Inferring to the Best Explanation from Uncertain Evidence

2025/09/10 by Finnur Dellsén · 1 voice · 1 citation
Computer Science · #Explainable Artificial Intelligence (XAI) #AI-based Problem Solving and Planning #Bayesian Modeling and Causal Inference

paper · pdf · doi:10.1017/psa.2025.10151

Abstract

Abstract This paper presents a new problem for the inference rule commonly known as “inference to the best explanation” (IBE). The problem is that uncertainty about parts of one’s evidence may undermine the inferrability of a hypothesis that would provide the best explanation of that evidence, especially in cases where there is an alternative hypothesis that would provide a better explanation of only the more certain pieces of evidence. A potential solution to the problem is sketched, in which IBE is generalized to handle uncertain evidence by invoking a notion of evidential robustness .

Citations

• On the relative motion of the Earth and the luminiferous ether

Cited by

• What Kind of Reasoning (if any) is an LLM actually doing? On the Stochastic Nature and Abductive Appearance of Large Language Models

Discussions

• A weirdly underappreciated problem about Inference to the Best Explanation is how it can handle uncertain evidence. This new paper, now forthcoming in Philosophy of Science, proposes a strategy for do [bsky, 24 points, 3 comments]

Related

• Uncertain Inferences and Uncertain Conclusions
• Uncertain Evidence in Probabilistic Models and Stochastic Simulators
• Evidence with Uncertain Likelihoods
• Evidence with Uncertain Likelihoods
• The Causal Uncertainty Principle