In the initial phase, we recruited 82 healthy young adults with either minimal or extensive piano experience. Participants were recruited from various advertisement sources (social media, flyers, etc.), while some expert musicians were specifically recruited through a snowball sampling method. As part of a broader study not detailed here, participants completed a comprehensive battery of tasks and were compensated with cash for their overall time. All Participants had completed at least 1 year of university-level education, and their demographic information is provided in Table 1. To conduct group comparisons on the effects of musical training, we defined subjects with >10 cumulative years of music training and > 4,000 h of lifetime practice as Musicians (N = 42), and subjects with <2 years of musical activity as Non-Musicians (N = 20) (Table 1).

Subjects provided informed consent and were compensated for their participation and time. All experimental procedures were approved by the McGill University Faculty of Medicine Research Ethics Board. All participants were screened to have normal or corrected-to-normal vision and reported no history of neurological disorders. Normal binaural hearing was confirmed by an audiometric test which measured pure-tone thresholds from 250 to 8,000 Hz (less than 25 dB SL). Participants with binaural hearing thresholds above 25 dB HL did not proceed with the study as deficiencies in the frequency range may influence their task performance. Out of the 82 participants from Experiment 1 who completed all parts of the study, 4 were excluded from the MINT analysis due to their inability to process basic musical content (with 2 or more out of 6 incorrect responses for the MINT task Control condition, see description below).

Prior to the testing session, participants confirmed eligibility and completed the Montreal Music History Questionnaire (MMHQ) (Coffey et al., 2011). The MMHQ provides the subject’s self-reported information regarding overall musical experience (instruments played, total cumulative practice hours), language proficiency, basic demographics, etc. The tasks were administered in the context of a larger test battery that will not be reported here. Each testing session began with an audiometry hearing test, followed by a series of behavioral tasks, including the AWM task (Albouy et al., 2017) and the MINT task (Coffey et al., 2019); see the following section for descriptions. The visual component of each task was presented on a computer screen and sounds were presented binaurally through headphones (ATH-M50x, Audio-Technica). A comfortable sound level set at 73 dB was determined during pilot testing and kept constant for all subjects and both tasks.

Data analyses were conducted using IBM SPSS Statistics (version 29.0.2.0) to perform correlation and mediation analyses. To examine the correlational relationships between cumulative practice hours, AWM, and MINT scores, both parametric (Pearson’s r) and non-parametric (Spearman’s rho) tests were conducted. For consistency with the mediation analysis, which uses raw values rather than ranks, only Pearson’s correlation coefficients are reported. Nonetheless, all tests produced comparable significant results (see Supplementary material for non-parametric correlations). Comparisons between Experiment 1 and Experiment 2 were performed using independent samples t-tests, while comparisons between Musicians and Non-Musicians in each experiment were conducted using the Mann–Whitney U test, concerning the non-normal distribution and small sample size.

PROCESS macro (version 4.2 beta release) for SPSS by A. F. Hayes was used for mediation analysis. PROCESS is an observed variable ordinary least square and logistic regression path analysis tool that provides estimation of direct and indirect effects within both single and multiple mediator models (Preacher and Hayes, 2004). It allows us to estimate the conditional indirect effects of AWM as a mediator between music training and MINT outcomes. All mediation models were tested for statistical significance through bootstrapping with 5,000 resamples, conducted with replacement, with significance determined by confidence intervals (Preacher and Hayes, 2008). Bootstrapping is a common procedure in mediation analysis that enhances statistical power and robustness against non-normal distributions, small sample sizes, and outliers.

Descriptive statistics for AWM score (% correct) for all participants, as well as for the Musician and Non-Musician groups, are presented in Table 2. Pearson correlation indicates a significant relationship between cumulative hours of practice and AWM task performance (r = 0.399, p < 0.001; Figure 3A; see Supplementary Table S1 for Spearman’s results). Mann–Whitney U test indicates a significant difference between Musicians and Non-Musicians groups on AWM score (U = 792, p < 0.001) (Figure 3C).

Descriptive statistics for overall MINT performance are presented in Table 2. The mean accuracy scores for each MINT sub-condition were: Baseline (Pitch) = 80.94 (SD = 11.95), Rhythm = 63.85 (SD = 14.56), Spatial = 84.02 (SD = 10.49), Visual = 90.60 (SD = 10.27), and Prediction = 90.85 (SD = 8.80). The mean accuracy scores for each SNR level were: SNR 0 = 84.77 (SD = 10.74), SNR −3 = 83.79 (SD = 9.32), and SNR −6 = 77.59 (SD = 9.93). Pearson correlation analysis between cumulative practice hours and overall MINT task performance revealed a significant correlation, with a r-value of 0.363 (p < 0.001; Figure 3B; see Supplementary Table S1 for Spearman’s results). Cumulative hours of practice were also correlated with the Baseline (Pitch) (r = 0.22, p = 0.025), Prediction (r = 0.26, p = 0.010), Rhythm (r = 0.28, p = 0.007), and Visual (r = 0.29, p = 0.005) sub-conditions. In addition, cumulative hours of practice correlated with all SNR levels: SNR 0 (r = 0.24, p = 0.019), SNR −3 (r = 0.31, p = 0.003), and SNR −6 (r = 0.29, p = 0.006). Mann–Whitney U test shows a significant difference in MINT performance between Musicians and Non-Musicians (U = 696, p < 0.001) (Figure 3C).

Pearson correlation analysis evaluated the relationship between performance on the AWM and MINT tasks. The AWM scores significantly correlated with the overall MINT scores (r = 0.584, p < 0.001) (Figure 3D). The AWM correlated with all MINT sub-conditions, as listed in Table 3A (see also Supplementary Table S2A). Moreover, AWM was correlated with all the SNR levels, as presented in Table 3B (see also Supplementary Table S2B). Fisher’s test performed to compare the differences between the z-transformations of each pair of correlations demonstrated that none of the correlations were significantly larger than the others.

Regression analyses with bootstrapping were performed to assess each component of the proposed mediation model. First, it was found that cumulative music training hours were positively associated with both MINT performance [R = 0.36, F(1, 76) = 11.56, p = 0.001] and AWM performance [R = 0.40, F(1, 76) = 14.42, p < 0.001]. It was also found that the mediator, AWM ability, was positively related to the MINT test score [R = 0.58, F(1, 76) = 39.42, p < 0.001]. Lastly, multiple regression analysis was conducted to examine the effects of hours of musical training (X₁) and AWM (X₂) on MINT performance (Y). Results indicated that the overall regression model was significant [R = 0.601, F(2, 75) = 21.24, p < 0.001], with VIF = 1.19, MSE = 37.88, and η² = 0.362 (Figure 4). Both predictors contributed to better MINT performance (β₁ = 0.16, p = 0.129; β₂ = 0.52, p < 0.001).

Because the general model, the a-path (music training to AWM), and the b-path (AWM to MINT) were significant, mediation analysis was tested using the bootstrapping method with bias-corrected confidence estimates (refer to Methods and Materials section Data Analysis; Preacher and Hayes, 2004). The 95% confidence interval of the indirect effect was obtained with 5,000 bootstrap samples and confirmed the significant mediating role of AWM in the relationship between music training and MINT task performance (Figure 4). Regression results also indicated that the direct effect of music training on MINT becomes non-significant (p = 0.11) when controlling for AWM, thus suggesting full mediation. Moreover, confidence intervals derived from bootstrapping mediation analysis revealed mediation effects of AWM on the Baseline (Pitch), Prediction, Rhythm, and Visual sub-conditions. Our results also indicate significant mediating effects of AWM on MINT performance at the SNR 0 and SNR −6 levels.

In the replication phase, we recruited 73 subjects with a distributed range of music experience and expertise (Table 4). Recruitment methods and compensation were consistent with those used in Experiment 1. All participants had completed at least 1 year of university-level education. On average, subjects from Experiment 2 have fewer practice hours than those in Experiment 1; t(143) = −1.84, p = 0.034. Within the 73 subjects, 19 were categorized as Musicians according to the same criteria as above, and 18 were Non-Musicians (Table 4).

All procedures and screening criteria remained consistent with those in Experiment 1 and were approved by either the McGill University Faculty of Medicine Research Ethics Board or Western University Non-Medical Research Ethics Board. Out of the 73 subjects who completed all components of Experiment 2, 3 who could not process basic musical content were excluded from the MINT analysis.

Refer to Experiment 1 Materials and methods section Procedure.

Refer to Experiment 1 Materials and methods section Measures and Behavioral Tasks.

The mean accuracy score (% correct) for the AWM task in the second sample was 66.33 (SD = 15.85, range: 41–100, N = 70). Results from a one-tailed Pearson correlation test indicated a trend toward significance in the association between musical training and AWM task performance (r = 0.191, p = 0.057). Potential outlier effects were suspected through examination of the data distribution, prompting the use of Spearman’s rank-order correlation, which is more robust to extreme values. The Spearman’s test revealed a significant monotonic relationship between AWM scores and cumulative hours of practice (ρ = 0.324, p = 0.003). The discrepancy between the rank-order and parametric test results suggests that the data may have been affected by extreme values. Upon comprehensive examination of the total 148 qualified subjects from Experiment 1 and 2 using linear regression (practice hours versus AWM performance), we identified two subjects from Experiment 2 with performance significantly deviating from the model’s predictions. Specifically, one subject had a standardized residual of −2.45 and the other −2.40, while the standardized residuals for the remaining 146 subjects ranged between −1.67 and 1.70. Consequently, these two subjects are considered outliers and were excluded from subsequent analysis.

By removing the two outliers, the adjusted mean AWM accuracy score in the second sample is presented in Table 5. Independent samples t-test indicates a significantly lower AWM performance for the subjects in Experiment 2 compared to Experiment 1; t(144) = −4.17, p < 0.001. A significant relationship between AWM score and cumulative hours of practice is demonstrated with Pearson’s test (r = 0.370, p < 0.001) (Figure 5A; see Supplementary Table S3 for Spearman’s results). In addition, the Mann–Whitney U test also indicates a group difference in AWM between Musicians and Non-Musicians (U = 302, p < 0.001; Table 5; Figure 5C).

Descriptive statistics for overall MINT performance are presented in Table 5. Independent samples t-test between Experiment 1 and 2 indicates a significantly lower MINT score for the subject in Experiment 2; t(144) = −5.36, p < 0.001. The mean accuracy scores for each MINT condition were: Baseline (Pitch) = 72.45 (SD = 16.25), Rhythm = 59.51 (SD = 12.95), Spatial = 71.76 (SD = 13.97), Visual = 83.43 (SD = 15.77), and Prediction = 81.47 (SD = 15.19). The mean accuracy scores for each SNR level were: SNR -3 = 79.29 (SD = 14.38), SNR −6 = 74.00 (SD = 12.51), and SNR −9 = 67.88 (SD = 13.04).

A correlation between cumulative practice hours and MINT task performance was tested with r = 0.293 (p = 0.008; Figure 5B; see Supplementary Table S3 for Spearman’s results). Cumulative hours of practice were also correlated with the Baseline (Pitch) (r = 0.21, p = 0.040), Prediction (r = 0.29, p = 0.009), and Visual (r = 0.29, p < 0.009) sub-conditions. In addition, cumulative hours of practice correlated with SNR −3 (r = 0.34, p = 0.003) and SNR −9 (r = 0.21, p = 0.043). Mann–Whitney U test shows a significant difference in total MINT performance between Musicians and Non-Musicians (U = 281, p < 0.001; Table 5; Figure 5C).

Pearson correlation analysis evaluated the correlation between performance on the AWM and MINT tasks. AWM score significantly correlated with the overall MINT score (r = 0.573, p < 0.001; Figure 5D). The correlations between AWM also significantly correlated with all MINT sub-conditions and SNR levels, as presented in Table 6 (see also Supplementary Table S4). Fisher’s test performed to compare the differences between the z-transformations of each pair of correlations demonstrated that none of the correlations are significantly larger than the others.

Regression analyses were conducted to assess each component of the mediation model proposed in Experiment 1. Linear regression with bootstrapping revealed a positive association between cumulative music training hours and both AWM performance [R = 0.370, F(1, 66) = 10.48, p = 0.004] and MINT performance [R = 0.293, F(1, 66) = 6.18, p = 0.015]. AWM ability as the proposed mediator was also positively related to MINT test scores [R = 0.573, F(1, 66) = 32.24, p < 0.001]. Subsequent multiple regression analysis was performed to assess the effects of musical training hours (X₁) and AWM (X₂) on MINT performance (Y). The overall regression model was significant [R = 0.579, F(2, 65) = 16.42, p < 0.001], with VIF = 1.16, MSE = 83.62, η² = 0.336, and predictors contributing to improved MINT performance (β₁ = 0.093, p = 0.394; β₂ = 0.538, p < 0.001).

Given that the multiple regression model and the paths were significant and consistent with Experiment 1, mediation analysis was conducted using the same bootstrapping method (refer to Methods and Materials section Data Analysis). A 95% confidence interval for the indirect effect was derived from bootstrap samples and demonstrated a significant mediating role of AWM in the relationship between music training and MINT task performance. Results also show that the direct effect of music training on MINT became non-significant (p = 0.394) when controlling for AWM. Additional bootstrapping analysis also revealed a mediating effect of AWM on the Baseline (Pitch), Prediction, Spatial and Visual sub-conditions, and across all SNR levels.

The findings from both Experiments 1 and 2 provide compelling evidence that supports our hypothesis of a musician’s advantage in both AWM abilities and music-in-noise perception. Importantly, the musician advantage was consistently observed across two distinct samples, which differed in overall musical experience, proportion of musicianship, and average performance on both tasks.

A meta-analysis by Talamini et al. (2017) demonstrated that musicians outperform non-musicians across various memory domains, including long-term, short-term, and working memory, with a particularly pronounced advantage for tonal stimuli. To investigate this tonal aspect of AWM in which musicians excel, the task in this study required participants to detect a local pitch change between two tonal patterns that differed in temporal order. This AWM task not only captured auditory retention capabilities but also assessed the ability to mentally manipulate the stimuli (i.e., serial order processing; Albouy et al., 2017; Foster et al., 2013), along with related cognitive skills such as decision-making, attention, processing speed, etc. Correlational analyses between cumulative practice hours and AWM task performance from Experiment 1 indicated a positive association between music experience and AWM abilities (Figure 3A), a finding that was replicated in Experiment 2 (Figure 5A). Moreover, the group comparison underscores this advantage, as musicians from both studies consistently outperformed their non-musician counterparts on the standardized measures of AWM (Figures 3C, 5C). These results are supported by existing literature, which consistently demonstrates behavioral, electrophysiological (event-related potential), and neuro-oscillatory evidence for the superiority of musicians in AWM abilities (Albouy et al., 2017; Foster et al., 2013; George and Coch, 2011).

Analyses of the overall MINT task performance in Experiment 1 in relation to cumulative practice hours suggests a clear association between musical experience and improved music-in-noise perception (Figure 3B). Although subjects in Experiment 2 showed an overall poorer performance on the MINT—potentially due to differences in musicianship and/or increased task SNR difficulty—the correlation between musical experience and MINT performance remained consistent and significant (Figure 5B). In other words, the relationship between musical experience and music-in-noise perception is stable across different signal-to-noise levels tested. Furthermore, significant musician advantage on music-in-noise perception was also observed in both studies when comparing the musician and non-musician group differences in the overall MINT performance (Figures 3C, 5C). These results are in line with the findings from Coffey et al. (2019) original MINT study and the subsequent MINT results by Hsieh et al. (2022), further validating the MINT’s reliability and supporting the cognitive benefits of musical expertise amid varying levels of noise interference.

Musicians’ music-in-noise benefits may arise from improvements in both auditory perception and cognitive processing. On the perceptual side, musicians demonstrate an increased sensitivity to fundamental acoustic features critical for music perception, such as pitch discrimination and temporal fine structure (Micheyl et al., 2006; Mishra et al., 2015). Cognitively, studies have shown a connection between musicianship and enhancements in cognitive faculties including working memory and attention (Bidelman and Yoo, 2020; Yoo and Bidelman, 2019), which may be linked to stream segregation improvements.

Evidence suggesting that AWM plays a crucial role in music-in-noise perception stems from the strong correlation between performance on the AWM task and the overall MINT score, observed in both Experiment 1 (Figure 3D) and Experiment 2 (Figure 5D). This finding replicates the original MINT study (Coffey et al., 2019), and is consistent with the majority of the SIN literature which suggests that working memory for phonological or tonal information is linked to improved speech segregation abilities (Bidelman and Yoo, 2020; Escobar et al., 2020; Lad et al., 2020; Mattys et al., 2012; Yoo and Bidelman, 2019).

The mediation analysis conducted in Experiment 1 supports our hypothesis that AWM ability significantly mediates the relationship between musical experience and music-in-noise perception (Figure 4). This mediation model was successfully replicated in Experiment 2, which included a more heterogeneous musician population. The comparable results from Experiment 2 reinforce the reliability and generalizability of our initial findings. Overall, our results suggest that musicians’ enhanced AWM skills are a crucial driving force behind their enhanced MINT performance, and that musical training is associated with improvements in the performance of auditory stream segregation tasks through the enhancement of AWM capabilities. This mediating effect of AWM in music-in-noise performance parallels the mediation model proposed for AWM’s role in SIN performance (Kraus et al., 2012; Parbery-Clark et al., 2009b). Parbery-Clark et al. (2009b) demonstrated that musicians possess superior AWM skills, which those authors identify as a significant factor behind the group’s improved SIN performance. In addition, Bidelman and Yoo (2020) found that the relationship between musicianship and performance on a complex SIN task did not remain significant after controlling for working memory, which is associated with the listener’s year of musical training. This finding supports the concept that auditory stream segregation superiority is driven heavily by the enhanced working memory capacity, likely developed through musical training, although aptitude may also play role.

Although evidence supports the importance of AWM in overall stream segregation, the precise mechanisms underlying its contribution remain unclear. The predominant literature on SIN has focused on the role of AWM in facilitating the understanding of linguistic context (Kraus et al., 2012). For example, the Ease of Language Understanding (ELU) model by Rönnberg et al. (2013) posits that working memory enables the listener to hold a schematic representation of speech while processing contextual information, using linguistic knowledge to compensate for missing information in adverse listening environments. In addition, the ELU model states that individuals with enhanced working memory capacity can apply more mental resources to resolve the phonological and semantic aspects of a listening task (Rönnberg et al., 2013). It follows that the advantage offered by AWM in aiding SIN processing may depend largely on the redundancy of linguistic contextual cues (e.g., phonological, lexical, syntactic, and semantic information) of the speech signal tested (Gordon-Salant and Cole, 2016). However, given the consistent relationship between AWM and MINT performance—which is not influenced by linguistic factors—our study provides evidence that the benefits of AWM in stream segregation extend beyond the speech domain, pointing to more fundamental mechanisms that are more generally involved in stream segregation processing.

One advantage of the MINT over standard SIN tests is its ability to assess specific cues and auditory sub-skills related to stream segregation (Coffey et al., 2019), offering insights into how AWM may interact with the perceptual and cognitive elements involved in this process. The original MINT study indicated that AWM has the most significant contribution to the Prediction condition, and the relationship between musical training and the Prediction task diminished in significance when AWM performance was factored in as a covariate in the analysis (Coffey et al., 2019). Prior research also supports the role of AWM in musical predictive processing, highlighting its importance in top-down schematic expectations—the concept that knowing the pattern to be segregated a priori facilitates subsequent detection (Bey and McAdams, 2002).

However, contrary to earlier findings, here we did not observe a more important contribution of AWM to the MINT Prediction condition compared to other conditions. Instead, there were significant and consistent correlations between AWM and all MINT sub-tasks in both Experiments (Table 3A; Table 6A). This finding suggests that AWM’s contribution is only one among many factors that modulate stream segregation situations.

One possible explanation for the contribution of AWM to general stream segregation is that enhanced AWM allows a more precise representation of acoustic signals in the mental workspace (Kraus et al., 2012). Research suggests that working memory is linked to improved performance on a rhythm synchronization task, where participants are required to reproduce the temporal structure of the presented rhythms (Bailey and Penhune, 2010). It is also indicated that individuals who can effectively retain auditory source properties, such as frequency and temporal fluctuations over time have a perceptual advantage in SIN tasks (Lad et al., 2020; Lad et al., 2024). It is therefore plausible that the ability to maintain acoustic information accurately aids the sequential segregation processes essential for stream intelligibility (Bregman, 1990).

Another perspective involves attention. Dalton et al. (2009) manipulated the working memory load during a distractor interference task, demonstrating a causal role for the availability of working memory in auditory selective attention. In addition, it is suggested that segregating auditory streams from background noise draws upon attentional resources (Heinrich et al., 2008), and accomplishing such tasks necessitates the allocation of one’s limited cognitive resources to balance the competing demands of attention, processing, and storage (Wingfield and Tun, 2007). It is therefore plausible that enhanced AWM proficiency promotes the maintenance and encoding of auditory signals, which in turn allows for more efficient use of attention resources to extract and recall the target stream.

In addition, the advantages of AWM can be understood through the temporal aspects of information processing: temporal integration and serial order processing. On the one hand, it is proposed that working memory aids the linkage between recent past and imminent future events, thus serving both a retrospective role in information retention and a prospective role in anticipation (Fuster and Bressler, 2012). Specifically, prior literature proposes that working memory is important for minimizing distractor interference through the active maintenance of current stimulus-processing priorities (Dalton et al., 2009; Lavie, 2005). In stream segregation, AWM may therefore enable individuals to hold fragments of auditory information while processing, integrating, and anticipating degraded target signals.

On the other hand, the AWM task used in this study, which requires temporal reversal, captures item-based retention and serial order processing, which have been shown to be distinct processes. Serial ordering, in particular, is thought to be a domain-general process based on positional codes, as observed in verbal and musical working memory studies (Hurlstone et al., 2014; Majerus, 2019; Gorin, 2022). Since melodic retention and prediction did not appear to play a special role in stream segregation, serial ordering may serve as an alternative key factor, contributing to tracking the sequence of items over time and thereby enhancing the ability to organize auditory streams. It will be of interest in future research to study the contribution of AWM when measured with tasks that do not require serial order processing, such as for example musical transposition.

The dorsal stream of auditory processing, which involves the parietal lobe, dorsal premotor cortex, and dorsolateral frontal regions, is central to higher-order cognitive auditory functions. It supports the manipulation of sound patterns in working memory, auditory-motor integration, abstract temporal representations, and predictive coding (Rauschecker and Scott, 2009; Zatorre, 2024). Neuroimaging studies highlight the dorsal stream’s key role in AWM, with activations in parietal regions associated with various kinds of mental transformation and manipulation processes (Foster et al., 2013; Zatorre et al., 2010). Moreover, Albouy et al. (2017) have observed that sustained evoked activity in the bilateral dorsal streams, particularly through long-range theta phase locking and increased local theta power in the IPS, is associated with successful AWM manipulation. Furthermore, when theta power is boosted in the dorsal stream via rhythmic brain stimulation (Albouy et al., 2017) or via flickering visual rotating stimuli (Albouy et al., 2022), AWM performance is also enhanced.

While perceiving auditory signals in background noise heavily engages primary and non-primary auditory regions (Holmes et al., 2021; Kell and McDermott, 2019; Puschmann et al., 2019), research indicates that motor and somatosensory areas are also more actively recruited under challenging listening conditions (for review, see Skipper et al., 2017). This suggests a compensatory mechanism of dorsal steam activity for reduced processing specificity in the auditory system (Du et al., 2014). Importantly, a study comparing musicians and non-musicians found that the benefits of musical training on SIN perception in difficult listening contexts were related to activity in the motor cortices of the auditory dorsal streams (Du and Zatorre, 2017).

Further research has shown that music training enhances functional connectivity within the dorsal auditory stream (Jünemann et al., 2023). Musicians also exhibit greater structural connectivity in the white matter tracts of the dorsal stream (i.e., arcuate fasciculus and superior longitudinal fasciculus; Halwani et al., 2011; Oechslin et al., 2010). Differences in the microstructural plasticity of dorsal white matter are suggested to underlie musicians’ improved SIN perception (Li et al., 2021). Considering the role of the auditory dorsal stream in AWM and SIN perception, we thus infer that the musician enhancements in these abilities may be rooted in this stream, although the exact mechanisms warrant further exploration.

Auditory functioning is one of the most prevalently affected sensory modalities in the elderly population (Yamasoba et al., 2013; Davis et al., 2016). In addition, older adults show deficits in speech recognition in noisy environments and AWM (Dubno et al., 1984; Humes and Floyd, 2005). Previous studies demonstrated that older musicians exhibit enhanced performance in AWM and SIN perception compared to their non-musician counterparts, suggesting that musical experience may mitigate age-related hearing challenges (Zhang et al., 2021; Zendel et al., 2019).

Recent longitudinal studies assigning older adults to musical activities (piano/choir) have also demonstrated behavioral, neurophysiological, and neuro-oscillatory evidence of improvements in SIN perception (Worschech et al., 2021; Hennessy et al., 2021; Dubinsky et al., 2019; Gray et al., 2022). Shedding light onto the mediating role of AWM in stream segregation, we propose that future music programs designed to address hearing challenges in older adults should focus on enhancing AWM to achieve optimal intervention outcomes.

Limitations of the current study include reliance on self-report music history questionnaire responses and the challenge of precisely controlling for the nuanced variations of individual musical experiences and expertise (e.g., learning styles, extent of practice). Moreover, the correlational design of the study does not address issues regarding self-selection and the direction of causality, particularly considering evidence suggesting that auditory and musical expertise arises from a combination of genetic predispositions and experience-driven plasticity (Schellenberg, 2015; Zatorre, 2013). The inherent predispositions for AWM or stream segregation abilities could potentially influence one’s path toward musical engagement, an aspect that warrants further investigation.

Longitudinal studies with school-aged children (as well as the elderly, as described in the preceding section) provide evidence that music instruction is in fact causally associated with moderate benefits in SIN and AWM abilities (Slater et al., 2015; Nie et al., 2022), but of course, this does not mean that predisposing factors do not exist. Moreover, research has demonstrated a relationship between music and language performance in elementary school children (Zuk et al., 2013), which is primarily driven by temporal processing (Andrade et al., 2024). These findings suggest some shared contributions, yet the extent of transfer effects from music to speech and phonological processing remains unclear. Therefore, future research directions entail conducting longitudinal studies to examine the development of both speech-in-noise and music-in-noise perception, further unraveling the relationship between musical training, AWM, and overall auditory stream segregation. Such endeavors will also help elucidate experience-dependent plasticity in the auditory domain and contribute to a deeper understanding of the development of higher-level auditory cognitive mechanisms.