Consider the following parallel passages from Berkeley’s Principles and Dialogues:
so long as men thought that real things subsisted without the mind, and that their knowledge was only so far forth real as it was conformable to real things, it follows, they could not be certain that they had any real knowledge at all. For how can it be known that the things which are perceived, are conformable to those which are not perceived or exist without the mind? (PHK sect. 86)
It is your opinion, the ideas we perceive by our senses are not real things but images or copies of them. Our knowledge therefore is no farther real, than as our ideas are the true representations of those originals. But as these originals are in themselves unknown, it is impossible to know how far our ideas resemble them, or whether they resemble them at all. We cannot therefore be sure we have any real knowledge. (DHP, L&J p. 246)
It is usually thought that in these two passages Berkeley is assuming some sort of internalism about justification. That is, he is assuming that we can’t gain knowledge by means of the senses unless we know that the senses are reliable. On this reading, Berkeley is arguing that representative realism leads to general skepticism, because of the impossibility of a non-circular justification of trust in the senses. Reid probably read Berkeley this way, and this was probably the reason why Reid thought that externalism about justification would allow him to escape Berkeley’s argument.
Now, I don’t want to deny that internalist assumptions may be in the background at many points in Berkeley’s writings, but I do want to point out that, as the bolded phrases show, these texts make no such assumption. The structure of the argument in these two passages is rather the following:
- If representative realism is true, then we gain knowledge by means of the senses only if our perceptions match mind-independent objects.
- We cannot know that our perceptions match mind-independent objects.
- If representative realism is true, we cannot know that we gain knowledge by means of the senses.
Therefore,
In other words, representative realism engenders second-order skepticism; it prevents us from knowing that we know. Externalism is not a way of escaping from this argument. Unless the externalist-representative-realist wants to allow knowledge of the reliability of the senses to rely directly or indirectly on the senses themselves (see Van Cleve), it would seem that she is stuck accepting the second-order skeptical thesis. Berkeley, however, finds the second-order skeptical thesis unacceptable.
It is in fact not surprising that much of Berkeley’s discussion should take place at the second-order. After all, the structure of the dialectic, both between Berkeley and his real-world opponents and between his fictitious characters Hylas and Philonous, is a debate about whether ‘the vulgar’ or ‘the mob’ or ‘the illiterate bulk’ have knowledge of familiar objects like apples, tables, and cherry trees, and if so how. Berkeley’s complaint against his opponents is that, on their theories, it cannot be proved that the gardener knows his cherry tree. He claims that his own theory does not have this defect: the philosopher who has grasped Berkeley’s arguments thereby comes to know that the gardener knows that his cherry tree exists.
(Cross-posted at blog.kennypearce.net.)
Hi Kenny,
I see the second-order aspect of Berkeley’s argument, but I think it should be pointed out that your reconstruction of his argument is invalid. I also think that it is not an accurate representation of what is going on in the text.
The argument in both passages is designed to show, as the bolded sentences indicate, that if representative realism is true, then we cannot know that our knowledge is real. The reason Berkeley gives relies, first, on a definition of real knowledge, which comes straight out of Locke’s Essay (4.4.3: “Our Knowledge therefore is real, only so far as there is a conformity between our Ideas and the reality of Things.”) This can be put in the following form:
(1) To say that our knowledge is real is to say that our ideas conform to the reality of things.
Berkeley then assumes (without stating) a simple principle about knowledge:
(2) If to say that P is to say that Q, then we can know that P iff we can know that Q.
From this, he infers
(3) We can know that our knowledge is real iff we can know that our ideas conform to the reality of things.
Finally, Berkeley claims
(4) If representative realism is true, then we cannot know that our ideas conform to the reality of things.
The reason Berkeley gives for (4) is that if representative realism is true, then there is a veil of ideas between us and the things that those ideas represent. But representation is a matter of resemblance, and thus in order to know that our ideas conform to external things under representative realism, we need to know that our ideas resemble those external things. But in order to know that our ideas resemble those external things, we need to be able to compare our ideas with the external things directly. But in order to compare our ideas with external things directly, we need direct access to those external things, which, according to representative realism, we cannot have.
From (3) and (4), Berkeley then infers
(5) If representative realism is true, then we cannot know that our knowledge is real.
This is a kind of second-order claim, because the consequent of (5) is a proposition about knowledge about knowledge. But it should be distinguished from what you appear to think Berkeley is after, which you characterize as follows:
(5*) If representative realism is true, then we cannot know that we know.
Berkeley does not state (5*), and on my view he does not argue for (5*). This is not a surprise, really. For what, according to Locke (the arch-representative-realist), is knowledge anyway? It’s perception of an agreement or disagreement between two ideas. So, according to Locke, whether I know that I know reduces to whether I know that I am perceiving an ideational (dis)agreement. And it’s pretty clear, after all is said and done, that (to both Locke and Berkeley) there is no problem here: because I have unmediated introspective access to my perceptions when I am having them, I can and do know that I am perceiving an ideational (dis)agreement when I am perceiving it. Indeed, it seems to me that Locke endorses what we now think of as the KK-principle:
(KK) If X knows that P, then X knows that (s)he knows that P.
And I see no reason for Berkeley to disagree with this. What Berkeley notes is not that representative realism entails that we cannot know that we know, but rather that representative realism entails that we cannot know that our knowledge is real.
Sorry, this comment came out as more long-winded than I intended when I started writing it. Anyway, I hope it helps to clarify matters.
cheers,
Sam
Hi Sam,
I agree that my argument relies on an unstated principle about knowledge, which is something like your (2). I had in mind some kind of closure principle, but the restrictions that need to be put on a closure principle in order to make it plausible (e.g. not to require logical omniscience) would require some tweaking to my formulations.
I also think it is quite clear that Berkeley has that quotation from Locke in mind in both of the passages I cited. Insofar as the argument is ad hominem against Locke, something like what you say is probably what’s going on. However, Berkeley has to reject Locke’s definition of knowledge because Berkeley believes that we can know verbal propositions (sentences) which contain words that don’t correspond to ideas. Note that ‘the cherry tree exists’ is such a proposition, since there is no idea of existence. As a result, I don’t think Berkeley would endorse nearly as much of the Lockean line of thought as you say. Since Berkeley rejects this view of knowledge, he doesn’t have Locke’s reasons for endorsing the KK principle (he may have other reasons). Also, Locke’s category of ‘sensitive knowledge’ is famously problematic, and we don’t have a lot of evidence as to how exactly Berkeley understood it. Finally, Locke is not the only target; Berkeley thinks his skeptical arguments gain traction against Descartes, Malebranche, and others, so although Berkeley is clearly alluding to Locke in his formulation, I don’t think we ought to read the argument in such a way as to depend on the details of Locke’s epistemology.
You are certainly right that Berkeley appears to be contrasting ‘real’ knowledge with some other kind of knowledge. I suspect that what Berkeley means by ‘real knowledge’ is knowledge of real things, as opposed to imaginary things. (I’m assuming that Locke’s ‘mixed modes’ will count as ‘imaginary’ in the relevant sense unless and until we know that they have instances.) This is actually in line with Locke’s own discussion in the opening sections of 4.4, especially section 1. Berkeley’s real target, here as elsewhere, is the representative realist’s criterion of reality. (This is exactly how Locke frames the issue in 4.4.3 as well.) Something like this is perhaps going on: I know that the cherry I’m perceiving is red. The thing I don’t know, if the representative realist is right, is whether I’m perceiving a real cherry or an imaginary one. Hence I don’t know whether my knowledge is real — that is, I don’t know whether it is about a real thing. I guess you are right that this doesn’t prevent me from knowing that I know that the cherry I’m perceiving is red.
Nevertheless, there is an important first-order/second-order contrast. According to the representative realist, I have knowledge about real things iff the ideas which are the immediate objects of my knowledge match some mind-independent objects. Further, according to the representative realist, my ideas do match. So, if we grant the representative realist this much, then we will have to concede that, on the representative realist’s view, I know about real things. However, Berkeley now objects, I won’t know that I know about real things, because I can’t know that the matching obtains. I know the perceived cherry is red, and in fact the perceived cherry is a real cherry, so I know of a particular real cherry that it is red, but I don’t know that I know of a real cherry that it is red, since I don’t know that the object of my first-order knowledge is real and not imaginary. This is similar to your (5), but I think it shows more clearly the way in which this is a second-order claim.