Hello,
A few thoughts on this question.
If you mask with C-R then, implicitly, you are confining yourself to areas
in which activity in C is intermediate between E and R. I suspect that you
would not find this ideal since, presumably, areas in which activity was
significantly greater in E than C, but did not differ between C and R,
would be of as much interest to you. If you mask with E-R then there isn't
this danger but you cannot discount the possibility that you are
identifying some areas in which activity in R is intermediate between E and
C. Perhaps a happy medium would be to use a (E + C) minus R mask within
which you explore your E minus C contrast. A danger of this is that certain
areas, which show a E minus C effect but no C minus R effect would be
diluted in the analysis and excluded from the mask and, therefore,
ultimately lost in your masked E minus C comparison.

As far as I can see, there are some pitfalls to masking. In addition, I
strongly feel that there are some brain regions in which the idea of a
'Rest' condition reflecting 'baseline' brain activity is unconvincing. The
silent soliloquies and daydreams that accompany 'rest' in the scanner may
produce large activations, e.g. frontally. Nancy Andreasen has referred to
the rest condition as Random Episodic Silent Thought. I think that her
acronym should be borne in mind when treating rest as a baseline (in
certain brain regions at least).

Sorry not to be of any practical help.

Paul Fletcher





At 19:56 02/02/01 +0900, you wrote:
>Dear Expert,
>I have conducted a fMRI experiment with 3 conditions, Experimental, Control,
>and Rest.What I want to do is to compare the Experimental condition and the
>control condition, i.e., E-C. But I have come to know that I should do
>masking to avoid misinterpretation of the results. That is, I was told (from
>a local SPM expert) that the deactivated areas, compared to the Rest, should
>be excluded from the analysis.That makes sense. For Inclusive Mask, there
>are two possibilities:   Experimental-Rest or Control-Rest. Which is
>preferable? It seems to me that since I want know about the activation in
>the Experimental condition, E-R would be a good choice. But I was told that
>when one wants to do A-B, inclusive masking should be B-Rest. That means
>C-R, rather than E-R, would be better. I am wondering which path I should
>go.
>
>any comments are welcome, and I thank you in advance.
>
>Chikashi Michimata, Ph.D
>Cognitive Psychology Lab.
>Psychology Department
>Sophia University
>7-1, Kioi-cho, Chiyoda-ku
>Tokyo 102-8554
>Japan
>[log in to unmask]
>
-----------------------------------------------------------------
Paul Fletcher,
Research Department of Psychiatry,
University of Cambridge,
Addenbrooke's Hospital,
Hills Road,
Cambridge,
UK
CB2 2QQ

Tel 01223 336 988
Fax 01223 336 581