Dear Mina and Christopher,
thank you very much for your replies!
> The model you have described will allow you to test connections modulated
> by the main effect of stimulus (your question 1) but not the interaction
> (your question 2). One way to test your two questions simultaneously
> would be to create 2 regressors, one corresponding to the main effect of
> stimulus as you describe above (all A trials) and the other encoding the
> interaction (i.e. all A trials in condition 1). You may find it useful to
> create a third regressor encoding alll trials ('photic') as proposed in
> the example dataset.
This is precisely what I did, but our reviewers argued that in the
context of a 2x2 factorial design, the modulatory regressor encoding the
trials of condition A1 does not specify for the interaction, but rather
just for the simple main effect of A1.
In my previous email I was describing an approach that tried to
reproduce the models described in e.g. Stephan KE et al, J. Biosci.
31(4), October 2006, or Heim S et al. Human Brain Mapping, Published
Online: Dec 19 2007.
But as far as I understand, also based on your comments, with such an
approach you cannot test both the main effect and the interaction at the
same time.
So, what about a third type of approach with:
- 1 regressor encoding all trials -> 'ALL'
- 1 parametric modulation regressor encoding for the main effect of
stimulus (A: +1; B: -1) -> 'STIMULUS'
- 1 parametric modulation regressor encoding for the main effect of task
(1: +1; 2: -1) -> 'TASK'
- 1 parametric modulation regressor encoding for the interaction
(STIMULUS.*TASK) -> 'INTERACTION'
In the DCM:
ALL is the only input to the model, while 'STIMULUS', 'TASK', and
'INTERACTION' are all allowed to modulate the specified intrinsic
connections.
Is this a valid approach?
Thank you and all the best,
Marco
--
Marco Tettamanti, Ph.D.
San Raffaele Scientific Institute
Facoltà di Psicologia
Via Olgettina 58
I-20132 Milano, Italy
Tel. ++39-02-26434888
Fax ++39-02-26434892
Email: [log in to unmask]
