[Eeglablist] recover a dataset without triggers recorded

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Mon Sep 18 18:27:14 PDT 2017

Hello Xiaonan, a few quick thoughts below, good luck!

That's a good idea, but not recommended. It's tricky to synchronize those
different events streams from different clocks without having anywhere from
~50ms to 200+ ms of variability, even if you are very lucky and smart. It
might be possible if you knew times when you are sure the clocks were
synched, or times where it is likely that they were synched. It might be
possible if you have an external event (such as notes and times from a
researcher taken during recording).

If you have an ERP paradigm, I would suggest just learning from this
problem, and forgetting these two subjects. If you have a non-ERP
continuous recording paradigm, then you might be able to at least mark of
the likely beginning and end of multi-minute periods, such as the start and
end of blocks in your protocol.

If you're very good at signal processing, and can estimate single trials
EEG and ERP in these subjects, you might also be able to somehow pull the
ERPs out, using the EEG dynamics as a way to guess when the onsets of the
stimuli were. For example, perhaps temporal ICA could help you isolate
event/stimulus-onset dynamics for your 2 subjects.

Last, and this is only if you are a genius, you could use all your other
data from all other subjects (EEG+exact timings of events) to stimulate and
predict when events occurred, and then use that as a model to guess when
events occurred in your two 2 subjects.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20170918/9d2dac50/attachment.html>

More information about the eeglablist mailing list