Absorbed in Inertia: Activation Analysis for Computer-Use Agents

2026-09-29

AI Summary

This study addresses the inertia dilemma in computer-use agents, where invalid actions frequently trap them in repetitive failure loops. For the first time, this work maps agent inertia into a high-dimensional activation space, revealing through activation analysis that its underlying mechanism constitutes an absorbing state. Accordingly, an R3 context reset strategy is proposed to disrupt execution loops and restore task progression. The findings demonstrate that modifying the context proves substantially more effective in eliminating inertia than directly manipulating activation values. By establishing an activation measurement protocol and a high-dimensional state analysis methodology, this research achieves a 17%–55% reduction in inertia across multiple models, significantly improving overall task completion rates.

Abstract

Computer-use agents have become increasingly capable of executing tasks on live desktops through natural-language instructions, based on trajectories of screenshots, actions, and reasoning. We discover that they can stealthily exhibit inertia, in which they repeat fruitless actions despite recognizing that these actions are ineffective. We hypothesize that inertia is reflected in the agent's internal state, i.e., the activation values of the agent's underlying model, and propose a protocol to measure the relationship between the two. Extensive analysis of high-dimensional activation states shows that inertia corresponds to an absorbing region of activation space, where activation values become stale across actions and even after attempts to steer them. We conjecture that drastically changing the agents' activations by re-initializing them is necessary to escape inertia. Specifically, we propose R$^3$ (Reset, Reroute, Restore), which temporarily resets the agent's context trajectory to escape the absorbing region and then restores the historical context to effectively complete the task. Our approach yields 17-55% lower measured inertia across models relative to unmodified agents. These results suggest that changing the context can interrupt recurrence more effectively than directly steering the resulting activations. Our code is available at https://anonymous.4open.science/r/vlm-agent-defense-D076

Problem

• Computer-use agents
• Inertia
• Activation analysis
• Absorbing region

Innovation

• Computer-use agents
• Inertia
• Activation analysis
• Absorbing region
• Context reset