Behavioural dataset for "Perceptual decision-making: extending a standardized rodent task to humans"

This dataset contains trial-level behavioural data and participant information from a human visual decision-making task. It is inspired by the data format used by the International Brain Laboratory (e.g. here). The release includes 80 participants. For more information, please see our preprint and GitHub repo. File Unit of observation Contents processed_data_2026.csv One row per trial Task events, responses, stimulus information etc. participants_info.csv One row per participant Session start time, gender, age, handedness, and questionnaire responses. The files can be linked using subject (read this identifier as text to retain leading zeros). Dataset overview This dataset includes 80 participants (65 female, 13 male, and 2 nonbinary participants; mean age = 19.99 years, SD = 3.14, range = 18–35 years) and 48000 trials in total. The instruction condition is stored as instructions in both files. Its code-to-condition mapping and group sample sizes: instructions Condition Participants 0 Not Instructed 41 1 Instructed 39 1. Trial file: processed_data_2026.csv Each row corresponds to one trial. Column Description and values subject Participant identifier, stored as a three-character string (e.g., 006). session Session identifier (e.g., 22121620). Each participant has one session in this dataset. trial Trial number within the session; between 0 and 599. instructions Instruction condition: 1 = instructed, 0 = not instructed. probabilityLeft Probability of the target assigned to the left side in the current block: 0.2, 0.5, or 0.8. These sequences were pre-generated, see the Methods section of the preprint. contrastLeft, contrastRight Rescaled contrast on each side: 0, 0.1, 0.25, 0.5, or 1, corresponding to displayed contrasts of 50%, 52%, 55%, 60%, and 70%, respectively. choice Response: 1 = left, -1 = right, 0 = no recorded response. firstMovement_times_from_stim Time (s) from stimulus onset to the first recorded movement. response_times_from_stim Time (s) between stimulus onset and the completion of the response, when the stimulus crossed the midline of the screen; missing for no-response trials. feedbackType Recorded feedback code: 1 = correct, -1 = incorrect, missing = no response. stimContrast Sum of the rescaled left and right contrasts (contrastLeft + contrastRight); observed values: 0, 0.1, 0.25, 0.5, 1. sideContrast Signed contrast, calculated as contrastLeft - contrastRight. Positive values indicate greater contrast on the left; negative values indicate greater contrast on the right. Observed values: -1, -0.5, -0.25, -0.1, 0, 0.1, 0.25, 0.5, 1. stimSide Side on which the target stimulus was presented: 1 = left, -1 = right. cursorPosition Recorded cursor positions over the trial, stored as an array-like string in the CSV. Each value corresponds to a sample in cursorTime. cursorTime Times at which cursor positions were sampled, stored as an array-like string in the CSV. dot.started Recorded onset time of the visual stimuli, in seconds from the start of the session. 2. Participant file: participants_info.csv Each row corresponds to one participant. An unanswered question may appear as a missing value when the CSV is read. Column Description subject Participant identifier, matching the trial file. instructions Instruction condition: 1 = instructed, 0 = not instructed. session_start Recorded session start date and time. gender Selected gender response, if provided. age Age response (years). handedness Selected handedness response, if provided. textbox1_text Free-text response. Please describe how you think the game worked and what the rules were. textbox2_text Free-text response. How well do you think you did in the game? textbox3_text Free-text response. What strategies did you use during the game? textbox4_text Free-text response. Did you notice any patterns in the game? note Additional information about the participant’s session, where applicable. Loading the data For example, use python pd.read_csv(path, dtype={'subject': str}) with pandas. In the CSV, cursorPosition and cursorTime are stored as strings representing arrays and must be parsed before numerical analysis, for example with: import ast import pandas as pd data = pd.read_csv( 'processed_data_2026.csv', dtype={'subject': str, 'session': str}, converters={ 'cursorTime': ast.literal_eval, 'cursorPosition': ast.literal_eval, }, ) Pupil and facial video data Eye-tracking data (gaze and pupil measurements) and facial video recordings were also collected during the task. Pupil data are described in Johnson et al. (2025) and are planned for inclusion in a future update of this release. For privacy reasons, facial video recordings are not included in the public release and can be obtained from the corresponding author on request.