DEFT: Demystifying VLN Failures via a Unified Dual-View Explainability Framework for LLM-based Agents

Large Language Models (LLMs) have emerged as central planners in Vision-and-Language Navigation (VLN), yet their complexity increasingly obscures their internal decisionmaking. Existing interpretability methods typically isolate temporal criticality from feature salience, creating an alignment gap and failing to account for the behavioral instability of black-box agents. To address this, we propose DEFT, a unified dual-view framework that demystifies agent behavior by jointly analyzing when a decision is pivotal and what visual evidence grounds it. Featuring a dualhead architecture with a shared latent representation, DEFT employs a Mask Head for counterfactual-based criticality detection and an Action Head that leverages an ensemble of surrogates to recover robust visual cues. Extensive experiments on MatterPort3D across three LLM-based agents demonstrate that DEFT outperforms baselines in both temporal and feature fidelity. User studies further validate its utility, showing 78% alignment with human intuition.