Hi Laura,
- comments inserted in your text -
>The contrast for when I use the hrf+time derivatives has got, in fact,
>more columns:
>To see only the effect of task A for example (which is repeated twice,
>as
>also tasks B and C) I used the contrast:
>[0 0 0 0 1 -1 1 -1 0 0 0 0]
>With "B C A A C B" I meant the order in which I acquired the fMRIs and
>herefore for consistency I put the data from those sections in the same
>order in the large final density matrix.
If I understand you right, this is a very unusual thing to do. Rather than
modelling each block (A-C) as a single session (which is - if I understand
correctly - what you did) it is common practice (there might be
advantages...) to create one colum of onsets for each block, even if it is
repeated more than once. So, e.g. you could define three colums, one for A,
B, and C each. But this is a different story.
>I meant to say that I have got a large density matrix which includes 6
>sessions, and therefore 6 small matrices. In the case of only hrf every
>small matrix has got just one column or regressor (apart from the
>constant
>always put at the end by SPM).
>In the case of hfr+time derivative the small matrix has got in fact two
>columns.
>I am not sure what you mean when you ask if I included the dispersion, I
>guess I didn't otherwise I would have known ... At least I hope.
Okay, forget the dispersion thing for now.
>I must say that the change of perspective you mention is quite
>interesting.
>But it means I haven't got a clue of how brain really works.
If you know who has, please forward to me ;-)
>Does it really happen sometimes that it is more active at the
>"off-times"
>instead of at the "on-times", or is it just an indication that something
>has
>gone wrong in the subtraction between data on the on and off condition?
Again, what is 'off', and what is on, is relative. But - if anything - you
consider what someone's brain is doing when the person is not engaged in an
explicit task, this might be considered as 'rest', and this is what Paul
hints at when citing the 'default mode' (Marc Raichle's name is usually
associated with this): there are in fact brain areas that are more active
during rest than during perception and action and - going towards less
activity - sleep, anaesthesia, vegetative state, generalized spike and wave
discharges [ :-) ] etc. Consequently, when the brain leaves this state of
rest (which by the way is quite variable in itself, too), the particular
"default mode brain areas" would than be less active ('deactivate') during
[almost] any given task.
But you do not even have to refer to the default mode of brain function: say
you define 'lights are off' as a condition, then - compared to when this
condition is not active - the lights are deactivated. If you define your
condition as 'lights are on', then they would be more active during this
condition but you would also either define your regressor differently - or
your design, which both means a change in perspective since the actual
'paradigm' of lights switched off or not has not changed.
I guess this is enough to cause confusion,
Sorry,
Helmut
-----Original Message-----
From: Helmut Laufs [mailto:[log in to unmask]]
Sent: 01 July 2005 08:28
To: Mancini,Laura
Cc: SPM (Statistical Parametric Mapping)
Subject: Re: deactivation
Hi Laura,
could you please clarify what your regressors in your model are (you say
you
have 3 conditions and model either hrf or hrf+ time derivatives [incl.
dispersion?], but it appears you always have 6 colums in your design). I
cannot work it out, even should you have put one colum per block (B C A
A C
B), which gives six colums, including the derivatives should give you
more...
In principle: onbe [/another] way to look at 'deactivations' is this:
change
your perspective and see it as: the set of brain areas that
'deactivates'
during your respective condition is in fact more active at the
"off-times"
of your condition.
Maybe this is a start?
Helmut
----- Original Message -----
From: "Laura Mancini" <[log in to unmask]>
Sent: Thursday, June 30, 2005 6:52 PM
Subject: deactivation
> Dear SPMers,
>
> I wonder if you can help soving a puzzle.
>
> I have 6 fmri acquisition, two for conditions A versus rest, two for B
> versus rest, two for C versus rest. The order of the acquisitions is:
> B C A A C B A = only hand movement
> B = only speech production
> C = hand mov + speech prod
> I did only t-tests, not F-tests.
> If as basis set for the SPM analysis I consider only the hrf I have
the
> following.
> When I consider the contrast [0 0 1 1 0 0] I see a huge activation in
the
> motor cortex, while with [0 0 -1 -1 0 0] there is little or no
> (de)activation
> When I consider [1 0 0 0 0 1] or [0 1 0 0 1 0], I see little
activation in
> some parts of the brain, while with [-1 0 0 0 0 -1] or [0 -1 0 0 -1 0]
I
> see
> large (de)activation almost everywhere in the brain.
>
> If as basis set for the SPM analysis I consider only the hrf+time
> derivatives I have the following. When I consider the contrast [0 0 1
> 1 0 0] I see much less activation in the
> motor cortex, while with [0 0 -1 -1 0 0] there is maybe more
> (de)activation
> When I consider [1 0 0 0 0 1] or [0 1 0 0 1 0], I see much more
activation
> in some parts of the brain, and with [-1 0 0 0 0 -1] or [0 -1 0 0 -1
0] I
> see less (de)activation.
>
> I am quite puzzled. Has anybody find something similar?
> Is it more likely that there was a problem during the acquisition or
> the analysis of the data, or that it is a real behaviour of the brain?
>
> Many thanks,
> Laura
>
>
> **********************************************************************
> This email is confidential and intended solely for the person or
> entity to
> whom it is addressed. If this email was not intended for you please
> notify the UCLH Mail Administrator at [log in to unmask]
> This footnote confirms that the email and attachments contained no
viruses
> when they left UCLH.
>
**********************************************************************
This email is confidential and intended solely for the person or entity
to whom it is addressed. If this email was not intended for you please
notify the UCLH Mail Administrator at [log in to unmask]
This footnote confirms that the email and attachments contained no
viruses when they left UCLH.
