Expert reasoning typically has special characteristics:
- Use of specialized representations
appropriate to the domain and specialized
problem-solving methods based on those representations.
- Translation of observables into specialized
terminology and representations (e.g., ``person has turned blue'' into
``patient is cyanotic'').
- Use of empirical rules of thumb (e.g., ``to
blow out a tree stump, use one stick of
dynamite per 4 inches of stump diameter''[Parker, T., Rules of Thumb,
Boston, MA: Houghton Mifflin Publishers, 1983.]).
- Use of empirical correlations (e.g., certain
bacteria have been observed to be likely to cause infection in burn patients).
- Use of ``incidental'' facts to discriminate
cases (e.g., ``a snake that swims with its head out of the water is a
water moccasin''). Such discrimination depends on the
sparseness of the domain (only certain snakes are possible).