Rethinking Saliency Maps: A Cognitive Human Aligned Taxonomy and Evaluation Framework for Explanations

TL;DR

This paper introduces a new taxonomy and evaluation framework for saliency maps, emphasizing the importance of aligning explanations with human reasoning and diverse user queries, and critiques existing metrics for their limited scope.

Contribution

It proposes the RFxG taxonomy for classifying explanations and develops four new faithfulness metrics to better evaluate saliency methods across different explanation types.

Findings

Existing metrics favor pointwise faithfulness over contrastive reasoning.

The evaluation reveals limitations in current saliency methods across RFxG dimensions.

The framework promotes user-centric evaluation for more meaningful explanations.

Abstract

Saliency maps are widely used for visual explanations in deep learning, but a fundamental lack of consensus persists regarding their intended purpose and alignment with diverse user queries. This ambiguity hinders the effective evaluation and practical utility of explanation methods. We address this gap by introducing the Reference-Frame $\times$ Granularity (RFxG) taxonomy, a principled conceptual framework that organizes saliency explanations along two essential axes:Reference-Frame: Distinguishing between pointwise ("Why this prediction?") and contrastive ("Why this and not an alternative?") explanations. Granularity: Ranging from fine-grained class-level (e.g., "Why Husky?") to coarse-grained group-level (e.g., "Why Dog?") interpretations. Using the RFxG lens, we demonstrate critical limitations in existing evaluation metrics, which overwhelmingly prioritize pointwise faithfulness while neglecting contrastive reasoning and semantic granularity. To systematically assess explanation quality across both RFxG dimensions, we propose four novel faithfulness metrics. Our comprehensive evaluation framework applies these metrics to ten state-of-the-art saliency methods, four model architectures, and three datasets. By advocating a shift toward user-intent-driven evaluation, our work provides both the conceptual foundation and the practical tools necessary to develop visual explanations that are not only faithful to the underlying model behavior but are also meaningfully aligned with the complexity of human understanding and inquiry.