[Eeglablist] epochs needed for ICA analysis

[Tara D. Reed]

Hi,

I am designing an experiment using auditory ERPs (P2, N400, P300).  I realize that the more data I collect, the more stable the ICA decomposition will be.

On a previous project, we got good results for group data (over 1000 epochs per condition), but I didn't have enough sweeps (50 epochs/individual) for reliable individual ICAs.  

In the future experiment, I want to increase the number of epochs so that I can get reliable individual data.  To obtain an N1-P2 is easy, as sweeps can be obtained quickly, but it takes much more time for cognitive tasks eliciting the N400 or P300.

With that in mind, does anyone have any thoughts about how many epochs are needed for reliable individual ICAs for cognitive potentials, assuming I have clean data?  For instance, would 500 epochs per individual be enough?  Any thoughts would be appreciated.

Thanks,
Tara

Tara Reed, MCD, CCC-A
University of Texas at Dallas
School of Behavioral and Brain Sciences
Texas Auditory Processing Laboratory