This was going to be a comment on the X-Phi post, but it became too long.
Suppose baseball bats come to be made of a special wood so that, as we’d be prone to say, how fragile they are varies greatly with how humid it is: when the humidity is very low, they become quite brittle, and break much more easily; while they become very resistant to breaking when the humidity is high. Managers would find themselves saying things like this to their teams: “Remember that your bats will be very fragile tonight. The humidity is very low.” Here the manager uses “fragile” to describe how easy it is to break the bats right now (or at least that evening), in the current conditions.
But there would also be a quite different, but also very natural, way to use “fragile.” The bat-maker at the factory might say this, after making an especially good batch of bats: “Today’s batch is excellent. They’re much less fragile than the bats we made yesterday.” The bat-maker doesn’t seem to be describing how easy it would be to break the bats right now, given the humidity of the air currently around them, but rather how easy it will be to break them under a variety of different humidity conditions. (We may suppose that today’s batch of bats is still at the factory, where the humidity happens to be extremey low, so that the bats would be very easy to break right now, while yesterday’s bats have already been shipped out to various high-humidity locations, so they are right now hard to break, though today’s bats would be harder to break than yesterday’s bats if the both sets were exposed to the same humidity conditions. Still, what the bat-maker says seems very natural.)
I think a similar thing happens with our descriptions using the likes of “confident.” Stakes are (at least roughly) to “confident” as humidity is (in our imagined situation) to “fragile.” I don’t know if this use is ultimately correct, but it sure is at least tempting to use “confident” in such a way that we typically become less confident of the truth of various propositions as we enter high-stakes situations. On this use, confidence seems to track (though I doubt it’s reducible to) certain “local dispositions,” we might say, to (un)confident behavior: Right now, in the subject’s current practical environment, is the subject disposed to act on the supposition that p is true without checking further into the truth of p?, is she disposed to flat-out assert that p should it be conversationally relevant?, etc. On this use, as the stakes go up, we tend to become less confident of the truth of various propositions – and this is revealed by such facts as that we will seek to verify certain propositions before acting on them, whereas we would have just acted on them without further checks in low-stakes situations.
But there’s another way to use “confident” – a way in which our “confidence” levels typically don’t vary much as we move between low- and high-stakes situations. On this use, “confidence” seems to track different disposition to (un)confident behavior. It describes how disposed we are to act on p, flat-out assert that p, etc., in a variety of different practical environments, not just the practical environment we happen to find ourselves in at the time in question. On this use, our confidence typically is not reduced as we move into high-stakes environments. Rather, we say that the same level of confidence that we had when the stakes were low is no longer confidence enough to produce confident behavior, given the current high stakes. (It’s of course *possible* for one’s “confidence” to be reduced by high stakes even on the current, more stable, use of “confident.” If the situation has shaken me up so badly that I won’t even produce confident behavior once I again find myself in low-stakes situations, then my confidence has been reduced on both ways of construing “confidence.” But it’s unusual, I think, for one’s confidence, on the second construal, to be so significantly affected by high stakes. That’s why I say that on this construal, our “confidence” is *typically* not affected much by stakes.)
That second use is how I usually find myself using “confident” and its cognates, at least while doing epistemology. I find it very natural to talk as Ram does, for instance, in comment #71 of the X-Phi comment thread:
Now you ask me to bet $10 at 1:1 odds that Obama will win. Fine by me. But now suppose you ask me to bet my life savings at 1:1 odds that Obama will win. No way! Given the massive cost to me of being wrong, I better go out and gather more evidence before I make such a bet! And that’s NOT because I have less than .8 confidence, or because I ought to have less than .8 confidence. It’s because I’m not willing to take a .2 chance of losing my life savings!
But I’ve found that others seem to think of confidence in the first way. This has come up in discussions of my old bank cases. Toward the end of the high-stakes case, I write:
Remaining as confident as I was before that the bank will be open then, still, I reply, “Well, no [I don’t know the bank will be open]. I’d better go in and make sure.”
Most readers have (or at least express) no trouble with that description. But some are baffled, and react in something like this way:
How can it be that he is “remaining as confident” as he was before?!!! Before he was aware of the high stakes of his situation, he was happy to just assume that the bank would be open the next day, to flat-out assert that it would be open, and to act on the assumption that it would be open, all without feeling any need to “make sure” he’s right. It seems that, even without further checks, he took himself already to be sure. But once he’s aware of the high stakes, all of that seems to change, and we’re now told that he is going to do something he’s willing to call “making sure” that the bank will be open before he’ll go back to the confident attitude he had before he was aware of the high stakes of his situation. So it certainly seems as if his appreciation of the high stakes of his situation has diminished his confidence quite considerably indeed. So I’m not sure how to make sense of the statement that his confidence is not diminished.
I think these baffled readers are thinking of confidence in the first, less stable way — the construal on which confidence levels do typically vary greatly as subjects move back and forth between low- and high-stakes situations.
Building on Ram’s betting scenario, we can bring out the differences between these two ways of using “confident.” Suppose Ram & I are equally rich/poor, that we have roughly the same evidence concerning whether Obama will win, that we both become more hesitant to bet that Obama will win as the stakes get very high, but that we both believe Obama is likely to win. I interpret the evidence as being more favorable to an Obama victory than Ram does, and in keeping with that, I will accept every bet in favor of Obama winning that Ram will accept, but there are many such bets that I’ll accept that Ram won’t. As it happens, Ram and I have each just been offered bets — but very different ones. Ram has just been given the chance to bet $100, even odds, that Obama will win. I have the same opportunity, but the bet I’ve been offered is for $200,000. Like Ram, I don’t have the kind of money involved in the bigger bet; to lose would be to go pretty deeply into debt. Now, look at our current behavior as we respond to the bets we’re offered. Ram immediately (and very confidently!) replies (just as I would if I were offered the $100 bet): “Yes, I’ll take that bet! Get ready to pay up, sucker!” Meanwhile, I say, very hesitantly (just as Ram would): “Umm, I don’t know. It’s very tempting. Can I take the rest of the day to decide whether to accept this bet?” Being told that I can delay my decision, I spend the day looking very carefully at all the evidence I can get my hands on, before deciding whether to bet that Obama will win. Meanwhile, Ram has just (confidently!) accepted the bet without doing any further checking. (His “sucker” has offered him the chance to take the rest of the day to decide; Ram brashly shot back (as I would have in his situation): “I don’t need any more time to decide!”) Which of us is more confident that Obama will win? If you’re thinking of confidence in the first way, you’ll say Ram is more confident that Obama will win (as well as being more confident about what he should do). After all, he just immediately took the bet. Look how confident he is! Listen to his brash and confident trash-talking! Look at Keith, worrying & checking, checking & worrying! He sure sounds very uncertain of himself: “Umm, I think,” “I guess”! But if you think of confidence in the second way (as I’m usually inclined to), you’ll say I’m more confident than Ram is that Obama will win (though less confident about what I should do), and the fact that Ram is displaying more confident behavior than I am is not because he is more confident than me that Obama will win, but because my level of confidence that Obama will win, while even a bit greater than Ram’s, is not confidence enough to produce confident behavior, given the high stakes of my situation.
I’m not saying here that both of these uses are correct – or that they’re not both correct. But there certainly seems to be a lot of potential for “slippage” here. Which is why I worry about how well I understand the self-reports of subjects in psychological studies about how “confident” they are, or respondents’ judgments about how “confident” some subject rationally should be in a certain situation. On the first use of “confidence” and the like, our “confidence” levels ought to vary greatly as we move between low- and high-stakes situations; on the second construal, they shouldn’t.