Hi Jeff,
[color=#000000]Thank you for your reply. I agree with your understanding about gPPI. In fact, there is indeed a fixation condition between the each task conditions in my experiment. However, I didn't explicitly model this fixation condition as a baseline in the design matrix to avoid over-parameterization (Pernet, 2014). I can specify a new design matrix including the fixation condition for the purpose of calculating gPPI, however, I'm not sure how to set this fixation condition as baseline in CONN. It seems that I can only select the conditions of interests in the analysis step of CONN. I don't know if the unselected condition would be automatically used as a baseline.[/color]
[color=#000000]Bingjiang[/color]
[i]Originally posted by Jeff Browndyke:[/i][quote]Bingjiang,
Does your task model have resting blocks? If not, then what would gPPI use to model a baseline? Without a baseline (rest) condition, it is my understanding that your best bet would be to stick with a GLM analysis and can only correctly report or state contrasts between condition blocks. Connectivity maps in a GLM without a resting condition for each block condition would be somewhat arbitrary to an implied null condition (but not present in the data), but connectivity differences between conditions would still be valid. At least this is my understanding...
Jeff[/quote]
[color=#000000]Thank you for your reply. I agree with your understanding about gPPI. In fact, there is indeed a fixation condition between the each task conditions in my experiment. However, I didn't explicitly model this fixation condition as a baseline in the design matrix to avoid over-parameterization (Pernet, 2014). I can specify a new design matrix including the fixation condition for the purpose of calculating gPPI, however, I'm not sure how to set this fixation condition as baseline in CONN. It seems that I can only select the conditions of interests in the analysis step of CONN. I don't know if the unselected condition would be automatically used as a baseline.[/color]
[color=#000000]Bingjiang[/color]
[i]Originally posted by Jeff Browndyke:[/i][quote]Bingjiang,
Does your task model have resting blocks? If not, then what would gPPI use to model a baseline? Without a baseline (rest) condition, it is my understanding that your best bet would be to stick with a GLM analysis and can only correctly report or state contrasts between condition blocks. Connectivity maps in a GLM without a resting condition for each block condition would be somewhat arbitrary to an implied null condition (but not present in the data), but connectivity differences between conditions would still be valid. At least this is my understanding...
Jeff[/quote]