Hi,
Sounds right! Multiply the beta weight by a 1D waver output file if you want to see what shape your data fit over several instances of a particular trial type. If your only interested in a single trial response, you can also just multiply the Cox Special by the beta weight.
All this really will tell you is what is the size of the assumed hemodynamic response shape (Cox special) present in your data. The response shape of the gamma function and the Cox special are fixed. If you were interested in seeing what kinds of differences in the temporal pattern of activation existed in your data, you'd need a set of basis functions. This allows the response shape to vary among different types of events.
-jim