Models and Tools for Studying Infants’ Attention

From birth, infants actively control where they look, long before they gain any significant motor control over other body parts. This early emergence of attentional preferences has allowed psychologists to use infants' gaze to gain insight into the developmental origins of perception and cognition. Understanding infant gaze therefore is critical both for understanding early development, and for interpreting decades of literature in developmental psychology. This thesis studies the functions of infants' looking behavior, and introduces novel tools to accelerate its study. Chapter 1 is a theoretical review which challenges the notion that learning in infancy is primarily incidental and passive. I outline ways in which infants use their gaze to learn, as well as form and manage social relationships. Chapter 2 demonstrates that, indeed, infants' looking behavior is better understood as an active sampling process. I describe a computational model that posits that infants' gaze is optimized to maximize expected information gain from noisy perceptual input, and show through large-scale behavioral experiments that infant looking is well described by this model. Chapter 3 then confronts the methodological challenges of studying infant gaze empirically: to obtain and process data from a single infant in an infant looking time experiment takes about 2 hours per infant. I describe a workflow in which we reduce this time to about 5 minutes per infants by a) using asynchronous, instead of in-lab, testing, b) training parents, rather than experimenters, to control the flow of experiments, and c) replacing manual gaze coding with automatic annotation using modern computer vision tools. Finally, synthesizing the preceding chapters, Chapter 4 describes outstanding challenges for the empirical and computational study of infant attention.

Name thesis:degree_name

Doctoral

Department dc:contributor.department

Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2025

Author dc:creator

• Raz, Gal

Advisor dc:contributor.advisor

• Saxe, Rebecca R.

Statement dc:rights

• Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
• Copyright retained by author(s)

Licence dc:rights.uri

• https://creativecommons.org/licenses/by-nc-nd/4.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/159205

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/159205