Barry Hall has done considerable work on mutations in bacteria which
appear to occur at a higher rate when they are "useful" than when they
are not. His papers include a very thorough consideration of all
the reasons this might be happening (including boring ones such as
experimental error) and tend to come to the conclusion that we don't
know why, or how, but the phenomenon is reproducible. It appears to
happen only in stationary-state cultures (where the bacteria have
grown up to their population max and are sitting around waiting for
food and slowly dying).
Several years ago there was an article in _Scientific American_ by
Allan Wilson about the possibility that intelligence gives an organism
access to evolutionary possibilities that are not open to less
intelligent organisms: he talked mainly about birds, and the fact
that the ability to learn to, say, open milk bottles with your beak
is a useful precondition for evolving a beak that is good at it.
You might take a look at this article's reference list.
Mary Kuhner mkkuhner at genetics.washington.edu