PaperlayerView exact revision on arXiv

Source-linked AI summary

Multi-channel EEG recordings during a sustained-attention driving task

Zehong Cao, Chun-Hsiang Chuang, Jung-Kai King, Chin-Teng Lin

arXiv:1809.06534v1eess.SPq-bio.NC

TL;DR

Fatigue and drowsiness increase crash risk, making driver behavior and brain dynamics relevant to sustained-attention driving research. The paper describes a large EEG dataset collected during a virtual-reality lane-departure task, intended to support research on driver decision-making under stress and cognitive fatigue.

  • Problem

    Fatigue and drowsiness increase crash risk, motivating study of how sustained-attention driving is reflected in driver behavior and brain dynamics.

    Source: Background & Summary · p. 1

  • Method

    The authors collected synchronized 32-channel EEG and vehicle-position data during a 90-minute event-related lane-departure task in a virtual-reality driving simulator.

    Source: Background & Summary · p. 2, Experimental paradigm · p. 3

  • Results

    The paper describes a large EEG dataset intended to support study of driver behavioral decision-making under stress and cognitive fatigue in complex operational environments.

    Source: Background & Summary · p. 2

  • Takeaways & Limitations

    The dataset is intended for reuse in studying interactions among brain, behavior, sensory input, and performance during cognitively demanding driving.

    Source: Background & Summary · p. 2

    The released EEG data are raw and require preprocessing, including manual removal of severe signal contamination.

    Source: EEG validation · p. 5

Abstract

from arXiv · show

We described driver behaviour and brain dynamics acquired from a 90-minute sustained-attention task in an immersive driving simulator. The data include 62 copies of 32 channel electroencephalography (EEG) data for 27 subjects that drove on a four lane highway and were asked to keep the car cruising in the centre of the lane. Lane departure events were randomly induced to make the car drift from the original cruising lane towards the left or right lane. A complete trial includes events with deviation onset, response onset, and response offset. The next trial, in which the subject has to drive back to the original cruising lane, occurs from 5 to 10 seconds after finishing the current trial. We hope that this dataset will lead to the development of novel neural processing assays that can be used to index brain cortical dynamics and detect driving fatigue and drowsiness. This publicly available dataset is beneficial to the neuroscientific and brain computer interface communities.

Title · Background & Summary

The paper describes a large EEG dataset from a sustained-attention driving task and presents it as a resource for studying driver behaviour and brain dynamics under stress and cognitive fatigue.

Title

  • The paper is titled “Multi-channel EEG recordings during a sustained-attention driving task.”Source: Title · p. 1

Background & Summary

  • Fatigue and drowsiness can suppress driver awareness, recognition, and directional control, while diminishing arousal and information processing in unusual or emergency situations.The paper frames these risks as motivation for studying sustained-attention driving.Source: Background & Summary · p. 1
  • The research program aims to characterize how drivers’ neurocognitive states vary with sustained-attention demands and to develop computational methods for assessing their state and performance.The computational work includes statistical modelling, data visualization, and analytic and algorithmic approaches.Source: Background & Summary · pp. 1–2
  • The experiment used a VR dynamic driving simulator, recording 32-channel EEG and vehicle position simultaneously while licensed participants sustained attention.Participants had no history of psychological disorders.Source: Background & Summary · p. 2
  • Prior studies using this dataset examined kinaesthetic effects, mind-wandering, and EEG-based prediction of driver alertness and response times.The work also assessed the feasibility of online brain-network assessment and rectification during cognitive challenges.Source: Background & Summary · p. 2
  • The dataset is intended to support joint study of brain, behavioural, sensory, and performance dynamics and development of individualized real-time neuroergonomic systems for driving.The authors identify situational awareness, decision making, and human-system performance as potential applications.Source: Background & Summary · p. 2

Methods · Participants

The study collected 62 EEG recordings from 27 university students or staff aged 22–28 during a 90-minute sustained-attention driving task. Participants met vision, health, sleep, and pre-experiment preparation requirements, and the study included a pre-test, ethics approval, and informed consent.

Participants

  • 27 volunteers aged 22–28, all students or staff of National Chiao Tung University, participated in the 90-minute task.Source: Participants · p. 2
  • The 27 participants contributed 62 EEG recordings across sessions held at multiple times on the same or different days.Source: Participants · p. 2
  • Participants had normal or corrected-to-normal vision, reported no recent sleep deprivation or drug-abuse history, and maintained a normal work-rest cycle.Source: Participants · p. 2
  • For the week before testing, participants were asked to sleep about eight hours nightly and not stay up later than 11:00 PM.Source: Participants · pp. 2–3
  • Participants abstained from alcohol, caffeinated drinks, and strenuous exercise the day before each experiment.Source: Participants · p. 3
  • A pre-test explained the instructions and checked for simulator-induced nausea; the study had institutional approval, obtained informed consent, and compensated each session at about USD $20.Source: Participants · p. 3

Virtual-reality driving environment · Experimental paradigm

The study used an immersive VR highway simulator for a continuous night-driving task, with randomly induced lane departures prompting steering responses while EEG was recorded. The paradigm defined each departure and recovery as a trial, with a 5–10-second interval before the next trial.

Virtual-reality driving environment

  • Six synchronized highway scenes projected across a nearly 360° field of view around a vehicle mounted on a six-degree-of-freedom motion platform.The screens were set at viewing angles of 0°, 42°, 84°, 180°, 276° and 318°.Source: Virtual-reality driving environment · p. 3
  • The simulator depicted a monotonous, traffic-free night drive on a straight four-lane highway, with a Ford Probe frame on the motion platform and a scenario refresh rate for 100 km/hr cruising.Lane width was 60 units on a 0–255 road-distance and vehicle-trajectory scale; the driver's view was recorded.Source: Virtual-reality driving environment · p. 3

Experimental paradigm

  • The event-related paradigm measured reactions to perturbations during a 90-minute drive without breaks, asking drivers to cruise in the centre of the third lane as the car drifted equally likely left or right.Drivers' activities were monitored by a dashboard surveillance camera.Source: Experimental paradigm · p. 3
  • Lane-departure trials came from experiments conducted at National Chiao Tung University, Taiwan, between 2005 and 2012.Source: Experimental paradigm · p. 3
  • Lane departures were randomly induced toward either side of the original cruising lane, and participants were instructed to compensate quickly by steering the wheel.Participants did not control the accelerator or brakes.Source: Experimental paradigm · pp. 3–4
  • Each lane-departure trial comprised a baseline period, deviation onset, response onset and response offset, with EEG recorded simultaneously.The steering direction corresponded to the direction of the required correction.Source: Experimental paradigm · p. 4
  • The next trial began 5–10 seconds after the current one, during which participants drove back to the third-lane centre line; no feedback alerted participants if they fell asleep.Source: Experimental paradigm · p. 4

Tutorial and code availability · Data Records

A 59-page tutorial and MATLAB code for EEG preprocessing and data analysis are available through figshare.

Tutorial and code availability

  • The figshare record provides a 59-page EEG tutorial and MATLAB code for preprocessing and data analysis.The tutorial is titled “Tutorial Data Analysis for Multi-channel EEG Recordings during a Sustained-attention Driving Task.pdf,” and the code archive is “Code-availability.zip.”Source: Tutorial and code availability · p. 4

Data recording and storage · EEG signals

The recording workflow combined synchronized EEG, event, and vehicle-position data, with the resulting files available for analysis in EEGLAB. The paper describes the EEG acquisition setup and how signals and task events are represented in the dataset.

Data recording and storage

  • The stimulus computer logged vehicle trajectories and events, while sending synchronized triggers to the Neuroscan system, and the two files were integrated to align event timing with behavior.The combined event file was imported into EEGLAB in MATLAB.Source: Data recording and storage · p. 4
  • EEG was recorded with a wired cap holding 32 Ag/AgCl electrodes—30 EEG and two opposite-mastoid references—placed using a modified international 10–20 system.Electrode-skin impedance was kept below 5 kΩ.Source: Data recording and storage · p. 4
  • The Scan SynAmps2 Express digitized EEG at 500 Hz with 16-bit resolution, and raw recordings could be saved as .cnt files on a PC and server.Data acquisition used Neuroscan's Scan 4.5 software.Source: Data recording and storage · p. 4

EEG signals

  • Raw files can be read in EEGLAB, and uploaded set files contain 32 EEG signals plus one vehicle-position signal in EEG.data.The vehicle-position signal describes the simulated vehicle's position.Source: EEG signals · pp. 4–5
  • The 32 EEG channels map to the named scalp and mastoid sites, with A1 and A2 identified as mastoid reference electrodes.The channel list includes frontal, temporal, central, parietal, and occipital sites.Source: EEG signals · pp. 4–5
  • Dataset event types identify deviation onset (251 or 252), response onset (253), and response offset (254).These event marks distinguish the task's lane-departure and response phases.Source: EEG signals · p. 5
  • Figure 3 shows an example of behavior performance and EEG signals alongside their associated events.Source: EEG signals · p. 5

Technical Validation · Behavioural validation · EEG validation

Technical validation combines EEG acquisition and preprocessing guidance with partner-group findings consistent with estimating changes in driver arousal, fatigue, and vigilance from behavioural and neurocognitive performance. The passages describe the participant sample and recording setup, and recommend artifact removal before analysis of the raw data.

Behavioural validation

  • The dataset includes 27 university students and staff with normal or corrected-to-normal vision and no reported psychiatric, neurological, or drug use disorder history.Source: Behavioural validation · p. 5
  • Before the experiment, participants wore a recording cap, read instructions for 5–10 minutes, and completed a participant information questionnaire.Source: Behavioural validation · p. 5
  • EEG was recorded from a 32-channel cap at 500 Hz with 16-bit quantisation, using Ag/AgCl electrodes and impedance calibrated below 5 kΩ.Thirty electrodes followed a modified international 10–20 arrangement, with two mastoid references.Source: Behavioural validation · p. 5

EEG validation

  • Stored EEG data remain raw; researchers are advised to apply a 1–50 Hz bandpass filter, down-sample to 250 Hz, and manually remove severe contamination before analysis.Filtering, baseline correction, and artifact rejection used during trials were not applied to the stored data.Source: EEG validation · p. 5
  • Ocular and muscle artifacts can mask cortical signals and bias interpretation, and independent component analysis is described as a tool for suppressing artifacts.The passage notes that blind source separation techniques have been proposed for EEG preprocessing.Source: EEG validation · pp. 5–6
  • Findings shared with UCSD and DCS Corporation were consistent with their results, supporting technical validation of estimating driver arousal, fatigue, and vigilance shifts from behavioural and neurocognitive performance.The passage does not provide quantitative validation results.Source: EEG validation · p. 6

Usage Notes

The dataset is publicly available on figshare, and the authors provide guidance and tools for analysing and reusing it. They recommend EEGLAB for EEG processing and describe a workflow from loading data to extracting event-related epochs.

Usage Notes

  • Researchers can download “Multi-channel EEG recordings during a sustained-attention driving task” from the public figshare repository.The authors identify the project as publicly accessible through figshare.Source: Usage Notes · p. 6
  • EEGLAB supports continuous and event-related EEG processing, including ICA, time/frequency analysis, and artefact rejection; the authors recommend version 5.03 on Windows 7 or Linux.EEGLAB is a MATLAB toolbox with a graphical user interface and tutorials.Source: Usage Notes · p. 6
  • A data-analysis tutorial and MATLAB code are provided as references for EEG preprocessing and analysis to help researchers reuse the dataset.The authors direct users to the materials at https://doi.org/10.6084/m9.figshare.6427334.v2.Source: Usage Notes · p. 6
  • The suggested workflow loads an existing dataset, checks EEG fields such as sampling rate, channel locations, events, and signals, then extracts event epochs and optionally removes each epoch’s mean baseline.EEGLAB also supports plotting trial or epoch averages, comparing ERP images, working with ICA components, time/frequency decomposition, and combining datasets.Source: Usage Notes · pp. 6–7

Competing interests

Z. Cao, C.H. Chuang, J.T. King, and C.T. Lin declared no competing financial interests.

Competing interests

Figure Legends · Tables · Data Citations

The data citation identifies the 2018 figshare fileset of multi-channel EEG recordings during a sustained-attention driving task [1].

Data Citations

  • The cited fileset is attributed to Cao, Chuang, King, and Lin (2018) and is hosted on figshare [1].DOI: 10.6084/m9.figshare.6427334.v2.Source: Data Citations · p. 9

Figures

The figures depict a virtual-reality lane-departure driving paradigm with simultaneous EEG and behavioral recording. They also show example behavior and EEG signals, and the EEG cap layout and electrode impedance.

Figures

  • The event-related lane-departure paradigm took place in a virtual-reality driving simulator, with EEG and behavior recorded simultaneously.Source: Figure 1 · p. 10, Figure 2 · p. 11
  • An example figure shows behavioral performance and EEG signals alongside associated events.Source: Figure 3 · p. 12
  • The EEG cap layout includes 10–20 and additional electrodes, with skin-contact impedance kept below 5 kΩ.Source: Figure 4 · p. 13

Related papers

Loading 1809.06534v1…