# Behavioral data All of the files documented below reside in the `/behavior` folder. ## Group data * `questionnaires.txt`: is a tab-delimited text file containing responses to the questions asked before and after the task. The questions are the column headers; the exact questionnaire is in `/materials/questionnaire.pdf`. These data were not analyzed. * `RT.csv`, `SDT.csv`, and `motivation_aversion.csv`: These contain tables with task performance (Response Time, Signal Detection Theory [A', Hit Rate, False alarm Rate]) and self-report (motivation and aversion ratings) data per participant and task block. They were created from the raw data (see [Individual subject data]) by the `MFBrain.Rmd` analysis script. * `MFBrain_behav_data.mat`: is a MATLAB data file containing the summary of the task/ratings data, used to make plots for the paper. ## Individual subject data Each participant in the study has their own folder, named in the following way: "/`SUBJECT`_`DATE`", where `SUBJECT` is a participant identifier (e.g. _S01_, _S02_), and `DATE` is the _YYYY-MM-DD_ date the data were collected. Each folder should have 5 text files, all created by the task program: * `SUBJECT`_aversion.txt: plain-text file with 10 ratings on a 1--7 scale, indicating the level of _aversion_ towards continuing the task. * `SUBJECT`_motivation.txt: plain-text file with 10 ratings on a 1--7 scale, indicating the level of _motivation_ to continue the task. * `SUBJECT`_pertrial.txt: tab-delimited text file with 2400 rows and 6 columns. Each row is a trial; the columns are the following: 1. Block number (range: 1--120) 2. Trial number within a block (range: 1--20) 3. `1` if target (short line) was present; `0` otherwise (long line) 4. `1` if participant responded a target was present; `0` otherwise (no response) 5. response time in milliseconds 6. either `Hit`, `Miss`, `False alarm` or `Correct rejection` * sp_id_`SUBJECT`_`DATE`_cpt-stable-PEST.txt: This is the output from the Parameter Estimation for Sequential Testing procedure, to estimate the length of the short line that should be used throughout the experiment. * sp_id_`SUBJECT`_`DATE`_cpt-stable.txt: This is the output from the main task, containing some summary statistics on response time and perceptual sensitivity, as well as the per-trial data (as also in `SUBJECT`_pertrial.txt). __N.B.__: * For participant _S02_, the _[...]_PEST.txt_ file is missing. * During data collection, _S18_ was unintentionally entered as "S19". For this reason, some of the file contents and names in _S18_s folder contain "S19" instead of "S18". Files that were used during data analyses were renamed to "S18" after the fact. * the `SUBJECT` identifier in the folder names uses leading zeros (e.g. _S01_) so that the folders are arranged in chronological order. The file names inside each folder do not use leading zeros (e.g. _S1_); this is how the subject identifier was originally entered. * similarly, the `DATE` identifier in the folder name is _YYYY-MM-DD_, while it is _DD.MM.YY.HH.MM_ in the file names. ---------- # EEG data Below all data files in `/EEG` are described in decreasing order of data processing (raw data last). ## EEG data for statistics ### Folder organization The tabular data in the text-files in the `/statistics` folder were extracted from the fully processed EEG data by the `src/func/MFBRain_group_stats.m` script. They contain the data averaged over time windows, frequency windows, and electrodes of interest. These are read in by the `MFBrain.Rmd` notebook that produces all the statistical results in the paper. Each dependent variable has their own folder inside `/statistics`: * `/ITPC` contains the theta inter-trial phase clustering data. * `/power` contains the theta power data. * `/N1` and `/P1` contain the ERP data. * `/lat_index` contains the alpha lateralization data. These folders contain text files with the following file name: "`BIN`\_`METHOD`\_`DV`\_`ELECTRODES`\_[`TIME`]ms_[`FREQ`]Hz.txt", where: * `BIN` indicates whether the data were split into 10-minute bins (correct rejections, `10m`) or 20-minute bins (hits and misses, `20m`) * `METHOD` indicates whether data were simply averaged in the window of interest (`average`), or whether a smaller average was taken around a subject-specific peak (`peak`) * `DV` indicates the dependent variable (`ITPC`, `power`, `latindex`, `ERP`). * `ELECTRODES` indicates the electrodes the data came from. For example, the right PO pool of three electrodes mentioned in the paper would be `PO8P6P8` * `TIME` indicates the time window the data came from; e.g. `150_500` for the ITPC data from 150--500 ms post-stimulus * `FREQ` indicates the frequency window, e.g. `3_7` for the ITPC data from 3--7 Hz. If the data are ERPs (which are not frequency resolved), this is fixed to `1_1` ### File organization The rows inside each text file are participants. The first column `subnum` contains the row numbers from 1--21; note that these __do not__ correspond to the participant identifiers used elsewhere (e.g. `subnum` = 1 is not necessarily `S01`). Each column header reflects the condition and slice of the data and is structured as follows: "`ELECTRODES`_`CONDITION`_t`TIME`_f`FREQ`", where: * `ELECTRODES` is the same as in the file name (e.g. `PO8P6P8`) * `CONDITION` indicates the task block and trial type the data came from. This format is the same as the `condition_labels` variable described under [Helper variables] below * `TIME` indicates the time window, e.g. `150500` for data from 150--500 ms post-stimulus * `FREQ` indicates the frequency window, e.g. `37` for data from 3--7 Hz. For ERP data, this is fixed to `11` If the data were extracted around subject-specific peaks (i.e. `METHOD` in the file name is `peak`), each data column is followed by two additional columns: 1. "`ELECTRODES`_`CONDITION`_t`TIME`_f`FREQ`_freq" is the frequency of the peak response inside the window defined by `TIME` and `FREQ`. 2. "`ELECTRODES`_`CONDITION`_t`TIME`_f`FREQ`_time" is the time point of the peak response inside the window defined by `TIME` and `FREQ`. ## Fully processed EEG data Each participant has two "`SUBJECT`\_tf\_`BIN`.mat" files in the `/level1` folder (where `SUBJECT` is their participant ID). `BIN` is either: * "10m": Data split into eight 10-minute blocks. This binning was used to analyze _correct rejections_. This dataset does not contain hits/misses/false alarms, as there were too few trials to split into 10-minute blocks. * "20m": Data split into four 20-minute blocks. This binning was used to analyze _hits_ and _misses_. This dataset does not contain false alarms, as these were too scarce even for 20-minute bins. This dataset does not contain correct rejection trials, as these were always analyzed in 10-minute bins. Each file contains the same set of variables. Variables either hold the data ([Data variables]) or help identify what's in the data variables ([Helper variables]). Two more files exist in this folder: "seed_`BIN`_`DATETIME`.txt", where `DATE` is a YYY-MM-DD_HH-MM-SS date-time string signifying when the file was created, and `BIN` is either "10m" (correct rejection data) or "20m" (hits and misses). These files contain a single number which was used as the seed for the trial subsampling process (see `trials2use`). They are read in by the `src/MFBrain_level1.m` script, such that it's possible to exactly recreate the results (using the same "random" subsets of trials). ### Data variables * `ERPs`: 3-dimensional array of trial-averaged ERP data (voltage values): 1. 8 conditions (see `condition_labels`) 2. 64 EEG channels/electrodes (see `chanlocs` for channel name and other info) 3. 2559 samples (see `ERP_time` for corresponding timestamps). * `tf`: 5-dimensional array of time-frequency data: 1. 8 conditions (see `condition_labels`) 2. 64 EEG channels/electrodes (see `chanlocs` for channel name and other info) 3. 30 frequencies (see `frex` for frequency values) 4. 201 samples (see `times2save` for corresponding timestamps) 5. 2 time-frequency outcomes 1. trial-averaged power 2. inter-trial phase clustering ### Helper variables * `chanlocs`: 1x64 structure with information on each EEG channel; e.g. the name of the channels are in the `label` field. Indices correspond to 2nd dimension of `ERPs` and `tf`. * `condition_labels`: cell array of strings indicating which task block and trial type EEG data is from. For example, `Correct_rejection_40` means data from correct rejection trials in the 4th 10-minute block (average of trials from 30 minutes to 40 minutes into the task). Likewise, `Hit_20_40` means data from hit trials in the 2nd 20-minute block (average of trials from 20 minutes to 40 minutes into the task). Indices correspond to 1st dimension of `ERPs` and `tf`. * `ERP_time`: vector of timestamps from -2000 ms to 3000 ms. Time 0 is the onset of the stimulus (target/non-target). The sampling rate was 512 Hz, so the difference between each timestamp is 1/512 s. Indices correspond to 3rd dimension of `ERPs`. * `frex`: vector with frequency of wavelets used in the time-frequency analysis. Indices correspond to 3rd dimension of `tf`. * `n`: vector with number of trials per task block (8 columns). Note that these counts reflect how many trials were left after preprocessing. Fewer trials were actually used in the analyses; see `trials2use`. * `times2save`: same as `ERP_time`, but for time-frequency data (these were downsampled after time-frequency decomposition, to have 15 ms between each timestamp). Indices correspond to 4th dimension of `tf`. * `trials2use`: vector of 8 cells (one per condition; see `condition_labels`). Each cell contains an `s` by `reps` matrix with trial indices. These trials were used to create the trial-averaged ERP and time-frequency results. The block with the least amount of trials `s` was identified (corresponding to the lowest number in `n`). Trials were then subsampled such that all blocks had the same amount of trials , i.e. trials in blocks with more than the minimum number of trials were discarded. This procedure was repeated `reps` times (by default, `reps` = 1000), to avoid a biased subsample. The subsampling procedure was done separately for correct rejections on the one hand, and hits and misses on the other (so there are always as many hit trials as miss trials for a given block, but usually more correct rejection trials). ## Preprocessed EEG data These files in the `/processed` folder were produced by preprocessing the data in `/raw` according to the description in the paper. They are read in by the `src/MFBrain_level1.m` script that produces a fully processed (ERP and time-frequency data) data file per participant. Each participant has one "`SUBJECT`_ready4level1.mat" file, where `SUBJECT` is their participant ID. These contain an EEGLAB `EEG` structure. The actual data is in the `data` field, and is a 3D channels by time points by trials array. ## Raw EEG data The files in the `/raw` folder are the unmodified BioSemi data files recorded in the lab. Each participant has one "`SUBJECT`.bdf" file, where `SUBJECT` is their participant ID.