Memtrain MEG

# Memtrain MEG: raw recordings Raw MEG recorded while participants viewed object images before and after eight weeks of web-based mnemonic discrimination training. These are the unprocessed continuous recordings behind the analyses reported in Rong et al. (in preparation). ## What is here - 31 participants, 2 sessions each, giving 62 recordings. - ses-1 is the pre-training session, ses-2 the post-training session. - One continuous recording per session, stored as split FIFF files (split-01, split-02, ...). MNE-Python reads the chain from split-01. - No filtering, no MaxFilter, no epoching has been applied. The only header change is anonymisation. ## Recording - Elekta Neuromag TRIUX, 306 MEG channels (204 planar gradiometers, 102 magnetometers). - 1000 Hz sampling, acquisition band-pass 0.03 to 330 Hz, 50 Hz line frequency. ## Task Passive viewing with an orthogonal target-detection task. 96 object images per session, 48 trained and 48 untrained, each shown centrally for 500 ms at 7 degrees visual angle on a grey background with a fixation cross, with a 250 to 450 ms inter-stimulus interval drawn at random in 50 ms steps. Each object was repeated 30 times per session, spread over 15 runs of about 263 s. Every 4 to 6 presentations a target image (a hammer) appeared and required a button press, followed by a longer 1500 ms interval to allow blinking. Target trials are marked catch in events.tsv and are not part of the analysed set. ## Events events.tsv lists every stimulus presentation and every block marker. - onset is taken from the photodiode-gated trigger, which marks the true screen onset. - trial_type is trained, untrained, catch, block_onset or block_end. - object_id is the 0 to 95 object identity. 0 to 47 are trained objects, 48 to 95 are untrained. - value is the STI101 code. Stimulus codes are 2^15 minus the image number, so object_id = 2^15 - value - 1. - response rows are button presses recorded on the second trigger channel. They were made during catch detection and are the only behavioural measure collected in the scanner. The epoched dataset on OSF does not contain them. ## Anonymisation Participant identifiers and birth dates were removed from the FIFF headers, and every recording date was shifted back by the same fixed offset. Absolute dates are therefore meaningless, while the interval between a participant's two sessions is preserved exactly. ## Related data The epoched, temporally binned arrays used for the analyses, together with the figure source data, are on OSF at https://osf.io/q84bz/ (folder MEG_epochs). Those files are derived from the recordings here. ## Coverage Participants: 01, 02, 03, 04, 08, 11, 13, 18, 20, 21, 22, 26, 29, 30, 34, 35, 38, 39, 49, 51, 52, 53, 55, 56, 57, 62, 63, 65, 66, 67, 69. All have both sessions. Contact: Boyan Rong <[e-mail]>.