Thomas -
> We have modeled the hdr to two classes of events (A and
> B) with a canonical hdf with temporal and dispersion
> derivatives. We thus have 2x3 regressors in our design
> matrix. Previous papers, mails on the list, and the
> respective manuals state that we could now assess with
> t-contrasts (a) the magnitude of activation differences
> between A and B by a contrast of the canonical hdr
> parameters, (b) latency differences with contrasts of the
> temporal derivatives, and (c) duration differences with a
> contrast of the dispersion derivatives.
This is not strictly correct - one cannot interpret differences
in the temporal/dispersion derivative parameter estimates
in terms of latency/duration, without additional information
about the canonical HRF parameter estimate - see, for example,
Henson et al (2002), Neuroimage, 15, 83-97.
Indeed, an F-contrast of the type you detail below is, in fact,
what is normally performed. You can interpret any differences
in the F-map as regions that show a different HRF shape between
your two conditions (at least, "shape" that can be captured by
the three basis functions). To get an idea what form that difference
takes (eg activation/deactivation in peak response, latency
difference),
you can look at the fitted responses for each voxel. (However, to
make inferences about specific differences (eg latency) you still
need to conduct further tests like that described in Henson et al
(2002)).
> However, we are now wondering whether it would be feasible
> to assess "activation differences" between A and B with an
> F-contrast, including all three single contrast
> specifications given above:
>
> F-Contrast:
> [A + (temp. + disp.) derivatives] > [B+ (temp. + disp.)
> derivatives]
> 1 0 0 -1 0 0
> 0 1 0 0 -1 0
> 0 0 1 0 0 -1
>
> We know that it is statistically feasible to do this, but
> we are wondering whether we could conceptually interpret
> the results as activation differences, given the onset and
> duration differences between the hdr at different voxels
> and for different events.
--
---------------------------------------------------------
DR RICHARD HENSON
Institute of Cognitive Neuroscience
& Wellcome Department of Imaging Neuroscience
University College London
17 Queen Square
London, WC1N 3AR
England
EMAIL: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~rhenson
TEL1 +44 (0)20 7679 1131
TEL2 +44 (0)20 7833 7472
FAX +44 (0)20 7813 1420
MOB +44 (0)794 1377 345
---------------------------------------------------------
--
|