PI: Megan Peters
we use neuroimaging and computational modeling to study how the brain represents and uses uncertain information and uncertainty itself
our brains continuously filter, quantify, categorize, and make "us" aware of a deluge of sensory information with remarkable accuracy and precision. we behave adaptively in our environments and we learn. in many cases, the neural computations underlying these abilities appear to be mathematically optimal (but maybe not always). how do brains do this? how is noisy, ambiguous information represented in neuronal activity and neural connections? how does a brain know whether it has interpreted incoming information correctly? how does it learn what to expect, and when to update those expectations? what can we learn from human and animal neural processing that will be beneficial to development of artificial systems?
these are the types of questions we try to answer in the lab. we use an interdisciplinary approach drawing insights and methodologies from bioengineering, computational neuroscience, cognitive science, psychology, and even philosophy. check out the projects page for more info.
generative models, monte carlo simulation, Bayesian decision theory, signal detection theory, neural networks, machine learning