[FieldTrip] Granger analysis

Schoffelen, J.M. (Jan Mathijs) jan.schoffelen at donders.ru.nl
Sat Aug 12 21:41:28 CEST 2017

Hi Matt,

I have never done this myself, but I think that in principle it should be possible to fit a set of models, and then compute the residuals’ variance ratio by hand.
In other words, assuming data1 and data2 to be variables that consist of a single channel:

mvar1 = ft_mvaranalysis(cfg, data1);
mvar2 = ft_mvaranalysis(cfg, data2);
mvar12 = ft_mvaranalysis(cfg, ft_appenddata([],data1,data2));

GC1to2 = log(mvar2.noisecov(1,1)./mvar12.noisecov(2,2));
GC2to1 = log(mvar1.noisecov(1,1)./mvar12.noisecov(1,1));

Best wishes,

J.M.Schoffelen, MD PhD
Senior Researcher, VIDI-fellow - PI, language in interaction
Telephone: +31-24-3614793
Physical location: room 00.028
Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands

On 12 Aug 2017, at 00:05, MATTHEW I BANKS <mibanks at wisc.edu<mailto:mibanks at wisc.edu>> wrote:

Greetings. What is the easiest way to compute non-spectrally resolved Granger causality, i.e. log(var(e_x)/var(e_xy)), where e_x and e_xy are the AR model error terms?


Matthew I. Banks, Ph.D.
Associate Professor
Department of Anesthesiology
University of Wisconsin
1300 University Avenue, Room 4605
Madison, WI 53706
office tel. (608)261-1143<tel:%28608%29261-1143>
lab tel. (608)263-6662<tel:%28608%29263-6662>
fax (608)263-2592<tel:%28608%29263-2592>
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20170812/ab439b2d/attachment-0002.html>

More information about the fieldtrip mailing list