Reality Conditions

Monday, October 02, 2006

Pictures of the week, links of the week


Not much time or inclination to write a proper post, so here are some pictures of Vancouver sightseeing, and interesting links for you to check:








View of Vancouver; Capilano Suspension Bridge; Victoria (Vancouver Island); Whale Watching near Vancouver Island.

Now the links:

-Cosmic Variance discussion on the best popular science books on quantum mechanics. I'll have to be unoriginal and go for Feynman's The Character of Physical Law and QED, the merits of whic I don't need to promote here. Two more idiosyncratic preferences of mine are Alberto Clemente de la Torre's Física cuántica para filo-sofos and Bernard D'Espagnat's In Search of Reality. The first one is an excellent introductory explanation of quantum mechanics and its philosophical problems, probably untranslated into English. The second one is not really "popular science" but a rather heavy (if very interesting) philosophical discussion, but contains a superb explanation of the Bell inequalities.

-Jason Rosenhouse explains how the World Chess Championship is jeopardized by Vladimir Krammik's frequent excursions into his private bathroom.

-Brandon has an excellent post explaining why "Bayesian arguments" can't be used to prove anything, let alone God's existence, even if Bayesian epistemology is sound. The only thing one could prove with these arguments, he says, is that one is being rational in one's beliefs. To my mind this is still a large problem in Bayesianism (at least in its "pop" version), because it seems that almost any beliefs could be justified as rational by it.

-Doesn't this remind anybody of this?

Helen: You've got to lead our protest against this abomination!
[shows newspaper article]
Marge: Mm, but that's Michelangelo's David. It's a masterpiece.
Helen: [gasp] It's filth! It graphically portrays parts of the human body,
which, practical as they may be, are evil.

-On the Friendly Blogs Circle: my nonintelligent friend re-creates the Sargent Pepper's cover; my sinful friend imagines how The Da Vinci Code would have been directed by other filmmakers; my cynical friend lists ten hateful things; my rock star friend doesn't like to be called "sir"; my suspicious friend reviews United 93; and my tsumani-rowing friend seems to have quitted after only three posts.

-Now I've seen it all: Agatha Christie anime.

4 Comments:

  • not quitted yet... just taking a break.

    By Blogger alejo, at 8:18 PM, October 02, 2006  

  • To my mind this is still a large problem in Bayesianism (at least in its "pop" version), because it seems that almost any beliefs could be justified as rational by it.

    This is a good point. We can make a distinction between conditional rationality and rationality as such. You are conditionally rational if, when you make a new assessment of the evidence and probabilities, you change or don't change your beliefs in a rational way. It's 'conditional' because it describes how rational you are being given the assessment. But you are rational as such only if your assessments are also rational. Even assuming that Bayesian epistemology is true, the most a Bayesian argument could do is show that you are conditionally rational.

    It's possible, of course, that your assessments could be rational but the conclusion is horribly, terribly, utterly, and (to other people) obviously wrong, and there are crazy conditions and circumstances under which this might be possible (for instance, even a completely rational person will end up with bizarre beliefs if he is systematically lied to by everyone he knows.). Bayesian epistemology, as a general account of all possible rational belief change, has to be general enough to allow for cases like that, no matter how rare or bizarre. But that's another reason why 'Bayesian arguments' are generally no good.

    By Blogger Brandon, at 1:07 AM, October 03, 2006  

  • Brandon,

    Thanks for the comment. I see your point, but I wonder whether even being conditionally irrational is a real possibility. Of course it is possible "in theory"; if you start believing (with some degree) "p" and "p implies q", and find out that "not q" and not update your probabilities, you are being conditionally irrational by definition. But I doubt whether this has any relevance to the messy way we form and change our beliefs in practice. I suspect that after a realistic belief change one might always spin a Bayesian story to show that it was rational. For example, it may be that before finding the new evidence the subject had not even considered the conditionals relating the old beliefs to the new evidence and of course had not asigned them any credences, so it is easy to asign post facto the numbers that make the desired conclusion come out. (E.g.: if the subject decided to not change his mind about p, this is because his degree of belief in p => q was low. The assumption of conditional rationality is unfalsifiable in Popper's sense.) I'd say that something similar is going on in these arguments for God's existence; the numbers seem rigged to obtain a pre-determined conclusion.

    Of course in a scientific context it may be that all the probabilities are stated explicitly at the outset, and then this criticism wouldn't apply.

    By Blogger Alejandro, at 8:36 AM, October 03, 2006  

  • I pretty much agree; it's one of the reasons why I'm not a Bayesian. Bayesians, I think, would deny that you can spin the numbers after the fact -- the rational agent for them is one who actually thinks in a way described by Bayes' Theorem rather than someone who appeals to Bayes' Theorem. But, as you say, it's very difficult to see how this can be a good account of most of the ways of approaching difficult questions, since they can be very messy, even when they are not obviously irrational. And I think your suggestion in the last sentence of the first paragraph is quite right: the Bayesian account does seem uncomfortably conducive to this post facto rigging.

    By Blogger Brandon, at 12:46 AM, October 05, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home


 
/body>