Hello all, I have a 20-subject sample who performed a sparse-sampling auditory perception task. I modelled the task-hrf using the time derivative and looked at the contrasts across three conditions A,B, C, with respect to baseline. I preprocessed the dataset using two different co-registration steps. In one, I used the functional task means as reference and the structural t1 image as the source to be modified, following which I normalised all realigned functional images to the co-registered structural (without segmentation) in the MNI space.In the second method, I used the structural t1 image as the reference and functional task means as the source and realigned all other functional images for the coregistration step. Following this, I normalised the coregistered functional images to the coregistered task means functional image in the same MNI space, once again without segmentation. I used the same model for both first level and second level analysis in both methods. All other parameters were kept same. However, I get vastly different results in the group activation. From what I understand, both the above methods of coregistration should theoretically yield similar results. However, I do not seem to be getting that. Could anyone comment/explain this discrepancy? Thanks. Megha Sharda -- Megha Sharda Graduate Student National Brain Research Centre Manesar, Gurgaon -122050 India Ph: +919810943202