A logical concept of representation independence is developed
for nonmonotonic logics, including probabilistic inference rules.
The general framework then is used to examine default logic and
propositional probabilistic logics with respect to their
representation independence. In the latter case our investigation
leads us to modified inference rules with greater representation
independence.
(This talk is about a new paper of mine I have submitted
to KR'96. You can have a look at the paper at
~jaeger/TeX/Papers/RepInd_paper/RISubmit.ps .)