Cognitive Decisions under Partial Information

Abstract

Since Shannon’s original work about information theory the notion of information and later on the notions of information content, semantic information, information value etc. have run through many definitions. Carnap (1966), Hintikka (1970), and others defined semantic information in a way (tracing back to Laplace (1812)) which virtually states that the more probable (or less surprising) an information is, the less valuable it is. This view is closely connected with Popper’s (1959) work, in which less probable hypotheses are considered more valuable, more informative, and more easily accessible to falsification. The ‘Surprise Thesis’ of information theory was questioned mainly by Levi (1969), Menges (1972), and Marschak (1974). The latter, in particular, emphasized the importance of the distinction between prior and posterior probabilities.