Pages

Friday, June 19, 2009

FMER, Leuven, June 10-12, cont'd

Day 3, after lunch

Lara Buchak (joint with Branded Fitelson who couldn't make it to the conference) -Is it rational to have faith? - Lara was trying to cash out what having faith commits one to, and on the analysis she presented faith in X requires that one not actively look for further evidence for the truth or falsity of X. This move seems to collide with the expected utility theory of rationality. Then, she argued that the claim that expected utility maximisers should always perform cost-negligible experiments neglects the phenomenon of risk aversion. It turns out that for individuals who take risk into account in a certain way it is sometimes rational to refrain from gathering further evidence.

One issue was raised, if I remember well, by Joshua - namely, if this is the way you understand faith, Richard Swiburne doesn't have faith, for he actively looks for evidence pertaining to the truth of religion. Perhaps (now that I think of it) this can be circumvented by saying that faith in X would require one not to actively look for further evidence for the falsity of X (it depends how one construes Swiburne's thought, but one way to see this is to think that he does actively look for evidence in support of religion, but doesn't actively look for evidence against religion).

Another issue is that the notion of evidence in/against religion is quite elusive. Lara used an example of prayers: pray for something and see if you get it. But it seems that (at best) what you're testing this way is the conjunction of some of your religious beliefs and the claim that your will agrees with the will of God). Also, what counts as test or evidence for or against religion is highly theory-dependent. You can go Swinburnian about this and think that no actual event whatsoever is evidence against religion, because given certain considerations, everything that is happening should be happening if (his version of) theism is true. You can go more Tooleyan about this, and count every event that prima facie should not happen as evidence against the truth religion. Perhaps it's just me being confused, but I don't think we have a good understanding of a test that both a theist and an atheist would agree upon, so that the cost of its performing is negligible (given the negligibility requirement, for instance, Hick-style die-and-see-what-happens is out of the question).

The conference started with a slightly apologetic talk by Swinburne. It ended with a rather atheistic talk by Herman Philipse. He gave a series of short arguments against the claim that there is a C-inductive argument from the Big-bang to the existence of God. Briefly,i f h is theism, e is the occurence of big bang, and k is tautological background knowledge, Swinburne argues that Pr(e|h&k)>Pr(e|k). The first point that Philipse makes is that given that God would want to create humans and the fact that given the cosmic singularity, the probability that it will result in there being humans is quite low, it seems that Pr(e|h&k) is very low. Another point raises pertaining to Swinburne's claim that Pr(e|k) is very low. Since Pr(e|k)=Pr(e|h&k)Pr(h|k) + Pr(e|~h&k)P(~h|k) we need to know the prior probability Pr(e|~h&k) and Pr(~h|k). Philipse argued that Pr(e|~h&k)> Pr(e|h&k). He also attacked Swinburne's use of simplicity criterion. Since this was directed against Swinburne who was present, quite an interesting discussion followed.

As for Bayesian-style arguments for/against God's existence, I'm rather sceptical. The problem is, even if the math adds up, they all rest on primitive assessment of probability of things like "big bang occurs" relative to the existence of God, or relative to the negation of his existence, and many other probabilities of this sort. When asked questions like "what's the probability that intelligent beings like humans exist given the hypothesis that there is no god and no multiverse?" or "What's the probability that Big Bang occurs given the hypothesis that God exists", I'm just inclined do say: I have no idea. I would love it if (a) someone explained to me the notion of probability at play, and (b) showed me how on this notion of probability, the probability claims involved can be assessed without hand-wavy and practically untestable extra assumptions. Perhaps I'm just a frequentist and haven't seen too many worlds being created. My bad.

No comments: