MSL906B Analyses using AI-enabled Computational Neuroscience

The brain contains many neurons. No one really knows exactly how many, but an adult human brain contains somewhere around 100 billion. A significant amount of research in neuroscience over the past century has involved the use of electrodes to measure brain activities, and then recording how their activity changes in response to different sensory or cognitive variables, medications, and so on. One key area in Computational Neuroscience is that of frequency-domain analyses, which are used to investigate processes such as neural oscillations. The advantages of frequency-domain analyses are that they are computationally fast and are ubiquitous in many branches of science, engineering, and communications technologies.

A second important area of analyses is that of Action Potentials. The term “spike” is colloquially used to refer to action potentials. Most people use those terms interchangeably. The action potentials of individual neurons are fast events, lasting a few milliseconds. They are so fast that researchers often treat them as binary events. That is, you can imagine a Boolean or logical time series that contains mostly zeros (no spike at that time point) and some 1s (a spike occurred at that time point). In this course, students will learn how the shape of the time course of the action potential can be used to isolate action potentials from different neurons that are measured by the same electrode.