Dear UAIers, on the history of "plausibility-like" measures, it may be of interest to see a fascinating piece by Henry Kyburg Jr. in SIPTA Newsletter, vol. 1, pp. 2-5, 2003; this issue of the newsletter is at http://leo.ugr.es/sipta/news/jan2003.ps.gz There is a lot of historical information there.
Cheers, Fabio Cozman On Thu, 20 Feb 2003, Peter McBurney wrote: > Hello all -- > I have just seen a 1959 paper by the late Australian philosopher and > computer pioneer, Charles Hamblin, which defines a measure he calls > "plausibility": > C. L. Hamblin [1959]: "The Modal "Probably" ", <Mind>, New Series, 68 > (270): 234-240. > Hamblin defines the plausibility of an event E as a real number between > 0 and 1 (inclusive) such that: > (i) the plausibility for a given event is assigned without regard to the > number and plausibilities assigned to contrary events; and > (ii) the plausibility assigned is not limited by the usual probability > additivity rule (that probabilities of the whole space should add to 1). > He then defines a simple combination rule: the plausibility of a > disjunction is equal to the largest of the plausibilities of the > components of the disjunction. > This would appear to be the first published formal definition of a > plausibility measure. In the ISI Citations Index, I have only been > able to find 2 papers since 1981 which cite this paper, and neither > relates to uncertainty formalisms. > Does anyone know of any earlier formal treatment of non-probabilistic > uncertainty? Hamblin cites Shackle's work on uncertainty in economics, > but I understand that treatment was not formal. > With thanks, > -- Peter McBurney > Department of Computer Science > University of Liverpool