it was generally agreed that something is given, indeed, that something must be given if there is to be any empirical knowledge at all. Lately, however, what people have called "the" given has come under attack from many different angles. In general the attacks have been of two sorts: epistemnlogical and metaphysical. The problem of countering the skeptic effectively about our knowledge of the external world has been one source for the attack on the given. It has been thought by many that if nothing is certain then nothingjsprobable and the only thing which is empiricaland certain is the given. But since the given was thought to be each individual's unique sensory experience, it was not seen how to make knowledge of the external world even probable relative to that. Recent attacks on this kind of skepticism have generally been of two sorts. The first denies that only sense experience is given. Indeed it claims that some statements about the external world are themselves certain.1 The second claim is that nothing at all is given, but this does not justify skepticism because empirical knowledge does not require a foundation with certainty at its base.2

The metaphysical attacks on the given have also been of two sorts. One has been in defense of scientific realism, the view that what there is just what science says there is, and the other has been in defense of materialism, which denied that there is anything mental over and above what is physical. Wilfrid Sellars, in defense of his version of scientific realism, has attacked what he calls "the myth of the given," and Richard Rorty in defense of eliminative materialism has also attacked what he gives the same name.3 Indeed, Rorty seems to think that the "myth" he identifies is the only thing standing in the way of success of his version of eliminative materialism, which I have called the "sensation" elimination theory.4 And because he thinks that this given is a myth, eliminative materialism is, or some day will be, established. In this paper one of my primary tasks is to examine this view I have attributed to Rorty. However, because there is not merely one conception of the given, there are several possible myths of the given, and so I shall first try to distinguish and discuss the most important and relevant sorts of given.

Four Kinds of Given

We can distinguish four general kinds of given, namely, the phenomenologically given, the factually given, the epistemologically given, and the linguistically given. Furthermore, all of these have distinct species which must be delineated. I shall, then, list and define each species. What is essential to something being given phenomenologically, is that something being given in this way neither entails that it exists nor that it does not exist. I find there to be two important senses of phenomenologically given, the second of which entails the first:

  1. P's are given = df. In perceiving, some perceivers are ostensibly presented with P's.
  2. P's are given = df. In perceiving, some perceivers are ostensibly presented with P's noninferentialiy and without interpretation.

The second sort of given I have called "factual" for two reasons. Unlike the phenomenological given, it does imply the existence of what is given, and it makes no reference to anything linguistic or epistemological. Again, there are two species, paralleling the two species of phenomenologically given:

  1. P's are given = df. P's are directly perceived by perceivers.
  2. P's are given = df. P's are presented to perceivers noninferentialiy and without interpretation.5
There are also two kinds of epistemologically given which correspond somewhat to the two factual givens. The first has to do with entities about which there is supposedly knowledge of a certain sort, and the second has to do with certain beliefs or statements which supposedly lie at the foundation of empirical knowledge.6
  1. P's are given = df. Fs are entities about which there is certain and noninferential (perceptual) knowledge.
  2. P's are given = df. Beliefs (statements) about P's are known with certainty and noninferentially, and are that upon which all empirical beliefs (statements) which are known are based.

The linguistic givens have come into discussion primarily since the advent of linguistic philosophy and its discussion of terms of language rather than nonlinguistic entities. These givens have been thought to have some consequences for the epistemological givens, but, once defined, we shall see that this is false of the four we shall discuss. The first three of these are purely linguistic because what is given is explicated solely in terms of linguistic entities. They differ, however, in that the first concerns those terms that can be used to report perceptual experiences, while the other two, which are relevant to the myth which Sellars combats, concern terms which are essential to the meaningfulness of language. The fourth, which Rorty attributes to me because of my objection to his materialism, concerns nonlinguistic entities which are experienced independent of and unaffected by language, but which are essential to learning language. The first corresponds roughly to the first species of epistemological given, and the last three resemble, in different ways, the second species.

  1. P's are given = df. P-terms are correctly used by perceivers to give a direct response to (report) their perceptual experiences.7
  2. P's are given = df. P-terms are terms from which the meaningfulness of all empirical terms derives.
  3. P's are given = df. P-terms stand for the extracoconceptual and extralinguistic entities from which the meaningfulness of all empirical terms derives.8
  4. P's are given = df. P's are the entities which are experienced even before learning and independent of language, and with which a language must be associated if it is to be learned.9

Before turning to an examination of any significance of the various givens for materialism, I wish to consider how plausible it is to think that something is given in each of these ways, and, in those cases where this is plausible, how plausible it is that sensations are given. I am using the term 'sensation' here very broadly to cover not only phenomenal objects, such as red sense-data or percepts, which I shall call "sensa," but also any event of someone sensing, whether this be construed as a relation between a person and a sensum or as an objectless adverbial sensing event, such as a person red-sensing or sensing red-ly. I shall claim that there is good reason to think that something is nonlinguistically given in senses (1) and (3), but that it is at least questionable whether anything is given in any of the other nonlinguistic ways. Furthermore, the only sense of given for which it is clearly reasonable to claim that sensations are given is linguistic sense (7), although it is somewhat plausible to hold they are given in epistemological senses (5) and (6) if the reference to certainty is omitted. This might give some comfort to materialists, but, I shall argue, not enough to prevent the refutation of Rorty's form of eliminative materialism. Indeed, I shall maintain that even granting that sensations are not given in any of these ten ways except (7), my objection, to Rorty's materialistic theory remains unaffected.

Phenomenological Givens

I think it is clear that perceivers are ostensibly presented with objects in perceptual experiences, whether veridical, illusory, or hallucinatory. We might call these objects the contents of the experiences, where something can be a content yet not exist. Perceivers are ostensibly presented with, have as content of their perceptual experiences, colored objects of various sizes and shapes which smell, taste, sound and feel various ways. Surely this is how most perceivers would describe the content of their perceptual experiences. But these ostensibly presented objects are not sensa, but rather physical objects.10 Thus even if sensa should be given factually, they are not phenomenologically given in this first sense, and, therefore, not in the second sense either.

While it is clear that some things, but not sensa, are given in sense (1), it is far from clear that anything is given in sense (2), but to see this we must first clarify 'without interpretation' and 'noninferentially'. If the former implies that none of the content of any perceptual experiences is dependent on the perceiver's attitudes, interests, and ways of conceptualizing, then it is doubtful that anything is given in sense (2).11 Such states and dispositions of perceivers often affect the sensory imput from the senses and thereby affect what results as particular content. If, however, it implies only that at least some elements in some of the contents, such as an ostensibly presented brilliant red color, are unaffected by such dispositions and states, or that such features are not altered merely by conceptualizing them differently or changing one's attitude toward or interest in them, then the case against an uninterpreted given is considerably weakened. This is, however, a question to be answered by physiologists and psychologists, either by discovery or by postulation which is justified in terms of fruitfulness of explanation.

The question of whether some processes of inferring precede or are causally relevant to all or at least some part of the contents of perceptual experience is also open and to be answered scientifically. Intuitively it seems that we rarely go through stages of inferring prior to being presented some sense-content. It may also seem that there must be at least something in the content of a perceptual experience independent of any inferring process, because if inferring occurs there must be something with which each inference begins. However, it may be that the stimulation of sense organs always, or at least sometimes, results in a process of inferring which precedes and, perhaps, causally affects what results as content, but which is not an inference from the content of any perceptual experience. However, even if it were to turn out that something, perhaps even sensations, are given in this way, this alone would have no consequences for something nonphenomenologically given. It has no consequences for the existence of something factually given, for whether anything is known with certainty and noninferentially, for whether language must be learned a certain way, and for whether certain terms are essential to the meaningfulness of all empirical terms. Furthermore, both of these kinds of phenomenological givens are consistent with materialism. What is phenomenologically given may not exist, and so it is possible that everything is physical even if sensations are phenomenologically given.

Factual Givens

I have claimed that something is given in sense (3). This is because if something is perceived, as some things certainly are, then something is directly perceived in a factual sense, that is, roughly, something is perceptually experienced without some causal intermediary being experienced.12 Whether physical objects or sensa are given in this sense is, of course, the debate between direct realists and indirect realists, but this does not affect the claim that something or other is given in this sense. Furthermore, because the debate between the two sorts of realists is unresolved, we cannot now decide the issue of what in particular is directly perceived. It might be, for example, that theoretical requirements of science or the unification of science would force the postulation of sensa.13 But, no matter what is given in this sense, it has no consequences for the linguistic and epistemological givens. However, sensa being given in this sense might be helpful for explaining why we are phenomenologically given physical objects of various colors, textures, smells, and tastes.

There is little to say about the given in sense (4) which has not already been said about senses (2) and (3). Whether something, such as a sensum, is presented in perceptual experiences, whether a perceiver's attitudes, interests, and modes of conceptualization affect what it is, and whether it is presented only after a process of inferring is to be decided primarily by scientific theories of perception. Of course, it does seem that certain brain processes are produced non-inferentially in perception, and if they were identical with sensa, then they would be presented, as sensa, noninferentially. However, it would still be doubtful that a person's attitudes, interests, and conceptualizing, or at least his physical realizations of these phgnomexia, would have no affect on these brain processes.14 Furthermore, if some forms of direct realism are correct, then physical objects are presented, and while it may be that some inferring process occurs before the objects are presented, they are not affected by interpretation. Of course, the way they appear may be so affected, but if there are no sensa, the way a thing appears is not some entity presented to someone. It is merely a modification of a perceptual experience. As with the previously discussed givens, this one has no consequences for linguistic or epistemological givens. It might, however, help explain why something is given in the two phenomenological ways, and, if sensa are given in sense (4) or in sense (3), problems arise for materialists. We must, then, return to these two givens later.

Epistemological Givens

An initial problem for characterizing the first sort of epistemological given is that, generally, propositional knowledge is what is said to be inferential or not, and certain or not. For example, we might say that John perceptually knows with certainty or noninferentially that his chair is red. The problem is how to make sense of saying that John knows the red chair noninferentially or with certainty. We can begin to get an understanding sufficient for our purposes by using the following initial rough definition:

P's are known (with certainty, noninferentially) = df. There is (certain, noninferential) knowledge of P's.
There are, however, two ways to interpret the preceding definiens, the first of which is at least initially plausible but the other is not. For the second, consider Berkeley's view when he says that individual sensations "are perfectly known, there being nothing in them which is not perceived."15 Berkeley would seem to agree that everything about sensations can be known with certainty and noninferentially. His reason would seem to be that everything about them is fully presented and manifest to perceivers. However, this claim is at least open to doubt because whether or not sensations are identical with brain processes and thus whether sensations have properties not manifestly presented in perception is far from perfectly known. It is more reasonable to claim:
There is (certain, noninferential) knowledge of P's = df. Something about P's is known (with certainty, noninferentially).

Once we accept this interpretation of (5), we can see that it is entailed by (6) which goes beyond it in its claim of a foundational theory of empirical knowledge. We are still not finished unpacking (5), however, because we must interpret 'certainty' and 'noninferential'. Let us define the first as:

S knows that F with certainty = df. S knows that F and nothingjs more reasonable for S to believe than that F.16
We can construe the second as:
S knows that F noninferentially = df. S knows that F and S would know that F even if he should not base it on any other evidence.17

I have said that it is somewhat plausible to claim that some things, indeed, sensations, are given in sense (5) if the reference to certainty is omitted. The stronger claim might be argued, however, that something must be noninferentially known if anything is known. This is because a person knows something, that F, inferentially only if he is justified in believing the evidence, eit on which it is based and ex is true. But if he is justified in believing e1 only if he bases it on some evidence e2, and for any such en, he is justified in believing it only if he bases it on some evidence en+1, then he does not know that F because he never becomes justified in believing e1. Thus if there is knowledge, which we can assume here, there is noninferential knowledge. However, this conclusion does not follow. What follows is that if someone has knowledge then there is some true evidence which he is justified in believing without basing it on others evidence, but, as recent discussion has shown, justified true belief is not sufficient for knowledge. However, while this regress argument is unsound, it may be that knowing that F implies knowing the evidence for F. If so, the argument could be made sound. I shall not debate this but leave it until, if ever, the elusive "fourth" condition of knowing is uncovered. It is sufficient to say here that noninferential justification is required for knowledge and that whether noninferential knowledge is required remains an open question.

Is there noninferential justified belief or knowledge about sensations? It has been thought not only that there is but that there must be it there is any empirical knowledge at all, because no other empirical beliefs are noninferentially justified. This is one reason given for a foundational view of empirical knowledge, such as expressed in (6), that puts sensation statements at the base of all knowledge, whether as the only epistemologically basic statements or merely as one essential subset of such statements. The second reason is that sensation statements are the only empirical statements known with certainty, and there is empirical knowledge only if something empirical is known, or at least justified, with certainty. The claims involved here are too complicated to consider now, but several points can be made in passing. As defined above, neither noninferential justification nor justification with certainty entails the other, and while it is reasonable to hold that some empirical statements, including many sensation statements, are noninferentially justifiable, there is more doubt about justification with certainty. It might be claimed, for example, that it is always more reasonable for me to believe that I either have a sensation or I do not have a sensation, than that I have a sensation. This, however, casts no doubt on whether I am sometimes noninferentially justified in believing that I have a sensation. And, while many sensation statements are among the clearest empirical candidates for justification which is both noninferential and certain, it is also clear that some sensation statements are not so justified and that there are plausible candidates of other kinds. The sensation sentence, 'That is the two thousandth twinge of pain I have had' is neither certain nor noninferentially justified. The physical object sentence, 'This is my hand' is, on many occasions, a candidate for noninferential and certain justification. It is, then, plausible to argue that the foundational view of knowledge which allows only sensation statements to be basic is mistaken. And, even if they are known both with certainty and noninferentially, if other sorts of statements are also known non-inferentially, then sensation statements need not be even an essential part of the foundation of empirical knowledge. Of course, some people argue that knowledge has no foundation, because, for example, a large group of statements purporting to describe and explain what there is, is justified together rather than stepwise from foundational statements.18 This is, however, compatible with some of the statements within the group being known with certainty and noninferentially. It is not compatible with those statements being needed to justify all other empirical statements.

Both epistemological givens have important consequences. Neither implies that something is phenomenologically, factually, or linguistically given, but if something is given in either epistemological way, then it exists or occurs. This poses a definite threat to eliminative materialism and makes the reduction of sensations to physical phenomena required by reductive materialism very dubious if some of the noninferential or certain knowledge would be about the phenomenal properties of sensa. It is these properties, such as the intensity or throbbing character of pains, that pose the crucial problem for such reductions.19 Consequently, a materialist must beware of both forms of epistemological given. Fortunately for him, with the many current attacks on these givens, the likelihood of his success is considerably better than it might have seemed not too long ago.

Linguistic Givens

I have listed four different sorts of linguistic givens and have claimed that sensations are given in sense (7). I find that this sense has no consequences for any other of the givens, or materialism in general, or scientific realism. However, if we add that only sensations are given in this sense, at least for certain sorts of experiences, then materialism faces problems. If we must use sensation terms to report some perceptual experiences correctly, then some sensation sentences are true, and both forms of materialism face problems. But if we understand reporting or directly responding to an experience as making an identifying reference to the experience, then while sensation terms can be used for this purpose, they need not be used. Observation terms, sometimes combined with the appearing terminology, or even theoretical terms, such as 'the entity in one-to-one simultaneity correspondence with my presently firing C-fibers', can be used. Thus although sensations are given in sense (7), no problems arise for materialism because they are not the only things given in this sense.

It is interesting to note that Sellars, a principal foe of the given, does not seem to oppose sensations and only sensations being given in this sense for certain sorts of inner, that is, nonperceptual experiences. It is observation terms playing such a role for perceptual exeriences that he rejects. This is because Sellars' scientific realism requires that the best description of what there is be provided by those theoretical terms of a ufied science that are needed for the best scientific explanations. For Sellars, sensation terms, unlike observation terms, are theoretical terms (although also reporting terms), because, very roughly, they are used to explain, but not describe, observable human behavior. The problem here for Sellars, however, is that although sensation terms are theoretical, they do not seem needed for scientific explanation, as Quine points out.20 If sensation terms explain behavior, then those physicalistic terms that stand for the physical correlates of sensation explain it at least as well. Thus if sensations are uniquely given in sense (7) in any situations, and the Quinean claim about explanation is correct, as seems reasonable, Sellars' scientific realism is mistaken. He should beware of both observation terms and sensation terms.

Sellars opposes anything being given in sense (9), and thus it might seem that he also opposes anything being given in sense (8). But this is mistaken, because he holds that "priority in the order of concept formation must not be confused with ontological priority."21 For Sellars the meanings of theoretical terms, including sensation terms, are derived analogically from observation terms, and so, although sense (8) is true of and only of observation terms, he can still maintain that they do not stand for anything at all as his scientific realism requires. Furthermore, it is plausible to claim that "observation entities," by which I mean entities corresponding to observation terms, are given in sense (8), although it is questionable whether only they, and not sensations, are so given. It is surely debatable whether sensation terms are analogically derived from observation terms as Sellars maintains.22 *

Sellars must reject sense (9) for observation terms. If observation terms are basic in the order of concept formation because they stand for, or express, or picture nonlinguistic entities from which all empirical meaning derives, then there is a descriptive need for observation terms contrary to Sellars' scientific realism. He takes, however, a very extreme position in justifying his rejection of sense (9), namely, that no predicates at all stand for anything "independent of thought and language," because they stand for attributes and all attributes are conceptual or linguistic entities.23 I shall not consider here Sellars' complicated attempt to establish this thesis, but I would like to point out that he need not use such a strong and initially dubious claim to justify rejecting (9).24 He need only show that observation terms, and perhaps sensation terms, do not stand for or express any independent attributes. This would allow him to agree that there are independent attributes, but to argue, much as he did in his discussion of universals, that although we may now refer to these attributes, perhaps misleadingly, by observation terms, it is plausible to hold that they do not stand for or express such attributes.25 This permits what is required for his scientific realism, that the theoretical predicates of the "final" unified science will express or at least come closest to expressing those attributes. Thus Sellars can grant that some terms stand for independent attributes, but he cannot agree that these are the terms from which all empirical meaning derives because of his view that sense (8) is true of observation terms. Furthermore, I think that this position is at least plausible. There is an open question about what kinds of terms, if any, express independent attributes, even granting that observation terms, and perhaps sensation terms, are basic in the order of concept formation.

It is not clear that Sellars agrees with Rorty that nothing is given in sense (10), although he would claim that observation objects are not given in this way. His reason for this would seem to parallel that for his rejection of sense (9). If there is pre- or extralinguistic awareness of observation entities to which language must be "fitted" in order to be learned, and if this fitting entails that the observation terms express or stand for such entities, then Sellars is faced with those consequences of (9) he wishes to avoid -- that observation terms are not replaceable. However, his problem does not arise because these terms are necessary for language learning or meaningfulness. Terms can be basic in the order of "concept learning," and also concept formation, but actually stand for nothing, if the association mentioned in (10) does not entail that they stand for nonlinguistic entities. Because there certainly seems to be no such entailment, Sellars could reject sense (10) while accepting something being given in the following sense:

(10a) P's are given = df. P-terms are used to refer to the entities which are evnerienced before learning and independent of language and with which a language must be associated if it is to be learned.

The crucial difference between (10) and (10a) is that being given in sense (10) entails that there are P's, but being given in sense (10a) entails only that there are P-terms and some entities referred to by P-terms. Something can be given in sense (10a) yet not be given in sense (10). Furthermore, (10a) as applied to sensations raises no problems for materialism, but (10) with its commitment to the existences of sensations raises problems for both kinds of materialism. We can, then, forget (10a). I shall assume it plausible to reject sense (10) with its restriction on which entities are needed for language learning, unless Rorty is correct in thinking my objection to his version of eliminative materialism requires me to accept (10) as applied to sensations. We must consider this possibility next. I shall also consider whether my reply requires sensations to be given in any of the other nine senses we have examined.

The Elimination of Sensations and Some Givens

I have called Rorty's version of eliminative materialism "the 'sensation' elimination theory" because he claims that, although sensation terms are used to explain, predict, report, and describe, they are not needed for any of these purposes and thus we are justified in eliminating sensation terms and sensations.26 The reason given for why these terms are not needed for these four purposes is that physicalistic terms either have, or can come to have, or can do a better job in these four roles, and thus sensation terms can be dispensed with except for convenience. In my article I granted, for the sake of argument, that sensation terms are not needed for explanation, predication, and reporting, and that the ability to make an accurate, precise description of what we report using sensation terms would not be lost if sensation terms are eliminated.

If Rorty is right, however, it must also be true that no sensation terms are needed to make any descriptions which are either true or accurate. It is not clear from his articles just how he proposes to show this, but it seems to be in one of two ways. My objection is that neither way succeeds.27 The first, which I have taken to be his approach, is to claim that purely physicalistic terms can accommodate the true descriptive roles now played by sensation terms. Thus they must be usable not only to describe the same entities as sensation terms, but also to provide at least the same descriptions of those entities. This requires that for any descriptive sensation sentence that entails a sentence, there is some physicalistic sentence that also entails it. This is because any sentence that takes over the descriptive role of another must have at least all the entailments of the second. The problem for this attempt is that all available evidence points to no physicalistic sentence having all the entailments of a sensation sentence such as 'John has an intense, sharp, throbbing pain', which seems to be true all too often. Of course, if such sentences are never true, Rorty need not consider them but that claim seems quite unreasonable, and Rorty offers nothing to make it reasonable.

The second way to construe Rorty's approach, which I believe he confuses with the first, is to claim that the descriptive roles of sensation sentences which we now consider true will not be taken over by, but rather will be replaced by the different roles of certain physicalistic sentences. On this view, sensation sentences are to the relevant physicahstic sentences much as the sentence 'It is somewhat likely that it will rain tomorrow' is to more precise sentences of the form 'The probability that it will rain tomorrow on evidence e, is n'. In neither case must the second have all the entailments of the first and, supposedly, in neither case is the first needed for either true or accurate descriptions. However, the problem facing Rorty on this interpretation is how to justify the claim that seemingly true descriptive sensation sentences are inaccurate attempts to provide descriptions that are more accurately provided by certain physicalistic sentences. For example, what 'John has an intense, sharp, throbbing pain' ascribes to John is so different from that ascribed by proposed physicahstic replacements, such as 'John's C-fibers are firing rapidly,' that we would have to conclude that a person's own inner experience is not even remotely like the way it seems to him when he believes he has an intense, sharp, throbbing pain. Not only would he have no special epistemological status about his own experiences, he would not even have any true beliefs whenever he expresses his beliefs using sensation terms. These consequences of this approach to the 'sensation' elimination theory are so implausible that both the consequences and the approach itself should be rejected.

The question before us is whether in my two-pronged attack on Rorty's theory I am "taking for granted that there is a sort of pre-linguistic giveness about e.g. pains, which any language which is to be adequate must provide a means of expressing," as Rorty claims.28 Am I even assuming any sort of given? I certainly claim that pains are given in senses (1) and (7) if these givens are amended to apply to inner experience instead of perceptual experiences. Rorty would at least agree about sense (7). Furthermore both Rorty and I need sensations to be given in sense (7), because a person's directly responding to, or reporting, his inner experience in sensation terms is central to his version of eliminative materialism and to my rejection of it.29 While my rejection of the second approach can utilize sensations being given in sense (1), because the implausibility of replacing sensation terms with physicalistic terms can be supported in part by the plausibility of people sometimes correctly describing at least the content of their inner experience in sensation terms, it does not need any claim about such content.

Nothing I say implies anything about the orders of concept learning and concept formation, and while I do believe that some of the same sorts of things are experienced betore learning and independent of language, this is not required by my objection to Rorty. The crux of my objection is threefold: no physicalistic sentences have the same entailments as many sensation sentences that seem to be true, it is implausible that none of those sensation sentences are true, and it is implausible that all sensation sentences are at best remote approximations to true descriptions which can be approached much more closely by physicalistic sentences. None of these claims requires that sensations such as pains, or their attributes are pre- or extralinguistic entities with which language must be associated to be learned. All three are consistent with pains never being experienced independently of language or before language is learned. Furthermore, the claims are compatible with the view that sensation terms are not basic in the order of concept learning, and with the view that language need not be associated with sensations to be learned. These claims are also compatible with Sellars' thesis that sensation terms are not basic in the order of concept formation, and thus they are compatible with sensations not being given in senses (8) and (9).

It might be thought that my two claims about implausibility at least commit me to the extralinguistic existence of sensations and some of their attributes as stated in (10). While I agree that this is true of sensations and their instances of properties, my two claims do not require it. Indeed my claims are even compatible with the view that no entities we are aware of are extralinguistic. Consider Rorty's view that "what appears to us. or what we experience, or what we are aware of, is a function of the language we use. . . . If we got in the habit of using neurological terms in place of 'intense,' 'sharp,' and 'throbbing,' then our experience would be of things having those neurological properties, and not of anything, e.g., intense."30 While this extreme linguistc idealism or relativism allows for the possibility that in the future everything we experience will be purely physical because we will have been weaned from sensation terms and to purely physicalistic terms, this mere possibility does not affect my two claims. Only if this possibility is shown to be likely would they be refuted, but there are reasons for its being unlikely. Custom and convenience are quite likely to keep us using sensation terms, and so, given this linguistic idealism, it would be at least as likely we would continue to experience sensations and their properties. Thus, as my claims state, it would also be implausible that no sensation descriptions would be either true or accurate.

My objection to Rorty's "sensation" elimination theory does not rely on linguistic givens (8), (9), and (10), and it is clear it does not rely on anything being phenomenologically given in sense (2) or factually given in sense (4) with their references to experience without inference or interpretation. Furthermore, it does not require that sensa are directly perceived in perception or even that sensations such as pains are directlyexperienced in inner experience. Although the thesis that persons indirectly experience their own pains is at least very odd, my two claims about implausibility would be unaffected if it were true. It is also clear that my objection does not require anything to be epistemologically given in sense (6) because it does not require a foundational view of knowledge. It may be claimed, however, that it does require pains to be epistemologically given in sense (5), and, furthermore, that the person having a pain is the final epistemological authority about it.31 But, first, my claims are justified if it is implausible to claim that no sensation descriptions are true or accurate, and this does not require knowledge, certainty or even noninferential justification about sensations. Second, my claims would be strengthened, rather than weakened, by a view that a person can be overruled about his own pains once science establishes contingent correlations between brain states and pains, because once such correlations are established it would be most implausible to claim that no sensation sentences are true. Some would be true whenever their correlate physicalistic sentences are true. Furthermore, these physicalistic correlates would be just those sentences usually proposed as replacements for the sensation sentences, but the reasons for finding them to be implausible candidates for this job are not affected by the confirmation of empirical correlations.

Summary and Conclusion

Of the ten sorts of given we have examined, my objection to Rorty relies only on sensations being given in linguistic sense (7), although it would be helpful if pains should be given in something like phenomenological sense (1) or factual sense (3). And while sensations being given in sense (7) is not accepted by those who take sensation terms to be either merely expressive or pure theoretical terms, it is not only quite reasonable to construe them as reporting terms but Rorty's view requires that they be so construed. It is clear, then, that eliminative materialism, and also reductive materialism, which entails the identity theory, are compatible with sensations being given in sense (7). Thus my attack on Rorty relies on no conception of a given which a materialist must reject, although some sorts of materialist, such as postulation elimination theorists, do reject sense (7).32 Thus if all these givens, except sense (7), prove to be myths about sensations, my objection will be unaffected.

We can also summarize which givens must be myths about sensations if materialism is to succeed. With the demise of analytical behaviorism which allows sensation sentences to be true because analyzable into behavioral sentences, and the refutation of the "sensation" elimination theory, an eliminative materialist cannot allow any descriptive sensation sentences to be true, because they would be irreplaceably true. But if sensations are given in either factual sense, either epistemological sense, or in linguistic senses (9) or (10), then some sensation descriptions are true. Thus all six of these givens must be myths if some form of eliminative materialism is to avoid refutation. However, as I have argued elsewhere, there are reasons which do not depend on any of these givens and which are sufficient to refute the extant forms of eliminative materialism.33 Reductive materialism, however, can survive if none of all ten givens are myths about sensations, because by identifying sensations with brain phenomena it allows for certain sensation sentences to be true and known with certainty and noninferentially. The problem for a reductive materialist is how to handle phenomenal properties of sensations. such as being intense, or sharp, or throbbing. If they are not reduced or eliminated in some way, then although sensations may be identical with certain brain phenomena, they would not be reduced to something physical as reductive materialism requires. Consequently, if some phenomenal properties are given in any of the preceding six ways and are not reducible to physical properties, reductive materialism is also refuted. Phenomenal properties, then, pose the crucial problem for the materialist, because if that problem is not resolved then there seems to be no version of his thesis that resists refutation.

A materialist, we can conclude, can accommodate sensations, and thus some things being given in any of the ten ways we have examined, but he must be wary of the ways in which he allows phenomenal properties to be given. He would, then, be much safer if nothing was given in any of the senses. We have, however, found good reason to accept something being given in senses (1), (3), (7), and (8), but these, and also (2), (4), and (10a) which are more questionable, pose no problem for him. We also found it questionable whether anything is given in linguistic senses (9) and (10). This leaves just the two epistemological senses to worry the materialist. Of course, if only physical objects are given in either sense he need not resort to the extreme measures taken by some, that is, to deny that any knowledge is both certain and noninferential, and to reject any foundational view of knowledge. One problem for this defense of materialism, however, is that it is far from clear whether it is based on fact or only on myths popular today but correctly discarded tomorrow. A defense of a venerable metaphysical thesis deserves better grounds than that.34


1. See A. M. Quinton. "The Problem of Perception," in Perceiving, Sensing, and Knowing, ed. by J. Swartz (New York: Anchor Books, 1965), pp. 511-518.

2. See W. V. Quine, From a Logical Point of View (New York: Harper & Row, 1963), Chap. 2; and B. Aune, Knowledge, Mind, and Nature (New York: Random House, 1967), passim.

3. See W. Sellars, Science, Perception and Reality (New York: Humanities Press, 1963), pp. 157, 161-170; and Philosophical Perspectives (New York: Charles C Thomas, 1967), p. 353. See also R. Rorty, "In Defense of Eliminative Materialism," Review of Metaphysics, 24, No. 1 (1970), 117-121.

4. See my "On the Elimination of 'Sensations' and Sensations," Review of Metaphysics, 22 (1968), 15-35, and Materialism and Sensations (New Haven: Yale University Press, 1971), Chap. 5.

5. See C. I. Lewis, Mind and the World Order (New York: Dover Publications, 1956), pp. 45-53.

6. See R. M. Chisholm, Theory of Knowledge (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1966), Chaps. 2 and 3; Sellars, Science, Perception and Reality, pp. I64ff.

7. Cf. Sellars, Philosophical Perspectives, pp. 352-353.

8. Ibid., pp. 351 and 362.

9. See Sellars, Science, Perception and Reality, pp. 161-162: Rorty, "In Defense," p. 118; and I. Scheffler, Science and Subjectivity (Indianapolis: Bobbs-Merrill), pp. 23-26.

10. It should be noted that I am claiming that physical objects, not ostensible physical objects, are phenomenologically given. It may be, however, that ostensible physical objects, i.e., are factually piven and this may help explain why physical objects are phenomenologically given. Cf. R. Firth, "Sense-Data and the Percept Theory." in Schwartz, ed., Perceiving, Sensing, and Knowing, pp. 219-225.

11. See Lewis and Scheffler, pp. 28-31.

12. For a more detailed examination of direct perception, see my "On Direct Perception" (unpublished).

13. For a discussion of this, see my "Sellars, Scientific Realism, and Sensa," Review of Metaphysics, 24 (1970), 417-451, and Sellars' reply, "Science, Sense Impressions and Sensa: A Reply to Cornman," Review of Metaphysics, 25 (1971), 391-447.

14. For a different view, see B. Russell, Philosophy (New York: W. W. Norton, 1927), p. 207,

15. Berkeley, Principles of Human Understanding, sec. 87.

16. For this definition of 'certainty', see R. Chisholm, Perceiving (New York: Cornell University Press, 1957), pp. 19-21. For other views, see Quinton, pp. 511-512, and H. Frankfurt, "Philosophical Certainty," Philosophical Review (1962), pp. 317-327.

17. Cf. Chisholm, Theory of Knowledge, pp. 26-27,

18. See Quine and Aune.

19. For a detailed discussion, see my Materialism and Sensations, Part I.

20. See W. V. Quine, Word and Object (Cambridge: MIT Press, 1960), pp. 264-265; and The Ways of Paradox (New York: Random House, 1966), pp. 208-214.

21. W. Sellars, "Rejoinder" (To Aune), in Intentionality, Minds, and Perception, ed. by H. Castaneda (Detroit: Wayne State University Press, 1967), p. 296 n.

22. See Sellars, Science, Perception and Reality, pp. 47-48 and pp. 91-95.

23. Sellars, Philosophical Perspectives, p. 362.

24. For Sellars' attempt, see "Abstract Entities," Review of Metaphysics, 17 (1963), 627-671.

25. See Sellars, Philosophical Perspectives, pp. 370-375.

26. See Rorty, "Mind-Body Identity, Privacy, and Categories," reprinted in Modern Materialism: Readings on Mind-Body Identity, ed. by J. O'Connor (New York: Harcourt, Brace and World, 1969), pp. 147-174.

27. See my "On the Elimination of 'Sensations,'" pp. 28-30.

28. Rorty, "In Defense," p. 117.

29. See Rorty, "Mind-Body Identity," p. 151.

30. Rorty, "In Defense," p. 117.

31. I have discussed final epistemological authority in Materialism and Sensations, Chap. 1.

32. See Quine for an example of such a theory. I discuss his theory in Materialism and Sensations, Chap. 4.

33. In Materialism and Sensations, Part II, I have discussed and rejected the extant forms of eliminative materialism, namely, analytic behaviorism, Rylean behaviorism, and the postulation and "sensation" elimination theories.

34. I have done detailed work on a defense of one form of materialism, which I call "adverbial" materialism because it utilizes the adverbial construal of sense experience, in Materialism and Sensations, Part III.