Possible Duplicate:
Card doubling paradox
I came across the following preferences paradox:
Suppose you have two identical boxes $A$ and $B$. One of them contains an unspecified number of coins. Another contains twice this amount. The first one you touch is the one you open. Which box do you choose?
Suppose I choose Box $A$. Suppose $E(A) = x$. Then the value of box $B$ is $x$ or $\frac{1}{2}x$. Thus $E(B) = x(\frac{1}{2})+(2x)(\frac{1}{2}) = 1.25x$. One can do a similar analysis when starting with Box $B$. It seems that both both boxes should have the same expected value a priori. Is it wrong to assume that $E(A) = x$ or $E(B) = x$? The value of these boxes may be $x$, but that doesn't imply that their expected values if $x$?
Or maybe our definition of expected value if flawed? Should we look at log utility? This seems similar to the Monty Hall Problem.