Munich Personal RePEc Archive

The Bayesian Fallacy: Distinguishing Four Kinds of Beliefs

Khalil, Elias (2008): The Bayesian Fallacy: Distinguishing Four Kinds of Beliefs.


Download (277kB) | Preview


This paper distinguishes among four kinds of beliefs: conviction, confidence, perception, conception. Conviction concerns self-ability:“I can build these stairs.” Confidence also concerns the self—ut focuses on the assertion of will in the face of weakness of will. Perception is about the environment such as weather prediction. Conception is also about the environment—but usually couched with context. While convictions are noncognitive and nonevidential beliefs, the other beliefs are either cognitive, evidential, or both. This paper uses the terms “cognition” and “evidentiality” as axes to distinguish the four beliefs. While “cognitive beliefs” are about one’s environment, “noncognitive beliefs” are about one’s self. While the cognitive/noncognitive divide is unconventional, it generates a payoff in light of the evidentiality axis. While “evidential beliefs” are correctable via Bayes’s rule, “nonevidential beliefs” are not. However, when the nonevidential belief is about the environment, the evidence can at least make the belief more (or less) warranted—where “warrantability” is a weaker criterion than “correctability.” And when the nonevidential belief is about the self, i.e., a conviction, the evidence cannot even make the belief more (or less) warranted. The evidence itself develops when one tries to test a conviction. This paper highlights that convictions are the basis of tenacity—crucial for entrepreneurship and economic growth. This paper further demonstrates how three major theories of action—standard rationality, normative theory, and procedural rationality—fail to distinguish the four kinds of beliefs. They, hence, commit, although in different ways, a set of confusions called here the “Bayesian fallacy.”

Atom RSS 1.0 RSS 2.0

Contact us: mpra@ub.uni-muenchen.de

This repository has been built using EPrints software.

MPRA is a RePEc service hosted by Logo of the University Library LMU Munich.