iNCog-EEG (ideal vs. Noisy Cognitive EEG for Workload Assessment) Dataset

iNCog-EEG: A Multitasking EEG Dataset for Cognitive Workload Assessment with Controlled Task DifficultyThis dataset, named iNCog-EEG (Ideal vs. Noisy Cognitive EEG for Workload Assessment), contains EEG recordings from 40 participants engaged in a multitiered multitasking protocol designed to emulate realistic cognitive workload environments. Unlike conventional controlled datasets, iNCog-EEG integrates both clean and artifact-contaminated EEG recordings, enabling researchers to evaluate algorithm robustness under ideal and noisy conditions.Key CharacteristicsNoise Sources Captured (for 10 subjects): Eye movements/blinks (EOG), muscle activity (EMG), cardiac activity (ECG), respiratory artifacts, and power line interference.Hardware Used: EEG signals were acquired using the KT88-3200 digital EEG system with 16 scalp electrodes at a 200 Hz sampling rate, positioned according to the international 10–20 system.Participants: 40 healthy individuals. Ethical approval was granted by the Institutional Review Board of Evercare Hospital, Dhaka Ltd. (IRB-1964105-1).Ground Truth Labels: Derived from task difficulty metadata and subjective feedback, yielding a two-level annotation scheme:Binary classification: No Workload (rest) vs. Workload (task phases)Hierarchical classification: Workload further divided into Low, Moderate, and HighTask and Recording DesignResting Phase: 5 minutes (baseline condition).Multitasking Phases: Three stages of escalating difficulty (easy, medium, hard), each lasting 5 minutes.Breaks: 10 minutes between phases to simulate recovery and workload transitions.Concurrent Cognitive Tasks:Math problem-solvingN-back memory matchingObject trackingResponse inhibitionThese were distributed across screen quadrants to induce simultaneous cognitive load.Total Duration: ~40 minutes per participant (5 min rest + 3 × 5 min multitasking + 2 × 10 min breaks).File Naming ConventionEach participant’s recordings are stored in a dedicated folder named subXX (e.g., sub01, sub02, …, sub40). Inside each folder, four .EDF files represent the workload conditions:subxx_nw.EDF → No Workload (resting state) subxx_lw.EDF → Low Workload (easy multitasking) subxx_mw.EDF → Moderate Workload (medium multitasking) subxx_hw.EDF → High Workload (hard multitasking) Subjects 01–30: Clean EEG recordingsSubjects 31–40: Noisy EEG recordings with real-world artifactsThis structure ensures straightforward differentiation between clean vs. noisy data and across workload levels.ApplicationsThis dataset can be applied to a wide range of research areas, including:EEG signal denoising and artifact rejectionBinary and hierarchical cognitive workload classificationDevelopment of robust Brain–Computer Interfaces (BCIs)Benchmarking algorithms under ideal and noisy conditionsMultitasking and mental workload assessment in real-world scenariosBy combining controlled multitasking protocols with deliberately introduced environmental noise, iNCog-EEG provides a comprehensive benchmark for advancing EEG-based workload recognition systems in both clean and challenging conditions.iNCog-EEG：一款用于认知负荷评估且可控任务难度的多任务脑电图（Electroencephalogram, EEG）数据集。 本数据集命名为iNCog-EEG（即用于负荷评估的理想与噪声认知脑电图数据集），包含40名参与者的脑电图记录，这些参与者参与了旨在模拟真实认知负荷环境的分层多任务范式。与传统可控数据集不同，iNCog-EEG同时涵盖了干净与含伪迹的脑电图记录，可支持研究人员在理想与噪声环境下评估算法的鲁棒性。 ## 关键特性 ### 捕获的噪声源（针对10名受试者） 眼动/眨眼（Electrooculogram, EOG）、肌肉活动（Electromyogram, EMG）、心脏活动（Electrocardiogram, ECG）、呼吸伪迹以及电力线干扰。 ### 所用硬件 采用KT88-3200型数字脑电图系统采集脑电图信号，配备16个头皮电极，采样率为200Hz，电极位置符合国际10-20系统标准。 ### 受试者概况 共40名健康个体参与本研究。该实验已获得达卡Evercare医院有限公司机构审查委员会的伦理批准（IRB-1964105-1）。 ### 真实标签 基于任务难度元数据与主观反馈生成，采用两级标注方案： - 二分类任务：无负荷（静息状态） vs. 有负荷（任务阶段） - 分层分类任务：将负荷进一步划分为低、中、高三个等级 ## 任务与记录设计 - 静息阶段：5分钟（基线条件） - 多任务阶段：三个难度递增的阶段（简单、中等、困难），每个阶段持续5分钟 - 休息间隔：阶段之间设置10分钟休息时间，用于模拟恢复与负荷过渡过程 - 并行认知任务：包括数学解题、N-back记忆匹配、目标追踪与反应抑制，这些任务分布在屏幕四个象限中，以诱发同步认知负荷 - 总时长：每名参与者的总记录时长约为40分钟（5分钟静息 + 3×5分钟多任务 + 2×10分钟休息） ## 文件命名规范 每名参与者的记录存储于以subXX命名的专属文件夹中（例如sub01、sub02……sub40）。每个文件夹内包含四个.EDF格式文件，分别对应不同负荷条件： - subxx_nw.EDF：无负荷（静息状态） - subxx_lw.EDF：低负荷（简单多任务） - subxx_mw.EDF：中负荷（中等多任务） - subxx_hw.EDF：高负荷（困难多任务） 其中受试者01-30的记录为干净脑电图数据，受试者31-40的记录为带有真实世界伪迹的噪声脑电图数据。该文件结构可清晰区分干净与噪声数据，以及不同负荷等级的数据。 ## 应用场景 本数据集可应用于众多研究领域，包括： - 脑电图信号去噪与伪迹剔除 - 二分类与分层认知负荷分类 - 鲁棒性脑机接口（Brain-Computer Interface, BCI）的开发 - 在理想与噪声环境下对算法进行基准测试 - 真实场景下的多任务与心理负荷评估 通过将可控多任务范式与刻意引入的环境噪声相结合，iNCog-EEG可为在干净与复杂环境下推进基于脑电图的负荷识别系统提供全面的基准测试平台。