Biomedical Engineering Reference
In-Depth Information
Chapter 7
Methods for Studying Functional Interactions Among
Neuronal Populations
Nandakumar S. Narayanan and Mark Laubach
Abstract
How do populations of neurons work together to control behavior? To study this issue, our group
simultaneously records from populations of neurons across multiple electrodes in multiple brain regions
during operant behavior. Here, we describe methods for quantifying the relationship between neuronal
population activity and performance of operant behavioral tasks. We describe statistical techniques, based
on time- and trial-shuffling, that can establish the significance of correlations between multiple and simul-
taneously recorded spike trains. Then, we describe several approaches to studying functional interactions
between neurons, including principal component analysis, cross-correlation analysis, analyses of rate cor-
relations, and analyses of shared predictive information. Finally, we compare these techniques using a
sample data set and discuss how the combined use of these techniques can lead to novel insights regard-
ing neuronal interactions during behavior.
Key words:
Neural coding, neural ensemble, population activity, correlated variability, JPSTH,
multi-electrode recordings.
1. Introduction
Recent developments in electrode recording technology have
facilitated recordings of neuronal population activity
(1, 2)
.Such
techniques can generate data from hundreds of simultaneously
recorded neurons in multiple brain areas
(3, 4)
. To analyze such
data, a multivariate approach is applied to firing rates of simulta-
neously collected neurons
(5)
. This approach can generate stable
predictions from neuronal ensemble data about a behavioral vari-
able or can be used as control signals for prosthetic devices
(3)
.
Analyses of neuronal population data do not automatically
offer insight about how neurons interact during behavior. This is
