henhaobio
Sign in

bioRxivNeurosciencePreprint

The Q1K Integrated EEG and Eye-Tracking Experimental Test Battery for Open Autism Science

Christian O’ReillyiD, James A. Desjardins, Gabriel Blanco-GomeziD, Scott HubertyiD, Anthony Hosein Poitras Loewen, Sweety RamnaniiD, Diksha Srishyla, Inga Sophia KnothiD, Charles-Olivier MartiniD, Jonathan P. BatteniD, Stefon J. R. van NoordtiD, Noémie Hébert-LalondeiD, Fabienne Samson, Tim J. SmithiD, Baudouin Forgeot d’ArciD, Carl ErnstiD, Melissa T. CarteriD, Christine Lucas TardifiD, Julie ScorahiD, Mélanie M. CoutureiD, Pascale AbadieiD, Ridha JooberiD, Sébastien JacquemontiD, Guy Armand RouleauiD, Sarah LippéiD, Mayada ElsabbaghiD

PDFDOICode
In-depth full-text digest available · Jump to digest ↓

Abstract

The Quebec 1000 Families (Q1K) platform has been designed to recruit, phenotype, and collect biospecimens from a large cohort of families with at least one member with autism spectrum disorder or a related neurodevelopmental condition. As part of the Q1K protocol, an experimental test battery was developed and validated using simultaneous high-density electroencephalography (EEG) and eye tracking (ET). We report on the general approach and design principles, multimodal EEG/ET integration, the tasks, and their validation, providing a blueprint for implement such a project. It also describes the cohort and its methods as a reference for future studies using this dataset. By releasing openly this experimental test battery, we aim to support task standardization and multi-project data pooling in autism and related neurodevelopmental disorders.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (20 Jul 2026) · CC BY 4.0

An open, instruction-free EEG plus eye-tracking task battery for autism families (344 to 587 participants per task so far), with timing-validated integrated recordings that enable gaze-locked ERP analysis.

Autism's etiological and phenotypic heterogeneity makes small studies poorly replicable, and the authors describe datasets combining deep multimodal phenotyping, the whole spectrum and relatives as rare despite earlier large MRI, clinical and genetic autism cohorts. Q1K's battery reuses established, instruction-free tasks to span ages 5-89 and ease pooling, and records high-density EEG and eye tracking simultaneously with synchronized, shareable outputs.

Key results

  1. ShownThe battery has so far been completed by several hundred people per task, including probands and their relatives, recruited at three Montreal sites; recruitment is still ongoing.

    Snapshot dated July 8th, 2026 (Table 3): 344 to 587 completers depending on task, of whom 89 to 200 are probands. More than half of probands have data from one relative, and about 10% have data from two. Sites: HSJ, MNI and NIM. Tasks were chosen so that participants aged 5-89 could do them. · 2.4 Sample size; Neuroinformatics; Whole-spectrum inclusion

  2. ShownVisual steady-state stimulation produced strong EEG entrainment at the driving frequencies and their harmonics. Inter-trial coherence was largely consistent between the two main sites, which supports the timing precision of the acquisition setup.

    Inter-trial coherence (ITC), HSJ N=266 vs MNI N=30 (NIM N=7 excluded). The first three harmonics of the 6 Hz and 15 Hz conditions were strongly significant after an early onset response. That onset response appeared in single conditions but not in their contrast. Between-site cluster permutation showed a difference in the 6 Hz condition after stimulation, which the authors judge probably a false positive. · 3.1 Timing precision; Figure legends

  3. SuggestedIn the Gap Overlap task, locking ERPs to gaze onset (taken from the eye tracker) instead of stimulus onset gave larger responses. This analysis needs synchronized EEG and eye-tracking data.

    Event-related potentials (ERPs) were computed for the overlap, baseline and gap conditions, locked either to stimulus onset or to gaze onset. Gaze-onset ERPs were about twice the amplitude (HSJ N=213, MNI N=118, NIM N=6). The text reports no statistical test of this amplitude difference. · 3.2. EEG/ET integration and synchronization; Figure legends

  4. ShownPupillary light reflex responses co-varied with EEG event-related potentials at some scalp channels above chance. This shows the data can be used to relate autonomic and central responses.

    Per-channel correlation between the average pupil response and the ERP. Significance used cluster permutation with one-sample t-tests (clusters at p<0.01). Sample: HSJ N=153, MNI N=99, NIM N=6. ERP is shown for channel E70 (O1). The text does not describe where the significant clusters lie. · 3.2. EEG/ET integration and synchronization; Figure legends

  5. ClaimedPilot testing pushed stimulus choices toward a balance between participant comfort and signal strength, for the pupillary light reflex task and both steady-state tasks.

    PLR trials were cut from 40 to 12 because the task was uncomfortable and average responses were already clear with fewer trials. Auditory: white noise gave higher contrast but tones were more comfortable; 40 Hz was clearest; 10 Hz was dropped. Visual: low spatial resolution, high contrast and luminance inversion were most salient but least comfortable (unpublished); 200 dots at 100% contrast chosen. · Visual Steady-State Response; Auditory Steady-State Response; Pupillary Light Reflexes

Methods, data and reproducibility

A one-hour battery of established tasks that need no instructions (resting state, tone oddball, gap overlap, visual and auditory steady-state, social preference, pupillary light reflex, visual search). It was recorded with simultaneous high-density electroencephalography (EEG) and eye tracking. Photodiodes and sound detection mark events for synchronization. Three tasks were analysed to check data quality.

  • Q1K EEG/eye-tracking cohort: probands with autism or related neurodevelopmental conditions plus relatives (snapshot July 8th, 2026) · human · n = 344 to 587 per task (89 to 200 probands)
  • Visual Steady-State task, inter-trial coherence analysis · human · n = HSJ N=266; MNI N=30
  • Pupillary Light Reflex task, pupil-ERP correlation analysis · human · n = HSJ N=153; MNI N=99; NIM N=6
  • Gap Overlap task, stimulus-locked vs gaze-locked ERP analysis · human · n = HSJ=213; MNI=118; NIM=6
Methods & tools
Experiment Builder (SR Research), Eyelink 1000 Plus (SR Research), Net Station 5, Net Amps 400 EEG amplifier (Magstim EGI), Cedrus StimTracker, MNE-Python, PyLossless, Brain Imaging Data Structure (BIDS), RedCap, LORIS / EEGNet, C-BIG repository, GLOBUS
Software & environment
Hardware and software: a Windows display PC running Experiment Builder; an Eyelink 1000 Plus host PC; an iMac running Net Station 5 with a Net Amps 400 amplifier; a Cedrus StimTracker with two photodiodes. No Python package versions or containers are given.
Assay conditions & compute
Annotation and release preparation run on Narval (Digital Research Alliance of Canada). No time or memory figures are given.
Reproducibility gaps
The open data-sharing process is still being implemented, and no data accession exists yet. C-BIG archiving and versioned releases are in progress. The data-release library is to be described in a manuscript in preparation.
Limitations (6)
  • In the Visual Search task, the distance from fixation to target was not balanced across conditions and strongly affects search time. Analyses must control for it, and the authors advise future reuse to pseudo-randomize it. · stated by the authors
  • The authors did not expect every participant to finish the full one-hour protocol, especially young children and people with more severe profiles, so core tasks run first and optional tasks last. · stated by the authors
  • PyLossless's 'lossless' annotated state currently still applies minimal low- and high-pass filtering; the authors plan to fix this in a future version. · stated by the authors
  • Showing the cartoon film Luna during the auditory steady-state task may add cognitive activity. The authors accepted this risk to improve compliance and do not expect it to interact with the steady-state rhythms. A film (Bao) is also played during the tone oddball task, where this risk is not discussed. · stated by the authors
  • Validation is limited to group-level task effects. The text reports no test-retest reliability and no autism vs relative comparisons, and the between-site timing comparison rests on N=30 at MNI vs N=266 at HSJ. · noted in this summary
  • The roughly doubled amplitude of gaze-locked ERPs is described only qualitatively, with no statistical comparison or effect size. · noted in this summary

Read this if you design EEG or eye-tracking studies in autism or other neurodevelopmental conditions and want an instruction-free task protocol ready for pooling, a recipe for synchronizing EEG with eye tracking, or upcoming Q1K data.

Citations
2

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISLevin '20Avrămiea '24Khuntia '19Darrell '25Williams '26Kim '19
60% semantic matchNeuroscience
April R. Levin, Adam J. Naples, Aaron Wolfe Scheffler+12 more·Frontiers in Integrative Neuroscience·Apr 30, 2020· 76 citations