Inferring and Explaining

23 superfcial appearance of being clear and iden- tifable. Models of knowledge that substitute criteria for epistemic justifcation must be pre- pared to state some new criterion for distin- guishing unfounded belief from a promising theory and from established knowledge. Te contemporary literature ofers many intrigu- ing possibilities—some highly formal and some quite commonsensical—but none that have won anything approaching consensus. I suggest that we understand the idea of epistemic justification in terms of evi- dence. Te things that we know are those true beliefs for which we have very, very, very good evidence—what a lawyer calls proof beyond a reasonable doubt. Good evidence is something that we are all familiar with and something that we can learn to reliably spot. I will be ofering in the chapters to follow a model of—or a kind of formula for testing for—good evidence. I hope to convince you that this model captures almost everything we care about when we assess the quality of a person’s evidence or for that mat- ter, their claims to knowledge. Let’s transform the standard analysis of knowledge in light of all this into the following: J knows P only if: i. J believes P . ii. P is true. iii. J has exceedingly good evidence for P . An Unsolved Problem If you were reading very carefully, you may have noticed a slight diference in the way I stated the standard analysis of knowledge at the end of preceding section and the section immediately before that one. You are all smart enough to see the obvious change in condi- tion iii , but can you fnd the other diference? Te way the philosophic tradition has defned knowledge is to articulate necessary and suf- fcient conditions for knowing something. Te standard analysis of knowledge claims that the three necessary conditions are, taken together, sufcient for knowing something. In my state- ment of a “transformed” analysis, I wimped out a bit. I claimed that my three conditions were all necessary—that’s what the “only if ” signifes— but I lef it open whether the three conditions were sufcient. Here’s why. Consider the following little thought experi- ment. My wife and I have spent the last hour collaborating on our special spaghetti sauce. Just as we are getting ready to serve dinner, we discover that we are out of Parmesan cheese. We divide responsibilities—she will toss the salad and serve dinner; I’ll make the emer- gency run to the store. While at the store, I meet a colleague doing research in contemporary epistemology—she wants an example of knowl- edge. I suggest that I know there is a spaghetti dinner sitting on our dining room table right now. And as luck would have it, it’s true that a spaghetti dinner is on the table. I believe it, it’s true, and I’m justifed in believing it. All is well. Well, maybe not. Afer I lef, our German shep- herd, Guido, got rambunctious and knocked the pot of simmering spaghetti sauce on the dirty kitchen foor. My wife considered violence against the dog, but before anything could hap- pen, a neighbor arrived with a pot of lefover spaghetti sauce, announcing that she was leav- ing on vacation and it would surely spoil before she returned. Tus the spaghetti sauce that the ConCePt of knowledge