Dear colleague
We are working on an 'attention' paradigm. The number of subjects is not quite enough yet, but the session on one subject is divided into several runs so that the number of files is at present 30.
On the 2nd-level analysis, we went into the 'Informed basis set'. The degree of freedom is therefore 87. Results are rather promising, some cluster showing p<0.05 (FWE). However, on certain aspect of attention, the 'Contrat estimates and 90%C.I.' showed poor canonical HRF contribution but larger contribution by derivatives. I wonder if these results are contradictory because these clusters have been detected by using some form of HRF.
Any comments or advice will be greatly appreciated on whether I should accept or reject the results with p<0.05 (FWE), on the basis of poor contribution by canonical HRF.
Kazuhiro Yamada
Department of Neurophysiology,
University of Oita Faculty of Medicine.
Oita, Japan
|