[Eeglablist] interpretation of weights on scalp maps
Scott Makeig
smakeig at ucsd.edu
Thu Feb 27 16:09:58 PST 2003
You are correct. The plotting convention in EEGLAB is that 0 = green.
Thus, component map projections colored red and blue may be equal in
amplitude (but opposite in polarity). The (back)projection of a
component to the scalp data (e.g., in uV) is the product of this map
with the component 'activation' time course. So, pos*pos=pos,
neg*neg=pos,pos*neg=neg,... The same applies to any linear decomposition
of the data, properly considered.
Scott Makeig
Jed Meltzer wrote:
> I have a conceptual question about the weights in an
> ICA decomposition. As far as I understand it, each
> point in time on each channel is multiplied by the
> inverse weight and sphere matrices to output each
> point in time in the component activations. Some of
> these weights are negative. Does this mean that a
> positive deflection in the component corresponds to a
> negative deflection in the channel?
>
> The reason I ask is because I don't know how to
> interpret the scalp maps of components. Are red parts
> and blue parts equally important? That is, do they
> both represent that the component maps strongly onto
> the channels in the colored region? That would mean
> that the green regions are the ones that are not
> affected strongly by the component. Furthermore, one
> could then infer that if a red region and a blue
> region are close together, then the underlying source
> may be modelled by a single dipole located in between
> the two colors.
>
> Or am I understanding this incorrectly? Does blue
> mean minimal projection?
>
> -Jed Meltzer
> Yale University
>
> __________________________________________________
> Do you Yahoo!?
> Yahoo! Tax Center - forms, calculators, tips, more
> http://taxes.yahoo.com/
> _______________________________________________
> Eeglablist mailing list
> Eeglablist at sccn.ucsd.edu
> eeglablist page: http://www.sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsub at sccn.ucsd.edu
More information about the Eeglablist
mailing list