Agenda

Ryszard Auksztulewicz and Krzysztof Basiński

Multi-lab collaboration to run the experiments proposed in the Neural Mechanisms review paper with scalp EEG in humans

Meeting Recording

Meeting Notes

Human EEG Mismatch Negativity Project Design: Ryszard and Krzysztof Basiński presented the collaborative initiative to record human EEG data using adapted mismatch negativity paradigms, with Jerome, Sam, and Alexander discussing design choices, controls, and cross-species comparability, aiming for broad data collection across multiple labs and pre-registration of protocols.

Project Aims and Scope: Ryszard explained the project's goal to create a human EEG dataset using a subset of paradigms from a previous review, targeting broad sampling across several labs, pre-registration, and open data sharing, with a focus on feasibility and complementarity to existing mouse data.

Paradigm Selection and Adaptation: The team decided to implement two conditions—standard oddball and sequential oddball—excluding sensory-motor and duration mismatch paradigms for feasibility, adapting stimulus timing and session length to suit human EEG constraints and ensure single-session recordings.

Control Block Design: Jerome raised the importance of control blocks, referencing mouse experiments with sandwiched control-oddball-control blocks to assess adaptation, and Ryszard discussed splitting blocks and matching deviant probabilities to address learning and baseline issues in human EEG.

Stimulus Randomization and Orientation: The group debated whether to randomize stimulus orientation across participants, with Ryszard noting stronger responses for certain gratings in pilot data and Jerome highlighting fixed orientations in mouse studies, ultimately considering both approaches for comparability and artifact reduction.

Pilot Data Collection and Logistics: Ryszard shared that pilot data collection is underway in Maastricht, with 15 participants tested and plans to expand to more labs, aiming for at least 15–30 subjects per site, and described recruitment strategies, session duration, and ethical considerations.

Analysis Strategies and Cross-Species Comparisons: Ryszard detailed the planned EEG analyses, mapping them to previous mouse studies, and discussed with Jerome, Alexander, and Krzysztof Basiński how to maximize comparability across species, including event-related potentials, decoding, and population tuning, while addressing modality-specific limitations.

EEG Analysis Mapping: Ryszard reviewed the four main analysis types from the preprint, explaining which can be directly applied to EEG—such as event-triggered averages, mass univariate ERP analysis, and multivariate decoding—and which are limited by EEG's spatial resolution.

Baseline Correction and Artifacts: The group discussed baseline correction methods, with Ryszard and Krzysztof Basiński explaining the challenges of stimulus-evoked baseline shifts in EEG and the need to keep pre-stimulus activity consistent, contrasting with single-neuron recordings in mice.

Population Tuning and Predictive Coding: Ryszard described how population tuning curves, variance partitioning, and predictive activity analyses can be adapted for EEG using multivariate techniques, and Alexander suggested representational similarity analysis (RSA) for cross-modal comparisons.

Layer-Specific and Connectivity Analyses: Jerome and Ryszard discussed the feasibility of inferring layer-specific activity and connectivity from EEG using Bayesian models, noting validation studies and the limitations compared to invasive recordings.

Exploratory and Individual Difference Analyses: Ryszard outlined plans for exploratory analyses linking excitability traits, resting-state EEG metrics, and behavioral thresholds to mismatch responses, using questionnaires and psychophysics to assess individual differences.

Collaboration, Data Sharing, and Recruitment: Ryszard, Krzysztof Basiński, and Jerome discussed ongoing recruitment of labs for the project, data sharing logistics, and criteria for participation, with Sam inquiring about lab requirements and Alexander proposing strategies for linking resources and analyses across species and modalities.

Lab Recruitment Criteria: Sam asked about requirements for joining the collaboration, and Krzysztof Basiński clarified that labs need a setup for visual stimulus presentation and preferably a 64-channel EEG system, with differences in systems modeled as site effects in analysis.

Data and Code Sharing Platforms: Ryszard and Krzysztof Basiński described plans to use OpenNeuro, Zenodo, and BIDS format for data sharing, with code hosted on GitHub, and discussed the advantages of standardized formats for transparency and global accessibility.

Cross-Species Analysis and Resource Linking: Alexander and Jerome emphasized the importance of linking analyses and resources across mouse, primate, and human studies, proposing website modifications and encouraging joint analyses by students and postdocs to foster concrete cross-species comparisons.

Collaboration Status and Expansion: Ryszard reported confirmed collaborations with labs in Maastricht, Oslo, and ongoing discussions with labs in France and Germany, inviting further participation and outlining the timeline for data collection and publication.

Experimental Procedures and Participant Instructions: Ryszard and Krzysztof Basiński explained the experimental procedures for human EEG sessions, including participant instructions, psychophysics, and measures to prevent drowsiness, with Jerome and Sam clarifying aspects of subject awareness and session logistics.

Participant Instructions: Participants are instructed to passively view sequences of visual stimuli, with explicit instructions only for the psychophysics portion, ensuring naivety regarding the experimental design.

Preventing Drowsiness: To prevent participants from falling asleep during passive EEG recordings, the team asks them to listen to podcasts, paralleling the use of silent movies in auditory mismatch paradigms.

Technical Implementation and Software Choices: Jerome, Ryszard, and Krzysztof Basiński discussed the technical implementation of stimulus presentation, noting the need to adapt Bonsai code for EEG labs and the preference for PsychoPy or PsychToolbox, with considerations for compatibility and reproducibility.

Stimulus Presentation Software: EEG labs typically use PsychoPy (Python-based) or PsychToolbox (MATLAB-based) for stimulus presentation, requiring adaptation of Bonsai code originally used in mouse experiments for compatibility with human EEG setups.

💬 Join the discussion for this page on GitHub