Agent-ScanKit: Unraveling Memory and Reasoning of Multimodal Agents via Sensitivity Perturbations
This work addresses the reliability issue for multimodal agents in real-world GUI tasks, highlighting a fundamental limitation in current methods.
The paper tackles the problem of unreliable reasoning in multimodal agents for graphical user interfaces by proposing Agent-ScanKit, a probing framework that reveals mechanical memorization often outweighs systematic reasoning, with results showing limited generalization across 18 agents in five benchmarks.
Although numerous strategies have recently been proposed to enhance the autonomous interaction capabilities of multimodal agents in graphical user interface (GUI), their reliability remains limited when faced with complex or out-of-domain tasks. This raises a fundamental question: Are existing multimodal agents reasoning spuriously? In this paper, we propose \textbf{Agent-ScanKit}, a systematic probing framework to unravel the memory and reasoning capabilities of multimodal agents under controlled perturbations. Specifically, we introduce three orthogonal probing paradigms: visual-guided, text-guided, and structure-guided, each designed to quantify the contributions of memorization and reasoning without requiring access to model internals. In five publicly available GUI benchmarks involving 18 multimodal agents, the results demonstrate that mechanical memorization often outweighs systematic reasoning. Most of the models function predominantly as retrievers of training-aligned knowledge, exhibiting limited generalization. Our findings underscore the necessity of robust reasoning modeling for multimodal agents in real-world scenarios, offering valuable insights toward the development of reliable multimodal agents.