Hi Britta, dPSD is not the delay but the finest temporal resolution. If you are going to assume an HRF when you analyze the data, you can set this to (almost) any fraction of the TR (eg, 1/4). If you are not going to assume a shape (ie, use an FIR model), then you should set it to your TR, unless you want to use sub-TR estimation (which you probably dont).

doug

Britta Holzel wrote:

For a rapid event-related design, how do I choose my dPSD, i.e. the delay of onset? I have a TR of 2 seconds and 45 trials for one stimulus type. Can I divide 2 seconds by 45 in order to get the maximum fine grained resolution? Or would it be better to have a larger dPSD, to make sure that I can average over several trials? Thanks! Britta