First data set

1) Compare t-maps for the different headcoils with and without projector

2) Do regional analysis to compare headcoils and with/without projector - do this by only looking at areas that we know are active in the task - do that ideally by doing conjunction analysis in SPM, or summing thresholded maps, or 'AND'ing thresholded maps from all four runs.

3) Do regional analysis to compare false positive rate for different headcoils and with and without projector - to do this must identify areas NOT involved in the task, perhaps by converse of (2) above but with very low threshold.

4) Did all the data normalize well - you want to check how well the mean functional images are aligned with each other. Different headcoils may alter alignment.

5) Does using the motion corrected (MoCo) runs produce better results? Is this consistent across the four runs?

Second data set

1) Normalizing data from the 2.5 mm voxel condition will be a challenge - to do this we will probably need to create a mask since the data only covers part of the brain. Will need to look up how to do this for SPM. Note that SPM has a listserver called SPM central where you may be able to look up prior enquiries about this.

2) We will need to carefully assess the registration and normalization of the different runs. There are likely to be differences.

3) The design matrix is going to be more complex to create a requires different events for each run. Mike can explain the files. You should concentrate on attended locations first.

4) One option is to define a statistical test in SPM that is sensitive to any changes between the seven conditions. Another is to pull out seperate t-tests for each of the conditions. Someone should try each of these.

5) Again we will want to do both regional analyses and whole brain maps to compare the conditions.