Auditory-motor entrainment and listening experience shape the perceptual learning of concurrent speech

27

Plasticity from auditory experience shapes the brain’s encoding and perception of 28 sound. Though prior research demonstrates that neural entrainment (i.e., brain-to-acoustic 29 synchronization) aids speech perception, how long- and short-term plasticity influence 30 entrainment to concurrent speech has not been investigated. Here, we explored neural entrainment 31 mechanisms and the interplay between short- and long-term neuroplasticity for rapid auditory 32 perceptual learning of concurrent speech sounds in young, normal-hearing musicians and 33 nonmusicians. 34

Participants learned to identify double-vowel mixtures during ~45 min training sessions 35 with concurrent high-density EEG recordings. We examined the degree to which brain responses 36 entrained to the speech-stimulus train (~9 Hz) to investigate whether entrainment to speech prior to 37 behavioral decision predicted task performance. Source and directed functional connectivity 38 analyses of the EEG probed whether behavior was driven by group differences auditory-motor 39 coupling. 40

Both musicians and nonmusicians showed rapid perceptual learning in accuracy with 41 training. Interestingly, listeners’ neural entrainment strength prior to target speech mixtures 42 predicted behavioral identification performance; stronger neural synchronization was observed 43 preceding incorrect compared to correct trial responses. We also found stark hemispheric biases in 44 auditory-motor coupling during speech entrainment, with greater auditory-motor connectivity in 45 the right compared to left hemisphere for musicians (R>L) but not in nonmusicians (R=L). 46

Our findings confirm stronger neuroacoustic synchronization and auditory-motor 47 coupling during speech processing in musicians. Stronger neural entrainment to rapid stimulus 48 trains preceding incorrect behavioral responses supports the notion that alpha-band (~10 Hz) 49 arousal/suppression in brain activity is an important modulator of trial-by-trial success in 50 perceptual processing. 51

52 Neural entrainment; Concurrent speech perception; Functional Connectivity; Auditory-motor 53 interactions 54 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint

55 Everyday listening involves complex auditory scenarios in which listeners must isolate information 56 from one talker in the presence of other talkers and background noise. Though difficult, many 57 listeners successfully navigate these types of “cocktail party” listening environments. In particular, 58 an extensive body of literature demonstrates perceptual advantages in speech-in-noise and 59 “cocktail party” listening among highly-trained musicians (Bidelman & Yoo, 2020; Maillard et al., 60 2023; Parbery-Clark et al., 2009; Puschmann et al., 2018; Zendel & Alain, 2009). Despite evidence 61 for a musician speech-in-noise advantage, the exact mechanism(s) underlying these 62 enhancements are still under investigation. Enhanced sensory processing (Bidelman et al., 2011; 63 Koelsch et al., 1999; Strait et al., 2010), attention (Román-Caballero et al., 2020; Strait & Kraus, 64 2011), and working memory/executive function (Kraus et al., 2012; Pallesen et al., 2010; Zuk et al., 65 2014) all might explain musicians’ superior figure-ground speech perception abilities. 66 One potential facilitatory mechanism that might enhance auditory perception, including 67 noise-degraded and concurrent speech perception, is neural entrainment. Neural entrainment, or 68 the yoking of ongoing neural oscillations to external stimuli, plays a strong role in governing the 69 perceptual parsing of speech (Vanthornhout et al., 2018) and musical sounds (Doelling et al., 70 2019). Entraining to speech facilitates its intelligibility both in quiet and noise (Riecke et al., 2018). 71 On the contrary, electrophysiological studies have shown that poorer entrainment in clinical 72 populations (e.g., listeners with auditory processing disorder) parallels behavioral deficits in 73 concurrent speech listening tasks (Gilley et al., 2016; Momtaz et al., 2021; Momtaz et al., 2022). 74 Collectively, these studies suggest the robustness of the brain’s neuroacoustic entrainment might 75 play an important role in successfully parsing concurrent speech signals. 76 In addition to their improved speech-in-noise performance, studies demonstrate musicians 77 have stronger neural entrainment to musical stimuli in the beta range (~20 Hz) that are associated 78 with rhythmic processing and forming temporal predictions (Doelling & Poeppel, 2015). Such 79 rhythmic brain oscillations have also been linked to sensorimotor synchronization ability (Arnal et 80 al., 2014; Krause, Pollok, et al., 2010; Krause, Schnitzler, et al., 2010). Indeed, a relationship 81 between enhanced neural entrainment strength and successful sensorimotor synchronization 82 (such as tapping along with a beat) has also been demonstrated on an individual level (Nozaradan 83 et al., 2016). In addition to the beta band, modulations in alpha entrainment (~10 Hz) can also 84 influence speech perception. Increases in alpha brain rhythms are traditionally associated with 85 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint internal reflection or decreased attention to a given task (Klimesch, 2012). Related to task 86 performance, phase-related changes and suppression of alpha activity have been shown to predict 87 successful speech intelligibility in quiet and noise (Obleser et al., 2012; Strauß et al., 2015; Weisz et 88 al., 2011). Germane to our current study, Puschmann and colleagues found that when attending to 89 continuous speech in quiet, the amount of participants’ music training positively correlated with 90 the strength of alpha-band phase locking between the primary auditory cortex and dorsal and 91 ventral auditory pathways, suggesting alpha-band entrainment to speech across the cortex is 92 influenced by music training (Puschmann et al., 2021). Cortical alpha states also influence 93 brainstem speech encoding through dynamic fluctuations in arousal and attention, and 94 consequently, are relevant to speech processing at multiple stages of the auditory system (Lai et 95 al., 2022). Thus, as evidenced by changes in several prominent time-frequency signatures of the 96 EEG, studies support the notion that musicianship might enhance the brain’s ability to entrain to 97 external acoustic sounds. Consequently, the current study aimed to determine whether musicians’ 98 extensive experience with sensorimotor synchronization and enhanced neural entrainment might 99 also enhance aspects of concurrent speech listening. 100 In addition to speech-to-brain entrainment, brain-to-brain interactions between the 101 auditory and motor systems might also aid the perception of “cocktail party” speech. Previous 102 studies have demonstrated engagement of the motor system (alongside the auditory system) to 103 enhance the neural representation of speech (Poeppel & Assaneo, 2020; Poeppel & Hickok, 2004). 104 Indeed, close coordination between the premotor and temporal cortices is used to track various 105 linguistic elements of the speech signal spanning the syllable, word, and phrase levels (Assaneo & 106 Poeppel, 2018; Ding et al., 2016; He et al., 2023; Keitel et al., 2018). Motor engagement is 107 particularly evident under noise degradation when efference copy must enhance speech 108 representations from the impoverished acoustic input (Du et al., 2014). Such top-down, cross-109 modal enhancement of auditory information might also be due to the ability of the motor system to 110 enhance temporal predictions of sensory stimuli (Dick et al., 2011; Morillon & Baillet, 2017). These 111 mechanisms could presumably improve degraded listening skills. One idea is that the enhanced 112 auditory-motor integration necessary for musicians may enhance auditory-motor connectivity, thus 113 enabling their more successful speech-in-noise comprehension (e.g., Du & Zatorre, 2017). 114 Functional connectivity between the auditory and motor systems (i.e., the degree of 115 coupling between regional activity) can be used to directly characterize auditory-motor signaling. 116 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Indeed, connectivity within the alpha band that tags the speech signal is stronger in musically 117 trained individuals (Puschmann et al., 2021). Functional connectivity enhancements gained 118 through long-term music training may even align with prevention of typical age-related declines in 119 speech-in-noise perception (Zhang et al., 2024). In addition to connectivity strength, the direction 120 of signaling (i.e., auditory-to-motor vs. motor-to-auditory) can provide insight into “bottom up” vs. 121 “top-down” mechanisms of auditory-motor involvement. Stronger connectivity in the auditory-to-122 motor direction could indicate greater reliance on sensory cue extraction and specific stimulus 123 features, whereas stronger motor-to-auditory signaling could indicate greater reliance on predictive 124 or anticipatory cues to perceive cocktail party speech. 125 In the present study, we reanalyzed the EEG data collected in our previously published 126 study on the neuroplasticity of concurrent speech sound learning in musicians and nonmusicians 127 (MacLean et al., 2024). In our prior work, we found that long-term plasticity (e.g., musicianship) 128 interacted with short-term perceptual learning (e.g., learning a task within one ~45 minute session) 129 in the perception of double-vowel speech stimuli. Musicians and nonmusicians demonstrated 130 different cortical (but not subcortical) learning trajectories which related to behavioral measures of 131 speech identification success. Fortuitously, our stimulus design included a rapid cueing speech 132 train which had the natural potential to induce neural entrainment prior to listeners’ behavioral 133 decision (Bidelman, 2015). Here, we performed novel analyses on this critical segment to 134 understand how neural entrainment and auditory-motor connectivity interact with long- and short-135 term auditory experiences during double-vowel learning. We hypothesized that alpha-band 136 entrainment to concurrent speech and auditory-motor connectivity strength would influence 137 behavioral success at the single trial level, and that these oscillatory processes would differ 138 between musician and nonmusician groups. Additionally, we investigated the direction of auditory-139 motor connectivity to understand how the relative signaling between the auditory and motor 140 systems relate to learning and behavioral performance. Our findings support the notion that alpha 141 suppression is critical for task success and bottom-up use of stimulus features for both groups, 142 while revealing differences in neural entrainment based on long-term music training. 143 144 145 146 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint

147 The current study represents a new analysis of neural entrainment from the EEG data reported in 148 MacLean et al. (2024). Evoked potential results including brainstem (FFR) and cortical (ERP) 149 responses to speech and how they are modulated by perceptual learning are reported in the 150 companion paper (MacLean et al., 2024). The reader is referred to the original manuscript for full 151 methodological details. 152 Participants 153 Twenty-seven young adults (ages 18-34; mean + SD: 23.68 + 4.22; 13 female) with normal hearing 154 thresholds (bilateral pure tone averages < 25 dB HL, octave frequencies between 250 and 8000 Hz) 155 participated in this study. All participants were fluent in American English and reported no previous 156 neurologic or psychiatric disorders. Participants gave written, informed consent in accordance with 157 a protocol approved by the Indiana University Institutional Review Board. 158 Participants were separated into musician (M; n = 13) and nonmusician (NM; n = 14) groups 159 based on their extent of formal music training. Musicians had at least 10 years of formal music 160 training starting at or before age 12, while nonmusicians had 5 or fewer years of lifetime music 161 training (Wong et al., 2007). Groups did significantly differ in amount of music training (M: 16.1 + 4.3 162 years; NM: 2.4 +1.7 years; t(25) = 10.93; p < 0.001), but were matched in age (t(25) = 1.58; p = 163 0.413), cognitive ability as assessed through the Montreal Cognitive Assessment (Nasreddine et al., 164 2005) (t(25) = 1.78; p = 0.088), self-reported bilingualism (X2(1, N = 27) = 0.022, p = 0.883), sex 165 balance (X 2(1, N = 27) = 1.78, p = 0.182), and handedness as assessed through the Edinburgh 166 Handedness Inventory (t(25) = -0.615; p = 0.544) (Oldfield, 1971). 167 Double-vowel stimuli and task 168 Concurrent vowel stimuli were modeled after previous studies (Alain et al., 2007; Assmann & 169 Summerfield, 1989, 1990; Bidelman & Yellamsetty, 2017). Stimuli consisted of synthesized, steady-170 state vowels (/a/, /e/, and /i/) which were presented in three unique vowel combinations (i.e., /a/ + 171 /e/; /e/ + /i/; /a/ + /i/). Vowels were never paired with themselves. Stimuli were created with a Klatt-172 based synthesizer (Klatt, 1980) coded in MATLAB (v 2021; The MathWorks, Inc., Natick, MA). Each 173 vowel was 100 ms in duration with 10-ms cos2 onset/offset ramping to prevent spectral splatter. The 174 fundamental frequency (F0) between vowels was 4 semitones (150 and 190 Hz), which promotes 175 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint segregation for most listeners (Assmann & Summerfield, 1990; Bidelman & Yellamsetty, 2017). F0 176 and the first two formant frequencies (F1a,e,i = 787, 583, 300 Hz; F2 a,e,i = 1307, 1753, 2805 Hz) 177 remained constant for the duration of the token. 178 The speech sounds were presented in rarefaction phase through a TDT RZ6 interface 179 (Tucker-Davis Technologies, Alachua, FL) controlled via MATLAB. Stimuli were presented binaurally 180 at 79 dB SPL through electromagnetically shielded (Campbell et al., 2012; Price & Bidelman, 2021) 181 ER-2 insert earphones (Etymotic Research, Elk Grove, IL). Prior to EEG testing, we required all 182 participants to identify single vowels with 100% accuracy. This ensured subsequent learning would 183 be based on improvements in concurrent speech identification rather than isolated sound labeling 184 ability. 185 We used a clustered stimulus paradigm (Bidelman, 2015) employing interspersed fast and 186 slow interstimulus intervals (ISIs) to collect speech-evoked potentials during the active perceptual 187 task (Figure 1). Speech-ERP/FFR data are reported in the companion paper (MacLean et al., 2024). 188 Each trial consisted of one of the three vowel combinations. During a trial, 20 repetitions of the 189 vowel pair were presented with a fast ISI of 10 ms to elicit the FFR. Thus, the corresponding 190 stimulus onset asynchrony (SOA) was 110 ms (i.e., 9.09 Hz). The ISI was then slowed to 1100 ms 191 and a single stimulus was presented to evoke the ERP and cue a behavioral response. Participants 192 then identified both vowels through keyboard responses following the isolated vowel pair. The next 193 trial began after the participants’ response and 250 ms of silence. Participants were asked to 194 identify both vowels as quickly and accurately as possible (no feedback was provided). Double 195 vowel pairs were randomized in order. This identical task was repeated over four learning blocks. In 196 total, each block included 150 stimulus trials. Each block took 10-15 min to complete. Participants 197 were offered a short (2-3 min) break after each block to avoid fatigue. 198 199 200 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Figure 1. Clustered stimulus paradigm to induce alpha-band speech entrainment. The stimulus 201 paradigm began with a rapid stimulus train presented at ~10 Hz, followed by a 1100 ms period of 202 silence before the isolated vowel pair which cued behavioral responses (Bidelman, 2015). Analyses 203 were performed on induced neural entrainment observed during the silent period. 204 205 To investigate neural entrainment induced by the preceding speech stimuli prior to listeners’ 206 behavioral response, we isolated neural activity to the silent stimulus portion (-1100 to 0 ms) 207 immediately following the rapid speech train. This allowed us to assess how ongoing brain rhythms 208 that have entrained to speech after its cessation modulate subsequent success in identification. 209 EEG recording and preprocessing 210 We used Curry 9 (Compumedics Neuroscan, Charlotte, NC) and BESA Research 7.1 (BESA, GmbH) 211 to record and preprocess the continuous EEG data. Continuous EEGs were acquired from 64-212 channel Ag/AgCl electrodes positioned at 10-10 scalp locations (Oostenveld & Praamstra, 2001). 213 Recordings were digitized at 5 kHz using Neuroscan Synamps RT amplifiers. Data were referenced 214 to an electrode placed 1 cm behind Cz during online recording. Data were re-referenced to 215 common average reference for subsequent analysis. Impedances were kept below 25 k. 216 Electrodes placed on the outer canthi of the eyes and superior and inferior orbit captured ocular 217 movements. Eyeblinks were corrected using a topographic principal component analysis 218 (Wallstrom et al., 2004). Responses were collapsed across vowel pairs to obtain an adequate 219 number of trials for analysis (Bidelman & Yellamsetty, 2017; Yellamsetty & Bidelman, 2018). 220 Responses exceeding 150 µV were rejected as further artifacts. We then bandpass filtered 221 responses from 7 to 12 Hz (zero-phase Butterworth filters; slope = 48 dB/octave) to isolate alpha-222 band activity (Alain et al., 2023; Bidelman, 2017; Lai et al., 2022), corresponding to the nominal rate 223 of our speech train stimuli. Data were then epoched during the silent portion of the stimulus 224 presentation (-1100 to 0 ms), baselined, and ensemble averaged to derive sustained response 225 waveforms for each condition per subject. For subsequent analyses, neural responses were 226 separated by listeners’ trial-by-trial response accuracy (correct vs. incorrect trials). 227 Fast-Fourier Transforms 228 To measure the strength of neural entrainment induced by the rapid stimulus train, we computed 229 the Fast Fourier Transform (FFT) in the -1100 to 0 ms time window separately for each block and 230 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint correct/incorrect trials. We measured the magnitude and frequency for the maximum spectral peak 231 within the alpha band (7-12 Hz) at the Cz electrode to quantify entrainment at the scalp level. 232 Functional connectivity 233 To resolve the underlying brain sources of entrainment effects, we measured directional flow of 234 information within auditory-motor networks using Granger Causality (GC) (Geweke, 1982; Granger, 235 1969). GC measures the degree to which Signal A “Granger-causes” Signal B and is computed 236 directionally in order to infer causal flow of information between brain regions. We computed 237 functional connectivity in the frequency domain between primary auditory (A1) and motor (M1) 238 cortices, bilaterally, using BESA Connectivity (v2.0) (Dhamala et al., 2008; Geweke, 1982). A1 and 239 M1 regions of interest (ROI) were defined via Talairach coordinates in template brain space (x, y, z 240 coords.: M1: ±44.8, -7.8, 38.24 cm; A1: ±50.4, -21.7, 11.5 cm). Frequency decomposition was 241 based on complex demodulation (Papp & Ktonas, 1977), which results in uniform frequency 242 resolution across the analysis bandwidth (i.e., sliding window FFT). The time-frequency analysis 243 initially spanned the entire epoch window (-3400 to 1000 ms), using a pre-stimulus baseline (-3400 244 to -3200 ms) over a bandwidth between 5-20 Hz (i.e., centered at the nominal alpha frequency). 245 However, we extracted GC within the post-stimulus train silence (-1100 to 0 ms) within the 7-12 Hz 246 band (collapsed across time and frequency) to examine alpha auditory motor coupling just prior to 247 the target cue and behavioral decision. We computed GC between A1 and M1 in both the forward 248 and reverse directions (A1 → M1 and M1 → A1, respectively) to assess directed “bottom-up” and 249 “top-down” neural signaling between auditory and motor system. 250 Statistical analyses 251 Unless otherwise noted, we analyzed dependent variables using mixed model ANOVAs in R (version 252 4.2.2) (R-Core-Team, 2020) and lme4 package (Bates et al., 2015). Behavioral measures (percent 253 correct, reaction time) were analyzed with fixed effects of group (2 levels), block, (4 levels), 254 Entrainment strength was analyzed with fixed effects of group (2 levels), block (4 levels), behavioral 255 response (2 levels, correct or incorrect), and random effect of subject. Additionally, we included a 256 covariate for the number of trial counts for correct and incorrect responses. Granger Connectivity 257 was analyzed with the same fixed and random effects as above, with two additional fixed effects of 258 hemisphere (2 levels; left vs. right) and direction (2 levels; forward: A1 → M1, reverse: M1 → A1). 259 Effect sizes are reported as partial eta squared (𝜂𝑝2) and degrees of freedom (d.f.) using 260 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Satterthwaite's method. Multiple pairwise comparisons were adjusted using Tukey method. Linear 261 contrasts were adjusted using the Sidak method. 262 Initial diagnostics indicated heavy tailed distributions for both neural measures. 263 Consequently, we used the Box-Cox procedure (Box & Cox, 1964) to transform the data and satisfy 264 normality assumptions necessary for parametric statistics. This procedure transforms the data 265 according to y’=(yλ – 1)/λ, where λ = 0.071 and λ = -0.11 where determined empirically for 266 entrainment strength and connectivity, respectively. 267 268

269 Behavior 270 Figure 2 displays behavioral results for both M and NM groups across all four training blocks. An 271 ANOVA on behavioral accuracy revealed a main effect of training block [F(3, 75) = 12.13, p < 0.001, 272 𝜂𝑝2 = 0.33], where both groups improved in accuracy with training block (linear contrast: M: t(75) = 273 4.34, p < 0.001; NM: t(75) = 3.01, p = 0.0035). These data demonstrate that both groups improved in 274 speech identification accuracy with training. 275 276 Figure 2. Behavioral accuracy increased with training block. Accuracy in identifying both 277 concurrently presented vowels improved with increasing training block for musicians and 278 nonmusicians. Error bars = +1 S.E.M. Reprinted from MacLean et al. (2024), with permission from 279 Oxford University Press. 280 281 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Neural entrainment to speech preceding behavior 282 Figure 3 displays alpha-band waveforms during the entire stimulus period. For subsequent 283 analyses, we focused on the silent period just prior to the target double-vowel presentation that 284 cued listeners’ behavioral response. 285 286 Figure 3. Alpha-band entrainment during the stimulus time course. Alpha-band (7-12 Hz) 287 entrainment waveforms for musicians and nonmusicians preceding correct (a) and incorrect (b) 288 behavioral responses. Gray boxes represent stimulus in the rapid stimulus train (see Fig. 1). To 289 assess true entrainment, analyses were performed during the silent portion of the stimulus 290 paradigm (yellow) just prior to the behavior-cueing token at t = 0 (black box). 291 292 Figure 4 shows alpha-band entrainment amplitude in the pre-stimulus silence period (i.e., 293 just prior to the target cue) for correct and incorrect trials. An ANOVA on alpha-band entrainment 294 amplitude revealed a 2-way interaction between group x trial accuracy [F(1, 172.38) = 4.52, p = 295 0.035, 𝜂𝑝2 = 0.03] (Fig. 4b), driven by larger spectral amplitudes for musicians than nonmusicians 296 preceding incorrect trials [t(29) = 2.20, p = 0.036]. Groups showed similar response amplitudes 297 before correct trials [t(29) = 1.011, p = 0.32]. There was also a block x response accuracy interaction 298 [F(3, 172.38) = 4.014, p = 0.0086, 𝜂𝑝2 = 0.07]. A linear contrast revealed this interaction was due to a 299 steady increase in response amplitude across blocks for incorrect trials [t(177) = 4.18, p < 0.0001], 300 regardless of group. Responses were invariant across blocks for correct trials [t(177) = -0.41, p 301 =0.68]. 302 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint 303 Figure 4. Neural entrainment following a rapid speech stimulus train predicts subsequent 304 behavioral identification accuracy for double-vowel mixtures. (a) FFTs are displayed for 305 musicians and nonmusicians for silences preceding correct and incorrect trials. Insets show time 306 waveforms of the post-train period (see yellow shading, Fig. 3). (b) Musicians had stronger 307 entrained responses preceding incorrect trials than did nonmusicians, despite similar responses 308 preceding correct trials. Error bars/shading = +1 S.E.M. 309 310 Auditory-motor connectivity 311 Figure 5 depicts time-frequency plots of source-level waveforms from auditory (A1) and motor (M1) 312 cortex in the left (LH) and right hemispheres (RH) per group. Spectrographic maps were used to 313 calculate Granger Connectivity (GC), reflecting directed neural signaling between ROIs, for correct 314 and incorrect trials per hemisphere and group (Figure 6). 315 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint 316 Figure 5. Source time-frequency responses reflecting neural entrainment within the auditory-317 motor network. Each spectrogram demonstrates spectral density within the alpha band range 318 stemming from auditory (A1) and motor (M1) cortex. Hot colors, %-increase in activity relative to 319 baseline; cool colors, %-decrease activity. t = 0 denotes the onset of the double-vowel mixture that 320 cued listeners’ behavioral response. Note the power at ~10 Hz reflecting phase-locking to the rapid 321 stimulus train (see Fig. 1a) which is also stronger in musicians. L/R = left/right hemisphere. 322 323 A mixed-model ANOVA on GC strength revealed several two-way interactions (Figure 7). We 324 found an interaction between hemisphere and group (Fig. 7a) [F(1, 766.72) = 9.07, p = 0.0027, 𝜂𝑝2 = 325 0.01]. This was driven by stronger GC values in the right hemisphere compared to the left 326 hemisphere for musicians only (pairwise comparison: M: t(766) = -4.69, p < 0.0001, NM: t(766) = -327 0.555, p = 0.58). An interaction between group and block (Fig. 7b) [F(3, 769.72) = 3.64, p = 0.013, 𝜂𝑝2 328 = 0.01] was driven by greater connectivity with block for musicians only (linear contrast: M: t(791) = 329 3.16, p = 0.0050; NM: t(779) = 1.51, p = 0.35). We also observed an interaction between group and 330 response accuracy (Fig. 7c) [F(1, 766.72) = 5.030, p = 0.025, 𝜂𝑝2 < 0.01] which was driven by stronger 331 GC during incorrect trials in both groups but especially musicians (pairwise comparison: M: t(766) = 332 -13.69, p < 0.0001; NM: t(766) = -11.11, p < 0.0001). Finally, an interaction between block and 333 response accuracy (Fig. 7d) [F(3, 766.72) = 4.55, p = 0.0036, 𝜂𝑝2 = 0.02] was driven by increasing 334 connectivity with block for incorrect trials only (linear contrast: correct: t(787) = 0.12, p = 0.99; 335 incorrect: t(787) = 4.60, p < 0.0001). All other interactions were non-significant. Additionally, we 336 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint observed a main effect of direction (Fig. 7e) [F(1, 766.72) = 4.53, p = 0.034, 𝜂𝑝2 < 0.01], attributed to 337 higher auditory-to-motor (i.e., A1 → M1) compared with motor-to-auditory (i.e., M1 → A1) 338 connectivity in both groups. 339 340 Figure 6. Auditory-motor coupling varies by group, hemisphere, and trial-wise accuracy. 341 Granger connectivity values were strongest for musicians in the right hemisphere preceding 342 incorrect trials. Both groups had weaker connectivity in the motor to auditory direction. Errorbars = 343 +1 S.E.M.344 345 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Figure 7. Significant interactions and main effects on Granger Connectivity. (a) Hemisphere x 346 group, (b) block x group interaction, (c) group x response accuracy, and (d) block x response 347 accuracy interactions. (e) Main effect of direction. Connectivity was stronger in the auditory-motor 348 (bottom-up) vs. motor-auditory (top-down) direction. Errorbars = +1 S.E.M. 349 350

351 By analyzing EEG entrainment during perceptual learning of double-vowel mixtures in musicians 352 and nonmusicians, we found: (i) stronger alpha-band power preceding incorrect responses, 353 especially for musicians, (ii) greater learning-related changes in connectivity for musicians, 354 especially in the right hemisphere and preceding incorrect responses, and (iii) stronger bottom-up 355 (auditory-to-motor) than top-down (motor-to-auditory) connectivity for both groups. 356 Musicians displayed stronger modulation of alpha activity that varied with trial success 357 Musicians displayed stronger alpha-band (7-12 Hz) activity preceding incorrect trials in our double-358 vowel identification task than nonmusicians. As induced entrainment here (~9 Hz) overlaps with the 359 alpha range, increased alpha activity could indicate stronger persistent stimulus entrainment after 360 the sound has stopped (exogenous activity) or reduced outward attention to the stimulus and 361 greater inward reflective processing (endogenous activity) (Klimesch, 2012). Given that increased 362 alpha-band power was observed prior to incorrect responses, the latter interpretation is more 363 plausible. One potential explanation for increased activity associated with reduced attention to 364 task may be that increased alpha power reflects decreased neuronal excitability (e.g., inhibition), 365 which leads to reduced stimulus encoding at the sensory level. As a result of the stimulus not being 366 encoded properly, task performance becomes poorer (Iemi et al., 2022). Increases in alpha activity 367 in the pre-stimulus period could indicate that participants were “tuning out” the trial and therefore 368 responded incorrectly (Klimesch, 2012). 369 Trial-dependent changes in alpha power were stronger for musicians than nonmusicians 370 preceding incorrect trials, despite similar levels of activity between groups preceding correct 371 responses. One explanation for this finding could be that musicians are greater “modulators” of 372 alpha activity. Previous studies have shown that changes in alpha activity are associated with 373 improved task performance (Klimesch, 2012; Lai et al., 2022; Pfurtscheller & Lopes da Silva, 1999; 374 Price et al., 2019; Strauß et al., 2015). We have also recently demonstrated listeners who show less 375 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint stimulus-related changes in their alpha (i.e., “low alpha modulators”) achieve poorer performance 376 on speech-in-noise perception tasks (Price et al., 2019). Alpha desynchronization in sensory, task-377 relevant brain areas may even be paired with alpha synchronization over task-irrelevant areas 378 where inhibition is necessary (Mazaheri et al., 2014). Alpha power is also associated with 379 attentional biasing during auditory processing, including tasks involving the perception of difficult 380 and ambiguous speech (Alain et al., 2023). Greater alpha activity preceding incorrect trials may 381 thus reflect changes in task-related inhibition and/or attentional gating. Indeed, broad increases in 382 prestimulus neural activation predict speech recognition errors (Vaden et al., 2015; Vaden et al., 383 2022). And consistent with our electrophysiological data, alpha power can be stronger preceding 384 incorrect responses (Samaha et al., 2020). Regardless of which interpretation accounts for the 385 alpha effects observed here, it is clear musicians recruit greater changes in alpha power between 386 successful and unsuccessful trials (Fig. 4b). Given musicians’ faster performance in double-vowel 387 identification (MacLean et al., 2024), it would appear that a more dynamic alpha range in brain 388 activity is advantageous. This could reflect musicians’ greater flexibility in deploying attentional 389 resources during speech perception (Strait & Kraus, 2011). This notion converges with previous 390 findings showing that acoustic-phonetic properties of speech indexed by alpha rhythms are 391 amplified in musicians and support more robust categorization in speech perception tasks 392 (Bidelman, 2017). 393 Relatedly, under the attentional interpretation of alpha, increased alpha activity in 394 musicians may be the result of greater “tuning-out, ” or reduced attentional gating to the task during 395 incorrect trials. Musicians’ greater alpha desynchronization for successful trials could reflect 396 stronger attentional modulation or even greater resources for redirecting attention when needed (or 397 desired). Indeed, both groups showed increased behavioral performance with block, but it is 398 conceivable that performance may have become more automatized in musicians during the time 399 course of learning. This is supported by our functional connectivity data. Musicians showed 400 increased alpha-band auditory-motor connectivity with training block, whereas nonmusicians did 401 not. In this vein, prior studies have also shown greater selective auditory attention for musicians in 402 concurrent speech or “cocktail party” scenarios (Brown & Bidelman, 2023; Clayton et al., 2016; 403 Strait et al., 2010), but see (Baumann et al., 2008), and there is also evidence that inhibitory 404 attentional control is stronger and more efficient in musically trained individuals (Medina & Barraza, 405 2019). 406 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Further support for interpretation of increased alpha power as reduced attention to task is 407 supported by findings relating alpha power to creativity (Stevens & Zabelina, 2019). Similarly, 408 greater auditory-motor connectivity for musicians preceding incorrect trials may be the result of 409 increased internal reflection. Reflective processing coincides with decreased arousal/attention 410 observed through alpha desynchronization (Klimesch, 2012). Internal reflections or “daydreaming” 411 prior to incorrect trials could indicate more widespread, inefficient processing unrelated to the task 412 (Fink & Benedek, 2014). Previous studies link stronger alpha activity and resting-state functional 413 connectivity with creativity in long-term trait (Bazanova & Aftanas, 2008; Beaty et al., 2014) and 414 short-term task-related (Stevens & Zabelina, 2019) contexts. Understanding increased alpha 415 activity as reduced attention to task goes hand in hand with greater “tuning out” or internal 416 reflection, though our task did not measure this phenomenon explicitly. 417 Determining the facilitatory or inhibitory role of alpha in concurrent speech listening, as well 418 as how this role may be modulated by long-term music training, could inform future interventions 419 to improve everyday complex listening skills (Gray et al., 2022). For example, the overall power and 420 ability to modulate alpha activity to suppress irrelevant information declines in older listeners, 421 which may render pre-target entrainment weaker and less viable as a mechanism for attentional 422 gating (Klimesch, 1999; Vaden et al., 2012; Wöstmann et al., 2015). As implied by prior behavioral 423 and neuroimaging studies, music engagement might help offset these age-related declines in 424 auditory processing and help fortify the sensory-attentional mechanisms necessary for parsing 425 complex speech mixtures (Bidelman & Alain, 2015; Lu et al., 2022; Zendel & Alain, 2009, 2012; 426 Zendel et al., 2019). 427 Our study only examined induced brain-to-speech alpha entrainment following a cueing 428 rhythmic speech train. Further exploration of the role of such induced (endogenous) alpha 429 entrainment, both to external speech and between brain areas, and how it interacts with stimulus-430 related speech phase-locking (Puschmann et al., 2018) is needed in order to understand 431 neuroplastic changes in neural entrainment and how it benefits concurrent speech perception. In 432 this vein, neurostimulation studies have already demonstrated that enhancing cortical entrainment 433 causally improves comprehension including performance for noise-degraded speech (Guilleminot 434 & Reichenbach, 2022; Wilsch et al., 2018). 435 436 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Auditory-motor connectivity during concurrent speech listening differs based on long-term 437 music training 438 During our active double-vowel perception task, musicians showed greater auditory-motor 439 connectivity in the right hemisphere, whereas nonmusicians displayed similar connectivity in both 440 hemispheres. These results are in line with emerging findings suggesting long-term music training is 441 associated with stronger functional connectivity in the right hemisphere that is associated with 442 preserved speech-in-noise capabilities with age (Zhang et al., 2024). Right hemispheric brain 443 pathways are dominant for pitch and fine spectral processing (Zatorre et al., 2002; Zatorre et al., 444 1992). Thus, greater RH engagement in musicians may indicate their greater “cue-weighting” of 445 pitch-based cues to distinguish vowels during our concurrent speech task, in line with our previous 446 ERP findings of the same data (MacLean et al., 2024). Relatedly, other studies suggest that 447 musicians have stronger right hemisphere entrainment to speech within the alpha band 448 (Puschmann et al., 2021). Nonmusicians’ similar patterns of connectivity between left and right 449 hemispheres may indicate that neither a left-biased linguistic (Hickok & Poeppel, 2007; Mankel et 450 al., 2022) nor right-biased pitch strategy was preferred. Given musicians’ greater speed in the task 451 (MacLean et al., 2024), a pitch-based, spectral strategy may have been advantageous which could 452 explain the larger recruitment of RH activity observed in our data. 453 Auditory-motor connectivity is stronger in the bottom-up vs. top-down direction 454 We found both musicians and nonmusicians had stronger connectivity in the auditory-to-motor 455 compared to motor-to-auditory direction prior to double-vowel identification. The directionality of 456 connectivity provides insight as to whether concurrent speech stimuli were processed in a bottom-457 up (auditory-to-motor) or top-down (motor-to-auditory) manner. Here, greater auditory-motor 458 connectivity preceding behavior may indicate more reliance on the extraction of stimulus-specific 459 features than anticipatory motor representations of the speech stimuli (Morillon & Baillet, 2017; 460 Tian & Poeppel, 2012). One idea is that the motor system becomes involved in speech perception 461 when listening becomes difficult, such as when acoustic input is sparse (Osnes et al., 2011) or 462 speech is presented in noise (Du et al., 2016). As we observed stronger bottom-up connectivity for 463 both groups, stimulus-based feature extraction may be more advantageous than anticipatory 464 timing during our task. That is, the repetitive stimulus train and simultaneous onset for both vowels 465 may have decreased the need for reliance on top-down anticipatory motor-system strategies (Wu 466 et al., 2014). Alternatively, if the task became more automatic with learning, this would tend to 467 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint evoke more bottom-up signaling, which is also enhanced in musicians (Bidelman & Krishnan, 2010; 468 Bidelman et al., 2014; Musacchia et al., 2007; Parbery-Clark et al., 2009; Puschmann et al., 2018). 469 Our stimulus train was also periodic and predictable. It is possible that changes to the timing 470 and/or predictability of speech sounds (e.g., jittered stimulus train) may differentially recruit 471 auditory-motor engagement and alter the direction of connectivity during speech processing (cf. 472 Momtaz & Bidelman, 2024; Morillon & Baillet, 2017). Future studies are needed to test these 473 possibilities. 474 AUTHOR CONTRIBUTIONS 475 Jessica MacLean (Design, Data Collection, Data Analysis, Writing), Jack Stirn (Data Collection, 476 Writing), and Gavin Bidelman (Design, Data Collection, Data Analysis, Writing). 477 478 FUNDING 479 This work was supported by the National Institute on Deafness and Other Communication 480 Disorders (R01DC016267 to G.M.B.). 481 Conflict of interest statement: None declared. 482 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint

483 Alain, C., Goke, K., Shen, D., Bidelman, G. M., Bernstein, L. J., & Snyder, J. S. (2023). Neural alpha 484 oscillations index context-driven perception of ambiguous vowel sequences. iScience, 485 26(12), 108457. https://doi.org/10.1016/j.isci.2023.108457 486 Alain, C., Snyder, J. S., He, Y ., & Reinke, K. S. (2007). Changes in auditory cortex parallel rapid 487 perceptual learning. Cerebral Cortex, 17(5), 1074-1084. 488 https://doi.org/10.1093/cercor/bhl018 489 Arnal, L. H., Doelling, K. B., & Poeppel, D. (2014). Delta–Beta Coupled Oscillations Underlie 490 Temporal Prediction Accuracy. Cerebral Cortex, 25(9), 3077-3085. 491 https://doi.org/10.1093/cercor/bhu103 492 Assaneo, M. F ., & Poeppel, D. (2018). The coupling between auditory and motor cortices is rate-493 restricted: Evidence for an intrinsic speech-motor rhythm. Sciences Advances, 4(2), 494 eaao3842. https://doi.org/10.1126/sciadv.aao3842 495 Assmann, P . F ., & Summerfield, Q. (1989). Modeling the perception of concurrent vowels: Vowels 496 with the same fundamental frequency. Journal of the Acoustical Society of America, 85(1), 497 327-338. https://doi.org/10.1121/1.397684 498 Assmann, P . F ., & Summerfield, Q. (1990). Modeling the perception of concurrent vowels: vowels 499 with different fundamental frequencies. Journal of the Acoustical Society of America, 88(2), 500 680-697. https://doi.org/10.1121/1.399772 501 Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using 502 lme4. Journal of Statistical Software, 67(1), 1-48. https://doi.org/10.18637/jss.v067.i01 503 Baumann, S., Meyer, M., & Jancke, L. (2008). Enhancement of auditory-evoked potentials in 504 musicians reflects an influence of expertise but not selective attention. Journal of Cognitive 505 Neuroscience, 20(12), 2238-2249. https://doi.org/10.1162/jocn.2008.20157 506 Bazanova, O. M., & Aftanas, L. I. (2008). Individual measures of electroencephalogram alpha 507 activity and non-verbal creativity. Neuroscience and Behavioral Physiology, 38(3), 227-235. 508 https://doi.org/10.1007/s11055-008-0034-y 509 Beaty, R. E., Benedek, M., Wilkins, R. W., Jauk, E., Fink, A., Silvia, P . J., Hodges, D. A., Koschutnig, K., 510 & Neubauer, A. C. (2014). Creativity and the default network: A functional connectivity 511 analysis of the creative brain at rest. Neuropsychologia, 64, 92-98. 512 https://doi.org/10.1016/j.neuropsychologia.2014.09.019 513 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Bidelman, G. M. (2015). Towards an optimal paradigm for simultaneously recording cortical and 514 brainstem auditory evoked potentials. Journal of Neuroscience Methods, 241, 94-100. 515 https://doi.org/10.1016/j.jneumeth.2014.12.019 516 Bidelman, G. M. (2017). Amplified induced neural oscillatory activity predicts musicians' benefits in 517 categorical speech perception. Neuroscience, 348, 107–113. 518 https://doi.org/10.1016/j.neuroscience.2017.02.015 519 Bidelman, G. M., & Alain, C. (2015). Musical training orchestrates coordinated neuroplasticity in 520 auditory brainstem and cortex to counteract age-related declines in categorical vowel 521 perception. Journal of Neuroscience, 35(2), 1240 –1249. 522 https://doi.org/10.1523/JNEUROSCI.3292-14.2015 523 Bidelman, G. M., Gandour, J. T., & Krishnan, A. (2011). Cross-domain effects of music and language 524 experience on the representation of pitch in the human auditory brainstem. Journal of 525 Cognitive Neuroscience, 23(2), 425-434. https://doi.org/10.1162/jocn.2009.21362 526 Bidelman, G. M., & Krishnan, A. (2010). Effects of reverberation on brainstem representation of 527 speech in musicians and non-musicians. Brain Research, 1355, 112-125. 528 https://doi.org/10.1016/j.brainres.2010.07.100 529 Bidelman, G. M., Weiss, M. W., Moreno, S., & Alain, C. (2014). Coordinated plasticity in brainstem 530 and auditory cortex contributes to enhanced categorical speech perception in musicians. 531 European Journal of Neuroscience, 40, 2662-2673. https://doi.org/10.1111/ejn.12627 532 Bidelman, G. M., & Yellamsetty, A. (2017). Noise and pitch interact during the cortical segregation of 533 concurrent speech. Hearing Research, 351, 34-44. 534 https://doi.org/10.1016/j.heares.2017.05.008 535 Bidelman, G. M., & Yoo, J. (2020). Musicians show improved speech segregation in competitive, 536 multi-talker cocktail party scenarios [Brief Research Report]. Frontiers in Psychology, 537 11(1927), 1-11. https://doi.org/10.3389/fpsyg.2020.01927 538 Box, G. E. P ., & Cox, D. R. (1964). An analysis of transformations. Journal of the Royal Statistical 539 Society: Series B - Statistical Methodology, 26(2), 211–252. https://doi.org/10.1111/j.2517-540 6161.1964.tb00553.x 541 Brown, J. A., & Bidelman, G. M. (2023). Attention, Musicality, and Familiarity Shape Cortical Speech 542 Tracking at the Musical Cocktail Party. bioRxiv, 2023.2010.2028.562773. 543 https://doi.org/10.1101/2023.10.28.562773 544 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Campbell, T., Kerlin, J. R., Bishop, C. W., & Miller, L. M. (2012). Methods to eliminate stimulus 545 transduction artifact from insert earphones during electroencephalography. Ear and 546 Hearing, 33(1), 144-150. https://doi.org/10.1097/AUD.0b013e3182280353 547 Clayton, K. K., Swaminathan, J., Yazdanbakhsh, A., Zuk, J., Patel, A. D., & Kidd, G., Jr. (2016). 548 Executive function, visual attention and the cocktail party problem in musicians and non-549 musicians. PloS One, 11(7), e0157638. https://doi.org/10.1371/journal.pone.0157638 550 Dhamala, M., Rangarajan, G., & Ding, M. (2008). Analyzing information flow in brain networks with 551 nonparametric Granger causality. Neuroimage, 41(2), 354-362. 552 https://doi.org/10.1016/j.neuroimage.2008.02.020 553 Dick, F ., Lee, H., Nusbaum, H. C., & J Price, C. (2011). Auditory-motor expertise alters “speech 554 selectivity” in professional musicians and actors. Cerebral Cortex, 21, 938-948. 555 https://doi.org/10.1093/cercor/bhq166 556 Ding, N., Melloni, L., Zhang, H., Tian, X., & Poeppel, D. (2016). Cortical tracking of hierarchical 557 linguistic structures in connected speech. Nature Neuroscience, 19(1), 158-164. 558 https://doi.org/10.1038/nn.4186 559 Doelling, K. B., Assaneo, M. F ., Bevilacqua, D., Pesaran, B., & Poeppel, D. (2019). An oscillator 560 model better predicts cortical entrainment to music. Proceedings of the National Academy 561 of Sciences of the United States of America, 116(20), 10113-10121. 562 https://doi.org/10.1073/pnas.1816414116 563 Doelling, K. B., & Poeppel, D. (2015). Cortical entrainment to music and its modulation by expertise. 564 Proceedings of the National Academy of Sciences of the United States of America, 112(45), 565 E6233-6242. https://doi.org/10.1073/pnas.1508431112 566 Du, Y ., Buchsbaum, B. R., Grady, C. L., & Alain, C. (2014). Noise differentially impacts phoneme 567 representations in the auditory and speech motor systems. Proceedings of the National 568 Academy of Sciences of the United States of America, 111(19), 1-6. 569 https://doi.org/10.1073/pnas.1318738111 570 Du, Y ., Buchsbaum, B. R., Grady, C. L., & Alain, C. (2016). Increased activity in frontal motor cortex 571 compensates impaired speech perception in older adults [Article]. Nature 572 communications, 7(12241), 1-12. https://doi.org/10.1038/ncomms12241 573 Du, Y ., & Zatorre, R. J. (2017). Musical training sharpens and bonds ears and tongue to hear speech 574 better. Proceedings of the National Academy of Sciences of the United States of America, 575 114(51), 13579-13584. https://doi.org/10.1073/pnas.1712223114 576 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Fink, A., & Benedek, M. (2014). EEG alpha power and creative ideation. Neuroscience and 577 Biobehavioral Reviews, 44, 111-123. https://doi.org/10.1016/j.neubiorev.2012.12.002 578 Geweke, J. (1982). Measurement of linear dependence and feedback between multiple time series. 579 Journal of the American Statistical Association, 77(378), 304-313. 580 https://doi.org/10.2307/2287238 581 Gilley, P . M., Sharma, M., & Purdy, S. C. (2016). Oscillatory decoupling differentiates auditory 582 encoding deficits in children with listening problems. Clinical Neurophysiology, 127(2), 583 1618-1628. https://doi.org/10.1016/j.clinph.2015.11.003 584 Granger, C. (1969). Investigating causal relations by econometric models and cross-spectral 585 methods. Econometrica, 37(3), 424-438. https://doi.org/10.2307/1912791 586 Gray, R., Sarampalis, A., Başkent, D., & Harding, E. E. (2022). Working-Memory, Alpha-Theta 587 Oscillations and Musical Training in Older Age: Research Perspectives for Speech-on-588 speech Perception [Mini Review]. Frontiers in Aging Neuroscience, 14. 589 https://doi.org/10.3389/fnagi.2022.806439 590 Guilleminot, P ., & Reichenbach, T. (2022). Enhancement of speech-in-noise comprehension 591 through vibrotactile stimulation at the syllabic rate. Proceedings of the National Academy of 592 Sciences of the United States of America, 119(13), e2117000119. 593 https://doi.org/10.1073/pnas.2117000119 594 He, D., Buder, E. H., & Bidelman, G. M. (2023). Effects of syllable rate on neuro-behavioral 595 synchronization across modalities: Brain oscillations and speech productions. 596 Neurobiology of Language, 4(2), 344-360. https://doi.org/10.1162/nol_a_00102 597 Hickok, G., & Poeppel, D. (2007). The cortical organization of speech processing. Nature Reviews 598 Neuroscience, 8(5), 393-402. https://doi.org/10.1038/nrn2113 599 Iemi, L., Gwilliams, L., Samaha, J., Auksztulewicz, R., Cycowicz, Y . M., King, J.-R., Nikulin, V . V ., 600 Thesen, T., Doyle, W., Devinsky, O., Schroeder, C. E., Melloni, L., & Haegens, S. (2022). 601 Ongoing neural oscillations influence behavior and sensory representations by suppressing 602 neuronal excitability. Neuroimage, 247, 118746. 603 https://doi.org/10.1016/j.neuroimage.2021.118746 604 Keitel, A., Gross, J., & Kayser, C. (2018). Perceptually relevant speech tracking in auditory and motor 605 cortex reflects distinct linguistic features. PLoS Biology, 16(3), e2004473. 606 https://doi.org/10.1371/journal.pbio.2004473 607 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Klatt, D. H. (1980). Software for a cascade/parallel formant synthesizer. Journal of the Acoustical 608 Society of America, 67(3), 971-995. https://doi.org/10.1121/1.383940 609 Klimesch, W. (1999). EEG alpha and theta oscillations reflect cognitive and memory performance: a 610 review and analysis. Brain Research Reviews, 29(2), 169-195. 611 https://doi.org/10.1016/S0165-0173(98)00056-3 612 Klimesch, W. (2012). Alpha-band oscillations, attention, and controlled access to stored 613 information. Trends in Cognitive Sciences, 16(12), 606-617. 614 https://doi.org/10.1016/j.tics.2012.10.007 615 Koelsch, S., Schroger, E., & Tervaniemi, M. (1999). Superior pre-attentive auditory processing in 616 musicians. Neuroreport, 10(6), 1309-1313. https://doi.org/10.1097/00001756-199904260-617 00029 618 Kraus, N., Strait, D. L., & Parbery-Clark, A. (2012). Cognitive factors shape brain networks for 619 auditory skills: spotlight on auditory working memory. Annals of the New York Academy of 620 Sciences, 1252, 100-107. https://doi.org/10.1111/j.1749-6632.2012.06463.x 621 Krause, V ., Pollok, B., & Schnitzler, A. (2010). Perception in action: The impact of sensory 622 information on sensorimotor synchronization in musicians and non-musicians. Acta 623 Psychologica, 133(1), 28-37. https://doi.org/10.1016/j.actpsy.2009.08.003 624 Krause, V ., Schnitzler, A., & Pollok, B. (2010). Functional network interactions during sensorimotor 625 synchronization in musicians and non-musicians. Neuroimage, 52(1), 245-251. 626 https://doi.org/10.1016/j.neuroimage.2010.03.081 627 Lai, J., Price, C. N., & Bidelman, G. M. (2022). Brainstem speech encoding is dynamically shaped 628 online by fluctuations in cortical α state. Neuroimage, 263, 119627. 629 https://doi.org/10.1016/j.neuroimage.2022.119627 630 Lu, J., Moussard, A., Guo, S., Lee, Y ., Bidelman, G. M., Moreno, S., Skrotzki, C., Bugos, J., Shen, D., 631 Yao, D., & Alain, C. (2022). Music training modulates theta brain oscillations associated with 632 response suppression. Annals of the New York Academy of Sciences, 1516(1), 212-221. 633 https://doi.org/https://doi.org/10.1111/nyas.14861 634 MacLean, J., Stirn, J., Sisson, A., & Bidelman, G. M. (2024). Short- and long-term neuroplasticity 635 interact during the perceptual learning of concurrent speech. Cerebral Cortex, 34(2). 636 https://doi.org/10.1093/cercor/bhad543 637 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Maillard, E., Joyal, M., Murray, M. M., & Tremblay, P . (2023). Are musical activities associated with 638 enhanced speech perception in noise in adults? A systematic review and meta-analysis. 639 Current Research in Neurobiology, 4, 100083. https://doi.org/10.1016/j.crneur.2023.100083 640 Mankel, K., Shrestha, U., Tipirneni-Sajja, A., & Bidelman, G. M. (2022). Functional plasticity coupled 641 with structural predispositions in auditory cortex shape successful music category learning. 642 Frontiers in Neuroscience, 16. https://doi.org/10.3389/fnins.2022.897239 643 Mazaheri, A., van Schouwenburg, M. R., Dimitrijevic, A., Denys, D., Cools, R., & Jensen, O. (2014). 644 Region-specific modulations in oscillatory alpha activity serve to facilitate processing in the 645 visual and auditory modalities. Neuroimage, 87, 356-362. 646 https://doi.org/10.1016/j.neuroimage.2013.10.052 647 Medina, D., & Barraza, P . (2019). Efficiency of attentional networks in musicians and non-648 musicians. Heliyon, 5(3), e01315. 649 https://doi.org/https://doi.org/10.1016/j.heliyon.2019.e01315 650 Momtaz, S., & Bidelman, G. M. (2024). Effects of stimulus rate and periodicity on auditory cortical 651 entrainment to continuous sounds. eNeuro, 11(3), 1-13. 652 https://doi.org/10.1523/eneuro.0027-23.2024 653 Momtaz, S., Moncrieff, D., & Bidelman, G. M. (2021). Dichotic listening deficits in amblyaudia are 654 characterized by aberrant neural oscillations in auditory cortex. Clinical Neurophysiology, 655 132(9), 2152-2162. https://doi.org/10.1016/j.clinph.2021.04.022 656 Momtaz, S., Moncrieff, D., Ray, M. A., & Bidelman, G. M. (2022). Children with amblyaudia show 657 less flexibility in auditory cortical entrainment to periodic non-speech sounds. International 658 Journal of Audiology, 1-7. https://doi.org/10.1080/14992027.2022.2094289 659 Morillon, B., & Baillet, S. (2017). Motor origin of temporal predictions in auditory attention. 660 Proceedings of the National Academy of Sciences of the United States of America, 114(42), 661 E8913-E8921. https://doi.org/10.1073/pnas.1705373114 662 Musacchia, G., Sams, M., Skoe, E., & Kraus, N. (2007). Musicians have enhanced subcortical 663 auditory and audiovisual processing of speech and music. Proceedings of the National 664 Academy of Sciences of the United States of America, 104(40), 15894-15898. 665 https://doi.org/10.1073/pnas.0701498104 666 Nasreddine, Z. S., Phillips, N. A., Bedirian, V ., Charbonneau, S., Whitehead, V ., Collin, I., Cummings, 667 J. L., & Chertkow, H. (2005). The Montreal Cognitive Assessment, MoCA: A brief screening 668 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint tool for mild cognitive impairment. Journal of the American Geriatrics Society, 53(4), 695-669 699. https://doi.org/10.1111/j.1532-5415.2005.53221.x 670 Nozaradan, S., Peretz, I., & Keller, P . E. (2016). Individual Differences in Rhythmic Cortical 671 Entrainment Correlate with Predictive Behavior in Sensorimotor Synchronization. Scientific 672 Reports, 6(1), 20612. https://doi.org/10.1038/srep20612 673 Obleser, J., Wostmann, M., Hellbernd, N., Wilsch, A., & Maess, B. (2012). Adverse listening 674 conditions and memory load drive a common alpha oscillatory network. The Journal of 675 Neuroscience, 32(36), 12376-12383. https://doi.org/10.1523/JNEUROSCI.4908-11.2012 676 Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. 677 Neuropsychologia, 9, 97-113. https://doi.org/10.1016/0028-3932(71)90067-4 678 Oostenveld, R., & Praamstra, P . (2001). The five percent electrode system for high-resolution EEG 679 and ERP measurements. Clinical Neurophysiology, 112, 713-719. 680 https://doi.org/10.1016/S1388-2457(00)00527-7 681 Osnes, B., Hugdahl, K., & Specht, K. (2011). Effective connectivity analysis demonstrates 682 involvement of premotor cortex during speech perception. Neuroimage, 54(3), 2437-2445. 683 https://doi.org/10.1016/j.neuroimage.2010.09.078 684 Pallesen, K. J., Brattico, E., Bailey, C. J., Korvenoja, A., Koivisto, J., Gjedde, A., & Carlson, S. (2010). 685 Cognitive control in auditory working memory is enhanced in musicians. PloS One, 5(6), 686 e11120. https://doi.org/10.1371/journal.pone.0011120 687 Papp, N., & Ktonas, P . (1977). Critical evaluation of complex demodulation techniques for the 688 quantification of bioelectrical activity. Biomed Sci Instrum, 13, 135-145. 689 Parbery-Clark, A., Skoe, E., & Kraus, N. (2009). Musical experience limits the degradative effects of 690

noise on the neural processing of sound. Journal of Neuroscience, 29(45), 691 14100-14107. https://doi.org/10.1523/JNEUROSCI.3256-09.2009 692 Pfurtscheller, G., & Lopes da Silva, F . H. (1999). Event-related EEG/MEG synchronization and 693 desynchronization: Basic principles. Clinical Neurophysiology, 110, 1842-1857. 694 Poeppel, D., & Assaneo, M. F . (2020). Speech rhythms and their neural foundations. Nature Reviews 695 Neuroscience, 21(6), 322-334. https://doi.org/10.1038/s41583-020-0304-4 696 Poeppel, D., & Hickok, G. (2004). Towards a new functional anatomy of language. Cognition, 92(1-697 2), 1-12. https://doi.org/10.1016/j.cognition.2003.11.001 698 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Price, C. N., Alain, C., & Bidelman, G. M. (2019). Auditory-frontal channeling in α and β bands is 699 altered by age-related hearing loss and relates to speech perception in noise. 700 Neuroscience, 423, 18-28. https://doi.org/10.1016/j.neuroscience.2019.10.044 701 Price, C. N., & Bidelman, G. M. (2021). Attention reinforces human corticofugal system to aid 702 speech perception in noise. Neuroimage, 235, 118014. 703 https://doi.org/10.1016/j.neuroimage.2021.118014 704 Puschmann, S., Baillet, S., & Zatorre, R. J. (2018). Musicians at the Cocktail Party: Neural Substrates 705 of Musical Training During Selective Listening in Multispeaker Situations. Cerebral Cortex, 706 29(8), 3253-3265. https://doi.org/10.1093/cercor/bhy193 707 Puschmann, S., Regev, M., Baillet, S., & Zatorre, R. J. (2021). MEG Intersubject Phase Locking of 708 Stimulus-Driven Activity during Naturalistic Speech Listening Correlates with Musical 709 Training. The Journal of Neuroscience, 41(12), 2713-2722. 710 https://doi.org/10.1523/jneurosci.0932-20.2020 711 R-Core-Team. (2020). R: A language and environment for statistical computing. In R Foundation for 712 Statistical Computing, Vienna, Austria. https://www.R-project.org/. 713 Riecke, L., Formisano, E., Sorger, B., Başkent, D., & Gaudrain, E. (2018). Neural entrainment to 714 speech modulates speech intelligibility. Current Biology, 28(2), 161-169. e165. 715 https://doi.org/10.1016/j.cub.2017.11.033 716 Román-Caballero, R., Martín-Arévalo, E., & Lupiáñez, J. (2020). Attentional networks functioning 717 and vigilance in expert musicians and non-musicians. Psychological Research. 718 https://doi.org/10.1007/s00426-020-01323-2 719 Samaha, J., Iemi, L., Haegens, S., & Busch, N. A. (2020). Spontaneous Brain Oscillations and 720 Perceptual Decision-Making. Trends in Cognitive Sciences, 24(8), 639-653. 721 https://doi.org/10.1016/j.tics.2020.05.004 722 Stevens, C. E., & Zabelina, D. L. (2019). Creativity comes in waves: an EEG-focused exploration of 723 the creative brain. Current Opinion in Behavioral Sciences, 27, 154-162. 724 https://doi.org/10.1016/j.cobeha.2019.02.003 725 Strait, D. L., & Kraus, N. (2011). Can you hear me now? Musical training shapes functional brain 726 networks for selective auditory attention and hearing speech in noise. Frontiers in 727 Psychology, 2, 113. https://doi.org/10.3389/fpsyg.2011.00113 728 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Strait, D. L., Kraus, N., Parbery-Clark, A., & Ashley, R. (2010). Musical experience shapes top-down 729 auditory mechanisms: Evidence from masking and auditory attention performance. Hearing 730 Research, 261(1-2), 22-29. https://doi.org/10.1016/j.heares.2009.12.021 731 Strauß, A., Henry, M. J., Scharinger, M., & Obleser, J. (2015). Alpha Phase Determines Successful 732 Lexical Decision in Noise. The Journal of Neuroscience, 35(7), 3256-3262. 733 https://doi.org/10.1523/jneurosci.3357-14.2015 734 Tian, X., & Poeppel, D. (2012). Mental imagery of speech: Linking motor and perceptual systems 735 through internal simulation and estimation [Review]. Frontiers in Human Neuroscience, 736 6(314). https://doi.org/10.3389/fnhum.2012.00314 737 Vaden, K. I., Kuchinsky, S. E., Ahlstrom, J. B., Dubno, J. R., & Eckert, M. A. (2015). Cortical Activity 738 Predicts Which Older Adults Recognize Speech in Noise and When. The Journal of 739 Neuroscience, 35(9), 3929-3937. https://doi.org/10.1523/jneurosci.2908-14.2015 740 Vaden, K. I., Teubner-Rhodes, S., Ahlstrom, J. B., Dubno, J. R., & Eckert, M. A. (2022). Evidence for 741 cortical adjustments to perceptual decision criteria during word recognition in noise. 742 Neuroimage, 253, 119042. 743 https://doi.org/https://doi.org/10.1016/j.neuroimage.2022.119042 744 Vaden, R. J., Hutcheson, N. L., McCollum, L. A., Kentros, J., & Visscher, K. M. (2012). Older adults, 745 unlike younger adults, do not modulate alpha power to suppress irrelevant information. 746 Neuroimage, 63(3), 1127-1133. https://doi.org/10.1016/j.neuroimage.2012.07.050 747 Vanthornhout, J., Decruy, L., Wouters, J., Simon, J. Z., & Francart, T. (2018). Speech intelligibility 748 predicted from neural entrainment of the speech envelope. Journal of the Association for 749 Research in Otolaryngology, 19, 181-191. https://doi.org/10.1007/s10162-018-0654-z 750 Wallstrom, G. L., Kass, R. E., Miller, A., Cohn, J. F ., & Fox, N. A. (2004). Automatic correction of 751 ocular artifacts in the EEG: A comparison of regression-based and component-based 752 methods. International Journal of Psychophysiology, 53, 105-119. 753 https://doi.org/10.1016/j.ijpsycho.2004.03.007 754 Weisz, N., Hartmann, T., Müller, N., & Obleser, J. (2011). Alpha Rhythms in Audition: Cognitive and 755 Clinical Perspectives [Review]. Frontiers in Psychology, 2. 756 https://doi.org/10.3389/fpsyg.2011.00073 757 Wilsch, A., Neuling, T., Obleser, J., & Herrmann, C. S. (2018). Transcranial alternating current 758 stimulation with speech envelopes modulates speech comprehension. Neuroimage, 172, 759 766-774. https://doi.org/10.1016/j.neuroimage.2018.01.038 760 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint Wong, P . C., Skoe, E., Russo, N. M., Dees, T., & Kraus, N. (2007). Musical experience shapes human 761 brainstem encoding of linguistic pitch patterns. Nature Neuroscience, 10(4), 420-422. 762 https://doi.org/10.1038/nn1872 763 Wöstmann, M., Herrmann, B., Wilsch, A., & Obleser, J. (2015). Neural alpha dynamics in younger 764 and older listeners reflect acoustic challenges and predictive benefits. The Journal of 765 neuroscience : the official journal of the Society for Neuroscience, 35(4), 1458-1467. 766 https://doi.org/10.1523/jneurosci.3250-14.2015 767 Wu, Z.-M., Chen, M.-L., Wu, X.-H., & Li, L. (2014). Interaction between auditory and motor systems 768 in speech perception. Neuroscience Bulletin, 30(3), 490-496. 769 https://doi.org/10.1007/s12264-013-1428-6 770 Yellamsetty, A., & Bidelman, G. M. (2018). Low- and high-frequency cortical brain oscillations 771 reflect dissociable mechanisms of concurrent speech segregation in noise. Hearing 772 Research, 361, 92-102. https://doi.org/10.1016/j.heares.2018.01.006 773 Zatorre, R. J., Belin, P ., & Penhune, V . B. (2002). Structure and function of auditory cortex: Music and 774 speech. Trends in Cognitive Sciences, 6(1), 37-46. https://doi.org/S1364661300018167 [pii] 775 Zatorre, R. J., Evans, A. C., Meyer, E., & Gjedde, A. (1992). Lateralization of phonetic and pitch 776 discrimination in speech processing. Science (New York, N. Y .), 256(5058), 846-849. 777 https://doi.org/10.1126/science.256.5058.84 778 Zendel, B. R., & Alain, C. (2009). Concurrent sound segregation is enhanced in musicians. Journal of 779 Cognitive Neuroscience, 21(8), 1488-1498. https://doi.org/10.1162/jocn.2009.21140 780 Zendel, B. R., & Alain, C. (2012). Musicians experience less age-related decline in central auditory 781 processing. Psychology and Aging, 27(2), 410-417. https://doi.org/10.1037/a0024816 782 Zendel, B. R., West, G. L., Belleville, S., & Peretz, I. (2019). Musical training improves the ability to 783 understand speech-in-noise in older adults. Neurobiology of Aging, 81, 102-115. 784 https://doi.org/10.1016/j.neurobiolaging.2019.05.015 785 Zhang, L., Ross, B., Du, Y ., & Alain, C. (2024). Accumulated reserves hold back age-related neural 786 compensation in speech-in-noise perception. bioRxiv, 2024.2005.2027.596069. 787 https://doi.org/10.1101/2024.05.27.596069 788 Zuk, J., Benjamin, C., Kenyon, A., & Gaab, N. (2014). Behavioral and neural correlates of executive 789 functioning in musicians and non-musicians. PloS One, 9(6), e99868. 790 https://doi.org/10.1371/journal.pone.0099868 791 792 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (whichthis version posted July 19, 2024. ; https://doi.org/10.1101/2024.07.18.604167doi: bioRxiv preprint