Is Berkeley best understood as a phenomenalist?

Introduction

Berkeley denied the existence of non-mental substances. Yet he claimed to accept the existence of “sensible objects” like houses, mountains and rivers, repeatedly emphasising that “we are not deprived of any one thing in Nature” by his theory. This raises the question of what he thought the existence of such objects involves.

This essay explores the possibility that he gave the phenomenalist answer, which is that it involves the truth of certain subjunctive conditionals about what would be experienced in various situations. After showing that he was in a position to give this answer, I argue that his discussion of creation forced him to do so, at least there. But I then consider three claims he makes which appear inconsistent with phenomenalism: that the existence of objects we do not perceive depends on divine perception; that only perceived things (and perceivers) exist; and that objects are ideas or collections thereof. These suggest that he gave an idealist account of sensible objects, identifying them with collections of human and divine ideas. Many commentators, including Pitcher (1977) and Bennett (2001), read this as his primary account, dismissing his phenomenalist explication of creation as marginal. I endorse this reading, but rather than simply dismiss the phenomenalist passages as inconsistencies (which should be a last resort) I shall argue that his slip into phenomenalism was natural, because his idealism came very close to this. It fell short only because he mistakenly felt the need to identify objects with something in his basic ontology. This was a natural mistake for someone in an age before the techniques involved in phenomenalist analyses were familiar. It makes Berkeley’s theory less elegant than phenomenalism, which is one of several reasons he would be best off understood as a phenomenalist, as Mill famously argued. This partly explains the temptation among sympathetic commentators to interpret him as one. However, I shall argue that he is best understood as an idealist, albeit one closer to phenomenalism than is usually appreciated.

What Berkeleyan phenomenalism would involve

As promised, I begin by showing that Berkeley was in a position to be a phenomenalist: he could accept the truth of the subjunctive conditionals which phenomenalism deems necessary and sufficient for the existence of sensible objects. P3 offers just such a conditional: “if I were out of my study I should say [the table there] existed, meaning thereby that if I was in my study I might perceive it”. Whether or not this betrays a commitment to phenomenalism, it does show that Berkeley thought these conditionals could be true. There might be no causally efficacious material substances or free-standing causal laws to ground their truth in his system, but there was a God ready to give Berkeley sensations of a table should he wander into his study. Berkeley could thus avoid what has been felt to be a weakness of phenomenalism: positing the ungrounded, unexplained and unconnected truth of a plethora of subjunctive conditionals. Whereas atheistic phenomenalism cannot even explain groups of phenomena as the results of general laws (because it typically treats these laws as abbreviated descriptions of the groups), Berkeley’s theism explains all these conditionals as part of God’s benevolent plan.

Stoneham claims that Berkeley accepted phenomenalist analyses in terms of subjunctive conditionals without adopting what he calls the ontological aspect of phenomenalism, which is that objects consist of possible but non-actual ideas (rather like Mill’s “permanent possibilities of sensation”). He thinks Berkeley grounded the conditionals in God’s decrees in the way I have just described. So far, this is a description of what I shall henceforth call ‘theistic phenomenalism’, but, as Stoneham himself recognises, his account merely “brings [Berkeley] very close to phenomenalism”. This is not, as he suggests, because phenomenalists must employ unperceived ideas in their ontology: they do not need to provide this sort of substitute for sensible objects. It is because Stoneham’s account equates objects with collections of actual ideas, which is characteristic of idealism. (Stoneham’s suggestion that these ‘collections’ can exist when none of their members do, because phenomenalist conditionals still hold, does not change this.)

Return now to the first conditional I quoted from P3. It falls short of the conditionals employed by phenomenalists in two respects. Firstly, phenomenalists would suggest neither that the table’s being perceptible to Berkeley should he enter his study was sufficient for its existence (since it would also have to be perceptible to others, for example), nor that it was necessary for this (since it isn’t specified that, for example, the study is lit). They offer numerous conditionals covering all possible circumstances in which the table could and could not be perceived. But the fact that Berkeley gave an over-simple example doesn’t show that he missed these details, or wasn’t a phenomenalist: he was a terse writer, and his books couldn’t fit full phenomenalist analyses of “Berkeley’s table exists”.

Secondly, a proper phenomenalist analysis would not involve conditionals with antecedents like “if I was in my study” unless these could themselves be described in terms of mental events themselves. But Berkeley is committed to this being possible on most interpretations, even if those in which divine perceptions feature among those mental events violate the spirit of modern phenomenalism.

This understanding of what theistic phenomenalism would involve will help us appreciate how close Berkeley came to it. As Berman notes, it is quite distinct from the phenomenalism of Mill (1865) and Ayer (1936) because Berkeley believed in “spirits which cannot be known or experienced by sense.”

The rival passages

So Berkeley could have been a phenomenalist. To decide whether he was, we must examine his writings.

As I noted, these contain both apparently phenomenalist and apparently idealist passages. Postponing individual discussion of these, let me give a partial list. Apparent phenomenalism occurs at: PC98, PC282, PC290, PC293-293a, PC302, P3, P58, 3D251 and Berkeley’s letter to Percival of September 6^th, 1701. Apparent idealism can be found at: P1, P33-34, P38, P48, 2D212, 2D214-215 3D230-231, 3D235, and 3D262.

As you can see, many of the phenomenalist passages occur in Berkeley’s notebooks, which clearly contain some views he never accepted, or abandoned before publishing his doctrines, since some notes contradict his published works (and each other). He may have recorded these phenomenalist thoughts for further consideration, and then rejected them. Seemingly phenomenalist passages do occur in Berkeley’s published works, but they are outnumbered by idealist ones.

However, the interpretive debate cannot be settled by counting how many lines support each position and then dismissing the less common one – as Pitcher and Bennett come close to doing. Nor can one set of passages be dismissed as the result of misleading phrasing: the idealist passages are too numerous, the phenomenalist passages are too neatly prefigured by the notebooks, and both sets span books published enough years apart for Berkeley (generally a careful philosopher) to spot poor phrases. Berkeley claims to give a precise analysis of existence-claims about sensible objects and ascriptions of sensible qualities. P3 speaks explicitly of “attend[ing] to what is meant” by these (which is a reason to take its phenomenalist element seriously when considering this question). Many of Berkeley’s arguments rested on claims about meaning, and later works like Alciphron show that Berkeley had an abiding interest in philosophy of language. Furthermore, PC408 echoes other notes: “I must be very particular in explaining wt is meant by things existing … wn not perceiv'd as well as wn perceiv'd. & shew how the Vulgar notion agrees with mine when we narrowly inspect into the meaning & definition of the word Existence”.

This counts against my suggestion that Berkeley offered two different theories of sensible objects, though it favours my attempt to show that these are closely related. But no other account is available, for attempts to reconcile his apparently idealist claims with phenomenalism fail, and we shall now see that his discussion of creation is intractably phenomenalist, for reasons he would have taken seriously. This scotches claims that he offered only an idealist theory.

Creation

Late in the Dialogues, Hylas understands Philonous to “make the existence of sensible things consist in their being in a mind”. This is an idealist position, and Philonous argues that it cannot make sense of the scriptural account of creation: if an object’s existence requires a created mind to perceive it, then Genesis’ claim that plants existed before animals must be false; whereas if it suffices that an object exist in God’s immutable mind, all objects must exist eternally since “God knew all things from eternity”. It is hard to see how the devout Berkeley could have remained a consistent idealist after noticing these consequences, and Philonous does not dispute that they are unacceptable. Instead, he offers a phenomenalist account of physical objects which can avoid them. He does not repudiate Hylas’ understanding of him, but – unless he is brazenly accepting two incompatible theories – this suggests that he sees his earlier claims that objects exist only in relation to perceiving minds (to which Hylas is referring) as closely related to his phenomenalist account, as I’ll argue.

According to this account, things began their existence “in respect of finite spirits … when God decreed they should become perceptible to intelligent creatures” in appropriate circumstances. If perceptibility is the condition for something beginning to exist, then it – and not actually being perceived – is all existence requires. Philonous acknowledges that this existence-in-respect-of-finite-spirits is “relative or hypothetical”. This leads Pitcher to distinguish it from the “first-class kind of existence that belongs to things that are actually perceived”, which is a mistake: the real world, for Berkeley, is that which God created for us to move and act in. As Winkler notes, existence-in-respect-of-finite-spirits is “real existence” for Berkeley, whereas archetypal existence in God’s mind is not, because God had an idea of plants before they really existed.

Philonous’ account of creation is clearly phenomenalist, and those who interpret Berkeley as a phenomenalist often dismiss it as a “brief interlude” or “‘ad hoc’, ‘tentative’, and ‘marginal’”. But Berkeley accepted it over a long period: it was prefigured in his notebooks, and he gave an identical account of how “sensible things … became perceptible (i.e. were created) in the same manner and order as is described in Genesis” in a letter written to Percival four months after the Principles’ publication. Accepting scripture was no marginal concern for Berkeley, so a phenomenalist account of creation was forced on him by his belief that Genesis required him to treat objects as existing just when they were perceptible to us (even if we did not perceive them). Perhaps Bennett is right that it would be pointless for God to make things perceptible before there were humans, but it would be equally pointless to create a world before creating the humans it is made for, and Berkeley clearly believed God did this.

Divine perception

We now turn to Berkeley’s three apparently idealist claims, which appear to rule out phenomenalism. The first of these is that objects not perceived by us must be perceived by God.

Thomas notes that this claim is only made explicitly at 2D212 and 3D235-236, and he argues that these passages should not be taken too seriously because they occur in an “apologetic context”. Thomas is, I submit, making the old mistake (discussed in my other essay) of not taking the Dialogues, and Berkeley’s attempt to reconcile his theory with religion and common sense, seriously enough. His argument for doing so is that, at 3D241, Berkeley maintains that God “perceives nothing by sense”, but this is not inconsistent with the earlier passages: in Berkeley’s terminology, ‘perception’ includes imagination and conception as well as sense-perception. Pitcher shows that Berkeley’s “purely active” God cannot undergo passive sense-perception, but that ‘perception’ in 2D212 and 3D235-236 probably means ‘conception’. So Berkeley required that God have ideas of objects we don’t perceive, but not that he passively sense them. (As we shall shortly see, Winkler shows that this requirement is not strictly inconsistent with phenomenalism.)

Pitcher reads Berkeley as claiming that divine perception of objects suffices for their existence. This is inconsistent with phenomenalism, but the passages Pitcher cites do not support his interpretation. The first set of passages actively suggest that God must not only perceive real objects, but also intend to make us perceive them in appropriate circumstances (as on theistic phenomenalism); the second set say only that divine perception is necessary for existence, not that it is sufficient. A typical example of the first set is 2D214-215’s claim that ideas of real things must “exist in some other mind, whose will it is they should be exhibited to me”; this suggests that when the second set of passages say, for instance, that things “are necessarily perceived by an infinite mind”, they stop short of saying that this perception is sufficient for objects’ existence because God’s intention to exhibit them to us is also necessary. However, while compatible with the phenomenalist interpretation, this second set of passages poses two problems for it: they at least imply that nothing besides being divinely perceived is necessary for being a real thing; and it is sometimes claimed that phenomenalism wouldn’t even consider it necessary that objects be actually perceived.

The first problem is not insuperable. The omission of a phenomenalist element in lines like “every unthinking being is necessarily, and from the very nature of its existence, perceived by some mind” may imply (i.e. suggest) that no such element is also necessary, but it does not entail this. The second problem is more serious, for it claims that requiring objects to be perceived is inconsistent with phenomenalism. However, Winkler suggests that Berkeley thought that God must perceive all things because He could not will that they be perceptible to us without Himself perceiving them. This suggestion is compatible with Winkler’s phenomenalist interpretation of Berkeley: let’s now examine it.

Blind volition

Winkler demonstrates that almost all philosophers of Berkeley’s time thought that one could not will ‘blindly’, i.e. without understanding what one willed, and that Berkeley’s notebooks show agreement on this point. PC674 says without “perception there is no volition”. PC743 and PC841-842 echo this, and PC812 applies it to God: “The propertys of all things are in God … He is no Blind agent & in truth a blind Agent is a Contradiction”. PC812 could be read as explicitly stating that God perceives “all things” because he is not a blind agent. This suggests that Berkeley’s God could only understand what it is for someone to have an idea by having that idea Himself. Against the charge that this is overly anthropomorphic, Winkler could point to Berkeley’s claim that it as an advantage of theism that we could understand God’s existence by understanding what it was like for our minds to exist. And Winkler’s reading need not be too anthropomorphic; he could capitalise on PC675’s claim that God, like us, can understand “painfull” ideas without feeling pain.

Winkler suggests the denial of blind agency is implicit in Berkeley’s arguments that God must perceive what we do not. He thinks 2D214 relies on this denial, for otherwise Philonous would not hastily conclude from the fact that “things … exist independently of my mind, since I know myself not to be their author” that “they must therefore exist in some other mind, whose will it is they should be exhibited to me”. Berkeley’s very haste here (and at 2D230, which makes a similar move) does suggest that he found it obvious that if God willed us to have ideas, He would have such ideas himself; i.e. that, as Leibniz put it, “He perceives them, because they proceed from him”. (Strictly speaking, denying blind agency would show only that God must have such ideas as He creates in us when he creates them. To explain Berkeley’s claim that God perceives objects when we are not perceiving them, Winkler would have to say that he thought God must perceive the ideas He intends to give us should it become appropriate. But this is a natural extension of the denial of blind agency: if it is plausible that we must understand what we will, it is also plausible that we must understand what we intend to will in appropriate circumstances.)

Tweyman and Bennett criticise Winkler’s claim that Berkeley rejected blind agency in the Principles and Dialogues; they note that neither work makes the rejection explicit. Winkler offers two reasonable responses: Siris, a later work, continues to deny blind agency; and Berkeley might simply not have mentioned the widely-accepted denial of blind agency in his in-between works. Tweyman and Bennett both suggest that Berkeley saw that his P28 account of imagination creating previously non-existent ideas involved blind agency, but Bennett concedes that this is a “guess”. Berkeley gives no indication of seeing this problem, or the other problem with denying blind agency which Tweyman suggests he would have seen, namely that it robs God of control over the ideas in his mind by requiring that they exist there before he wills them to.

Tweyman also argues that blind agency was not impossible for Berkeley’s God, because this would limit God’s power, and Berkeley deems Him “above all limitation or prescription whatsoever”. But this line is part of Philonous’ denial that God needs instruments to act, and so does not, as Tweyman needs to maintain, imply that God can do anything which doesn’t violate analytic truths. Such a strong view of God’s omnipotence might let Him be a blind agent, since Tweyman’s claim that it is not analytic that God is unable to will blindly is plausible. But even this is not certain, and Tweyman provides no direct evidence for his attribution of this position to Berkeley, whereas we have seen evidence that Berkeley denied divine blind agency.

All this shows that Winkler’s understanding of Berkeley’s views on blind agency is at least plausible. What is implausible is his suggestion that Berkeley required God to perceive objects we do not because he denied blind agency, rather than because he was an idealist. Winkler’s claim that Berkeley studiously avoided implying that divine perception was sufficient for an object’s existence (rather than simply because the mind which causes our ideas of them must itself perceive similar ideas) strains credulity. Yes, P48 says only that objects we don’t perceive “may” be perceived by another mind, but 2D212 says “there must be some other mind wherein they exist”. This doesn’t prove Berkeley wasn’t a phenomenalist (pace Broad), but it does suggest this. It seems that his requirement that objects be perceived at all times followed not from a denial of blind volition, but from his famous dictum that their esse is percipi.

Esse is percipi

As presented in P3, “esse is percipi” is a claim about meaning, and one which applies to objects as much as ideas. It sits uneasily with the phenomenalism suggested by P3’s “if I were out of my study I should say [the table there] existed, meaning thereby that if I was in my study I might perceive it”, because this does not require that the table is actually perceived. This suggests that we should read the continuation of this line – “or that some other spirit actually does perceive it” – as Berkeley’s preferred, idealist meaning. But that goes against the line’s natural reading, and the phenomenalist discussion of creation, so we should see whether it can be reconciled with phenomenalism.

Winkler simply suggests that Berkeley’s requirement that things are actually perceived resulted from his denial of blind agency. But it is less plausible that Berkeley would have left readers to spot this hidden premise here than at 2D214 and 2D230, for it is even less obviously at work. He seems instead to think that ideas, like thoughts and passions, obviously cannot “exist without the mind” – and that it follows that objects cannot either (which it would, on idealism). If his phenomenalist discussion of creation – which states that real existence is relative to finite perceivers, and that some objects existed before perceivers were around to perceive them – can be reconciled with his claim that only perceivers and perceived things exist, it must be in some other way.

Perhaps there is such a way, for at the final stage of analysis theistic phenomenalism does speak only of perceivers and perceived things. Unperceived objects, like the first plants in the Garden of Eden, are admitted only at a less basic level of discourse, so Berkeley’s claim that esse is percipi or percipere can still be taken to hold at the most basic level; just as some atomists claim that there exist only atoms and the void while allowing everyday speakers to claim that macroscopic objects exist.

This interpretation could survive what Bennett describes as the “anti-phenomenalist skirmish” of 3D234, which sees Philonous responding to Hylas’ suggestion that sensible things need only be perceivable rather than perceived by asking: “And what is perceivable but an idea? And can an idea exist without being actually perceived?” This is not necessarily an anti-phenomenalist skirmish, for Hylas is not advocating phenomenalism; more likely, he is suggesting that – for all Philonous has shown – matter could still exist, so long as it is necessarily perceivable. Philonous’ answer would then only show that ideas are all we really perceive when perceiving sensible objects, which is what phenomenalism says too.

The account described above would let Berkeley avoid contradicting his phenomenalist account of creation. But instead, he contradicts himself, saying that even objects must always be perceived. This introduces stricter, idealist criteria for the existence of objects than the phenomenalist criteria operative in his discussion of creation.

Objects as collections of ideas

The most serious hurdle for a phenomenalist interpretation is Berkeley’s repeated equations of objects with “collections” or “combinations” of ideas. These occur at P1/12/26/109/148 and 1D195; P1’s claim that “collections of ideas constitute a stone, a tree, a book, and the like sensible things” is typical.

Bennett argues that Berkeley rejected phenomenalism in favour of idealism because only idealism provided these direct substitutes for sensible objects: phenomenalists must offer complex reinterpretations of any sentences which refer to objects, whereas idealists can replace references to objects with references to collections of ideas. Bennett claims this would have put off anyone of Berkeley’s time. And indeed, such sentence-wide reinterpretations would have been unfamiliar then. Though Berkeley was happy with radical reinterpretations of everyday speech, urging readers to “think with the learned, and speak with the vulgar”, he does seem to assume he must equate objects with some part of his basic ontology.

Some confirmation for this comes from the fact that the early notes (PC41/52/98/282/290/293-293a/302) in which Berkeley came closest to phenomenalism do provide a direct substitute for objects: God’s powers to grant us appropriate sensations of those objects. PC293a is the example which best exhibits this account’s kinship with phenomenalism: “Bodies taken for Powers do exist wⁿ not perceiv’d but this existence is not actual. wⁿ I say a power exists no more is meant than that if in y^e light I open my eyes & look that way I shall see it i.e y^e body &c.”

Marc-Wogau, Barnes and Grayling provide evidence that Berkeley abandoned this account: 1) He marked these early notes with ‘+’, which Luce (1944:xxv-xxvi) controversially thinks signals cancellation. 2) Idealism dominates after PC280 (occurring explicitly or implicitly at PC301/347/377/408/427a/429/437/472-4a/477a/517-8/550/606/609/775/792/801-802/823). 3) PC802 resolves “Not to mention the Combinations of Powers but to say the things the effects themselves to really exist even wⁿ not actually perciev’d but still with relation to perception”. Not mentioning a theory is distinct from abandoning, but the published works clearly put a different theory in its place by identifying objects with collections of ideas (which are the “effects” of God’s powers). 4) The Principles and Dialogues consider objects directly perceptible by us; ideas can be, but divine powers cannot.

When Berkeley moved to an idealist account, he again provided a substitute for sensible objects. This time it was collections of ideas, not powers: P4, for example, asks “what are [sensible objects] but the things we perceive by sense, and what do we perceive by sense besides our own ideas or sensations?”

All this makes it very unlikely that Berkeley was a phenomenalist, but here is the best response I can imagine: “To render phenomenalism comprehensible, Berkeley had to ‘identify’ objects with something, so he chose the ideas which manifest the relevant subjunctive conditionals’ truth. P4 claims that objects are immediately perceptible, and only ideas are immediately perceptible for Berkeley. But in perceiving ideas, we perceive that those ideas are perceptible in certain situations, i.e. that some phenomenalist conditionals are true. This ‘propositional perception’ meets 1D174’s definition of immediacy: the conditionals are neither inferred from the idea nor mediately represented by it.”

Quite apart from its questionable account of immediate propositional perception, this response would be a stretch: it turns passages like P4 from clear expressions of idealism into opaque (and misleading) expressions of phenomenalism. It is better to accept their apparent meaning, and conclude that even if Berkeley sometimes had phenomenalist inclinations, he abandoned rather than disguised them in the larger part of the Principles and Dialogues.

How close Berkeley came to phenomenalism

We have now seen that three major elements of Berkeley’s philosophy can only be reconciled with phenomenalism by very strained readings of the texts. We should thus conclude that, outside his discussion of creation (and some portions of his notebooks), Berkeley was an idealist. But, lest this be obscured, I will finish by showing how close his idealism nonetheless was to phenomenalism.

The key point is that Berkeley always emphasised the role played by God’s intentions as to which sensations He would present in which circumstances. We identify “ideas of sense” as members of one object rather than another based on the patterns in which God presents them. And God presents ideas just as a phenomenalist’s conditionals would predict they’d be presented (i.e. according to “the set rules or established methods … [which] are called the Laws of Nature”), because He knows this will make us conceptually collect them together into ‘objects’ whose effects we can predict. At 2D245, Philonous explains how we classify collections of ideas as objects when they are “observed to have some connexion in Nature, either with respect to co-existence or succession”. (We have to pay attention to these connections between ideas in determining which objects they belong to, because ideas belonging to different objects can have indistinguishable intrinsic characteristics if the objects sometimes look the same. This is, I think, a powerful argument against overly-simple idealisms.) For Berkeley, objects are collections of ideas distinguished by the divinely ordained patterns in which they appear.

So Berkeley thinks that statements about objects are made true by God understanding the ideas to be perceived in various situations and granting us those ideas. Theistic phenomenalists think this too, if they require that God perceives what we do not because “He is no Blind agent”. Berkeley disagreed with such phenomenalists about what statements about objects mean, not about what ultimately grounds their truth. He thought that these statements were descriptions of our ideas, and perhaps God’s ideas. This is implausible: we don’t seem to be describing anyone’s actual ideas when we say a table exists in an unoccupied study (particularly if we are atheists). But while Berkeley might be better off understood as a phenomenalist, that was not what he advocated outside his discussion of creation.