Estimating Central, Peripheral, and Temporal Visual Contributions to Human Decision Making in Atari Games
arXiv:2604.04439v2 Announce Type: replace Abstract: We study how different visual information sources contribute to human decision making in dynamic visual environments. Using Atari-HEAD, a large-scale Atari gameplay dataset with synchronized eye-tracking, we introduce a controlled ablation framework as a means to reverse-engineer the contribution of peripheral visual information, explicit gaze information in the form of gaze maps, and past-state information from human behavior. We train action-prediction networks under six settings that selectively include or exclude these information sources
This research builds on advancements in large-scale AI datasets and eye-tracking technology, enabling deeper analysis of human cognitive processes in dynamic environments.
Understanding how humans make decisions visually can inform the development of more sophisticated and human-like AI agents, especially for tasks requiring rapid perception and action.
The ability to reverse-engineer human visual contributions to decision-making provides a new framework for AI training, potentially leading to more efficient and robust agent design.
- · AI Researchers
- · Gaming Industry
- · Robotics Developers
- · Cognitive Science
- · AI approaches lacking human-centric design
Improved action-prediction networks and AI agent performance in dynamic visual tasks.
Development of AI systems that can better anticipate human actions and intentions by mimicking human visual processing.
Ethical considerations regarding AI systems with highly human-like perception and decision-making capabilities, particularly in sensitive applications.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG