This definition seemed to reflect the common use of the term through history. However, in more recent times, Edmund Gettier provided counterexamples showing that if a belief relies on a false premise, then it won't commonly be regarded as knowledge. For example, you may believe the correct time by looking at the clock on your wall. But your belief doesn't count as knowledge if your tacit assumption that the clock is working is mistaken.
My amendment to Plato's definition is to recognize that for a conclusion to be known, any essential premises also need to be known. This is a slightly stronger formulation than William Lycan's excellent no-essential-false-assumptions analysis.
My proposal for the definition of knowledge: For S to know that P is true means that:
(1) P is justified
(2) P is true
(3) P is believed
(4) S knows that the essential premises of P (if any) are true
One intuition for this definition is that knowledge is constructed hierarchically like building blocks. If a low-level building block fails, then building blocks that sit on it are not supported and so also fail.
There's an interesting example from someone after Gettier (Kripke perhaps) of a farmer who goes to his field and looks at what he thinks is his cow (though its actually a bush) and based on this knows that his cow is in his field. Now the cow happens to actually be in the field (hidden from the farmer's sight by a tree). So the condition (2) is met, but not through (1).
ReplyDeleteIt seems a little like a legal framework, the general principle (say, no theft) can be expressed by a few conditions, but the endless specific situations lead to endless caveats or, better, interpretations. Especially since knowledge is (IMO) more like fuzzy or Bayesian than absolute. Knowledge is confidence, not certainty.
I think the single intuition that Gettier-style examples are exploiting, including the "farmer's cow" example, is that one or more premises in the knowledge claim fail the JTB test, thus undermining the knowledge claim. So condition (4) incorporates that intuition by using deep recursion to impose a JTB test on every premise.
DeleteWith the "farmer's cow" example, the farmer's conclusion depends on his false premise that "he is looking at his cow". This causes condition (4) to fail and so the farmer's knowledge claim also fails.