What will the lab study?

My lab will focus on understanding how we learn to derive meaning from what we see.  This work sits at the intersection between vision science, developmental psychology, and machine learning, and I welcome mentees with interests in any of these areas! The lab will primarily focus on behavioral studies with infants and children and use computational models as both tools for analyzing large, naturalistic datasets as well as testbeds for modeling early learning.  Our overarching goal is to create a grounded theory of visual concept learning — one that incorporates realistic estimates of what infants’ see, formalizes visual recognition and learning mechanisms, and makes concrete predictions of how visual concepts evolve throughout early development and into adulthood.  To do so, we'll focus on three main research areas:

Learning environment:  What are the regularities in children’s everyday learning environment? And what are the consequences of these regularities for models and theories of visual concept learning? This line of work will quantify what's-in-view for infants and young children using video data from head-mounted cameras (see Long et al., 2023; Behavioral Research Methods), and develop parental surveys that can capture variability in what young children experience.

​

Learning mechanisms: What are mechanisms that help children build robust and generalizable visual concepts? And what kinds of models can learn as children do — from this same kind of everyday learning environments vs. large, curated datasets of photographs? How can we construct models that better approximate the actual learning process of young children?

​

Representations: How do children's visual concept change throughout childhood? What are the major factors that influence how children represent visual information? How can we best measure graded changes in children's representations of everyday visual concepts?

​