Thursday, February 9, 2012

A method for testing definitions

I have a new method for testing definitions. Read a definiens to someone, out of context, and ask her what she thinks the definiendum is. If she doesn't come up with something pretty close to the definiendum, you've got reason to think the definition is bad.

One can also do this as a thought experiment, though it's probably less effective that way. What does "justified true belief with no false lemmas" define? Answer: nothing other than justified true belief with no false lemmas. (Maybe you were trying to define knowledge?) What does "Sex between two people at least one of whom is married and who are not married to each other" define? Answer: adultery. (Right!)

7 comments:

Alexander R Pruss said...

Here's another example: "What do we call rules that a bunch of intelligent and rational delegates would come to a consensus on after their minds have been cleared of knowledge of their comprehensive views and idiosyncratic preferences and needs?"

The tempting answer, in light of truisms about how committees operate and in light of how stultifying such mind-emptying is likely to be, is "Stupid rules".

And there should be no temptation to answer: "Rules of justice."

Kenny Pearce said...
This comment has been removed by the author.
Edgar Foster said...

In my opinion, a number of factors come into play when one is trying to infer or derive a definiendum from the definiens. Would not the hearer or reader's presupposition pool or personal lexicon (idiolect) also make a difference? After all, before I studied Plato or epistemology, there would have been no way that I would have defined "knowledge" as "justified true belief" etc.

Kenny Pearce said...

CORRECTED:
Ever played the board game Blurt? They have cards with definitions of everyday words from Webster's on them, and someone reads a definiens, and the other players all try to be the first to call out the definiendum. The funny thing is, some of these very common everyday words can be very difficult to guess. I'm not sure if this should be taken as evidence that your procedure is bad, or evidence that Webster's is not a good dictionary.

Alexander R Pruss said...

Kenny:

Good point, that. Maybe, though, it is evidence against a definition when people plausibly guess something else as the definiendum?

Alexander R Pruss said...

Edgar:

Maybe that's a sign of the fact that knowledge isn't justified true belief, or anything like it?

Kenny Pearce said...

It occurs to me that perhaps the reason many of the cards in Blurt are difficult to guess is that they are words usually defined ostensively. For instance, they define 'staple' as 'a U-shaped metal fastener.' I would never think of defining 'staple' as anything other than 'one of those things.' So perhaps your criterion works for more abstract terms, like 'knowledge' or 'adultery' or 'justice', that have to be learned by a different mechanism than words like 'staple'.