On the question of evaluating the effect of a parametric modulation of
a block design:
Thank you Donald and Jonathan on you help so far.
The way I am approaching my question now is to set up three separate
fMRI models (with estimation, contrast, results in a batch).
The results on a single session are illustrated in the attached pdf. I
am setting this up like an event-related design
with parametric modulation for each brain volume.
Left-most, the parametric modulation of the original block consists of
a string of zeros and ones just like a simple block design.
In the middle, the parametric modulation of the original block
consists of a custom vector representing the subjective stimulus
The important aspect is that both models are convolved with the HRF.
(If I had chosen to enter one of them as a regressor, this one would
not be convolved with the HRF)
On the right, I have entered both options as two conditions (even
though there really is only one condition present). I am thinking that
this way I will be
able to ask the question whether my parametric modulator (that
represents subjective stimulus intensity) helps to explain the BOLD
data after adjusting
for the design of the experiment.
Does this sound reasonable?
On May 29, 2009, at 10:24 AM, Jonathan Peelle wrote:
> Donald correctly points out I was too quick to agree to your
> mean-centering summary...if you have a vector X mean centering is just
> X - mean(X), so the range will depend on the distribution of the
> values you had in X.
> X = [1 1 1 1 1 1 1 5 10];
> mcX = X - mean(X);
> ---------- Forwarded message ----------
> From: MCLAREN, Donald <[log in to unmask]>
> Date: Fri, May 29, 2009 at 4:18 PM
> Subject: Re: [SPM] Model Contrast Regressor
> To: Jonathan Peelle <[log in to unmask]>
> (1) Mean-centering to a range of -5 to 5 would assume that you have
> equal frequencies of all stimuli, if you have more stimuli above 5
> than below; then the range would not be -5 to 5.
> (2) on the second point your correct.
> On Fri, May 29, 2009 at 7:11 AM, Jonathan Peelle <[log in to unmask]>
>> Hi Michael
>>> To center them, should they range from -5 top 5? Is that what mean
>> Yes, exactly.
>>> About the question of the orthogonal contrast:
>>> Is this referring to the situation where my custom regressor has a
>>> shape as the block for the block design or if they are strictly
>>> then they would be redundant (explain the same variability)?
>> Right---if two regressors ( = 2 columns in your design matrix) are
>> related at all, it's hard to know if data is being properly
>> assigned...with correlated regressors there is no way to know the
>> "right" regressor to assign activity to. However I think Donald's
>> point was more practical (correct me if I'm wrong!). If you have two
>> columns that are actually orthogonal, then the parameter estimate for
>> 1 doesn't depend at all on the presence of the second. This makes
>> your 'model comparison' easier, because (in the orthogonal case) if
>> you run one model with just your task regressor, and a second with
>> task regressor and your custom regressor, the parameter estimate for
>> the 1 0 contrast (task regressor) will be identical. Thus, you only
>> need to run the model that has the task regressor with parametric
> Best Regards, Donald McLaren
> D.G. McLaren
> University of Wisconsin - Madison
> Neuroscience Training Program
> Office: (608) 265-9672
> Lab: (608) 256-1901 ext 12914
> This e-mail contains CONFIDENTIAL INFORMATION which may contain
> PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED
> and which is intended only for the use of the individual or entity
> named above. If the reader of the e-mail is not the intended recipient
> or the employee or agent responsible for delivering it to the intended
> recipient, you are hereby notified that you are in possession of
> confidential and privileged information. Any unauthorized use,
> disclosure, copying or the taking of any action in reliance on the
> contents of this information is strictly prohibited and may be
> unlawful. If you have received this e-mail unintentionally, please
> immediately notify the sender via telephone at (608) 265-9672 or