I'm with Sumitra on this issue.  
   
Insisting that economics is about how agents make choices may widen its scope in some
respects, but  it (i) narrows its scope in other ways (i.e. social phenomena not evidently
grounded in choice are out of bounds); and (ii) narrows the range of legitimate methods
(if you're not analyzing a problem in choice theoretic terms, you're not doing economics).
Fred's definition would, for example, cut Marx, or a good part of what he was about, out
of the discipline; Sraffa.
   
Obviously most of what constitutes modern economics IS indeed about how agents and
communities make choices; and, equally obviously, economists of all stripes ought to be
interested in such issues. But why should that be the only lens through which economists
look at social phenomena?
   
What's a non-material choice, anyway?  
   
Gary