by Dave Maier
A reader writes in to ask a question about skepticism. Enrique asks:
I get confused with the dream argument. It confuses me to read the argument of the dream because I see several interpretations. Do they refer to the dreams we all experience while sleeping? Or do they refer to a class of dreams that has nothing to do with our sleeping bodies?
Enrique goes on to note that philosophers like Barry Stroud seem to equivocate when they talk about this argument, saying both that the argument refers to ordinary dreams (that is, the ones we have while sleeping) and at the same time that ordinary dreams do not pose any particular skeptical problem (although "philosophical dreams" do). In either case it seems that the dreaming argument is not where the real action is.
I agree with Enrique that the dream argument does not seem to get at what is really driving the skeptical worry. It’s true that a lot of philosophers address that argument specifically, which makes it look more important than it is, but Stroud is right that that’s not the real issue. And indeed there are a number of perfectly good answers to the dreaming argument in the literature, none of which put to rest the skeptical problem as a whole. So I advise noting the historical importance of the argument, but not to take it (or its refutations) too seriously. But since understanding the skeptical dialectic is essential for understanding modern philosophy, let’s go past that particular formulation to say more about skepticism in general. (See also an earlier post on the subject.)
One way of making progress here is to distinguish different kinds or aspects of the Cartesian commitments driving the modern skeptical dialectic. First, of course, we know that traditional Cartesian metaphysics divides the world into mind (res cogitans) and matter (res extensa), and we naturally think of the epistemological puzzles as following from that: if we are constituted as knowers by one sort of substance, how do we come to know about another sort of substance entirely? We are used to regarding our senses as delivering knowledge about the physical world, and indeed how could we not? But on Descartes’s mechanical conception of the physical body, our senses themselves are physical processes outside the mind, and thus on the other side of the metaphysical (and thus epistemological) gap. This is one thing making it look like dreaming is a good example: our purported sensory perception during the dream is itself (as it happens, and as we see when we wake up) a merely mental process – one which is, unfortunately for our knowledge, subjectively indistinguishable from “real” perception. But of course as we’ve noted this way of thinking of it leads to further puzzles.
We might do better to put the metaphysics to one side for now (but bring it back in at just the right time). The skeptical conclusion seems most dramatic when it concerns our knowledge of an “external world”, but its lasting significance has depended on considering it purely as an epistemological puzzle – that is, independently of what exactly it is that we seem both to know and (thanks to skeptical considerations) not to know. After all, the ancient skeptics didn’t need Descartes’s dualistic metaphysics to get their argument going, and they didn’t attack knowledge of the “external world” in particular.
Descartes himself has another famous argument which is more to the point here – that of the famous “evil demon” (malin génie), but its contemporary analogue is the “brain in a vat” scenario, in which my apparent sensations are caused not by the world, but instead by an appropriately programmed computer hooked up to my brain, which is floating in a vat. From the inside, the skeptic notes, we cannot distinguish between sensations caused by the computer, and those caused by, say, actual trees. Here everything is physical – brain, vat, computer – so no questionable metaphysics is required to get the argument going (or so it seems …).
This may look like simply more of the same, and let me tell you, a great deal of the massive literature on this issue is indeed not particularly helpful, but some of the discussion following Hilary Putnam’s influential presentation of this example (and related work by Saul Kripke) eventually led to some valuable insights, not simply into skepticism, but into the metaphysical assumptions which underly it as well.
Let’s leave that for another time (although see here for some of it). The important move – even if made in a variety of ways by various people, with varying results – was to recognize a role not merely for epistemology (as in the original skeptical paradox) and metaphysics (as already noted) but also for semantics.
The main insight of “semantic externalism” broadly construed is an anti-skeptical one (and was explicitly so described in Putnam’s original formulation in Reason, Truth, and History). The idea of the BIV example was to portray someone in a particularly poor epistemic situation. But is the BIV really so badly off? And even if so, are we really sure we can intelligibly describe that purported possibility?
According to Putnam, since the BIV has never been in the sort of causal contact with actual trees which it would need in order for its word “tree” to mean “actual tree,” when it says “there is a tree in front of me,” it means something which is in fact true, and perfectly well known: that there is, as we would say, a vat-tree in front of its apparent body, with all that this entails, e.g. that if it “walks forward,” it will “bonk its head” on the “tree” and thus experience (real!) pain.
This suggests that the BIV is not nearly as badly off, epistemically speaking anyway, as the skeptic makes it sound, having as it does plenty of perfectly reliable knowledge about its actual surroundings, and ignorant only about the ultimate (or higher-level, at any rate) nature of reality – which sounds a lot like you and me, unless you happen to know the ultimate nature of reality, because I don’t; and the knowledge that I care about is about that thing there and things like it, not some allegedly realer reality which I can apparently have no contact with and can barely even conceive.
Putnam’s argument doesn’t stop there. If I am to consider, as the skeptic demands, the alleged possibility that I am, as he puts it, a “brain in a vat,” I have to ask myself what these words mean. For me, in the vat, “vat” means “one of those” (that is, a vat-vat), and I can perfectly well see that I am not in that situation. After all, I’m out here, and not in there. This renders the skeptical possibility unintelligible – if it were true, we would lack the concepts it requires for its formulation, and only intelligible statements can be true [QED] – and thus not worth our trouble.
One problem with Putnam’s semantic argument, as critics point out, is that even if it works against the global “external world” skeptic, it doesn’t deal with we like to call "recent envatment." If you put my brain into the vat last night, it seems that all is okay semantically: “tree” means “real tree,” and any subsequent belief that I might be standing in front of one will thus have enough intelligible content to count as false, as the skeptic suggests may be the case for all I know. So we are back in the skeptical soup.
Second, all this argument really does is add to the BIV’s epistemic difficulty a further semantic one: not only does it not know the actual situation, it can’t discuss or even consider the possibility intelligibly, as it lacks the relevant concepts with which to do so. This somehow fails to dispel the skeptical notion that we ourselves might be, if not in that same situation, at least a relevantly analogous one – one which we cannot, as per the argument, intelligibly discuss, but a disturbingly possible one all the same. (Something like this is why responses to skepticism which amount simply to attacks on the intelligibility of purported skeptical possibilities aren’t particularly satisfying in general.)
A more familiar example, even if less picked over in the literature (see e.g. this collection), is the popular science-fiction film The Matrix. (I will assume familiarity with it here: if you haven’t seen it, stop reading this right now and watch the movie instead, as it is much more entertaining.)
At the beginning of the movie we see a hacker. Let’s call that guy Thomas Anderson. Later we see someone flushed out of his, well, vat, and brought into the “real world”; let’s call that guy Keanu. Later still we see a leather-clad guy in shades defeating the evil Agents and saving the day for humanity; let’s call that guy Neo. With me so far?
Anderson’s semantic situation turns out to be hard to describe, being somewhere on the scale between “always-BIV” and recent envatment. He has always been a coppertop and has, as Morpheus points out to Keanu later, never used his eyes (an assertion by which he is surprised, but has no trouble understanding, given the circumstances). On the other hand, the AIs modeled the Matrix on an actual past Earth, and the referential chain from word to world might very well be taken simply to have a few extra steps. This allows us to see Keanu’s assertion (should he make one, which in the actual film he doesn’t) that he has indeed used his eyes plenty of times, as intelligible but false. “Eyes,” in the language they are speaking, means those actual organs which are even now acclimating to light, and which prompt Morpheus’s true observation that Keanu has never used them.
But that sounds odd too. Presumably (at the moment of his disenvatment anyway) Keanu speaks the same language as Anderson does (did). In fact it is natural to consider all three names as referring to the same person. By taking the red pill, Anderson comes to know that he is Keanu and subsequently Neo as well. But regardless of the referential history of the word “eyes,” and of whatever he will later say about the matter, Anderson has only ever used that word to refer to, as we would now say, things which were never eyes at all. This brings out the weirdness, as utilized by Putnam in his anti-skeptical argument, of counting the vast majority of Anderson’s beliefs as false, rather than taking them to be true and known, and counting only the one big deception as a single false belief (that he is not in the Matrix, whatever that is).
We have also simply stipulated that the names “Anderson,” “Keanu,” and “Neo” refer to the same individual. But now let us ask some pointed questions.
— Does each know how to play chess?
— … know what a rainbow looks like/has seen a rainbow?
— … know what strawberries taste like?
— … speak English?
— Has each met Trinity/spoken with her?
— Does Trinity love each of them?
If the three are the same person, that suggests that for most if not all of these questions, the answer must also be the same, whether yes or no. But that’s going to start to sound funny. How does Anderson know how strawberries taste if he’s never tasted them (as one character in the movie points out)? How does he know how to play chess if he’s never picked up and moved a chess piece? Or is that even what we want to say in the first place? At issue here is not simply the oddness of the Matrix scenario but also what exactly we think names are for, which certain philosophers seem to have simply, shall we say, underthought.
Another distinctive feature of the Matrix scenario is more significant. One reason the BIV scenario is epistemically worrying is that the “reality” the BIV thinks it knows is, from our point of view, ontologically fragile. It seems that the AI controlling the BIV’s sensations is constrained only by its whims (ok, programming). But the Matrix is not like that. A key difference between the Matrix and BIV scenarios is that when the BIV takes itself to communicate with other humans, that’s just part of the illusion. But the Matrix is a group illusion, created all at once as the environment for all of us. When you are in the Matrix, apparently, you really are communicating with whoever else is in there with you. This allows us to (as Davidson would say) “triangulate” with our interlocutors concerning our shared environment, even though that environment is in a certain way illusory. I say “in a certain way” because now, as a shared environment, acting as the third point of our mutually interpretive interaction, it is in another way perfectly objective – and thus ontologically stable enough to support not only semantic relations to it but epistemic ones as well.
We might have thought to dodge the issue of, for example, whether Anderson can play chess (“real chess”?) by emphasizing the abstract nature of chess as a formal system. One can “play chess” with imaginary pieces just as well as with real ones. Presumably Anderson can “add” and “subtract” as well, for the same reason. But the issue comes up with concrete objects as well, rendering it unavoidable. At one point in the film, while in the Matrix with the others, Neo sees a cat and says so, referring to it, naturally enough, as a “cat”. So does “cat” mean “matrix-cat” in the language he’s speaking? But if so, what should he call any such creature (that is, any real creature of the same sort as that on which the matrix-creature was modeled) he may see after they return to the real world? Not a “cat” – that means “matrix-cat,” as we said. But you can’t go around in the real world calling everything “real” this and “real” that – your interlocutors would beat you with sticks.
Again, this might not matter if we were simply observing the solo BIV merely seeming to converse with what are in fact illusory companions. But in this case, not simply in the real world but also in the Matrix, you can see what I am pointing at and can refer to it yourself. I take the semantic bottom line to be that just as the word “elf” can perfectly well, depending on context, mean “apparent elf” or “actor who plays an elf” — non-metaphorically, without asterisks – so too can “cat” mean either “real cat” or “matrix-cat” — or, crucially, “cat, matrix-cat, whatever” (again, without metaphor or asterisk). This interesting result has repercussions well beyond the epistemological problem with which we began.
The film brings up other philosophical issues as well, e.g. personal identity, but my head is spinning too much, plus we’re out of space for today. Thanks to Enrique for his provocative question!