SOURCE: “Berkeley's Idealism: Yet Another Visit,” in Berkeley's Metaphysics: Structural, Interpretive, and Critical Essays, edited by Robert G. Muehlmann, The Pennsylvania State University Press, 1995, pp. 23-37.

[In the following essay, Allaire surveys the ontology of Berkeley's philosophy of idealism, and why it fails.]

In these remarks, I try to show that Berkeley's idealism was inevitable and that its failings continue to be instructive. Being too long indifferent to epistemological matters, I have only recently come to believe the former. All ontological solutions to the problems of knowledge issue in idealism or a variant of it, I think.

Berkeley's ontology contains two basic kinds: minds and ideas, as he calls them. The former—also called spirits, souls, and selves—are acknowledged to be (mental) substances (PR [A Treatise Concerning the Principles of Human Knowledge] 7); the latter, except for some simple ideas that are deemed qualities, though not relative to substances, are not otherwise categorized by Berkeley. Why that is so is of concern later.

I want now to develop a simple representational device in order neatly to exhibit some features of Berkeley's ontology. Let lowercase exes with subscripts stand for minds; lowercase efs with subscripts, for simple ideas. That “vocabulary” suggests that the schema has only one combination rule, ‘fx’. That cannot be. Berkeley distinguishes between simple ideas and collections of them, the collections serving primarily to assay trees, mountains, chairs, and so on ontologically. Berkeley also distinguishes between sense, imagination, memory, and emotion ideas.

Let ‘s’, ‘i’, ‘m’, and ‘e’ be used as superscripts of ef signs, thereby enabling the schema to reflect the different kinds of simple ideas. Also, let a pair of brackets, between which are ef signs, stand for a collection of simple ideas. Now for several comments: (1) Some ef signs agree in subscript, differ in superscript. Speaking loosely, a simple idea—a shade of red, say—may be a sense as well as an imagination idea. (2) Every bracket expression contains more than one ef sign, and all the ef signs in a bracket expression have the same superscript. Again speaking loosely, an idea composed of the idea red and the idea square is not half imaginary, half real. (3) Different bracket expressions may contain the same ef signs. That is, different collections may contain one and the same simple idea. (4) I ignore the rules for specifying which combinations of ef signs are permissible within a pair of brackets. I ignore a great deal. For example, one would have further to distinguish between ef signs by using, for example, more superscripts in order to reflect the differences between a color, a shape, an odor, a texture, and so on, differences that would enable the schema to ground, for example, that everything colored must have a shape. Much that I ignore is interesting, but what is of concern on this occasion is dialectically independent of what I ignore.

I turn now to combination rules. Bracket expressions are signs, notwithstanding that they contain simple signs. In other words, a bracket expression cannot stand alone to represent a thing or complex or whatever. Accordingly, one formation rule is ‘x[]’. Another is ‘xf^e’. Neither a solitary ‘f^m’ nor a solitary ‘fⁱ’ may combine with an ex sign. I shall also assume that ‘xf^s’ is not a combination rule, an assumption to be remarked on later.

The schema used to exhibit Berkeley's ontology is plainly odd, not at all of the customary sort. First, it contains two levels of vocabulary, the bracket expressions being made from simple ef signs, using rules, even though the use of them yields signs, not sentences. Second, there are two formation rules but only one “tie,” if I may so speak. That is, the connection between a mind and an emotion idea is the same as the connection between a mind and a collection idea, be the collection a sense, an imagination, or a memory collection. Somewhat differently, ‘f^e’ and ‘[]’ expressions stand for items of the same ontological kind.

Since Berkeley categorizes minds as substances, one is likely to insist that ‘f^e’ signs and bracket-expression signs stand for accidents or qualities or properties and that the connection between a mind and an emotion or a mind and a collection is that of inherence. That, to be sure, would nicely explicate Berkeley's claim that sensible objects cannot exist without the mind or unperceived by the mind. There is, of course, something right about using ‘inherence’ as the word for the connection, but there is something wrong about it as well. For now, Berkeley's dependence claim is best left as merely reflected by the schema. Nothing is gained at this stage by insisting on introducing a word for the connection. Indeed, no word is wholly suitable, just as no traditional category term is wholly suitable for ideas. Berkeley himself prefers ‘experience’ for the connection, though that word is particularly unsuitable for ontological purposes.

The foregoing schema neatly allows for depicting the truth makers of such (“true”) sentences as ‘Michelle imagines a red square’, ‘Michelle remembers a red square,’‘Michelle sees a red square’, and ‘Michelle feels pleasure’. (The examples are, of course, rough and ready. I take ‘Michelle’ to stand for a mental substance and let a red square serve for a tree or table or whatever. The liberties are harmless, I believe.) The differences between seeing, remembering, imagining, and feeling are all on “the side of the object” and reflected by the differences in superscripts that encode the differences in kind of the simple ideas, thus serving to capture Berkeley's talk of liveliness (PR 30), or vivacity, to use Hume's term. That is as it should be: Berkeley does not have mental acts, either as relations or as qualities.

The reason for not using ‘accident’ or ‘quality’ as the category word for emotions or sensible objects is that Berkeley would construe the above examples to express that Michelle experiences this or that and not to attribute a quality to Michelle. Statements about minds are not like statements that express, for example, what color an object is. Somewhat differently, ‘Michelle is in (or has a) pain’ is best understood as ‘Michelle feels, experiences, pain’. Berkeley thus prefers to use ‘perceive’ or ‘experience’ as the word for the relation between a mind and “its object.” He nonetheless wishes to insist that sensible objects cannot exist without or independently of a mind; and that insistence is rooted in his being driven to categorize sensible objects as accidents of the mind even though they are not, in everyday talk, attributed to the mind. In a substance ontology, there is room for only one genuine relation: namely, inherence. (By a “genuine relation” I mean one that is not represented by means of a sign that is classified as a sign for an accident. That the only genuine relation in a substance ontology is a “tie” is pleasantly ironic.) What, however, is the basis for the dependence claim? Why is Berkeley an idealist?

On the variant of the ideal-language method I am presently using, one designs a schema in order to represent the truth makers of the sentences that are assumed to describe the world, and to represent the truth makers such that the ontology of the world is shown by features of the representation. The schema is for ontological, not logical, purposes. The nature of validity is, so to speak, of no concern whatsoever. Relatedly, the aim is not to “translate” ordinary talk into the schema or symbolism.

A far as concerns the description of the world, that is a datum; and an ontology is given for a world; but what the world is, for which the ontology is given, is not an ontological issue. (Whether there are lions or tigers is not an ontological issue, though whether they are substances or congeries is.) I thus do not challenge Berkeley's description of the world, except to remark on how impoverished the world of the Principles is.

The world for which Berkeley provides an ontology is one whose description is exhausted by statements about what minds experience. (Statements, suitably recast, about trees, tables, and so on are construed to “refer to” that which is experienced!) One class of statement conspicuously absent is that to the effect that someone knows he or she is in pain or knows he or she is seeing a white wall or the like. (Also absent are statements to the effect that a person is thinking or believing or judging such and such.) The absence is striking; after all, Berkeley is intent on providing an account of knowledge. He is aware of the absence, though. He waves at but does not wrestle with it (PR 30, 89; and part II of the Principles is absent). That is understandable: ‘xx[]’ is not a combination rule. To put the point simply, a mind cannot experience itself as experiencing an emotion or sensible object. I could, of course, have simply noted that Berkeley cannot account for a mind's knowing that it exists (‘xx’ is not a combination rule); but that failure is not the interesting one. Berkeley cannot accommodate one of the necessary conditions for a mind's knowing that it sees a tree or a chair or a stream.

Let me now explain why I exclude ‘xf^s’ as a formation rule. Sentences such as ‘Michelle senses red’ and ‘Michelle senses the odor of a banana’ do not occur in the description of the world for which Berkeley gives an ontology. That may strike one as mad, since it plainly conflicts with what Berkeley says in the very first section of the Principles. Let me explain my willingness to be mad. (1) If such sentences do occur in the description of the world, Berkeley's discussion of abstraction runs aground. (2) PR 1 is used in Berkeley's argument for idealism, and that argument is not the heart of the matter. About that, more later. (3) PR 1 is a clear invitation to Kant; it tempts one to dream up a causal or quasi-causal explanation for how a mind constructs or makes sensible objects. Berkeley is best spared that kind of nonsense. Besides, such “explanations” have nothing at all to do with ontology. (4) Berkeley's use of ‘idea’ in PR 1 is wholly unanchored.

I need now to make several more framework comments. (1) One can do without superscripts on the ef signs, except for ‘e’, by using ‘s’, ‘m’, and ‘i’ as superscripts on bracket expressions. (2) I ignore how God is to be differentiated from “finite” minds; indeed, I construe Berkeley's world and its ontology as if there were no God.¹ (3) I ignore Berkeley's causality and activity talk; it is vacuous as far as concerns ontology. (4) I ignore for now how Berkeley might individuate sensible objects and how he might solve the problem of identity through change of sensible objects. (5) I have given Berkeley a solution to the one-many problem by allowing the same ef sign to occur in different bracket expressions. (6) I have also allowed for different minds to experience the same emotion or the same sensible object. Have I in (5) and (6) been too generous? I think not, though I confess that my largess flows from my having no sympathy at all for the view that Berkeley is a nominalist. Apart from nominalism's being foolish—fashionable though it is these days, all decked out in tropes—there is no point in burdening Berkeley with the bewildering complications that are unleashed by committing him to nominalism.²

Berkeley's thought is driven by his preoccupation with skepticism: more specifically, by his attempt to defend what he takes to be a truism, namely, that people—gardeners and carpenters—sometimes know that an object is a rose or a lily, a chair or a table. His defense of that truism is guided by what I call the comparison model of knowledge (CMK), the first appearance of which is in Theaetetus (191c-195b). The model dominates the tradition and underlies much of what goes on in Descartes and Locke. The structure of the model is as follows: A person applies a word to an object. The word is true of the object provided that the object and the standard for (or the meaning of) the word match, or agree, in the appropriate way; and a person knows the word is true of the object—knows that the object is W—provided that he or she discerns the match.

Here is an illustration of CMK. A “yardstick” is a standard for ‘one yard (long)’, as well as for a family of measurement words. Suppose a person applies ‘one yard’ to an iron rod, says of the rod that it is one yard. The word/phrase is true of the rod provided that the end points of the stick and those of the rod coincide; and the person knows that the stick is one yard (in length) provided that he or she discerns the coincidences of the end points.

On the CMK, a necessary condition for knowing is that the knower experience the object about which he or she has knowledge. Doubt is present or can, at any rate, be engendered whenever the object about which a claim is made is not experienced, not in plain view.

In the tradition, the model was so construed that the standard for a word was “in the mind” and was called a concept or an abstract idea or an idea or a notion. Though few philosophers paid much attention to the word, their thought was nonetheless guided by the model.³ The model, I should say, has a deep hold on us quite apart from its presence in the tradition. For example, alleged cases of ESP are puzzling because they are exceptions to the CMK.

Berkeley interprets Locke and Descartes to hold that everyday physical-object words (‘chair’, ‘apple’) are applied to objects that cannot be experienced. For them, judgments to the effect that something is a tree or table express that that which is experienced or sensed has a correlate, the latter matching or failing to match the former (if indeed there is a correlate). That which is experienced (sensed, perceived) is taken to be both a sensed object and in effect a standard. Berkeley's strategy is simple: a physical-object word is applied to that which is experienced (PR 1), and that which is experienced is not a representative entity, for it cannot be like anything different in kind from what is experienced (PR 8).

Skepticism is rooted, supposedly, in representationalism; and to preclude the former, one must reject the latter by collapsing the idea-thing dichotomy. Berkeley collapses it to the side of ideas. Why? The simple answer is that he merely follows his predecessors in using ‘idea’ as the category word for sensed objects and thus drifts or stumbles into idealism. That answer, though somewhat suggested by the early sections of the Principles, is not very illuminating, if only because Berkeley himself is of at least two minds concerning his use of ‘idea’. His idealism stems instead from his categorizing minds as substances.

Recall that ‘x[]’ is a combination rule. In order to avoid idealism, that is, to collapse the idea-thing distinction to the side of things, one would, first, have to allow that instances of ‘[]^s’ may stand alone in representing part of the world and, second, have to introduce a relation such that ‘xR[]^s’ is a rule (‘R’ being interpreted as, say, seeing). A substance ontology, however, has no room or place for a relation other than inherence, as noted above, and as Berkeley himself knew (PR 89). If one insists on having the object of knowledge related to the knower and the knower categorized as a substance, then idealism is inevitable.

One might imagine that Berkeley could have allowed instances of ‘[]^s’ to stand alone and could also have allowed ‘x[]^s’ to be a rule. But that would make no sense. If a sensible object is to be related to a substance, the sensible object has to be thought of as an accident of, and thus as dependent on, the substance. One cannot allow for an entity's being both dependent and independent in an ontology that countenances substances. That is a moral of Spinoza's Ethics, one that he neatly expresses as Axiom 1: “All things that are, are either in themselves [substances] or in something else [modes, affections].”⁴

Representationalism and thus skepticism were latent throughout the tradition, becoming manifest only after Galileo and Descartes managed to give plausibility to the claim that the cause of that which is sensed need not be like that which is sensed in order “causally to explain” why one senses what one does. On a substance ontology, the only items that can be experienced are accidents; therefore, knowledge of objects must be indirect and representational. Skepticism was a crisis waiting to happen; idealism was its inevitable offspring.⁵

Berkeley's idealism is a consequence of his attempt to defeat skepticism while clinging to a substance ontology. In order to depict the truth maker of a physical-object judgment as something the mind experiences or with which the mind is “directly acquainted,” Berkeley has no choice but to insist that the truth maker is mind-dependent and thus in effect an accident of the mind. However, he cannot explicitly categorize the truth maker an accident: chairs and trees are not ascribed to the mind.

The rudimentary structure of a substance ontology is as follows: Given a subject-predicate sentence, the subject of which is used to refer to an individual, the subject is said to refer to a substance, the predicate to an accident, and the sentence expresses that the instances of the kinds are related by inherence. (I ignore the complications deriving from distinguishing between accidents and forms.) Berkeley knows that that which is expressed by ‘Michelle sees a white horse’ cannot be analyzed in the way that that which is expressed by ‘Michelle is walking’ may be analyzed. And that is so even when ‘seeing a white horse’ is construed to stand for a complex having a suitable mark of vivacity. The fact is, Berkeley does not quite know what to do with that which is expressed by ‘Michelle sees a white horse’. He knows only that he cannot analyze a white horse as a substance having accidents, since that leads to representationalism and thus to skepticism.

Berkeley does not, I believe, break with the substance ontology. True, he gives a nonsubstantialist, immaterialist, analysis of a tree or a chair; but he does so only in order to make the object mind-dependent and thus in effect an accident. Whether or not he did so wittingly, Berkeley, I submit, embraced Spinoza's first axiom.

A more generous, perhaps more illuminating way of stating the point just made is to say that Berkeley tries to use a substance ontology for a purpose that cannot be served by such an ontology. The ontological depiction given by Berkeley for the truth maker of, say, ‘Michelle sees a green leaf’ does not mirror the grammatical structure of the sentence. The sentence is relational, and thus no substance ontology can handle it. Berkeley's predecessors managed, however, by construing the sentence as ‘Michelle is seeing a green leaf’ and then analyzing the truth maker as a substance having an accident of the “idea of” sort; but that leads to skepticism. Berkeley must thus analyze the truth maker of the sentence as a substance having an accident, which accident cannot, however, be ascribed to the mind.⁶ For Berkeley, then, the ontological representation in no way mirrors the grammar of the sentence. Berkeley deserves celebration, not scorn: he begins the assault on the conviction that grammar is a reliable (the only?) guide to ontology. That Berkeley fails to appreciate the consequences of that assault should not distress us: seminal philosophers rarely know their offspring.

Berkeley is no happier with his use of ‘idea’ as the category term for sensible objects (PR 38). He sometimes remarks that his unhappiness is owing to the fact that his use of ‘idea’ yields nonsense: for instance, ‘A chair is an idea’. That is not much of a reason; mixing ontological talk with everyday talk always begets nonsense. He has a much deeper reason for his unhappiness, though. One premise of the argument that opens the Principles is that an idea cannot exist unperceived or without the mind. He is entitled to it only if he is using ‘idea’ in the “idea of” sense; but he is well aware that is not how he uses ‘idea’.⁷ For him, an idea is not of anything. He thus uses ‘idea’ in a new, philosophical sense, shorn of ‘of’, and thus must acknowledge that on that use he has no ground for the above premise. In the third dialogue (D [Three Dialogues Between Hylas and Philonous] 235) Berkeley remarks that he used ‘idea’ because a necessary relation to the mind is understood to be implied by that term. In so remarking, he suggests that he has available an argument for idealism that does not make use of ‘idea’. The argument is hard to find; however, if one reads PR 86-95 with no more than a hint of sympathy for the dialectic, one sees there the only argument that Berkeley could give; and it is the nakedly transcendental one I have given above: idealism is the only way out of skepticism.

Berkeley's organization of the Principles is puzzling in the light of the dialectical flow of his thought. Surely he would have been better off to begin with a discussion of skepticism and representationalism. Had he done so, he could have explained his odd use of ‘idea’ and spared himself the bad arguments that depend on that use. (PR 1-7 is a torturous stretch of writing. It needs an exhaustive commentary.) Of course, had Berkeley so organized the Principles that it began with a discussion of his central problem, thus making the dialectic transparent, he would have been an exception to what seems to have long been the rule: first things last.

Why did Berkeley categorize minds as substances? Why did he not stick with the congeries analysis he entertained in the notebooks? And could he have avoided idealism had he insisted on a nonsubstantialist account of mind? These are rather daunting questions, and on this occasion I shall do no more than broadly sketch some guesses. (1) Berkeley is not a Humean when it comes to cause. He needs an agent, something that can be said to be active. Only substance will do.⁸ (2) A congeries account of mind allows one to avoid idealism only if one can analyze experience as a relation that is external to the relata and such that the relata can exist independently of the relation. Berkeley is not able to master such a view of relations; indeed, no one even thought that they could do so until this century. (3) Had Berkeley allowed a sensible object to be a part of a “mind congeries” and also held that the part could exist independently of such a congeries, he would not have avoided idealism, at least not in any more than a verbal sense: that which can be a part of mind has to be categorized as mental. More accurately, Berkeley would not have been made any happier by the claim that chairs are mental than he was by the claim that chairs are ideas (and thus cannot exist unperceived or without the mind).

I return to Berkeley's idealism. There is nothing wrong with it as far as it goes, which is not, as I have indicated, very far; and, as I show shortly, it goes even less far than initially said. Nonetheless, Berkeley's ontology is able to exhibit truth makers for such sentences as ‘Spina sees a red ball’, ‘Tietjens imagines Sylvia's face’, and ‘Valentine remembers Mark's face’. The superscripts are powerful devices indeed. Furthermore, the schema can be supplemented such that the individuation problem is solved—the problem, that is, of depicting the truth maker of, say, ‘Spina sees two red squares’ (same size, etc.). Bare particulars will do nicely! (It is sheer prejudice to insist that the members of a collection must be of the same kind, a prejudice that stems from the thought that the ontological depiction of an object must mimic the grammatical form of the sentence used to describe the object, such that ‘This is red’, about a square, must be so analyzed that the demonstrative refers to the particular, ‘red’ to a quality.)⁹ Berkeley's ontology can be made to accommodate a good deal, but it cannot accommodate an account of knowledge. Berkeley's idealism simply cannot do what it was designed to do.

Berkeley's idealism secures only what appears to be a necessary condition for one's knowing that a sensed object is a chair or table or horse. In the Principles, Berkeley seems to believe that error is possible only if representationalism is the case; or perhaps he thinks that the only kind of error that needs to be handled is error of the mislabeling sort, calling something a horse when it is a donkey, an error that could be explained as due to carelessness in comparing the standard for a word and the object to which the word is applied. Some errors result, however, from a thing's appearing other than it is, as Berkeley himself comes to realize in the Three Dialogues. It is errors of this sort that undermine his account of the truism that “we” sometimes know that an object is a tree or table.

Suppose that Spina asserts that the wall is red and later claims that he was wrong, explaining his mistake by claiming that the wall appeared red at the time he asserted that it was red. If Berkeley allows that Spina senses a collection that contains “red” (at the time he asserts it is red), and allows, as surely he must, that there is no mark that differentiates veridical from nonveridical sensings, then that which is experienced cannot ground the true application of a physical-object judgment. (I ignore the difference between applying ‘wall’ and ‘red wall’ and assume that one can be mistaken about an object as well as an object's color. A mirror image of a red wall should do.) The truth maker for a physical-object judgment is not experienced. Accordingly, ‘x[]^s’ does not actually represent the truth maker for someone's seeing an object that grounds the application of a physical-object word; it grounds at best the fact that someone experiences a sensible “something.”¹⁰

Berkeley is not helped, I should add, by being read as a phenomenalist or, more generously, as an incipient phenomenalist. First, phenomenalism does not permit experiencing the ground for a physical-object judgment. Second, that which is experienced is the ground for a judgment that has no ordinary use; it is a judgment contrived as part of a philosophical explanation of a kind of error. That is, the claim that Spina experiences a collection containing red is a philosophical claim, as is the claim that Spina could have made or did make an everyday assertion that was about it and made true by it. If Spina speaks everyday language, none of his judgments are about the sorts of items posited by the phenomenalist.

Let us assume for the sake of the argument, however, that Berkeley is headed toward phenomenalism in the Dialogues and that phenomenalism is able to provide an account of “our” knowledge of trees, walls, chairs, and so on. Berkeley would still be an idealist, for phenomenal collections would have to be accidents of the mind, at least insofar as his account of knowledge—however Pickwickian the knowledge claims—rests on the comparison model; and if it does not rest on that model, what does it rest on?

Berkeley's ontology fails, first, because it cannot accommodate “our” purported knowledge of chairs, tables, streams, and so on and, second, because it cannot accommodate knowledge at all—at least not if the account of knowledge is shaped by the CMK. The former, resulting from the fact that “things sometimes appear other than they are,” is daunting; the latter is disastrous and results from the fact that ‘x[]’ is not a combination rule. That means that the ontology cannot represent a truth maker for, say, ‘Michelle knows that she senses a red something’ or, more generally, for ‘I know this is red’ asserted by anyone.

One could try to skirt the disaster by claiming that the ontology can at least accommodate the truth maker for, say, ‘Michelle knows the ball is red’. After all, the ontology allows for, say, ‘x_m[X]^m & x_m[Y]^s’. (I use ‘x_m’ to stand for the referent of ‘Michelle’, ‘X’ for the content of what serves as the standard, and ‘Y’ as the content of the sensed collection.) Quite apart from the question of what role is to be assigned to the mind and quite apart from how one is to exhibit the matching relation such that one is not simply knowing the same object over and over again, the salvage operation saves little, if anything. What can be done with a situation in which Michelle senses two objects of different colors and is said to have knowledge of the color of but one? In this situation, one must account for the discerning of the match between the standard and the appropriate sensed object; but the ontology lacks the resources for that. That is, the ontology lacks the resources to represent a mind's attending to one of several sensed objects—not to mention that there is no way to represent the discerning of a relation between a standard (assuming it can be marked as such) and a sensed object. Accordingly, Berkeley's ontology cannot accommodate more than the necessary conditions for knowing a something, however one categorizes it over and above its being an “accident” of the mind. True, one could retreat to the only dry ground possible: namely, that a mind senses only one collection at a time and that when the mind does, the appropriate standard is always also an accident of the mind. Actually, one could push the poverty one step farther, claiming that knowing is experiencing.

The foregoing attack on the failure to accommodate ontologically the CMK has something in common with the later Wittgenstein's skeptical attack on the model, an attack that assumes that the object of purported knowledge has favored status as “in the mind.” Wittgenstein, insisting that knowledge involves putting the right word to the object, presses the problem of how the ostensible knower knows that he or she has got the right standard for the word and how she or he can be certain that the matching has been properly carried out. The aim of Wittgenstein's attack is to produce a reductio of the comparison model itself, not merely of phenomenalism or, if one will, mentalism.¹¹ My attack is meant to show that even if one thinks one can surmount the skeptical attack, one cannot devise an ontology within which to locate or ground the CMK.

I want now both to sharpen and to broaden my attack on idealism by reflecting briefly and structurally on Gustav Bergmann, an ontologist as heroic as the tradition has known. Bergmann's ontology has, among other kinds of entities, particulars, universals, and relations. Further, he has two basic kinds of “atomic facts,” mental ones and nonmental ones, the difference resulting from his having two kinds of radically different universals, propositions and nonpropositions. The former are used to accommodate “thinking something is green,” “believing something is square,” and so on; the latter are used to accommodate shapes, colors, tones, and so on.

Bergmann's ontology leads to an utter catastrophe: it is unable to get mind and nonmind together, unable to get a mental act related to a nonmental state of affairs, notwithstanding that he, unlike Berkeley, has relations other than inherence or exemplification at his disposal. (1) It will not do, even though the syntactical rules allow for it, for Bergmann to make experience a relation between a mental particular and a nonmental particular. Not only would that violate his Principle of Acquaintance, the basis for his insistence that particulars always exemplify at least one universal, it would also provide no ground for knowing that something is, say, red: that which is the object of acquaintance must contain the ground for the truth of the sentence, not merely for the application of the subject of the sentence. (2) Bergmann cannot allow the mental particular to be related to a state of affairs, for in order to do that, one would have to introduce a so-called pseudorelation. Such a relation is objectionable not because of its odd syntactical nature (‘xRfy’) but rather because a relation cannot be exemplified without there being the relata, and the relatum associated with ‘this is red’ may not exist, since the sentence can be false. On Bergmann's conception of the ideal-language method, sentences of the natural language are mapped onto strings of the formalism, but none of the strings need be true. That means that for Bergmann the mere possibility that something is not the case puts the nonmental world beyond the reach of mind. Somewhat differently, Bergmann is forced to analyze an act of sensing on analogy with an act of thinking. Notwithstanding his insistence on the principle of acquaintance, Bergmann's ontology has no room for acquaintance, at least not as a genuine relation. All that mind (mental acts) experiences are, so to speak, its own propositional characters. The nonmind world is lost, utterly lost, to mind.¹²

Bergmann's fate is not, upon reflection, particularly surprising. His mental acts are particulars and as such not that different from Berkeleian minds. After all, a Berkeleian mind is in effect an enduring particular; the Bergmannian mental act, a momentary Berkeleian mind. Further, though Bergmann has relations, the relata of all of them are particulars; and no particular provides a ground for a knowledge claim. Finally, Bergmann's method dictates that sensing is as problematic as believing. Just as false belief is possible, so too false sensing is possible.

My ontological representation of Berkeley was not designed to make translation possible. My representation was designed to depict only the truth makers of true sentences. I was thus able to represent a mind as acquainted with, experiencing, a complex entity, a “something” having properties. Berkeley is spared the problem of false sensing, but that does not save his account of knowing.¹³

Notes

1. If God is included in the ontology, he can at best serve to experience sensible objects.

2. What reasons could Berkeley have for nominalism? He needs it neither for his attack on Locke's doctrine of abstraction nor for his attack on representationalism.

3. Descartes's search for a “new” criterion of truth is inspired by his belief that the CMK does not accommodate the sort of knowledge he demands. His clarity-distinctness criterion, even if accepted, will not provide for the kind of knowledge Berkeley wishes to account for.

4. The Ethics and Selected Letters, trans. Samuel Shirley and Seymor Feldman (Indianapolis: Hackett Publishing Co., 1982), p. 32. Wittgenstein makes a similar-sounding claim in the Tractatus (2.0122), but its meaning is altogether different. Spinoza's claim expresses merely that if one uses ‘substance’ in the traditional way, then that which is not a substance but combines with a substance must be an affection or mode—it must be what the Aristotelian-Scholastics labeled an “accident,” that is, an inessential feature possessed by, and depending for its existence upon, a substance.

5. The peculiar status of sensed items and the latent skepticism in the Aristotelian tradition were ignored because before Galileo there was no appreciation of the difference between direct realism and the view that the cause of what is sensed “completely” resembles what is sensed. Curiously, in the Third Meditation Descartes remarks that he once held the latter, representationalist view; but he seems not to have realized that it alone suffices to bring on a skeptical crisis.

6. PR 49 encapsulates Berkeley's agonies.

7. That an idea cannot exist without the mind is at best a grammatical truth and, as such, supports no ontological claim. However, when ‘idea’ is shorn of ‘of’, there is not even a grammatical truth that Berkeley can think is of use to him. PR 49 is also to the point here; in fact, it bears on many points central to Berkeley's position.

8. Though “causal” explanations have no place in ontology, Berkeley believes they do. Indeed, his primary reason for categorizing mind as substance seems to be that he needs a cause for imaging, sensing, and so on.

9. The prejudice gains a foothold early. It is present in Aristotle's Categories and dominates the thinking of all the analytical ontologists who take this inspiration from Russell and the early Wittgenstein. It is rather amusing that the analytical ontologists, having liberated themselves from grammatical form, submit to domination by logical form, especially so since the difference between the two does not come to that much.

10. Berkeley's failure on this score is elegantly laid out by Richard H. Popkin in “The New Realism of Bishop Berkeley,” in George Berkeley: Lectures Delivered Before the Philosophical Union of the University of California in Honor of the Two Hundredth Anniversary of the Death of George Berkeley, Bishop of Cloyne (1685-1753), University of California Publications in Philosophy, no. 29 (Berkeley and Los Angeles: University of California Press, 1957), pp. 1-19; reprinted in Popkin, The High Road to Pyrrhonism (San Diego: Austin Hill Press, 1980), pp. 319-38.

11. My critique of Berkeley does not require that the word be a feature of his account. I presuppose merely that experience is only a necessary condition of knowing and that a standard, however characterized, must be brought into play.

12. Bergmann's difficulties come to the surface in “Realistic Postscript,” in Logic and Reality (Madison: University of Wisconsin Press, 1964), pp. 304-40, and stalk him for the rest of his career.

13. If there is anything of worth in this essay, it is due, I should think, to others. Virtually all of my thoughts on Berkeley have grown out of conversations—many of them from long ago, though some of them quite recent—with members of the “Iowa School”: Bracken, Popkin, Bergmann, Watson, Cummins, Van Iten, Muehlmann, Flage, Alan Hausman, and David Hausman. I have also benefited over the years from conversations with, among others, Pappas and Winkler. If only I could remember the things I have read!

Bibliographical Note

Cross-references to this anthology are by parenthesized page numbers: (#) or (#-#). Except for those to Berkeley's writings, all other (external) page references are preceded by p. or pp.

For Berkeley's writings, we have used A. A. Luce and T. E. Jessop, editors, The Works of George Berkeley, Bishop of Cloyne, 9 volumes (London: Nelson, 1948-57); hereafter referred to as Works. More specifically, references to Berkeley's writings in this volume include the following abbreviations:

A: Alciphron (Works III). References are by dialogue, section, and page number.

D: Three Dialogues Between Hylas and Philonous (Works II); hereafter, Dialogues. References are by page number.

DM: De Motu (Works IV). References are by section number.

IN: Introduction to the Principles (Works II). References are by section number.

NTV: An Essay Towards a New Theory of Vision (Works I). References are by section number.

PC: Philosophical Commentaries (Works I). References are by entry number.

PR: A Treatise Concerning the Principles of Human Knowledge (Works II); hereafter, Principles. References are by section number.

S: Siris (Works V). References are by section number.

TVV: The Theory of Vision Vindicated and Explained (Works I). References are by section number.