PERCEPTUAL RELATIVITY MARK PASTIN

In this paper I make a proposal concerning the meaning of physical statements. I do not attempt conclusive support for this proposal. Given the important implications of the proposal, it deserves a hearing. If this proposal is correct, the current abandonment of traditional empiricism in epistemology is not justified by two considerations that motivate it. The traditional empiricism I have in mind maintains that knowledge of physical reality is based just on a person's present and recalled sensory experiences and inductive generalizations therefrom. Traditional empiricism is often thought to entail some form of phenomenalism, the view that physical statements are in some sense equivalent to sensory statements. Philosophers endorsing this entailment often take the Argument from Perceptual Relativity (APR) to refute phenomenalism and traditional empiricism. Other philosophers reject traditional empiricism because they think it presupposes a false theory of meaning which, via phenomenalism or not, requires that statements by themselves, not in the context of a complete theory, be confirmable or infirmable. "The dogma or reductionism survives in the supposition that each statement, taken in isolation from its fellows, can admit of confirmation or infirmation at all. My counter-suggestion.., is that our statements about the external world face the tribunal of sense experience not individually but only as a corporate body. ''1 On the proposal to be outlined the APR does not refute traditional empiricism. A variant of the proposal shows the compatibility of traditional empiricism and corporate theories of meaning. I first examine the APR. I then outline the proposal. Finally objections to the proposal are considered. 1. The APR and Responses. The APR is directed at phenomenal. ism. In order to discuss phenomenalism I adopt some conventions concerning statements. Statements, as I shall understand them, are 341

M A R K PASTIN

ordered triples of sentences, propositions, and contexts such that the sentence can be used to express the proposition in the context. Talk of propositions is a device for discussing what different sentences used to assert the same thing have in common. I shall not explore the ontological implications of this device. Sensory statements have sentence components which contain only terms for sensory experiences, persons having sensory experiences, and times at which sensory experiences occur. Sentence components of physical statements contain terms for physical objects other than person and time indices for experiences, terms such as 'book,' 'cloud,' and 'pi-meson.' The characterization of sensory and physical statements is discussed in Section 2. Phenomenalism claims that for each physical statement there is a sensory statement having the same proposition component. The APR is this: Consider statements P and S such that S is the sensory counterpart or a conjunct of the sensory counterpart of P on some phenomenalist view. 2 For any such P and S there is a statement C describing conditions of observation such that it obviously is possible that P&C be true but S false. I f P & C does not entail S, then P does not entail S. Since a physical statement must entail the conjuncts of its sensory counterpart, the original phenomenalist view is mistaken. For example, let P be 'This is a doorknob' and S be 'If one should have the experiences of reaching before one, one would have the experiences of touching something doorknob-like. '3 That P does not entail S can be seen by considering C, 'The observor's bodily coordination is abnormal.' P&C obviously does not entail S, so P does not entail S. If a condition stating that there are no sensory indications of abnormal coordination is added to S, other conditions can be described to show that P does not entail this new statement. One way to meet the APR is to hold that, despite appearances, P&C entails S. Roderick Firth takes this approach.~On his view one sentence such as 'This is a doorknob' can be used to make several logically independent statements P~, P2, P3, etc., the p-family of statements. For each member of the P-family there is a sensory statement having the same proposition component. P~ may entail S~ and $3, P2 entail S~ and $2, and P3 entail $2, $3, and $4. It is useful to have one sentence with which these evidentially linked statements can be made. Firth claims that the APR exploits the ambiguity. Suppose one uses the sentence 'This is a doorknob' to make the statement P2, which entails S~ and $2. If asked whether this statement might be true in condition C where S~ is false, the phenomenalist is likely to respond affirmatively. His contradiction is only apparent. For he probably wishes to retain other statements 342

PERCEPTUAL RELATIVITY

made with the sentence 'This is a doorknob.' This is reflected in his saying, 'Although $1 is false, this is a doorknob.' Firth's view h-as the correct form for a response to the APR. It denies the basic contention of the APR while accounting for the linguistic data supporting the contention. But many philosophers find the required pervasive ambiguity of physical discourse an unacceptable consequence for a theory of meaning. Lewis concedes that physical statements do not entail sensory statements of the above sort. He holds that physical statements entail sensory statements of a more complex form. P ('This is a doorknob') entails Sp ('If one should have the experience of reaching before one, in all probability one would have the experience of touching something doorknob-like'), The idea is that while P&C does not entail S, it entails Sp since the conditions C describes are rare. But so long as probability-qualified sensory statements are interpreted as contingent statements, the APR will apply to Lewis's view. Let C' be the statement 'Whenever anyone reaches before one, his coordination becomes abnormal.' It is clear that P&C' can be true and Sp false, so that P does not entail Sp.S I assume that Lewis's response to the APR involves introduction of a probability qualification within sensory statements. This accords with the usual interpretation of Lewis as a phenomenalist, and with most of his writings. Lewis's response to Chisholm's formulation of the APR suggests another reading. Probabilities are relative to the premises.., from which they are determined. And in consequence of this fact it can be - and frequently is - true that on the premise 'P' alone something, 'R', is highly probable, but on the premise 'P and Q', 'R' is highly improbable... In my account of perceptual knowledge, it is probability-consequences of objective statements which are in question. 6 (emphasis in original) Lewis here suggests that physical statements do not entail sensory statements, but are related to sensory statements by necessary truths of the f o r m ' S is probable on P.' 'S is probable on P' and 'S is not probable on P&C' are compatible. I develop the idea suggested here by Lewis. 2. Meaning and Warrant. My proposal is that certain epistemic relations of physical statements are constitutive of the meaning of the statements and of epistemic expressions. Meaning accounts of physical statements reporting just these relations are partial. Distinct 343

MARK PASTIN

physical statements could have the same epistemic relations of the specified kind. Such meaning accounts can, however, be "epistemicatly complete." For a meaning account to be "epistemically complete" for a statement is for the account, along with meaning accounts for other statements, to determine the conditions under which a person is warranted in believing the (proposition component of the) statement. I characterize epistemic completeness more precisely in Section 3 and urge that meaning accounts should be epistemically complete in Section 4. One sort of meaning account, the sort proposed by phenomenalists and other reductionists, provides an alternative way of making the statements that are the objects of the account. They provide statements having the same proposition components as the oject statements, but systematically different sentence components. The point of such accounts may be to enhance understanding of the object statements (their proposition components are the same as those of better understood statements), to show that the object statements do not carry commitments (the apparent commitment is a function of the sentence component), or simply to reveal the expressive equivalence of two sorts of sentences. Other meaning accounts provide statements which entail or are entailed by the object statements, but do not yield equivalences. Meaning accounts of the sort I propose do neither of these things. The proposed accounts reveal "conceptual connections" between physical and sensory statements (more exactly, between their proposition components). If we were constructing a language, these accounts could be said to state "meaning postulates." Since we are interested in statements in a natural language, I speak of "meaning constituents." Meaning constituents describe the most fundamental relations among propositions, relations constituting the logic of the propositions in a sense of 'logic' not restricted to entailments. I hope to clarify the specific "conceptual connections" at issue in my proposal, however unclear the general notion of a "conceptual connection" is. The relation instantiated in meaning constituents is warranting. An important principle of warranting is that statement P warrants statement Q if anyone warranted in believing (the proposition components of) P and Q at a given time would be warranted in believing (the proposition component of) Q at that time even if he had no evidence but that necessary for (the proposition component of) P to be warranted for him. 7 Briefly, P warrants Q if anyone warranted in believing P at a given time is, ceteris paribus, warranted in believing Q at that time. Warranting is analogous to a probability relation in 344

PERCEPTUAL RELATIVITY

that P may warrant Q although P&R does not warrant Q. Understanding of the notion of warranting obviously depends on understanding of the presupposed notion of warrant or epistemic justification. I can not explore this notion here.8 The notion of warranting is at least partly explicated by the meaning constituents into which it enters. The elements of meaning constituents are simple sensory statements (abbreviated '$1 ,' '$2 ,' etc.), connective sensory statements (abbreviated '$1>$4' and 'S2&S4>S~,' etc.), and physical statements (abbreviated 'PI ,' 'P2 ,' etc.) Simple sensory statements attribute a sensory experience of a certain phenomenal character to a person at a time. While the deficiencies of natural languages for making such statements are lamented by some and dwelled on by others, I think the sentences 'Tom now feels pain,' 'It looks to me as if I see something red,' and 'Yesterday I tasted something bitter' can be used in some contexts to make simple sensory statements. 9 I shall not discuss qualms reasonably felt about the existence of sensory statements since my proposal is relevant to views construing "basic perceptions" differently. (See Section 5.) Connective sensory statements are conditionals. The antecedents and consequents of their sentence components are subjunctive counterparts of sentence components of simple sensory statements. The sentence 'If one should have the experiences of reaching before one, one would have the experiences of touching something doorknob-like' can be used h~ some contexts to make a connective sensory statement. The type of conditional involved is the same as that of such independent-offactual statements as 'If someone should strike this match, it would ignite.' Physical statements are contingent, non-sensory statements made with a physical vocabulary. Interpret 'physical vocabulary' broadly. This can make explicating the meaning of physical statements no easier. A meaning account M for physical statement P includes meaning constituents in which P or not-P are warranters and others in which P or not-P are warrantees. M includes constituents of the fomas 'P warrants S~ >$3 ,' 'P warrants not-(S6>Ss ),' and 'not-P warrants S~>$9.' If P is 'This is a doorknob,' St>S3 might be 'If one should have the experiences of reaching before one, one would have the experiences of touching something doorknob-like.' The sensory statements warranted by P, P's "warrant consequences," may be about past, present, or future experiences. M may also include constituents stating that P or not-P conjoined with sensory statements warrant other sensory statements. I doubt that most physical state345

MARK PASTIN

ments by themselves warrant simple sensory statements. The experiences one has in the presence of a physical object usually are a function not only of the condition of the object and perceiver, but also of the actions undertaken by the perceiver. Constituents stating that P and not-P warrant sensory statements entail that certain sensory statements warrant P or not-P. If P warrants $1 >$3, SI & not-S3 warrants not-P. M also includes constituents explicitly asserting that sensory statements warrant P or notP, constituents of the forms 'S~>SI~ warrants P' and '$4>$2 warrants not-P.' Simple sensory statements may have a greater role warranting physical statements than as warrant consequences of physical statements. If P is 'This is red' and $7 is 'It looks to one as if one is seeing something red,' M may include '$7 warrants P.' I think meaning accounts of physical statements include enough meaning constituents of these forms to be epistemically complete. I underscore a peculiarity of the constituents considered here. They relate P or not-P, possibly in the context of sensory statements, but never in the context of other physical statements, to sensory statements. This simplification is not essential to my proposal or the main issues on which it bears. (See Section 4.) 3. Is Completeness. 1 said that for a meaning account to be epistemically complete (e-complete) for a statement is for the account, along with meaning accounts for other statements, to determine the conditions under which a person is warranted in believing the statement. This notion can be made more precise. In assessing the e-completeness of a meaning account for statement P we focus upon warranting statements involving (the sentence component of) P, statements of the forms ' . . . P . . . warrants Q' or 'Q warrants 9 P . . . ' . Meaning account M is e-complete for statement P if every self-consistent warranting statement Ir involving P is logically entailed by a consistent set of elements of M and other statements not containing non-logical constituents of P. The e-completeness of M for P depends not only on what M says about P, but also on what other meaning accounts say about other statements in P's "semantic field."~~ This distinguishes the e-completeness aspired to by meaning accounts of the proposed kind from that aspired to by verificationist accounts9 This account of e-completeness requires clarification. Statement P contains non-logical constituents of statement Q if the sentence component of P contains non-logical constituents of the sentence component of Q. Thus criteria for identifying constituents of sentences and for distinguishing logical and non-logical constituents are 346

PERCEPTUAL RELATIVITY

needed. The account is oversimple. It does not note strength or mode of warranting, and does not explicitly require that a meaning account determine the conditions under which a person is categorically warranted in believing a statement. While these are serious shortcomings, they do not preclude use of the notion of e-completeness in assessing the proposed meaning accounts. In application, identification and classification of sentence constituents is not problematic. And there is no reason to think that a meaning account e-complete in the present sense is not e-complete by a more rigorous standard. The warranting relations of a fion-basic statement determine whether a person is warranted in believing ~he statement in various circumstances. I can not demonstrate that physical statements have e-complete meaning accounts of the proposed kind. To do so would require that the accounts be fully presented. I argue in Section 4 that, whether or not statements have the proposed sort of meaning accounts, these accounts can not be presented in the format outlined above. I shall assume that physical statements have the proposed sort of meaning accounts to indicate how such accounts might satisfy the conditions for e-completeness. The main thing to show in establishing the ecompleteness of M for P is that P's warranting relations to other physical statements are derivable in the required way. The strongest doubt about M's ability to account for P's warranting relations to sensory statements presupposes that these relations are mediated by warranting relations to physical statements. H Consider two examples of how M might account for P's warranting relations to physical statements. Physical statement Q, 'On each of many occasions there has been a desk here,' warrants physical statement P, 'There is now a desk here.' For M to be e-complete for P 'Q warrants P' must follow from elements of M and statements not containing non-logical constituents of P. The present case may be unproblematic since 'Q warrants P' is an instance of an inductive principle, a statement not containing non-logical constituents of P. In other cases P's warranting relations may not be easily subsumed by an inductive principle. I examine the "deeper structure" of this case to show how more complex cases are to be treated. On my proposal there will be meaning constituents affirming that Q has various warrant consequences. Some consequences will be like those of 'There is a desk here at (time) t.' E.g., Q may warrant 'If one should have the experiences of being here and looking before one, one would have the experiences of seeing an extended, fiat 347

MARK PASTIN

surface.' Q will warrant sensory statements St>S2, $4>$6, Ss>Ss, and so on, such that "$1>$2 warrants P' belongs to M, '(SI>S2)& (S3>Ss) warrants P' belongs to M, or Sa>S~ warrants $8 and 'Ss warrants P' belongs to M. This does not ensure that Q warrants P. 12 What is crucial is that the set of Q's sensory warrant consequences warrants P. If this set of Q's meaning-determined consequences warrants P, Q warrants P. So 'Q warrants P' follows from (1) meaning constituents stating that Q warrants sensory statements, (2) meaning constituents stating that sensory statements warrant P, and (3) general warranting principles. 13 Only the statements in (2) need be elements of M. The statements in (1) will be elements of a meaning account for Q. The warranting principles of (3) apply to both sensory and physical statements and decide such questions as: Does set of statements A warrant statement B given that some members of A warrant B, some warrant not-B, and there are warranting relations among members of A? The strategy is to take sensory statements as the epistemic medium of exchange between physical statements. Consider another example. Suppose that P&Q,'X is red and X is in daylight,' warrants R, 'X will appear red if observed.' It probably is the case that P&Q must be supplemented (by evidence concerning the way red things appear in daylight) to warrant R. I ignore this complication. The meaning constituents of P and Q state that they warrant sensory statements. These consequences of P and Q determine a set of sensory statements as warrant consequences P&Q. This set may not include all sensory statements warranted by P or Q and may include statements warranted by neither. The key point is that this epistemic intersection of P and Q warrants R. This set includes statements of the forms 'If one should have the experiences of looking at this in daylight, this probably would appear red' and "Visual experience of the surroundings would indicate the presence of daylight. 't4 "P&Q warrants R' will then follow from meaning constituents for P, Q, R and possibly other physical statements and general warranting principles. These considerations show how rich meaning accounts of the proposed sort might, within traditional empiricist means, satisfy the conditions for e-completeness. That physical statements have such meaning constituents is consistent with empiricism. A constituent stating that P warants S is true if the proposition component of P warrants the proposition component of S. Some proposition warrants the proposition component of S if S is self-consistent. So the truth of 'P warrants S' is a matter of what the sentence component 348

PERCEPTUAL RELATIVITY

of P expresses, a matter of meaning. More on this in Section 4. The required warranting principles are within empiricist means. Principles for weighing warranting factors are needed to determine warranting conditions of physical statements on any view. 4. Assessment. I shall consider some of the main objections to my proposal. It may be objected that even if physical statements had the proposed sort of meaning accounts, this would not support traditional empiricism. Some epistemologists try to bridge the gap between sensory and physical statements by positing "special warranting principles" beyond the principles of inductive generalization accepted by empiricists. These epistemologists agree with traditional empiricism that knowledge of physical reality is based on present and recalled sensory experiences. They posit special principles because they hold that physical statements can not be expressed as sensory statements and that consequently inductive generalizations from sensory statements can not warrant physical statements. The posited principles axe special in that they are employed to bridge the sensory/physical gap but not at higher levels of a theory of knowledge. Examples of such principles are Locke's Principle of Resemblance between Effect and Cause, Price's Principle of Confirmability, and Chisholm's Principle of Perceptual Acceptance. Is There are objections specific to particular principles. More important, the local application of the principles suggests ad hoc-ness. This is enhanced by the fact that often the only rationale for these principles is that they by-pass skepticism. While empiricists generally are not skeptics, they do not accept principles solely because they by-pass skepticism. The objection is that my proposal sneaks in, none too stealthily, special principles in the guise of meaning constituents. The objection is not to principles for weighing warranting factors. These principles apply at all levels of a theory of knowledge, and do not, by themselves, sustain inferences from sensory to physical statements. While the meaning constituents considered here differ from actually proposed special principles, there could be a view endorsing special principles to the same effect as these constituents. The essential difference between the proposed view and special principles views is not in the constituents/principles, but in the role of the constituents in a comprehensive theory of knowledge and meaning. This difference is reflected in the sort of justification appropriate to special principles views and to the proposed view. Special principles are justified as alternatives to skepticism. Assuming that physical statements are not expressible as sensory statements and that consequently inductive generalizations from sensory statements can not 349

MARK PASTIN

warrant physical statements, it is argued that special principles provide the only warranting route from sensory to physical statements. Avoiding skepticism is not a sufficient justification for meaning constituents. Meaning constituents must be tested by methods generally appropriate for testing meaning accounts. These methods are not independent of epistemological considerations. A theory of meaning not compatible with an adequate theory of knowledge is physically unacceptable. The connection between theories of knowledge and meaning is closer in the present case since meaning constituents affirm warranting relations. Truths of meaning concerning warrant are, or follow from, basic principles of warrant, principles constituting a theory of warrant. Thus my proposal must be justified as part of a comprehensive theory of knowledge and meaning. The justification of the proposal depends on how well the theory accounts for a broad range of data concerning knowledge, warrant, and physical statements, intuitions about skepticism constituting but one item. Skepticism can be evaluated only as part of a theory of knowledge and meaning resulting from investigation of the concepts of knowledge, warrant, and physical reality. It can not be precluded at the outset of investigation of these concepts that the investigation will lead to unanticipated outcomes, e.g., skepticism. 16 My view differs from the view recently defined by Pollock on the issue of special principles. 17 Pollock's view, like mine, incorporates warranting conditions for a statement into its meaning. But Pollock incorporates whatever principles are needed to avoid skepticism into the meanings of physical statements. This requires commitment to the falsity of skepticism prior to epistemological investigation. If the present view avoids skepticism, this may motivate investigation of it, and given certain assumptions about languages and conceptual schemes, lend plausibility to it. But my view does not presuppose a prior commitment to the falsity of skepticism. To summarize: My proposal is an installment of a theory of knowledge and meaning. A significant part of that theory is the traditional empiricist account of warrant sketched earlier. An important corollary of the embededness of meaning constituents is that identification of the proposition component of a statement and determination of the warranting relations of the statement are reciprocal tasks. Special principles views are offered as alternatives to skepticism. Skepticism is judged false either via an epistemological theory or pre-theoretically. In the former case, special principles are ultimately justified by the principles against which skepticism is judged false. This seems not to be the intent of special principles 350

PERCEPTUAL RELATIVITY

views, which take the falsity of skepticism as a "prior truth." In the latter case, the result of an investigation is dogmatically prejudged or, at best, judged on the basis of insufficient data. A theory capable of deciding the truth of skepticism has broad implications concerning knowledge and meaning which must be weighed against appropriate data. It may be objected that, anti-skeptical considerations aside, there is little reason to accept my proposal. There is indirect support for the proposal. Philosophers of diverse views, e.g., Lewis (sense meaning), Quine (stimulus meaning), and Sellars (language-entry transitions), hold that physical sentences must be semantically tied to experience to be applied in sensuously presented situations. This tie has not been adequately characterized. Lewis's attempt to secure the tie via phenomenalism runs into the APR. Behavioristic accounts fail to explain how the experiences which guide use of physical sentences pertain to the warrant of statements made with them. I think the proposed constituents more adequately characterize this tie. Supposing that physical statements have some of the proposed constituents, is there reason to think they have a supply sufficient for e-completeness? To deny that physical statements have such a supply of constituents is, within an empiricist framework, to allow that it may be impossible in principle to determine whether a person is warranted in believing a physical statement at a given time. The point is not that if physical statements do not have a sufficient supply of meaning constituents, an empiricist will be forced to skepticism. Skepticism and anti-skepticism are equally unsupportable if physical statements do not have e-complete meaning accounts. My proposal is a minimal venture for traditional empiricism concerning the meaning of physical statements. It may be objected that the proposed meaning accounts are too "subjectivistic." I emphasize that while the proposed accounts are to be e-complete, they do not claim to be complete meaning accounts. This is another difference between my account and Pollock's. Pollock holds that meaning accounts of physical statements describing their warranting relations are complete. Even the proposed partial accounts are realistic in one sense. To be warranted in believing a physical statement is to be warranted in believing connective sensory statements affirming "factual connections" among not just our own experiences. Since connective statements may be true even if the described experiences do not occur, one must be warranted in believing something true independent of the existence of minds. Some will find this realism too weak. It does not specify that physical objects exist. More on this below. 351

MARK PASTIN

It may be objected that the proposed view not only presupposes a meaning-theory distinction. It presupposes the "museum myth," that each statement has its own parcel of meaning. I am committed to talk about what sentences express and the objects of the propositional attitudes and to whatever, if anything, is ontically presupposed by such talk. The "museum myth" charge is less clear. The meaning constituents considered are for physical statements having logically simple sentence components. This restriction on the form of constituents is not essential to the basic proposal that physical statements have e-complete meaning accounts based on meaningconstitutive warranting relations to sensory statements. On one alternative the minimal units of physical discourse having meaning-constitutive warranting relations are statements having complex sentence components expressing a theory, part of a theory, or a belief system. Let T be such a unit. My proposal requires that there be an e-complete meaning account for T based on T's meaning-constitutive warranting relations to sensory statements. A more plausible alternative is that physical statements with logically simple sentence components have meaning-constitutive warranting relations, but e-complete meaning accounts can not be based on just these relations. E-complete accounts must include constituents for physical statements having complex sentence components, possibly expressing a theory, part of a theory, or a belief system. This view also comports with my proposal. It may be objected that since meaning accounts of the proposed sort deal with whole statements, providing such accounts for all statements makeable in a language would be an interminable task. I concede this. I do not propose providing meaning accounts for physical statements by enumerating constituents of the forms considered. I propose that meaning accounts for physical statements logically entail such constituents. To systematically provide meaning accounts for physical statements one must provide meaning accounts for constituents of their sentence components such that the proposition expressed by the sentence is a function of the meaning of these constituents. Many philosophers hold that dividing sentence components of statements into meaning-functional parts is closely related to determining the constituents of the proposition component and the ontological import of the statement. The connection has not been made clear. My proposal is not based on a view of the ontological import of physical statements. I think that phenomenalist, standard realist, and other accounts of what the sentence components of physical 352

PERCEPTUAL RELATIVITY

statements express are compatible with physical statements having e-complete meaning accounts of the proposed sort. Something about the ontological import of physical statements (beyond the modest realist implication noted above) or the structure of their proposition components may follow from their having such e-complete meaning accounts. But it is not obvious that this is the case, or, if it is, that what follows decides any significant ontological issue. If statements having phenomenalist, standard realist, or other proposition components have the same meaning-constitutive warranting relations, the differences among these views are not reflected in empirical warranting conditions of statements. The conclusion that differences between such views as phenomenalism and standard realism are not empirically reflected is not novel. On one reading Kant holds that in so far as we have empirical knowledge, that knowledge is inductively based on sensory experiences. On this basis we know physical reality and dispositions of physical reality as sensuously indicated. Judgements that there is something beyond what is sensuously indicated can not be empirically supported or refuted. Lewis sometimes claims that his views about the warranting conditions of physical statements do not commit him to phenomenalism or any ontological view. For Lewis one important type of meaning is sense meaning. He sometimes claims that his views about the relations between physical and sensory statements concern only sense meanings. 18 This accords with my proposal if we take meaning con. stituents to specify a type of meaning. Lewis seems not to take this line seriously, instead seeking a logical equivalence between physical and sensory statements. 5. Conclusion. The proposed view shows that the idea that traditional empiricism is incompatible with perceptual relativity or corporate theories of meaning is incorrect. The key feature of the proposed view, that it takes certain warranting relations of physical statements to be meaning-constitutive, can be exploited by epistemologists who reject some elements of traditional empiricism. Epistemologists who reject the traditional empiricist identification of foundational statements may hold that meaning-constitutive warranting relations between foundational and non-foundational statements play an important role in explaining warranting conditions of non-foundational statements. The argument from perceptual rela-

353

MARK PASTIN tivity and the c o r p o r a t e view o f m e a n i n g , r a t h e r t h a n r e f u t i n g empiricism, s h o w h o w it m a y be d e v e l o p e d in new d i r e c t i o n s , t9 INDIANA UNIVERSITY BLOOMINGTON, INDIANA 47401 U.S.A.

NOTES 1

W.V. Quine, "Two Dogmas of Empiricism," reprinted in From a Logical

Point o f View (New York, 1953), p. 41. 2

3

4 s 6 7 s

9

354

I use italicized capital letters to designate statements and single quotation marks to form names of both sentences and statements (by placing them around the sentence component of the statement). P entails Q iff it is necessarily the case that if the proposition component of P is true, then the proposition component of Q is true. P logically entails Q i f f P entails Q and that P entails Q can be demonstrated by logical operations on the sentence components of P and Q. The distinction between entailment and logical entailment is important later, especially in connection with epistemic completeness (Section 3). The phrases 'of reaching before one' and 'of touching something doorknob-like' must be taken to phenomenologicaily characterize kinds of experiences. Presumably these kinds of experiences can be characterized in ways which avoid apparent reference to physical objects. I sacrifice pure formulation for brevity. "Radical Empiricism and Perceptual Relativity," The Philosophical Review, 59 (1950), 164-183 and 319-331. There is a detailed discussion of Lewis's view in Firth's "Radical Empiricism and Perceptual Relativity," op. cit., 178-183. "'Professor Chisholm and Empiricism," The Journal o f Philosophy, 34 (1948), 522-523. In what follows 1 delete references to proposition components of statements in discussing objects of propositional attitudes. I examine the sort of counterfactual involved in the notion of warranting in "Counterfactuals in Epistemology," forthcoming in Synthese. 1 provide an account of the presupposed notion of warrant in "A Decision Procedure for Epistemology?," forthcoming in Philosophical Studies. I offer a "descriptive account of warrant stating that one is warranted in believing something if one would believe it in certain ideal empirical circumstances. The need to refer to contexts indicates the point of including a context variable in the notion of a statement.

PERCEPTUAL RELATIVITY 10

~

12

13

14 Is

16

17

la t9

The "semantic field" of a statement is the set of statements with which the original statement has meaning-constitutive relatioons. This notion requires precise characterization. I borrow the jargon from Keith Lehrer, Knowledge (Oxford, 1974), p. 115 who attr~utes it to John Lyons, Structural Semantics (Oxford, 1963), Chapter 4. I do not preclude the possibility that M includes constituents relating P to other physical statements. It is P's warranting relations to physical statements that are not directly meaning-determined in this way that must be accounted for. A may warrant B and B warrant C while A does not warrant s,C. Suppose P is evidence, like the evidence most of us have, warranting the statement N, there are exactly nine planets. R is evidence, of the sort ! am told astronomers actually have, about a bump in Pluto's orbit. P&R warrants the statement T, there are exactly ten planets. P&R warrants P, P warrants N but P&R does not warrant N. Roderick Chisholm called my attention to this case. Constituents of the forms 'S~ warrants R ' and 'R warrants S 9>S~ol where $9 >$1o warrants P, may also come into play. Thus warranting relations between sensory statements mediated by constituents for physical statements may have a role in establishing e-completeness. Physical terms must be eliminated from the sentence components of these statements. See footnote 3. One objection to the Price and Chisholm principles is that they fail to give the content of sensory experience a warranting role. See Roderick Firth "Chisholm and the Ethics of Belief," The Philosophical Review, 68 (1959), 504-506, Herbert Heidelberger, "Chisholm's Epistemic Principles," Nous, 1 (1969), 81-82, and Roderick Chisholm, Theory of Knowledge (Englewood Cliffs, New Jersey, 1966), pp. 11-14 and 24-37. 1 have profited from discussion of methodological issues with Hector-Neri Castafieda and from Chapter 3 of his Thinking and Doing (Dortrecht, 1975). John Pollock, Knowledge and Justification (Princeton, 1974). An Analysis of Knowledge and Valuation (La Salle, Illinois, 1946), p. 201. An earlier version of this paper appeared as "Meaning and Perception" in The Journal o f Philosophy, 73 (1976), pp. 571-585. Criticisms of earlier drafts presented at Wayne State, Ohio State, and Indiana Universities have been helpful. Roderick Firth, Romane Clark, and Larry Powers have provided useful comments. ! thank the National Endowment for the Humanities for support.

355