A computational tutorial at MIT
In April 2019, Dr. Costilla Reyes, then a postdoctoral researcher in Earl Miller’s laboratory, taught “Neural decoding of spike trains and local field potentials with machine learning in Python,” a computation tutorial of MIT’s Department of Brain and Cognitive Sciences. It took place in the McGovern Seminar Room in Cambridge, Massachusetts, was open to the public, and asked participants to bring a laptop with Anaconda Python 3.7. Afterward he thanked the attendees and shared Jupyter notebooks, data, and slides, and the session was later published as a video.
What neural decoding is
Neural decoding uses machine-learning models to predict a stimulus, a behavior, or a cognitive state from recorded brain activity, which is useful both for understanding what neural populations represent and for building brain-computer interfaces. The tutorial worked with two kinds of signals: spike trains, which record when individual neurons fire, and local field potentials, slower electrical signals that reflect the combined activity of many neurons near the electrode. The working dataset contained recordings from six cortical areas of the macaque brain, spanning frontal to occipital cortex, with examples focused on prefrontal cortex (PFC) and visual area V4.
The recording demonstrates how wideband neural signals are processed into spike and local-field-potential data and then used in prediction tasks. The hands-on workflow covers data preprocessing, model selection and optimization, and cross-validation, which estimates how well a model performs on data it was not trained on.
Public learning materials
MIT OpenCourseWare links the tutorial references and videos from its Brain and Cognitive Sciences computational tutorials collection, and MIT’s Center for Brains, Minds & Machines publishes the recording and transcript, so readers can follow the presentation and its discussion. The tutorial is a teaching resource that demonstrates methods, and it complements his research at MIT on neural signals recorded while animals perform cognitive tasks.