Carlos R. Ponce, MD, Ph.D., Assistant Professor in the Department of Neuroscience at Washington University School of Medicine discusses his work studying conscious understanding.

 

Dr. Ponce's research program seeks to explain and elaborate on how various brain regions interact to solve motion processing as well as visual object recognition, by utilizing a combination of reversible deactivation and sophisticated microstimulation techniques, and computational modeling. Dr. Ponce holds an MD-PhD from Harvard Medical School and a Bachelor of Science from the University of Utah.

 

Dr. Ponce talks about what motivated him to study and research the brain, cells, and signals, and he explains how the interaction happens that creates a consciousness of vision. Building upon fifty to sixty years of work in vision, Dr. Ponce seeks to learn more about the brain's hierarchy of visual areas—groups of neurons. As he explains, cells in the occipital lobe respond to basic information, lines and dots, but these cells communicate with other cells that respond to slightly more complicated information such as corners and curvature, and then these cells project to other cells that respond to texture, and so on, in a systematic way that ultimately provides conscious understanding. 

 

The physician-scientist explains the process and the stages that take place as light enters the eye, from light hitting the retina, to the center of the brain, to the primary visual cortex cells, and eventually the temporal cortex, which completes visual comprehension. He expounds upon his thoughts on how his research is related to AI, and how neural networks respond to complex shapes. He details how neural networks receive information and how they learn, and compares that to how the human brain learns. As he explains, the human brain has much-unsupervised learning, which is different than AI. Dr. Ponce explains his thoughts about the human brain, and how it is a form of a neural network. 

 

Dr. Ponce cites some examples that explain his theories and research. He details generative adversarial networks (GANs), which are a class of machine learning systems, and he explains evolutionary algorithms, and then outlines how they utilize this combination to find and extract information that is hidden by a cell. 

 

The Harvard Ph.D. explains the state of machine learning, and he discusses some subtleties of machine vision. And Dr. Ponce cites examples of early experimentation in machine vision, discussing symmetry and other factors.