Modelings decisions in AI:re-thinkning Linda in terms of coherent lower and upper conditional previsions

In this paper, the model of coherent upper and lower conditional previsions is proposed to represent preference orderings and equivalences between random variables that human beings consider consciously or unconsciously when making decisions. To solve the contradiction, Linda’s Problem (i.e., conjunction fallacy) is re-interpreted in terms of the probabilistic model based on coherent upper and lower conditional probabilities. Main insights of this mathematical solution for modeling decisions in AI are evidenced accordingly.