Realistic Postcript

The realism-idealism issue has been the cross of modern philosophy. The efforts that went into it since Descartes are out of proportion. Their fruits are scant and bitter. Other issues were neglected, beclouded, distorted. With not a few the preoccupation became a block. The intellectual torment it caused to many more is appalling. Of this torment I have had my share.
Structurally, whatever they may have said, the logical positivists, my first teachers, are all either phenomenalists or materialists. None of them saw another alternative. Neither, therefore, when I began to philosophize, did I. So I chose what I thought and after a fashion still think the lesser evil, phenomenalism. Yet I was restive. Eventually I broke away, recognized that there are acts and proposed an assay of them which, in its main outlines, I still think adequate. Structurally, that made me a realist. But my ontology was cluttered with phenomenalist dross. Every distinction and every argument I have presented since then has made the realism a bit more explicit, got rid of a bit of the dross. Every distinction and every argument I have presented since then has made the realism a bit more explicit, got rid of a bit of the dross. In this essay I propose to present some further distinctions and some further arguments which should resolve some of the perplexities of classical realism to which I have as yet not attended and, at the same time, get rid of the last bit of dross.
The way to the solution of the philosophical problem leads, we know, through the explication of the philosophical uses of some key words. Once explicated, such a use is no longer philosophical. In this way we come to distinguish among several ordinary uses of a single word. The ***** ¹ This essay could not have been written without the encouragement and help I received in many discussions with Edwin Allaire. Reinhardt Grossmann also has contributed by steadily nudging me toward realism during the last few years.

302

ordinary man, though he uses the word, is not aware of the distinctions. Can he grasp them when they are pointed out to him? Obviously, he will not appreciate them, even if he can grasp them, unless he himself has suffered philosophical torment, in which case he is not the man I am talking about. The question, therefore, is not whether he can appreciate the distinctions but, rather, whether he can grasp them. With one crucial exception the answer is Yes. (The ordinary man I am talking about is of course the proverbial man on the street. But the phrase sticks in my throat.
Common sense has a soft fringe and hard core. The soft fringe is science. The hard core is what we attend to when we adopt the phenomenological attitude. If that makes me a phenomenologist and my realism phenomenological, then I accept the label, though only after having reminded the dispensers of labels that one of the longest essays in this book dissects the errors which drove the founder of the phenomenological movement to idealism.
Attending to the hard core, we sometimes find ourselves in situations which cannot be accurately described without one of the several commonsensical (ordinary) uses of a key word. An ordinary man never finds himself in a situation of this sort; or, again, he is not the man I am talking about. This one particular use he will therefore not grasp, even if it is pointed out to him. We may of course try to instruct him how to attend to the hard core. But the attempt may fail. This is the exception. It makes a difference for what the ordinary man on the one hand and the philosopher on the other can do about the meaning of a key word. That is why it is crucial.
Ask an ordinary man what a word he uses means. What will happen depends on the word. If you pick 'is', all you can reasonably expect is puzzlement. If it belongs to one of two very large classes of words, you will receive, or at least you may reasonably expect to receive, an answer. 'Horse' belongs to one of the two classes, 'green' to the other. Depending on the class to which the word you picked belongs, the answer you will receive is one of the two kinds. If the word belongs to the 'horse'- class, the answer is, however roughly, a definition. If it belongs to the 'green'-class, all you will get and all you can reasonably expect is examples.²
If a word has several ordinary uses, one may be a 'green'-use; the others, 'horse'-uses. A use of a key word may be recognizable as a

***** ² 'Green' is a color' is neither a definition nor mentions examples. Yet it is a likely answer. Oxford philosophers may think that this is an objection. I leave it to them to resolve or rejoice in.

303

'horse'-use if and only if one knows how to attend to the hard core. That is the difference between what an ordinary man and we can do about the meaning of a key word. It is crucial because there are cases where one cannot spot an insidiously elusive philosophical use without first having discovered and distinguished from others that 'green'-use we make of it only when attempting to describe accurately some feature of the hard core. An example will help.
For an ordinary man 'mental' and 'nonmental' are 'green'-words. If you ask him what they mean, what you get and all you can reasonably expect to get is examples. He will tell you that the tree in front of him is nonmental while his seeing it is mental; that yesterday's sunrise is nonmental while your remembering it now is mental; and so on. Ask him, then, whether the dragon which at this very moment you imagine is mental or nonmental. Or ask him the same question about the spot Lady Macbeth cursed. You may or may not give him a moment's pause. In either case you will be told that the dragon and the spot are both mental. (The actual words may be that they are both merely in the head.) If there is that pause, you are catching the moment of transition from an ordinary to an insidiously elusive philosophical use of 'mental' and 'nonmental'. This particular philosophical use is at the root of some of the perplexities in which the classical realists bogged down. To ferret it out, one must recognize it as a 'horse'-use, which in turn cannot be done without first anchoring the 'green'-use of the two words in the hard core. But I am getting ahead of my story.
What exists? What reason do we have to believe of anything that it exists? The first question leads into ontology; the second, into epistemology. Ontology is primary; epistemology is but the ontology of the knowing situation. That is the gist of my ontologism. Epistemologism reverses the order of precedence. The point has been made more than once in this book. The reason for making it again is that the three hundred years of anguished preoccupation with the realism-idealism issue coincide with the reign of epistemologism. Why this is so you will, if I am right, understand completely after you have read this essay. But it may help to read it if I now try to state the connection in one paragraph.
The things we perceive are nonmental. Some of them are real, some aren't. Perception, that is, is sometimes erroneous. At the time of Descartes, the new science and, not unconnected with it, the resurgence of skepticism, had made perceptual error a major problem, if not the major problem, of philosophy. Such concern is the mark of epistemologism. By that time, the Aristotelian-Thomistic assay of perception, which is realistic in structure, had collapsed, partly because it could

304

not account for perceptual error. Yet, the Aristotelian-Thomistic formula by which a mind can only know what is "in" it had not lost its power. ³ Assume now that in this ambience one uses 'mental', 'real', and 'exist' so that (1) to be "in" the mind and to be mental is one and the same thing, and (2) what is nonmental and not real does not exist. Such a one will easily be led to wonder how he can know that anything nonmental exists, and, eventually, to deny that it exists. That shows how the realism-idealism issue arose as well as why the three centuries during which it had such a baleful fascination coincide with the reign of epistemologism.
The next paragraph gets ahead of the story. But it, too, may help you to read it.
Concerning (1), I shall argue that with the hard-core use of 'mental' and the one clear use of 'being in the mind', being mental and being in the mind are two and not one. Concerning (2), I shall argue that with the hard-core use of 'nonmental' and the one clear use of 'real', to be real and to be nonmental are two and not one. To make either (1) or (2) true thus required some very special uses. Going unrecognized, these special uses became the philosophical uses at the roots of the perplexities with which I am here concerned.
The traditional dialectic of the realism-idealism issue is a tangle of ontological and epistemological questions. That may make you suspect that my emphatic ontologism prevents me from doing justice to the issue. It may again help if, in a paragraph, I try to allay the suspicion. That will be the last preparatory remark.
"You insist on ontology being primary. I take this to mean that in systematically expounding your philosophy you begin with what you call general ontology, then, proceed to what you claim is the ontology of our world. Why should I accept this claim?" Suppose that one who harbors that suspicion asks me this question. I answer with a comparison. There is the game of chess a sense of 'winning' in which the last and only the last move is the winning move. This move, though, must be preceded by many others. A philosopher's ontology is like the preceding moves; his epistemology (ontology of the knowing situation), like the winning move. If his last move does not win, you have indeed no reason to accept the claims he makes for his ontology. Since our issue is a tangle of both kinds of questions, which I believe is a major cause of its sorry state, the successive distinctions and arguments of this essay will be of both kinds, some ontological, some epistemological.

***** ³ This is explained at several places in this book and implicit in several others. But see p. 264.

305

As it happens, the very first distinction I shall make, concerning the phenomenological uses of 'mental' and 'nonmental', will be introduced by an epistemological question: What reason do I have to believe of anything that it exists? But I shall not bother to say of each step to which of the two kinds it belongs. As long as one knows what he is doing and does not improperly mix the two, no harm will come to him.
The key words of the realism-idealism issue are 'mental', 'nonmental', 'real', 'dependent' in such expressions as 'mind-dependent' and 'independent of mind(s)', and, of course, in view of the ontological involvement, 'exist'. 'Physical' and 'phenomenal' occur as synonyms or would-be synonyms of 'nonmental' and 'mental', respectively. Until much later I shall avoid them both. About 'exist' much has been said in this book. But it will be well to recall, very concisely, six points which will play a role.
1. 'Exist' is a 'green'-word for everyone, ordinary man and philosopher alike. Notice that I speak of a 'green'-word, not of a 'green'-use. That leads to the next point. 2. 'Exist' is univocal. When I read in a philosophical book that there are several kinds of existing, I am tempted to put it aside. If I don't, it is because I hope that reading on I shall discover that what is so unfortunately expressed is the proposition that there are several kinds of existents. If this hope is disappointed then I know that I could not possibly understand the book. So I put it aside. 3. The philosopher's concern is only with the most general kinds of existents. These are the ontological modes, categories, subcategories. 4. In my world there are quite a few ontological kinds. There are subsistents and simple things. The latter are either individuals, which are bare particulars, or universals. There are complexes. And so on. There are also two modes, actuality, and possibility. (The modes will play a role in this essay. 'Actual' I haven't used before. But the proper contradictory of 'possible' is of course 'actual', not 'real'.) 5. All the philosophical uses of 'exist' are narrower than the univocal commonsensical one. That fits with the second point. To use 'exist' philosophically is to adopt unwittingly what I call a pattern, i.e., a sentence one does not allow to be falsified. By one such pattern, only individuals exist; by another, only simples; by a third, only facts. And so on.
6. An ontological kind is "in" the entity to which the latter belongs. A thing's individuality or univocality, for instance, is a subsistent which is "in" it. What is "in" and entity is, as one says, "internal" or "intrinsic" to it. Both these words have been used philosophically. So I shall avoid them. A tree's color is "in" it; its distance from another tree is not; nor is color itself. What is not "in" a thing is, as one says, "external" or "extrinsic" to it. I shall again avoid these two words, for

306

the same reason as before, use 'contextual' instead. A classification based on contextual criteria does not yield ontological distinctions . That is of the very essence of ontology.
The only primary reason I have to believe of anything that it exists is that it is presented to me.⁴ The first two of the following five comments unpack the formula. The third introduces an assumption that will simplify the exposition. The last two not only show the import of the formula; they also will, once more, run ahead of the story in order to make it easier to follow.
1. Seeing footprints in the sand, I believe that someone walked there. The primary reason for the belief is my seeing the footprints. That unpacks 'primary'.
2. To be presented (to me) is the same as to be the intention, or "in" the intention, of an act (of mine). The act's species makes no difference. It may be direct (immediate) acquaintance, or perceiving, or believing, or disbelieving, or any other. That unpacks 'presented'.
3. Speaking of the reasons I have rather than of those we have provides an opportunity for introducing the assumption that will simplify the exposition. I believe that the group of philosophical questions I set out to answer in this essay can and must be answered without ever mentioning another group of such questions, namely, those concerning our knowledge of other minds. The later Husserl⁵ and his followers hold that this belief is false. If, as I shall suppose, it is true, then I shall neither prejudge nor distort anything but merely disentangle the two groups of questions by assuming, as I shall, that the intention of an act (of mine) contains a constituent which is mental if and only if its species is direct acquaintance. The assumption is contrary to fact of course, but it merely keeps out what can and must be kept out.
4. If we accept the formula, are we not dialectically forced also to accept its converse, i.e., that whatever is presented exists? I believe that we are. But I also have another reason, which is as weighty as any, for accepting the converse. No realistic ontology, we know, is adequate unless it provides an ontological ground for the connection between a thought (*P*) and its intention (P) (asterisks are used in lieu of corner brackets - sb). In my world, the connection is the intentional tie (M) in *P*MP. Since M and *P*MP both exist, the former as a subsistent, the latter as a formal fact, the connection is

***** ⁴ This is what Descartes' Cogito has long meant to me. In a recent essay parts of which I greatly admire, Jaakko Hintikka argued that it meant this also to Descartes. See his "Cogito, Ergo, Sum ; Inference or Performance?" The Philosophical Review , 71, 1962, 3-32. See also fn. 37.
⁵ I refer to the strange idea that our knowledge of rocks, trees, chairs, and so on, structurally depends on that of other minds. See p. 222.

307

grounded. But I do not see how *P*MP could exist unless P has some ontological status. That is the second reason for accepting the converse.
5. If all intentions are to exist, how about those of (the thoughts in) false beliefs and false memories, which, as one says, do not "exist"? How about perceptual error? The double quotes around 'exist' mark the philosophical use that limits existence to the actual. In my world, we know, the intention of, say, a false belief is a possibility (p-fact) and as such has ontological status (exists), even though it is neither actual nor, as we shall see, real. In all cases but one (and those which depend on it) the mode of possibility provides the way out of the difficulties the realists encounter in assigning, as I believe they must, some ontological status to the intentions of all acts. The recalcitrant case, as we shall see, is perceptual error of the kind called existential. In the traditional dialectic it looms very large. The classical realists were more anguished by the arguments drawn from it than by any others. I have as yet not attended to the recalcitrant case. In this essay I shall. (Psychologists may attribute the delay to a reaction against the classical realists' frenzy.)
What is mental, what nonmental? The right answer is the one that keeps one out of philosophical trouble. To find it, one must attend to the hard core. There, we shall see, the dichotomy mental-nonmental "corresponds" to another. Yet they are two and not one. Nor does one depend on the other. Unless a philosopher grasps that firmly he gets into trouble. It should make it easier for you to keep these two dichotomies apart if I first direct your attention to the second. That requires that you recall the schema of the act and, from the opening essay (p. 27), how and why I use 'conscious state'.
The several species, perceiving, believing, doubting, and so on, are simple characters. Hence, if I know them at all, I also know them from each other. To ask how or why makes no more sense than to ask how or why I know red from green. Sameness and difference are primary. Similarly, it makes no more sense to ask how I know that a character is a species than to ask how I know that green is a color and square is not. 'Remembering is a species' is exactly like 'green is a color'. Both are synthetic apriori. As for species so for thoughts. To ask how I know that a simple character is a thought makes no more sense than to ask how I know that round is a shape. Remember what has been said in the preceding essay about the dimensions. Thought and species are two dimensions; and it is a synthetic apriori truth that every individual exemplifying a member of one of these two dimensions also exemplifies a member of the other. (These individuals I call awarenesses.)

308

Schematically,⁶ a conscious state is an act, e.g., a perceiving, i.e., an awareness that exemplifies perceiving and a thought. But it is a conscious state only by virtue of a second act whose intention it is.⁷ The second act is not "in" the conscious state. Yet this second act can and often does itself become a second conscious state through what psychologists call a shift in set. Every act by virtue of which another act is a conscious state is a direct (immediate) acquaintance. Direct acquaintance is as unmistakably a species different from all other species as green is a color different from all other colors.
When I perceive something, my conscious state is an act of perceiving. Yet "my attention centers" on the intention of the act. As for perceiving so for all other species. The expression between the double quotes is a metaphor. I know no better way of directing attention to what I am trying to express. Husserl expresses it by saying that we "live in" the intentions of our acts. That, too, is a metaphor.
Direct acquaintance is not just a species among species; it has a "flavor" that sets it apart from all others. That is again a metaphor. A comparison may help. Black, white, and the several greys have a flavor that sets them apart from all other colors. (Hence the familiar use of 'color' that excludes them.) If the comparison does not help, recall the metaphor of the "lisere" Proust uses in the passage that is the epigraph of the opening essay. The two metaphors may illuminate each other. When my conscious state is a perceiving, I "am" this direct acquaintance. In this respect there is no difference. The difference is that when my conscious state is a perceiving, there is between me, i.e., the act which I "am," and its intention that "lisere" which is not there when I "am" a direct acquaintance. Philosophers, speaking about these matters, have used four adjectives. One species is said to yield acquaintance which is "direct" or "immediate." All others yield only "indirect" or "mediate" acquaintance with the intentions of the acts whose species they are. Direct acquaintance, then, is one thing. All other species, the several kinds of indirect acquaintance, are another thing. This is the dichotomy to which I wanted to call attention first.
In establishing the dichotomy direct-indirect no reference has been made either to the differences, if any, among the intentions of the dif-

***** ⁶ I say schematically because dialectically it makes no difference whether or not a conscious state also contains sense data. Nor is there any reason why one conscious state shouldn't contain several acts or be what the classical psychologists call a fusion of them. See also fn. 26.
⁷ More precisely, it is the intention of the thought in the act. But I shall permit myself this contradiction.

309

ferent kinds of acts or to anything else except the species themselves and their "flavors." That makes it an ontological distinction among species. Call now an act's intention direct (indirect) if its species is direct (indirect). That saves words and does no harm as long as it is kept in mind that this classification of intentions, being contextual, is not, like the dichotomy between the two kinds of acquaintance, ontological. There is, however, a dichotomy among intentions, based on what is "in" them and hence ontological, that is coextensive with the contextual classification.
The dichotomy mental-nonmental is primarily one among simple characters. An individual, being bare, obviously is neither mental nor nonmental. Nor, even more obviously, is a subsistent. Thoughts and the several species are mental; all other simple characters (of the first type) are nonmental. This is one way of stabilizing the hard-core use of 'mental'. Another way is to call thought and species themselves the two mental dimensions. The first way puts mental and nonmental in the second type; the second, in the third. The difference makes no difference whatsoever.
With the sole exception of the temporal characters, no individual exemplifies two simple characters of the first type one of which is mental while the other is nonmental. Nor am I acquainted with any simple character of the second type exemplified by both mental and nonmental characters of the first type; and so on. This is due to a remarkable pattern of dependencies and exclusions (p. 295), all of which are synthetic apriori.
I have said that the hard-core use of 'mental' is a 'green'-use. In other words, I believe that mental itself is a simple character. Perhaps you disagree; propose to define 'being mental' as 'being either a thought or a species'; believe that I am reifying a "flavor" produced by that remarkable pattern of dependences and exclusions. Again, the disagreement makes no difference whatsoever as long as you agree that I have succeeded in calling attention to the hard-core use of 'mental' and 'nonmental'; that this use can never get us into philosophical trouble; and that the dichotomy it represents is ontological.
All uses of 'mental' and 'nonmental' share a feature that gives no trouble and which it will therefore be safe for us to preserve. A complex is called mental if and only if the simple characters "in" it, if they are of the first type, are either temporal or mental; or, if they are of the second type, are exemplified only by the mental characters; and so on. A complex is called nonmental if and only if all the simple characters "in" it, if they are of the first type, are nonmental; or, if they are of

310

the second type, are exemplified only by nonmental characters; and so on. This is the derivative use of 'mental' and 'nonmental'. All other complexes may be called mixed. Mixed intentions we may safely ignore. That is the import of the assumptions by which we disentangled the realism-idealism issue from that of our knowledge of other minds (and, as we now see, also from the mind-body problem). If we keep that in mind and adopt the derivative use then we can state the "correspondence" between the dichotomies mental-nonmental and direct-indirect very perspicuously.
A species (acquaintance) is either direct or indirect. An act's intention is either mental or nonmental. These are the two ontological dichotomies. An act's intention is mental if and only if its species is direct acquaintance; it is nonmental if and only if the act's species is indirect. That is how the two dichotomies "correspond" to each other. Close as the correspondence is, we have seen that the two dichotomies are equally fundamental, so that neither of the two depends on the other. To appreciate how important that is, suppose that mental-nonmental depends on direct-indirect. This is the one of the two possible dependencies which, if one is not accurate, may seem plausible. In this case, a distinction within the "seen" would depend on one between kinds of "seeing." That blurs this distinction between "seeing" and the "seen." One who blurs this distinction is, we know, on the way to idealism. That is why I have so emphatic, almost prolix.
The phenomenalists hold that we acquire the "idea of external existence" from the "coherence" of what I, though not they, call the intentions of several acts. Presently we shall hear much more of this coherence. For the moment I merely avail myself, for a limited purpose, of what everyone who will read this book knows.
I hold that when I am presented with a nonmental intention, i.e., when I "am" an act which is not a direct acquaintance, then I am eo ipso presented with the "idea of external existence." For this "idea" is nothing but the "idea" of the nonmental. Or, perhaps, it is Proust's lisere. The difference makes no difference. In other words, 'external existent' is not the name of an ontological kind. Or, if you prefer it that way, it is merely a synonym of 'nonmental'. Hence, even if per absurdum my Self exhausted itself in a single act and this act were, say, a doubting or a disbelieving - I deliberately pile absurdity upon absurdity - I could yet have the "idea of external existence" without, as the phenomenalists hold, having to acquire it from a plurality of conscious states. That alone shows the realistic impact of grounding the dichotomy mental-nonmental in the hard core. It also frees the notion

311

of coherence for the use I shall presently make of it by relieving it of a burden which it cannot bear.
"Some things are mental; some, nonmental. (Things of both kinds exist.)" This is a familiar statement of the classical realists' claim. If one construes the two key words in it with their hard-core use and recognizes that a certain proposition is a hard-core truth, the statement itself becomes such a truth. Why, then, call it a claim? Why argue for it? The answer is that the classical realists understood neither the hard-core use of 'mental' nor the proposition in question. ⁸ The latter is the converse of the one from which we started. "If anything is presented (to me), so is its existence." If one doubts that this is a truth of the hard core, let him ask himself how he knows what 'exist' means. ⁹ Why the classical realists failed to grasp this truth we understand already. There are the difficulties they could not overcome because they did not recognize that possibilities exist. There is also the recalcitrant case, perceptual error of the kind called existential.
We just saw that unless the two key words in a familiar statement of the classical realists' claims are used in a special way, the statement is a hard-core truth rather than anything to be worried about and argued for or against. We could of course explore that special use directly, but it will pay if we proceed indirectly.
The controversy is about stones, trees, chairs, and so on. We must first argree on a suitable name for their kind. 'Thing' I so often use technically that it is preempted; even 'ordinary thing' does not recommend itself. 'Physical object' and 'perceptual object' I want to avoid because we shall need them later. I think it will be best if we agree to call stones, trees, chairs, and so on, ordinary objects.
"Ordinary objects are real." This is another familiar statement of the claim. It rings in the third key word, 'real'. The idealists deny that ordinary objects are real. Everything depends on how the two sides use 'real'.
An entity is real if and only if it is actual, i.e., not merely a possibility, and, in case it is not mental, mind-independent. That explicates how both sides use 'real'. I, too, shall use it this way. The materialists, we shall see,use it differently. As far the idealists are concerned, the explication is a bit schematic. Yet it does justice to the analytic center of their claim and the speculative accretions do not interest me at all.

***** ⁸ From what I know about him, Meinong, for all his tenacity and all his profundity, is no exception.
⁹ This is one of the main points of Essay Three.

312

The explication makes all actual mental entities real.¹⁰ 'Mind-independent' is the last of the four key words. We must answer three questions. What does it mean for an entity to be mind-independent? What reason do we have to believe that there are such entities? If there are any, are ordinary objects among them? The answers depend on how one ontologically assays ordinary objects and the situation in which he perceives them. Since ontology is primary this is as it ought to be.
A substance is a simple individual, is a continuant, and has a nature. Or, rather, this is the "idea" of a substance. Its several parts are irreconcilable with each other.¹¹ In my world there are therefore no substances. The structurally consistent alternative to them is bare individuals. The particulars of my world are bare individuals; but they are momentary,¹² not continuants. The reason is that we cannot recognize particulars are such. ¹³ That may seem to exclude continuants on epistemological grounds; yet it really doesn't. Epistemology asks how we can know what we do know. Ontology accounts for what in fact we do know. Since we do not recognize particulars as such, we do not in fact know continuants (in the relevant sense).¹⁴ That virtually determines my assay of ordinary objects. An ordinary object is a temporal succession of particulars each of which exemplifies certain nonmental (nonrelational) properties and stands in certain nonmental relations to every other. ('Temporal succession' merely refers suggestively to the temporal ones among these relations.) That makes table, even schematically, a very complex character. Beyond the schema we cannot and need not go.¹⁵
A perceiving is an act among acts. Perceiving is a single species. 'Perceiving' has three uses; we shall distinguish them by subscripts. Since perceiving, like all species, is a simple character, the distinctions

***** ¹⁰ Upon the broad use of 'realism' and 'idealism', materialism is a variety of realism; phenomenalism, one of idealism. I stay here with the narrow use. See fn. 32.
¹¹ For an argument in support, see the Leibniz-essay in Meaning and Existence .
¹² The dialectic of moments, instants, and the specious present we can here safely ignore. See, however, Essay Five.
¹³ This is one of the major burdens of Essay Nine.
¹⁴ For all we know, some physical entities may be continuants. See the concluding essay of Meaning and Existence. Concerning the notion of physical entities, see Essay Six and below.
¹⁵ For some relevant features of the schema see the essay on Logical Atomism in Meaning and Existence. In that essay the particulars of the schema are still sensa. Very significantly, that makes no difference whatsoever except in the one respect which is being considered in the present essay. See below.

313

are contextual, not ontological. That is so important that I want to emphasize it by stating succinctly what these distinctions are before describing them in detail. The intention of a perceiving₁ is narrower than that of a perceiving₂. A perceiving₃ is the last of a series of perceivings₂.
Every time I look up from my desk I perceive a table. These are all perceivings₂. Take one of them. The thought "in" it is the-thought-that-this-is-a-table. (*This is a table*; briefly:*P*). *P* is simple. Its simplicity accounts for a striking feature. All thoughts are simple.¹⁶ In the case of perception, some philosophers sensitive to the feature insist that what I am presented with is the table and nothing else. They are right, of course; but the way they express themselves easily leads to the misunderstanding that the intention (P) of the thought (*P*) is itself simple. We know that all awareness is propositional; hence, no intention is simple. That is why I rather express what these philosophers try to express by insisting that the intention in question is the fact¹⁷ represented by 'This is a table'. How shall we assay this intention? It will pay if we postpone the answer for a moment.
When I perceive₂ the table, my "attention does not center" on any of the particulars "in" it that may or may not on this occasion be presented to me. This is but a third way of expressing, by means of a metaphor already familiar, what the philosophers just mentioned insist upon. But we all are, when perceiving₂, capable of shifting to another act whose intention is the fact of some particular (or particulars) exemplifying some properties (and relations). These particulars are all "in" the table; and they are all simultaneous with the awareness, i.e., with the particular "in" the (second) act. The species of the latter is again perceiving. Its intention, accordingly, presents me with the "idea of external existence." Yet it is characteristically "narrower" than that of the preceding perceiving₂. An act with such an intention I call a perceiving₁. Psychologists call the shift in question a shift of set. In painters it is supposed to be a professional habit. The impressionists made it the corner stone of their technique. To us it provides a cure for assaying the intentions of perceivings₂.
Return to the example. How are we to assay its intention (P), i.e., the fact represented by 'This is a table'? Let 'Mp' be the sentence representing the fact that would have been presented to me if the act had been a perceiving₁ instead of a perceiving₂. Let 'a₁,..., 'a_n' stand for

***** ¹⁶ For the dialectic that springs from this feature see in particular the first section of Essay Seven.
¹⁷ Or possibly (p-fact). We are not concerned with this distinction.

314

the particulars "in" it and, whenever it helps, write 'Mp(a₁,...a_n)' instead of 'Mp'. The letter 'M' in 'Mp' is to remind us that the fact is molecular. The letter 'p' is to recall 'part'. For I assay P (This is a table), which is the intention of the perceiving, with which we are concerned, as a conjunction of two (part) facts. One is Mp; the other I call Op. That makes 'P' an abbreviation for 'Mp & Op'. The letter 'O' is to remind us that 'Op' contains operators. That makes 'P' an abbreviation for 'Mp & Op'. 'Op', the operator part of P, is very complex indeed. Fortunately we need no details. The idea is easily grasped. 'Op' schematically states that there are all the particulars which must be there, that these particulars have all the properties they must have, and that they stand in all the relations in which they must stand, among themselves and to the particulars in the molecular part, if the latter is to be "in" a table. One may indicate this complexity by expanding 'Op' into '[a]Op(a₁,...a_n, a)' {'a' is used where Bergmann uses the Greek alpha - sb}, where the bracket stands not for one but for many operators, some existential, some general; and the Greek letter {'a' in this document - sb} serves to avoid commitment to either number or type of the bound variable.¹⁸
Details, I said, don't matter. One, though, must be singled out. For the statement that a certain particular, one of those which as I just put it must be there, is in fact there, or that it exemplifies a certain character, to become deducible from 'Op', some synthetic a priori truths (facts) about the relevant dimensions, particularly about space and time, must be included among the parts of Op. That may well be one of the sources of the need, felt by so many philosophers, for setting apart the truths they call synthetic a priori. Thereby hangs another point too good to be missed. Some now believe, very strangely, that synthetic apriori truths are "linguistic." Nothing being both red and green (all over) is one such truth. Everything red being extended is another. Neither is molecular. That spots a likely source of the strange belief. Those holding it not only are sure that there are no sensings but also fail to distinguish¹⁹ between sensing and perceiving₁. Thus they miss, or ignore, all molecular intentions.
Even the most fleeting glimpse I have of I know not what is a perceiving. One perceives when he dreams or hallucinates; otherwise he would not be dreaming or hallucinating. We all make perceptual errors; may have mistaken for round a tower that is square. We have all been victims of cleverly devised illusions; may have mistaken a dummy for a

***** ¹⁸ Write now 'Table (x₁,...,x_n)' for 'Mp(x₁,...x_n) & [a]Op(x₁, ...x_n, a)' and you have the predicate which (in this case) represents the complex character table. For this as well as some related details, see the essay cited in fn. 15.
¹⁹ The distinction will of course be made below.

315

man. In this sense, all perceiving is problematic. In principle, it forever remains so. Practically, when we check a perceiving, we are much less likely to be, or remain, deceived. One checks a perceiving by making it the first of a series of perceivings, all of the same ordinary object (if there is one). I walk toward and then around the tower; I touch the dummy; and so on. When I am satisfied, I stop and the series breaks off. Its last member I call a checked perceiving or perceiving₃. That makes 'perceiving₃' what some now call an achievement verb. There is no harm in the label as long as it does not cause one to ignore the two other kinds or not to recognize that the species is the same in all three. Perceiving₃ involves remembering. We may safely ignore this circumstance. Again, one cannot neatly really separate perceiving₃ from the "coherence" which is our next topic. But, again, nothing will be lost, and, the traditional dialectic being what it is, something will be gained by proceeding as I do.
We are ready for the three questions. What does it mean for an entity to be mind-independent? What reason do I have to believe that there are such entities? If there are any, are ordinary objects among them?
An entity is mind-independent if and only if it exists while it is not in the intention of an act (of mine). That explicates how both sides to the classical controversy use 'mind-independent'. I, too, shall use it this way. 'While' is temporal. What exists while it is presented is in time. At this point I draw on my ontology. The only entities literally in time are (momentary) particulars. That allows for a rephrasing, without substantial loss, of the last two questions. What reason do I have to believe that a particular not presented to me exists? What reason do I have to believe that a particular not presented to me exists? What reason do I have to believe that there are all the particulars not mentioned in 'Mp' which, according to 'Op', must be there if the table I perceived is to be mind-independent? We need not, we see, bother with the rephrased second question. If we can find the reason asked for in the rephrased third, we have also found the one asked for in the second, except for one pedantry which it will be best to dispose of right now. Taken literally, the explication implies that a₁..., a_n, the particular in Mp, are not mind-independent, simply because they are presented (to me). If you noticed that, modify the explication so that they, too, become mind-independent provided only that we can find the reason asked for in the rephrased third question.
(One pedantry bring to mind another. G. E. Moore often makes much of the distinction between good and bad reasons. Sometimes he has a good reason for that. I here always speak of reasons. I mean good reasons, of course. Presently I shall in a long footnote give some

316

plausible structural reasons why the good reason I am about to give in answer to the rephrased third question seems bad to some. For the moment, keep in mind that, given the relevant uses, if this reason is a good reason, then we have also found a good reaon for believing that ordinary objects are real.)
The only reason I have to believe that ordinary objects are real is coherence. That means three things. (1) 'Op', we remember, deductively implies statements to the effect that there are certain particulars exemplifying certain characters which were not presented to me when I perceived the table. Whenever I am in a position to perceive such a particular I do in fact perceive it. More precisely, I perceive it in a very large proportion of all cases provided the original perceiving was a perceiving₃. (2) There is a huge body of generalities (laws) about the changes ordinary objects undergo in their properties and relations to each other. None of these laws mentions anything mental. In a very large proportion of cases in which either we do not perceive or the particulars we do perceive do not exemplify the characters which we expect them to exemplify, these laws "explain" the discrepancy. When I last saw these leaves they were green; now they are yellow; fall has arrived. (3) The remaining cases in which we do not perceive what according to (1) and (2) we ought to perceive as well as the cases in which our perceivings ₂ deceive us are "explained" by another large body of laws. These laws do mention mental things. The stick partially immersed in water seems bent. We know that it is straight. If waking from a dream I remember it, I know that it was a dream.
To separate sharply (2) from (1) is no more important than so to separate from each other the notions of perceiving₃ and coherence. Another distinction is most important. The laws mentioned in (3) are one thing; the job of ontologically assaying perceptual error, dreaming, and so on, is quite another thing. To believe that those laws do this job is the crudest kind of scientism.
Do our perceivings in fact cohere without residue? If they have so far, could they not cease to cohere from now on? The answer is familiar. In principle, coherence, like perception, is forever open or problematic. This is their very nature.²⁰

***** ²⁰ Or, what amounts to the same, perception and coherence share the characteristic feature of induction. There is, of course, a sense of 'problematic' in which induction is not problematic at all. Neither, therefore, are perception and coherence. Some nevertheless reject the above analysis because of this feature. G. E. Moore, for instance, asserts that 'P' follows deductively from either 'I perceive that P' or 'I know that P' (though not from either 'I judge that P' or 'I believe

317

My perceivings of the table cohere. The table is real. The spot I saw on my blotter while having an aferimage is not real. I had an hallucination, saw a ghost. The ghost is not real. What has been said explicates this use of 'real'. The key idea is coherence. That makes the notion contextual. Hence, real is not an ontological kind. (In speaking about ordinary objects I shall use 'real' in no other way.)
Characters are either mental or nonmental; particulars, being bare, are neither. Yet, as it has been convenient to call a particular an awareness in order to indicate its being "in" a mental fact, so it will help

***** that P'. If my assay of the act is adequate, then Moore is radically wrong. If you substitute in 'f(a)&*P*(a) ---> P' (f has been used in place of Greek phi; arrows are used instead of the horseshoe - sb) the name of a species for 'f', the resulting sentence will be synthetic in all cases, including those of 'perceiving' and, if it be a species, of 'knowing'. What is analytic is, rather, '*P*MP', and this irrespective of any species, which is not even mentioned in it, and of whether 'P' is true or false.
If you use 'real' as I do and accept my assay of P, then if what Moore asserts were true , the table I perceived would be real (mind-independent) on the sole ground that I perceived it. Clearly, this is a short-cut to realism. It is true, although not analytic, that what we perceive is nonmental. But the (actual) nonmental does not, we shall see, coincide with the real. The realistic urge behind the short-cut is only too transparent.
Knowing is not a species. Believing, on the other hand, is not a single species but a series of such. The series is produced by the relational character more-certain-than. The higher in the series a believing is, the more inclined we are to call it a knowing. But there is a complication. Just as there is perceiving₂ and perceiving₃, i.e., checked perceiving₂, so there is believing and checked believing. The complication is that when one says he knows something he implies not only that he believes it with a high degree of certainty but also that he has checked the belief in a manner appropriate to it, while, on the other hand, we also hold beliefs with very high degrees of certainly without having checked them.
The only reason Moore gives for his assertion is that it merely makes explicit the way 'perceiving' and 'knowing' are used. As to 'perceiving', I have my doubts. As to 'knowing', I have no doubt whatsoever. We all say such things as "How could he have known it if it isn't the case?" It does not follow that the sentence (or sentences, if you include 'perceiving') which Moore asserts to be analytic are analytic. Checked beliefs, like checked perceptions, are very reliable. This more than suffices to account for the use. That creates the following situation. Moore's assertion and my assay of the act contradict each other. His assertion rests on no more than a certain use. My assay rests on the hard core. Also, the use in question is easily accounted for in a manner consistent with my assay of the act. The situation shows how dangerous it is to introduce uses as reasons into philosophical arguments. Those who now agree with Moore on what he asserts about knowing and perceiving and also accept his reason for it, no doubt satisfy their realistic urges. So, we saw, did Moore, although he did not, as these latter-day disciples, either miss or prejudge virtually all the distinctions and analyses which are the heart of the realism-idealism issue. But, then, most of his arguments were better than theirs, which are all from use.

318

to have a word for the particulars "in" nonmental facts. Let us call them external particulars . The expression will remind us that each nonmental intention presents us with the "idea of external existence." Let us agree to call external particulars and molecular facts real if and only if they are "in" real ordinary objects. (This is but a slight and most natural extension of the above use of 'real'.) Are all external particulars real? In order to answer we must assay perceptual error. To this task I now turn.
Take perceiving; let its intention be P.²¹ Perceptual error occurs if and only if 'P' is false. 'P' is an abbreviation of 'Mp&Op'. The conjunction is false if and only if at least one of its terms is false. That provides us with a schema for the cases which must be considered and the comments which must be made.
First Case. I perceived a coin. For some reason I reached for it. Touching it and weighing it in my hand. I discovered that it was a cardboard imitation of the kind used in games. I also saw that the imitation of the visual appearance of a coin was very accurate. In other words, the 'Mp' of the original perceiving is true; its 'Op' is false.
First Comment. If 'Op' is false. Op is a mere possibility. The conjunction of two entities at least one of which is a possibility is itself merely a possibility. Possibilities, though, have ontological status. P exists. (The same holds mutatis mutandis in the other cases.)
Second Comment. There are now quite a few who would object to my saying I saw a coin, would insist that what I saw merely looked like a coin or seemed to be one.²² Those who would make this objection either confuse or tend to confuse the task of ontologically assaying perception with the quite different and most tedious one of accurately describing the relevant uses of 'looking', 'seeming', and 'appearing'? What is the ontologically ground of looking, seeming, appearing? The mere question suffices to spot the confusion.
Second Case I perceive₂ an oval coin. Surprised that there should be such a thing, I reach for it, examine it, perceive₂ it to be round. The 'Op' of the original perceiving ₂ is false. How about its 'Mp'? Let a be the particular (area) "in" it that was presented to me. 'a is oval', false. The latter is a conjunction term of 'Mp'. That makes Mp, too, a mere possibility. But the external particular in Mp is real.
Third Comment. Some believe, mistakenly, that the particular pre-

***** ²¹ No harm will be done if I continue to use 'P', 'Mp', 'Op' even though the examples will be different.
²² For the ground of this objection see fn. 20.

319

sented "in" Mp when I perceive a round coin is in fact often oval. Probably they have been misled by a familiar bit of science. The irrelevance of science at this point is obvious. But there is also a dialectical blur, or apparent difficulty, which is easily cleared up. Having shifted from a perceiving₂ of a round coin to a perceiving₁, one sometimes is presented with an area that is oval. The spurious difficulty arises when one fails to realize that the intentions of the two acts are two and not one. Turn back to the statement on p. 314 which introduces the notion of Mp and you will read: Let 'Mp' be the sentence representing the fact that would have been presented if the act had been a perceiving₁ instead of a perceiving ₂.
Fourth Comment. Perceiving a coin, I think and perhaps say; "This, though it looks oval, since it is a coin, is round." The sentence represents the intention of an act (of mine) which is not a perceiving. This act - of judgment, as one says - is the last of a series. The first of the series is an instance of the second case. The series may be run through very quickly. We often recognize "immediately" a familiar object even though we are "fleetingly" aware of its unusual color due to an unusual illumination. If you take that to be a single act of perceiving, you will be in trouble; for, 'Mp' ('a is green') is false, yet 'P' ('This is my blue easy chair') is true. Since recognition was "immediate," you may wonder. Was there really a series? How many members did it have? What were they? Did they fuse? One who worries about the answers to these questions will get lost in what I call the wrong kind of phenomenology because, as I have repeatedly insisted in this book, it is dialectically irrelevant and a philosopher must therefore know how to keep away from it.
Third Case. Suppose I saw a ghost. Let 'a is white' exhaust 'Mp'. The operator part of 'P' is again false. In this respect the three cases are alike. In two others they differ. 1. In the first two cases, the particulars "in" 'Mp' are all real. In this case, the external particular in Mp is not real. 2. Since all particulars are "momentary," we cannot "check" any.²³ Yet, if a particular is real, we can check the ordinary object "in" which it is. If we do that, coherence provides us with the only reasons we could possibly have for believing that a certain particular not presented to us exemplifies certain characters. Since in the first two cases the particulars in Mp are all real, we had reasons for believing that the 'Mp' of the first case was true; that of the second case, false. For

***** ²³ For the significance of such "checking," also in connection with the now fashionable idea of criterion, see Essay Eleven, p. 238.

320

the 'Mp' of the third case we have no such reason for believing either that it is true or that it is false. Presently we shall see that there is another reason for counting is true.
The first cases cover schematically all the perceptual errors of the kind called qualitative. One must recognize that possibilities exist. The third case covers schematically all the perceptual errors of the kind called existential. This is the recalcitrant case, the one which undid the classical realists (and to which I have as yet not attended). In assaying it adequately a second hurdle has to be overcome.²⁴ One must recognize that some external particulars are not real. The classical realists identified the external with the real, either because they did not make them correctly. So they were undone.
The resistance against recognizing the critical particulars for what they are must have been very great. Its source is not hard to spot. On the one hand, these particulars are not mind-dependent; on the other hand, they are external, i.e., they are "in" nonmental facts. That has an appearance of paradox which is a plausible source of the resistance. Notice that I called the critical particulars "not mind-independent" rather than "mind-dependent." As we use 'mind-independent', it stands for the one clear notion that goes by the name. But there are several ways in which entities may depend or be thought to depend, on a mind. The thing to do, therefore, is to describe accurately and distinguish from all others the way in which a critical particular depends on the act which intends it. Then the appearance of paradox will disappear.
1. A classical substance "creates" its attributes. The created "depends" on its creator. An act (mind) may be thought to create its intentions. That makes the latter depend on the former. Since this use of 'create' is irremediably anthropomorphic, this notion of dependence remains blurred. I mention it nevertheless because, according to the idealists, a mind creates its intentions. An idealist thus need not deny that nonmental entities exist. He merely claims that, since they are mind-dependent, they are not real.
2a. Let 'A' and 'B' be the descriptions of two temporal cross sections (momentary states), at times t₁ and t₂ respectively, of a system that undergoes a process. Both A and B are facts, of course. B being the system's state at t₂ follows deductively from the law of the process in

***** ²⁴ This is not to imply that the classical realists all overcame the first hurdle. Hardly any did.

321

conjunction with the premiss that A is its state at t₁. That makes the earlier of the two states a "cause" of the later; the later, an "effect" of the earlier.²⁵ An effect "depends" on its cause. Process laws are synthetic generalities.
2b. There are also cross-section laws; for instance, the laws connecting features within a cross section of a process; hence the name. The Pythagorean theorem is such a law. The synthetic a priori truth that everything colored is extended is another. The (parallelistic) connection between a conscious and bodily state is a third. The facts in an instance of a cross-section law "depend" on each other. Cross-section laws, too, are synthetic generalities.
3. Let A and B be two facts such that 'A ---> B' is a formal truth (analytic). 'A' logically implies 'B'. B "depends logically" on A. Process and cross-section laws being synthetic, neither 2a nor 2b are logical dependencies. That makes it convenient to lump them together as two kinds of causal dependence (in a broad sense of 'causal').
Let now 'A' be an abbreviation of 'perceiving (b) & *B*(b)'. The fact A is an act of perceiving; b, the awareness "in" it; B, its intention. The only connection between the two facts A and B is the formal fact *B*MB. Hence, even if you identify the act (A) with the thought (*B*) "in" it, the connection is not one of either logical or causal dependence. Since '*B*MB' is analytic, B does not causally depend on A, even though, as in 2b, the awareness b is simultaneous with the particulars in B. Since '*B*MB', even though it is analytic, is not a conditional, B cannot be deduced from either A or *B*. Hence, B does not logically depend on A. The connection, we see, is sui generis.
Let B stand for 'a is white'. If a is real, there is no good reason for calling the connection a dependence of B on A. If anything, the occurrence of A depends causally, among other things, on B. If, however, a is the particular presented to me when I saw the ghost, then, if the act, i.e., my perceiving the ghost, did not exist, neither would a. This is a good reason for calling a as well as the fact of its being white mind-dependent. Only, this is a fourth kind of dependence which is unique in that it obtains only between some acts and their intentions. Also, it is in a certain sense mutual, as in 2b.
External particulars which are not real are mind-dependent.²⁶ Unless

***** ²⁵ For a detailed analysis of the notions that appear in 2a and 2b, see the second chapter of Philosophy of Science.
²⁶ Suppose I imagine a dragon. Or suppose I have a memory image of a mermaid in Lake Michigan. In either case I am presented with critical particulars (i.e. particulars which are external but not real). Nor is there any good reason to believe that the particulars I am presented with in a true memory accompanied

322

some acts existed, they would not exist. There being such particulars is part of the intentional nature of mind. To see that there is nothing paradoxical about this one merely has to recognize the dependence for what it is. Then one will not be tempted to say of any act either that it creates its intention or that the latter either causally or logically depends on it.
A sentence represents a fact or a possibility (p-fact) depending on whether it is true or false, respectively. If you noticed that I just called the entity which in the case of the ghost is represented by 'a is white' a fact, then you probably also remembered that I promised to give a reason why, in the absence of reasons from coherence, this sentence should be counted as true. This reason is negative, as it were. Since a is mind-dependent, i.e., since it would not exist unless it were presented, it is plausible, to say the least, that it does exemplify the characters with which it is presented. If you don't think this is a good reason, I shall merely say that, if you agree with the way in which I am resolving the classical dialectic in this essay, then it does not matter at all whether the Mp of the ghost case is a fact or a possibility. The reason I nevertheless try to convince you that one may at least plausibly hold it to be a fact that this plausibility does matter, is indeed, as we shall see, of crucial importance in connection with the

***** by an image are "in" the ordinary object I remember. The conscious states which we "are" when we make a perceptual error of the kind called existential are not, we see, the only conscious states in which we are presented with critical particulars. As philosophers have used 'image', the conscious states of this sort are those in which we are presented with an "image." The word is dangerous, but it is also suggestive.
The adequate assay of perceptual errors of the kind called existential provides us with schemata for so assaying all conscious states in which we are presented with critical particulars that we can answer all the dialectical questions that have been asked about them. This is all that matters. Take a conscious state which is a fusion of three acts; one an act of remembering; one an act of perceiving "in" whose intention are critical particulars; the third an act of believing that the intentions of the first two resemble each other. I have little doubt that this fusion will do as a schema for rememberings accompanied by an image; and I have no doubt whatsoever that if it shouldn't, we have the means to construct another that will. Is the perceiving in this schema a perceiving₁ or a perceiving₂? Dialectically, what does it matter? If you are fascinated by such details you are in danger of getting lost in the wrong kind of phenomenology.
If there are sense data, of which more anon, then there are still other possibilities. In the above schema for rememberings accompanied by images, for instance, the perceiving could be replaced by sense data. Even the conscious state which I "am" when perceiving something could be fusion of an act of perceiving with sense data. And so on.

323

so-called "infallibility of sense data" which is one of the pivots of the classical dialectic we must resolve.
The critical particulars are external and not real. A particular is external if and only if it exemplifies nonmental characters. It is not real if and only if it is not "in" a real ordinary object. Both criteria are contextual. The critical particulars are therefore not an ontological subkind (of the kind particular). Occam's razor is primarily a weapon against redundant ontological kinds. Yet, the resistance against the critical particulars is so strong that one should know how to defend them against attempts to excise them with the razor. I shall next examine one likely attempt.
A critic argues as follow: "First you made possibilities exist, which was bad enough. Now you insist that there are particulars of the kind you call critical, which crowds your world still more (although not, I grant, your categories). Why don't you at least make those particulars possibilities?" I answer that, as I explicate the ontological use of 'possible', a possible entity is essentially a complex. Take sentences, which of course represent complexes. To say that what a false sentence represents is an entity in the mode of possibility is merely another way of saying that its being well-formed has an ontological ground. So explicated, the ontological use of 'possible' makes sheer nonsense out of the expression 'possible particular'. Nor can I think of any explication that doesn't. But assume, for the sake of the argument, that there is one. Then actual and possible would be two ontological (sub-)kinds of particulars. Hence, there would have to be two subsistents. All I can say and all that I need say is that I have never been presented with either. At this point the critic, who is eager to use the razor on something, tries to strike a bargain. If I permit him to excise the mode of possibility, he will let me have the critical particulars and permit me to make every well-formed sentence represent an actuality. I reject his proposal; refer him for my reasons to Essay Seven. Their gist is that the only way of doing what he proposes without being overcome by immediate dialectical disaster is Frege's.
Ordinary objects exist independently of the minds which may or may not perceive them. If you don't believe this truism you are mad. Believing it does not make you a realist. A realist is a philosopher who propounds an ontology which perspicuously reflects the truism. An ontology cannot be adequate unless it can be defended against all dialectical challenges. The recalcitrant case presents a challenge to the realist. If he fails to meet it, he may be driven to idealism. If he cannot

324

stomach idealism, he must choose phenomenalism. (I ignore for the time being materialism, which is the only remaining alternative sufficiently articulate structurally as well as important historically to deserve serious consideration.) To the phenomenalist perception, including the recalcitrant case, causes no trouble. We must understand accurately why this is so.
The time has come to talk about sense data. Are there such entities? The question is controversial. I have long sided with those take the affirmative. Nor do I now see any cogent reason to change sides. That, though, is entirely beside the point. The point, which I have not grasped accurately until recently even though fortunately I have sensed it for quite some time, is, rather, that dialectically it makes no difference whatsoever whether or not there are sense data. If one can show that, one can dismiss the question. Since I hope to show it, I put the question aside, start from another. If there are entities of the kind philosophers call sense data, what are they?
Everyone can make the shift from perceiving₂ to perceiving₁. Having made it, he is no longer presented with an ordinary object but with a molecular fact simultaneously with the awareness intending it. (This is an abbreviated way of speaking. Literally, no fact is in time; only the particulars "in" it are. Nor does either an act or the awareness "in" it literally intend anything; only thoughts do. But no harm will be done.) The molecular fact intended by a perceiving₁ (of mine) still presents the "idea of external existence." There is the lisere between it and the act which I "am." Or, to say the same thing differently, this conscious state is what it is by virtue of the act of perceiving₁ which is "in" it. Whatever else may or may not be "in" it, the molecular fact, its intention, certainly, certainly is not.
To claim that there are sense data is to claim that, just as we can shift from perceiving₂ to perceiving₁. If you made this shift, there is no longer an act "in" the conscious state which you "are." If the intention from which you shifted is Mp, call this conscious state {square brackets will here be used in place of bars over the contained expressions - sb}[Mp]. [Mp] is exactly like Mp. Only, the lisere has disappeared₂! This conscious state [Mp] is a sense datum.²⁷ One radical difference between such conscious states and all others is that there is no act "in" them. In

***** ²⁷ This is not to say that all philosophers who use the expression use it consistently as it is here explicated. This is a very major source of trouble. Even G. E. Moore uses it so ambiguously that sometimes it stands on the same page once for an entity which is in the mind, once for one which isn't. The cause of this ambiguity, which is so unusual with him, is of course his overpowering urge toward that adequate realism at which, alas, he never arrived. See also fn. 50.

325

another respect they are like all others. A fact, we remember, is a conscious state only by virtue of a direct awareness which intends it but is not "in" it. A direct awareness which intends a sense datum is not a species. When I am, as one says, sensing something, I "am" not the sensing but what I sense. This is only another way of stating the radical difference. But it is of course very easy to shift from a sense datum to a direct awareness of the sensing that makes it a conscious state.
There is still another radical difference between as [Mp] and all other conscious states. Call a fact intentional if and only if there is a thought "in" it. (The appropriateness of the word is, I trust, obvious.) All acts are intentional; all nonmental facts, nonintentional. So far I have proceeded as if all the atomic facts "in" a conscious state were intentional. Now we see that this is so only if there are no sense data. Sense data are nonintentional facts "in" conscious states. The real does not coincide with the nonmental. The distinction proved crucial. We achieved it by starting from, and holding fast to, the hard-core meanings of "mental" and "nonmental." Now we are ready for a second distinction of equal import. Are sense data mental? If we stay with the hard-core use, as of course we shall, the immediate answer is No. Those answering Yes get into trouble because they have without noticing it abandoned the hard-core use. Yet they try to express something which is important. I shall first provide ourselves with a safe way of expressing it, then show that it important.
(1) An entity is "in the mind" if and only it is "in" a state of consciousness. (2) An entity is "in the mind" if and only if it is the intention of a direct awareness which, depending on its intention, may or may not be a sensing. If you make (1) the explication of 'in the mind', (2) becomes a very fundamental truth. If you make (2) the explication, (1) becomes the truth. I shall make no other use of 'in the mind', (2) becomes a very fundamental truth. If you make (2) the explication, (1) becomes the truth. I shall make no other use of 'in the mind'. Those who call sense data mental try to express that they are in minds. Speaking as accurately as one must if the classical dialectic is to be resolved, one sees that a sense datum is a nonmental fact which is a mind . Every accurate terminology leads to some such clashes with the covertly philosophical way we often speak. To apologize for them is as foolish as it is for a soldier to be ashamed of his battle scars.
The entities in our minds are those which correspond parallelistically to brain states. They are never mentioned in the physical and biological sciences. They are those which interest the psychologist, who calls them "subjective." This is indeed the one clear use of 'subjective' I know. This not only shows the importance of the notion of being in a

326

mind; it also shows how easily the notion is confused with that of being mental.
Are there sense data? To ask this question, we now see, is to ask another. Can we make the shift from perceiving₁ to (as one says) sensing which I mentioned when explaining what sense data are if there are any? Or, if you believe that you can make the shift, is it as clear and distinct as it would have to be if your Yes is to be completely unproblematic? The answer depends on phenomenological subtleties which belong to the wrong kind of phenomenology. Hence, if I am right about these subtleties, the answer depends on phenomenological subtleties which belong to the wrong kind of phenomenology. Hence, if I am right about these subtleties the answer is dialectically as irrelevant as that huge catalogue of all the subtleties of all ordinary uses which is now being dreamt of by the worshipers of the Oxford English Dictionary.
Are there sense data? In this century (and, alas, in my own thought) the question has been so important that I undertook here to demonstrate its dialectical irrelevance. To have shown that it belongs to a certain kind of questions, although engaging, is not yet the demonstration itself but at best only its first step. Presently I shall take the others. Let us first attend to a historical situation.
With the rise of what now at Oxford passes for philosophy the battle against sense data grew fiercer and fiercer. Why all the sound? Why all this fury? The answer is as simple as it is ironic. Those who now attack sense data are all more or less explicit materialists. That is, they deny, more or less covertly, with or without bad intellectual conscience, that there are minds (conscious states). Sense data, we saw, are indeed problematic in that they are perhaps never as clearly and distinctly presented to us as are, on countless occasions, rocks, chairs, trees, and so on, which are all ordinary objects, as well as rememberings, perceivings, thinkings, and so on, which are all acts. Acts, that is, are not in the least problematic, whether or not sense data are. And if there are acts, there are minds. Why, then, have those who recently attacked mind by attacking sense data been so successful? Why have the defenders of sense data been so timid, even though they had at least the good sense to hang on to mind by hanging on to sense data? Once more, the answer is simple and ironic. The attackers and the defenders had both lost the act. That left them only the choice between (covert) materialism and (covert) phenomenalism. The attackers chose the former; the defenders, the latter. As a man grows older, if he believes that he has slowly been learning a few things, he becomes less and less certain even of what he believes to have learned well. Of this historical diagnosis, though, I am certain as the child was when it exclaimed that the emperor wore no clothes.
As we agreed to use 'external', sense data being nonmental facts, the particulars in them are external. Yet they are in minds, or, to bring

327

out the potential verbal clash for once, they are "internal to minds." You see again that when I asked you to agree on how to use certain words, I ignored, quite deliberately, the complications due to sense data. The adjective in 'external particular' was to remind us that the intention "in" which these particulars occur present us with the "idea of external existence." Sense data don't present us with it. That makes the clash so awkward that we must do something about it. Considering the way philosophers speak, the best thing to do is to introduce a new word. That will kill two birds with one stone by explicating the relevant use of the new word as well as spotting the verbal sources of certain confusions. Call a thing (particular or simple character) phenomenal if and only if it is in a mind, i.e., to repeat, if it is "in" a conscious state.²⁸ Call a fact phenomenal if and only if all the simples in it are phenomenal. Now we can without danger of confusion say such things as that the particulars "in" sense data are phenomenal (and external); that the (external) particulars in the intentions of all indirect acquaintance are nonphenomenal; that if there are no sense data,²⁹ then all external particulars are nonphenomenal; that acts are intentional phenomenal facts; sense data, nonintentional phenomenal facts; and so on. The criterion for being a phenomenal entity is contextual. Hence, phenomenal is not an ontological category. That alone shows that 'phenomenal' and 'mental' do not stand for the same. Yet philosophers have used the two words as if they did.³⁰
What is a mind (Self)? Whenever there is a mind, there is a temporal series of conscious states which are in certain ways connected among themselves as well as (parallelistically) with a body. E.g., some are memories of some others. Conversely, whenever there is such a series, there is a mind. That much is obvious. If it weren't, I wouldn't know what 'mind' means. Are a mind and the series which belongs to it two or are they one? Since there are no substances in my world, I agree with those who hold that they are one. But it makes no difference for what follows if we are wrong.
There are countless ordinary objects such that the particulars "in"

***** ²⁸ Some philosophers use 'experience' as I use 'conscious sate'. But the word has been batted about so irresponsibly that, since it is expendable, I avoid it.
²⁹ I ignore, as throughout this essay, certain phenomenal things and facts which the classical psychologists from James Mill to Wundt, taking their cue from the phenomenalists, call affective. These facts are "like sense data"; most importantly, they are molecular and nonintentional. See fn. 33.
³⁰ If so much "mere terminology" bores or even irritates you, let me remind you that to build, check, and recheck in this way a terminology that starts from the hard core is to track down the blurs and ambiguities which lead to philosophical uses.

328

them which are simultaneous with my writing this line are not "in" the intention of any act and therefore, a fortiori, not in any mind.³¹ If all minds now ceased to exist, ordinary objects would not cease to exist. There were such objects before any minds existed. Ordinary objects are not phenomenal entities. The table I now perceive, for instance, is not in my mind. Nor is it in any other.
All phenomenalists claim that all things are phenomenal. Hence, they must consistently deny that ordinary objects exist. That distinguishes them from the idealists who, we saw, need only claim that, being mind-dependent, ordinary objects, though existents, are not real.³² Typically, phenomenalists also claim that there are no intentional entities, or, equivalently, that all phenomenal facts are either sense data or "like sense data."³³ Both claims are absurd. Each suffices to dismiss phenomenalism. But it will pay if we first attend to three questions. How do phenomenalists assay the conscious state I "am" when perceiving a table? How do they account for the "idea of external existence"? How do they assay perceptual error?
First. The phenomenalists and the classical psychologists who took their cue from them all rely on analytical introspection. To introspect a conscious state analytically is not to inspect it but, rather, to produce under a special set a series of other such states. The idea is that the original state is uniquely determined by the series, and that, being so determined, it is a fusion of states exactly like those in the series and of nothing else. If you state this idea as of course a typical phenomenalist wouldn't, in terms of intention, replacing the original state by the intention of an original act, the series by the intentions of a series of acts, then, taken as a structural possibility the idea is not at all unreasonable. Yet it cannot be realized.
Take my perceiving the table. Its intention, we know, is 'Mp & Op'. That makes [Mp] the obvious candidate for the first member of the series. [Mp], you will remember, is exactly like Mp except that it is a phenomenal fact. Unfortunately, [Mp] does not uniquely determine the

***** ³¹ I put it this way, in terms of particulars, partly because of the old gambit that those countless objects are at this very moment intended by a thought of mine. The intention of *There are tables* is (Ex₁)...(Ex_n) Table (x₁,...,x_n) {for typographical reasons an uninverted 'E' is used in place of the standard existential quantifier - sb} and contains no particulars! See fn. 18.
³² That this way of drawing the line between idealists and phenomenalists also makes sense historically is shown by the case of Berkeley, whom it puts with the idealists, since for him an ordinary object is a cluster of universals which, though it exists, depends on a substantial mind to exemplify it. See E. B. Allaire, "Berkeley's Idealism," Theoria, 29, 1963, 1-16. ³³ See fn. 29.

329

original state. If it did, what would be the difference between my perceiving the table on the one hand and my thinking of it, in case the thought is accompanied by an image on the other? The many futile attempts at overcoming the difficulty all lie between two extremes. At one extreme, one adds to [Mp] a huge class of other sense data, which the mind supposedly produces by association, including all those, past, present, and future, that might be caused or might have been caused by the table. This is the absurd "solution" of James Mill. At the other extreme, one adds to [Mp] a series of alleged phenomenal entities which, whatever they may be called, structurally are like [Op]. Psychologists may want to call some of them expectations. The trouble with this "solution" is that sense data are essentially molecular facts. Thus, even if there were such entities, they would not be "like sense data" but, if anything, like acts.³⁴
Second When perceiving something, I "am" the perceiving. When sensing something. I "am" not the sensing but the sensed. Sense data (if there are any) are not perceived. When perceiving either the table or the ghost, I am presented with the "idea of external existence." Yet the table is real; the ghost is not. To be real (mind-dependent) and to present (if presented at all) the "idea of external existence" are two and not one. This is one of the crucial distinctions we have achieved. Phenomenalists cannot make it, must compound the two notions into a single one, which therefore remains blurred. In the next paragraph, 'idea of external existence' stands for the compound.
If the intentions of my perceivings cohere, so do the sense data which, if the phenomenalists were right, these perceivings would be. The inadequacy of the phenomenalists' assay of perceiving does not, we see, prevent them from utilizing the notion of coherence. They do in fact take full advantage of it, which permits them to explain how we come by the "idea of external existence."³⁵ To say this table is an external existent (or, synonoymously for them, that it is real or mind-independent) is, according to them, to say no more nor less than that the sense data which, as we would say, are caused by the table, cohere. Hence, if our perceivings did not cohere we would not, according to the phenomenalists, have the "idea of external existence." Contrast that with what has been said earlier (p. 311).
Third. We know the two hurdles the realist has to overcome in his

***** ³⁴ All this is very succinct, selective, and allusive. There is of course much more to be said, some of which is quite interesting as well as instructive. But why flog dead horses? For some details, see the essay on "The Problem of Relations in Classical Psychology" in The Metaphysics of Logical Positivism.
³⁵ I do not ask the embarrassing question what a sense-data assay of the "idea" itself would be like.

330

assay of perceptual error. There is the real particular exemplifying one character (e.g., round) which is presented as exemplifying another (e.g., oval). This was our second case. There is also the (nonphenomenal) external particular which is not real. This critical particular, though, may plausibly be believed to exemplify the characters it is presented as exemplifying. Or, if you prefer it that way, neither the belief that it does exemplify them nor the belief that it doesn't will ever conflict with beliefs based on coherence (p. 323). This was our third case, the recalcitrant one.
The phenomomenalist in his assay of perceptual error encounters no hurdles. For one, he believes that all the particulars presented to him are in mind; hence none can be critical. (In a world without acts nothing could be literally "presented." It would therefore be more accurate to say that a phenomenalist would hold this belief if he could consistently speak of anything being presented to him. But I shall continue to take advantage of the word.) For another, the phenomenalist has no reason to believe that any particular does not exemplify the characters it is presented as exemplifying. In the third case that follows from what has been said about the critical particulars. In the second case, one merely has to remember that the psychological laws (clause (3), p. 317) which enter into the notion of coherence explain why our sense data are what they are or, as one says, why they are what they seem.
A sense datum is indeed what it seems (is presented as). Many consider that as much a part of the notion of sense datum as its being in a mind. This is the famous infallibility of sense data . I have no quarrel with the notion, merely add two comments. 1. If being infallible is made part of the notion, then of course sense data are infallible. Necessarily so, as one says, but also trivially so. The only question that remains is whether there are sense data. As I introduced the notion, infallibility is not part of it. 2. To say that sense data, as I introduced the notion, are in fact infallible, we just saw, but another way of saying that our perceivings cohere. And coherence we know to be a fact which, however sweeping and reliable it may be, partakes in principle of the openness of induction.
I am about ready for a comment in structural history which, however, will be more convincing if we first glance at a world in which there are not only sense data but also acts. This is the world of the atypical phenomenalist. Acts and sense data are the two kinds of fact "in" the intentions of direct awareness.³⁶ This, though, is not the only feature

***** ³⁶ That ignores the great binder of the mental and the nonmental, time. An awareness and a particular in a sense datum being simultaneous is a phenomenal fact. But no harm will be done.

331

they share. Acts, too, are infallible in exactly the same sense in which sense data are! Coherence sometimes leads us to believe that we perceived erroneously or remembered falsely. But the perceiving itself and the remembering itself can always be accommodated in the pattern which coherence imposes upon the world of the atypical phenomenalist.
"Perception is fallible." The formula can stand for the skeptical critique of perception whose rival heralded the Cartesian revolution. "A mind can only know what is in it." The survival of the Aristotelian-Thomistic formula was major cause of the drift away from realism, toward either idealism or phenomenalism, which began at the time of the revolution. Acts and sense data are in the mind. Acts and sense data are infallible. Phenomenalism, including atypical phenomenalism,³⁷ satisfies the intellectual needs represented by the two formulae. Structurally, that accounts for two of its major strengths. Historically it stands to reason that these two major strengths of phenomenalism are also among the major causes of its rise.
The time from the Cartesian revolution to the present is also, alas, the age of epistemologism. Phenomenalism has a third feature which in that age became its third major strength and therefore, plausibly, the third major cause of its rise. Aphoristically and a bit paradoxically, it can be stated in one sentence: The phenomenalists discovered the right ontological assay of ordinary objects. We must next unpack the aphorism.
(1) An ordinary object is a temporal succession of particulars each of which exemplifies certain nonmental properties and stands in certain nonmental relations to all others. That makes 'table' a very complex character (p. 313). (2) The intention of my perceiving the table is Mp&Op. (3) There are no substances nor even continuants. (4) The right assay of causation is Humean in style. (1) is the right assay of ordinary objects. (1) and (2) entail each other. The structural connections between any two of these four propositions are very close. We need not on this occasion inquire what entails what. That these propositions are true is of course not argued but assumed in this essay. But it should perhaps be said that I call an assay the right one only if I believe that it is the only one compatible with an adequate ontology.

***** ³⁷ As I read Descartes, the hard-core base from which he tried to arrive at realism was indeed atypical phenomenalism. That is why he couldn't be certain that an entity existed unless it was presented to him in immediate awareness and why, therefore, since his concern was, classically, with acts and not at all with sense data, which are a later discovery, he fastened on a species, cogitare. The inference from cogitare to ens cogitans is a different matter. It rests on the old ontology of substance and attribute which only the later British critique showed not to be the "truth of reason" for which Descartes still mistook it. But I do not pretend to be a scholar. That is why I said: "as I read Descartes." See also fn. 4.

332

Return to the phenomenalist. Nothing will be lost if we let them all be represented by those who propose [Mp]&[Op] as the assay of the act which I "am" when perceiving the table. As an assay of the act that is hopelessly wrong. But one merely has to remove the bars {that is, square brackets - sb} in order to obtain the right assay of its intention. So we must ask: What does it mean either to add or remove the bars? If 'a' stands for a particular, '[a]' stands for one that is phenomenal. If 'f₁' stands for a color, 'f₁[a]' stands for a phenomenal fact.³⁸ '[S]' is obtained from 'S' by putting bars over all names³⁹ in it. If one removes the bars from [Mp]&[Op], one therefore merely drops the claim that ordinary objects are phenomenal entities. Everything else, as far as these objects are concerned, remains unchanged. Structurally, that unpacks the aphorism. In their peculiar way, the phenomenalists did discover the right assay of ordinary objects.
Historically, the credit for the discovery does not belong to the phenomenalists alone but, rather, to the succession from Locke to Hume, the movement of thought known as British Empiricism. But there are three reasons why the synecdoche does not go beyond what may pass in an aphorism. For one, the whole movement is under the spell of the formula that a mind can only know what is in it. For another, Locke's critique of such "abstract" entities (ideas) as substance and of the alleged truths of reason in which they occur could not but spread to powers (causes). Nor, thirdly, did anyone understand that there can be acts in a world without powers.⁴⁰ These three reasons make typical phenomenalism the inevitable consummation of the movement.
'Empiricism' is now used so broadly and so vaguely, so often merely honorifically or ideologically, that the word has lost all its savor. Originally, it stood for two important ideas, a specific thesis and an attitude. The empirical thesis asserts that our knowledge of ordinary objects comes to us through the senses. The empiricist attitude demands that all "abstract" entities as well as the alleged truths of reason in which they occur be securely grounded in what is presented to us.

***** ³⁸ As you see, it would not be necessary to put the bar over the names of universals. This is the strongest structural reason against there being sense data. But it is far from cogent. If you wonder what sense it makes to speak of structural reasons in matters of fact, I shall remind you that the "facts" of the wrong kind of phenomenology are peculiarly elusive.
³⁹ I.e., about the marks representing either particulars or simple characters. That presupposes that 'S' contains no defined descriptive terms. The connectives and operators are constituents of the world's form. Phenomenal-nonphenomenal is a dichotomy among the world's things.
⁴⁰ The argument for this compatibility is one of the major burdens of Essay One.

333

The thesis and the attitude are the two cornerstones of every epistemology worthy of the name. During the reign of epistemologism their combined strength was at least as great, if not greater, than that of any other force which shaped the classical dialectic. Thomas, I take it, upheld the empiricist thesis. So did Leibniz. Nihil est in intellectu quod non prius fuerit in sensu. The famous qualification, nisi intellectu ipse merely safeguards the act. Yet the men in the British succession were the only ones to embrace vigorously both thesis and attitude. Recalling the close connections among the propositons (1), (2), (3), (4), you will see that the phenomenalist' assay of ordinary objects involves them both. That makes the peculiar "rightness" of this assay their third major strength.
Are there sense data? I promised to demonstrate that dialectically it makes no difference whether the answer is Yes or No. If you have forgotten that promise, let me remind you of it. If you haven't, perhaps you still wait for the demonstration. In case you do, let me present you with three propositions which, if true, jointly are this demonstration. First. In case there are no sense data, phenomenalism is sheer nonsense. Second. In case there are sense data, it is absurdly false; atypical phenomenalism because ordinary objects are not phenomenal entities; typical phenomenalism also because there are acts. Third. In either case, the phenomenalist assay of ordinary objects becomes the right one if the particulars it mentions are taken to be nonphenomenal entities. The first two propositions are the destructive part of the demonstration; the third is the constructive part.
Are ordinary objects phenomenal entities? Some tell us that we must not ask the question. Nor must we ask whether there are sense data. Both questions are "metaphysical" and therefore "meaningless." The right question to ask instead is: Can one in principle talk about ordinary objects in a sense-data language? The answer, we are told, is Yes. This "linguistic" defense of sense data against the recent attacks by the ordinary-language philosophers surely is the weakest and meekest of all. One can and one must explicate the relevant use of 'phenomenal' and 'sense data'. Then both questions - Are ordinary objects phenomenal entities? Are there sense-data? - become factual. The only thing peculiar about the second is the peculiar elusiveness of the wrong kind of phenomenology. The linguistic defense, we see, is not only weak and meek; it is completely futile. Yet there is an idea behind.
A scheme is an ideal language if and only if (1) everything can in principle be said in it, and (2) all philosophical problems can be solved by talking commonsensically about it. Both (1) and (2) are familiar; both need unpacking; but we need not unpack them once more. I

334

merely restate them in a way that suits the immediate purpose. (1') There is a "coordination" of a certain kind between certain sentences of the language we actually speak on the one hand and the formulae of the schema into which these sentences are said to be "transcribed" on the other. (2') The schema perspicuously reflects an adequate ontology.
In a schema that contains such expressions as 'Mp&Op' everything about ordinary objects can in principle be said. Thus it partially satisfies (1). Such a schema is called a sense-data language if and only if (a) it contains no names for mental characters (species and thoughts) and neither 'M' nor the corners {here we use asterisks - sb} (*...*); and, (b) it names all stand for phenomenal entities. To restate (b), making use of an earlier device, the bars {brackets - sb} have been added to the names in the schema. But, then, either adding or removing the bars obviously does not affect the possibilities of a "coordination" of the kind mentioned in (1'). That shows the peculiar way in which everything about ordinary objects can be said in a sense-data language. Not surprisingly this peculiar way corresponds exactly to the peculiar way in which the phenomenalists' assay of ordinary objects is right. A sense-data language also partially satisfies (2). One can by talking commonsensically about it solve many philosophical problems. Many, but not all. To see that at a glance, switch from (2) to (2'). In the ontology the schema perspicuously represents there are neither acts nor nonphenomenal particulars.
The last paragraph merely restates in the "linguistic way" what we knew already. Yet this recourse to artificial languages serves two purposes. For one, ironically, it shows that if one wants to demonstrate the complete futility of the "linguistic" defense of sense data, one merely has to state it, correctly, in the "linguistic way." For another, it provides the cue for attending to some unfinished business.
The schema which which until now I claimed to be the ideal language reflects perspicuously the ontology of atypical phenomenalism.⁴¹ That makes

***** ⁴¹ Or, to say the same thing differently, I insisted for the names of the ideal schema on the Principle of Direct Acquaintance. The principle implicit in this essay is one of Presentation. Making it explicit is hardly worth the trouble. Another point, though, deserves unpacking (or, more accurately, rehearsing).
Since the obeisance I have paid to the incomplete schema has been a rather empty gesture for quite some time, it did no harm. But, then, the emptiness of this ritual was not just a piece of good luck. Or, if it was, there is a lesson in it. The fundamental dualism is intentional-nonintentional. Compared with it, the dualism between what is and what is not in minds is shallow. That is the way the lesson is put in the opening essay. Within minds "in" which there are acts, sense data are an adequate model of what is not in any mind. That is the way the lesson is put in Essay Six. Structurally one is a realist if one knows how to assay that act. That is the way the same point is made, concisely, in the second paragraph of this essay.

335

it adequate in all respects but one. Hence it is incomplete and cannot be the ideal language. I now complete it by adding to it names for nonphenomenal particulars. Whether or not names for phenomenal particulars which are not awarenesses (i.e., for particulars in sense data) are retained makes no dialectical difference. Thus I get rid of that last piece of phenomenalist dross which is mentioned in the second paragraph of this essay. I also understand now, in retrospect, why this last step took so long. One cannot make it unless he has first purged his thought from the last taint of epistemologism. For one who comes from where I started this is very difficult.⁴²
Unlike idealism and phenomenalism materialism, the last of the four classical ontologies, is not only absurd but also very dull. Yet, since we have assembled all the tools and since it will not take long, we might as well render materialism a service by distinguishing what is absurd in it from what very probably is true. We shall be rewarded by a clear view of an instructive similarity between idealism and phenomenalism on the one hand and what passes wrongly for two variants of an ontological position on the other.
Materialism₁ (metaphysical materialism) is an absurd ontology. Materialism₂ (scientific materialism) is not really an ontological position but merely the best scientific truth available. Science is not ontology. (That is why, in the paragraph which precedes the last, 'wrongly' is italicized.) In a familiar scene, no scientific truth is more than very probable, least of all one as sweeping as the truth in question which, because of its sweep, I would rather call a frame of reference. (That is why, in same paragraph, 'very probably' is italicized.) Yet, we have no good reason to doubt this frame of reference. Philosophers who

***** ⁴² There are many among us who have come by their ontologies very easily. Mostly they derive it from the later Husserl. Tomorrow they shall perhaps derive it from the later Wittgenstein. Such ontologism, alas, is merely the battle cry of the latest rebellion against dialectical discipline in general and epistemological discipline in particular. Looking at this spectacle, I remember with a very warm feeling the debt of gratitude which I owe to the logical positivists.

336

pin their hopes on its collapse are merely quixotic. Good philosophers take the world as they find it.
The word 'real' which I used, deliberately of course, promotes the illusion that the claim of scientific materialism is ontological. Without that word, the claim is simply that minds depend on bodies. That is what both common sense and its long arm, science tell us. The dependence is of the kind 2b, parallelistic. There is of course much more to be said. This is the dialectic of the mind-body problem, which requires and deserves a book of its own.⁴²
Materialism₁ and phenomenalism are mirror images of each other. How could one who insists on taking the "linguistic way" express this feature? The answer is again instructive. Just as in principle one can say everything about ordinary objects in a sense-data language, so one can, if parallelism is true, in principle say everything about minds in what the logical positivists call a thing language, i.e., in a schema without names for phenomenal entities. Smith perceives a table. He is in this conscious state if and only if his body exemplifies a very complex property which, being nonphenomenal, can be represented by an equally complex predicative expression of the thing language.⁴⁴ 'Being in the (bodily) state-of-perceiving-a-table' may be introduced into this language as a definitional abbreviation for that predicative expression. Having provided yourself with a sufficiency of such definitions, you can after a fashion say everything about ordinary objects in a sense-data language. Yet there are two differences. That is why I say after a fashion.
First, The phenomenalist can in his way say everything about ordinary objects because he discovered (what I believe is) the right assay of them. Whether or not this assay is in fact right is an ontological question. The materialist can in his way say everything about minds because (as I believe) parallelism is true. Whether or not parallelism is in fact true is a scientific question. Second. The phenomenalist, having introduced his definitions, can in the sense data language say 'This is a table'. After the materialist has introduced his definitions, he cannot

***** ⁴³ A very large part of this dialectic belongs to the philosophy of psychology about which I have already said many things. Some of these things I can now say more simply; some others, more accurately; still others should be added. Perhaps I shall be able to undertake this task.
⁴⁴Very probably, the (perceptual) thing language must for this purpose be supplemented by that of physics. This, though, makes no difference for the argument. For the distinction perceptual-physical, see the next paragraphs.
⁴⁵ See fn. 18.

337

in the thing language say 'Smith perceives a table' but merely 'Smith's body is in the state-of-perceiving-a-table', where the hyphens indicate that there are no substitutions salva veritate for any of the words in the phrase. If this were not so, one could transcribe an intentional context into an extentional language.⁴⁶
These two differences alone will convince any philosopher that dialectically materialism is very dull. That is why, if faced with the unhappy choice between phenomenalism and materialism, as I was when I started, he will choose phenomenalism. Materialism, alas, is a philosophy for nonphilosophers. That is why scientism has undone so many.
We are ready for the last turn of the dialectical screw. It starts from the distinction between perceptual and physical objects. Until now I deliberately avoided both words. To be an ordinary object and to be a perceptual object is one and the same. Science (physics) replaces the perceptual object by the physical object. This is a long story. It is told in Essay Six. So I merely recall its gist.
The table I perceive is colored. The physical table lost its color at the time of Descartes.⁴⁷ This, though, was only the beginning. By now, as I understand the quantum theory,⁴⁸ there is not even space in "physical reality." To find out what physical reality is like is the scientists' business. Our concern is merely with their use of 'real' and its derivatives.
Science replaces the perceptual object by the physical object. To appreciate the import of the italicized prefix, introduce 'reconstruct'. Science reconstructs the perceptual object. One cannot replace or reconstruct what is not there. According to the phenomenalists, the perceptual object, which is not there, is a construction out of sense data. According to the "positivistic" analysis of physical theory, the physical object is a construction out of perceptual objects. If you accept both views, phenomenalism and the "positivistic" analysis of physical theory, the physical object is a construction out of sense data and not, as I hold, a reconstruction of the perceptual object. That explains the affinity between those two views.

***** ⁴⁶ See Reinhardt Grossmann, "Propositional Attitudes," The Philosophical Quarterly, 10, 1960, 301-12; also the essay on "Sameness, Meaning, and Identity" in Meaning and Existence and May Brodbeck, "Meaning and Action," Philosophy of Science, 30, 1963.
⁴⁷ The historical (causal) connection between the rise of science, that of epistemologism, and the Cartesian revolution is plausible, to say the least.
⁴⁸I.e., there is no physical space if the entities of quantum mechanics are taken to be physical reality. About this feature of the scientists' use of 'real', see "The Logic of Quanta," reprinted in H. Feigl and May Brodbeck, Readings in the Philsosophy of Science (Appleton Century Crofts, 1953).

338

Perceptual objects present us with the two ideas of external existence and reality (mind-independence).⁴⁹ Only because we are thus provided with these two ideas, can we combine them with that of the physical object. Scientists use 'real' as if to be real, to be an existent not in a mind, and to be a physical entity were one and the same. With science, the ambiguity does no harm. In ontology it leads to disaster.
We do not know physical objects, in the sense in which we know perceptual ones. The gap is as unbridgeable as that between that mythical entity, the "percept" of representative realism, and the entity it allegedly "represents." It is indeed the fatal weakness of representative realism that it displaces this gap between the perceptual and the physical object, puts it between the mind and the perceptual object, and then tinkers with the latter in order to resolve the dialectic of perception. Just think of G. E. Moore who considered, seriously even though reluctantly, such "Pickwickian" possibilities as the coin being a bundle of sense data,²⁰ round, oval, and so on, from which the perceiving mind selects one. This kind of tinkering is the business of science, not of philosophy. The proper place for it is the gap, which is in the nature of things, between the perceptual and the physical object rather than the alleged gap between a mind and what it perceives.
For the scientist there is nothing not in minds which is literally the color of the perceptual object. As for colors, so for all nonmental characters. That makes the perceptual object "mind-dependent" in what, where and when I grew up, they called das Weltbild der Wissenschaft. This "picture" is not an ontology. Ontology is phenomenological. The rest is merely science. Such is the nature and such is the limits of human knowledge.
We can of course think of physical entities. Such thoughts may be accompanied by "images" of all kinds. Yet they remain peculiarly "abstract." For instance, we are never presented with a physical entity in a manner allowing us to name it. Nor is that the only striking peculiarity of such thought. Inaccurately one may express them all by saying that there is no intentional tie between a physical entity and the

***** ⁴⁹ This is not to deny that what we know to be of strategic importance, that a single act of indirect awareness presents us with the "idea of external existence." But, then, perceptual objects are of course very prominent among those which do present us with this idea.
⁵⁰ These are of course passages in which Moore's "sense data" are not in minds, which merely increases the confusion even though it also shows his sound distrust of representational realism. See fn 27. The rejection of representational realism is, quite probably, the sound core of the ordinary-language philosophers' attack on sense data. If so, why don't they say so? Representative realism and the existence of sense data do not, in spite of the historical connections, entail each other. And philosophy, I submit, is still the dialectical art of distinction.

339

thought which intends it. Imagine now an ontologist who, while not a materialist, is yet so preoccupied with science that in his world all things are either phenomenal or physical. In such a world, we just saw by means of a deliberate inaccuracy, the intentional is easily lost sight of. If this tie is lost (and the Aristotelian-Thomistic assay of perception seems no longer viable), then, we know, realism is lost. Our ontologist will eventually become either an idealist or a phenomenalist or, if he has the stomach for it, a materialist.
At the time of the Cartesian revolution, when the trend toward epistemologism and away from realism began, philosophy found itself in the situation of the ontologist whom I just asked you to imagine. Dialectically that closes the circle. Such closure is encouraging. The cross of the realism-idealism issue may be lifted off our shoulders. The reign of epistemologism may be drawing to a close.
Ontology, I said, is phenomenological. The rest is merely science. Such is the nature and such are the limits of human knowledge. I do not want these sentences to be misunderstood. There is now a rising wave of talk about the Lebenswelt. What is true in this talk is not philosophical; what is philosophical in it is not true; the bulk of it is a mixture, a la francaise, of ignorance, antiscience, and mediocre literature. Today is supported by the spurious authority of the later Husserl, tomorrow perhaps by the equally spurious one of the later Wittgenstein. Any similarity between this sort of talk and the sentences I do not want to be misunderstood is purely incidental, just as I did not endorse either scientism or philosophical behaviorism or any other kind of materialism when I gave science its due. Philosophy, I also said, takes the world as it finds it. The most likely present threat to first philosophy is an unholy alliance among those who now propound these two opposite kinds of error. But, then, first philosophy, the fundamental dialectic, never attracted any but the happy few. Thus its position has always been precarious. There is no more cause for alarm than usual.
Having said that much, I shall say one more thing in order to forestall another misunderstanding. The largest group of contemporary philosophers who stand against that likely threat are the heirs of Thomas and Aristotle. With them I share the realism and the recognition of the act. Yet, fundamental as these two agreements are, they are but two islands in a sea of disagreement. Der Starke ist am machtigsten allein.

340

Home

REALISTIC POSTSCRIPT1

REALISTIC POSTSCRIPT¹