There is a shadow over the second half of the course in thinking through which alternatives must one be able to rule out to count as having knowledge. (Recall from week 1, in the pub quiz, not being able to tell John from Edward ‘Jedward’ precludes knowledge that that one is Edward: a merely lucky true belief.)
In Barn Façade County, Henry’s inability to tell a mere façade from a real barn means that his lucky position by the only real barn in the county does not grant him knowledge that that’s a barn: a merely lucky true belief, again.
But in Swaledale, does the mere possibility of a barn façade undermine knowledge? (And does it matter whether Barn Façade County is real but across the Atlantic or unreal and a mere sceptical hypothesis?) Outside a philosophy class, we would say that innocent visitors can recognise barns on the hillside. They do know a barn when they see one. But what is the principle?
This challenge comes up in the discussion of reliabilism: how reliable does a method have to be to yield knowledge? The worry is: less than 100% and it looks to be mere luck. 100% seems unrealistically too high. But worse, it can’t just be a method which, by luck, is 100% successful. (Imagine that I can’t tell a coal tit from a great tit but by chance never see a great tit, common though they are.) So we might say 100% reliable in this world and nearby possible worlds (in which case, my inability to spot a great tit will score less than 100%) which is even more unrealistically too high.
Nozick offers a recipe. Knowledge has to be a true belief which also tracks the fact. Hence condition 3: If it were not true, you would not believe it. (Condition 4: And if it were, then you would.) So for the knowledge claim in question, go to the nearest possible world in which it is not true and see whether you would have believed it. If not: knowledge (in the actual world). If so, not knowledge (in the actual world).
It turns out that we do track ordinary claims. (In the nearest world in which I am not in Preston, I am at home ill and I thus do not believe I am in Preston.) But we do not track (the falsity of) sceptical hypotheses (in the nearest world I am a brain in a vat, I do not believe I am a brain in vat, the fact scepticism trades on).
So it seems that we do know that we are in Preston but we don’t know that we are not brains in a vat (and thus not in Preston). Since the sceptic argues like this:
1. I don’t know that not-H (the sceptical hypothesis)
2. If I don’t know that not-H then I don’t know O (what I ordinarily think I know)
3. So I don’t know that O.
Nozick seems forced to deny 2. He denies ‘epistemic closure’. And thus he is forced to hold, in DeRose’s phrase, ‘abominable conjunctions’ such as:
· I know that I have hands, but I don’t know that I am not a handless brain in a vat.
* “How about the world in which I am in bed in Kendal dreaming I am in Preston? Might that not be the closest possible world in which I am not in Preston and thus give a different answer? If so, in the closest world in which I am not in Preston, I still think I am in Preston and so condition 3 fails and I do not know, in the actual world, that I am in Preston.”
A really good objection. But what of the world in which one is asleep not dreaming of Preston? To be like this! would require the kind of dream we do not actually have and hence a greater departure from actuality. (I am not very happy with this answer since I have been forced to appeal to similarity between a pw and the content of a dream.)
NB: Nozick grants the sceptical hypothesis that we are dreaming in distinguishable dreams the status of something we don’t know to be false. Ie just like the brain in the vat case.
** “If we were a brain in a vat, we might not think that we were a human in a world. We might think nothing”
True, but that isn’t then a sceptical hypothesis / ringer which is experientially indistinguishable. So we can ignore it. The one we can’t ignore is the one in which we would still think we were humans in a world.