In Defence of Post-hoc Explainability

TL;DR

This paper defends post-hoc explainability methods in machine learning as valuable tools for scientific discovery, emphasizing their interpretative role and the importance of empirical validation despite their approximative nature.

Contribution

It introduces a philosophical framework supporting post-hoc explanations as legitimate scientific tools and demonstrates their utility through biomedical ML applications.

Findings

Post-hoc explanations can generate new hypotheses.

Proper integration of explanations advances understanding.

Empirical validation is crucial for reliability.

Abstract

This position paper defends post-hoc explainability methods as legitimate tools for scientific knowledge production in machine learning. Addressing criticism of these methods' reliability and epistemic status, we develop a philosophical framework grounded in mediated understanding and bounded factivity. We argue that scientific insights can emerge through structured interpretation of model behaviour without requiring complete mechanistic transparency, provided explanations acknowledge their approximative nature and undergo rigorous empirical validation. Through analysis of recent biomedical ML applications, we demonstrate how post-hoc methods, when properly integrated into scientific practice, generate novel hypotheses and advance phenomenal understanding.