information-theoretic Spike-Triggered Average and Covariance (iSTAC) estimator for neural receptive fields
Description: Estimates a set of linear filters that best capture a neuron's input-output properties, using an information-theoretic objective that optimally combines information from the spike-triggered average and spike-triggered covariance. The filters can be considered as the first stage in a linear-nonlinear-Poisson (LNP) model of the neuron's response. They are sorted by informativeness, providing an estimate of the mutual information gained by the inclusion of each filter.
- Command line: clone the repository from github (e.g.,
git clone firstname.lastname@example.org:pillowlab/iSTAC.git)
- Browser: download zipped archive: iSTAC-master.zip
Launch matlab and cd into the directory containing the code (e.g.
Examine the script
test_iSTAC_script.mfor a line-by-line tutorial on how to use the code contained in this package, which goes through several simulated examples.
The primary function used for estimating the filters is