Hi Mark,

Thanks for the reply. I will try and use both cope and/or z-stat images in my machine learning algorithm.

One more question, when I used the cope images in the group level analysis in SPM8 using the same peak and extent threshold(as I did for feat), I could not replicate the fsl finding. I know feat's higher group level analysis should be more complex than just plain ttest on the cope images, but I would still like to know what goes in FEAT's group level analysis if you don't mind.

Thanks again,
Binod