Thank you for your comments. Please see some responses below.
Gang Chen wrote:
> This seems to create some inconvenience for estimating FWHM
> kernel sizes. As multiple comparison correction makes sense
> only in the original EPI space, estimating FWHM would have to
> be done on the EPI dataset instead of tlrc tranformed one. And
> smoothing on the tlrc'ed data after individual subject analysis
> makes the process a little inconvenient.
>
I believe that you use AlphaSim to simulate the dataset that you want to apply a particular statistical threshold to, i.e. the t-stats from the group analysis. That is, you are calculating how likely it is that you would see clusters of a certain size by chance in your group average data (after sampling to talairach, smoothing, and averaging across subjects). Thus the smoothness value should correspond to that group average dataset, not the original raw EPI data. Since smoothness values from raw EPI data will be much lower than those after group analysis, you would be greatly overestimating the statistical significance of your results.
> Ideally you would estimate smoothness for each
> condition/constrast separately based on each condition or
> contrast's EPI data.
>
But why should you penalize a condition with lots of real activations that will result in higher FWHM values? Furthermore, since we are simulating noise, why should the cluster size limit vary across conditions?