[In the following essay, Foster examines two apparently contradictory views in Berkeley's philosophy: that all reality exists solely in the mind and that a physical world does indeed exist and follows set laws.]

I

Berkeley's philosophy of the physical world is built around two central claims. The first is his claim that reality is ultimately purely mental, consisting solely of minds (spirits) and what exists or occurs within them. The second is his claim that there is a physical world and one which (more or less) answers to the specifications of our ordinary beliefs. These two claims appear to be incompatible: it seems that to accept the existence of a physical world, with its three-dimensional space and its solid extended occupants, is precisely to accept the existence of something outside the mind. But Berkeley believed that the two claims could be reconciled. He thought that the physical world itself is wholly constituted by the things which exist and the facts which obtain in the mental reality. It is the nature of this reconciliation which I want to explore in the present essay.

Before we look more closely at Berkeley's own position (or positions), it will be useful to distinguish two ways in which someone who accepts a mentalist doctrine of ultimate reality might try to preserve the existence of the physical world. On the one hand, he might claim that, while the ultimate reality is purely mental, the physical world itself is part of that reality—that it is, quite literally, composed of mental entities and mental facts. For example, he might take physical objects to be collections of minds and take the spatial arrangement of these particle-minds to consist in the qualitative relations between their experimental states.¹ On the other hand, he might concede that the ultimate mental reality is wholly non-physical, but claim that the facts which obtain in this reality in some way logically suffice for the existence of a physical world (and I am here using the term ‘logically’ in a broad sense to cover not only a priori but also, as some people term them, metaphysical necessities). For example, he might, following a phenomenalistic course, claim that, while physical objects are not as such mental, their existence and spatial arrangement is logically sustained by the organization of human sense-experience. Now these two ways in which a mentalist might try to preserve the existence of the physical world are quite different. The first way, which takes the physical world to be part of the ultimate reality, is a form of physical realism. It differs from our common-sense view of the physical world not in respect of the ontological status of physical entities, but only in respect of their intrinsic nature. It takes the physical world to be ultimately real, but to be mental in its substance and character. The second way, in contrast, is a form of reductionism. It concedes that at the level of ultimate reality the physical ontology disappears, but insists that statements about physical entities can still be counted as true in virtue of facts about minds. It preserves the physical world not by locating it within the ultimate mental reality, but by construing it as something derivative, which this ultimate reality sustains.

The distinction between these two approaches—between mentalistic realism and mentalistic reductionism—is not one which Berkeley himself explicitly drew, nor, I think, fully appreciated. But the distinction will help to put his philosophical views in a clearer perspective. In particular, it will help to clarify the difference between his accounts of the physical world in the Principles and the Dialogues. Neither of these accounts, it must be admitted, is internally consistent. But as I interpret them, Berkeley's true position in the Principles is a form of reductionism, while his true position in the subsequent Dialogues is a form of realism.

II

First and foremost, of course, Berkeley is a mentalist. On this he never wavers or equivocates, despite his varying accounts of the physical world. This mentalism stems directly from his doctrine of esse est percipi—the doctrine that, within the domain of unthinking things, to exist is to be perceived. In effect, the doctrine restricts the ultimate ontology to two sorts of entities, namely minds (spirits) and their ideas.

Berkeley accepts this restriction because he thinks that it is impossible to conceive of a counter-example. Thus suppose we try to conceive of some type of object which exists in the ultimate reality, but is neither a mind nor composed of ideas. To have any positive conception of the intrinsic nature of this object, we are, Berkeley thinks, forced to clothe it with sensible qualities—the kind of qualities, like colour and visual or tactual extension, which feature in the content of sense-experience. In other words, we are forced to conceive of the object either as a concretion of sensible qualities or as a substance in which such qualities inhere. But Berkeley thinks that it is both self-evident and demonstrable by a variety of further arguments that sensible qualities can have no ultimate realization except as the internal objects of perception, i.e. as ideas existing in the mind. So if we conceive of the object as a concretion of sensible qualities, we have to construe it as a collection of ideas. And if we conceive of it as a substance in which such qualities inhere, we have to construe it as a mind. Either way, we fail to achieve the conception we were seeking—of something which would serve as a counter-example to Berkeley's mentalist doctrine.

I think that Berkeley's reasoning here is basically sound, though at certain points his arguments need to be amended or supplemented.² Where I think he goes seriously wrong is in assuming that the conclusion of this reasoning is enough to vindicate his mentalist doctrine. What the reasoning establishes is that if the ultimate ontology is not confined to minds and ideas, the additional entities are ones of whose intrinsic natures we can form no positive conception. But it by no means follows from this that there are no such entities nor even that it would be irrational to postulate them. Where this becomes crucial, of course, is over the issue of the physical world. Berkeley's opponent—the ‘materialist’—is claiming that physical objects are both ultimate and mind-independent, and Berkeley is challenging him to provide some positive and coherent account of what these objects are like in themselves. But it is far from clear why the materialist should feel obliged to meet this challenge. Why should he not be happy to concede that he can offer no positive specification of physical objects beyond a description of their structural, causal, and dispositional properties—a description which does not reveal their intrinsic content.³ Of course, he can only be happy with this if he can also offer some reason for believing that such objects exist; and, clearly, if they have no sensible qualities, they are not directly perceptible. But the materialist will argue that the postulation of these insensible objects can be justified by an explanatory inference. Their postulation is justified because, together with the postulation of certain psychophysical laws, it best explains the orderly character of human sense-experience. In effect, it best explains why human experience is organized as if a physical world obtains.

If Berkeley was wrong to assume that the materialist need feel embarrassed over his incapacity to specify the intrinsic nature of these mind-independent objects, he also had an independent argument against the materialist's case for postulating them. The materialist claims that we can justify their postulation by an explanatory inference—an inference which takes us from observed experiential effects to their unobserved physical causes. But Berkeley argues that, even if there were such objects, they would be causally inert and thus incapable of playing the explanatory role which the materialist assigns to them. For he thinks that the only kind of causation which we can conceive of is that of which we have an introspective understanding through the exercise of our own volition—through such volitional activities as the framing of a mental image and the attempt to move some part of one's body. And he thinks that the kind of causation thus revealed to us is not something which we can abstract from its volitional setting and envisage as operating independently of the will. In short, Berkeley thinks that all causation is, and has to be, volitional, and that, consequently, we cannot make sense of the materialist's claim that our experiences are caused by external unthinking objects. And if we cannot make sense of this claim, then we cannot justify the postulation of such objects by their role in explaining experience.

In claiming that all causation is volitional, Berkeley is, of course, assuming that our concept of cause involves the notion of some genuine agency or necessitation. Thus if someone were to define causation in a Humean fashion, as mere constant conjunction in the perspective of our inductive propensities, Berkeley would consider this definition inaccurate, but happily concede that, thus defined, non-volitional causation is possible. His claims are simply that volition is the only form of genuine agency (or causal necessitation), that only by invoking such agency can we provide a genuine explanation of human sense-experience, and that only a theory which is genuinely explanatory can be rationally inferred from our experiential data. Whether these claims are correct is another matter, and certainly Berkeley does not provide anything approaching an adequate defence of them. The first claim is particularly controversial, and in two distinct ways. On the one hand, it is far from clear that we do not possess a coherent notion of non-volitional agency; on the other hand, it is questionable whether, in the case of volition, introspection reveals anything more than constant conjunction. However, for the purposes of our discussion, we must leave these issues on one side.

Berkeley's conclusion is not that human sense-experience has no explanation, but that it must be explained in terms of the causal agency of God. Clearly, our sense-experiences are not the product of our own volition (unlike the mental images we frame) and their orderly character is not of our own making. Berkeley concludes that these experiences are the product of divine volition and that their orderly character is the product of the consistent volitional policies which God adopts:

But whatever power I may have over my own thoughts, I find the ideas actually perceived by sense have not a like dependence on my will. … There is therefore some other will or spirit that produces them.

The ideas of sense … have likewise a steadiness, order, and coherence, and are not excited at random, … but in a regular train or series, the admirable connexion whereof sufficiently testifies the wisdom and benevolence of its Author. Now the set rules or established methods, wherein the mind we depend on excites in us the ideas of sense, are called the Laws of Nature: and these we learn by experience, which teaches us that such and such ideas are attended with such and such other ideas, in the ordinary course of things.

(Principles, 29-30.)

Given his assumptions, I think that Berkeley's position here is basically well-founded. There is no denying that our sensory ideas do have a certain order and coherence—one which precisely accords with the hypothesis that we are the perceptive inhabitants of a certain kind of physical world. And, given that all causation has to be volitional, it is reasonable to account for this sensory order in terms of the volitional strategy of some controlling mind. In effect, it is reasonable to conclude that our experiences are directly caused by God and deliberately selected with the aim of making them conspicuously amenable to physical interpretation.

The only point where Berkeley goes astray here is in thinking (or so it seems) that the sensory order is reducible to certain phenomenal regularities in the ordinary sense, whereby, as he puts it, ‘such and such ideas are attended with such and such other ideas, in the ordinary course of things’. In actual fact, the sensory order can only be adequately specified by reference to the physical theory which it prompts us to accept (or at least by reference to something which preserves the structure of that theory). It is not just that the order accords with the hypothesis that we are the perceptive inhabitants of a certain kind of physical world. It is also that the order can only be fully and exactly specified in terms of that hypothesis (or something isomorphic). The order consists, precisely and irreducibly, in the fact that our experiences are systematically as if that hypothesis were true. This presupposes, of course, that the hypothesis postulates a physical world which is conspicuously orderly in its own terms. For even if our experiences were purely random, they would match the specifications of some kind of physical world. What makes our experiences genuinely orderly is that they can be taken to reflect, in an orderly fashion, the character of an orderly world. It is in this sense that they are conspicuously amenable to physical interpretation.

The fact that the sensory order can only be fully and exactly specified by reference to some physical theory does not mean that it is wholly concealed at the phenomenal level—that, outside the perspective of its physical interpretation, human experience appears totally random. Clearly, this is not so, and, from an epistemological standpoint, it is important that it should not be. For presumably it is those aspects of the sensory order which we can discern at the phenomenal level which provide us with our entry into the physical theory. Presumably, it is only because there are at least some crude phenomenal regularities which are conspicuous in their own terms, but which invite physical interpretation, that we ever come to acquire any physical beliefs. But the point is that the order which we discern at the phenomenal level is only a fraction of the order which we come to discern in the perspective of the physical theory. And it is this latter order which must be taken as the expression of God's volitional strategy.

What this means, in effect, is that Berkeley's conception of God's selection procedures, with respect to the causation of human experience, was too atomistic. He recognized that God's overall aim was to make human experience conspicuously amenable to physical interpretation, but he wrongly supposed that this aim was implemented by the adoption of a set of specific volitional rules (revealed, from our viewpoint, as the laws of nature) ensuring certain regularities at the phenomenal level. The fact is that, just as the sensory order cannot be detached from the perspective of some physical (or strictly, psychophysical) theory with which our experiences accord, so God's selection procedures cannot be detached from his overall strategy of securing an accord with that theory.

In a sense, these points help to strengthen Berkeley's case. For if the sensory order could be fully specified in purely phenomenal terms, the need to explain it by appeal to some external reality would be diminished. Indeed, it might be possible to find a perfectly adequate explanation in terms of certain autonomous phenomenal laws, without having to postulate anything outside the realm of experience to explain the laws themselves. It is because the sensory order can only be specified by reference to some physical theory, so that the experiential realm is, as it were, nomologically unmanageable in its own terms, that there is a pressing need for an externalist explanation. And if the explanation has to be volitional, then Berkeley's theistic hypothesis is surely the most plausible. In another way, of course, these points only serve to increase the pressure on Berkeley to make good his basic assumptions. If human experience is organized as if a certain kind of physical world obtains, and if that is precisely how the organization is defined, then the most natural explanation is that such a world does obtain. Berkeley regards this explanation as doubly defective—both defective in its postulation of a mind-independent world and defective in its acceptance of non-volitional causation. But on neither point are his arguments compelling.

III

As a restriction on the composition of the ultimate ontology, Berkeley's doctrine of esse est percipi does not determine any particular theory of the physical world. Indeed, it leaves him, with respect to the status and nature of physical objects, with four options: (1) to construe such objects as minds or collections of minds; (2) to construe them as ideas or collections of ideas; (3) to deny that such objects feature in the ultimate reality, but admit them as things whose existence the ultimate facts in some way logically sustain; and (4) to deny that there are physical objects at all in any sense. Options (3) and (4) are respectively reductionism and nihilism. Options (1) and (2) are forms of physical realism, though constrained by the mentalistic metaphysic.

While all four options are available to him, Berkeley never seriously entertains either (1) or (4). He takes it for granted that our senses afford us adequate grounds for believing in the existence of a physical world and considers it a defect in a theory if it leads to scepticism. Likewise, he takes it for granted that physical objects are, by definition, unthinking: the suggestion that they are really collections of minds he would have dismissed as manifestly incoherent.⁴ On both these points, of course, Berkeley's position is in line with common sense. This leaves him with a choice between (2) and (3). And on this issue, his position is less clear. Part of the difficulty is that, even in the course of a single work, he says things which are clearly inconsistent, some of them favouring (2) and others favouring (3). A further problem is that, by failing to appreciate the distinction between the realist and the reductive approaches, Berkeley never explicitly addresses himself to the issue as I have formulated it. This means that, in trying to determine his true position, so much is a matter of the implications of his remarks, rather than of what he actually says.

There can be no doubt that Berkeley's basic instinct—his disposition prior to any deep philosophical reflection—was to accept option (2), which takes physical objects to be ideas or collections of ideas in the mind. This position is the direct application of the doctrine of esse est percipi to the physical world: it construes physical objects as things whose existence is to be perceived. I shall henceforth call this the Simple View. There are many passages in both the Principles and the Dialogues where Berkeley explicitly endorses this view. A famous example occurs in section 6 of the Principles, where he claims the view to be self-evident:

Some truths there are so near and obvious to the mind, that a man need only open his eyes to see them. Such I take this important one to be, to wit, that all the choir of heaven and furniture of the earth, in a word all those bodies which compose the mighty frame of the world, have not any subsistence without a mind, that their being is to be perceived or known …

Berkeley's basic instinct to accept the Simple View was partly due to his failure to appreciate the distinction between the realist and the reductive approaches. Without an explicit recognition of the possibility of reductionism, he finds himself automatically moving from his mentalist doctrine of esse est percipi to its direct application to the physical world. Thus, in defence of his preceding remarks, section 6 concludes:

To be convinced of which, the reader need only reflect and try to separate in his own thoughts the being of a sensible thing from its being perceived.

Whatever its merits, this argument is only directly relevant to the composition of the ultimate reality. There is no special difficulty in separating the being of a sensible thing from its being perceived, if the thing is accorded a lower ontological status—if it is taken to be, as I would put it, ‘ontologically derivative’ rather than ‘ontologically primitive’.⁵ The phenomenalist, for example, can allow the physical world to contain unperceived sensible objects, by reducing the existence of such objects to certain ‘possibilities of sensation’ or to other aspects of the way human experience is organized. That Berkeley feels able to apply the argument directly to physical objects shows that he has not taken the reductionist option into account, though, as we shall see, the theory which eventually emerges and predominates in the Principles is implicitly of a reductive kind.

If Berkeley was drawn to the Simple View partly through a failure to notice any alternative, it is also true that he found the view attractive in another respect—at least as he initially interpreted it. For he liked the way in which, in line with common sense, it made physical objects directly perceptible. It combined what philosophy had established about the nature of perceptual experience, namely that the immediate objects of perception are internal to the mind, with what we all accept prior to philosophical reflection, namely that we directly perceive portions and aspects of the physical world. Indeed, at the end of the Dialogues, Berkeley (through the mouth of Philonous) describes this combination as the summary of his whole position:

My endeavours tend only to unite and place in a clearer light that truth, which was before shared between the vulgar and the philosophers: the former being of opinion, that those things they immediately perceive are the real things; and the latter, that the things immediately perceived, are ideas which exist only in the mind. Which two notions put together, do in effect constitute the substance of what I advance.⁶

Although it has this epistemological advantage, it seems that, even from Berkeley's standpoint, the Simple View is open to a number of objections. I shall begin by mentioning the three most fundamental.

In the first place, if the physical world is entirely composed of ideas, there seems to be no way in which the separate sensory elements fit together to form a unified whole—at least, not a whole which would qualify as a physical world in any recognizable sense. The reason, of course, is that the elements are not arranged in a common space. An individual idea may include a spatial field (as, for example, in the case of visual experience); but there is no enduring and intersubjective field in which the ideas of different subjects and ideas at different times are collectively located. And, consequently, the Simple View seems to dissolve the physical world into a collection of disconnected fragments. Secondly, it seems that even if all the fragments could be pieced together, they would only cover a fraction of the spatio-temporal whole which we actually believe to obtain. For there would be no provision for the existence of unperceived objects or for perceived objects with unperceived aspects. Thus we would apparently be forced to say that there was no physical world before there were humans (or animals) to perceive it, that oranges have nothing inside them until they are peeled, and that my desk is annihilated every time I leave my study and recreated on my return. All this is clearly contrary to what we actually believe. Thirdly, the Simple View seems to exclude any genuine distinction between veridical perception and illusion. For if there are no external objects for them to represent, there seems to be nothing with respect to which our sensory ideas can be accurate or inaccurate. The drunkard's ‘pink rats’ and Macbeth's visionary dagger have, it seems, as much claim to count as part of the furniture of the world as the desk and paper I currently perceive. In effect, the physical world would appear to have lost its objectivity.

To these three objections we should perhaps add a fourth, though even its prima facie force is less clear-cut. I am thinking of the objection that the Simple View makes it impossible for different subjects to perceive the same physical item and thus conflicts with our belief that physical objects are publicly observable. The basis of this objection is clear enough: each subject only directly perceives his own ideas and these ideas, which represent nothing external, exist in his mind alone. What is not clear is that this prevents physical objects from being publicly observable in any adequate sense. For if physical objects are collections of ideas, different subjects could perhaps be said to perceive the same physical object by perceiving ideas which are elements of the same collection. Of course, since ideas are not located in some common sensory space, there is still the problem of finding some non-arbitrary basis for grouping them into the relevant collections. But this problem is already covered by the first objection: it is a special case of the more general problem of how the separate sensory elements fit together to form a unified sensible world.

These prima facie objections to the Simple View are ones which Berkeley cannot afford to ignore. If he is to preserve the physical world in any recognizable sense, he has to preserve it as something unified, objective, and public, and as something vastly more extensive and internally rich than our actual perceptions cover. In effect, he needs to preserve the physical world in a form which, apart from the unavoidable conflict with his metaphysical doctrines (i.e. his mentalism and his insistence on volitional causation), accords with our ordinary beliefs. Berkeley is aware of this, and it is essentially in his attempt to meet these requirements that the conflict in his thought arises.

IV

In his Three Dialogues, Berkeley's solution is to identify the physical world with the internal object of God's perception—with a complex idea which exists in God's mind and which, in partial and piecemeal fashion, our sensory ideas represent.⁷ This retains the Simple View, which equates the existence of a physical object with its being perceived; but, by making the physical world external to human minds, it avoids the objections to which this view seemed vulnerable. Since God's perception is not fragmented by time or distributed across different minds, the physical world perceived has the requisite unity: it is not a collection of isolated ideas, but one complex idea, which contains all physical items in a single spatio-temporal field. Likewise, since God's perception is not subject to the limitations of human perception, the physical world can be as extensive and replete as we ordinarily believe: there is no problem in supposing that there were mountains and rivers before there were humans to perceive them or that my desk continues to exist when there is no one in my room. Moreover, since there are external items for our sensory ideas to represent, there is, with respect to human experience, a genuine distinction between veridical perception and illusion. And, in the same way, physical objects are rendered publicly observable, since different human ideas can represent the same item in God's idea.⁸ In short, by locating the physical world in God's mind, it seems that Berkeley can solve all the problems at a stroke. He can retain the Simple View, but, at the same time, secure a physical world with a unity, objectivity, publicity, and repleteness to match the specifications of our ordinary beliefs.

Berkeley takes God to be a purely active being, not subject to any causal influence. Consequently, he thinks of God's perception of the physical world as an active conceiving rather than as a passive sense-experience;⁹ and, to stress the contrast with sense-experience, he often uses the terms ‘knows’ and ‘comprehends’ in place of ‘perceives’.¹⁰ The fact that God's perception is not sensory does not, of course, mean that what he perceives is not a sensible world—a world composed of the kind of sensible qualities and relations which feature in the content of human sense-experience. And it is doubly crucial for Berkeley that the object of God's perception should be sensible in this way. For otherwise the physical world would not meet the specifications of our ordinary physical beliefs; nor, indeed, would it be something of which we could even form a positive conception, and so would be no more respectable, as an ingredient of the ultimate reality, than the mind-independent world of the materialist.

What remains puzzling about this is why Berkeley should suppose that a sensible quality, or complex of qualities, can achieve some genuine realization by being divinely conceived. It is one thing to accept a sense-datum account of ordinary perception, which claims that the quality-patterns which feature in the content of sense-experience are genuinely realized (as sense-data) by featuring in that content. It is another thing to claim that such patterns are also genuinely realized simply by being conceived. And certainly, in the human case, this claim seems wrong. Surely I do not create a real colour-patch or a real sound simply by thinking of one.

Berkeley's answer would be that I do create a real colour-patch and real sound (though not real in the sense of physical) if I conceive of them by the framing of an image. For in that case there is, he would claim, something (the image) which is the internal object of my conception and which instantiates the sensible qualities in question. This could be disputed (indeed I would dispute it myself), but let us, for the sake of argument, suppose it to be true. Berkeley could then claim that God's conception of the sensible world is a form of imaging or something analogous. One reason for allowing it to be merely analogous to imaging is obvious: we cannot expect God's psychological operations to be quite like ours. But there is also another and more subtle reason. It is arguable that to frame an image of something is to represent oneself as sense-perceiving it. And, presumably, if God cannot have sense-experiences, he cannot (or at least would not) represent himself as having them either. At all events, it seems safer for Berkeley to appeal to the case of human imagination, as an example of an active conception with a sensible object whose esse is concipi, and then merely claim that God's conception of the sensible world is like human imagination in that respect.

One of the most curious aspects of Berkeley's new position is that, while retaining the Simple View, it commits him to a representative theory of perception. As he originally interpreted it, the Simple View gave us direct perceptual access to the physical world, by equating our own sensory ideas with the physical items we perceive. This, indeed, was one of the reasons why he found the view attractive. But by locating the physical world in God's mind, Berkeley is now committed to saying that our access to it is only indirect. We perceive physical objects, but only by perceiving certain other objects, i.e. our own ideas, which represent them. Berkeley himself, strangely, seems to be only half-aware that his position has changed. For although, in elaborating the theory, he stresses the fact that the physical world is now external to human minds, and speaks of God's ideas as the ‘archetypes’ of which our sensory ideas are the secondary copies, he still feels able, in the end, to claim that his theory preserves the direct realism of common sense. For in effect, as we saw, he claims that it is precisely this direct realism, combined with the sense-datum theory, which constitutes ‘the substance of what I advance’. No doubt this is what Berkeley had wanted to advance. But it was not the theory which actually emerged.

Part of what Berkeley liked about direct realism, as we noted, was that it puts our knowledge of the physical world on a firm foundation. For we can derive our physical knowledge from the character of our sense-experiences without having to make inferences to some external realm beyond what these experiences immediately present. This raises the question of whether physical knowledge is possible on Berkeley's new account. If the physical world is located in God's mind, do we have any way of discovering its character? Do we even have any reason to believe that such a world exists? Since Berkeley wants to avoid scepticism, it is essential to his project that these questions should be answered affirmatively.

Before we can deal with this epistemological issue, we need to get clearer about the relationship, as Berkeley conceives it, between our sensory ideas and the physical items in God's mind which they represent. On standard representative theories, a sensory idea is thought of as representing a particular physical object in virtue of two factors: first, a certain qualitative or structural resemblance between the idea and the object (or the object as it is projected onto the observer's viewpoint); and secondly, the fact that the physical object is, in some suitable way, causally involved in the production of the idea, with each represented object-feature having a special involvement in the production of the corresponding idea-feature. In Berkeley's theory, this second factor has to be replaced by something else, since he holds that volition is the only form of causal agency. What replaces it is the fact that, in the framework of God's volitional policies, the object, together with the ‘circumstances’ of the relevant human subject, serves as an occasion for God to cause the idea: ‘there is an omnipresent eternal Mind, which knows and comprehends all things, and exhibits them to our view in such a manner, and according to such rules as he himself hath ordained, and are by us termed the Laws of Nature.’¹¹ This sounds like Malebranche's position (that we see all things in God), but Berkeley is at pains to stress that the resemblance is only superficial.¹² The most important difference is that, in Malebranche's system, the physical things which God comprehends and exhibits to our view are external to him, as they are to us, while in Berkeley's system, they exist in God's mind, as the internal objects of his conception.

If God ensures that our sensory ideas accurately represent physical things, as he conceives them, then in one weak sense we are assured of physical knowledge. We have, that is, a reliable way of reaching the truth about the physical world by following the evidence of our senses. But, clearly, this does not settle the real epistemological issue. For what really matters is whether, on Berkeley's theory, we can have good reason for believing that our senses are reliable in this way. Do we have good grounds for believing that there is a world in God's mind and that God reveals it to us, in piecemeal and partial fashion, through the control of our sensory ideas?

On this point, Berkeley's own reasoning is woefully inadequate:

It is evident that the things I perceive are my own ideas, and that no idea can exist unless it be in a mind. Nor is it less plain that these ideas or things by me perceived, either themselves or their archetypes, exist independently of my mind, since I know myself not to be their author, it being out of my power to determine at pleasure, what particular ideas I shall be affected with upon opening my eyes or ears. They must therefore exist in some other mind, whose will it is they should be exhibited to me.¹³

This argument is clearly fallacious. The fact that our sensory ideas are not within our own volitional control does not, as such, establish that they or their archetypes (and it could only be their archetypes) exist in some other mind. Indeed, the whole argument seems to be based on a confusion of two senses in which the things we perceive may be independent of us: (1) the sense in which they are independent if we are not causally responsible for them, and (2) the sense in which they are independent if they do not exist in our minds. In effect, Berkeley reaches his conclusion by appealing to the first kind of independence and then reinterpreting it as the second.

Although Berkeley's argument is, as it stands, fallacious, it might still be possible to transform it into something sound and effective by inserting, between the premisses and the conclusion, certain further steps of reasoning. We could make a start in this direction by invoking another argument which Berkeley supplies, though to a weaker conclusion, and which we examined in our earlier discussion. I am thinking of his argument for God's volitional role—the argument that because our sense-experiences are not the product of our own volition and their orderly character is not of our own making, we should take them to be directly caused by God and their orderly character to be the reflection of God's volitional policies. If this argument is sound—and given Berkeley's basic assumptions, it seems very reasonable—what we need, to complete the whole argument, is some way of inferring God's perceptive role from his volitional role. It is the possibility of such an inference which we must now consider.

One thing which suggests that such an inference might be available is that God's volitional strategy already includes a conception of a certain kind of physical world. As we have seen, what renders our experiences orderly is that they are conspicuously amenable to physical interpretation, and God's overall strategy, in selecting the types of experience to cause, is to make them orderly in that way. In effect, God has to start with a conception of the kind of physical world whose existence he wants us to accept and then ordain that our experiences are to be such as to accord with, and invite our acceptance of, the hypothesis that such a world obtains. If we now equate the actual physical world with the internal object of this conception—with the complex sensible idea by which God identifies the kind of world in terms of which he frames his volitional strategy—it seems that the epistemological problems are solved: we are entitled, by means of an explanatory inference, to move from the sensory order we observe to the postulation of God's volitional strategy; and from the postulated strategy we can then deduce the existence and character of the physical world. In short, it seems that the physical world becomes epistemologically accessible to us, because God's perceptive (i.e. conceptive) role is implicit in his volitional role: we have good reason for believing that the physical world exists and that our sensory ideas accurately represent it, since the divine conception, by which it exists, forms an indispensable part of the divine strategy, which the sensory order reveals. Berkeley himself may be making this point when he remarks in his Notebooks (PC, [Philosophical Commentaries] 812):

The propertys of all things are in God i.e. there is in the Diety Understanding as well as Will. He is no Blind agent & in truth a blind Agent is a Contradiction.

This looks promising. And I think it is the best argument that we can construct on Berkeley's behalf. Nevertheless, I think that it can be shown to fail. For while we may be justified in explaining the sensory order in terms of God's volitional strategy and also justified in taking this strategy to involve a conception of a certain kind of physical world, we are not, I think, justified in supposing that this conception has an internal object of the appropriate sort. In fact, there are two problems here, as we shall now see.

Let us assume for the moment, in line with Berkeley's requirements, that the kind of world in terms of which God frames his volitional strategy is purely sensible, i.e. is a spatio-temporal arrangement of sensible qualities. Let us also assume, again in line with Berkeley's requirements, that it is possible for God to conceive of the arrangement by framing an image, or quasi-image, which instantiates it, thus providing an actual sensible particular, of the appropriate world-type, as the internal object of the conception. Even so, it is not clear what entitles us to suppose that it is this kind of conception which features in the volitional strategy. All that God's adoption of the strategy entails is that he, in some way or another, knows the character of the hypothetical world with which he intentionally makes our experiences accord. To suppose, in addition, that this knowledge is derived from some quasi-imagist conception, with an internal object which exemplifies the character in question, seems to be pure speculation. And without this supposition, there is no valid step of inference from God's adoption of the strategy to the existence of an actual physical world in his mind.

This point alone may well be decisive. But even if it is not, there is a second point which undermines the proposed argument at a more fundamental level. The only reason why we are even tempted to suppose that the conception within the volitional strategy is of a quasi-imagist kind is that we are assuming the kind of world in terms of which the strategy is framed to be purely sensible. And in making this assumption, we are implicitly assuming that the physical theory by reference to which the sensory order is defined is some version of naïve realism: we are assuming, in effect, that what renders our experiences orderly is their accordance with the hypothesis that we are located in a certain kind of sensible world (some concrete spatiotemporal arrangement of sensible qualities) and that our sense-experiences present to us portions of this world in the perspective of our current viewpoint. Now when it is contrasted with the claim that the sensory order can be specified in purely phenomenal terms, this assumption seems perfectly acceptable. It is, indeed, a move in the right direction, since vastly more of the sensory order can be captured at the level of this naïve physical interpretation than at the level of phenomenal regularities. However, our physical theorizing does not stop at naïve realism. It develops into the scientific theorizing of chemistry and physics, which postulate objects and properties of a non-sensible kind and, indeed, take these objects and properties to be the fundamental ingredients of the physical reality. Now the crucial point is that it is only by reference to the scientific theory that the sensory order can be fully and exactly specified. The order revealed by naïve realism considerably surpasses anything discernible at the phenomenal level, but it is crude and incomplete in relation to the order revealed by science. This does not mean merely that science postulates simpler laws. It means that the ways in which experience is orderly in relation to the physical world of science (or strictly, the physical world plus certain psychophysical laws) subsume and transcend the ways in which it is orderly in relation to the world of naïve realism. Certain aspects of the sensory order simply cannot be specified at all in the naïve realist's theoretical framework.

Since what calls for explanation is the whole sensory order, not just those aspects of it which are captured by naïve realism, we must suppose that the kind of physical world in terms of which God frames his volitional strategy is the kind of world which would be postulated by a hypothetically perfected science with access to all the actual and possible sensory evidence. It follows that, in framing this strategy, God has no need to form any conception of a sensible world. Indeed, if he confines himself to the perspective of science, his conception of the relevant kind of world—the kind which informs his strategy—will be ‘topic-neutral’: it will be a conception which specifies the structure and laws of the world without specifying its intrinsic content.¹⁴ Clearly, such a conception would not have an internal object of the appropriate sort—one which instantiates the structural and nomological properties thus conceived. It would not, indeed, have an internal object at all in the relevant sense.

V

In his Principles of Human Knowledge, which was published a few years before the Dialogues, Berkeley does not equate the physical world with the internal object of God's perception (conception). There are, admittedly, passages which seem to point in this direction. Thus in section 6, having claimed that ‘all those bodies which compose the mighty frame of the world, have not any subsistence without a mind’ and that ‘their being is to be perceived or known’, he adds:

… consequently so long as they are not actually perceived by me, or do not exist in my mind or that of any other created spirit, they must either have no existence at all, or else subsist in the mind of some eternal spirit …

And he seems to pick up this point in section 48, when dealing with the problem of what happens to physical objects ‘during the intervals between our perceptions of them’. But these passages fall short of the Dialogues theory in three crucial respects. First, the perceptive role of God is left as something merely hypothetical. The claim is not that God does perceive physical objects at times when we do not, but only that he may do so. Secondly, as well as being merely hypothetical, God's perception is only invoked as a supplement to human perception. The possibility which seems to matter to Berkeley is not that God perceives the whole physical world, but that he perceives those portions of it which we fail to perceive. Thirdly, and most crucially, however extensive God's perception, the physical world is thought of as something to which we also have direct perceptual access. Perhaps God keeps my desk in existence at times when I do not perceive it. But while I am perceiving it, my sensory ideas are genuine elements of that physical desk. They are not merely representations of some desk-like idea which exists in God's mind and which only God can directly perceive. Indeed, even the suggestion that God's perceptions may serve as a blueprint for his volitional procedures is one which, it seems, Berkeley finds ‘too extravagant to deserve a confutation’.¹⁵

Not only do the claims in sections 6 and 48 fall short of the Dialogues position, but they are also out of line with the theory which predominates in the Principles. Right at the outset, Berkeley explicitly allows for the possibility of unperceived physical objects. For in section 3 he says of his table: ‘if I were out of my study I should say it existed, meaning thereby that if I was in my study I might perceive it, or that some other spirit actually does perceive it.’ The clear implication is that even if no one is actually perceiving the table, it can still continue to exist solely in virtue of being, in principle, perceptible. Now this implication, of course, is not compatible with the Simple View, which equates the existence of a physical object with its being actually perceived. And, on the face of it, it is the Simple View which Berkeley is defending in the next few sections. This might lead one to dismiss the concession in section 3 as a momentary aberration. But this would be a mistake. For the same point occurs in section 58, with more elaboration and with the full awareness that the Simple View is being denied. The point is prompted by a hypothetical objector, who claims that Berkeley cannot accommodate the finding of astronomy that the earth moves, since ‘the motion of the earth is not perceived by sense’. Berkeley's reply is that the motion of the earth reduces to the fact that we would perceive it, if we were transported to a suitable viewpoint. Moreover, he claims that this reply is in agreement with the principles he has advanced.

In making this claim, Berkeley has more in mind than his brief remarks about the table in section 3. What he has primarily in mind are sections 29-36, in which he tries to show how his position preserves the existence of an objective physical world. In these sections, he does two things. First, he draws attention to the fact that our sensory ideas are not the product of our own volition, and that they collectively exhibit a ‘steadiness, order and coherence’ which is not of our making. From this he infers that they are the result of God's volition, and that their orderly character is a reflection of God's fixed volitional policies, which constitute, from our standpoint, the ‘laws of nature’. (All this we have already examined.) Secondly, he appeals to the first point to rebut the objection that ‘by the foregoing principles, all that is real and substantial in Nature is banished out of the world: and instead thereof a chimerical scheme of ideas takes place’, in which ‘houses, rivers, mountains, trees, stones’ are ‘but so many chimeras and illusions on the fancy’ (section 34). Berkeley's answer is that there is an objective world, and one which our perceptions directly reveal, simply in virtue of the orderly and thematic way in which human experience is divinely controlled. Our sensory ideas exist only in our minds; but they are not just ‘illusions on our fancy’, since they feature in an objective and independent sensory order sustained by God's volitional policies. They are, as it were, subjective in their existence, but objective in their organization. And it is the presence of this organization, divinely imposed, which allows us to speak of a physical world.

Taken to their logical conclusion, these points would yield a form of reductive phenomenalism, in which the physical world is excluded from the ultimate reality, but admitted as something whose existence is logically sustained by God's volitional strategy. In effect, a certain kind of physical world is held to be non-ultimately realized in virtue of the fact that our experiences are organized as if it were ultimately realized. This, of course, as well as according the physical world the requisite objectivity, unity, and publicity, allows for the existence of unperceived physical objects and of perceived physical objects with unperceived aspects. And, in the sorts of case which Berkeley considers, the organizational factors on which the existence of these unperceived items depends involve (though they are not exhausted by) our potential to have the appropriately perceptive experiences in suitable circumstances.

Berkeley himself, it must be admitted, does not seem to think of his position as fully reductive in this way. Indeed, in sections 29-36, he seems to think that our sensory ideas are, quite literally, elements of the physical world, although it is their organization which makes them that, by determining their spatio-temporal arrangement. In effect, this would involve giving a reductive account of physical facts, while retaining certain physical entities in the ultimate ontology. Of course, the account would have to be ontologically reductive with respect to unperceived objects. And this should have led Berkeley to adopt the reductive account quite generally. For it would be absurd to suppose that the physical desk (or its perceptible surface) is constantly varying in its ontological status according to whether it is currently perceived. Clearly, whether he realizes it or not, Berkeley's phenomenalistic approach commits him to adopting the fully reductive position, which excludes both physical facts and physical entities from the ultimate reality and takes them to be the logical product of the organization of human experience. And, for the purposes of our discussion, we may take this to be, in the Principles, Berkeley's true position.

Although this position is incompatible with the Simple View, it comes closer than the Dialogues theory to that version of the view which Berkeley initially favoured. For by making the existence of the physical world depend solely on God's organization of human experience, it comes closer to the claim that physical objects are things which we directly perceive. Strictly speaking, of course, this claim is rejected, since the immediate objects of perception are ideas and ideas are not to be taken as literal constituents of the physical world. But since the physical world is nothing over and above the organization of our experience and since this organization is reflected in the sensory order whose elements we directly detect, there is a sense in which the position preserves the spirit of the claim. Certainly it does so when contrasted with representative theories of perception, like that of Locke and like that of Berkeley himself in the subsequent Dialogues.

In preserving the spirit of this claim, it also, of course, avoids those problems for the Dialogues theory which we discussed in the previous section. Where the phenomenalist account itself seems vulnerable is in coming perilously close to physical nihilism. For there seems to be only a fine dividing line between saying that there is a physical world, but one which is wholly created by the organization of human experience, and saying that there is no physical world, but our experiences are organized as if there were. Indeed, on the face of it, the first formulation seems to be only a euphemism for the second. Berkeley might reply that what substantiates the claim that there really is a physical world is that the sensory organization is imposed by God with the intention of creating a world-for-us. But, in this respect, there seems to be only a difference in name and motivation between Berkeley's benevolent God and Descartes's malevolent demon. Why should we say that, by imposing the organization, God succeeds in creating a physical world rather than that he creates the illusion of such a world? Looked at objectively, is not the latter the more natural conclusion?

Modern phenomenalists have sometimes tried to avoid this problem by claiming that statements about the physical world can be analysed into statements about sense-experience, so that the existence of a certain kind of physical world can be logically deduced from a description of an appropriate sensory organization. It would then be incoherent to say that there is no physical world, but our experiences are organized as if there were, since in conceding the sensory facts, one would be implicitly acknowledging the physical facts as well. However, though Berkeley himself shows some leanings towards it (e.g. in the passage from section 3 quoted above), this analytical version of phenomenalism is clearly untenable. Even if the existence of the physical world is wholly sustained by the sensory organization, there is no denying that our physical language, with its distinctive ontological perspective, enjoys a certain conceptual autonomy—an autonomy which prevents any replacement of physical statements by sensory statements without some radical change in meaning. Maybe the ultimate reality is as Berkeley describes it. But the commitment of physical statements to an external and mind-independent world cannot be simply eliminated by semantic analysis. The suspicion remains, then, that, in denying the ultimacy of the physical world, Berkeley has, in effect, denied its existence altogether.

I think that Berkeley's best response to this problem would be to appeal to the distinction, which I have drawn in another context,¹⁶ between prospective and retrospective sustainment. If A is a set of facts and B some fact which A logically sustains, the sustainment is said to be prospective if and only if someone who knew A could, on that basis alone, establish its sustainment of B; and the sustainment is said to be retrospective if and only if it could only be established by someone who, in addition to knowing A, had independent knowledge of B. The relevance of this distinction to the present issue is that while the conceptual autonomy of the physical language undermines the claim that the existence of the physical world is prospectively sustained by the sensory organization, it does not undermine the claim of retrospective sustainment. It excludes the possibility of establishing the existence of a physical world on the sole basis of the sensory organization. But it leaves open the possibility of arguing that, given the existence of the physical world, it is by this organization that its existence is ultimately sustained.

We can illustrate this point by developing an analogy within the framework of physical realism. Let us suppose, from this realist standpoint, that science has shown the physical world to be ultimately composed of just space and time (or space-time), a stock of minute, insensible particles (characterized, say, as point-sources of causal influence), and certain laws controlling the spatio-temporal arrangement of these particles. It has also shown, let us suppose, that there are certain psychophysical laws, assigning experiential effects to certain ‘neural’ configurations of particles, and that these laws, together with the physical laws and facts about particle-arrangement, account for the sensible appearance of things to us. The question now arises as to what status we should accord to the sensible world of ordinary perception (the world of extended, coloured, and tangible objects) in the light of these scientific findings. Clearly, if we distance ourselves from the perspective of ordinary perception and just focus on the physical world and psychophysical laws as science specifies them, we cannot legitimately infer the existence of a sensible world of the kind we ordinarily accept. All we can infer is that the subjects whose experiences are controlled by the physical processes and psychophysical laws would find it useful and natural to think of themselves as the inhabitants of such a world. On the other hand (though leaving aside any independent sceptical arguments), it seems possible to start from the epistemological perspective of our ordinary beliefs and consider how the sensible world, whose existence we are now taking for granted, is metaphysically related to the underlying world which science reveals. And, proceeding in this fashion, the legitimate conclusion is that the existence of the sensible world is wholly sustained by the scientific world and the psychophysical laws. It might be objected that the scientific findings, if correct, would show the ordinary perspective to be illusory, so that our ordinary beliefs and the scientific account cannot be coherently combined. But this, I think, is not so. The scientific findings on their own fail to validate the ordinary perspective. But this does not mean that someone who enjoys the perspective has to repudiate it before he can accept the findings. Nor does there seem to be anything else which obliges him to do this. Indeed, if there were, the supposed findings would themselves be discredited. For, since we cannot adopt a God's eye view, it is only by theorizing with respect to the evidence available in this perspective that we can acquire knowledge of the scientific facts.

In an analogous way (though the analogy is not perfect) there are two distinct perspectives in which we can evaluate the claim that the sensory organization sustains the existence of a physical world. If we just consider in the abstract an ultimate reality composed of only God and a group of finite minds whose experiences he controls in the appropriate (world-suggestive) way, we can find no basis for accepting the existence of a physical world. At most, we are entitled to conclude that the postulation of such a world would, from both a descriptive and an explanatory standpoint, be useful and natural for the minds whose sensory organization we are contemplating. But this does not prevent us starting from the assumption that there is a physical world and then employing anti-realist arguments to establish its phenomenalistic status—so that the sensory organization is then viewed as the reality which underlies and wholly sustains this physical world, whose existence is taken to be independently secure. And if we can find such arguments and if, apart from these arguments, our acceptance of the physical world is well-founded, then this seems to be the correct procedure. The only objection might be that, since it is an essential part of our conception of a physical world that it be something external to, and logically independent of, human minds, the phenomenalist view is already excluded by the ontological perspective of our physical beliefs. But this can be met, I think, by distinguishing the two theoretical frameworks in which the question of the world's externality can be raised. Within the framework of the physical theory it is a conceptual truth that the world is external to human consciousness and (apart from its sensible appearance) independent of how human experience is organized. It is this, indeed, which gives the physical language its conceptual autonomy and prevents its analysis in sensory terms. But within the framework of the philosophical theory, which is concerned with the metaphysical status of the external world which the physical theory describes, it is a conceptual truth that the existence of this world is ultimately sustained by the sensory organization. I can see no incoherence here.¹⁷

Of course, if Berkeley were to follow the line I am suggesting, he could not employ his phenomenalism as a way of defeating a radical scepticism about the physical world—a scepticism which called all our physical beliefs in question and demanded their justification in terms of something else. For if the phenomenalistic sustainment is only retrospective, we have to accept the existence of the physical world before we can establish the sustainment. As it turns out, however, Berkeley would not be sacrificing anything here. For since we need to rely on our physical knowledge to obtain most of our information about the sensory organization (this is obvious in the case of our information about other minds, but also holds, to a large extent, in the case of a subject's information about himself¹⁸), phenomenalism (even with prospective sustainment) would not defeat the sceptic anyway. Whether anything would defeat him is a further question and one which I shall not pursue here.

No doubt it was because Berkeley thought his phenomenalist account was in danger of collapsing into nihilism that he came later to adopt his realist position. It must be stressed, however, that his revised account does not repudiate phenomenalism entirely. For although the physical world gets identified with a complex idea in God's mind, this idea only achieves its physical status by serving as the blueprint for God's volitional procedures. This is why, in discussing the Mosaic account of creation, Berkeley speaks of a ‘twofold state of things’, the one ‘archetypal and eternal’ (the world as it exists in God's mind) and the other ‘ectypal or natural’ (the world as God volitionally creates it for us).¹⁹ Berkeley is not here acknowledging two physical worlds, one to be construed realistically and the other to be construed reductively. But he is acknowledging that the thing which exists eternally in God's mind only qualifies as our world, and thereby as a physical world, because God empirically manifests it to us through the organization of experience. To this extent, even his revised account remains phenomenalistic. Perhaps this goes some way towards explaining how, at the end of the Dialogues, he manages to misinterpret his own position, in supposing that, despite its location in God's mind, our perceptual access to the physical world is still direct.

Notes

1. For a specific version of this see my The Case for Idealism (Routledge and Kegan Paul, London, 1982) ch. 11.

2. In effect, I have tried to provide an improved version of Berkeley's argument in The Case for Idealism, Part II.

3. For a full discussion of this see my The Case for Idealism, chs. 4-5.

4. I think it is incoherent, but not manifestly so. See the reference in note 1.

5. This is my terminology in The Case for Idealism.

6. Dialogues, 262.

7. This at least is how I interpret the position developed in the Second and Third Dialogues. See especially pp. 211-15, 230-1, and 234-5.

8. See Dialogues, 248.

9. See especially Dialogues, 240-1.

10. Berkeley's varying terminology with respect to God's perception (i.e. conception) of the physical world is nicely set out and clarified by G. Pitcher in Berkeley (Routledge and Kegan Paul, London, 1977), pp. 175-9.

11. Dialogues, 231.

12. Dialogues, 214. Malebranche elaborates his position in his De la Recherche de la Vérité.

13. Dialogues, 214-15.

14. On this notion of topic-neutrality, see my The Case for Idealism, chs. 4-5.

15. Principles, 71.

16. In The Case for Idealism, ch. 14.

17. See The Case for Idealism, ch. 15.

18. See The Case for Idealism, p. 228.

19. Dialogues, 250-6.

Abbreviations

The following abbreviations have been employed throughout the text:

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

PC: Philosophical Commentaries (Works, vol. i).

Principles: The Principles of Human Knowledge, Part I (Works, ii).

Principles, Intro.: Introduction to The Principles of Human Knowledge (Works, ii).

Dialogues: Three Dialogues between Hylas and Philonous (Works, ii).

Alc.: Alciphron or the Minute Philosopher (Works, iii).

References to PC are by entry number; to Principles, De Motu (Works, iv), and Siris (Works, v), by section number; to Dialogues and Alc., by page number. Other references to Works are by volume and page number.