Center of mass should be weighted by default. However there are various ways to do it. If you cluster an image consisting of activation magnitudes (like percent signal change), you will get it weighted toward the maximum magnitude. You could also cluster an image consisting of statistical values such as t-maps, and the weighted center may be somewhat different. It is a philosophical question which is better, I think most in the Afni camp would go with magnitude.

Finally, if you did want to compute the center of a cluster WITHOUT weighting by amplitude, you can trick the clustering program into doing that. Simply cluster the data first and then use 3dcalc with
-expr 'step(a)'

to convert the image into ones in the clusters and zeros elsewhere, then the center of mass should simply be the unweighted geometric center of the cluster.