Sequential neural dynamics underlie unconscious integration and conscious perception of visual stimuli

This repository contains the code and data necessary to reproduce the results reported in Menétrey, Herzog & Pascucci, “Sequential neural dynamics underlie unconscious integration and conscious perception of visual stimuli.” In this study, we analyzed EEG recordings from 18 participants performing the Sequential Metacontrast Paradigm (SQM), in which a stream of lines and vernier offsets is integrated over several hundred milliseconds without conscious awareness (for further details, see Drissi et al., 2021; https://doi.org/10.1038/s41467-019-12919-7). Five conditions were randomly presented: a no-vernier condition (NV; only a sequence of straight lines), three single-vernier conditions (V0, V2, V4; with a vernier in the central line, the second flanker, or the fourth flanker), and two dual-vernier conditions with opposite offsets (V0–AV2 and V0–AV4; consisting of a vernier in the central line paired with a vernier of opposite offset in the second or fourth flanker). Participants were asked to report the perceived vernier offset. Description of main folders Data folder The folder "1. Data" contains 18 subject subfolders. Each subject folder includes: EEG files (.set): Preprocessed EEG data (see manuscript for preprocessing details). Behavior files (.mat): Trial information. Key variables include: condition (tbl.labels; NV = 0, V0 = 1, V2 = 2, V4 = 3, V0–AV2 = 4, V0–AV4 = 5), central vernier offset direction (voffsdir; −1 = left, 1 = right), and participant response accuracy (hits; 1 = correct identification of the central vernier offset, 0 = incorrect). EEG decoder files (.mat): EEG and behavioral data formatted for decoding analyses. Analysis folder The folder "2. Analysis" contains all scripts required to reproduce the analyses, as well as outputs supporting the manuscript findings. 1. PrepareDataForDecoding) Script that prepares EEG and behavioral data for decoding analyses. EEG data are resampled, re-epoched (−0.2 to 1 s), and z-scored. The resulting .mat files are required for subsequent analyses and are also provided in the repository (EEG decoder files in 1. Data). 2. Scripts) Contains subfolders with all analysis scripts: 1. Behavioral: plots and compares performance across conditions. Performance is defined as the proportion of responses matching the direction of the first vernier offset in the stream. 2. EEG decoding: scripts for LDA with temporal generalization and cross-condition decoding analyses. 3 EEG activation patterns: scripts for estimating EEG activation patterns contributing to decoding, along with clustering analyses revealing distinct topographical maps. 4 ERP: script examining the relationship between topographical maps and classical ERP components. 3. Results) Contains all outputs supporting the manuscript findings: 1. Behavior: Excel file with average performance across conditions (used for Figure 1D). 2. Decoding: Results from all decoding analyses (Figures 2–6 in the main manuscript; Figures S1–S2 in the Supplementary Information). Results of linear discrimant analysis (LDA) with temporal generalization (.mat files): NV vs V conditions (Figure 2 and 3) NV vs V-AV conditions (Figure 4) V0 vs V-AV (Figure 5) Between V-AV conditions (Figure 5) correct vs incorret response in V conditions (Figure 6) V0 vs V2 or V4 and V2 vs V4 (S1 Fig) NV vs V-AV trials where either the first or second vernier was reported (S2 Fig) Results of LDA with cross-condition generalization (.mat files): Between V conditions (training: V2 or V4 vs NV, testing: V0 vs NV) (Figure 2) Between V0 and V-AV conditions (training: V0 vs NV, testing: V0-AV2 or V0-AV4 vs NV) (Figure 4) Between V0 and V-AV trials where either the first or second vernier was reported (training: V0 vs NV, testing: V0-AV2 (V0 reported), V0-AV2 (AV reported), V0-AV4 (V0 reported), or V0-AV4 (AV reported)) (S2 Fig) 3. Prototypes: Information on topographical maps temporally correlated with activation patterns identified in decoding analyses (Figures 5–6; S1 Fig). 4. ERP: MAT file used to reproduce ERP plots (S3 Fig). Function folder The folder "3. Function" contains all custom MATLAB functions required for analysis and visualization. --------------- **1. System requirements** Software, toolboxes, and functions required: MATLAB (MathWorks Inc., Natick, MA, USA): The scripts were executed using MATLAB R2022b. EEGLAB: The scripts were executed using EEGLAB v2024.2, which can be downloaded from the official EEGLAB website (https://sccn.ucsd.edu/eeglab/download.php). Custom-made functions written in MATLAB, which can be downloaded from this repository (EEGdecoding\_SQM/3. Functions). **2. Installation guide** To reproduce the results, download the full repository while preserving the folder structure. **3. Instructions for use** Before running the scripts in 2. Analysis, ensure that the main path is set to the location of the Data folder. In addition, add both EEGLAB and the 3. Function folder to the MATLAB path. --------------- Additional information can be requested by writing to [e-mail].