Dear Amit,
here are my replies to your questions:
1. No, intrinsic connections do not have to be significant to allow
for an interpretation of bilinear effects. See the
attention-to-motion example by Friston et al. 2003 NeuroImage where
the V1->V5 connection was basically zero except for the presence of
motion (as a bilinear effect) that massively enhanced this connection strength.
2+3. Note that in the DCM state equation the bilinear effects are
*additive*, not multiplicative. That means, at any point in time the
effective strength of a particular connection is the *sum* of the
corresponding entry in the A matrix and the corresponding entry in
the B matrix times the value of the modulatory input u at that time:
dz/dt = (A+uB)z + Cu
Best wishes,
Klaas
At 17:51 26/04/2006, you wrote:
>Hi all,
>
>I have two scenarios from some DCM analyses that I would like to get
>suggestions/thoughts on
>how to interpret...all help appreciated!!!
>
>1. Signficant modulatory (bilinear) effects on a pathway with
>non-significant intrinsic connectivity.
>Can one interpret the modulatory effects as activating a pathway, or
>does intrinsic connectivity
>have to be also significant to do so?
>
>2. Say the bilinear effect is positive and significant and the
>intrinsic connectivity is negative and
>non-significant. Does that mean an "inhibitory" pathway is
>activated, or because the intrinsic
>connectivity is non-significant, that the pathway is activated but
>of unkonwn excitatory vs
>inhibitory nature?
>
>2. What if the bilinear effect is positive and significant and the
>intrinsic connectivity is negative
>and significant, can one argue that an "inhibitory" pathway has
>therefore been activated?
>
>much, much thanks!
>
>Amit
|