I agree with your objection, but I think there might be a way of modifying it slightly to make Graham's counterexample work:
Suppose that, for some reason, Jim knows that Trudy could not have broken the vase. Say, for instance, that he knows she has a severe phobia or allergic reaction to vases and never would pick one up in the first place... but Susan, not knowing Trudy at all, doesn't know this. It seems to me that it might make sense in that situation to say that Susan has far too little information to have real knowledge that Judy broke the vase (even though she believes so based on what she has seen enough to report that to Jim), whereas Jim does have enough information to say that he "knows" Judy broke the vase.
Of course, in that case, Jim doesn't "know that p on the basis of S's testimony", since he knows it inferentially - on the basis of S's testimony he knows that either Judy or Trudy broke the vase (maybe that's testimony plus inference too), and then on the basis of this and his other knowledge, he infers that Judy broke the vase.
But yeah, I agree with the other people that I modus ponens-ed where Graham modus tollens-ed. I think your example makes it somewhat clearer, because instead of one of these unlikely defeaters (an unknown twin) we have a perfectly ordinary defeater (color).
In the original article, Jim does know that Trudy was in the other room. But, as easwaran usefully points out, for it to be testimonial knowledge per se, it cannot be the result of an inference from that knowledge. Graham thinks that it can function as a restorer for Jim without being an inference and with all the epistemic merit of the belief deriving from Jim's trust in the testifier. I think he's wrong about that, so I tried to redescribe the case without that component. What is clear is that Graham wants Jim's knowledge to be based on the testimony and not on any inference.
Ah, I see your and easwaran's point. I guess I'm pretty skeptical about being able to separate pure testimonial knowledge from inferential knowledge. I think any time someone listens to a testimony, they automatically interpret it in the context of the knowledge they already possess. (Some inferences are made just in deciphering the meaning of the statements made, for instance.) So it's hard to say what knowledge is gained on the basis of the testimony versus inference. But other than that, I agree. To the extent that it makes sense to talk about purely testimonial knowledge, it looks like the original principle holds up.
I think that's all correct. You'd be surprised by how many philosophers think that a sensible notion of pure testimonial knowledge, abstracted away from all inference, can be articulated. I am dubious of the possibility.
Suppose that, for some reason, Jim knows that Trudy could not have broken the vase. Say, for instance, that he knows she has a severe phobia or allergic reaction to vases and never would pick one up in the first place... but Susan, not knowing Trudy at all, doesn't know this. It seems to me that it might make sense in that situation to say that Susan has far too little information to have real knowledge that Judy broke the vase (even though she believes so based on what she has seen enough to report that to Jim), whereas Jim does have enough information to say that he "knows" Judy broke the vase.
Reply
But yeah, I agree with the other people that I modus ponens-ed where Graham modus tollens-ed. I think your example makes it somewhat clearer, because instead of one of these unlikely defeaters (an unknown twin) we have a perfectly ordinary defeater (color).
Reply
In the original article, Jim does know that Trudy was in the other room. But, as easwaran usefully points out, for it to be testimonial knowledge per se, it cannot be the result of an inference from that knowledge. Graham thinks that it can function as a restorer for Jim without being an inference and with all the epistemic merit of the belief deriving from Jim's trust in the testifier. I think he's wrong about that, so I tried to redescribe the case without that component. What is clear is that Graham wants Jim's knowledge to be based on the testimony and not on any inference.
Reply
Reply
Reply
Leave a comment