[Eeglablist] how to recover the cluster-average weight vector from the STUDY.cluster structure

[arno delorme]

The cluster average map stored for each cluster is

STUDY.cluster(x).topo

This is all detailed at

http://sccn.ucsd.edu/eeglab/clusttut/clustertut.html#cluster

> However, I
> can't find the weight vector that defines the scalp maps.  All I can  
> find is
> STUDY.cluster.topo, which seems to be the complete scalp map, ready  
> for
> plotting, but useless for spatially filtering the EEG data...

Averaging weight vectors does not make sense. Weight vectors (rows of a
ICA weight matrix) allows to compute component activity by using a
linear combination of multiple channel activities. However, we usually  
only visualize the column of the
inverse weight matrix (and not the weight vector itself) which is the  
back projection of a component to the
channel space. This vector contains values of how a component is  
weighted on each channel.

This is detailed on this page:

http://sccn.ucsd.edu/~arno/indexica.html


> Finally, the documentation states "the cluster.topoall cell array  
> contains
> one element for each component and each condition."  What  
> information for
> these elements contain?

Columns of the inverse weight matrices (one for each component).

Best,

Arno