We issued a warning about bug 1971 in January, 2017. This bug was detected when rejecting continuous portion of data. We originally thought the bug was more general than it actually is. It turns out that almost no users are affected.
The bug arises when rejecting continuous data in a script. For example calling eeg_eegrej and including sample point 0 generates the bug.
EEG = eeg_eegrej(EEG, [0 1604;6907 7896;9496 10225])
In this case, the first portion of data is rejected but the latency of the events is not accurately calculated (it is not subtracted from event latencies). The important points are
- This only happens from the command line when manually entering 0 as the beginning of the first region
- When copying EEGLAB history, this type of call is never present
- When using the EEGLAB graphic interface, it is not possible to encounter this bug
The bug has been fixed in EEGLAB 14 and later revisions of EEGLAB.