Fundamental Limitation on Applicability of Statistical Methods to Study of Living Organisms and Other Complex Systems

Abstract

A living organism is a complex system whose state is characterized by extremely large number of variables that far exceeds the number of individual organisms that can be experimentally studied. Since the relations between these variables and even their identities are largely unknown, the applicability of statistical methods of inference to the outcome of experiments in biomedical sciences is severely limited. Far from being a purely theoretical issue, this explains the recently proposed “Truth Wears Off” effect and sets a fundamental limitation on the applicability of machine-like approaches to the study of living organisms.