Aaaron,

To some extent I misunderstood your original post since you didn't mention modulation in that context. Now I get a better picture of what you're trying to achieve. Thanks for the clarifications.

> I've ran this model and gotten some results that fit our expectations.
> However, when running the analysis with one condition using the strength
> numbers as a modulating variable, what do I need to put in as the seed time series?
> ...
> The issue I'm running into is I have six of these -- one for each memory
> strength. When using 1dplot to look at these files they are all very similar,
> with only very slight differences. If I put all six of them into the correlation
> model 3dDeconvolve dies on me due to problems of multicollinearity.
>
> To get around this I averaged them together using 1deval and put in the
> averaged seed time series as the seed time series regressor. Is this a valid
> thing to do?

How about this: Run 3dSynthesize one more time (not 6 times) to isolate everything but memory strengths and then get a 1D file of the seed time series for all memory strengths of interest through subtraction?

Gang