CoChat

The DOI matches its Crossref record; 68 of the 68 references it cites were found in a registry or an index; none of the 46 sources its claims cite is checked for support yet.

Effects of bimodal divided attention on cortical representations of linguistic context during continuous speech perception in noise

Open access
Xie, Zilong · 2026 · bioRxiv (Cold Spring Harbor Laboratory)

Abstract

Speech perception often takes place in environments with competing sensory inputs, both within the auditory modality and across modalities; for example, following a conversation in a noisy cafe while simultaneously reading a menu. This study examined the extent to which dividing attention between auditory and visual modalities (bimodal divided attention) influences linguistic context processing across hierarchical levels during continuous speech perception in noise. Electroencephalographic (EEG) responses were recorded while participants listened to audiobook stories in multitalker babble as a secondary task, concurrently performing a demanding primary visual task that imposed either low or high cognitive load. Behaviorally, speech comprehension accuracy was significantly lower under high-load than low-load dual-task conditions. Multivariate temporal response function (mTRF) encoding models were used to predict EEG responses from information-theoretic measures (entropy and surprisal) indexing linguistic context at sublexical, word-form, and sentence levels. Significant neutral tracking was observed at the word-form and sentence levels, but not the sublexical level. Critically, neutral tracking of sentence-level linguistic representations was significantly reduced under high compared to low load, with effects emerging at latencies beyond 200 ms. In contrast, neutral tracking of word-form-level representations was unaffected by dual-task load. mTRF analyses further revealed that neutral tracking of acoustic features was not modulated by dual-task load. These findings indicate that bimodal divided attention selectively disrupts cortical representations of sentence-level linguistic context, while lower-level processing remains relatively preserved. Such impairments in higher-level linguistic processing may contribute to reduced speech comprehension during multitasking in noisy environments.

Overview

CoChat Summary

TL;DR

Bimodal divided attention during continuous speech-in-noise selectively reduced cortical tracking of sentence-level linguistic context, while word-form and acoustic tracking were largely unaffected. Higher visual task load also impaired behavioral speech comprehension and enhanced visual tracking.

Key findings

  • High visual load lowered behavioral speech comprehension: story understanding was 49.6% (SD = 23.4%) vs. 77.0% (SD = 18.0%) in low load, t(23) = -5.983, p < 0.001; perceived effort was 8.5 (SD = 1.50) vs. 4.83 (SD = 2.22), t(23) = 10.498, p < 0.001. Visual accuracy was lower and visual RT slower under high load.
  • Sentence-level linguistic tracking was significantly reduced under high vs. low load: Δ% explained mean = 0.006 (SD = 0.013) vs. 0.012 (SD = 0.012), t(23) = 1.818, p = 0.041, Cohen’s d = 0.499, BF = 1.771; mTRF amplitudes were smaller from 220 to 300 ms (p = 0.005).
  • Word-form tracking was robust but not significantly affected by load: t(23) = 0.618, p = 0.271, Cohen’s d = 0.176, BF = 0.511. Sublexical predictors did not significantly enhance model performance (tmax = 2.64, p = 0.220).
  • Acoustic tracking did not significantly change with load: envelope t(23) = 0.34, p = 0.365, d = 0.076, BF = 0.454; onsets t(23) = 1.498, p = 0.074, d = 0.332, BF = 1.147. Visual tracking was enhanced under high load: Δ% explained mean = 0.907 (SD = 0.504) vs. 0.573 (SD = 0.469), t(23) = 3.571, p < 0.001, d = 0.684, BF = 45.993.

Methods

  • 24 younger adults (20 females), aged 18–23, native English speakers with normal hearing and normal or corrected-to-normal vision; EEG recorded with 64 active electrodes at 5 kHz.
  • Dual-task paradigm: primary visuospatial n-back task under low load (0-back) or high load (3-back), while participants listened to ~60-s excerpts from *Alice’s Adventures in Wonderland* in multitalker babble at 5 dB SNR; 15 trials per condition, with story comprehension questions.
  • Multivariate temporal response function (mTRF) encoding models predicted EEG from visual, acoustic (envelope and onset spectrograms), and linguistic predictors (surprisal and entropy at sublexical, word-form, and sentence levels); models used 5-fold cross-validation and cluster-based permutation tests.

Limitations

  • Stated: the study used a fixed segment of multi-talker babble at a single SNR of 5 dB, so it is unclear whether findings extend to other noise types or more negative/favorable SNRs; dual-task costs may also depend on task pairing, such as visual-object vs. visual-spatial tasks.
  • Evident from the text: the sample was limited to 24 younger adults aged 18–23, mostly female, and auditory stimuli were presented monaurally to the right ear; the current study compared only low- and high-load dual-task conditions, with no auditory single-task condition.
  • The paper notes that sublexical tracking was not significant here, unlike prior quiet-listening studies, and suggests this may be because sublexical context processing is more vulnerable to noise interference.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.64898/2026.04.28.721419

Cites

68 of 68 references verified0 citation errors · 0 not checkable

Statistics

Figures

Image 1
CC-BY-NC-ND · From the PDF
Figure 2 Behavioral performance on visual and auditory tasks. (A) Accuracy for the low load (0-back) and high load (3-back) visual tasks, computed as hit rate (i.e., correct target responses) minus false alarm rate (i.e., identifying a non-target as being a target). (B) Median reaction time (RTs) for hits only (correct target responses) in the low- and high-load visual tasks. (C) Accuracy on the auditory task, quantified as the proportion of story comprehension questions answered correctly. Each line in panels (A to C) corresponds to an individual participant (n = 24).
Figure 2 Behavioral performance on visual and auditory tasks. (A) Accuracy for the low load (0-back) and high load (3-back) visual tasks, computed as hit rate (i.e., correct target responses) minus false alarm rate (i.e., identifying a non-target as being a target). (B) Median reaction time (RTs) for hits only (correct target responses) in the low- and high-load visual tasks. (C) Accuracy on the auditory task, quantified as the proportion of story comprehension questions answered correctly. Each line in panels (A to C) corresponds to an individual participant (n = 24).CC-BY-NC-ND · From the PDF
Image 3
CC-BY-NC-ND · From the PDF
Figure 4 Neural tracking of word-form level linguistic predictors across dual-task conditions . (A) The topographic map illustrates the increase in prediction power (expressed as Δ % variance explained) attributed to word-form level predictors (word-form surprisal and entropy). A statistically significant increase was observed within a single cluster, marked by the highlighted yellow electrodes. (B) Prediction power across dual-task conditions, with blue lines representing individual participants. A black asterisk marks whether the prediction power for
Figure 4 Neural tracking of word-form level linguistic predictors across dual-task conditions . (A) The topographic map illustrates the increase in prediction power (expressed as Δ % variance explained) attributed to word-form level predictors (word-form surprisal and entropy). A statistically significant increase was observed within a single cluster, marked by the highlighted yellow electrodes. (B) Prediction power across dual-task conditions, with blue lines representing individual participants. A black asterisk marks whether the prediction power forCC-BY-NC-ND · From the PDF
Figure 5 Neural tracking of the acoustic envelope of speech across dual-task conditions .
Figure 5 Neural tracking of the acoustic envelope of speech across dual-task conditions .CC-BY-NC-ND · From the PDF
Image 6
CC-BY-NC-ND · From the PDF
Figure 7 Neural tracking of visual stimuli across dual-task conditions . (A) The topographic map illustrates the increase in prediction power (expressed as Δ % variance explained) attributed to visual stimulus predictors. A statistically significant increase was observed within a single cluster, marked by the highlighted yellow electrodes. (B) Prediction power across dual-task conditions, with blue lines representing individual participants. A red asterisk indicates the statistical significance of the difference between conditions, while a black asterisk marks whether the prediction power for each condition was significantly above zero. Error bars reflect the 95% within-subject confidence interval (Loftus & Masson, 1994). (C) The global field power of the visual temporal response functions (TRFs). Each visual stimulus was displayed for a period ranging from 0 to 500 ms. Shaded regions represent within-subject standard errors around the mean (color coding as in panel B). Blue horizontal lines highlight time intervals where TRFs
Figure 7 Neural tracking of visual stimuli across dual-task conditions . (A) The topographic map illustrates the increase in prediction power (expressed as Δ % variance explained) attributed to visual stimulus predictors. A statistically significant increase was observed within a single cluster, marked by the highlighted yellow electrodes. (B) Prediction power across dual-task conditions, with blue lines representing individual participants. A red asterisk indicates the statistical significance of the difference between conditions, while a black asterisk marks whether the prediction power for each condition was significantly above zero. Error bars reflect the 95% within-subject confidence interval (Loftus & Masson, 1994). (C) The global field power of the visual temporal response functions (TRFs). Each visual stimulus was displayed for a period ranging from 0 to 500 ms. Shaded regions represent within-subject standard errors around the mean (color coding as in panel B). Blue horizontal lines highlight time intervals where TRFsCC-BY-NC-ND · From the PDF

Verification

Claims the paper makes

0 Valid · 0 Not valid · 24 Not checked · 55 Own results and interpretations

42 more claims

Create a free account to see all 79 claims and to verify the 40 references still open.

See all claims — free account
  • It may state two separate claims as one. 17
  • The quote only partly states this claim. 3
  • Probably not a substantial claim. 18

Create a free account to see these claims.

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Not confirmedt(23) = 0.34reported p = .365 · recomputed p = .737

    “t(23) = 0.34, p = .365”

  • Consistentt(23) = 10.498reported p < .001 · recomputed p < .001

    “t(23) = 10.498,p < .001”

  • Consistentt(23) = -5.983reported p < .001 · recomputed p < .001

    “t(23) = -5.983,p < .001”

  • Consistentt(23) = 8.791reported p < .001 · recomputed p < .001

    “t (23) = 8.791, p < .001”

  • Consistentt(23) = -6.866reported p < .001 · recomputed p < .001

    “t (23) = -6.866,p < .001”

5 of 9 · the rest aren’t loaded

4 more recomputed tests

Create a free account to see every test the Stats check recomputed.

See all tests — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 68 references

Research works printed but not linked: 2

  1. 10.1016/j.cub.2018.10.042 · DOI · Paper page
  2. 10.2139/ssrn.4921233 · doi:10.2139/ssrn.4921233

Cited by

Still being checked.

Metadata

Authors
Xie, Z.
Affiliations
Florida State University (US)
Corresponding author
Xie, Zilong
Venue
bioRxiv (Cold Spring Harbor Laboratory)
Publisher
openRxiv
Published
2026-05-01
OpenAlex
W7160077270
Cited by
0
References
68
Type
posted content
Language
en
Open access
Locations
bioRxiv (Cold Spring Harbor Laboratory) (submittedVersion, cc-by-nc-nd)
Funders
Florida State University
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neurobiology of Language and Bilingualism
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Phonetics and Phonology Research
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Multisensory perception and integration
Keywords
speech perception in noise, multi-talker babble, sentence-level representation, sublexical processing, cognitive load, neural encoding models, speech understanding, multitasking

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-19, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-19 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Effects of bimodal divided attention on cortical representations of linguistic context during continuous speech perception in noise. https://app.cochat.ai/papers/10.64898/2026.04.28.721419
The paper
Xie, Z. (2026). Effects of bimodal divided attention on cortical representations of linguistic context during continuous speech perception in noise. bioRxiv (Cold Spring Harbor Laboratory). https://doi.org/10.64898/2026.04.28.721419

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.