The point process component of an extracellular recording results from the spiking activity of neurons in a background of physical and biological noise. When a recording electrode measures action potentials from multiple cells, these contributions must be disentangled from the background noise and from each other before the activity of individual neurons can be analyzed. This procedure of estimating one or more single cell point processes from a noisy time series is known as spike sorting. When it succeeds, it can transform a weakness of extracellular recording, namely the inability to isolate changes in the firing rate of single neurons into one of its strengths—simultaneous measurement from multiple cells. A range of different approaches have been used to address this problem. Although the algorithmic approaches vary in their assumptions about noise statistics, incorporation of domain knowledge specific to the recording area, and the criteria for identifying single cells, most can be viewed as different implementations of a common series of steps. This chapter develops a framework for these steps and discusses the practical considerations of each level without reference to a specific computational approach. The transformations of the data are illustrated by an idealized example modeled on recordings taken from the mammalian retina.
Oxford Scholarship Online requires a subscription or purchase to access the full text of books within the service. Public users can however freely search the site and view the abstracts and keywords for each book and chapter.
If you think you should have access to this title, please contact your librarian.