Start Submission Become a Reviewer

Reading: Resemblance, Representation and Scepticism: The Metaphysical Role of Berkeley’s Likeness Pri...


A- A+
Alt. Display


Resemblance, Representation and Scepticism: The Metaphysical Role of Berkeley’s Likeness Principle


David Bartha

Humboldt University of Berlin, DE
X close


Berkeley’s likeness principle states that only an idea can be like an idea. In this paper, I argue that the principle should be read as a premise only in a metaphysical argument showing that matter cannot instantiate anything like the sensory properties we perceive. It goes against those interpretations that take it to serve also, if not primarily, an epistemological purpose, featuring in Berkeley’s alleged Representation Argument to the effect that we cannot reach beyond the veil of our ideas. First, in section 1, I raise some concerns about the traditional narrative concerning the likeness principle’s role in Berkeley’s argumentation. In section 2, I delineate an alternative narrative, arguing that there is no ‘missing premise’ in his alleged Representation Argument we need to explain simply because he advances no argument like that in the first place. In section 3–4, I provide a close reading of the relevant passages—first from the Principles, then the Dialogues—and their contexts, supporting textually a purely metaphysical interpretation of the likeness principle arguments. In section 5, I address some possible objections, based on the phrasing of the likeness principle passages and some related texts.

How to Cite: Bartha, D., 2022. Resemblance, Representation and Scepticism: The Metaphysical Role of Berkeley’s Likeness Principle. Journal of Modern Philosophy, 4(1), p.1. DOI:
  Published on 25 Jan 2022
 Accepted on 01 Nov 2021            Submitted on 21 May 2021

Berkeley’s famous, or perhaps infamous, likeness principle states that only an idea can be like an idea. In a Notebooks entry, Principles sections 8 and 9 as well as at the end of the first of his Three Dialogues, he relies on this principle in arguing against representative realists like Descartes and Locke, according to whom although we do not perceive the material objects themselves, these nonetheless instantiate sensory properties like the ideas we do perceive. Berkeley argues that since an idea cannot be like a material quality but only another idea, material substances cannot have sensory properties similar to the ideas we perceive.

Historians of philosophy often emphasize how much hangs on the likeness principle in Berkeley’s argumentation against representative realism, but they tend to be equally interested in the extent it both relies on and influences early modern debates on resemblance and representation. Nonetheless we are surprisingly far from reaching a consensus about the reason why Berkeley thought that an idea can resemble only another idea. According to Cummins’s (1966) classic interpretation, resemblance can only obtain between things that share at least one property of the same ontological kind. But since the sensible qualities we perceive are mind-dependent, they cannot have any property of the kind an unperceivable quality of a mind-independent object has. Following Winkler (1989: 145–49), many assume that Berkeley’s justification relies on the incomparability of our ideas and imperceptible material qualities, and perhaps that he was a verificationist who concluded from this consideration not only that the resemblance between ideas and material qualities cannot be discovered, but that we cannot even assert meaningfully that it exists in reality. Ryan (2006) proposes that for Berkeley, just like for Locke, resemblance is a mind-made relation, and since the mind cannot access both the relata, the resemblance relation itself cannot obtain between our ideas and the unperceived material qualities. More recently, West (2021) argues that Berkeley’s fundamental commitment to the transparency of our ideas excludes the possibility that an idea be like a non-idea, since it would require, after all, that our ideas have some properties we cannot perceive.

In what follows, I will not be engaging directly with the vexed question of why Berkeley thought the likeness principle is true. Rather, I am more interested in the purpose he used this principle for in the relevant passages as well as its role in his wider argumentative strategy. In contrast to the controversial why question, commentators generally agree that the likeness principle is deployed by Berkeley, like in the sceptic Foucher’s hands, as a premise in an epistemological argument against the feasibility of the representative relation supposedly connecting our ideas with the mind-independent reality. My aim is to challenge this all too widespread, if not standard, interpretation. On the proposed purely metaphysical reading of the relevant passages, the only actual conclusion Berkeley draws is that material objects cannot instantiate sensory qualities, that is, qualities like the ones we perceive by our senses. While my reading naturally complements Cummins’s ontological interpretation, even if he were to rely on some epistemological (such as Winkler’s verificationist) ‘bridging’ principle to justify the likeness principle, Berkeley reaches this metaphysical conclusion about the representative realists’ sensory understanding of matter without asserting or even suggesting any additional epistemological point about our incapability of representing mind-independent things.

On the proposed reading, the likeness principle argument fits seamlessly into Berkeley’s wider argumentative strategy, showing that the concept of matter is either inconsistent—when assuming that it has features even just similar to, let alone identical with, our perceptions—or intrinsically empty, indeed so thin that it is on the verge of losing all its content and meaning, as Berkeley argues subsequently. Moreover, if the materialist insists that our ideas represent something out there without revealing its intrinsic nature, while voicing no qualms about the possibility of this sort of extrinsic representative relation, Berkeley is keen to point out that it cannot avoid a miserable sort of scepticism. While it is not the epistemological sort of scepticism Berkeley, on the traditional reading, not only discovers in his opponents’ views but in fact embraces as an important step to his own idealist conclusion, it nonetheless forces the representative realists to acknowledge that there are no sensible objects. The material objects a realist might still assume to exist independently of our perceptions are neither perceivable nor can have anything like the sensible properties we perceive them having. The representative realists might make—in Berkeley’s opinion unsuccessful—last-ditched attempts to save something of our concept of matter, it is undoubtedly a huge blow to its appeal to our common-sense intuitions. As such, this narrative also allows us the see more clearly what sort of scepticism Berkeley did and did not target in his opponents’ views while seeking to show that, against all odds, immaterialism is the most faithful to common sense.

Furthermore, since, on my reading, the conclusion Berkeley draws from the likeness principle is restricted to the intrinsic nature of matter, a significant amount of pressure is lifted off him insofar as he does not need to justify the principle’s relevance to our minds’ capacity for being about things distinct from them or, more simply, to intentionality. As I will argue, it has nothing more to do with representation than assuming, at most, that the content of a representation cannot resemble the material quality it is supposed to represent without the manifest absurdity of the vehicle of the mental representation also being similar to the material quality in terms of its nature. But nowhere does he conclude, on this or any other basis, that a non-resembling idea cannot work, even if only in an extrinsic or purely formal way, as a representation of a mind-independent quality.

In section 1, I recap and raise some concerns about the traditional narrative concerning the role the likeness principle fulfils in Berkeley’s argumentation. In section 2, I delineate an alternative narrative into which the likeness principle fits much more seamlessly. In sections 3–4, I provide a close reading of the relevant passages—first from the Principles, then the Dialogues—textually supporting a purely metaphysical interpretation of the argument in which the likeness principle features as a crucial premise. In section 5, I address some possible objections based on the phrasing of the likeness principle passages and some related texts that commentators might cite to support the—in my view, mistaken—interpretation that Berkeley had an epistemological aim in these passages.

I. The Traditional Narrative, and the Mystery of the Missing Premise

The likeness principle is generally regarded as one of Berkeley’s most important philosophical convictions, serving as a crucial premise in his pivotal argument against representative realism. Indeed, the likeness principle argument has long been depicted as the final blow Berkeley delivered to Locke and Descartes’s position. Cummins, for instance, calls it his ‘fundamental objection’ to representationalism, and hence it is to be seen as a ‘fundamental element of his immaterialism’ (1966: 63–64). In a footnote, he cites Turbayne’s even more enthusiastic evaluation: ‘the argument against representationalism is the strongest weapon Berkeley has for attacking belief in material substances and indicates the importance of the Likeness Principle’ (Turbayne 1954: xii-xviii).1 As the traditional account has it, after attacking direct realism, that is, the view that we perceive the very qualities of material objects without any perceptual intermediary, Berkeley turns to the representative or indirect realist and tries to undermine the very possibility of our ideas resembling and, hence, representing anything beyond our minds.

Allaire (1982), Winkler (1989: 138–39), Dicker (1985, 2011: chap. 7), Bolton (1987, 2008), Dancy (1998), Jacovides (2009), Downing (2011), Hill (2011), Fasko and West (2020), and West (2021) are just a few other examples of those numerous commentators who seem to endorse this standard reading of the relevant passages. According to this interpretation, Berkeley used the likeness principle to prove two things: (i.) the metaphysical thesis that sensory qualities like our ideas cannot exist in material objects, and (ii.) the epistemological problem that seeks to undermine any theory holding that our ideas represent things outside our minds. Indeed, most commentators, like Hill (2011), focus almost exclusively on the latter, aptly called by him ‘Representation Argument’, and take the former, at best, as one of its premises.2 There is, of course, a crucial premise missing in this argument, so commentators point to Berkeley’s implicit conviction that representation requires resemblance and since there cannot be (or, at least, cannot be discovered) any similarity between our ideas and their material archetypes, it follows that our ideas cannot represent anything outside them. This way we end up with a rather worrying sort of skepticism, rendering us unable to reach out to the world from behind the veil of our perceptions. As the traditional story goes, Berkeley’s paradoxical solution to this miserable state is to full-heartedly embrace the position that we know only our ideas while giving up the assumption that there should be anything behind them to be represented in the first place. Hence, he turned the skeptical view into an idealist identification of the veil of perception with reality itself. However, there is something suspect not only with the broad-strokes of this general picture—concerns about Berkeley’s alleged endorsement of external world scepticism, for instance, I can hardly touch on here—but also with how this narrative construes Berkeley’s wider argumentative strategy and reads the likeness principle argument specifically. I will be asking, firstly, why Berkeley argued for an unnecessarily strong metaphysical conclusion if his main interest lies in the epistemological point he allegedly makes; secondly, why he still felt he need to reject a wide variety of concepts of matter proposed by representative realists; and, finally, why he did not even mention the notorious missing premise.

First, if the epistemological point of (ii.) is really the ultimate purpose of Berkeley’s argument, one might wonder why he also argued for the much stronger metaphysical conclusion of (i.), showing that matter cannot have properties similar to our ideas. It doesn’t seem to be a necessary step or premise in establishing (ii.), as he could have taken a much easier route and gone with a merely epistemological reading of the likeness principle, arguing that since our ideas cannot reveal to us any resemblance with things we cannot perceive, even if there are material objects with similar sensory qualities, our perceptions cannot represent them to us at all.3 One perhaps might insist that, according to the theory Berkeley attacks, X represents Y if and only if X resembles Y as a matter of fact, that is, regardless of whether we can ever discover this resemblance. And in this case, he indeed needs the stronger, metaphysical implication of the likeness principle, that is, (i.), to conclude (ii.). But if Berkeley had this ‘matter-of-fact’ or externalist reading in mind, interpreters surely cannot appeal either to the incomparability of our ideas and material qualities or the mind-dependence of relations as his implicit justification for the likeness principle. If he allowed that X might resemble, and hence represent, Y without us ever being able to tell or establish it, and, accordingly, realized that first he needs to show that this resemblance is in fact impossible between an idea and a material quality, it would be manifestly inconsistent to argue for this on the grounds that we cannot perceive both to compare them and discover (or establish) their similarity. Moreover, it does not seem to be the most plausible or straightforward understanding of the resemblance theory, as it takes away one of its appealing philosophical merits, namely, that it promises to account for representation in terms of internally accessible features: X represents Y through the similarity relation we can discover between X and Y. In fact, it would arguably make it even worse than other externalist theories in this respect, since while the causal relation those tend to appeal to can be established at least indirectly, the similarity cannot be discovered even in principle. Also, no reasonable theory of representation should allow that simply the similarity of X and Y is sufficient for X to represent Y (but not for Y to represent X). Famously, it leads to many of the problems associated with the resemblance theory of representation, for instance, that of reflexivity, symmetry and, perhaps, transitivity and, more generally, to the problem that far more things resemble one another than how many, we would normally think, represent other things. To circumvent some of these problems, it seems more promising and natural to say that our minds need to be aware of the (relevant sort of) resemblance to establish the representation relation. But this obviously leaves us in great difficulty, again, which, on the traditional reading, Berkeley wants to exploit in arguing against representative realism. In this case, however, it remains unexplained why Berkeley argued for the metaphysical impossibility of material qualities resembling our ideas, when the much less controversial epistemic reading of the likeness principle could have done the job perfectly.

Second, if we already have such a knockdown argument against the possibility of representing anything outside the narrow boundaries of our minds, and this horrendous scepticism can be defeated only by adopting idealism, it is equally unclear why Berkeley went on to reject so keenly a whole bunch of different concepts of matter. Whatever metaphysical, such as causal, argument representative realists might come up with to support their conception of matter, or however they might spell it out, Berkeley, on the traditional narrative, has already established its irrepresentability with the help of the likeness principle. Accordingly, no theory of matter will ever hold out any hope to be successful, invariably leading to external world scepticism and, hence, its only viable antidote, idealism. As such, all the subsequent discussion should be of secondary or merely hypothetical importance for Berkeley, presupposing the per impossibile antecedent ‘even if matter were representable …’. But, as it seems to me, this is not how Berkeley sees the dialectic of his argumentation at all, putting no less weight on the subsequent criticisms of those concepts of matter that are still very much in the running. One might think, as Cummins argues, that Berkeley wanted to prove that matter is not merely irrepresentable but impossible, motivating his metaphysical attack after the epistemological point pressed in the likeness principle passages. But, as we will see, his concern to prove its impossibility is at least as central to these early sections of the Principles or the First Dialogue—where he underlines the incoherence of matter supposedly instantiating sensory qualities—as it is in the subsequent considerations that actually end up with dismissing an empty concept of matter—a concept which, though (or rather due to) literally meaning ‘nothing’ to us, we cannot rule out as strictly impossible. Relatedly, if all representative realists are refuted by the Representation Argument, who is Berkeley arguing so ardently against in all these passages, for instance in the whole Second Dialogue? Of Berkeley’s predecessors and contemporaries, who actually believed that there are material objects, but that we cannot represent them in some way or another? Are the occasionalists or the scholastics—who seem to be the targets of some of the subsequent arguments—not also committed to a broadly representationalist paradigm, that is, to the view that the ideas, species or whatever it is through which we perceive the world are representations of external objects? If so, while some medieval philosophers can be classified as straightforward direct realists, the dominant scholastic view concerning the resemblance or, in fact, formal identity of the object and the (sensory) species as well as occasionalists like Malebranche seem to be just as appropriate targets of the epistemological attack commentators discover in the likeness principle passages as Descartes or Locke ever was.4

Third, in addition to these perhaps excusable discrepancies the traditional narrative causes in our understanding of what Berkeley is doing exactly in his wider argumentation, an even more curious thing about this standard story is the ‘mystery’ surrounding the Representation Argument’s missing premise. If it had such a crucial role in his argumentation for idealism, one cannot but wonder why Berkeley would not argue for the resemblance theory of representation or at the very least make his or his opponents’ commitment to it explicit—as did Foucher, for instance, in his Critique de la Recherche de la Verité (45–46, 51–52). To explain why he did not bother to mention it in any of his published works, let alone, in the context of the likeness principle, many like Jacovides (2009) suggest that Berkeley simply presumed—and justifiably so—that Locke accepted it, while others like Hill (2011) argue that the resemblance requirement of representation was so widely accepted among the Cartesians and their critics that he might have easily considered it in need of no explicit statement. Others (Fasko and West 2020) pointed to Berkeley’s intellectual context in Ireland with a similar conclusion—providing us with another piece of circumstantial evidence. On a similarly popular reading, however, the missing premise is actually a misapprehension on Berkeley’s part. The representative realists he is arguing against, that is Descartes and Locke, while indeed thinking that the primary qualities resemble our ideas of them, did not think that they represent their objects through this resemblance. Nor is it obvious, to say the least, that the non-resembling ideas of secondary qualities are said not to represent their causes. In fact, Locke, for instance, did not even claim that, due to the lack of resemblance, they do not represent them just as adequately or perfectly as the idea of primary qualities represent their objects. Accordingly, Bolton (2008: 76–87) argues that Berkeley begs the question against Locke when requiring resemblance for representation. And even Hill (2011: 76–77) acknowledges that Locke would not buy into this premise—which might set alarm bells ringing given that, as important as the Cartesian context is to Berkeley, Locke is surely one of the targets of this argument.5 So, on the traditional narrative, it is highly dubious if the alleged Representation Argument can take off and be raised against its targets, the actual proponents of the representative theory of perception. And even if Berkeley’s opponents were committed to the resemblance theory of representation or even if he was at least justified to read them this way, it is nonetheless a weakness of the argument that it clearly leaves the theoretical possibility open for adopting other theories of representation, for instance a causal or covariance theory.

II. The Purely Metaphysical Reading of the Likeness Principle Argument

All these puzzles haunting the traditional narrative seem to dissolve once we take care not to read more into Berkeley’s likeness principle passages than the argument we actually find there. As I try to prove textually in what follows, even if it was a verificationist or some other epistemological background assumption that made him endorse the likeness principle, the point Berkeley wants to establish with its help is exclusively the metaphysical impossibility of material objects’ instantiating sensory qualities or qualities like those our ideas instantiate, for instance—citing Berkeley’s own examples—the solidity, motion or extension we perceive in physical objects. On this alternative narrative, it is clear why, first, he argued for the strong metaphysical conclusion concerning the sensible concept of matter representative realists such as Locke believe in, second, went on to criticizing other—more abstract—concepts of matter with equal tenacity, and, third, did not mention a premise that is not missing from his argument, after all.

Moreover, reading its role in a purely metaphysical way, the likeness principle fits much more seamlessly in Berkeley’s whole argumentative strategy than the supposed Representation Argument, since it can be seen as an extension of his earlier arguments to the effect that the intrinsic nature of the directly perceived sensory qualities are essentially mental or mind-dependent, and hence cannot be instantiated by material objects. This is exactly the two-fold plan he set out in the Notebooks, namely ‘to Demonstrate the Principle i.e that neither our Ideas nor any thing like our ideas can possibly be in an unperceiving thing’ (Notebooks 379) with arguments that ‘must be proposd shorter & more separate in the Treatise’ (Notebooks 378a).6 One of these arguments is to prove that ‘None of our ideas can be in a thing wch is both thoughtless & senseless’ (no. 9. in the formulation from Notebooks 377–78); the other (no. 15) is that ‘nothing like an idea can be in an unperceiving thing’. As both the early sections of Principles and its forerunner in Notebooks 377–78 claim, the objects or qualities we directly perceive are nothing else but ideas, and are like sensations or thoughts (or ideas of imagination and passions, as he adds in Principles 3) insofar as their esse is just their percipi. And just as it is contradictory to say that our very ideas exist in unperceiving material objects, the likeness principle passages aim to show that it is equally incoherent to hold that qualitatively identical or even just similar qualities exist in them. While, as he says, there is ‘another demonstration of the same thing’ presented in Notebooks 378 which relies on the incomparability of ideas and non-ideas, the conclusion is the same as the one he draws based on the impossibility of property-sharing between matter and mind: if they are indeed similar to our ideas, the intrinsic nature of the sensory qualities supposedly existing materially should also be essentially mental or mind-dependent—which is an obviously incoherent thought.7 In fact, we might add that on the view targeted in the likeness principle argument, the supposed bearer of the essentially mind-dependent qualities is not only unperceiving (‘both thoughtless & senseless’, as Berkeley puts it in Notebooks 378) but unperceivable as well (see Principles 68, for instance, for pushing this latter point). So, while the unperceiving matter cannot uphold them, our minds cannot ‘support’ them through a perceptual relation either.

Accordingly, and in line with the whole argumentative strategy of the Principles as well as the Dialogues, the likeness principle argument is not to exclude the possibility of representing material objects. While matter might exist, and we might even represent its unknown qualities in some relative or extrinsic manner, the conclusion established at this point is only that it cannot instantiate properties like the sensory objects we perceive. Indeed, the other arguments in the vicinity both in the Principles and the Dialogues aim to show something similar: due to the relativity and varying nature of the properties we directly perceive, perception cannot reveal to us how they are supposed to exist on their own with their supposedly stable and absolute (‘settled and determinate’, as he puts it in section 14) intrinsic nature. The likeness principle is, of course, much stronger than these arguments borrowed from the skeptics, which, as Berkeley acknowledges in Principles 15, prove only that, in the absence of having any criterion for identifying one of our various perceptions as a true representation of how the material qualities exist in themselves, ‘we do not know by sense which is the true extension or colour of the object’.8 The basic argumentation culminating in the likeness principle, by contrast, proves that they cannot have any of, indeed, anything like those properties we perceive. Referring to this basic point, he says:

But the arguments foregoing plainly shew it to be impossible that any colour or extension at all, or other sensible quality whatsoever, should exist in an unthinking subject without the mind, or in truth, that there should be any such thing as an outward object. (Principles 15)

Or when introducing his infamous master-argument:

This easy trial may make you see, that what you contend for, is a downright contradiction. Insomuch that I am content to put the whole upon this issue; if you can but conceive it possible for one extended moveable substance, or in general, for any one idea or any thing like an idea, to exist otherwise than in a mind perceiving it, I shall readily give up the cause. (Principles 22, emphasis added)

On the present reading, the likeness principle argument indeed has a fundamental role and functions as a pivot in Berkeley’s dialectic. But, in contrast to the traditional narrative, it does not separate his supposedly epistemological attack on direct and representative theories of perception from a subsequent tirade against the metaphysical problems besetting the concept of matter. In fact, the difference between the likeness principle argument and the subsequent considerations Berkeley levels against representationalists like Locke and Descartes just as much as against the occasionalists and the scholastics is merely that their target is a different concept of matter. Specifically, up to the likeness principle argument, Berkeley attacks the concept of material sensory qualities (or matter identified with these qualities), before turning on the hypothesis of various forms of material substances or substrata bearing or supporting those, which—construing matter as an essentially non-sensible logical construct—is immune to the metaphysical attack on the alleged likeness between our ideas and the material sensory qualities. But, adding to its pivotal role in Berkeley’s argumentation, the likeness principle argument also connects the two main types of arguments he appeals to in attacking these different concepts of matter. One line is that, on the sensory readings, it involves a contradiction, culminating in the likeness principle as a proof of the incoherence of the view that matter instantiates properties even just similar to our ideas. In the passages just quoted from the Principles, Berkeley hammered away at this line, making the likeness principle’s metaphysical role very clear. But it also foreshadows the second sort of argument Berkeley employs against matter, serving as the first warning that taking the alternative route of endorsing a non-sensory concept of matter does not hold much promise either.

As Philonous is keen to point out after discussing the likeness principle at the end of the First Dialogue, the representative realist cannot help but to end up advocating a worrying sort of scepticism. The position Hylas is forced to take on from the beginning of Second Dialogue onwards is that there are no sensible objects, as the only material objects that might still exist are not only unperceivable, and hence we have no perceptual access to their nature, but are indeed nothing like our sensory perceptions. Berkeley would not miss an opportunity to underscore that if we are willing to bite this bullet with the materialist, we will lose all our hope to grasp or, if you like, represent the intrinsic nature of the mind-independent world. We might look for some relative, purely logical and abstract concept of matter, for instance that of an imperceptible support or substratum of sensible qualities (see Principles 16, immediately following a reference to the likeness principle), but it is definitely not what the common sense regards as ‘outward object’ based on what we perceive by our senses. Moreover, Berkeley seeks to show that not even philosophers can form any meaningful notion of either of the terms making up the term ‘material substance’, insofar as ‘there is no distinct meaning annexed to them’ (Principles 17). For various reasons, it will not help to treat matter merely as the mind-dependent power to cause our perceptions either. As he put the point in his late Theory of Vision Vindicated:

Whenever, therefore, the appellation of sensible object is used in a determined intelligible sense, it is not applied to signify the absolutely existing outward cause or power, but the ideas themselves produced thereby. (Theory of Vision Vindicated 12)

As Berkeley gradually points out in the parts following the likeness principle passages in the Principles and the (second of the) Dialogues, if we go down the route of supposing non-sensible material substances, we will not only realize its inevitable sceptical implications, flying in the face of common sense. But being pushed to construe it more and more abstractly—as, for instance, a mere instrument or occasion for God to create our ideas or an ‘unknown something’, as characterless and formless as the scholastics’ materia prima—we will eventually peel so much content of it that we will end up with a vague and empty concept that means nothing to us at all.

So, on the proposed minimalistic reading of the likeness principle argument, Berkeley’s explicit aim, as well as the conclusion he draws, has nothing to do with undermining the realist concept of intentionality, and hence he is not committed to endorsing, or even just attributing to his opponents, any theory of representation at all. As a consequence, we do not need to assume that there is a missing premise in one of his most crucial arguments for immaterialism, and, a fortiori, to look for a reason why he did not even bother to mention it, let alone argue for it, in any of his published works. Accordingly, on this simpler and more charitable reading, there is no ‘apparent gap’ in Berkeley’s whole argumentation against materialism or, more precisely, representative realism, overlooking the possibility that matter might be represented as the non-resembling cause of our perceptions, nor did he assume rather dubiously that by rejecting the resemblance theory he rejected the possibility of representing mind-independent objects altogether—just as he did not, contra Bolton’s reading, assume a probably mistaken reading of Locke’s (and Descartes’s) theory of representation. Indeed, it seems that commentators like Hill create the ‘apparent gap’ in Berkeley’s argumentative strategy by assuming a ‘missing premise’ in an argument Berkeley never actually formulated.

III. The Likeness Principle Argument in the Principles

To support this reading further textually, in what follows I will look more closely at what Berkeley is actually doing in the relevant passages and texts surrounding or referring back to them. Let us start with the locus classicus.

But say you, though the ideas themselves do not exist without the mind, yet there may be things like them whereof they are copies or resemblances, which things exist without the mind, in an unthinking substance. I answer, an idea can be like nothing but an idea; a colour or figure can be like nothing but another colour or figure. If we look but ever so little into our thoughts, we shall find it impossible for us to conceive a likeness except only between our ideas. Again, I ask whether those supposed originals or external things, of which our ideas are the pictures or representations, be themselves perceivable or no? If they are, then they are ideas, and we have gained our point; but if you say they are not, I appeal to any one whether it be sense, to assert a colour is like something which is invisible; hard or soft, like something which is intangible; and so of the rest. (Principles 8)

According to Hill’s (2011: 49) formulation, in these lines Berkeley proposed the following Representation Argument.

  1. If X is to represent Y, X must resemble Y.
  2. Ideas cannot resemble non-ideal things.
  3. Therefore, ideas cannot represent non-ideal things.

Does the text justify this reading? There is no question about premise 2, which is the likeness principle itself. Premise 1 is the notorious ‘missing premise’, so it is acknowledged to be lurking only between the lines. But not even 3 does justice to the conclusion Berkeley draws. Rather than being presented in this epistemological form concerning representation, it seems to be that material objects cannot have sensory qualities (i.e., qualities like our ideas), or, in the words Berkeley used to introduce the criticized view, it is impossible that ‘there may be things like them [i.e., our ideas] whereof they are copies or resemblances’.9 It is worth quickly quoting how Berkeley summarizes the argument immediately in the following section (remember a similar statement from Principles 15 and 22, quoted above):

[…] But it is evident from what we have already shewn, that extension, figure and motion are only ideas existing in the mind, and that an idea can be like nothing but another idea, and that consequently neither they nor their archetypes can exist in an unperceiving substance. Hence it is plain, that the very notion of what is called matter or corporeal substance, involves a contradiction in it. (Principles 9)

Accordingly, we should formulate the basic argument of Principles 8–9 (and the likeness principle argument generally) as follows.

  1. If the representative realist position under discussion is true, all (or, as discussed in the unquoted part of section 9, some) qualities of material objects are not ideas themselves but like our ideas.
  2. Only ideas can be like our ideas.
  3. Therefore, since qualities of material objects cannot be like our ideas; this representative realist position is false.

The argument concludes the impossibility of the representative realist view simply by applying to matter the metaphysical lesson of the likeness principle. In Principles 8, after the term ‘again’, Berkeley complements this basic argument with a dilemma, providing us with a fuller insight into his thinking:10

  1. If the representative realist position under discussion is true, all (or, as discussed in the unquoted part of section 9, some) qualities of material objects are not ideas themselves but like our ideas.
  2. The qualities of material objects are either perceivable or unperceivable.
  3. Anything that is perceivable is an idea.
  4. An idea cannot be a quality of material objects.
  5. Only a perceivable thing can be like our ideas.
  6. If the qualities of material objects are perceivable, they are ideas. (from 3)
  7. If the qualities of material objects are perceivable, they cannot be qualities of material objects. (from 4, 6)
  8. If the qualities of material objects are unperceivable, they cannot be like our ideas. (from 5)
  9. Therefore, contrary to this representative realist position, qualities of material objects cannot be like our ideas. (from 2, 7, 8)

Premise 1 is just stating the representative realist’s view under consideration—or more precisely the metaphysical picture underlying it. Premise 2 is the dilemma Berkeley proposes to show it cannot be maintained. From 6 to 9, it is a logically valid deduction. So, the work is done by premises 3, 4, and 5. Though premise 3—the thesis that whatever is perceived or even just perceivable is an essentially mind-dependent idea—is not made explicit here, it is a crucial consideration Berkeley thinks he has established earlier. As we will see below, he made it explicit in the Dialogues version, as he did early in his Notebooks (entry 50): ‘Nothing but ideas perceivable.’ We might also take premise 3 as an assumption shared by his opponents as well as Berkeley.11 Accordingly, premise 4—the view that matter cannot instantiate our (essentially mind-dependent) ideas—seems to be uncontroversial in this context. So, it all depends on premise 5, which looks like a version of the likeness principle. Though we do not need it for the argument, from 3 and 5, we can derive the standard version of it, namely that only ideas can be like our ideas. This shows that the likeness principle is a composite claim with a central intuition behind it: if something is to be similar to our ideas of sensory qualities, it has to share at least one property with our essentially perceivable idea and hence be perceivable itself. And, according to Berkeley’s esse is percipi principle, what is perceivable, that is, what is a possible object of perception, cannot exist without a mind, hence is an (essentially mind-dependent) idea.

One might say that my reconstruction ignores a crucial element in the argument, namely that Berkeley talks about colours and figures specifically, not merely the perceivability of such qualities. It suggests that his argument concerns the content of representation, pointing towards an epistemological use of the likeness principle, after all. But I believe both of these steps can be questioned. First, on my reading, when speaking about colours and figures in this passage Berkeley intends nothing else than that for two things to share some property, they need not only to be equally sensible or perceivable in general but to be perceivable through some particular modality, say, equally visible. In line with his theory of vision, maintaining the heterogeneity between the natures of visible and tangible objects, his likeness principle thus holds more specifically that in order for two things to be alike, they have to be perceivable through the same modality. But his fundamental point nonetheless concerns the perceivability of these supposed qualities, and that the representative realist description of matter is incoherent insofar as it construes it as a mind-independent thing instantiating sensory qualities while not being perceivable through any modality.12

Secondly, and much more importantly, be it as it may, it does not affect my main point. Indeed, it might very well be that the likeness principle argument concerns the content of representation as well as the nature of its vehicle. Berkeley’s implicit argument then is the following: the content of the representation cannot be similar to the quality it is supposed to represent if the vehicle of the representation is not similar to the quality in terms of its nature, too. Hence, if they are to be similar to the content of our representations, as the representative realists of this sort claim, material objects should ultimately instantiate ideas, leaving us with the same absurdity Berkeley aims to point out. Even on this reconstruction, then, his attack has nothing to do with the irrepresentability of the material qualities the content of our ideas might nonetheless represent somehow, but merely that there cannot be a material quality that is similar to the content and hence to the vehicle of our representations.

Of course, it needs a bit more work to convince the representative realists, and we might doubt that Berkeley had an argument or even the resources to do so without begging the question. In any event, what I want to emphasize now is not the soundness of the argument but that both the basic and the complemented argument conclude that matter cannot have sensory qualities like our ideas, building on and extending his most fundamental arguments for the mind-dependence of the perceivable objects and their qualities. Rather than having an epistemological conclusion concerning the irrepresentabilty of the external world, as we have seen, Berkeley ends section 9 by pointing only to the metaphysical incoherence of the concept of matter, more precisely in its sensible or sensory incarnation. To quote again, ‘neither they [i.e., our ideas] nor their archetypes can exist in an unperceiving substance. Hence it is plain, that the very notion of what is called matter or corporeal substance, involves a contradiction in it.’ As he promised in the Notebooks, this is indeed the reproduction of his basic principle presented there as follows: ‘neither our Ideas nor any thing like our ideas can possibly be in an unperceiving thing’ (Notebooks 379). Even much later in the Principles, when referring back to his main argumentation, he labels the representative realist position on matter just as absurd as the ‘vulgar’ direct realist one:

[…] philosophers having plainly seen, that the immediate objects of perception do not exist without the mind, they in some degree corrected the mistake of the vulgar, but at the same time run into another which seems no less absurd, to wit, that there are certain objects really existing without the mind, or having a subsistence distinct from being perceived, of which our ideas are only images or resemblances, imprinted by those objects on the mind. (Principles 56)

As Berkeley continues, the same sort of metaphysical impossibility (or, in his terminology, repugnancy) is to be found in the view that matter, which is nothing else but an inert, passive entity, causes our ideas as is to be found in the view that material objects have sensory qualities even just similar to our ideas:

But why they should suppose the ideas of sense to be excited in us by things in their likeness, and not rather have recourse to spirit which alone can act, may be accounted for, first, because they were not aware of the repugnancy there is, as well in supposing things like unto our ideas existing without, as in attributing to them power or activity. (Principles 57, emphasis added)

As I mentioned, on this purely metaphysical reconstruction of the role and aim of the likeness principle argument, it naturally paves the way for the following discussion concerning the relativity of our perceptions in the Principles, hammering away at the significantly weaker point that material objects cannot actually possess sensory qualities identical with or like all of those we perceive them to have. Similarly, before attacking the tenability of the distinction itself, in section 9 of the Principles, Berkeley argues that differentiating resembling primary qualities from non-resembling secondary qualities will not help, as the likeness principle argument applies to the primary qualities just as much as the secondary ones, excluding the possibility of matter instantiating anything similar to the sensory qualities of our ideas.

IV. The Likeness Principle Argument in the Dialogues

In the Dialogues, the context makes the metaphysical aim and significance of the likeness principle argument perhaps even clearer. At the end of the First Dialogue, Berkeley proposes three arguments in quick succession against the Lockean sort of representative realism: an epistemological argument, one based on the relativity and changing nature of our perceptions of physical objects, and one relying on the likeness principle. In the reverse order, these all appeared in the Principles, between sections 8 and 21. After accepting the conclusion of the epistemological argument, namely that neither perception nor reason can prove that material sensible objects exist, Hylas attempts to hold his ground by asserting that it is still possible that material objects exist and instantiate sensory properties like our ideas. The relativity and, especially, the likeness argument attack this very possibility.

We are already familiar with the relativity argument. It argues that since our ideas of sensory qualities are changing and constantly fleeting relative to perspective, situation, perceiver etc., and, at the same time, the qualities of material objects assumed to be stable and determinate, we can conclude that material objects do not have sensory qualities like our ideas. As he did in the Principles, so too here he realized the loophole in this argument, and, hence, added the problem of criterion. At this point, Berkeley is falling back to a skeptical epistemological point: we have no independent criterion to determine, hence any justifiable reason to say which, if any, of our ideas’ changing or constantly fleeting qualities is showing faithfully how matter is in itself. Of course, as the original metaphysical point intended, the relativity argument at least shows that material objects cannot instantiate all of those sensory qualities we represent them as having. But Berkeley clearly wants more than this, and after enthusiastically exclaiming that ‘But neither is this all’, he introduces his strongest, all-inclusive, argument against the ontological picture underlying this representative realist position. He straightaway presents the likeness principle argument in the form of the dilemma we saw in the Principles (what I called the complemented argument).

PHILONOUS. But neither is this all. Which are material objects in themselves, perceptible or imperceptible?

HYLAS. Properly and immediately nothing can be perceived but ideas. All material things therefore are in themselves insensible, and to be perceived only by their ideas.

PHILONOUS. Ideas then are sensible, and their archetypes or originals insensible.

HYLAS. Right.

PHILONOUS. But how can that which is sensible be like that which is insensible? Can a real thing in itself invisible be like a colour; or a real thing which is not audible, be like a sound? In a word, can any thing be like a sensation or idea, but another sensation or idea?

HYLAS. I must own, I think not.

PHILONOUS. Is it possible there should be any doubt in the point? Do you not perfectly know your own ideas?

HYLAS. I know them perfectly; since what I do not perceive or know, can be no part of my idea.

PHILONOUS. Consider therefore, and examine them, and then tell me if there be any thing in them which can exist without the mind: or if you can conceive any thing like them existing without the mind.

HYLAS. Upon inquiry, I find it is impossible for me to conceive or understand how any thing but an idea can be like an idea. And it is most evident, that no idea can exist without the mind.

PHILONOUS. You are therefore by your principles forced to deny the reality of sensible things, since you made it to consist in an absolute existence exterior to the mind. That is to say, you are a downright sceptic. So I have gained my point, which was to shew your principles led to scepticism. (Three Dialogues 206)

While this exchange follows its archetype from the Principles quite closely (as well as their common ancestor from the Notebooks), it displays some notable features. First, it suggests that the dilemma he proposed here as well as in the Principles is not a different consideration than, or the justification for, the basic argument, but rather it is the fully formed version of the point Berkeley wants to make. Asserting that ‘Properly and immediately nothing can be perceived but ideas’, this version also makes explicit Berkeley’s commitment to premise 3 (see above) he did not bother to put forward in the Principles version. He also put emphasis on this thesis in the Caesar passage at the beginning of his three-pronged attack on the Lockean sort of representative realist position, proclaiming ‘How! is there any thing perceived by sense, which is not immediately perceived’ (Three Dialogues 203), that is, an idea according to both Berkeley and the representative realists. More importantly, in line with the whole setup of the three-pronged argumentation, it makes even more conspicuous and perspicuous that Berkeley regarded the likeness principle argument as an attack on the metaphysical possibility of the concept of matter that underlies this sort of representative realism. Just like in the Principles, here too the likeness principle argument is presented as a metaphysical claim, concluding with the denial of ‘the reality of sensible things’ construed materially—a conclusion that has nothing to do with the supposed epistemological argument that we cannot in principle represent anything outside our ideas. Moreover, it explicitly connects it to his basic intuition of the mind-dependence of sensible objects, when indicating that the likeness principle’s introspectively self-evident nature is similar to realizing the impossibility of ideas existing unperceived. Appealing to the transparency of our perceptions, Berkeley emphasizes again that it is no less inconceivable that qualities similar to our ideas exist in matter than that our very ideas do. When Hylas concedes that it is impossible to conceive of either of those absurdities, Berkeley should not be understood to be simply appealing to the limits of our intellectual or conceptual abilities, but with his usual rhetoric he effectively claims, just as he did on various occasions in the Notebooks or the Principles, that it is inconceivable simply because it is a plain contradiction or repugnancy.13

The broader strategic or ‘pivotal’ role of the likeness principle argument is also easier to detect in the Dialogues, since it serves as the finale of the First Dialogue, with Hylas being forced to take the ‘downright sceptic’ position of ‘deny[ing] the reality of sensible things’ (Three Dialogues 206). According to the definition Hylas and Philonous settle on at the outset of that dialogue, a sceptic is one who is ‘distrusting the senses, […] denying the real existence of sensible things, or pretending to know nothing of them’ (Three Dialogues 173). After this point, Hylas cannot assume any knowledge of the intrinsic nature of matter as he is forced to accept that what he perceives as sensible objects are not and not even like the alleged material objects, effectively claiming that there are no real—that is, mind-independent—sensible objects. In other words, in the Second Dialogue, he can only raise those thinner and thinner, relative and extrinsic, but ultimately either contradictory or empty concepts of matter that have remained on the table after admitting that it cannot instantiate any sensory qualities. Note, again, that this sort of scepticism is not what, on the traditional epistemological reading, Berkeley attributes to his opponent and turns to his own idealistic advantage, but rather concerns the denial of the reality of sensory objects—something he rejects through and through.

V. Objections: Representation and Resemblance

In the last section, I want to address a couple of considerations that might be raised as objections to my reading. In the face of what strikes me as quite perspicuous textual support for a merely metaphysical reading, one might wonder why many, if not most, commentators still believe or at least let us believe that Berkeley proposed something like the Representation Argument among the lines of the likeness principle passages. Why do people assume Berkeley did anything else in the likeness principle argument than attempting to prove the metaphysical impossibility of matter instantiating qualities like those of our ideas? One motivation, I believe, is that despite not making one of its premises explicit in his published works (let alone in the context of the likeness principle argument itself), commentators, as we have already seen, often assume that Berkeley accepts all premises of the Representation Argument. But insofar as this objection refers to other (often unpublished) texts or his intellectual context, and clearly does not prove the separate claim that Berkeley indeed put forward such an argument, it is hardly a decisive consideration against my reading of the likeness principle passages specifically. Accordingly, I will not pursue this line of possible objection here any further, especially because Berkeley’s take on the resemblance theory of representation would deserve a detailed examination, expanding my argument unnecessarily. However, what I think is a more apt reason why people tend to read the likeness principle argument in the traditional way concerns the phrasing of the relevant texts.

In section 3, I have already discussed the issue that while proposing the likeness principle argument in Principles 8 (as well as in the Dialogues) Berkeley mentions colours and sounds specifically. In addition, though not unrelatedly, there are a couple of other issues with how Berkeley phrases his argument in the likeness principle passages or in close proximity that might lend themselves to an epistemological interpretation. Specifically, in Principles 8 and in the Dialogues passage introducing the representative realist position and his three-pronged attack on it, Berkeley characterizes the representative realists’ ideas as ‘pictures or representations’, and ‘images and representations’, respectively, of the material qualities they supposedly resemble (see for instance Winkler 1989: 148, for emphasizing this point). As we have seen above, Principles 56 also speaks about ‘images and resemblances’ when summarizing the likeness principle argument. More generally, Berkeley regularly phrased the symmetric relation of resemblance he, on my reading, actually attacks in terms of our ideas resembling the qualities. These can naturally deceive us into thinking that Berkeley has interest in the asymmetric representative relation between the representative realists’ ideas and their correspondent material qualities and even specifically in the resemblance theory of representation. But, as the context of the arguments makes clear, the emphasis is on the supposition that our ideas are resembling representations, representations instantiating sensory properties that are said to be similar to material qualities. As we have seen in the texts, the bit that is under Berkeley’s attack is not that our ideas are representations but that there are material qualities similar to them. So, even if his terminology suggests that, on Berkeley’s reading, the representative realists are committed to the resemblance theory of representation, it does not follow that his argument turns on this conviction.14 Indeed, despite how Berkeley phrased it, as I tried to argue, neither the likeness principle argument’s conclusion nor its role in his wider argumentation depends on any theory of representation. It requires only that (at least) the primary qualities of material objects are regarded by Locke and co. as similar to our essentially sensory mental representations of them.

Moreover, we should take into consideration the highly plausible possibility that Berkeley, in these passages, did not use the term ‘representation’ (or ‘pictures’ or ‘images’) in the sense of intentionality at all, but, as these instances (as well as Dialogues 205) seem to indicate, simply as synonymous with ‘resemblances’ (or ‘copies’), referring to things that are similar to an original.15 Accordingly, with this terminology, he might have not wanted to even suggest anything else than that, on the representative realist view under discussion, matter has qualities like our ideas. Rather, his interest lies firmly in their concept of matter, presuming, as Berkeley argues, the untenable similarity between ideas and material qualities, and he completely ignored as irrelevant the question whether the representative realists are committed to the view that ideas represent through this resemblance. As I tried to argue, since Berkeley does not care about this problem of representation in the likeness principle passages at all, what he actually concludes is compatible with the realists’ retreating to the perhaps unusual—but in the period not rare—view that ideas are non-resembling representations, that is, the material qualities they represent are fundamentally dissimilar to the extension, motion, etc. we perceive.

One, however, might want to point to a passage not far away where, re-phrasing Hylas’s Caesar example, Berkeley’s spokesperson, Philonous, seems to attribute the resemblance theory of representation to his opponent more straightforwardly:16

HYLAS. Yes, Philonous, in some sort there is [an object ‘perceived by sense, which is not immediately perceived’]. For example, when I look on a picture or statue of Julius Caesar, I may be said after a manner to perceive him (though not immediately) by my senses.

PHILONOUS. It seems then, you will have our ideas, which alone are immediately perceived, to be pictures of external things: and that these also are perceived by sense, inasmuch as they have a conformity or resemblance to our ideas. (Three Dialogues 203)

But a couple of things are to be clarified about this passage. Firstly, this is part of introducing the three-pronged attack on representative realism, not strictly speaking the likeness principle argument itself. So, even if it were an allusion to the resemblance theory of representation, it is not necessarily a true reflection of what happens in the likeness principle argument itself. Note that in the Principles we find nothing similar to the Caesar example. Secondly, it is far from clear that the term ‘conformity’ here is used as a synonym of ‘resemblance’.17 As in various authors from Locke to Régis, conformity might mean something much broader than picture-like resemblance. For instance, in Essay IV.iv.4, Locke makes clear that our simple ideas’ conformity lies in nothing else but that they enable us to differentiate and use the things they refer to, which he illustrates by ideas conforming or answering to the powers of the non-resembling secondary qualities. If it is more like a generic notion of covariance or correspondence for Berkeley too, one might argue that the pair of ‘conformity or resemblance’ refers back to ‘representation and picture’ as, arguably, distinct concepts used in introducing the representative realist position just before this passage (just as in Principles 87, where, again arguably, he listed the concept of ‘notes’ in addition to ‘images’). Though, as we have seen, Berkeley was often tempted to use ‘representation’ as synonymous with ‘resemblance’, this disjunctive approach would be fair to Locke’s and Descartes’ view, as it can differentiate representations of primary and secondary qualities, both of which are representations that correspond to an original power or cause, with only the former displaying resemblance to that. If this is right, then Berkeley here does not presuppose, either, that all representations are resemblances, let alone, that all ideas represent through resemblance—just as he acknowledged when qualifying the likeness principle argument in Principles 9 that, according to Locke and co., only some ideas are resemblances of material qualities (naturally allowing that some other representations are not). But even if he did ignore this ‘complication’ here, I have to stress the same point again that nothing in this text proves that resemblance matters to Berkeley due to its putative role in representation and not because, at this very step of his argumentation, he turns his attention to the sensory concept of matter endorsed by the likes of Locke. Not very long before attacking this theory with the likeness principle, undermining the possibility of the likeness between material qualities and our sensory ideas, in the Caesar passage, he first argues that even if matter were sensory, that is, similar to our ideas, since it is not perceived directly, it is not perceived, properly speaking, at all.

Also, regarding this passage as relevant to the likeness principle argument would lead to the curiosity that the representative realists Berkeley is attacking in the latter are committed to thinking that we ‘perceive by sense’ the material object as well as the ideas that represent them. But, as I suggested, Berkeley’s argumentative strategy in the likeness principle argument, both in the Dialogues and, albeit less explicitly, the Principles, rests on the point the Caesar passage actually concludes, namely that perception, properly speaking, can be only direct, taking mind-dependent ideas as its immediate objects. Consequently, the representative realist position against which he ultimately avails himself of the likeness principle is of the inferential variety. For Berkeley and for the representative realists (actual or imagined) he targets with the likeness principle, representation is not a perceptual concept, as it involves some higher-order cognition, like judgement or inference.18 While, until the likeness principle is invoked a bit later, Philonous’s immediate reply does not argue much against the possible inference the representative realist might want to make to the effect that there are material qualities like our ideas, the Caesar passage makes his conviction explicit that we cannot, properly speaking, (sense) perceive an external object via our ideas, which are the only objects we directly perceive. Even if the idea or, using a related scholastic term, the sensible species is qualitatively the same as, or formally identical with, the external object, since they are two numerically distinct objects, as Hylas proposes, we cannot perceive the object itself by sense. Or, using other scholastic jargon, we cannot contuit both the thing and the representing medium at the same time.19 For Berkeley, it seemed obvious that even if they are alike, the object itself is not presented to me in the medium in the same (perceptual) way as the medium is presented to us. As such, in the Caesar passage Berkeley is not as much interested in undermining the intentionality of the representative realists’ ideas as in arguing that even if we allowed them to be resembling representations we would not be perceiving the represented thing itself.

But even if I am wrong about this exchange, to stress it again and for the last time, it surely does not prove that the separate likeness principle passages should be read as an argument aiming to establish an epistemological point about representation rather than merely a metaphysical conclusion about the impossibility of sensory matter. To sum up, in his ambitious quest of going through the various concepts of matter available to his realist opponents, Berkeley used the likeness principle to attack Locke’s and Descartes’ representative realist position according to which there are material objects with qualities similar to our ideas. And, as I tried to argue in this paper, he attacked this position not by seeking to undermine the representability of such mind-independent qualities but by stressing that the metaphysics underlying this theory of representative realism is nothing short of being manifestly absurd.


1Cummins adds that ‘that the representationalist could retreat to the view that there are material substances, but that presented or sensory objects do not represent them. Hence, despite Turbayne’s claim, to refute representationalism is not to refute the claim that there are material substances’ (1966: 64). So, while Cummins rightly points out that the likeness principle passages do not refute all concepts of matter, he agrees with Turbayne—in my view, mistakenly—that it aims to do away with all theories of representative realism. 

2Jacovides (2009) calls a similar argument ‘Resemblance Argument’, which he locates in Three Dialogues 190. For a helpful criticism, see Marusic (2009). 

3See Pitcher (1977: 118–20) discussing and rejecting a weaker, epistemological interpretation of the likeness principle. One nonetheless might worry that not all the interpreters I mentioned above attribute (i.) to Berkeley, for instance those like Winkler (1989) who advocate an epistemological defence of the likeness principle. But I think it is not only true of the metaphysical interpretation proposed by Cummins (1966) or the ‘middle-way’ readings such as Ryan’s (2006) ‘new metaphysical’ and West’s (2021) conceptual reading but even Winkler’s interpretation is committed to (i.). As I understand him, his proposal is epistemological only with regard to the justification Berkeley has for the ultimately metaphysical principle concerning the impossibility of any likeness between an idea and a non-idea. As he puts it, ‘an epistemological defence of the likeness principle arrives at the conclusion that ideas and things cannot in fact be alike’ (1989: 141, emphasis added) and ‘Any genuine state of affairs is such that we can, at least in principle, verify that it obtains’ (1989: 147, emphasis added). In any event, even if verificationism does not get us to a robust metaphysical interpretation of the likeness principle, most of the numerous commentators I referred to realize that, whatever epistemological purpose it serves in his argumentation against representative realism, Berkeley in fact speaks about the actual impossibility of resemblance between ideas and non-ideas. Thanks to an anonymous reviewer for pushing me to clarify this point. 

4Actually, for Foucher and Huet, who are cited by Hill (2011) as the most probable sources of Berkeley’s Representation Argument, the initial targets were indeed Malebranche’s and the scholastics’ (species-based) theory of perception. Though Foucher might have misunderstood Malebranche a bit, Arnauld also raised the criticism against the latter that positing ideas in the mind of God as representations of material objects will not help us represent them at all. For a useful summary of the species-based medieval views as well as Arnauld’s criticism of Malebranche, see Adriaennsen (2017). 

5But, as Hill himself notes, not even Régis, Arnauld, la Forge, Rohault accepted the resemblance theory of representation. Given that Descartes and Malebranche are also controversial in this respect, not many Cartesians remain on our list, either. Among their critics, Hill mentions Huet, who argued against the possibility of likeness between objects and our perceptions. Nonetheless, he does not develop this point into an argument that representation is impossible because of the lack of resemblance. Moreover, even his attack on the likeness is very unlike Berkeley’s argument, as it attacks a material view of ideas, and hence does not rest on dualism, in sharp contrast to Foucher’s and Berkeley’s intuition. In any event, he probably had no influence on Berkeley’s likeness principle, as Huet’s relevant work, the Traité Philosophique de la Foiblesse de l’Esprit Humain, though written much earlier, was published only in 1723, 13 years after Berkeley’s Principles (see Hill 2011: 57). And with regard to the only possible source of his alleged Representation Argument, Berkeley probably did not know Foucher first-hand, either. Not even Bayle, who not only knew Foucher very well but disseminated many of his arguments, recapitulates the Representation Argument in his Dictionary, which Berkeley, as is well-known, heavily relied on. 

6While complaining that he failed to realize this plan, most commentators assume that this remark concerns the two justifications he gave for the likeness principle in the foregoing entry. But it seems rather to refer to all the 19 steps he mentions there, with only a couple being about the likeness principle. So, while he indeed did not discuss publicly any of his reasons for adopting the likeness principle, he did offer separate, if not necessarily shorter, arguments for the various points of this long and complex argument. At this point, I want to add a quick methodological note concerning the Notebooks: while I do not want to rest any argument merely on their dubious authority, I believe that some of the views Berkeley actually went on to endorse publicly are expressed there in more conspicuous and perspicuous ways. 

7As he puts it in Notebooks 299, if there were an assertible similarity between the perception and the material quality of extension, it would require an ‘idea i.e perception thought, or sensation […] to be in an unperceiving thing [which] is a Contradiction’. The incomparability argument also comes up this entry as well as in Notebooks 51, where he writes that ‘A man cannot compare 2 things together without perceiving them each, ergo he cannot say any thing wch is not an idea is like or unlike an idea.’ In my view, Berkeley was aware of the limitation of this argument, as it does not exclude the possibility that material objects are like our ideas, after all, showing merely that it is not a claim anyone can ‘say’ with any justification. Later, in his published works, he realized perhaps more clearly that he wanted to go further than this merely epistemic point. In any event, he clearly thinks that resemblance requires sharing a property of the same sort. It is the driving force behind the other justification given in Notebooks 378, but mentioned also in entry 496: ‘they can have nothing common & consequently no likeness.’ For the incoherence of the view attacked, see also as early as Notebooks 37–37a: ‘Extension to exist in a thoughtless thing A is a contradiction. or rather in a thing void of perception. Thought seeming to imply action.’ For the intuitively obvious nature of the likeness principle, see for instance the Theory of Vision Vindicated 11, claiming that ‘it seems evident that an idea can be only like another idea’. 

8While the relativity argument is traditionally taken to be an argument against direct realism, here Berkeley is not using it like that. It is rather meant to show that material objects cannot have all the ideas we perceive them having, regardless of whether, in perception, we are directly related to material qualities in the objects or representations of them in our minds. So, while representative realism might be a viable answer to the problem of conflicting appearances, which if identified with mental representations, as opposed to material qualities of a single object, are not in actual conflict with one another, it has not taken us even a step closer to know how things really are. In the Dialogues, the argument serves multiple purposes, in the Principles, Berkeley actually appeals to it against representative realists to show the latter point. 

9A similar thing is going on with Jacovides’s (2009) Resemblance Argument located in Three Dialogues 189. As Marusic (2009) quotes Berkeley’s actual conclusion, it is not to prove any point about what we can and cannot represent, but that, for any sensory quality, that is, directly perceivable ideas or ones similar to them, it is ‘impossible it should subsist in that which doth not perceive it’. 

10As many point out, the term ‘again’ helpfully divides the section into two parts worth discussing in turn. Some take it as introducing a separate argument, some as the only proof Berkeley published for the likeness principle simply stipulated in the first part. I think it is an extended version of the basic argument he gives earlier. 

11Though if a representative realist thinks that the representation perceived directly brings us into perceptual, albeit mediate, contact with the object itself, then she might disagree. As we will see, this is not how Berkeley understood their position, anyway. 

12I am grateful to an anonymous reviewer for pushing this line of objection. I might also add as a response that it is perhaps not accidental that Berkeley talks about the likeness between a colour and another colour, suggesting that what he cares about is not the particular content, say the particular colour or shade, the representation displays but merely the modality of the perception. As my reviewer also argued, that, as we have seen in note 7, Berkeley appeals to comparability in the Notebooks might more clearly suggest his interest in the content of the representation. This consideration, however, is conspicuously missing from his published arguments, possibly because he realized that it does not help him to the metaphysical conclusion he is actually getting at. 

13So, while it is controversial whether Berkeley ever used the inconceivability principle (see Holden 2019 denying it), that is, concluding the impossibility of a state of affairs from its inconceivability, on my reading, here he is committed only to the contraposition of the uncontroversial conceivability principle, holding that since something is impossible, it is inconceivable. 

14But let me add briefly that speaking in terms of pictures and images in itself does not commit one to any theory of representation. As in Descartes and Arnauld, it is quite common in other authors from the period that all ideas or mental representations are called figuratively images or pictures insofar as they are about intentional objects, without assuming any theory about the way they represent those objects. While Berkeley definitely uses these terms in the sense of resemblance, he might not fully be consistent. 

15Representation in this sense is more like ‘re-presenting’ something perceived earlier or rather an idea similar to that original. I think this is all Berkeley means when claiming for instance that ideas of imagination are re-presentations (or copies, images, pictures or resemblances) of earlier perceptions made present to the mind in their absence. It might also simply mean ‘presenting to the mind’ or, more specifically, revealing to it the true nature of things, whether through a direct or indirect perceptual relation. Among others, Descartes also used the term representation in a similarly manifold way, so we should be cautious about reading our notion of representation back to early modern philosophy. 

16Bracken (1974: 46–47) is one of the examples who explicitly make the connection between the Caesar passage and the likeness principle argument. 

17Carriero, who otherwise acknowledges that Locke denies that ‘sensory cognition of the world in general involves resemblance’ (2003: 27), thinks that this passage is all about resemblance or ‘formal identity between sensory and sensed thing’ not only because of the words ‘resemblance’ and ‘conformity’ but also the example of a statue depicting Caesar (2003: 40). In my opinion, neither should be seen as decisive evidence. 

18Cf. Dicker (2011), who thinks that the likeness principle obviously works for the inferential variety of representative realism, but not against the perceptual one. If I am right, in the likeness principle passages, which as I read them, argue specifically against a sensible concept of matter, Berkeley assumed that the latter is a non-starter. In the Dialogues, he makes this clear in the Caesar argument, showing that mediate perception cannot be anything else but an inference. Carriero (2003: 40–41) argues that the mediation at issue here is more like suggestion, like hearing a coach mediately, suggested by the sounds we actually hear. I do not think this is the right way to read this passage, as Berkeley seems to introduce the coach-analogy as a separate point, qualifying his main claim that nothing is perceived by sense except what is directly perceived by adding that ‘we may in one acceptation be said to perceive sensible things mediately by sense’, but this is not the cognition he talked about earlier with regard to the image of Caesar, a sort of cognition proceeding, as he says, ‘from reason and memory’. Also, a couple of sentences later he contrasts knowledge from mediate perception, that is, ‘reason and reflection’, with knowledge from immediate (sense) perception. In any event, the suggestion relation equally entails that what is only mediately perceived is not really perceived, as the suggested ideas are merely imagined, even if it is a sort of cognitive mechanism, unlike conscious inferences, we are normally not aware of doing. 

19For a helpful discussion on John Peckham’s attempt to use the notion of contuition to defend his species theory from the charge that we only perceive the species but cannot perceive in it what it represents, see Adriaennsen (2017: 225–27). 


I presented an early draft of this paper at Dominik Perler’s Theoretical Philosophy Colloquium at the Humboldt-Universität zu Berlin. I’m grateful to all members of this group for their challenging comments. I also want to thank Peter West and the two anonymous reviewers for helping me improve both the content and structure of the paper. I’m also grateful to the Alexander von Humboldt Foundation for financially supporting my postdoctoral research at the Humboldt-Universität zu Berlin, as well as the Human Abilities Centre for Advanced Studies in the Humanities, where I’m currently a fellow.

Competing Interests

The author has no competing interests to declare.


  1. Adriaennsen, Han Thomas. 2017. Representation and Scepticism from Aquinas to Descartes. Cambridge: Cambridge University Press. DOI: 

  2. Allaire, Edwin B. 1982. “Berkeley’s Idealism Revisited.” In Berkeley: Critical and Interpretive Essays, edited by Colin M. Turbayne, 197–206. Minneapolis: University of Minnesota Press. 

  3. Berkeley, George. 1948. Philosophical Commentaries In Berkeley 1948–57, 1: 9–104. [= Notebooks, referenced by entry number]. 

  4. Berkeley, George. 1948. The Theory of Vision, or Visual Language, Shewing the Immediate Presence and Providence of a Deity, Vindicated and Explained. In Berkeley 1948–57, 1: 251–76. [= Theory of Vision Vindicated, referenced by section number]. 

  5. Berkeley, George. 1948–57. The Works of George Berkeley, Bishop of Cloyne. 9 vols. Edited by A. A. Luce and T. E. Jessop. London: Thomas Nelson and Sons. 

  6. Berkeley, George. 1949. A Treatise Concerning the Principles of Human Knowledge. In Berkeley 1948–57, 2: 21–113. [= Principles, referenced by section number]. 

  7. Berkeley, George. 1949. Three Dialogues between Hylas and Philonous. In Berkeley 1948–57, 2: 165–263. [= (Three) Dialogues, referenced by page number]. 

  8. Bolton, Martha Brandt. 1987. “Berkeley’s Objection to Abstract Ideas and Unconceived Objects.” In Essays on the Philosophy of George Berkeley, edited by Ernest Sosa, 61–84. Holland: D. Reidel. DOI: 

  9. Bolton, Martha Brandt. 2008. “Berkeley and Mental Representation: Why Not a Lockean Theory of Ideas?” In New Interpretations of Berkeley’s Philosophy, edited by Stephen H. Daniel, 79–108. Amherst, NY: Humanity Books. 

  10. Bracken, Harry M. 1974. Berkeley. London: Macmillan. DOI: 

  11. Carriero, John. 2003. “Berkeley, Resemblance, and Sensible Things.” Philosophical Topics 31 (1/2): 21–46. DOI: 

  12. Cummins, Phillip D. 1966. “Berkeley’s Likeness Principle.” Journal of the History of Philosophy 4 (1): 63–69. DOI: 

  13. Dancy, Jonathan. 1998. Notes for A Treatise Concerning the Principles of Human Knowledge by George Berkeley, edited by Jonathan Dancy, 194–218. Oxford: Oxford University Press. 

  14. Dicker, Georges. 1985. “An Idea Can Be Like Nothing but an Idea.” History of Philosophy Quarterly 2(1): 39–52. 

  15. Dicker, Georges. 2011. Berkeley’s Idealism: A Critical Examination. Oxford: Oxford University Press. 

  16. Downing, Lisa. 2011. “George Berkeley.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. 

  17. Fasko, Manuel and Peter West. 2020. “The Irish Context of Berkeley’s ‘Resemblance Thesis’”. Royal Institute of Philosophy Supplement 88: 7–31. DOI: 

  18. Foucher, Simon. 1675. Critique de la Recherche de la Verité. Paris: Coustelier. [Referenced by page number] 

  19. Hill, Jonathan. 2011. “Berkeley’s Missing Argument: The Sceptical Attack on Intentionality.” British Journal for the History of Philosophy 19 (1): 47–77. DOI: 

  20. Holden, Thomas. 2019. “Berkeley on Inconceivability and Impossibility.” Philosophy and Phenomenological Research 98 (1): 107–22. DOI: 

  21. Jacovides, Michael. 2009. “How Berkeley Corrupted His Capacity to Conceive.” Philosophia 37 (3): 415–29. DOI: 

  22. Locke, John. 1975. An Essay Concerning Human Understanding. Edited by Peter H. Nidditch. Oxford: Oxford University Press. [= Essay, referenced by book, chapter and section numbers]. 

  23. Marusic, Jennifer Smalligan. 2009. “Comments on Jacovides ‘How Berkeley Corrupted His Capacity to Conceive’.” Philosophia 37 (3): 431–36. DOI: 

  24. Pitcher, George. 1977. Berkeley. London: Routledge & Kegan Paul. 

  25. Ryan, Todd. 2006. “A New Account of Berkeley’s Likeness Principle.” British Journal for the History of Philosophy 14 (4): 561–80. DOI: 

  26. Turbayne, Colin M. 1954. Introduction to Three Dialogues between Hylas and Philonous by George Berkeley, edited by Colin M. Turbayne, vii–xxviii. New York: Bobbs-Merrill. 

  27. West, Peter. 2021. “Why Can An Idea Be Like Nothing But Another Idea? A Conceptual Interpretation of Berkeley’s Likeness Principle.” Journal of the American Philosophical Association 7 (4): 530–48. DOI: 

  28. Winkler, Kenneth. 1989. Berkeley: An Interpretation. Oxford: Oxford University Press. 

comments powered by Disqus