I will reply to this a little more when I get some free time, but another paper which might be put into proposition 2 is here: < http://www.svcl.ucsd.edu/publications/journal/2009/pami09-Sal.pdf >. This, along with Bayesian surprise by Itti and Koch at USC, does a good job of showing how Bayesian models of focus-of-attention mechanisms in the mammalian brain actually perform well computationally and experimentally.