Decoding Neuronal Information
This How-To outlines the sequence of analysis steps that allows judging whether the firing of a population of cells is informative about the conditions in which they were recorded (e.g. attend stimulus A versus attend stimulus B). This classification, or Decoding, approach is showing how we utilize the excellenet functionality and framework of the The Neural Decoding Toolbox that is developed by Ethan Meyers at MIT !
The following will go through matlab code that answers the following questions
-
What is overall accuracy to classify attention conditions from the firing of all neurons across different time?
-
Which of three classifiers provide the highest classification accuracy ? How does cross validation work to judge classification accuracy ?
-
How much information are individual neurons contributing to the classification accuracy ?
-
How many neurons are required to obtain above-chance classification accuracy ?
-
Are the most-informative single neurons sufficient to obtain maximal classification accuracy ?
-
How many trials / observations are required to obtain a sufficient confidence in classification ?
What is overall accuracy to classify attention conditions from the firing of all neurons across different time
Overview
An overview of the NDT Components is linked here .
An overview of the NDT Classifiers is linked here .
Data Analysis
Click on the link below to see the source code and details of how the analysis was done.