Emerging Mind

How do newborns learn to perceive and understand the world? Although philosophers and psychologists have puzzled over the origins of the mind for centuries, two major barriers have hindered progress. First, humans can't be raised in strictly controlled environments from birth, so it has not been possible to examine how specific experiences shape the newborn mind. Second, newborn humans can’t be tested continuously for long periods of time, so it has not been possible to examine — with high precision — how perception and cognition emerge in the newborn brain.

To overcome these two barriers, we can turn to controlled-rearing studies of newborn chicks. We use newborn chicks as an animal model because they are uniquely suited for studying the earliest stages of visual learning. Unlike commonly-used animal models in psychology (e.g., rats, pigeons, and monkeys), chickens are a precocial species (mobile in the first day of life) and can be raised in strictly controlled environments immediately after hatching.

To study the origins of perception and cognition in chicks, our lab has developed a variety of automated controlled-rearing methods. These methods allow newborn chicks to be raised in strictly controlled virtual environments from the onset of vision. Using automated image-based tracking software, we record all of the chicks’ behavior (24/7), providing a complete digital record of each newborns’ behavior over their first few weeks of life. As a result, we can measure early emerging perceptual and cognitive abilities with an unprecedented degree of precision. Play the video below to take a short virtual tour of our lab and see our controlled-rearing chambers.

Our results show that newborn organisms can develop high-level visual abilities rapidly. For example, newborn chicks can build an abstract (view-invariant and background-invariant) representation of the first object they see in their life, and can recognize objects rapidly—within a fraction of a second. Newborn brains are therefore equipped with powerful visual processing machinery. We have also discovered, however, that these abilities only emerge when newborns receive certain types of experiences. Specifically, newborn chicks need experience with natural visual environments, containing objects moving slowly and smoothly over time across patterned backgrounds. Without natural visual experience, chicks develop abnormal or ‘incorrect’ object concepts. Our working hypothesis is that natural (slow and smooth) visual input optimizes the newborn brain for object perception, allowing for the emergence of abstract mental models of objects. These results illuminate the role of experience in the development of object perception and provide high-precision benchmarks for testing computational models of newborn vision.

One of our ultimate goals is to develop quantitative models of newborn brains. For each controlled-rearing experiment, we create a “virtual experiment” for testing the learning abilities of autonomous artificial agents (models). The artificial agents receive only pixels as input (akin to the retinal images received by newborn chicks) and are raised in virtual environments that simulate the controlled-rearing experiments. By raising newborn animals and autonomous artificial agents in the same environments, we can compare their learning abilities and build progressively more accurate models of newborn brains. We plan to make this testbed widely available to the scientific community to facilitate a community-wide effort to formalize the learning mechanisms in newborn brains.

Watch the video below for a talk presented to MIT's Center for Brains, Minds, and Machines Seminar Series.