Despite the effectiveness of convolutional neural networks (CNNs) in visual categorization, the logic behind their predictions is not human-understandable. While existing concept-based explainability methods reveal what a CNN sees, there is a need to understand how a specific concept is chosen (rather than another concept) for a prediction, aligning more closely with human perception. To address this challenge, we propose a novel contrastive paradigm to bridge the critical gap in global concept discovery by leveraging contrasts from cognitive sciences for discriminative concept retrieval. A new multiple-case concept retrieval method is proposed for improved local understanding of (dis)similar classification cases. We argue that a contrastive paradigm for concept retrieval and sanity checks is essential to an explainer's trustworthiness and integrate these missing ingredients into state-of-the-art concept-based explanation frameworks to foster a better human understanding through contrast. The proposed Contrastive Perceptual Inference and Sanity Checks for Concept-based CNN Explanations (CoPISan) framework accelerates salient concept retrieval. It evaluates explainer trustworthiness via sanity checks conducted under Frontdoor and Poisoning adversarial attacks. Experimental results demonstrate CoPISan's encouraging performance, mitigating issues related to duplication, entanglement, diminishing returns, and ambiguity of concept explanations. CoPISan is motivated by cognition and perception, offers theoretical justification and resilience, and is computationally efficient.

Convolutional neural networks (CNNs) have succeeded remarkably in various computer vision tasks. However, they are not intrinsically explainable. While feature-level understanding of CNNs reveals where the models looked, concept-based explainability methods provide insights into what the models saw. However, their assumption of linear reconstructability of image activations fails to capture the intricate relationships within these activations. Their fidelity-only approach to evaluating global explanations also presents a new concern. For the first time, we address these limitations with the novel transformative nonlinear concept explainer (TraNCE) for CNNs. Unlike linear reconstruction assumptions made by existing methods, TraNCE captures the intricate relationships within the activations. This study presents three original contributions to the CNN explainability literature: 1) an automatic concept discovery mechanism based on variational autoencoders (VAEs). This transformative concept discovery process enhances the identification of meaningful concepts from image activations; 2) a visualization module that leverages the Bessel function to create a smooth transition between prototypical image pixels, revealing not only what the CNN saw but also what the CNN avoided, thereby mitigating the challenges of concept duplication as documented in previous works; and 3) a new metric, the faith score, integrates both coherence and fidelity for comprehensive evaluation of explainer faithfulness and consistency. Based on the investigations on publicly available datasets, we prove that a valid decomposition of a high-dimensional image activation should follow a nonlinear reconstruction, contributing to the explainer's efficiency. We also demonstrate quantitatively that, besides accuracy, consistency is crucial for the meaningfulness of concepts and human trust. The code is available at https://github.com/daslimo/TrANCE.

In their natural form, convolutional neural networks (CNNs) lack interpretability despite their effectiveness in visual categorization. Concept activation vectors (CAVs) offer human-interpretable quantitative explainability, utilizing feature maps from intermediate layers of CNNs. Current concept-based explainability methods assess explainer faithfulness primarily through Fidelity. However, relying solely on this metric has limitations. This study extends the Invertible Concept-based Explainer (ICE) to introduce a new ingredient measuring concept consistency. We propose the CoherentICE explainability framework for CNNs, expanding beyond Fidelity. Our analysis, for the first time, highlights that Coherence provides a more reliable faithfulness evaluation for CNNs, supported by empirical validations. Our findings emphasize that accurate concepts are meaningful only when consistently accurate and improve at deeper CNN layers.