Hi, this is the response your email got last time > Hi - what was the error message? > > If your task C is filling up the "gaps" - is is in effect the > baseline, then you should exclude this, if this is a first-level > analysis, as the data and EVs all get demeaned at first level. > > Cheers, Steve. > For further clarification: for first-level analysis feat de-means all the data. The implication is that your EVs only need to model signals over and above the mean baseline signal C, i.e your design only needs A and B. A [1 0] contrast then is the 'where is signal under A > signal under C' contrast and similarly for B. A [1 -1] contrast test for A>B. hope this helps christian On 3 Jul 2006, at 21:47, Joongnam Yang wrote: > Hi all, > > I posted this question, but didn't get any response. > > I have three conditions, which are two tasks (A, B) and one control > (C). > I wanted to compare A vs. B, A vs. C, B vs. C. > I came up with three design files for each condition, and did a > contrast, say A vs. B, but FSL gave me an error. > But I found out that if you load only two design files for A and B > conditions, then A vs. B contrast was allowed to go. > > I don't understand what's going on with FSL model design. > Would you someone clarify on this? > > Thanks in advance. > > Nam. -- Christian F. Beckmann Oxford University Centre for Functional Magnetic Resonance Imaging of the Brain, John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/ Phone: +44(0)1865 222551 Fax: +44(0)1865 222717