Friday, June 19, 2009

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

Day 3, before lunch

Edward Wierenga talk titled Developing Molinism employed fairly complex modal stuff (you know, actuality, counterfactuals and all that) to help formulate Molinism, the view that God has a knowledge of propositions that are intermediate between being necessarily true and independent of God's will or creative ativity, and contingently true propositions dependent on God's will. These are contingent true propsitions not dependent upon God's will (in the intended interpretations: propositions about future but free actions of men).

This knowledge is often taken to be a knowledge about certain counterfactuals (like "If Adam were placed in the Garden of Eden, he would freely eat the forbidden fruit"). This knowledge would assist God in devising the world so that it is the best world possible without interfering with human free decisions. The technical problem is that it's difficult to find right truth-conditions for counterfactuals of this sort which satisfy all the desiderata. Wierenga first discussed his original view (that Plantinga's conditionals of world actualizations can do the job) and criticized it, and then presented another suggestion, employing tense considerations.

Paul Bartha talking about Many gods, many wagers, discussed in detail the many-gods objection against Pascal's wager. He then argued that given the evolutionary stability condition on probabilistic reasoning (roughly, the condition is that after making a bet, no further probability considerations of the state of affairs after making the bet will make you change your mind) the many-god objection doesn't raise any difficulties that the classical version of Pascal's argument already encounters.

David Glass - Can evidence for design be explained away? - An obvious way to counter a design argument is to provide an alternative explanation. For instance, evolution theory is taken to render inconvincing certain design arguments. The problem is, certain version of design arguments are compatible with alternative explanations - why accept both? Well, don't if there's no need to do that! The technical question, however, is when one explanation is good enough to render the other redundant. Are there cases where it is better to accept both explanations than only one of them?

David addresses these issues within the Bayesian framework. Even if two explanations are marginally independent they typically become negatively dependent when one conditions on the evidence they explain. So, if one explanation is found to be true, this lowers the probability of the other explanation. There are, however, two important possible outcomes. Say a design hypothesis D has a certain prior probability Pr(D). Next, suppose it receives confirmation from evidence E so that Pr(D|E)>Pr(D). Then, we find out that an alternative explanation A is true. This undermines Pr(D|E&A). In the first case, Pr(D|A&E) is not higher than Pr(D), and so the initial confirmation of D by E has been completely negated. In the second, Pr(D|E&A) is higher than Pr(D) but lower than Pr(D|E). Given than only the first kind of outcome counts as explaining away, it turns out that it is very difficult to come up with an alternative theory that completely explains away the evidence for design.

No comments: