Hi Maren,
Both files should be [B]exactly[/B] the same if you used the option “arithmetic average” in the average option, with the difference that the default options for the colormap scaling might be different. In the case of the files averaged from the sources of the individual trials, we have all the values and can get the real maximum of the file.
In the case of the “link” created automatically for the averaged recordings, the source values are never calculated explicitly for all the time points. The real maximum of the file is not calculated but roughly estimated, to have a faster file display. The default colormap bounds can be different than for the other file.
If you want to check if the values are the same on the cortex, you can set a fixed colobar range in the colormap options.
For one source model (= one channel file), the two ways of calculating the average should be exactly the same. The “link” attached to the average trial is already there, it does not require any additional calculation step, plus it opens and display faster and many advanced processes are optimized to work on this compact representation (recordings + imaging kernel). In this case, I do not recommend you re-calculate explicitly the average of the sources for the individual trials.
When you want to average sources from different subjects (or different sessions in MEG), you have different channel files/head positions, you need to mix different source models. In this case, the two approaches are not equivalent (average(sources) ~= sources(average)), to get an accurate result you need to average the source values.
Let me know if you need more explanations.
Cheers,
Francois