Hello,
while estimating a model in batch mode I have encountered a weird
problem: having included parametric modulation for one of three
conditions (see below), the design matrix I get contains parametrically
modulated regressors that only contain zeros. As if this was not enough
these regressors spread over all 4 sessions, although the length of the
corresponding vectors do not exceed the respective session.
Here is what I have programmed:
model(1) = struct( ...
. .
. .
. .
'replicated', 1, ...
'nsess', 4, ...
'nscans', [size(dg1,1) size(dg2,1) size(dg4,1) size(dg5,1)],
...
'files', {{dg1 dg2 dg4 dg5}}, ...
'conditions_nb', [ones(1,4).*3], ...
'conditions', [1:4], ...
'regressors_nb', [zeros(1,4)], ...
'regressors', [], ...
'parametrics_type', {{'other','other','other','other'}}, ...
'parametrics', [1:4], ...
'stochastics_flag', [zeros(1,4)], ...
'stochastics', [] ...
);
model(2)=...
.
.
parametrics(1)=struct(...
'name', {'rotpm'}, ...
'exp_type', {'linear'}, ...
'trials', [2], ...
'parameters', {dg_pasrh(:,3)}... %this is a column vector with the
same length as the corresponding onset vector
);
parametrics(2)= parametrics(1);
parametrics(2).parameters = {dg_paslh(:,3)};
.
.
.
conditions(1) = struct( ...
'names', {cond_names}, ...
'onsets', {{dg_pasrh(:,1), dg_pasrh(:,2), dg_pasrh(:,4)}}, ...
'types', {{'events','events','events'}}, ...
'bf_ev', [1 1 1], ...
'bf_ep', [0 0 0], ...
'volterra', 0, ...
'variable_dur', [0] ...
);
.
.
.
Any hints would be greatly appreciated.
Greetings,
Thomas
--
Thomas Wolbers (Dipl.-Psych.)
Universitaetsklinikum Hamburg-Eppendorf
Klinik fuer Neurologie
Cognitive Neuroscience Laboratory
Martinistr. 52
20 246 Hamburg
Germany
tel: +49-40-42803-5778
fax: +49-40-42803-9955
[log in to unmask]
|