… Also they should ideally be weighted to sum to nil (.5 -.5 -.5 .5). And re. your question, I think you're right – those two questions are expressed the same numerically. Double contrasts can be a little confusing. Chris
But does the contrast you suggested only get at the mean activation of A and D vs. the mean activation of B and C?
From: Christopher Benjamin <[log in to unmask]<mailto:[log in to unmask]>>
Date: Wed, 27 Apr 2011 21:40:17 -0400
To: FSL - FMRIB's Software Library <[log in to unmask]<mailto:[log in to unmask]>>
Subject: Re: [FSL] contrasting two contrasts
One other thing – the contrast
(A>B) > (C>D) -> (1 -1 -1 1)
Is ambiguous, as it could represent either
(A>B) > (C>D) -> (1 -1 -1 1)
Or
(A<B) < (C<D) -> (1 –1 –1 1)
So you need to inclusively mask the contrast (A>B) > (C>D) with (A > B) to make sure you really are looking at areas are greater in condition A than condition B.
Cheers
Chris
Christopher Benjamin
On Apr 27, 2011, at 5:11 PM, Lara Foland-Ross wrote:
Hi Cornelius,
Thanks so much for your response to my question. But does the contrast you suggested only get at the mean activation of A and D vs. the mean activation of B and C?
Thanks again,
Lara
Quoting Cornelius Werner <[log in to unmask]<mailto:[log in to unmask]>>:
Hi,
you have to setup this contrast on the first level. If your four conditions A, B, C, D are represented by a vector (1 1 1 1), then the contrasts are as follows:
A>B -> (1 -1 0 0)
C>D -> (0 0 1 -1)
(A>B) > (C>D) -> (1 -1 -1 1)
Hope that helps,
Cornelius
Am 27.04.2011 um 18:13 schrieb Lara Foland-Ross:
Dear FSLers,
I have a basic question, the answer for which I couldn't find anywhere on the email list or fsl website. My apologies if I missed the solution to my problem...
I have a simple task design with 4 conditions: A, B, C, D. I first contrasted A>B and C>D, and would now like to contrast (A>B)>(C>D). How does one accomplish this within feat?
Many thanks,
Lara
--
Lara Foland-Ross, Ph.D.
Laboratory of NeuroImaging
University of California, Los Angeles
635 Charles E. Young Drive South, Suite 225
Los Angeles, CA 90095-7334
Research Associate,
Harvard Medical School & Gaab Laboratory, Children's Hospital Boston
office: (617) 355 0400
waltham: (781) 216 1931
fax: (617) 730-0518
|