# Read me
## Acquisition
\
Acquisition done in NeuroSpin (CEA, INSERM):\
Commissariat à l’énergie atomique et aux énergies alternatives\
Centre de Saclay\
Bâtiment 145 – P.C. 156\
91191 Gif-sur-Yvette Cedex\
France\
\
MEG setup:\
- Elekta Neuromag 306, with...\
- 102 MEG magnetometers (MEGXXX1 sensors),\
- 204 MEG gradiometers (MEGXXX2/MEGXXX3 sensors).\
\
Acquisition setup:\
- Acquisition frequency: 1000 Hz\
- Online high-pass filter: 0.10 Hz\
- Online low-pass filter: 330 Hz
## Preprocessing
\
MaxFiltering options:\
- Visual bad channels detection and correction (monitored online during acquisition),\
- Automatic bad channels detection and correction (MaxFilter badlimit = 4).\
\
FieldTrip preprocessing options:\
- Epoching between -250 ms an 1 s with respect to stimulus onset,\
- DC filtering: 50, 100 and 150 Hz,\
- Semi-automatic reject of trials with motor artifacts and/or SQUIDs' jumps,\
- Low-pass filter: 30 Hz,\
- Downsampling: 250 Hz,\
- ICA: removing components characteristics of heartbeats, eye-blinks and eye-movements,\
- Epoch demean,\
- Baseline correction (-250 ms to 0 ms with respect to stimulus onset).
## Data files
\
"demographics.xlsx": an Excel file specifying the demographic data of the subjects as well as some experimental informations.\
\
"subjectXX.mat": a MATLAB file specifying the MEG data of a single subject.\
- "meg": a structure with:\
- "trial": a N trials x 306 sensors x 321 time-points matrix specifying the MEG amplitude (in Tesla).\
- "time": a 1 x 321 array specifying the time-stamps (from -250 ms to 1 s).\
- "label": a 1 x 306 array specifying the sensors' name.\
- "trialinfo": a N trials x 4 columns matrix with the columns specifying (in order):\
- col. 1: the trial number (in the context of the full experiment),\
- col. 2: the trial number (in the context of the block),\
- col. 3: the block number (1, 2, 3 or 4; see below),\
- col. 4: what was presented (1 for A, 2 for B).\
!!! the total number of trials (N) varies from one subject to another depending on the number of trials that survive the preprocessing procedure.\
- "seq": a 1 x 4 blocks cell array specifying the full (including missing observations in the MEG data due to preprocessing) binary sequence (1 for A, 2 for B, NaN for explicit prediction).\
- "seqidx": a 1 x 4 blocks cell array specifying the indices of sequence observations for which the corresponding MEG activity is available (i.e. trials that survive the preprocessing procedure).
## Nomenclature
\
"Block" refers to the experimental conditions (counter-balanced order across subjects):\
- Block 1 (repetition-biased): p(A|B) = 1/3, p(B|A) = 1/3;\
- Block 2 (fully stochastic): p(A|B) = 1/2, p(B|A) = 1/2;\
- Block 3 (alternation-biased): p(A|B) = 2/3, p(B|A) = 2/3;\
- Block 4 (frequency-biased): p(A|B) = 1/3, p(B|A) = 2/3.
