Topics Knowledge

Question It would seem to me that I don't have to have ever witnessed a particular phenomena to be able to recognize evidence of it. For example, if I were to see a set of footprints in the sand, and on every left footstep there's a small hole, I might explain this finding by hypothesizing a person walking across the beach with a nail stuck in their shoe. Of course I understand that it could be explained in an other way, but if that was actually what had happened and I'd never before seen a person walking with a nail in their shoe then I'd have recognized evidence of something that I'd never seen before. But I have experienced footprints and nails before so perhaps I'm mistaken. My question then is, Is it possible to recognize evidence of something I've absolutely no experience whatsoever of? And what are the implications to the idea of sense data being evidence of the external world (if our only evidence of the external world is our sense data, how can we hypothesize an external world to explain such data when we've never experienced it...if you catch my drift)? Wouldn't we then be using the very thing that we're attempting to explain as the only evidence for the explanation?

Accepted:
February 9, 2008

Comments

Let's start with the more general question: is it possible to recognize evidence of something that we had no experience of before? The answer seems pretty clearly to be yes, since we've frequently found good reasons to believe in various such things. We have evidence for black holes, for example. We have evidence of the existence of various exotic particles. Although the details are complicated, the way this works can be explained in this sort of way: We're interested in whether there are Xs. We ask: what would we expect to see if there were? What evidence E would Xs produce? And how likely is it that we'd find E if there were no Xs? Assuming that the "prior probability" of Xs isn't too low (assuming, for example, that Xs would have a sensible place in our larger scheme of things), assuming that the probability of E given X is appreciably greater than the probability of E given not-X, finding E might well give us good reason to think there are Xs. Or at least, this is a fairly common kind of story about how we come to reasonable belief in such things. (It's a story, by the way, that's modeled on Bayes' theorem in probability.)

A little more directly: it's reasonable to believe in hitherto unexperienced Xs if we have some evidence that would be surprising if there were no Xs, but not surprising if there were.

With this in mind, we can read the second question as asking whether this sort of reasoning could lead us from sense data to reasonable belief in the external world. But before we try to answer that, we might want to pause and ask what it has to do with our actual situation. Are things really as the question presupposes? Are we really in the position of having only "sense data" as evidence, and trying to find a way to infer the existence of a hypothetical external world? It's not obvious that we are. A story that seems at least as plausible is that we experience the world. "Sense data" as the supposed internal objects that we're directly acquainted with are themselves hypothetical entities: parts of an eminently contentious tale. It seems at least as reasonable to say that when I look around, I see people, tables, chairs, chunks of cheese and such. I don't infer the existence of an outer world by some process of reasoning from "sense data." And though using my senses can provide data for various inferences (as when I see the light under the door and infer that John is in his office), we needn't reify by turning "sense data" into inner intermediaries.

This alternative story isn't without its puzzles. When I look into the sky, I may "see" a galaxy that no longer exists, as Bertrand Russell pointed out. But whatever we say about this problem, the best approach isn't likely to be by building a theory of knowledge on sense data.

Of course all this may be a way of making part of your point. If we were to view knowledge of the external world as inference from sense data, it would be tough sledding. Part of the reason is that the Bayesian story I told above couldn't get going. We wouldn't have any basis for saying that our "sense data" would be less likely if there were no external world.