SOURCE: “Berkeley's Objection to Abstract Ideas and Unconceived Objects,” in Essays on the Philosophy of George Berkeley, edited by Ernest Sosa, D. Reidel Publishing Company, 1987, pp. 61-81.

[In the following essay, Bolton provides support and opposition for Berkeley's rejection of abstraction as well as his form of idealism.]

1. According to Berkeley's famous theory of perception, we see, feel and otherwise perceive nothing but ideas; the whole of the sensible world with its trees and rocks, sun and stars, consists of nothing but idea sequences. The oddity of the consequence, that we eat, drink and clothe ourselves in ideas, seems to have discredited this theory for the first two hundred years. But in this century, the theory has been applauded for its elegant economy and the daring way it cuts off speculation about the existence and nature of the sensible world. Many other early modern philosophers regarded sensory ideas as only the means by which we perceive other sorts of things; for them, ideas were instruments which, even if fully known in themselves, offer at best partial knowledge of objects that exist in the sensible world. Berkeley's stunningly simple move is to identify objects we apprehend by sense with aggregates of ideas, while still maintaining that ideas are fully accessible to the mind that has them. This is inter alia a thesis about the intentional objects of perceptual states. We do not perceive objects by means of ideas, because ideas do not represent other things. By identifying ideas and sensible things, Berkeley robs minds of cognitive access to perceptual objects that are not fully accessible to consciousness.

Recent admiration for Berkeley's theory of perception in matched, however, by widespread dismay over another aspect of his theory of ideas, his denial of the possibility of abstract ideas.¹ Strategically placed in the introduction to the Principles, the attack on abstraction is said by Berkeley to “anticipate the design” of the whole work. It exposes a general source of error which, Berkeley thinks, plays a more particular part encouraging the mistaken opinion that matter exists independently of mind (Principles, Intro., 6, 5).

This anti-abstractionist part of Berkeley's philosophy has been severely attacked on several counts.² For instance, even the most sympathetic commentators doubt that the attack on abstraction has the direct relevance to immaterialism Berkeley claims.³ In contrast, however, I think one only needs to understand Berkeley's general theory of ideas to see that it does and that the case is just as Berkeley's prefatory handling of the doctrine of abstraction suggests. At the outset of the Principles, he suggests (mainly by presupposing it) a general theory of ideas as they function, not only in sense perception, but also in derivative modes of thought: memory, imagination, dreams and reveries, the practice of geometry and other sciences. In this broad theory of ideas, Berkeley makes the same sort of move he makes, more famously, in the case of sense perception. That is, Berkeley takes the objects of various modes of cognition, objects his contemporaries typically regarded as things represented by ideas, simply to be ideas. Because Berkeley agrees with those who thought ideas are fully known, the result is strictly to limit the class of objects we are able to think about to things of a type we adequately know. Now this general theory of ideas is diametrically opposed to the view of ideas as instruments by which we think of other sorts of things and by which we have a partial knowledge of them. I will contend that Berkeley's theory of ideas is the basis of his rejection of abstraction.

My aim in this paper is to vindicate Berkeley's treatment of abstract ideas in the introduction and early sections of Principles. I do not want to defend the truth of Berkeley's account of ideas, but rather to defend him against familiar charges of misusing the rejection of abstract ideas in certain arguments and especially the charge of misjudging the relevance of the doctrine of abstraction. Far from being irrelevant, I want to show, Berkeley's anti-abstractionist theory of ideas grounds the unintelligibility of mind-independent sensible things. Berkeley's “Idealism” claims not just that sensible objects are mind-dependent collections of ideas, but further that it is incoherent to think sensible objects are independent of mind. Thus, Berkeley agrees with other “anti-realist” philosophers that it is incoherent to suppose there are things in the sensible world outside the scope of possible knowledge. But what exactly is Berkeley's basis for such a view? Some commentators suppose he means to argue that expressions like “unperceived matter” are meaningless, due to the Empiricist demand that significant noun phrases be linked with items of sense experience.⁴ But, as other scholars have shown, Berkeley does not hold that theory about language.⁵ My contention is that what grounds Berkeley's claim that mind-independent sensible things are unintelligible, and gives his Idealism its special stamp, is his anti-abstractionist theory of ideas, in particular his views about the intentional objects of ideas.

Let me explain before proceeding that of course ideas, or sensible things, are not the only objects of thought and knowledge Berkeley recognizes. There are also “notions”: minds, their actions and relations (see, e.g., Principles, 2, 27, 89, 142). My concern here is only with various modes of thinking about bodies or what exists in the sensible world. Berkeley holds notions and ideas to be mutually exclusive categories. Thought about bodies involves ideas, but not notions, whereas thought about minds concerns notions, not ideas. So, it is, I believe, possible to discuss his views about cognition of sensible objects with only passing mention of minds and other notions.

2. As a first step toward understanding Berkeley's attack on abstract ideas, it is useful to look at his view of the history of the unfortunate doctrine. He ranks it chief among the false principles responsible for confusion, uncertainty and absurdity in logic and metaphysics, mathematics and other sciences (Principles, Intro., 4-6; Draft Intro. 17). He took the doctrine to be central to all the main traditions regarding the theory and practice of the sciences, for he ascribes it to Aristotle, Locke, “ancient [and] modern logicians and metaphysicians,” to the Scholastics and, indeed, “all other philosophers” (Draft Intro. 17; Letter to Johnson, Works, ii. 293). This last is presumably an overstatement, but surely it indicates that Descartes, Malebranche and Arnauld are among the “masters of abstraction.”

It is not easy to identify a relevant view shared by this heterogenous group of philosophers. Surely they did not all hold to a single theory of ideas.⁶ But Locke and others such as Arnauld and Nicole took abstract ideas to be the means by which we think about kinds and the basis of all knowledge of general truths (e.g., Essay III.iii, and IV.vi; Logique, I.v). Berkeley links abstract ideas with genera and species (Draft Intro. 7; PC, 566, 703) and with accepted views about “science” and the “demonstration of general propositions” NTV, [An Essay towards a New Theory of Vision] 124; Principles, Intro., 15). His target seems to be a view I call “Essentialism” about knowledge of general propositions. The tenets of this view are: that essences determine kinds in that the things that belong to a kind are distinguished from others by having the essence of the kind; that the essence is the “source” of the features shared by all members of the kind; that, in addition to the essence and the features derived from it, members of the kind have various “accidental” features in which they do, or can, differ from one another; and, most important for Berkeley's purposes, knowledge of the essence is required for science or knowledge of universal propositions about a kind. Now I have purposely constructed this set of tenets so generally that they ignore a wide range of deep and important topics of dispute, including: the status of universals, the nature of essences, how (and whether) essences can be known, the objectivity of the distinction between essence and accident, the role of species-essence in the identity of individuals, how properties “flow from” essences, the semantic connection between essences and general terms. But it requires a broad doctrine that arches over these issues that had divided philosophers to sustain Berkeley's charge of responsibility for a continuing history of error in the sciences going back to Aristotle.

Berkeley objects particularly to one tenet of the Essentialist view: an account of essence gives a perspicuous, but only partial, characterization of particular things. It is incomplete relative to the things to which it applies, because it omits all features accidental to the kind. It is this assumption that we can conceive a partial characterization, and thereby have (partial) knowledge of all things that conform to it, that Berkeley rejects. Conceptions of essences would be abstract ideas, and they are among Berkeley's targets:

It is indeed a tenet, as well of the modern as the ancient philosophers, that all general truths are concerning universal abstract ideas; without which, we are told, there could be no science, no demonstration of any general proposition …

(NTV, 124; see also Principles, Intro., 15.)

Now the abstractionist model for understanding general propositions by a partial characterization to which any number of particulars may conform can be separated from other principles of Essentialism. But the model is embedded in that traditional view, and I think it is in that role that Berkeley supposes it had pernicious effects upon the sciences.

Berkeley's descriptions of the abstractions he opposes clearly show that abstract ideas are incomplete characterizations (relative to things that conform to the ideas) and that abstract ideas include conceptions of essences of kinds.⁷ There are two types of abstract ideas, he explains in Principles, Introduction (see also Principles, 99). In the first, “singling” type, the mind forms an idea of a single quality abstracted from other qualities which it cannot exist without, for instance, an idea of color or motion without extension. Notice that anything that conforms to the idea of color will have extension, so the idea of the “singled” quality conforms to the pattern of abstract ideas. The second type of abstract idea includes the features shared by several particular things and omits their various differences. The idea of extension abstracted from all particular sizes and shapes is one example; for another:

… the mind having observed that Peter, James and John resemble each other in certain common agreements of shape and other qualities, leaves out of the complex or compound idea it has of Peter, James, and any other particular man, that which is peculiar to each, retaining only what is common to all and so makes an abstract idea, wherein all the particulars do equally partake.

(Principles, Intro., 9.)

This favors a Lockean view of the (nominal) essences of human beings, rather than an Aristotelian or Cartesian one, but ideas of essences of any sort would be abstract ideas of this second type.

Finally, Berkeley allows that he can form abstract ideas of things of qualities that can exist apart:

Thus, I imagine the trunk of a human body without limbs, or conceive the smell of a rose without thinking on the rose itself. So far, I will not deny I can abstract; if that may properly be called abstraction which extends only to the conceiving separately such objects as it is possible may really exist or be actually perceived assunder.

(Principles, 5; see also Intro., 10.)

I will later have something to say about the use of this “test” for possible abstractions. For the present, I want only to suggest that the conceptions Berkeley allows are not partial characterizations of the objects that conform to them. A limbless body is part of a natural whole and a rose's scent, part of the collection of qualities that belong to a rose; but the object that conforms to an idea of a limbless body is not the natural whole, nor is the object of the idea of a rose's scent the rose. Compare the abstract idea of red that omits extension or a specific shade; any particular instance of red, anything that conforms to the idea, will be extended and determinate in shade. It is that type of partial characterization of things that conform to an idea that Berkeley rejects.

In light of his rejection of abstract ideas, Berkeley thinks it necessary to explain himself particularly on two topics: the signification of general terms and the knowledge of general propositions. Detailed discussion is impossible here, but a brief account is necessary background for Berkeley's main arguments. Locke urged that words acquire meaning by association with ideas, that a general term signifies an abstract idea of the things to which the term applies (e.g., Essay, III.iii). Berkeley maintains that a general term stands, not for one abstract idea, but for many particular ones; that is, in case the general term names sensible things, it signifies many particular things, i.e. ideas. (Commentators have noted that, unlike Locke, Berkeley seems unconcerned with explaining how general terms are linked to the particular things they name.) Further, it is not necessary that on every occasion that a word is used, it should raise in the mind of those who use it the idea of what it stands for (Principles, Intro., 18-20).

As for the knowledge of universal propositions, Berkeley maintains there are no universal conceptions, but a particular conception is “rendered universal” by coming to represent other particular things. For instance, to demonstrate a general theorem about triangles, one proves it of a particular triangle which “doth equally stand for and represent all … triangles whatsoever” (Principles, Intro., 15). How can we be sure that a property known to belong to one particular triangle belongs to others that are not exactly like it? Because, Berkeley replies, one sees that the proof does not depend on any feature peculiar to that triangle as opposed to any other. He adds:

And here it must be acknowledged that a man may consider a figure merely as triangular, without attending to the particular qualities of the angles, or relations of the sides. So far he may abstract. But this will never prove that he can frame an abstract, general, inconsistent idea of a triangle.

(Principles, Intro., 16.)

As many commentators have observed, Berkeley gives no clear account of what determines the features common to all things in a kind, for instance, no explanation of what is supposed to distinguish the things a particular triangle “stands for and represents” from those it does not. Still, something can be gleaned. One main point is that what an idea represents is not intrinsic to the idea, but rather the idea is “rendered universal” by the use to which it is put. Further, the account strongly suggests that one idea might represent quite different things in different circumstances: a right triangle would represent all triangles if shown to have interior angles equal to two right angles, but would represent only right triangles if shown to be inscribed in a semicircle, and so on. What the particular represents is a function of human activity (also see Alc., [Alciphron: or, the Minute Philosopher,] 303-308). Although Berkeley gives no full account of how actions establish a relation of representation between one idea and others, it seems to have something to do with our ability to attend selectively to certain features of an idea. This view about the basis of distinctions of kinds, which I have only sketched, is central to Berkeley's alternative to “Essentialism.”

3. Why is the doctrine of abstract ideas a “false principle”? Berkeley's main argument seems intended to apply generally to all abstract ideas and to show that they are impossible on logical grounds.⁸ The argument is actually made in a quoted passage from Locke's Essay, in which Berkeley seems to think the author so clear and candid that he reveals the absurdity of his own doctrine. Locke's example is the abstract idea of a triangle. It includes being a figure with three sides and angles, for these features are common to all triangles, but it omits angles and sides of any definite size or mutual relation, because triangles differ in these respects. Locke expresses the surprising view, which Berkeley eagerly endorses, that it is an idea of a triangle that “… must be neither oblique nor rectangle, neither equilateral, equicrural, nor scalenon; but all and none of these at once.” But, the argument goes, such a figure is logically impossible, and so the idea of it is impossible; or, as Locke writes of the idea: “… it is something imperfect that cannot exist, an idea wherein some parts of several different and inconsistent ideas are put together” (Principles, Intro., 13, quoting Essay, IV.vii.9). It has by now become a familiar point that there are two main objections to this argument. In the first place, why should we think an idea is impossible just because its object is? A triangle that is both right-angled and not right-angled is impossible, but why should it follow that an idea of such a thing is impossible? And in the second place, what entitles Berkeley to conclude that the object of the abstract idea is impossible? In particular, why should we agree that its object is neither obtuse, acute nor right angled? Of course, if its object is a triangle, it must be one of these, but why must it also be none of them? Berkeley assumes that the features omitted from the idea are lacking from its object, but there seems no need for this. It seems an idea of a triangle may simply fail to specify the length of a figure's sides without specifying that the figure has sides of no particular length. Why should we agree that an abstract idea must be formed in this second way and that its object must be supposed to have features it also lacks?

No doubt the generally accepted view is correct that Berkeley relies on a peculiar theory of ideas which perhaps need not be accepted by an advocate of abstraction. The problem is to say just what Berkeley's theory of ideas is. Neither of the two lines of thought dominant in the secondary literature seems to me entirely correct.⁹

One view ascribes to Berkeley a more or less deliberate tendency to think that an idea has the features of what it is the idea of: the idea of a triangle is triangular.¹⁰ This answers our first question, why Berkeley assumes an idea to be impossible if its object has mutually incompatible features; the idea must have the incompatible features of its object. But it does not answer the second question, why the object of an abstract idea is impossible. In particular, it does not explain why the abstract idea of a triangle represents a triangle that is neither acute, obtuse nor right-angled, nor a fortiori why it is “all and none of these.” The principle that the idea of F is F simply does not explain why the idea of F that omits G should be thought to be the idea of an F that is not G. For that, we should need two further principles: (i) that an idea that omits G is itself not G; and (ii) that an idea that is F and not G is the idea of an F that is not G. (The latter is needed, if the initial proposal is understood to be the conditional: if I is the idea of F, then I is F. It is not needed, if the proposal is understood to be biconditional: I is the idea of F if and only if I is F.¹¹) A further problem with this attempt to explain Berkeley's position on ideas is that it fails to say why Berkeley should have adopted the rather implausible thesis that an idea of F is itself F; for instance, the idea of motion is moving, the idea of heat is hot. Berkeley does think an idea has the features of its object, and only those features, but the key point is that he does so because he deliberately identifies ideas and objects.

It is often said that Berkeley's reasoning is explained by the fact that he takes ideas to be images of things, i.e. mental reconstructions of what it is like to see some thing, hear it, etc.¹² This may also explain why an idea is impossible if its object is. For, we say an image has the features of what it is an image of (although this may not be the best way of speaking¹³), so the image of something with inconsistent features might be said to have those features; moreover we do seem unable to form images of things with incompatible properties. Indeed, we seem unable to make images corresponding to a large number of abstract ideas, for instance, that of a triangle which is neither acute, obtuse not right-angled. But this image view of ideas does not explain Berkeley's general objection to abstract ideas, for images typically do, and perhaps must, omit some details of their objects; the image of a scalene triangle need not exactly specify the sizes of angles or sides.¹⁴ Furthermore, although the image theory explains why we cannot have ideas that omit certain features of their objects, it does not account for the assumption crucial in Berkeley's actual argument, that if an idea does omit a feature it is an idea of something that lacks that feature.

Berkeley took for granted that to think about an object is to have an idea in mind, but he maintains that the object is the idea. He wrote in the early notebooks: “By Idea I mean any sensible or imaginable thing” (PC, 775; also see 101 and 808). Thus, we sense and imagine ideas. He continued to regard ideas (or collections of ideas) as the objects of our cognitive activities, for he frequently asserts that we perceive ideas (e.g., Principles, 1, 4, 5, 7, etc., and especially 38 and 39); and sometimes he writes that ideas are objects of knowledge (Principles, 1), objects of geometrical demonstration (Principles, Intro., 12), and the things named by general terms for objects (Principles, Intro., 11, 18). Other idea theorists habitually write of ideas as representing the objects of cognitive acts, for instance, Arnauld and Nicole: “When I consider a body, the idea of it represents to me a thing or substance, because I consider it as a thing that subsists by itself …” (Logique, I.2; also, e.g., 5 and 6).¹⁵ On this view, an idea must be an idea of something else, for the idea just is the instrument by which thought is directed toward that other thing. Berkeley also uses the expression “idea of,” but whereas other idea theorists mean by it “idea that represents,” he typically means “idea, namely.”¹⁶ When Berkeley writes that someone who perceives something triangular has an idea of a triangle, he means the person perceives an idea that is a triangle. So, we can say that for Berkeley an idea is its own object; if there is an idea of x, then x itself is that idea.

It is fundamental that an idea is its own object, but there is more to Berkeley's theory of ideas than that. Although the intrinsic object of an idea is just the idea itself, Berkeley holds that an idea can come to represent something else, or as I shall say, it can have an “acquired” object. We saw earlier that one idea comes to represent another idea, only by the actions of a mind and only if the first idea (object) shares selected features with its acquired object. Berkeley's view departs markedly from that of Arnauld and Nicole, for whom each idea has an intrinsic object other than itself, and has it in virtue of features the conception “includes” rather than features the idea has. The notebooks show Berkeley's dissatisfaction with the standard account of the intentional objects of ideas:

Properly speaking Idea is the picture of the Imagination's making this is y^e likeness of & refer'd to the real Idea or (if you will) thing.

PC, 657a.

The referring Ideas to things w^ch are not Ideas, the using the term, Idea of, is one great cause of mistake, as in other matters, so also in this.¹⁷

PC, 660.

Whoever shall cast his eyes on the writings of Old or New Philosophers & see the Noise is made about formal & objective Being Will etc.¹⁸

PC, 781.

The distinction between Idea & Ideatum I cannot otherwise conceive than by making one the effect or consequence of Dream, reverie, Imagination the other of sense & the Constant laws of Nature.

PC, 843.

(Also see PC, 230, 523, 657, 672, 684, 823.) Berkeley deliberately asserts that an idea of F is itself F, because he cannot understand the notion of the “objective being of F” or an idea that intrinsically represents F. (One would like to know more about Berkeley's objection, but I have been unable to find a basis for more than speculation about this.) So, Berkeley rejects the view that an idea represents things in virtue of features the idea includes (vs. has) and the view that an idea has an object independently of the action of a mind. If a mind sets an idea up to represent something else, the idea must share certain features with its acquired object. Berkeley draws from this the important conclusion that the acquired object must be another idea (see, e.g., Principles, Intro., 15-16, 8, 9, 25, 27).

Now Berkeley does share with many other idea theorists the view that an idea is fully accessible to the mind that has it:¹⁹

As long as I confine my thought to my own ideas, divested of words, I do not see how I can easily be mistaken. The objects I consider, I clearly and adequately know.

(Principles, Intro., 22; also see Principles, 25 and PC, 606.)

He also writes about ideas that “since they and every part of them exist only in a mind, it follows that there is nothing in them that is not perceived” (Principles, 25; also 87). Two qualifications need to be made, however. Berkeley does not suppose it is impossible to err about your own ideas, but just that care and attention suffice to avoid it. Further, the “adequate” knowledge we have of an idea does not extend to its relations to other things (e.g., Principles, 89); perhaps this enables Berkeley to explain why one who has the idea of a triangle may not easily apprehend all theorems about triangles. For Berkeley, then, a mind not only “clearly and adequately” knows its ideas, but also adequately knows the (intrinsic) objects of those ideas, for the objects are identical to the ideas. In contrast, according to the instrumental view of ideas, what may be completely known is the features ideas “include” or how they characterize their objects, and typically the characterization is incomplete. So, here we have Berkeley's characteristic move: while maintaining that ideas can be adequately known, he collapses the ontological distinction between an idea and its (intrinsic) object, and thereby arrives at the position that the (intrinsic) object of an idea is adequately known.²⁰

This theory of ideas fully explains, I think, the argument against abstraction. If an idea has been identified with its own object, then evidently the idea is impossible if its object is, and the properties of the idea coincide exactly with those of its object. The abstract idea of a triangle omits any specific angular size, and so any particular size is “omitted” from the triangle; but this can only mean that the triangle lacks angles of any particular size. Presumably, however, any triangle must have determinate angles, so the object of the abstract idea is a triangle that both does and does not have angles of some particular size. The same reasoning applies to any abstract idea: an abstract idea is its own object and thus has all the properties its object does; but it must also lack some of those properties, because it is supposed to omit some properties of its object(s). Thus all abstract ideas are logically impossible.

Abstract ideas are possible on the view that ideas have objects in virtue of their intrinsic representational content, the features they “include.” Berkeley's attack against abstract ideas presupposes an alternative theory of ideas and their intentional objects. Berkeley's ideas do not “include” and “omit” features of their objects. An idea simply is an object of cognition. Moreover, an idea has no fixed representative function; but if an idea does acquire an object, a mind selects some features the idea has as basis of representation.

In his unfortunate account of the abstract idea of a triangle, Locke seems to have a foot in both camps.²¹ He clearly does not suppose the idea has a content, since on that view it simply includes triangularity and omits any specific ratio of sides. He seems rather to adopt the view that the idea is an object of cognition and represents by features it has. His predicament stems from the assumption that all features of the idea have an equal role in determining what the mind represents. Unlike Berkeley, he does not base representation on a mind's selection of certain features. Thus, he says the idea that represents all triangles cannot be, for example, a scalene triangle, because it represents triangles that are not scalene; but then the idea also must be a scalene triangle in order to represent those triangles that are. A general idea of a triangle is impossible, if the idea is supposed to represent a figure if and only if it has exactly the features the idea has. Berkeley must have thought Locke was only recognizing the truth in adopting the view that an idea represents in virtue of features it has (vs. includes). He must also have thought Locke erred in continuing to think of the general idea of a triangle as abstract, as intrinsically (vs. by activity of mind) providing a partial characterization to which all triangles conform.²² Locke's passage perfectly illustrates Berkeley's contention that ideas are objects which, although they may come to represent, cannot be abstract.

4. Having clarified Berkeley's objection to abstraction, we are in a position to consider the uses to which he puts the impossibility of abstract ideas. He maintains that the false doctrine of abstraction lends support to the opinion that material things exist unperceived, and further he uses the impossibility of abstract ideas as a positive argument for the mind-dependence of sensible things. Unlike many others, I think Berkeley is correct about this pair of connections. I think that, with one possible exception, his arguments are successful if he is granted his theory of idea-objects.

One of Berkeley's appeals to the doctrine of abstract ideas occurs in Principles, 5, and it is important to remember what he has already accomplished in the earlier sections. In swift order, he has established (as he supposes) the main tenets of immaterialism: sensible objects are collections of sensible qualities; sensible qualities are ideas; ideas cannot exist without a mind (Principles, 1, 3 and 4). Then Berkeley turns to the “opinion strangely prevailing among men” that trees, mountains and other sensible things have an existence distinct from a mind that perceives them:

If we thoroughly examine this tenet it will, perhaps, be found at bottom to depend on the doctrine of abstract ideas. For can there be a nicer strain of abstraction than to distinguish the existence of sensible objects from their being perceived, so as to conceive them existing unperceived?

The conception of existence abstracted from perception seems not to fit the paradigms of abstract ideas, in which some features of a thing are abstracted from others. However, Berkeley thinks he has shown that sensible qualities are ideas which depend upon perception; for example, the particular color a mountain has depends on how it is perceived. Given that doctrine, a mountain abstracted from the perception of it is taken apart from all its determinate sensible qualities. As Berkeley complains, “For my part, I might as well divide a thing from itself.” It seems an unperceived sensible thing must have determinate sensible qualities; but given the assumption that sensible qualities depend on perception, it cannot have any determinate qualities. The idea of an unperceived mountain is abstract and logically impossible for the same sort of reason the abstract idea of a triangle is.

It is a mistake to think Berkeley offers this as an argument for the conclusion that is is impossible for a mountain, etc., to exist unperceived.²³ That conclusion does not appear in the text, nor was that conclusion to be expected from the opening remark that suggests the strangely prevailing opinion is supported by the doctrine of abstraction (the argument would “deny the antecedent”); moreover, the mind-dependence of sensible qualities is presupposed in showing that the idea of an unperceived sensible thing is abstract. Then why does Berkeley bother to point out that the idea is abstract? In part, I think, to help explain why the view of the sensible world he claims is false has not always been recognized as such: it came under the protection of abstraction and the traditional model of scientific knowledge. In addition, Berkeley probably wanted to illustrate (vs. establish) the absurdity of unperceived sensible things by showing it to involve the inconsistencies already shown to plague abstract ideas.

Elsewhere, however, Berkeley does appeal to the impossibility of certain abstractions as a premise for arguing that sensible things cannot exist unperceived. In the most familiar case, the strategy is used against those who think figure, motion and other primary qualities exist unperceived although colors, heat and cold are sensations that depend on a mind:

But I desire any one to reflect, and try whether he can, by any abstractions of thought, conceive the extension and motion of a body without all other sensible qualities. For my own part, I see evidently that it is not in my power to frame an idea of a body extended and moving, but I must withal give it some colour or other sensible quality, which is acknowledged to exist only in the mind. In short, extension, figure, and motion, abstracted from all other qualities are inconceivable. Where, therefore the other sensible qualities are, there must these be also, to wit, in the mind and nowhere else.

(Principles, 10; also 11-13; also Dialogues, 193f.)

This argumentative strategy has been sharply criticized on logical grounds.²⁴ The problem turns on how Berkeley can support the claim that an idea of body without color or other sensible quality is impossible. Clearly, the fact that Berkeley, you or I try to frame an idea and fail does not prove the point, because future attempts may succeed and other minds may be better at forming ideas than we are. Berkeley needs something better than the “try and fail” test to establish impossibility of an idea. He does sometimes offer the principle that it is possible to form an idea of A without B if and only if it is possible for A to exist without B (assuming here that A and B are sensible qualities or things; e.g., Principles, Intro., 10, 5). But critics point out it is circular to appeal to this principle in the context of the present argument, where the impossibility of a certain abstract idea is invoked to prove that certain qualities cannot exist apart. In this context, is Berkeley able convincingly to show that the crucial idea is indeed impossible?

Furthermore, a critic can ask what is supposed to follow if it is impossible to conceive, say, a figure without color or other sensible quality. If the idea is merely beyond our conceptual powers, that does not show the object cannot exist. Surely, there may be things inconceivable to us. Nor does the impossibility of the idea show that an expression such as “unperceived figure” is meaningless; no Lockean in the theory of language, Berkeley admits meaningful expressions that are not linked with an idea of what they name (e.g., Principles, 2). Did he misjudge the significance of the impossibility of the idea?

Understanding Berkeley's attack on abstract ideas helps clear up these difficulties, because it supplies an argument against the possibility of certain putative ideas. Granted Berkeley's theory of ideas, any abstract idea is logically impossible; a description of its features is a formal contradiction. This is directly relevant to Berkeley's anti-materialist claims, because a Berkeleian idea is identical to its object. Berkeley does not rely on the weak “try and fail” test alone to establish that an idea-object is impossible.²⁵

In some cases where Berkeley appeals to the impossibility of abstractions to establish the impossibility of mind-independent sensible things, it is quite clear that he can rely on the logical argument against abstract ideas (e.g., Principles, 11-13). However, it is not clear to me that it is completely successful against the ideas of primary qualities central to the argument quoted above. It is true that a visible or tangible figure must have some determinate color or tactile “feel”; thus, an abstract idea of visible figure without color would be a figure that has a determinate color and does not have one. But it begs the question against the Cartesian idea of intelligible extension to assume a figure must be visible, tangible or otherwise perceivable. Perhaps Berkeley assumes in Principles, 10, that figure, motion and extension are sensible qualities and his argument there is not meant to show that the idea of intelligible extension is impossible. Or perhaps he thinks he can show there is no such idea by the meager gambit of “try and fail.” However, I doubt that he used that weak argument without having in reserve a more compelling objection to the idea's logical consistency. A passage in Three Dialogues shows clear appreciation of the logical requirements of the case:

… I do not deny the existence of material substance, merely because I have no notion of it, but because the notion of it is inconsistent. Many things, for ought I know, may exist, whereof neither I nor any other man hath or can have any idea or notion whatsoever. But then these things must be possible, that is, nothing inconsistent must be included in their definition …

(Dialogues, 232.)

Berkeley's general attack on abstract ideas fits perfectly into this lucid picture of his strategy against materialist doctrines.

5. The rejection of abstract ideas enters most fully into Berkeley's metaphysics in what has been dubbed “the master argument.”²⁶ In Principles, 22-23, Berkeley boasts he is “content to rest the whole” question whether sensible things can exist without the mind on a single test: whether the reader can conceive such a thing to exist unperceived. You cannot do so, he argues, because if you conceive, say, books in a closet with no one standing by perceiving them, still you do conceive the books, and so you do not conceive something that is unconceived (Principles, 22-23; also see Dialogues, 200).²⁷ This astounding argument does not mention abstract ideas, but I think that the rejection of abstraction is behind it.²⁸ Indeed, I think, the main thesis of Berkeley's metaphysics is an almost immediate consequence of his model of conception. What he is rightly “content to rest the whole upon” is the theory that identifies ideas and their objects and precludes abstraction.

It seems Berkeley makes a glaring error by overlooking the difference between what one conceives and the circumstances necessary for conceiving it. Pitcher offers the comparison of someone who denies it is possible to perform a play about Robinson Crusoe, stranded on a desert island, because of the presence of the audience. Of course, I can see a play “about” a man (in that he is a character in the plot) who is not thereby seen by me (as part of the plot); similarly, it seems I can conceive of a woman who is not (as part of the conception) conceived by me. Berkeley may be right that I cannot conceive books which are not thereby conceived, but that does not force me to include their being conceived in my conception. It does not show that a conception of books that leaves out their being conceived is logically impossible.²⁹

Berkeley's attack against abstract ideas gives him a devastating reply to this very natural objection. The “manifest contradiction” of supposing the books you conceive are not conceived, the objection goes, is no barrier to a conception of books that leaves out their being conceived. But given Berkeley's theory of ideas, this is a case of illegitimate abstraction. An idea of books that omits their being conceived is logically impossible; it would be books that are conceived (by you) and not conceived (being conceived is omitted from the idea-object). Berkeley undoubtedly thought the most compelling objection to his argument would take this form, and his general argument against abstract ideas provides the response to it. Of course, there may be further objections formulated in terms of other alternatives to his theory of idea-objects,³⁰ but Berkeley's attack on abstraction addresses the main historical competitor to his view. I am not attempting here to defend Berkeley's theory of ideas, but I do want to bring out that the shocking “master argument” is nothing more than a clever, unexpected application of his theory of cognition. According to that theory, there is no ontological difference between an idea and its (intrinsic) object and thus an idea does not “include” features that partially characterize its object.³¹

The master argument does not show that unconceived books are logically impossible, but only that it is logically impossible to conceive books that are unconceived. The former thesis, of course, does not follow from the latter. Nevertheless, Berkeley proclaims that the argument decides the case for immaterialism. The point is, I think, that we are unable to secure something (i.e. something sensible) as an object of thought unless it is an idea-object and an idea-object cannot exist unconceived on logical grounds. That is, I think, the characteristic thesis of Berkeley's Idealism. An unconceived, mind-independent sensible object is literally unthinkable. Either a mind fails in attempting to think about a sensible object or the mind has an idea of the sensible object and, in the latter case, it is logically impossible that the sensible object is unconceived. One cannot think about books that are not conceived. It might seem, nevertheless, that the books you succeed in thinking about might have existed even if they had not been conceived, but the reply is that you cannot think so. To think that books exist unconceived, you must form an idea of unconceived books and that idea is impossible on logical grounds.

Can we escape the conditions Berkeley lays down as necessary for cognitive contact with an object in order to examine the logical status of the object itself? Perhaps language offers means of referring to the object that circumvent Berkeley's identification of idea and object. In the section of Principles immediately following the “master argument,” Berkeley responds to this attempt to evade his theory of cognition:

It is very obvious, upon the least inquiry into our own thoughts, to know whether it be possible for us to understand what is meant by the absolute existence of sensible objects in themselves, or without the mind. To me it is evident those words mark out either a direct contradiction, or else nothing at all.

(Principles, 24.)

When using the words “unconceived books,” either I succeed in referring to an object (thinking about it) or I do not. In the former case, the conditions are created which make it logically impossible that the books in question are not conceived. In the latter case, the words are not used to “mark out” anything for thought, and I can make no judgement about the logical possibility of unconceived ideas. In sum, language does not secure reference when ideas do not. The argument comes around again to the demands of Berkeley's theory of idea-objects.

Berkeley's main metaphysical doctrine is thus a consequence of his revolutionary theory of ideas. That theory seems to have been motivated, at least in part, by obscurities in the established view that ideas have intrinsic representational contents; however, I have only hinted at Berkeley's place in the development of the theory of ideas and their intentional objects. The point I hope to have made is that his theory of ideas is central to both his attack on abstract ideas and his Idealist metaphysics. Abstract ideas are incompatible with the account of ideas Berkeley adopts. If that theory is granted, it follows that abstract ideas are logically impossible. That vindicates Berkeley's use of the attack on abstraction in some arguments against materialist claims. Moreover, abstract ideas form part of a theory of intentionality that countenances mind-independent objects of cognition. Thus, the attack against abstraction can be turned against the claim to be able to conceive sensible things that are unconceived, as in the “master argument.” Berkeley's anti-abstractionist theory of ideas limits things that can be secured as objects of cognition to things that are conceived, and that is the basis of Berkeleian Idealism.³²

Notes

1. See, e.g., Monroe C. Beardsley, ‘Berkeley on “Abstract Ideas,”’ Mind, 52 (1943); E. J. Craig, ‘Berkeley's Attack on Abstract Ideas,’ Philosophical Review, 77 (1968); Jonathan Bennett, Locke, Berkeley, Hume: Central Themes (Oxford: Clarendon Press, 1971), ch. 2; George Pitcher, Berkeley (London: Routledge and Kegan Paul, 1977), ch. 5; Willis Doney, ‘Is Berkeley's a Cartesian Mind?’ in Colin Turbayne, ed., Berkeley: Critical and Interpretive Essays (Minneapolis: University of Minnesota Press, 1982); J. O. Urmson, Berkeley (Oxford: Oxford University Press, 1982), pp. 23-31; Kenneth Winkler, ‘Editor's Introduction’ to Berkeley, A Treatise Concerning the Principles of Human Understanding (New York: Hackett, 1982), pp. xviif.

2. Compare the praise of David Hume, Treatise, I.i.7. The reversal in appraisals of the attack on abstract ideas and the doctrine of immaterialism was noted long ago by A. A. Luce, Berkeley and Malebranche (Oxford: Clarendon Press, 1934), p. 126.

3. See M. R. Ayers' introduction to George Berkeley, Philosophical Works (London: Dent, 1975), p. xx. Also Bennett, sect. 8; I. C. Tipton, Berkeley (London: Methuen, 1974), p. 157.

4. See especially Urmson, pp. 15-20; also Bennett, sect. 10.

5. The crucial point is that we have no ideas of spirits and other “notions,” but Berkeley clearly holds that there are words that name these things (see, e.g., Principles, 2); see especially Luce, pp. 20ff., on the importance of this thesis in the development of Berkeley's philosophy.

6. There was, however, a tradition regarding abstraction or the cognition of universals; see Julius Weinberg, Abstraction, Relations and Induction (Madison: University of Wisconsin Press, 1965), pp. 5-13.

7. Berkeley does not deny that we use and understand linguistic expressions that are incomplete descriptions. He opposes partial conceptions and denies that one who understands a partial description of a thing has in mind a partial characterization of it (see, e.g., Dialogues, 193).

8. Compare Craig, who maintains that the argument made in Locke's passage does not apply to the “singling” or the “common properties” types of abstract ideas. When the argument is correctly understood in terms of Berkeley's theory of idea-objects, however, it is clear that it holds for all abstract ideas. Throughout this paper, I mean by “abstract idea” abstractions of the sort Berkeley rejects, ideas that partially characterize objects that intrinsically (vs. by activity of a mind) conform to them.

9. Two main commentators who seem to me correctly to understand Berkeley's theory of ideas do not discuss its application to the attack on abstraction: Luce (see pp. 126-47 on abstract ideas) and Desirée Park, Complementary Notions (The Hague: Martinus Nijhoff, 1972), pp. 36-53 on ideas, and pp. 100ff. on concepts.

10. See, e.g., Bennett, sect. 6; Craig, pp. 435f.

11. One might reconstruct the reasoning this way: once an idea is reified, all its features have representational significance. If the abstract idea of a triangle had right angles, then it would be the idea of a right-angled triangle. The idea of a triangle (regardless of angle size) must then be a triangle, but cannot have angles of any particular size. I think this is the reasoning that ensnared Locke (see below, p. 000). Notice, however, that this argument does not depend on the thesis that the idea of F is F (i.e. if I is the idea of F, then I is F), but rather on the thesis that an idea represents whatever has all the features the idea has.

12. See, e.g., Craig, p. 430; Pitcher, pp. 70f; Urmson, pp. 28f; Tipton, pp. 142ff. Contrast Ayers' brief suggestion (p. xx) about Dialogues, 193; also Kenneth Winkler, ‘Berkeley on Abstract Ideas,’ Archiv für Geschichte der Philosophie, 65 (1983), pp. 75f.

13. See Bennett on “how not to reify ideas,” sects. 5 and 7.

14. The point is made by Bennett, p. 22; Pitcher, p. 70; Tipton, pp. 144f.

15. Also Des vraies et des fausses idées, pp. 209-10. Descartes also regards ideas as representations of their objects, comparing ideas to images and pictures, and writing that they “exhibit” or “represent” other things to mind; see, e.g., Haldane and Ross, pp. 159-65, passim.

16. See also Park, pp. 43ff.

17. The question here is our having no idea of spirit, as Berkeley sees it, because there is no resemblence between active spirits and passive ideas.

18. See, e.g., Descartes, ed. Haldane and Ross, vol. 1, pp. 162f.

19. Locke held that with attention we can completely know any of our own ideas, but it is not clear Descartes took this view of confused and obscure ideas. See, e.g., Meditations, III, on “false ideas”; also compare Principles, I, 45, and Essay, II.xxix, on confused ideas. Jessop's note to Principles, 25, oversimplifies the case.

20. In case a mind has an idea A that represents another idea B, the mind has adequate knowledge of A, but presumably may only know that B has those features in virtue of which it resembles A and is represented by it. Still, B is an idea and, Berkeley will claim, it cannot exist without some mind for which it is an intrinsic object and by which it is adequately known. All ideas are “in principle” knowable, but all ideas are not knowable to a given finite mind.

21. I do not want to discuss here Locke's settled views on abstraction. Berkeley's understanding of Essay, IV.vii.9, seems to me natural. Of course, it should not be taken in isolation from other passages as giving an accurate account of Locke's position, but I doubt that Berkeley takes it that way. I think he thought it recorded a lucid moment in which Locke briefly saw part of the truth about ideas and how ideas represent.

22. Some others have pointed out that an idea theorist can avoid Berkeley's attack either by strictly adhering to a doctrine of representational content or by agreeing with Berkeley that the representative function of an idea depends on the features it has plus the interpretive activity of the mind; see C. C. W. Taylor, ‘Berkeley's Theory of Abstract Ideas,’ Philosophical Quarterly, 28 (1978), sect. 5; Winkler, ‘Berkeley on Abstract Ideas,’ p. 74. It is a further question whether either of these moves would be fully consistent with Locke's actual doctrines.

23. Compare Doney, p. 279.

24. See Beardsley, sect. 4; Doney, sect. 2.

25. Also see Weinberg, pp. 13-24; but compare Willis Doney, ‘Berkeley's Argument against Abstract Ideas,’ Midwest Studies in Philosophy, 8 (1983).

26. André Gallois, ‘Berkeley's Master Argument,’ Philosophical Review, 83 (1974).

27. The text presents a puzzle, because Berkeley sets out to show that a sensible object cannot exist unperceived and ends with the conclusion that a sensible object cannot exist unconceived. Some critics have complained that at best the conclusion would not establish what Berkeley wants to show, that the existence of objects depends upon perception. I think, however, that Berkeley's conclusion suffices for his purposes. Berkeley's identification of ideas and their objects implies that an object exists if an idea of it exists; as he noted in PC, 473, he has a more inclusive conception of existence than the ordinary one. What distinguishes idea-objects that are part of the actual world from other objects that exist, but are only imagined or conceived, is roughly that the actual idea-objects are related to other idea-objects in the steady, general ways dictated by the laws of nature. Thus, we can say that an object is perceived if and only if (a) it is an intrinsic (vs. acquired) object for some mind and (b) it is related by appropriate laws of nature to other objects (idea-objects).

    Berkeley asks the reader to conceive books existing without the satisfaction of condition (a). For, if the reader conceives books actually existing, then automatically the object of conception meets (b). The contention that it also meets (a) rests on the point that the reader conceives the books. For the reader, the books are acquired (vs. intrinsic) objects of conception; but by Berkeley's account of ideational representation, the acquired object must be an idea-object and thus must be an intrinsic idea-object for some mind. Thus, the fact that the reader conceives books actually existing suffices to show that some mind must be conceived to perceive them.

28. See also Pitcher, p. 63; Gallois, pp. 64f.; J. J. Thomson, ‘G. J. Warnock's Berkeley,’ Mind, 65 (1956).

29. See, e.g., Gallois, pp. 56ff.; Pitcher, pp. 112ff.; Urmson, pp. 45ff. Also, A. N. Prior, ‘Berkeley in Logical Form,’ Theoria, 21 (1955), p. 122.

30. E.g., someone might phrase an objection in terms of some sort of “causal theory” of the intentional objects of cognition. This is not entirely unknown to Berkeley's predecessors: Locke, at least, suggests that ideas caused by secondary qualities refer the mind to the qualities that cause them (although he never writes of these ideas as ideas of secondary qualities); see Essay, II.xxx and xxxi.

31. Critics have suggested that the “master argument” proves too much, for I cannot conceive books which are not conceived by me and conceived when I do so; it should follow by Berkeley's reasoning that books cannot exist unless conceived by me, etc. The sting of this objection is removed, I think, by Berkeley's theory of ideational representation. Presumably I cannot now have an idea from the past, or one which belongs to another mind. But suppose I have an idea A which represents such an idea, B. Must the fact that I conceive A be part of the basis of the representation of B? I do not see that Berkeley is forced to say it must. B is represented by A in virtue of salient respects that may not include being conceived by me at a certain time. This means of thinking about B poses no threat to the thesis that I cannot conceive (even representationally) something that can exist unperceived, for the object represented by one idea must be another which, according to Berkeley's principles, cannot exist without a mind.

32. I want to thank Robert Bolton and Peter Klein for helpful comments on a draft of this paper. I benefitted a great deal from conversation with participants in the Newport Conference, and this is reflected in some revisions made afterwards. I am especially grateful to Robert Sleigh, who commented on the paper, Michael Ayers, George Papas and Kenneth Winkler.

Bibliography of Works Cited

Antoine Arnauld and Pierre Nicole, La Logique ou L'Art de Penser, reproduced from the 1662 Paris ed. (Hildesheim: Georg Olms, 1970).

Antoine Arnauld, Des vraies et des fausses idées, in Oeuvres de Messieur Antoine Arnauld, vol. 38 (Paris, 1683), pp. 177-365.

John Locke, An Essay Concerning Human Understanding, ed. P. H. Nidditch (Oxford: Clarendon Press, 1975).

The Philosophical Works of Descartes, trans. Elizabeth S. Haldane and G. R. T. Ross, 2 vols. (Cambridge: Cambridge University Press, 1970).

The Works of George Berkeley Bishop of Cloyne, ed. A. A. Luce and T. E. Jessop, 9 vols. (Edinburgh: Thomas Nelson, 1948-57).

Abbreviations

The present volume follows John Foster and Howard Robinson, eds., Essays on Berkeley: a tercentennial celebration (Oxford: Clarendon Press, 1985), for most of the abbreviations here listed.

Works: The Works of George Berkeley Bishop of Cloyne, ed. A. A. Luce and T. E. Jessop, 9 vols. (Edinburgh: Thomas Nelson, 1948-57). References, except to the works listed below, are by volume and page number.

PC: Philosophical Commentaries (Works, vol. i). References by entry numbers (Notebook B = entries 1-399; Notebook A = entries from 400 onwards).

NTV: An Essay Towards a New Theory of Vision (Works, vol. i). References by section numbers.

TVV: Theory of Vision Vindicated and Explained (Works, vol. i). References by section numbers.

Principles: The Principles of Human Knowledge, Part I (Works, ii). Referred to by section number.

Principles, Intro.: Introduction to The Principles of Human Knowledge (Works, ii). Referred to by section number.

Dialogues: Three Dialogues between Hylas and Philonous (Works, ii). References are to pages numbers.

Alc.: Alciphron or the Minute Philosopher (Works, iii). Referred to by page number.