Hume's Representational Conditions for Causal Judgment: What Bayesian Formalization Abstracted Away

TL;DR

This paper analyzes Hume's conditions for causal judgment, tracing their evolution through Bayesian and modern models, highlighting how contemporary AI models lack these representational features.

Contribution

It identifies and extracts Hume's three representational conditions, showing their abstraction in Bayesian formalizations and implications for AI models.

Findings

Hume's conditions are experiential grounding, structured retrieval, and vivacity transfer.

Bayesian models preserve updating but abstract away these conditions.

Large language models exemplify statistical updating without satisfying the conditions.

Abstract

Hume's account of causal judgment presupposes three representational conditions: experiential grounding (ideas must trace to impressions), structured retrieval (association must operate through organized networks exceeding pairwise connection), and vivacity transfer (inference must produce felt conviction, not merely updated probability). This paper extracts these conditions from Hume's texts and argues that they are integral to his causal psychology. It then traces their fate through the formalization trajectory from Hume to Bayesian epistemology and predictive processing, showing that later frameworks preserve the updating structure of Hume's insight while abstracting away these further representational conditions. Large language models serve as an illustrative contemporary case: they exhibit a form of statistical updating without satisfying the three conditions, thereby making visible requirements that were previously background assumptions in Hume's framework.