Advancing Neural Interfaces: Signal Processing, Optogenetics, and Beyond
February 22, 2011
2:45 PM - 3:45 PM
Electrical and Computer Engineering
George R. Brown Hall
The last decade's advances have increased the expectation that technology may play a critical role in repairing or replacing damaged portions of the nervous system. At the same time, advances in signal processing, embedded systems, and photonics have led to an increasing role for engineering in facilitating our study of the brain. Development of brain-machine interfaces has focused primarily on replacing damaged sensory input or motor output. We will begin by introducing our work on a motor interface, outlining a model-based approach for decoding the patterns of neural activity which accompany reaching movements. Specifically we will show how a hidden Markov model of neural activity provided a principled solution to an unsolved problem: the automatic detection of internal movement-related regimes of activity corresponding to reach planning and movement initiation.
While the repair of sensory and motor systems is important, perhaps the most fascinating frontier of brain-machine interfaces lies in those systems which interact with more cognitive processes like emotion, decision-making, and memory. The remainder of the talk will focus on this type of system, with a particular focus on interacting with the hippocampus - the neural circuit responsible for the formation, maintenance, and utilization of episodic memories. In order to reverse-engineer how the hippocampal circuit operates during behavior, we designed a system to record neural activity while strategically injecting information. We have used a lentivirus to express the light-sensitive ion channel channelrhodopsin-2 in targeted cells, which are subsequently activated during behavior using pulsed laser light delivered via optical fiber to the hippocampus. Using this and other approaches, we have found that behavioral state – specifically, an animal's speed – strongly modulates the balance of how information flows through and reverberates within the hippocampal circuit.
Characteristic brief patterns of neural activity which arise in the dynamics of the hippocampal circuit appear to correspond to the neural trajectories of memories of specific episodes. The final portion of the talk will focus on the future: engineering novel real-time signal processing and low-power embedded systems that will allow us to manipulate (e.g., weaken or enhance) memory trajectories in animal models over extended periods. Excitingly, the development of these systems, in addition to furthering our understanding of how memories are formed, will represent the first steps towards brain-machine interfaces which therapeutically interact with human memory.
Hosts: Behnaam Aazhang and Richard Baraniuk
Light refreshments will be served before the talk near Duncan Hall Room 1070.
Caleb Kemere is a postdoctoral fellow in the lab of Loren Frank at the University of California, San Francisco. He studies the neurobiology of memory and learning in rats, using tools from traditional multi-electrode recording to novel optogenetic activation. As a graduate student at Stanford University, he was mentored by Teresa Meng and Krishna Shenoy. His thesis work focused on signal processing algorithms for decoding the neural activity which controls arm movements. He is a Helen Hay Whitney and Sloan-Swartz fellow, and an alumnus of the University of Maryland, College Park.