Authors: Fauve Duquette-Laplante¹; Cyril Alex¹; Shivani Prabhu¹; Varghese Peter²; Mridula Sharma¹
¹Flinders University; ²University of the Sunshine Coast
Background: Cortical tracking of the speech envelope, measured using multivariate temporal response function (mTRF) analysis, is used to investigate neural speech processing in noise. While low levels of noise have been associated with enhanced neural responses (potentially reflecting stochastic resonance), repeated exposure to speech may also influence tracking through attentional or learning effects. This study examined the impact of noise on envelope features and the priming effect of noise on listening in quiet.
Method: In Experiment 1 (EXP1), 23 young adults listened to continuous speech in four conditions: QuietA, +5 dB SNR, QuietB, and −5 dB SNR. In Experiment 2 (EXP2), 9 participants followed a sequence (Quiet1, +5, Quiet2, -5, Quiet3, and Quiet4) to further investigate priming. EEG data were analyzed using mTRF envelope reconstruction, and predictive correlations were compared across conditions.
Results: Regarding the impact of noise, EXP1 showed significantly higher predictive correlations for +5 dB SNR than for QuietA (p < .001) and −5 dB SNR (p < .001). Regarding the priming effect, EXP1 revealed that QuietB had a significantly higher predictive correlation than QuietA (p < .001). Similarly, EXP2 showed that subsequent quiet conditions (Quiet3 and Quiet4) had significantly stronger predictive correlations than the initial Quiet1.
Conclusion: Although enhanced neural tracking in moderate noise (+5) may appear consistent with stochastic resonance, these findings suggest that priming effects or statistical learning cause preceding noise exposure to affect subsequent speech perception in quiet. This highlights the critical importance of presentation order and exposure effects when designing and interpreting EEG paradigms.
