On the Faithfulness of Post-Hoc Concept Bottleneck Models

2026-06-29

AI Summary

This work addresses a critical limitation in existing post-hoc concept bottleneck models, which rely solely on task accuracy and thus fail to detect semantically unfaithful concepts—such as predictive artifacts mistakenly treated as valid concepts. The study formally characterizes the mechanisms underlying such unfaithfulness, identifying two failure modes driven by covariate shift in auxiliary data and label noise from vision-language models. To overcome this, the authors propose a novel evaluation paradigm that decouples semantic faithfulness from predictive accuracy. Through theoretical analysis of error upper bounds, experiments on both synthetic and real-world benchmarks, and a new faithfulness metric, the proposed approach effectively uncovers semantic distortions entirely overlooked by conventional evaluation metrics, demonstrating its necessity and efficacy across multiple benchmarks.

Abstract

Human decision-making interprets the world through high-level concepts, such as recognizing a bird by its belly color. To bridge the gap between opaque deep learning representations and human understanding, Post-Hoc Concept Bottleneck Models (post-hoc CBMs) project latent features onto interpretable concept spaces using auxiliary datasets or vision-language models. However, relying on target task accuracy as the primary measure of post-hoc CBM success obscures whether the learned concepts are semantically meaningful or merely predictive artifacts. For example, random concept projections can achieve competitive accuracy despite being semantically meaningless. In this work, we analyze the learned projections directly and identify two failure cases: First, for concept projections learned from auxiliary data, covariate shifts can lead to unfaithful concept representations for the target task. In particular, we provide an upper bound on the error introduced by this shift. Second, systematic label noise in surrogate concept labels generated by vision-language models leads to unfaithful projections. After formalizing these failure modes, we introduce novel metrics that decouple concept faithfulness from predictive accuracy. Our empirical results across real-world and synthetic benchmarks confirm that these metrics identify unfaithful behaviors that standard accuracy-based evaluation fails to detect.

Problem

• Post-Hoc Concept Bottleneck Models
• Concept Faithfulness
• Semantic Meaningfulness
• Covariate Shift
• Label Noise

Innovation

• post-hoc Concept Bottleneck Models
• concept faithfulness
• covariate shift
• label noise
• interpretability metrics