Internalist Reliabilism in Statistics and Machine Learning: Thoughts on Jun Otsuka's Thinking about Statistics

TL;DR

This paper explores the philosophical foundations of statistics and machine learning, proposing a unifying internalist reliabilist perspective that links Bayesian, classical, and machine learning methods.

Contribution

It introduces a novel interpretation of classical statistics and machine learning as sharing an internalist reliabilist foundation, promoting philosophical unity.

Findings

Classical statistics and machine learning share reliabilist foundations.

A unified internalist reliabilist framework can encompass diverse statistical methods.

The perspective aligns with internalist epistemology in philosophy.

Abstract

Otsuka (2023) argues for a correspondence between data science and traditional epistemology: Bayesian statistics is internalist; classical (frequentist) statistics is externalist, owing to its reliabilist nature; model selection is pragmatist; and machine learning is a version of virtue epistemology. Where he sees diversity, I see an opportunity for unity. In this article, I argue that classical statistics, model selection, and machine learning share a foundation that is reliabilist in an unconventional sense that aligns with internalism. Hence a unification under internalist reliabilism.