regarding your first question:
It depends. It is possible to model small blocks in the context of an
ER-design. However, the duration of these blocks usually varies along a
certain dimension, for example and most often reaction times. Using
blocks of varibale duration has also been termed "variable epoch"
approch and may prove useful to model higher order cognitive processes
as single cell recordings provide evidence of sustanined activity in
"these" areas between stimulus presentation and behavioural output.
This is discussed in more detail here:
This may be a good starter to base your decision upon.
Regarding you second question:
Mean centering of regressors is automatically provided by SPM5/8. There
is no need for you to do this.
I hope this helps and good luck
Am 05.07.2011 10:02, schrieb Ben Becker:
> Dear SPMers,
> I’m analyzing an event related fMRI learning task. The first level matrix comprises of one regressor coding the correct responses with a parametric modulation (repetition). I’ve written a script to generate matfiles containing the onsets& durations for the correct responses and the parametric modulation. Now I have 2 questions:
> 1. Duration of the regressor coding the correct responses
> I’ve modeled the duration of the correct responses as the stimulus-response duration of 2500ms. In the context of an event related analysis would it be more reasonable to model the duration with duration = 0ms?
> 2. Parametric modulation& mean centering
> The script I’ve written codes the parametric modulation (repetition) with values from 0-5 (0 = first presentation, 5 = 6 presentation). Should the parametric regressor be mean centered (values -2 -1 0 1 2) or does SPM8 automatically mean center the parametric modulator?
> Thanks in advance& kind regards,