Post-intervention high-beta synchronization tracks stable-manifold alignment during stick-balancing learning

================================================================================ DATASET README: Virtual Stick-Balancing Task and EEG Recordings ================================================================================ 1. OVERVIEW -------------------------------------------------------------------------------- This dataset contains behavioral/dynamics data files(.csv) and preprocessed electroencephalographic (EEG) dataset files (.set) from 7 healthy participants who performed a virtual stick-balancing task across multiple experimental conditions. - Study Topic: Post-intervention high-beta synchronization and stable-manifold alignment during virtual stick-balancing learning. - Corresponding Publication: Shimokado et al., "Post-intervention high-beta synchronization tracks stable-manifold alignment during stick-balancing learning" (Neuroscience Letters). - Affiliation: Nomura Lab, Graduate School of Informatics, Kyoto University. 2. EXPERIMENTAL TASKS & CONDITIONS -------------------------------------------------------------------------------- Participants stabilized a virtual inverted pendulum displayed on a monitor using a 2-axis joystick operated exclusively in the horizontal plane (i.e., lateral tilting). Task difficulty was manipulated via artificial control delay (Delta_a) and process noise intensity (sigma_a): - Control / Skilled: Tasks with baseline parameters (Delta_a = 0 s, sigma_a = 0 Nm). - Delayed: Task performed with an artificial feedback delay of Delta_a = 0.07 s (sigma_a = 0 Nm). - Noisy: Task performed with a process noise intensity of sigma_a = 0.05 Nm (Delta_a = 0 s). 3. DIRECTORY & FILE STRUCTURE -------------------------------------------------------------------------------- The dataset is anonymized and organized per participant (sub1 to sub7): sub1/ ├── dynamics_data/ │ ├── Control/ # Dynamics data files(.csv) for Control task │ ├── Delayed/ # Dynamics data files(.csv) for Delayed task │ ├── Noisy/ # Dynamics data files(.csv) for Noisy task │ └── Skilled/ # Dynamics data files(.csv) for Skilled task │ └── preprocessed_eeg/ ├── Control/ # Preprocessed EEGLAB .set files (Control task) └── Skilled/ # Preprocessed EEGLAB .set files (Skilled task) sub2/ └── ... (The same structure for sub2 through sub7) README.txt 4. DATA FILE DESCRIPTIONS -------------------------------------------------------------------------------- - subName denotes the participant ID (sub1 to sub7) - taskName represents the experimental condition ('Control', 'Delayed', 'Noisy', 'Skilled') - trialName denotes the trial file identifier (e.g., `trial01`, `trial02`, ...). Note: Files are named sequentially, but due to data cleaning and quality checks (e.g., removal of severe artifacts or trigger issues), some trial indices may be omitted and are not strictly consecutive. A. Behavioral & Kinematic Data (path:./subName/dynamics_data/taskName/trialName.csv) - Time-series recordings of the virtual balancing task (sampled at 200 Hz). - Typical columns include: 1. time[s] 2. theta[rad] (Pendulum angle) 3. dtheta[rad/s] (Angular velocity) 4. delayed_input (-1 to 1 delayed joystick input) 5. torque[Nm] (Applied torque) 6. survive (Binary flag indicating whether balance is maintained: 1 for balanced, 0 upon fall) Note: While `delayed_input` and `torque` are generally proportional to each other, in the delayed task conditions, `torque` is further delayed by 0.07 seconds relative to the input. B. Preprocessed EEG Data (path:./subName/preprocessed_eeg/taskName/trialName.set) - EEGLAB dataset files containing continuous EEG signals prior to wavelet transformation. - File Naming & Indexing: Files are named sequentially (e.g., `trial01.set`, `trial02.set`, ..., up to `trialXX.set` depending on the total trial counts per condition). Note that some trials may be omitted due to data cleaning or hardware/trigger exclusion. - Preprocessing steps applied: * Line noise attenuation (60 Hz) and channel interpolation via PREP pipeline. * Band-pass filtering at 1–70 Hz and downsampling to 200 Hz. * Extended Infomax Independent Component Analysis (ICA) for artifact removal (retaining components classified as "Brain" by ICLabel). * Re-referenced to a robust average. - Synchronization Note: While the time-series in the dynamic CSV files starts from 0 s (representing the start of each recording file), a 2-second delay exists in the TTL trigger, meaning that the EEG triggers are recorded 2 seconds later relative to the behavioral time-series. During data analysis and segmentation, this 2-second offset (t = -2 s to 0 s relative to the behavioral timeline) is accounted for to ensure precise alignment with the EEG epoched data. ( The name of the event is "trialstart") - Data Completeness & Trial Exclusion: Some trials containing hardware issues (such as missing triggers or severe artifact contamination) were identified and excluded or noted during quality checks. Consequently, trial indices in the dataset may not be entirely consecutive. 5. SOFTWARE REQUIREMENTS & USAGE -------------------------------------------------------------------------------- - Software: MATLAB (R2022b or later recommended) - Toolboxes: EEGLAB (v2022.0 or later) To load and inspect an EEG dataset in MATLAB (Example)====== subName = 'sub1'; taskName = 'Control'; fileName = 'trial02.set'; filePath = fullfile('.', subName, 'preprocessed_eeg', taskName); EEG = pop_loadset('filename', fileName, 'filepath', filePath); %---------------------------------------------------------- % Generalization: % To load data from other participants, tasks, or trials, simply change % the 'subName', 'taskName', and 'fileName' variables accordingly. % Summary of available trials for EEG data per participants and condition: % sub1, task: Control, maxtrial: 39, omitted trials: [1, 26, 27, 28, 31] % sub1, task: Skilled, maxtrial: 05, omitted trials: [1] % sub2, task: Control, maxtrial: 15, omitted trials: [11] % sub2, task: Skilled, maxtrial: 06, omitted trials: None % sub3, task: Control, maxtrial: 18, omitted trials: [2, 9] % sub3, task: Skilled, maxtrial: 05, omitted trials: None % sub4, task: Control, maxtrial: 50, omitted trials: [6, 12] % sub4, task: Skilled, maxtrial: 11, omitted trials: None % sub5, task: Control, maxtrial: 26, omitted trials: [1] % sub5, task: Skilled, maxtrial: 05, omitted trials: None % sub6, task: Control, maxtrial: 37, omitted trials: None % sub6, task: Skilled, maxtrial: 05, omitted trials: None % sub7, task: Control, maxtrial: 50, omitted trials: [19, 36, 37] % sub7, task: Skilled, maxtrial: 06, omitted trials: None %---------------------------------------------------------- ============================================================ 6. CONTACT & SUPPORT -------------------------------------------------------------------------------- For any questions regarding the dataset, code, or experimental procedures, please contact both: - Haruki Shimokado (Email: [e-mail]) - Taishin Nomura (Email: [e-mail]) - Laboratory: Nomura Lab, Graduate School of Informatics, Kyoto University