|
|
|
|
|
by HerculePoirot
998 days ago
|
|
As a formal framework, probability theory relies on a series of axioms to hold (e.g. a measurable space of events), which enables a closure provided with interesting mathematical properties, but which is not necessarily representative of the way in which humans mentally form or process events. Given any description of the world we may always find a description which differs in some aspect from the previous one, adding any detail. As a modeling framework, the limitations of standard theory of probability to capture human reasoning is proven by the existence of several cognitive patterns (often named biases or fallacies) which do not follow what is predicted by the formal theory. The core limitation motivating the present contribution lies however in the mismatch between what humans see as informative and the definition of information given by Shannon, that triggered in the 90’s the introduction of Simplicity Theory (ST) . The present paper introduces a novel hypothesis concerning the theoretical bases which makes this cognitive model functional. Unexpectedness and Bayes’ Rule – Giovanni Sileno and Jean-Louis Dessalles https://cifma.github.io/Papers-2021/CIFMA_2021_paper_13.pdf |
|
I don't think probability theory (or more generally logic) tries to model human reasoning. To the contrary, it is a tool to make reasoning better by introducing rigour, and thus overcoming biases.