Dear all,<div><br></div><div><br></div><div>we are using DCM (DCM10, SPM version 4010) to investigate the motor network comprising primary motor cortex (M1), premotor cortex (PMC) and supplementary motor area (SMA). We are basically following the approach of [1], defining M1, PMC and SMA as a fully connected network for both intrinsic (DCM.a) and modulatory (DCM.b) connections.(For DCM.b, we set the self-connections to zero.)</div>

<div><br></div><div>In contrast to [1], we don&#39;t have a pacing cue for the hand movement. Instead, our protocol used auditory commands (&quot;Move&quot;,&quot;Stop&quot;). The commands were given just once at the beginning of the block, so it is not a pacing signal. Due to this lacking input region, we must use some of the motor regions as input (DCM.c). We defined different models using three kinds of input regions:1) PMC and SMA , 2) SMA  , 3) PMC.</div>

<div><br></div><div>We observed that the input regions seem to have a quite important role: in the models where a certain region is defined as input region, it gets inhibited from other regions. In contrast, this inhibitions are not observed in the models where the same region is not defined as input region. Although it is not surprising, that the input region is quite important and will determine the behavior of the propagating network dynamics, we would like to know, if there is a way to define the input in a more &quot;neutral&quot; fashion. </div>

<div><br></div><div>A possible answer to this question is to define all regions as input to equalize the influence over all regions. Unfortunately, the resulting modulatory connections for this approach became less significant (t-test for 20 subjects). Another possible solution is to define no direct input region. In this case, the DCM estimation seems not to be able to deal with this case (the estimation returns for all subjects the same value for all non-diagonal intrinsic connections and zero for all modulatory connections).</div>

<div><br></div><div>To summarize:</div><div>1) Is it expected that input regions tend to get inhibited ?</div><div>2) Is there a way to define input in a &quot;neutral&quot; way ?</div><div>3) Is there a relation between number of input regions and (significant) connection strengths ?</div>

<div>4) Is DCM supposed to deal with &quot;no direct input&quot; models in the future ?</div><div><br></div><div><br></div><div>Any comments or suggestions are highly appreciated.</div><div><br></div><div><br></div><div>
Best Regards,</div>
<div><br></div><div>Sebastian</div><div><br></div><div><br></div><div><div>[1] Grefkes, C., Nowak, D. A., Eickhoff, S. B., Dafotakis, M., Ku, J., Karbe, H., &amp; Fink, G. R. (2007). Cortical Connectivity after Subcortical Stroke Assessed with Functional Magnetic Resonance Imaging, 236-246. doi:10.1002/ana.21228</div>

<br class="Apple-interchange-newline"></div>