[Eeglablist] Spatial resolution

[Scott Makeig]

This is a good research question. However, the answer will depend on the 
statistics of the data: how many dimensions (channels) and time points, 
the data quality (how much true noise?), etc.

ICA can be quite sensitive: Decomposing a very clean set of grand-mean 
ERP data, I once found an ERP component pair that selected for stimulus 
color (red, blue)!

Scott


Lars O.D. Christensen wrote:
> Hi,
> 
> I am trying to understand ICA and its possible applications. It's very
> interesting.
> 
> One issue of importance for me is what the best spatial resolution one
> can obtain is, that is, e.g. in terms of anatomical distances how well
> can sources be separated (how close together can the sources be before
> they simply overlap/merge, if ever ...) ? A related question would, for
> example, be whether it is possible with ICA to separate sensory input to
> S1 from its effect on M1 (motor cortex) since these areas are clearly
> distinguisable (by the central sulcus) but anyway only separated by a
> few centimeters, or maybe only millimeters (at most).
> 
> If anybody could tell me something about this, please do.
> 
> Best regards,
> Lars