A new study from Dartmouth researchers reveals that the way you look at the world is so uniquely yours that artificial intelligence can identify you based solely on your eye movements, even more accurately than by analyzing the physical objects you're looking at. The findings, published in the Proceedings of the National Academy of Sciences, suggest that our eyes unconsciously seek out information that matches our personal, abstract interests, creating a visual signature as distinctive as a fingerprint.

What Makes Your Gaze Pattern Unique?

The research team, led by Caroline Robertson at Dartmouth, studied about 60 participants wearing virtual reality headsets as they explored everyday scenes including an auto repair shop, a public swimming pool, and an airport. Each person had 16 seconds to look around freely while eye-tracking technology recorded exactly where their gaze landed.

The key discovery centers on what researchers call "conceptual priorities," which are personal mental biases that determine what visually stands out to you. For example, an American flag and a football look completely different physically, but an AI language model can group them under abstract themes like "patriotism" or "sports culture." The study proved that our eyes naturally spend more time seeking out objects that match our unique internal thematic interests.

"From the earliest moments of taking in a new environment, we make radically different choices about what we pay attention to. This work suggests that our latent conceptual priorities are embedded in the signatures of our gaze," said Caroline Robertson, an associate professor of psychological and brain sciences at Dartmouth.

Caroline Robertson, Associate Professor of Psychological and Brain Sciences, Dartmouth

When researchers analyzed the eye-tracking data using machine learning and large language models (LLMs), which are the deep-learning systems powering modern artificial intelligence, they found something striking. The LLM, which specifically encoded participants' conceptual preferences, could identify individuals by their gaze patterns more accurately than a vision model that only tracked the physical features of objects they looked at.

How Does the Brain Process Visual Information?

The research revealed that regardless of personal interests, all human visual exploration follows a consistent three-stage biological timeline when entering a new space. Understanding these stages helps explain why some people notice different things in the same environment:

• Stage 1 (0 to 2 seconds): The brain scans the physical dimensions of the space, such as the room's horizon and center point, establishing basic spatial orientation.
• Stage 2 (2 to 8 seconds): Gaze shifts to prominent, physically striking objects and people that stand out visually from their surroundings.
• Stage 3 (8 to 16 seconds): The brain transitions into a deep conceptual mode, focusing heavily on what objects mean to you personally and thematically.

This progression explains why the LLM's analysis of conceptual meaning proved so powerful for identifying individuals. By the time people reach that third stage, their gaze patterns reflect their deepest personal interests and values, not just what happens to be visually prominent.

How Stable Are These Gaze Patterns Over Time?

One of the most striking findings was the temporal stability of gaze preferences. When about half of the study participants returned a week later to view an entirely new set of scenes, the AI models built from their previous week's data accurately predicted exactly what objects would catch their eye. This demonstrates that individual gaze choices are stable, personality-level traits that persist over time, much like other consistent personality characteristics.

"This suggests that individual differences in gaze patterns contain stable, personality-level preferences that extend beyond testing days," noted Caroline Robertson.

Caroline Robertson, Associate Professor of Psychological and Brain Sciences, Dartmouth

The consistency of these patterns raises important questions about privacy in an increasingly connected world. As virtual reality and augmented reality headsets become more integrated into daily life, eye-tracking data could become a powerful tool for understanding people's interests, beliefs, and psychological profiles.

What Are the Real-World Implications of This Research?

The findings have both concerning and promising applications. On the privacy side, advertisers could potentially use this technology to passively harvest information about your political leanings, hobbies, and psychological profile based simply on what you look at in virtual environments, moving far beyond traditional web-click tracking.

However, the research also opens doors to meaningful clinical applications. Dr. Amanda Haskins, the study's first author, noted that this marks the first time an LLM has been used to model human visual gaze in this way. The framework could revolutionize early autism screenings by determining whether a child's visual avoidance of faces stems from a visual processing difference or a deeper conceptual choice. This distinction could enable clinicians to diagnose autism closer to age two, lowering the current national average diagnosis age of four and enabling faster educational support.

The research demonstrates how advances in AI and eye-tracking technology can reveal hidden patterns in human behavior that were previously invisible to researchers. As these tools become more sophisticated and more widely available, understanding both their potential benefits and privacy risks will become increasingly important for individuals, researchers, and policymakers alike.