[FieldTrip] Average referencing and demeaning prior to beamformer

[Schoffelen, J.M. (Jan Mathijs)]

Hi Maris,

> Question 1: after I combine the the two cut-out time windows to calculate the common filter, do I need to average reference the combined data set prior to calculating the beamformer filter?

The referencing procedure operates across channels, not time points. Thus once the data has been average referenced (and as long as you don’t throw away channels in the mean time, or do something funky that affects the channel mixing, e.g. remove certain topographical patterns based on an ICA), the data does not need to be re-referenced anymore.

> Question 2: Do I need to separately average reference each of the two cut-out windows prior to applying the beamformer on each of the cut-out data segments? 

See above, the rereferencing is independent of time.

> Question 3 on demeaning: do the cut-out combined/separate time windows should also be demeaned or demeaning doesn't make any difference for beamformer source reconstruction?

The LCMV beamformer requires a covariance matrix. This is generally constructed by the user with ft_timelockanalysis. Unless you specify otherwise, the data will be demeaned by default prior to the covariance computation.

Best,
Jan-Mathijs


> 
> Thanks!
> 
> Maris
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip