Dear Kwok,

Apologies for the delay in replying, I have been away. Just to expand on Karl’s reply – you can extract ROIs using the design matrix you already have. Then for specifying the DCM, you may wish to build a new design matrix with a regressor called ‘task’, which includes every trial, and use that for the C-matrix (driving input). As Karl suggested, you could include a second regressor in your new design matrix modelling just the high load trials, which you would use as modulatory input in the DCM.

Best,

Peter

From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Friston, Karl
Sent: 04 April 2015 11:01
To: [log in to unmask]
Subject: Re: [SPM] DCM advise

Dear Kwok Fu Yu,

Question. As mentioned above (in the aims) I would like to find out the the effective connectivity for high load minus load load (baseline) condition. In this case, should I include both conditions (HL and LL) under the 'input specification)

I will assume that you just have two conditions in your Block design (and that you do not have a rest condition). Under this assumption, it is sufficient to include one input that encodes high load (or the difference between high load and low load). If you want to estimate the effect of memory load on connection strength, then this input will be modulatory and will mediate its effects through the B matrix. I am also assuming that you have visual or auditory stimuli that act as driving inputs (mediated through the C matrix).

In brief, it is only necessary to include the differences between high and low load as a single regressor or input.

With best wishes – Karl

From: #KWOK FU YU# [mailto:[log in to unmask]]
Sent: 04 April 2015 10:35
To: Friston, Karl
Subject: DCM advise

Dear Dr Karl Friston,

Hope this email finds you well. I was hoping you could provide me some advice on my DCM analyses if i've done anything wrong and i've some questions in red:

Background of the study:

• N=42
• Block Design fMRI, 2runs, 160scans/run
• 2 Conditions: High load (HL) and Low load (LL)
• 8 Models for DCM

Aim of doing DCM:

To understand the effective connectivity during verbal working memory task (High load minus low load to find out the actual effect of task modulation; low load acts as a baseline measure) using all 42 subjects

DCM analyses:

• At the preprocessing and 1st and 2nd analyses stage, I concatenated the 2runs together
• Pre-DCM

• I created the VOIs for all the subjects (for participants that has no activation within the VOI, we excluded them)

• Model Specification

• For model specification, it was fairly straightforward and I followed the steps documented in the SPM12 manual
• it