“An Idea Can Be Like Nothing But An Idea”—Another Try 1. Introduction The fundamental premise of Berkeley’s idealism is that the things we perceive by our senses are only ideas. But Berkeley knew that even philosophers who accept this premise would propose ways to avoid the idealistic conclusions that he drew from it. Chief among those ways was the representational theory of perception, commonly attributed to Locke and to Descartes, according to which ideas are representations of material things. The ideas represent the things, it was held, chiefly by virtue of resembling the things’ primary qualities. If this theory is right, then idealism is false. Accordingly, in Principles I sec. 8 and in a corresponding passage in Three Dialogues Between Hylas and Philonous, Berkeley attempts to demolish the representational theory of perception, by invoking what has come to be called his “Likeness Principle” (hereafter referred to as “LP”). The purpose of this paper is to examine this part of his attack on the representational theory. In 1965, David Armstrong described Berkeley’s argument in section PHK sec. 8 as “extremely ingenious, probably valid, but seldom considered.”i There have in fact been several discussions of LP since 1965. In 1966, Phillip Cummins published a classic paper “Berkeley’s Likeness Principle,” that attempts to explicate LP in terms of three metaphysical principles.ii George Pitcher argued in his 1977 book, Berkeley, that Berkeley has no cogent grounds for LP.iii I offered a reconstruction and a critique of Berkeley’s argument for LP in a 1985 paper, “An Idea Can Be Like Nothing But An Idea,” that also criticized both Pitcher’s and Cummins’s treatments.iv In his 1989 book, Berkeley: An Interpretation, Kenneth Winkler criticized Cummins’s account in much the 2 same way as I, and offered a different account of Berkeley’s argument for LP.v More recently (2006), Todd Ryan has published a paper in this journal, “A New Account of Berkeley’s Likeness Principle,” that criticizes my account as well as Cummins’s and Winkler’s, and that proposes yet another analysis of Berkeley’s case for LP.vi I shall not review this entire debate in this paper, for it seems to me that Ryan does a fine job of reviewing and supplementing my and Winkler’s critiques of Cummins’s account as well as of criticizing Winkler’s positive account. Rather, I shall present an improved version of my 1985 reconstruction and assessment of Berkeley’s argument. In section 2, I analyze the structure of PHK sec. 8 and evaluate what I take to be one of the two objections to representationalism that Berkeley offers in it. In section 3, I review my original reconstruction of Berkeley’s argument for LP, expound Ryan’s criticism of it, and present a revised version that avoids the difficulty he raises. In section 4, I offer a brief critique of Ryan’s positive account. In section 5, I show that the argument for LP is ineffective against any version of representationalism that distinguishes between immediate perception and perception tout court. 2. Berkeley’s two-pronged rebuttal of representionalism in PHK I sec. 8 Section 8 has two parts, which provide two different, complementary rebuttals of representationalism, and which I shall call “A” and “B” for convenience,vii A But, say you, though the ideas themselves do not exist without the mind, yet there may be things like them, whereof they are copies or resemblances, which things exist without the mind in an unthinking substance. I answer, an idea can be like nothing 3 but an idea; a colour or figure can be like nothing but another colour or figure. If we look but never so little into our thoughts, we shall find it impossible for us to conceive a likeness except only between our ideas. B Again, I ask whether those supposed originals or external things, of which our ideas are the pictures or representations, be themselves perceivable or no? If they are, then they are ideas and we have gained our point; but if you say they are not, I appeal to any one whether it be sense to assert a colour is like something which is invisible; hard or soft, like something which is intangible; and so of the rest. The key claim of part A is that an idea can resemble only an idea; the key claim of part B is that something perceivable cannot resemble something unperceivable. In the First Dialogue, Berkeley seems to amalgamate A and B, for he has Philonous say: But how can that which is sensible be like that which is insensible? Can a real thing in itself invisible be like a color; or a real thing which is not audible, be like a sound? In a word, can anything be like a sensation or idea, but another sensation or idea? (W2: 206) Here Philonous’ first two rhetorical questions assert B and his third rhetorical question asserts A, but his use of “in a word” runs the first two questions together with the third.viii Still, A and B are distinct points, so I shall discuss them separately. 4 I start with B, because it calls only for a brief discussion. B is plainly directed against any version of representationalism that holds that material things are unperceivable and that only ideas are perceivable. Locke sometimes seems to embrace such a view, and it was a commonplace throughout the modern period.ix It seems to me that insofar as Berkeley’s target is this version of representationalism, he is entirely successful in demolishing it. Something that is perceivable cannot resemble something that is unperceivable, at least if we take “unperceivable” to mean “unperceivable in principle” (unperceivable even with the aid of the most powerful instruments that could ever exist), as opposed to “unperceivable in fact” (unperceivable even with the aid of the most powerful instruments currently available). However, a representationalist need not accept Berkeley’s premise that only ideas are perceivable in this unqualified form; instead, she can hold that only ideas are immediately perceivable. Whether this qualified version of the premise can lead to the conclusion that only ideas are perceivable tout court, and hence to the further conclusions that material things are unperceivable and so cannot resemble ideas, depends crucially on what “immediately perceivable” means, and more will be said about the notion of immediate perception in section 4. But for the moment, it suffices to say that from the premise that only ideas are immediately perceivable, it does not straightforwardly follow that material things are not perceivable, nor therefore, that they cannot resemble ideas. So Berkeley’s B, while very ingenious and correct so far as it goes, does not by itself refute any version of representationalism that distinguishes between perception tout court and immediate perception. 5 Are matters any different when it comes to passage A? Here we should first note that if A is correct, then it demolishes virtually all, if not all, versions of representationalism. For virtually all, if not all, representationalist theories hold that ideas represent material things by virtue of resembling (at least the primary qualities of) those things. But passage A contains the following argument: An idea can be like nothing but an idea. ___________________________________ An idea cannot resemble a material thing. The premise of this argument is of course just Berkeley’s way of expressing LP. The importance of LP for Berkeley’s idealism is nicely captured in a remark by E. B. Allaire: Berkeley’s contemporaries … would not have been thrown off balance by Berkeley’s claim that sensible objects are composed of ideas and thus cannot exist unperceived; but they would have been deeply upset by the so-called likeness principle that Berkeley put forth in Sect. 8; for it is that principle that undercuts representative realism. The principle is thus an essential part of Berkeley’s argument for idealism.x Yet, passage A gives only the barest indication of why Berkeley thinks LP is true; it seems to treat LP as virtually self-evident or axiomatic. Thus, one may be tempted to think that LP rests simply on the idea that something mental cannot resemble something non-mental. As Phillip Cummins points out, however, this would not do justice to Berkeley’s thought: 6 It is not being claimed that something mental cannot resemble something non-mental, where “mental” is left vague, as “the only thing like an idea is an idea” suggests. Rather, Berkeley’s claim is a colour cannot be like a non-colour, a shape cannot be like a nonshape and a sound cannot be like a non-sound.xi Furthermore, if the claim that something mental cannot resemble something non-mental were all there was to LP, then the principle would be open to an obvious objection, which is well put by George Pitcher: Why can’t a mental image of a piano, say, be “like”—i.e., resemble—a piano, in the perfectly straightforward way in which a painted picture of a piano is like a piano? Looked at in this way, Berkeley’s dictum that an idea can be like nothing but an idea seems as false as the claim that a picture can be like nothing but a picture.xii Pitcher’s objection shows that LP is not obviously true; it needs to be argued for. In what follows, I propose to reconstruct and to assess Berkeley’s argument for LP. 3. Berkeley’s case for the Likeness Principle Why does Berkeley hold his Likeness Principle? Like Cummins, I take my clue from the only reason that Berkeley himself gave for it in his published work: “An idea can be like nothing but an idea; a colour or figure can be like nothing but another colour or figure [my italics].” The remark after the semicolon, it seems quite clear, is the premise of an (enthymematic) argument whose conclusion is “an idea can be like nothing but an idea”—LP. Evidently, then, Berkeley thought that he could derive LP from the 7 highly plausible proposition that a color can resemble only a color, a shape only a shape, a sound only a sound, a smell only a smell, a taste only a taste. This proposition can be put as a categorical one, as follows: (0) All sensible qualities of any sort are things that can resemble only sensible qualities of the same sort. In my 1985 paper, I claimed that the argument for LP could be built upon (0). But in order to keep the argument uncluttered, I started instead from the weaker premise that (1) All sensible qualities are things that can resemble only sensible qualities. It is clear that although (1) does not entail (0), (0) entails (1). So if (1) entails LP, so does (0). The main interpretive project of my paper was to show that (1), when combined with two other premises held by Berkeley, does entail (a slightly but harmlessly restricted version of) LP. Those two other premises are: (2) All sensible qualities are ideas and (3) All ideas of sense are sensible qualities. Propositions (2) and (3) constitute what I called Berkeley’s “Ideaism”—his identification of sensible qualities with (a certain range of) ideas. This doctrine is assumed from the outset of the Principles, though it is argued for at length in the First Dialogue. It is important to note, then, that there is nothing question begging or otherwise illegitimate in using (2) and (3) in reconstructing an argument that Berkeley does not advance until section 8 of the Principles. In trying to reconstruct Berkeley’s line of thought, it is 8 legitimate to use (2) and (3), since those are premises he assumes to be true at this stage of his overall argument for idealism. A note is in order about the qualification “of sense” in premise (3). This qualification is necessary, because Berkeley would not accept an unqualified version of (3), viz. that “all ideas are sensible qualities.” For some ideas, namely those that he says are “formed by help of memory and imagination” and those “perceived by attending to the operations of the mind” (PHK I, sec. 1), are not sensible qualities for Berkeley. It might be objected that Berkeley would not accept premise (3) even with its restriction to ideas of sense, because an idea had during a hallucination or a vivid dream would be an “idea of sense” but not a “sensible quality.” The answer to this objection is that Berkeley does not classify ideas had in hallucinations and dreams as “ideas of sense.” There are many passages where he indicates that “ideas of sense” are only those regular, coherent ones (caused by God) that for him constitute physical objects; for instance he says, “the ideas of sense are allowed to have more reality in them, that is, to be more strong, orderly, and coherent than the creatures of the mind” (PHK I sec. 33; see also PHK I secs. 29, 30, 36, 90). To cite such passages is of course not to show that Berkeley’s way of distinguishing between “ideas of sense” and “creatures of the mind” is unproblematic, But the passages do show that his distinction cannot be obviated merely by pointing out that at times ideas had in hallucinations and dreams are indistinguishable from those had in perception, For this ignores Berkeley’s way of making the distinction between “appearance” and “reality,” which is in terms of an idea’s relations to other ideas. Furthermore, it assumes that objects with different statuses (i.e. Berkeley’s “sensible things” or realities and his “creatures of the mind” or appearances) cannot be 9 phenomenologically indistinguishable—an assumption on par with assuming that a wax apple cannot look exactly like a real apple. Assuming then that Berkeley would accept premises (1)-(3), the interpretive project of my 1985 essay was to derive LP from this set of premises. Let me review the reasoning. Suppose we start by formulating LP as a categorical proposition: “All ideas are things that can resemble only ideas.” It is obvious that one cannot derive this proposition from premises (1)-(3), since the term “ideas” is distributed in it but not in them. But one can derive a version of LP that is restricted to the ideas of sense, in the following way. From (1) and (3), it follows that (4) All ideas of sense are things that can resemble only sensible qualities. But from premise (2) alone, it follows that (5) All things that can resemble only sensible qualities are things that can resemble only ideas. The inference from (2) to (5) is of course valid, since it is of the same form as the inference from “all horses are animals” to “all heads of horses are heads of animals,” or from “all cows are mammals” to “all creatures that resemble only cows are creatures that resemble only mammals.” But from (4) and (5), we can deduce the conclusion that (6) All ideas of sense are things that can resemble only ideas. Admittedly (6), unlike LP itself, is restricted to the ideas of sense. But this is no objection to the reconstruction, because the ideas of sense are the only ones for which LP needs to be demonstrated. For the only other sorts of ideas that Berkeley recognizes are ideas “excited in the imagination” (whether voluntarily as in fantasizing or involuntarily 10 as in dreams or hallucinations), and ideas of the “passions and operations of the mind” (Locke’s “ideas of reflection,” which are accessed by introspection). But the former are merely copies or images of the ideas of sense (PHK I, sec. 33). So the above argument extends to them as well: if ideas of the imagination can resemble only ideas of sense and ideas of sense can resemble only ideas, then ideas of imagination can resemble only ideas. And Berkeley would have regarded it as obvious without proof that an idea had by introspection can resemble only another idea. Before turning to Todd Ryan’s objection to this reconstruction, I must note that it has a flaw. As just mentioned, Berkeley holds that the ideas of imagination are copies of ideas of sense. This entails that ideas of sense can resemble something other than sensible qualities, namely, ideas of imagination. But in that case, step (4)—“All ideas of sense are things that can resemble only sensible qualities”—is false, at least by Berkley’s lights.xiii Further, since (4) is derived from (1) and (3), at least one of those propositions must be one that he would regard as false. The amended reconstruction that I shall offer below will not have this flaw. I turn now to Ryan’s critique. Ryan argues that although the derivation of LP from premises (1)-(3) is valid and those premises are ones that Berkeley “probably accepts,” my reconstruction “fails to explain why Berkeley endorses LP.”xiv The reason why it fails, he writes, is that “Locke would certainly deny (1), since his position is precisely that it is possible for a sensible idea (say, the idea of extension) to resemble something that is not itself a sensible idea.” Now this remark, which at first glance might seem not to contradict (1), really does contradict (1), because the relation of resemblance is symmetrical, so that if a sensible idea can resemble something that is not itself a 11 sensible idea, but presumably a sensible quality instead, then a sensible quality can resemble something other than a sensible quality, namely an idea of a sensible quality— which contradicts (1)’s claim that a sensible quality can resemble only a sensible quality. Thus Ryan’s point, I take it, is that (1) simply begs the question against Locke, because Locke would say that a sensible quality certainly can resemble something that is not a sensible quality, namely an idea of that quality. If this is what Ryan means, then he is right; for Locke’s representationalism holds precisely that the ideas of certain sensible qualities--the ideas of primary qualities--do resemble those qualities. Ryan also elaborates on his objection, as follows: The problem, I think, is that Dicker begins with what is relatively non-controversial, viz. that a figure can be like nothing but a figure, but this initial plausibility is lost in the translation to (1), since the crucial dispute between Berkeley and the representationalist realist concerns whether, for example, all figures or shapes must be sensible. It is at least not obvious that this must be the case. George Pitcher puts the latter point nicely, observing that [Locke] could agree that a figure can be like nothing but another figure—and then add: But of course Lockean material objects do have figures … and so, as I have always maintained, our ideas of sense do resemble material objects with respect to figure and the other primary qualities. 12 Thus, although Dicker does provide a valid deduction of LP from a set of principles which Berkeley probably accepts, his explanation is unenlightening in so far as it offers no insight into Berkeley’s reasons for subscribing to the argument’s crucial premise.xv This elaboration digresses from Ryan’s charge that premise (1) begs the question against Locke. Furthermore, it seems to me that Berkeley has a ready and powerful response to the objection that some figures or other sensible qualities might be insensible (unperceivable). He would respond by recurring to his point, made in passage B and by Philonous in the First Dialogue, that if the figures and other primary qualities possessed by Lockean material objects are really unperceivable, then they cannot resemble ideas, because something that is (in principle) unperceivable cannot possibly resemble something that is perceivable. That is the insight on which his rebuttal in passage B of forms of representationalism that subscribe to the doctrine of the unperceivability of material things turns, and I find it hard to dispute. Let me return, then, to what I regard as Ryan’s more important objection, namely that Locke and his ilk would just reject premise (1). I concede that this objection shows that (1) is not a premise on which Berkeley can build a successful argument for LP. However, he could start instead from an amended version of (1): (1a) All sensible qualities are things that can resemble only sensible qualities and ideas of sensible qualities. This proposition captures what is plausible in Berkeley’s premise that a “colour or figure can be like nothing but another colour or figure,” and it begs no questions against Locke. Furthermore, an argument for LP that closely parallels the above reconstruction can be 13 built on (1a), as follows. Just as Berkeley would accept premise (2) above—that “all sensible qualities are ideas”—so he would surely accept the premise that (2a) All sensible qualities and ideas of sensible qualities are ideas. As we have seen, it also a fundamental the premise of his thought that (3) All ideas of sense are sensible qualities. Now from (1a) and (3), it follows that (4a) All ideas of sense are things that can resemble only sensible qualities and ideas of sensible qualities. Notice that (4a), unlike (4) (“All ideas of sense are things that can resemble only sensible qualities”), allows that ideas of sense can resemble ideas of imagination, since those are precisely the “ideas of sensible qualities” referred to in (4a). Thus, the use of (4a) corrects the flaw in (4), viz. that (4) overlooks the point that for Berkeley ideas of sense can resemble ideas of imagination. But now the reconstruction can be completed on the same model as my original one. From (2a) alone, it follows that (5a) All things that can resemble only sensible qualities and ideas of sensible qualities are things that can resemble only ideas. Finally, from (4a) and (5a), it follows as before that (6) All ideas of sense are things that can resemble only ideas. It is possible to give a structurally similar but simpler reconstruction of Berkeley’s argument for LP, by interpreting his remark that “a colour or figure can be like nothing but another colour or figure” more freely than I have done so far. The remark could be interpreted as implying that (A) What is perceivable can be like only something perceivable. 14 Such an interpretation is not far-fetched, especially in light of what Berkeley says in passage B: “I appeal to anyone whether it be sense to assert a colour is like something invisible; hard or soft, like something intangible; and of the rest.” I have said that this remark should be seen as a direct attack on the doctrine that ideas can resemble things that are unperceivable in principle. However, when interpreted in the light of (A), it also allows us to deduce that the only thing an idea can resemble is another idea. For Berkeley’s “Ideaism” can be concisely formulated as asserting that: (B) Whatever is perceivable is an idea and (C) Whatever is an idea is perceivable.xvi But LP can be straightforwardly deduced from these three premises. For from (A) and (C), it follows that (D) Whatever is an idea can be like only something perceivable. Further, from (B) alone it follows that (E) Whatever can be like only something perceivable can be like only an idea. Finally, from (D) and (E) it follows that (F) Whatever is an idea can be like only an idea which is equivalent to Berkeley’s dictum that “an idea can be like nothing but an idea.” In summary, then, LP can be derived from a very plausible premise about resemblance suggested by Berkeley’s aphoristic remark that “a colour or figure can be like nothing but another colour or figure,” conjoined with his “Ideaism”--his doctrine that sensible qualities are identical with (a certain range of) ideas. Alternatively, it can be 15 derived from an equally plausible premise about resemblance suggested by the same remark taken together with his rhetorical questions, “whether it be sense to assert a colour is like something invisible; hard or soft, like something which is intangible; and so of the rest,” conjoined with “Ideaism,” concisely formulated as the doctrine that all and only ideas are perceivable. 4. Ryan’s account of Berkeley’s argument for LP Ryan’s interpretation of Berkeley’s case for LP turns on a metaphysical doctrine that Ryan claims to find in Locke, namely the doctrine that relations are essentially minddependent. Ryan quotes a number of passages from Locke’s Essay that seem to endorse such a view, notably this one: “The nature therefore of relation, consists in the referring, or comparing two things, one to another; from which comparison, one or both comes to be denominated” (Essay II.xxv.5). Underscoring what Locke seems to be saying here, Ryan writes: “Notice that Locke’s claim is that it is part of the very nature of relations that they involve a mental act of comparison between our ideas, and thus are, in an important sense, mind-dependent entities.”xvii Ryan adds that In turning to Berkeley’s account of relations, we find that he has disappointingly little to say about them either in his published or unpublished works. However, what little he does say suggests that, at least with respect to the essential mind-dependence of relations, he is wholly in agreement with Locke. At PHK 142 Berkeley argues that it is impossible to have an idea of a relation because the latter involves an act of mind, and as Berkeley argues at length, ideas are altogether passive and inert and therefore cannot represent a thing that is active. He writes: “all relations 16 including an act of the mind, we cannot so properly be said to have an idea, but rather a notion of the relations or habitudes between things.” Ryan then says: Given this account of relations, an alternative reading of [Berkeley’s] argument suggests itself. If resemblance is taken to be a relation between two objects, and if relations essentially involve a mental act of comparison, then a necessary condition of two things resembling one another is that they either are or have been the object of a mental act of comparison. If one can compare only what one directly perceives, and the only thing one directly perceives are one’s own mind-dependent ideas, it follows that an idea cannot resemble an unperceived material object, because only ideas can stand in a relation of resemblance; or, in other words, ‘an idea can be like nothing but an idea.’xviii This reasoning can be formulated as follows: (1) Resemblance is a relation between two objects. (2) Relations can obtain only between things that are or were compared. (3) Resemblance can obtain only between things that are or were compared. (1, 2) (4) One can compare only things that are directly perceived. (5) Only ideas can be directly perceived. (6) Resemblance can obtain only between ideas. (3, 4, 5) 17 As Ryan notes, this argument would show not just that two things having been compared is “a necessary condition for knowing that the two things resemble, but that it is a necessary condition for two things resembling at all.”xix Although I have other reservations about this reconstruction, I shall confine myself to one criticism.xx This is that a representationalist need not accept (2), because she need not accept the mind-dependence of relations. Furthermore, despite the passages that Ryan quotes from Locke, and other passages in the same vein in the Essay, it is far from clear that Locke accepts such a view. There are of course metaphysical issues about the status of relations, stemming from the fact that they are neither substances nor (monadic) properties or Lockean “modes,” and Locke shows his own awareness of these issues when he says such things as that “Relation [is] not contained in the real existence of Things, but something extraneous, and superinduced” (Essay II.xxv.8). But it does not follow from this that Locke was prepared to relegate relations to the dustbin of the mind. After all, he tells us that “One single man may at once be concerned in, and sustain all these following Relations, and many more, viz. Father, Brother, Son, Grandfather, Father-in-Law, Son-in-Law, Husband, Friend, Enemy, Subject, General, Judge, Patron, Client, Professor, European, English-man, Islander, Servant, Master, Possessor, Captain, Superior, Inferior, Bigger, Older, Younger, Contemporary, Like, Unlike, etc. to an almost infinite number…” (Essay II.xxv.7). I find it difficult to believe that Locke believed that a man must await being compared by some mind with the appropriate correlative(s) before actually bearing any of the above relations. Furthermore, Locke uses phrases that imply the objectivity of relations, such as “perceives the Relation,” “overlooks … a relation,” and “[cases where] the Relation is not always so easily taken notice of” (Essay 18 II.xxv.2). Finally, he classifies both causality and identity as relations, and he surely did not believe that impulses cause motion or that the sapling is identical with the grown oak only if someone “compares” or “has compared” them. I think, then, that when Locke talks of relations in terms of making comparisons, he means to say only that we come to know of relations by comparing things, not that acts of comparison are constitutive of relations. Perhaps one reason he uses language that suggests otherwise is that it seems very difficult to give a non-circular definition of “relation” without alluding to some act of comparing. But this does not entail that the things compared do not stand in various relations independently of being compared. As for Ryan’s point that Berkeley may have held that all relations essentially involve an act of mental comparison, it can carry no weight against Locke, or more generally against representationalists, if they are not really committed to that (dubious and rather idiosyncratic) doctrine. 5. The Significance of the Likeness Principle What is the significance of Berkeley’s Likeness Principle, in light of my reconstructions of his argument for it? Has Berkeley shown that representational theories of perception are untenable? Berkeley has shown, I believe, that one type of representational theory is untenable. This is the familiar version, commonly but somewhat unfairly attributed to Locke, according to which only ideas can be perceived, and the material things that these ideas supposedly resemble are in principle unperceivable. I have already said that passage B provides a knockdown objection to this form of representationalism. I now add that Berkeley also provides a refutation of this version of representationalism in passage A, because that version of the theory is committed to premises (2a) and (3) of the first 19 reconstruction and to premises (B) and (C) of the second reconstruction. But I have shown that those premises lead easily to LP, and consequently to the conclusion that ideas cannot resemble material things. Insofar as Berkeley’s target was representationalism of this sort, his arguments in Principles I section 8 are devastating. It does not follow, however, that Berkeley provides a general refutation of representationalism; on the contrary, he quite fails to do so. For representationalists do not have to accept the “ideaistic” premises used in Berkeley’s arguments for LP; indeed, it is doubtful that Locke, though be clearly held some form of representationalism, was firmly committed to those premises. In this section, accordingly, I shall indicate how a representationalist can formulate his position in such a way as to avoid Berkeley’s refutation by the Likeness Principle. I shall not try to settle the question of whether Locke’s own version of representationalism should be equated with the one Berkeley so successfully attacks or with the one that I shall sketch, because I think that Locke is simply ambiguous on this point, so that neither answer to this question would be wholly correct. Further, the philosophically interesting question is whether any version of representationalism can withstand Berkeley’s attack in section 8. I think that the answer to that question is yes. The key to formulating a version of representationalism that does not fall prey to Berkeley’s Likeness Principle is to distinguish clearly between perception tout court and immediate perception. A representationalist is of course committed to the view that we immediately perceive only ideas. But, as previously noted, she is not thereby necessarily committed to the view that we perceive only ideas. To see this more clearly, it is necessary to determine the meaning(s) of the crucial term, “immediately perceive.” Now 20 there is one sense of this term—a sense that can be found in Berkeley and other modern philosophers as well--on which it is arguably true that we can immediately perceive only ideas. This is an epistemological sense, which I define as follows: D1: X is immediately perceivede = df X is perceived in such a way that its existence and nature can be known solely on the basis of one’s present perceptual experience.xxi Given this definition, it can be shown, by appealing to hallucinations and dreams, that the only possible objects of immediate perception are ideas. The definition, as it were, “selects” ideas as the only possible objects of immediate perception: if such objects exist at all, they can only be ideas.xxii But notice that there is no conflict at all between saying that only ideas are immediately perceivede, and that material things are perceived. Upon viewing an envelope from a certain angle, for example, I immediately perceivee (perceive in such a way that I can know its existence solely on the basis of my present visual experience) a rhomboidal, whitish patch, and I perceive (though not in such a way that I can know its existence solely on the basis of my present visual experience) an envelope. There is also another sense of “immediately perceived,” on which representationalists, at least, are committed to saying that we immediately perceive only ideas. This is what may be called the “no-intermediary sense,” and it can be defined roughly as follows: D2: X is immediately perceivedo = df X is perceived, and it is false that X would be perceived only if some item that is not identical with X and that is not a part of X were perceived.xxiii 21 Berkeley introduces this sense of the term early in the First Dialogue, and it also figures prominently in his works on vision.xxiv But notice that saying that only ideas are immediately perceivedo is perfectly compatible with saying that material things are perceived: even if it be true that I do not perceive the envelope unless I immediately perceiveo the rhomboidal patch, it does not follow that I do not perceive the envelope. Thus, it is open to the representationalist to say that I perceive both the rhomboidal patch and the envelope, though I perceive them in different ways. One might object that this falsifies the phenomenological facts, because I do not then see two objects, but only one. The representationalist may reply that admittedly, I do not see two objects separated in space; indeed, the rhomboidal object that I see is not in physical space at all. Yet, she may say, I can see the envelope only by also seeing the rhomboidal object. The representationalist will want to add that (a) the envelope is the cause of my immediately perceiving the rhomboidal patch, and that (b) the rhomboidal patch in some important way (presumably involving resemblance) represents the envelope. I am not saying that such an account of perception is correct: obviously much would have to be said to clarify (a) and especially (b), and my own preference is for an adverbial view of perception that does away with objects of immediate perception altogether. All I wish to claim is that a version of representationalism that distinguishes between perception and immediate perception eludes Berkeley’s objections in section 8. By this I do not mean that even if LP were established, it would be ineffective against such a theory: LP obviously rules out any resemblance between ideas and material things, regardless of whether only ideas are perceived, or only ideas are immediately perceived.xxv My claim, rather, is that the argument for LP becomes invalid when its premises are amended so as to accommodate 22 the perception/immediate perception distinction. So by building this distinction into her theory, the representationalist can protect the theory from refutation by LP, since the distinction deprives LP of the argumentative support that it requires. To see this, let us see what happens when we modify the premises of the reconstructed arguments for LP to accommodate the perception/immediate perception distinction. The first reconstruction would have to be replaced by the following argument: (1a') All immediately perceived representations of sensible qualities are things that can resemble only sensible qualities and ideas of sensible qualities. (2a') All immediately perceived representations of sensible qualities and ideas of sensible qualities are ideas.xxvi (3') All ideas of sense are immediately perceived representations of sensible qualities. (4') All ideas of sense are things that can resemble only sensible qualities and ideas of sensible qualities. [from (1a') and (3')] (5') All things that can resemble only immediately perceived representations of sensible qualities and ideas of sensible qualities are things that can resemble only ideas. [from (2a')] (6) All ideas of sense are things that can resemble only ideas. [from (4') and (5')] But it is obvious that the inference from (4') and (5') to (6) is invalid. The second reconstruction would have to be formulated as follows: 23 (A') Whatever is immediately perceivable can be like only something perceivable. (B') Whatever is immediately perceivable is an idea. (C') Whatever is an idea is immediately perceivable. (D) Whatever is an idea can be like only something perceivable. [from (A') and (C')] (E') Whatever can be like only something immediately perceivable can be like only an idea. [from (B')] (LP) Whatever is an idea can be like only an idea. [from (D) and (E')] But again, it is obvious that the inference from (D) and (E') to (LP) is invalid. However, Berkeley might here reply that even the more sophisticated form of representationalism succumbs at least to the second reconstruction, provided that we build into the argument a principle that he repeatedly insists upon in the Three Dialogues Between Hylas and Philonous. This is what I call his Principle of Perceptual Immediacy (PPI), which says that whatever is perceived by the senses is immediately perceived.xxvii For then, letting “perceivable” mean “perceivable by the senses,” this principle implies and is implied by: (F) Whatever is perceivable is immediately perceivable. But once (F) is added, the second reconstruction can be validly completed as follows: (G) Whatever can be like only something perceivable can be like only something immediately perceivable. [from (F)] (H) Whatever is an idea can be like only something immediately 24 perceivable. [from (D) and (G)] (LP) Whatever is an idea can be like only an idea. [from (H) and (E')] Whether this attempt to refute more sophisticated forms of representationalism is successful turns on whether, on either sense of immediate perception on which the representationalist must agree that we immediately perceive only ideas (and thus accept (B')), (PPI) is true. Those two senses are the epistemological one (“immediately perceivede”) and the no-intermediary one (“immediately perceivedo”). Suppose then that the term has its epistemological sense. Then PPI means PPIe: Whatever is perceived by the senses is perceived in such a way that its existence and nature can be known solely on the basis of a given present perceptual experience. But PPIe is false: when I look an envelope in my hands in good light, I cannot know that it exists or is an envelope solely on the basis of my present perceptual experience, since I could have the same experience in a hallucination or a dream. Yet, it does not follow that I do not then perceive the envelope by sight (that I do not see it); rather, all that follows is that I do not see it in the privileged epistemic sense of “immediately perceivede”. Thus, if “immediately perceived” has the epistemological sense, then the representationalist can safely reject premise (F) (“whatever is perceivable is immediately perceivable”). Alternatively, suppose that the term has its “no-intermediary” sense. Then PPI means PPIo: Whatever is perceived by the senses is perceived in such a way that it is false that it is perceived only if some item that is not identical with it and that is not a part of it is perceived. 25 But a careful representationalist will hold that PPIo is false: she will hold that whenever one sees an envelope, one sees a colored patch that is not identical with the envelope or any part of it, but that this is perfectly compatible with saying that one sees the envelope. Thus, if “immediately perceived” has the no-intermediary sense, then again the representationalist can safely reject premise (F). Is there any sense of the term “immediately perceived” on which PPI is true? I think that there is. In the First Dialogue, Berkeley has Hylas say, “in truth the senses perceive nothing which they do not perceive immediately: for they make no inferences” (W2 174-75). This suggests the following, psychological definition of “immediately perceived”: D3: X is immediately perceivedp = df X is perceived without (the perceiver’s) performing any (conscious) inference.xxviii I have included the word “conscious” parenthetically in the definition, because Berkeley is not concerned here with what one might call “unconscious inference”--i.e. mental processing or activity that occurs without our being aware of it. Rather, he is concerned only with explicit, conscious reasoning: the sort of process that occurs when one follows an argument step-by-step, or solves an equation, or figures out the likely consequences of a proposed course of action. I suspect that for Berkeley, the definiens of “immediately perceivedp” is extensionally equivalent to the definiens of “immediately perceivedo;” in other words, that Berkeley believes that the things that answer to D3 are all and only the things that answer to D2. His thought seems to be that since a thing is perceived mediately (= not immediately) when it is perceived “not without [= with] the intervention of others,” those other things must be items or data from which one draws an inference or 26 reasons to a conclusion that would not have been reached without the data. Thus I am inclined to agree with Kenneth Winkler’s remark that “the subsequent exchanges show clearly that both Hylas and Philonous take perception without intervention to be the same as perception without inference.”xxixStill, these two concepts are different, and shortly we will see that they are not really extensionally equivalent. If we take “immediately perceived” in this psychological sense, then PPI means: PPIp: Whatever is perceived by the senses is perceived without any (conscious) inference on the perceiver’s part. Now I think that PPIp is true. For the experience one has when one perceives something by one’s senses—when one sees, touches, hears tastes, or smells something--is quite unlike the experience of going through a process of reasoning or drawing an inference. Sense perception has a certain psychological or phenomenological “immediacy” or directness that marks it off from reasoning or inferring. Roderick M. Chisholm puts the point this way: It is sometimes said that to perceive something is to “make an inference” or to “frame a hypothesis” . . . But surely no perceiver, on opening his eyes in the morning, can be said to “infer” that he is surrounded by familiar objects . . . If we do use the words “inference” and “hypothesis” in this context, we cannot take them in the ordinary sense--in the sense in which a physician, studying symptoms, may be said to “make an inference” or “frame a hypothesis” about the disorders of his patients. Use of the psychological terms “unconscious inference” and “interpretation,” in this context, serves only to obscure the fact that perceiving is not an 27 inference, in the ordinary sense of the word “inference.”xxx The truth of PPIp, however, does not salvage the argument for LP. For now the representationalist may object there is no reason why she should accept premise (B') (“Whatever is immediately perceivable is an idea”). For why should the representationalist agree that the only things perceived without any conscious inference are ideas? Representationalism does hold, to be sure, that the only things perceived without any intermediaries—the only things immediately perceivedo—are ideas, but from this it does not follow that material objects are not immediately perceivedp, and thus not perceived by the senses. To return to my example, suppose that when I look at a rectangular envelope from a certain angle, I can see it only if I see a rhomboidal patch of color that is neither identical with nor part of the envelope, so that I do not immediately perceiveo the envelope. It does not follow that to see the envelope I must infer its existence from that of the rhomboidal patch. It is open to the representationalist to say that I immediately perveiveo the rhomboidal patch and immediately perceivep the (rectangular) envelope. Here again one might object that this falsifies the phenomenological facts since I do not then see two objects but only one. But again the representationalist may accommodate the phenomenological facts by admitting that I do not see two objects separated in space, since the rhomboidal object that I see is not in physical space at all. Yet, she may say, I can see the envelope only by also seeing this rhomboidal object, but this does not mean that I must then infer the envelope’s presence from the patch. Rather, I immediately perceivep the envelope, though I do not immediately perceiveo the envelope: contrary to what Berkeley thought, “immediately pereivedo” and “immediately perceivedp” are not extensionally equivalent. You may say 28 that these are odd ways of talking, and I would agree. My own view is that they are misleading ways of reporting the fact that I see a rectangular envelope that looks rhomboidal from a certain angle. But I can see no good reason why someone who holds a representationalist theory of perception, and who reifies appearances in the way representationalists characteristically do, should not talk in these ways. In summary, I have argued for the following assessment of Berkeley’s Likeness Principle. The argumentative support that this principle requires is available if one grants that the only objects of sense perception are ideas; therefore the Likeness Principle provides a refutation of the cruder forms of representationalism, which hold that we can perceive ideas but that material things are in principle unperceivable. But the argumentative support that the Likeness Principle requires is not available if one only grants that the only immediate objects of perception are ideas; therefore the Likeness Principle does not refute more sophisticated forms of representationalism, which hold that we immediately perceive only ideas but also perceive (though we do not immediately perceive) material things. Georges Dicker The College at Brockport State University of New York 29 Notes i David M. Armstrong (ed.), Berkeley’s Philosophical Writings (New York: Macmillan, 1965), p. 17. ii Phillip D. Cummins, “Berkeley’s Likeness Principle,” p. 65. Journal of the History of Philosophy 4 (1966), pp. 63-69. This article is reprinted in C. B. Martin and D. M. Armstrong (eds.), Locke and Berkeley: A Collection of Critical Essays (Notre Dame: University of Notre Dame Press, 1968), pp, 353-363, and in Walter Creery (ed.), George Berkeley: Critical Assessments, Volume II (London and New York: Routledge, 1991), pp. 359-366. iii George Pitcher, Berkeley (London, Henley and Boston: Routledge & Kegan Paul, 1977). iv Georges Dicker, “An Idea Can Be Like Nothing But An Idea,” History of Philosophy Quarterly 2 (1985), pp. 39-52. Reprinted in Walter Creery (ed.), George Berkeley: Critical Assessments, Volume III (London and New York: Routledge, 1991), pp. 162176. v Kenneth G. Winkler, Berkeley: An Interpretation (Oxford: Oxford University Press, 1989). vi Todd Ryan, “A New Account of Berkeley’s Likeness Principle,” British Journal for the History of Philosophy, 14:4 (November 2006), pp. 561-580. vii Here and in the rest of this chapter, “section 8” refers to section 8 of Part I of Berkeley’s Principles of Human Knowledge. For good evidence that section 8 contains two different lines of argument, see Todd Ryan, “A New Account of Berkeley’s Likeness Principle,” British Journal for the History of Philosophy, pp. 562-64. Ryan shows that the 30 two parts of section 8 are foreshadowed in two arguments Berkeley offers, and considers as distinct arguments, in his Notebooks (Philosophical Commentaries). viii For brevity’s sake, here and below I use “A” and “B” to refer to the key claims made in passages A and B. ix Passages where Locke seems to embrace this view include these: SINCE the Mind, in all its Thoughts and Reasonings, hath no other immediate Object but its own Ideas, which it alone does or can contemplate, it is evident, that our Knowledge is only conversant about them. (Essay IV.i.1) ALL our Knowledge consisting, as I have said, in the view the Mind has of its own Ideas ... (Essay IV.ii.1) How shall the Mind, when it perceives nothing but its own Ideas, know that they agree with Things themselves? (Essay IV.iv.3) Notice how, in the first passage, Locke moves from saying that only ideas are immediately perceived, to saying that only ideas are perceived: we can “contemplate” only ideas. x Edwin B. Allaire, “Berkeley’s Idealism Revisited,” p. 203. In Colin M. Turbayne (ed.), Berkeley: Critical and Interpretive Essays (Minneapolis: University of Minnesota Press, 1982), pp. 197-206. xi Phillip D. Cummins, “Berkeley’s Likeness Principle,” p. 65. Journal of the History of Philosophy 4 (1966), pp. 63-69. This article is reprinted in C. B. Martin and D. M. Armstrong (eds.), Locke and Berkeley: A Collection of Critical Essays (Notre Dame: University of Notre Dame Press, 1968), pp, 353-363, and in Walter Creery (ed.), George 31 Berkeley: Critical Assessments, Volume II (London and New York: Routledge, 1991), pp. 359-366. xii George Pitcher, Berkeley (London, Henley and Boston: Routledge & Kegan Paul, 1977), p. 116. It is not so clear that a picture of a piano is like a piano in a “perfectly straightforward” way. The picture is flat but the piano is not (at least not in the same sense of “flat”!), the piano can make music but the picture cannot, and so on. There are complex issues about the nature of representation beneath the surface here. I shall ignore such issues in what follows. xiii xiv xv This was pointed out to me by Daisy Radner. Ryan,“ A New Account of Berkeley’s Likeness Principle,” p. 565. Ryan, Ibid., p. 565. Ryan’s quotation is from Pitcher, Berkeley, p. 117. xvi In formulating these premises, I have deliberately followed Berkeley’s frequent practice of omitting the qualification, “immediately,” which, it might be said, ought to precede “perceivable.” The implications of reintroducing this qualification will be taken up in the next section. xvii Ryan, “A New Account of Berkeley’s Likeness Principle,” p. 575. xviii xix xx Ibid. Ibid., p. 576. My other reservation concerns premise (4), which, in light of what I shall say about immediate (“direct”) perception in the next section, seems to me to be untrue. xxi I introduced this definition in Perceptual Knowledge: An Analytical and Historical Study (Dordrecht: Reidel, 1980), p. 56 (though there I used “a given perceptual experience” instead of “one’s present perceptual experience”). I have discussed the 32 definition, and argued that the sense of “immediately perceived” that it captures can be found in Berkeley, in a number of articles, most recently in “Berkeley on Immediate Perception: Once More Unto the Breach,” Philosophical Quarterly 56: 225 (October 2006), pp. 517-535, especially section III. My other discussions of it can be found in “The Concept of Immediate Perception in Berkeley’s Immaterialism,” in C. M. Turbayne (ed.), Berkeley: Critical and Interpretive Essays, (Minneapolis: University of Minnesota Press: 1982), pp. 48-66, and in “Berkeley on the Immediate Perception of Objects,” in Phillip D. Cummins and Guenter Zoeller (eds.), Minds, Ideas, and Objects: Essays in the Theory of Representation in Modern Philosophy, North American Kant Society Studies in Philosophy, Volume 2, (Atascadero: Ridgeview Publishing Company, 1992), pp. 201213. xxii There is a full-scale defense of this claim in Georges Dicker, Perceptual Knowledge: An Analytical and Historical Study (Dordrecht: Reidel, 1980), Chapters 3 and 4 (though there the term “sense-data” is used rather than “ideas”). xxiii For a highly detailed and refined discussion of this kind of definition of immediate perception, see George Pappas, Berkeley’s Thought (Ithaca: Cornell University Press, 2000), pp. 148-160. xxiv For evidence that Berkeley uses the term in his sense, see Pappas, Berkeley’s Thought, pp. 148-151. xxv I am indebted to Deborah Modrak and Paul Weirich for clarification on this point. xxvi The phrase “immediately perceived representations of sensible qualities and ideas of sensible qualities” is not a mere redundancy, because “immediately perceived 33 representations of sensible qualities” can refer to ideas had in perception and “ideas of sensible qualities” can refer to ideas had in imagination and in memory. xxvii It is important to note that this principle is no mere tautology, saying that whatever is perceived by sense is perceived by sense, or that whatever is immediately perceived is immediately perceived, or the like. Rather, the subject term “perceived by sense” and the predicate term “immediately perceived” have different meanings. The former refers to sense-perception, the latter needs to be defined, and the truth of the principle depends on how it is defined. Passages where Berkeley affirms this principle in Three Dialogues Between Hylas and Philonous include at least W2 174-75, 183, 194-95, 203, 204, 205, 215. For discussion of these passages, see my “Berkeley on Immediate Perception: Once More Unto the Breach,” pp. 530-531. In that article, pp. 531-535, I also argue that Berkeley ultimately accepts a qualified version of PPI that allows for what he calls “suggestion.” Although I shall not argue the point here, I believe that substituting the qualified version of PPI into the reconstruction I am about to offer would not change the evaluation of it that I shall give. xxviii I defend view that Berkeley uses “immediately perceive” in this sense in the papers cited in note 16, xxix xxx Kenneth G. Winkler, Berkeley: An Interpretation, p. 150. Roderick M. Chisholm, Perceiving: A Philosophical Study (Ithaca: Cornell University Press, 1957), pp. 158-159.