It might be good to distinguish the beta weights
from whether the statistics show significance.

Gang's point about the center/mean value of the
modulation parameters is directed at the meaning
of the beta weights for those terms. That is
important when interpreting the betas or when
using them in a subsequent group analysis.


Christine's question (2) about whether activity
follows the modulation parameters seems directed
at the significance of the parameters (and might
go along with her interest in the correlation
coefficients).


> 2) For the unmodulated effect, does that mean
> that those are voxels where activity was equal
> (i.e., not different) across all trials of
> interest?

No. Having an unmodulated effect simply means
that there is a mean effect (which is why the
AM2 parameters were demeaned). It does not say
anything about whether there is also a modulated
effect. They are orthogonal.

When deciding whether the BOLD effect follows the
modulation parameters, the average/mean effect
could be significant or not. It could be positive,
negative or zero.

Conversely, having a modulated effect does not
imply anything about whether there is a mean
effect. Of course that might seem strange. But
it would be from a psychological/physiological
perspective, not from a mathematical one.

Just hoping to confuse the discussion...

- rick