Linear modeling of brain activity during selective attention to continuous speech: the critical role of the N1 effect in event-related potentials to acoustic edges

Mai, Adrian; Hillyard, Steven A.; Strauss, Daniel J. · 2023 · Crossref

DOI: 10.1101/2023.07.14.548994

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the relationship between traditional event-related potentials (ERPs) and modern linear regression-based modeling techniques used to analyze cortical tracking of continuous speech. Specifically, it addresses whether speech-evoked ERPs, particularly the N1 component enhanced by selective attention (the N1 effect), drive the performance of linear models in decoding attended speech. The authors hypothesize that the consistent enhancement of the N1 component improves the signal-to-noise ratio (SNR), thereby facilitating the neural tracking of selected auditory streams. The researchers re-analyzed a publicly available EEG dataset from a selective auditory attention task involving 18 participants. Participants focused on one of two competing speakers presented at ±60° azimuth in an anechoic environment. The study employed two primary analytical approaches. First, EEG data were segmented based on "acoustic edges" derived from speech onset envelopes to extract speech-evoked ERPs. The consistency of these ERPs was quantified using wavelet phase synchronization stability (WPSS) in the theta band. Second, linear forward and backward models were applied to estimate temporal response functions (TRFs) and perform stimulus reconstruction (SR), respectively. The forward models mapped speech features to neural responses, while the backward models reconstructed speech envelopes from EEG data to assess tracking accuracy. The results demonstrated a strong spatiotemporal correspondence between the modeled TRFs and the extracted speech-evoked ERPs, both exhibiting clear P1-N1-P2 complexes. Crucially, the TRFs and ERPs showed enhanced N1 amplitudes for attended streams compared to ignored ones. The study found that stimulus reconstruction accuracies were driven by this consistent attention-related enhancement of the N1 component. Furthermore, the WPSS analysis revealed that the stability of the instantaneous ERP phase in the theta band—interpreted as a measure of sustained SNR enhancement—correlated with decoding performance. The attended speech elicited more phase-locked N1 responses, resulting in higher reconstruction accuracy than the ignored speech. These findings establish a direct link between classical ERP components and linear modeling techniques in speech processing. The authors conclude that the N1 effect is critical for achieving accurate tracking of selectively attended speech. By improving the SNR through consistent attentional enhancement of the N1 component, the brain facilitates the higher-order processing of the selected auditory stream. This work bridges the gap between transient ERP paradigms and continuous neural tracking models, suggesting that the mechanisms underlying selective attention in naturalistic listening scenarios are rooted in the same gain-control operations observed in traditional auditory ERP research.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• attention selective divided