Easiest Gettier Cases to Deny

Some of the recent discussion of the Gettier problem has suggested that we simply abandon the intuitions that Gettier cases are not cases of knowledge. The plausibility of doing so depends on how obvious it is, regarding a particular case, that it is a case of knowledge. On this score, some of the examples in the literature are less plausible than others, and it might be interesting to find out what epistemologists think are the least compelling.

Among the least compelling, as I see it, are some of the social cases, such as Harman’s Jill/assassination case. Harman questions how Jill could know since she lacks some evidence that all those around her have. Maybe that’s right, but it must also be noticed that the evidence all those around her have is concocted and misleading evidence. I don’t find this point compelling against the case, but it makes me suspicious of it.

Lycan has an explanation in Judgement and Justification for why he is suspicious of both this case and the fake barn case, but I don’t remember the explanation and don’t have the book here at home to include the explanation. But it’s worth noting that Bill views both of these cases as not compelling. Are there other cases that epistemologists view suspiciously?


Comments

Easiest Gettier Cases to Deny — 7 Comments

  1. The fake barn case can be less than compelling, I find, if I’m imagining that while there are many fakes around, the agent (Henry, usually) hasn’t seen & been taken in by any of the fakes. So, when presenting the case, I usually include that Henry has seen & been fooled by many fakes before now encountering the only real barn in the region. With that added, it becomes pretty compelling, at least to me, that Henry doesn’t know. & if the fakes were all good enough, Henry’s still justified in his belief.

  2. I don’t know whether people would count them as Gettier cases, but I think that where X is a standard Gettier victim (e.g. someone who inferred the true belief that p from the false justified belief that q) and X tells Y that p, it can be very easy to think that Y knows that p. It’s particularly easy if the focus of attention isn’t on p itself, but on something Y inferred from p. For example, X has a Gettiered belief that the meeting starts at 12, X tells Y that the meeting starts at 12, Y sees it is 12.15 and comes to believe she’s late for the meeting. I think this is knowledge, and hence she knows the meeting starts at 12. But I’m probably in a minority here.

  3. Keith, now that you mention this, I see the point. I’ve often added the dimension you mention when meeting resistance in students, but hadn’t connected it with some skepticism about whether it was a gettier case nor that I was sympathetic!

    Brian, I think there are cases like this, too. Maybe not where the Gettiered belief itself is what you believe on the basis of testimony, but at least where you use the Gettiered information for inferences. For example, suppose I rationally believe that you are either in Thailand or Berkeley this week, and i’m told by a reliable source who’s gettiered (let’s make it one of your colleagues) that you’re at Stanford giving a talk. I also know that there was some issue about whether you were going to go to Stanford or Berkeley, and rationally believe that early last week it was settled that if you weren’t going to Thailand for fun, you’d go to Berkeley to give a talk. Then I know that you’re not in Thailand for fun, even though the testimony isn’t knowledge and even though I don’t believe what you’re colleague is telling me.

  4. Of course, just how deniable you will find any given Gettier case may depend significantly on your ethnic background *ahem* — but putting that aside for a moment, I find that students have a sort of ‘naturalness of inference’ bias concerning whether or not a given case is really an instance of JTB. Namely, they sometimes deny that the agents in the original cases really have justification, because the agents draw such odd inferences in those cases. They will acknowledge the validity of disjunctive weakening, but still deny that in these instances that inference form was justification-transferring. As a result, I tend to use existential generalization cases, instead of disjunctive weakening cases, in teaching Gettier these days.

  5. Good point, Jonathan, a kind of reservation about classical logic! Very nice, and I can remember being suspicious of that too when I first read Gettier’s own cases: the first one about the coins in the pocket was straightforward, but the brown in barcelona case didn’t make any sense initially.

  6. It doesn’t just have to be suspicion of classical logic (or indeed logic that licences v-intro). It could be that students are prima facie willing to accept Crispin Wright style worries about transmission of justification, even if they think the argument is truth preserving.

  7. My pocket hypothesis about the students’ hesitation in these cases is less, er, philosophically laden: the inferences in question are just very unmotivated, ‘unnatural’. There’s a sort of “serves him right for drawing such a weird inference from a premise that turned out to be false!” flavor.

    Here’s one possible test: give one set of subjects a bare “Brown in Barcelona”, and another set a slightly elaborated version where the agent’s interest in the disjunction is made psychologically plausible. E.g., the agent is trying to determine whether the disjunctive antecedent of a conditional is fulfilled, like, “If Brown is in Barcelona or in Brest-Litovsk, his mail should be forwarded to Division X3 for further processing.” I predict that subjects in the second condition will be more likely to attribute justification to the agent than subjects in the first. This is just a rough idea, of course, and at a minimum runs a risk of interference with the phenomenon Jon mentioned in comment 3 (even, I think, if the experiment is clearly asking about the disjunction itself, and not the consequent of the conditional).

Leave a Reply

Your email address will not be published. Required fields are marked *