Hi Aditya,

Sorry I'm not Alfonso, but I'm currently running a dyn-ICA analysis, so thought I might be of help:

1) I don't know of a way to do this within CONN. However, a relatively simple possibility is to use the Minimum Description Length (MDL) algorithm in the GIFT toolbox available here: http://mialab.mrn.org/software/gift/. Just set up your ICA analysis in GIFT, respond yes to "Do you want to estimate the number of independent components?", and then GIFT will perform an MDL algorithm and produce a scree plot that will let you know how many independent components you should use before you start to overfit your data. Further info about the algorithm available here: http://onlinelibrary.wiley.com/doi/10.1002/hbm.20359/full

2) It sounds like what you are looking for is SD in seed-based functional connectivity. In this case, you could add a "temporal decomposition" or sliding time window in the conditions screen, extract beta values in each window from the resulting 'SubjectXXX_ConditionXXX.mat' files, and then it would be fairly straightforward to write a matlab script to extract each subject's beta SD. OR, if you want to use the dynamic ICA approach that CONN uses (which I believe takes the approach outlined in Allen et al., 2012; https://academic.oup.com/cercor/article/24/3/663/394348), then you can pull up the temporal components screen in the second level window in CONN, and look at the "variability in factor XXXX" as a function of your conditions or second-level covariates.

Hope this helps!

Cheers,
Jeremy