Dear Andrew,
> We would like to compare the effects of free viewing, and 2 types of
> instructed encoding, on images that vary along 2 dimensions (valence and
> complexity).
>
> The instructions are kind of complicated, and so I wondered if it would be
> possible to directly compare activations between different runs of several
> blocks. So, we would have a run of ~ 4 mins of free viewing, with 30s
> blocks of two types of pictures (say a, b), so we would have a run
> a1b1a1b1a1b1 etc. Then we would have a run of 1 type of instructed on the
> same images a2b2a2b2 etc, and then a run of a different type of instructed
> encoding a3b3 etc. We then have a final block of instructed encoding, but
> only for neutral (a) images, so this would look like a4a4a4a4 etc..
>
> My question is this. If we counterbalance the order of the runs, is it
> valid in SPM to directly compare levels of activation between runs? i.e.
> compare all blocks in a run of instructed encoding, versus all blocks in a
> run of free viewing? Or, would one have to alternate 30 secs of differing
> encoding instructions within the same run?
In the following, I assume that you mean by a 'run' a session. You have
a 2x2 factorial design, where one factor is the image type and the other
the action. The latter factor has 4 levels, free viewing, inst. coding
1, instr. coding 2 and instr. coding 3, where you have omitted to scan
'b4'.
My feeling is that it is ok to scan each level of the factor action in a
counterbalanced fashion over subjects. However, it seems to depend on
the question you have, whether you should include an explicit rest
condition here. The reason is, if you want to find e.g. a difference
between encoding 1 and encoding 2 irrespective of the factor image, then
it might be that with your current design this effect will be captured
by the session effect. On the other hand, if you're after the effect of
image a vs. b or the interaction between image and action, you should be
doing fine.
>
> The reason we would like to do this is that the instructions are quite
> complex and we don't want too much "switching" going on within a run.
> However, if we can't directly compare across different runs, is there a
> disadvantage in lengthening task blocks to say 60 secs, rather than 30 in
> terms of power?
I don't really know, whether 60 sec epochs will be disadvantegeous as
compared to 30s epochs. Your paradigm frequency is then shifted to a
lower frequency and if there are more noise components around this
frequency, you will naturally experience a loss in sensitivity. But of
course, this might only be a marginal loss, depending on your effect
size. Others might want to comment on this.
> Alternatively, should we interleave 30 sec blocks of task
> with 30 sec rest periods, within the same run, to allow us to present
> instructions on the screen and to make it easier for subjects to task-switch?
If time allows, I would try this. Not only that you can scan different
tasks in one session, but also avoid the above mentioned potential
problem.
Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-(0)20-7833-7478
FAX : -7813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|