These are not stupid questions - rather, they go to the heart of what you want to get from the data analysis. The purpose of the analysis is to test some hypotheses. Each hypothesis in 3dDeconvolve is expressed as a statistical test of some regression parameter, or combination of parameters, being different from zero. 3dDeconvolve itself doesn't know what parameters or combinations have meaning to you. So you have to select. The program does output some default tests --- the "-tout" option outputs t-statistics for each individual regression parameter, the "-fout" option outputs F-statistics for each separate "-stim_file" input as well as an F-statistics for the effect of all stim_files together, etc.

These partial and total F's are probably a good place to start. You can also test combinations of these that the program doesn't calculate unless you tell it. For some examples, see AFNI Howto#2: [afni.nimh.nih.gov]

You have to learn to think in terms of what you are testing, in the language of regression coefficients (which is what 3dDeconvolve does). You may want to know where the brain responds to stimulus A. Then the partial F for that stimulus would be a good place to start. You may want to know where the result for stimulus A is signficantly different from the result for stimulus B. This would require a -glt to set up the proper "contrast" between the regression parameter for the A stim_file and the B stim_file.


bob cox