Open craddm opened 2 years ago
Currently if you average over multiple recordings from a single participant, the way to handle the recording column is unclear, as clearly you don't want (usually) want to average sets of trials separately for each recording!
Currently if you average over multiple recordings from a single participant, the way to handle the recording column is unclear, as clearly you don't want (usually) want to average sets of trials separately for each recording!