Biased evidence assimilation under bounded Bayesian rationality

<h1>Biased evidence assimilation under bounded Bayesian rationality </h1>

<br><a href="..">Brendan O'Connor</a>
<br>M.S. Thesis, Symbolic Systems
<br>Readers: Jonathan Bendor and James McClelland
<br>September 2006

<h2>Abstract</h2>
I explain evidence assimilation bias as the result of agents trying to maintain cognitive consistency. This can be interpreted as a boundedly rational 
inference method -- local search for a maximally likely world model. Given 
a sufficiently complex network of beliefs, such an approximate Bayesian can 
display systematically non-Bayesian behavior. These arguments are first 
sketched via connectionist Hopfield networks, in line with previous psychology literature, and then illustrated and analyzed in more detail with 
probabilistic graphical models -- Bayesian networks and Markov random 
fields. 

<P>
<a href=msthesis-revised1.pdf>PDF</a>