Research

Information seeking in infant–caregiver play

NIH Blueprint / BRAIN Initiative D-SPAN Award (F99/K00)

How do infants use their gaze for learning? In this study, we ask whether infants' visual attention in a naturalistic play paradigm is guided by the same principles of information seeking that guide visual attention in screen-based paradigms. We aim to understand how visual attention changes in response to children's own actions, as well as the broader informational landscape — that is, the words and actions of informative caregivers.

Investigating the influence of conversational AI on learning

Huo Family Foundation Postdoctoral Fellowship

By elementary school, children have acquired complementary tools for information seeking. In addition to seeking information with their eyes (i.e., "visual question asking"), they begin seeking information with words (i.e., "verbal question asking"). How does real-time attention and question-asking unfold as school-aged children explore museum exhibits with AI tools? And what is the impact of active information seeking on longer-term learning and curiosity?

Convergence and divergence in naturalistic movie watching

How do two people watching the same scene come to build similar — or different — viewing strategies? This project shows adults, children, and infants a shared set of naturalistic movie clips that vary along several dimensions: how heavily produced versus "raw" the footage is, how quickly scenes cut, how much social versus nonsocial content is on screen, and how clearly the material is aimed at a young audience. We're asking what visual and semantic features cause viewers to look in similar versus different ways, and how that pattern evolves over development.

Individual signatures in naturalistic gaze behavior

NIH Blueprint / BRAIN Initiative D-SPAN Award (F99/K00)

When people explore a real-world scene, where they look leaves a trace of their informational priorities. My dissertation asked what drives this selection, and why individuals prioritize different information from one another. Using a novel paradigm — head-mounted virtual reality that lets viewers actively explore immersive, naturalistic scenes — I found that active visual attention is an information-seeking process, disproportionately guided by meaningful semantic content compared to passive viewing. Drawing on natural language models to characterize abstract, conceptual-level scene information, I then identified "gaze fingerprints": individually specific, reliable patterns of visual attention that generalize across diverse scenes. [Paper]

Gaze differences in autism

NIH Blueprint / BRAIN Initiative D-SPAN Award (F99/K00)

My dissertation also examined how these individual gaze signatures differ in autism. I found that conceptual priorities guide visual attention among autistic adults much as they do in neurotypical adults, and that models of these conceptual gaze patterns contain classifiable information about a viewer's diagnostic status.