CAREER: Characterizing Object Recognition Machinery in a Newborn Visual System

University Of Southern California, Los Angeles CA

Investigators

Abstract

How does early experience shape how we process and interpret visual information? Two major limitations have made this question difficult to answer. First, researchers can typically collect only a few data points from newborns, which prevents precise measurement of the infants' visual cognitive abilities. Second, human infants cannot ethically be raised in controlled environments from birth, which prevents researchers from studying how specific experiences shape the newborn mind. To overcome these limitations, Dr. Wood has developed a new controlled-rearing method using a non-human animal model. This method can be used to measure all of a newborn's behavior (24 hours/day, 7 days/week) with high precision (9 samples/second) within strictly controlled environments. With support from this NSF CAREER award, Dr. Wood will use the new controlled-rearing method to characterize how newborns recognize objects at the onset of visual object experience. Dr. Wood's laboratory will use a two-pronged approach. First, the lab will perform a series of controlled-rearing experiments with newborn chickens. Studies of chickens can inform human cognitive development because chickens and humans have similar neural processing systems for sensory information. These controlled-rearing experiments will reveal how specific visual experiences shape newborns' object recognition abilities. The findings will provide the foundation for a new, publicly-accessible database that describes how specific sensory experiences relate to specific behaviors in a newborn organism. Second, the lab will build biologically-inspired computational models of newborns' object recognition behavior, using state-of-the-art techniques from artificial intelligence. These models will make predictions that can be compared to the data from the controlled-rearing experiments. This will help identify how the visual system processes objects. This approach integrates ideas from developmental psychology, vision science, and computational neuroscience, providing a unified framework for studying the origins of object recognition and other visual cognitive abilities.