Pat Gunning wrote: 
> What more sensible means would you suggest for building 
> models of interacting human beings than to use logic and 
> to assume that the beings you are modeling also use logic? 

Although I sympathize with Pat's point that such methods are 
useful, his (rhetorical?) question goes much too far.  There   
are many other frameworks for thinking usefully about human  
interaction.  I would include for example much social 
psychology (e.g., I do not think the "logical point of view" 
helps us improve our understanding of the Milgram experiment
http://en.wikipedia.org/wiki/Milgram_experiment ).
Of course economic imperialists will try to propose tastes 
and beliefs to cram this into their preferred modeling 
framework.  Stigler and Becker, IMO, tore that approach 
apart thirty years ago.

Cheers,
Alan Isaac