Stroud is very concerned with the question whether the skeptic means the same thing by "knowledge" as we ordinarily mean by it. (See pp. 35-6 in his Chapter 1, and pp. 40-42 in his Chapter 2.) Stroud imagines someone who distorts the meaning of "There are no physicians in New York," by setting the requirements for being a "physician" absurdly high. Some philosophers think something similar is going on with the skeptic's claim that "We can't know anything about the external world." They think the skeptic's claim only comes out true when you distort the meaning of "knowledge" into some special philosophical sense.
For example, some skeptical arguments (like Unger's) turn on understanding "knowledge" to require absolute certainty. But it's not really clear whether the ordinary person ordinarily understands knowledge to require that.
Even if we set aside skeptical arguments that demand certainty, we've still seen that the skeptic requires us to know we're not dreaming, in order to know anything about the external world. And in everyday life, or in a legal trial or a science lab, we don't ordinarily insist that you know you're not dreaming before allowing that you know anything. So does Descartes' requirement, enshrined in Stroud's Principle or in one of the Closure Principles, raise our ordinary standards for knowledge in an artificial way?
Austin thought that the skeptic does distort our ordinary standards for knowledge. The skeptic makes them out to be more stringent than they really are. Stroud's Chapter 2 sets out and responds to Austin's views about this.
Stroud mentions three threads in Austin's writings about skepticism.
On pp. 45-5, he discusses Austin's claim that, to raise a legitimate doubt about someone's knowledge claim, you have to suggest some specific way in which the subject might be mistaken. (Descartes does seem to do this.)
On pp. 46-8, he discusses Austin's claim that dreams are qualitatively distinguishable from waking experiences. (According to Austin, dreams always have a special "dream-like quality.")
On pp. 48-53, he discusses the most important thread in Austin's writings. This is the claim that you only need to rule out some possibility if there's some special reason to believe that it does now obtain.
For instance, if you have positive evidence that you are a brain in a vat (like the guy in the Pollock and Cruz selection we read at the start of term), then you need to rule that possibility out, before you can know anything about the external world. But the ordinary person doesn't have any evidence that he is a brain in a vat, so he can know things about the external world (assuming those things are true), without having to rule out the possibility that he is a brain in a vat.
Here's a useful way to think about this proposal. The skeptical arguments we've been looking at so far have basically gone like this:
1. You can't eliminate or "rule out" the possibility that you're a BIV, or in the Matrix, or deceived by an evil demon, or..., even though those are possibilities you know to be incompatible with P.
2. You can only know that P after you've "ruled out" all the alternatives you know to be incompatible with P.
3. So you don't know that P.
A Relevant Alternatives Theorist is a philosopher like Austin who denies 2. He says you don't have to rule out all the alternatives to P--or even all the things you know to be alternatives. He says you just have to rule out the alternatives that are somehow "relevant." According to Austin, this means: you only have to rule out those alternatives to P that you have some evidence to believe really are true.
So now we have two accounts of knowledge.
Descartes' requirement on knowledge: In order to know anything about the external world, you have to rule out the possibility that you're dreaming.
Austin's weaker requirement: If there's some special reason to believe you are dreaming, then the dreaming possibility is "relevant" and you have to rule it out; otherwise not.
Which one should we accept?
Stroud favors Descartes' requirement (see Stroud pp. 57-64). He denies that this requirement distorts our ordinary standards or requirements for knowledge.
To understand Stroud's argument for this, we need to distinguish between different things that go on when you speak.
Suppose that you and Barry are good friends who often tease each other about your clothes. At the moment, you're sitting in a dark classroom. Barry pokes his head into the classroom, looking for you, You know he can't see you in the dark, so you say to him "That's a real ugly necktie you're wearing." By uttering those words you've accomplished several things:
When you communicate things in the latter ways, without literally saying them, philosophers say that you conveyed those things or implied them or got them across, without literally saying them.
The contrast between literally saying something and merely conveying it suggests a second distinction, between:
Some statements can be true even though it wouldn't be appropriate to say them. For instance, suppose that you know that Mary is hard at work in the office. Then it's true that she's either hard at work in the office, or still home in bed. But if the boss asks you where Mary is, it wouldn't be appropriate for you to say "She's either hard at work in the office, or still home in bed." That's true but in that situation it wouldn't be appropriate to say it. Saying it implies that you don't know where Mary is, or that she often stays home in bed when she ought to be working, or something like that.
In addition, some things can be appropriate to say even if they're not, strictly speaking, true. Some philosophers argue that no physical surface is ever really flat, because it will always have microscopic bumps and irregularities. The same philosophers argue that no physical container is ever really empty, because it will always contain bacteria, specks of dust, and the like. But even if it's not strictly speaking true that the blackboard is flat, it can often be appropriate to call it "flat." Suppose you're flying a plane, and the engines fail, and the copilot says, "We need to find a flat place to land the plane," wouldn't it be appropriate to think of things like some empty highway in Kansas as being "flat"? Even if it's not strictly speaking true that they're flat. If you move all the goods out of the warehouse, wouldn't it be appropriate to call the warehouse "empty"? Even if there are still some old beer cans and cigarette butts laying around? It's often appropriate to use "flat" and "empty," even if, by the standards these philosophers insist on, it's not really true that the things we're talking about are flat or empty.
Now much of Austin's argument for his weaker requirement for knowledge goes as follows. He argues that it would normally be inappropriate to challenge a person's claim to know that P, on the grounds that she hasn't ruled out some remote possibility like dreaming. That's the evidence that Austin cites in favor of his account of knowledge.
Stroud and the skeptic accept Austin's point that this would normally be inappropriate. But they argues that it might nonetheless be a requirement for knowing P, that you rule out possibilities like dreaming. When someone satisfies Austin's weaker requirements, that may make it appropriate or reasonable to say that they know--and inappropriate to deny that they know--without yet really making it true that they know. This may be a case like "flat" and "empty." You never really know, because you're never really in a position to rule out all the alternative possibilities. But often it's appropriate to say you know, even though it's not strictly speaking true.
Stroud suggests that there will often be good practical reasons why we say that people "know," even though they haven't ruled out all the possibilities that he thinks, strictly speaking, they need to rule out in order to have knowledge. He illustrates this with the example of the airplane spotters (pp. 67ff.).
In that example, spotters are trained to watch planes flying overhead and classify them as Es or Fs. If the planes have features x, y and w, they class them as Es; if they have features x, y, and z they class them as Fs. However, nobody bothers to tell the spotters that Gs also sometimes have features x, y, and z. It's not important that they know this, since Gs rarely fly overhead, and if they do, it's not a big deal if they get mistaken for Fs. Plus from the ground it's basically impossible to tell Fs from Gs, anyway.
Now if a spotter sees a plane with features x and y, clearly that is not enough for him to know it's an F. It might also be an E. He has to inspect the plane to see whether it also has feature w or feature z. Stroud says that even if the spotter determines that the plane has feature z, though, that is still not enough for him to know it's an F. It might also be a G. He would have to rule that possibility out, too, if he's really to know that the plane is an F.
Of course, since no one told the spotter about the Gs, it may be reasonable for him to think that the place he sees is an F. It may also be reasonable for him to think he knows it's an F. But we do know about the Gs. Given what we know, what should we say?
Well, perhaps the airplane spotter is close enough to knowing it's an F for our present intents and purposes. This may make it reasonable for us to say that he knows the plane is an F. But Stroud thinks that even if we do say it, we'd wouldn't really count it as being true. We wouldn't really regard the spotter as knowing that the plane is an F. Any more than we regard him as knowing a plane is an F when he hasn't yet ruled out the possibility that it's an E.
Now, the skeptic sees our position with respect to his skeptical hypotheses (involving dreaming, evil demons, or brains in vats) as analogous to this. There may be good practical reasons why we don't try to eliminate, or even consider, far-out possibilities like the ones the skeptic discusses. But strictly speaking, those possibilities do have to be ruled out, if we're really to know the things we say we know. Just as the airplane spotter would have to rule out the possibility that the plane he sees is a G, if he's really to know that it's an F.
If this story is right, then the skeptic is not distorting our ordinary understanding of "knowledge." He's just relying on a phenomenon which is already implicit in cases like the airplane-spotter case. He's not changing the meaning of "knowledge," or imposing any new or artificially high standards. He's just pointing out that sometimes we think it's reasonable to say someone knows (we might say it about the spotter), even if in fact there are possibilities that person hasn't ruled out, and so it's not really true that they know. The skeptic thinks this is exactly our situation with respect to our beliefs about the external world.