Forty-eight middle-aged adults (45–65 years, mean age 56 ± 5 years) participated. All subjects had normal hearing for octave frequencies from 0.125–4 kHz bilaterally ≤20 dB HL, pure-tone average ≤10 dB HL. Participants had no history of neurological or learning disorders, did not have asymmetric pure-tone thresholds (defined as ≥15 dB difference at two or more frequencies between ears) and demonstrated normal click-evoked ABRs (wave V latency ≤6.8 ms at 80 dB SPL). No participant reported a history of chemotherapy, taking ototoxic medications, major surgeries, or head trauma. In addition, all participants were native English speakers and had normal non-verbal IQ: Abbreviated Wechsler's Adult Scale of Intelligence's matrix reasoning subtest, (Wechsler, 1999). All experimental procedures were approved by the Northwestern University Institutional Review Board and participants provided informed consent.

Twenty-three subjects were categorized as musicians, having started musical training before the age of nine and consistently engaged in musical activities a minimum of three times a week throughout their lifetimes. For information relating to participants' music practice histories, see Table 1. Twenty-five subjects were categorized as nonmusicians with 17 having had no musical training and eight having fewer than 4 years of musical experience. The groups did not differ in age, hearing thresholds, sex, or non-verbal IQ (all P > 0.1; Table 2). Participants were also matched on measures of physical activity [F_{(1, 47)} = 0.032, p = 0.858], assessed by asking participants to describe the type and quantity of physical activity they engaged in each week. To account for varying types of physical activity, “walking” and “biking” were given half values while “running,” “weight training,” and other more vigorous activities were given full values. From these values, the total hours of physical activity per week was calculated for each participant. Participants were then assigned a value based on their overall activity level: 0 (<1 h/week), 1 (1–2 h/week), 2 (2–3 h/week), 3 (3–4 h/week), or 4 (>4 h/week); Table 2. Two musicians and three nonmusicians were left-handed. In terms of alcohol consumption, 4 musicians and 6 nonmusicians reported never drinking.

All statistical analyses were conducted in SPSS Version 18.0 (SPSS Inc., Chicago, IL). Repeated measure analyses of variance (RMANOVA) were used for group (musician vs. nonmusician) × condition (quiet vs. noise) comparisons for latency, spectral representation, stimulus-to-response correlations, envelope encoding, and response consistency. Univariate analyses of variance were used for behavioral measures. Post-hoc tests were used when appropriate. To assess relationships among variables, Pearson r correlations were used. Levene's test was used to ensure homogeneity of variance for all measures and the Shapiro-Wilk test was used to ensure that all variables were normally distributed. Bonferroni corrections for multiple comparisons were applied as appropriate; p-values reflect two-tailed tests. The SSQ (self-reported hearing in noise ability) was the only test that violated the assumption of normality. Neither log nor reciprocal transforms rendered these data normal. As such, we only used these data to quantify group differences using the non-parametric Mann-Whitney test; correlations with other variables were not explored.

Musicians demonstrated greater speech-in-noise perception [HINT: F_{(1, 47)} = 20.276, p < 0.005; musicians mean: −3.16, SD 0.61; nonmusicians mean: −2.34, SD 0.63] and rated themselves as having less difficulty hearing in noise than nonmusicians as assessed by the SSQ (Table 3). Musicians exhibited more robust neural encoding of speech in both quiet and noise. Musicians had earlier neural response timing, greater neural representation of the stimulus harmonics as well as more precise phase-locking to the stimulus both in terms of temporal envelope and stimulus-to-response correlations. Musicians also demonstrated less neural response degradation in noise evidenced by smaller neural timing shifts and smaller decreases in neural response consistency. We also found that specific neural measures such as earlier neural response timing and more robust brainstem responses to speech correlated with better speech-in-noise performance as measured by HINT.

Musicians demonstrated enhanced onset and transition timing in quiet and limited degradative effects of background noise for all aspects of neural timing. To quantify effects of musicianship and noise on neural response timing, we divided the neural response into three time regions: onset, transition, and vowel. We performed a mixed-model repeated-measures ANOVA (RMANOVA) 2 group (musician/nonmusician) × 2 condition (quiet/noise) with latencies in the three distinct time regions entered as dependent variables. Noise delayed peak timing across all time regions [onset: F_{(1, 42)} = 98.008, p < 0.001; transition: F_{(1, 46)} = 19.113, p < 0.001; vowel F_{(1, 46)} = 2.375, p = 0.025]. Musicians demonstrated earlier neural response timing for both the onset [F_{(1, 42)} = 11.080, p = 0.002] and the transition [F_{(1, 46)} = 13.219, p < 0.001] but not for the vowel [F_{(1, 46)} = 1.471, p = 0.185]. A significant group-by-condition interaction was found for all three time regions [onset: F_{(1, 42)} = 4.822, p = 0.034; transition: F_{(1, 46)} = 3.668, p < 0.019; vowel F_{(1, 46)} = 2.053, p = 0.050]. Post-hoc tests revealed that musicians had significantly earlier responses in both quiet and noise conditions for the onset and transition [Onset_quiet: F_{(1, 42)} = 4.521, p = 0.039; Onset_noise: F_{(1, 42)} = 12.720, p = 0.001; Transition_quiet: F_{(1, 46)} = 10.459, p < 0.001; Transition_noise: F_{(1, 46)} = 11.786, p < 0.001], whereas for the steady-state musicians and nonmusicians were equated in quiet but musicians were earlier in noise [Vowel_quiet: F_{(1, 46)} = 1.423, p = 0.205; Vowel_noise: F_{(1, 46)} = 1.912, p = 0.071]. In summary, musicians demonstrate earlier response timing in quiet for the onset and the transition but not the vowel. We also find that the addition of background noise delays neural responses for both groups, but that musicians' responses shifted less than those of nonmusicians (Figure 2).

Hearing in noise relies on the ability to distinguish and track the target voice from the background noise, and recognizing the distinct timbral signature of a person is a key way to achieve this. Envelope and harmonic cues contribute to timbre (Krimphoff et al., 1994; McAdams et al., 1995), making them an important component of the neural code.

Envelope encoding, like stimulus-to-response correlations, is thought to represent the neural encoding of mid-to-high frequency neurons (Dau, 2003; Parbery-Clark et al., 2009a; Ruggles et al., 2012), thus providing a direct link between envelope encoding, the neural representation of higher harmonics and timbre. Furthermore, robust envelope and stimulus-to-response correlations are behaviorally relevant in that they facilitate listening in complex environments such as in background noise (Parbery-Clark et al., 2009a; Swaminathan and Heinz, 2012) or reverberant environments (Ruggles et al., 2012). Our results indicate that middle-aged musicians have stronger representation of envelope, stimulus-to-response and harmonic encoding than nonmusicians and we believe that the strengthened encoding of these spectral features may afford musicians the ability to better discern and segregate voices, giving them an advantage for speech-in-noise perception (Parbery-Clark et al., 2009a,b, 2011a; Zendel and Alain, 2011; Strait et al., in press). Throughout their training and subsequent musical experience, musicians spend countless hours attending to spectrally rich musical sounds, learning to use subtle differences in acoustic cues to discriminate instruments. Spectral information is of great behavioral relevance for musicians, with young adult musicians detecting slight harmonic differences as well as having a greater neural representation of harmonics (Koelsch et al., 1999; Shahin et al., 2005; Musacchia et al., 2008; Lee et al., 2009; Parbery-Clark et al., 2009a; Zendel and Alain, 2009). Our results indicate that older musicians also have a greater neural representation of the harmonics than nonmusicians suggesting that musical experience maintains spectral encoding despite the general trajectory of decline in the ability of the nervous system to represent spectral cues across the lifespan (Clinard et al., 2010; Ruggles et al., 2011, 2012; Anderson et al., 2012).

Middle-aged musicians demonstrate enhanced neural timing of speech in both quiet and noise—as has been found in child musicians (Strait et al., in press), whereas young adult musicians (Parbery-Clark et al., 2009a) only exhibit these enhancements in the more challenging of the two conditions—in noise. In explaining the developmental trajectory between child musicians to young adults, we propose that musical training during childhood accelerates the developmental trajectory of neural mechanisms underpinning the neural encoding of sound, as demonstrated by earlier neural response timing in child musicians, whether it be in the presence or absence of background noise (Strait et al., in press). By young adulthood, we suggest that nonmusicians have “caught up” with the musicians in that both groups are equated for response timing in quiet, even though musicians are still earlier in noise (Parbery-Clark et al., 2009a; Strait et al., in press). Here we extend this proposal to suggest that on the other side of the life cycle—that of aging—musical experience prevents declines in neural mechanisms that underlie neural encoding irrespective of the listening environment.

Our results highlight faster response timing in middle-aged musicians for the onset and transition—two parts of the response that decline with both age (Anderson et al., 2012; Parbery-Clark et al., 2012) and the introduction of noise (Cunningham et al., 2002; Parbery-Clark et al., 2009a; Anderson et al., 2010), are the most challenging in terms of perception (Miller and Nicely, 1955) and neural encoding (Anderson et al., 2010). Importantly, in quiet, there were no group differences for the vowel, indicating that the middle-aged nonmusician's neural responses are not globally delayed for response timing; rather, these effects were exclusively found in the response to the most complex portions of the sound (Parbery-Clark et al., 2012). The addition of background noise did result in a general delay for both groups; still, musicians' responses were delayed to a lesser extent. Musicians' decreased neural response degradation in noise was further evidenced by more consistent neural responses. Taken together, our results provide evidence for musical training across the life span having a pervasive effect on sensory and neural processing, maintaining neural function both in quiet and noisy conditions.

To date, the majority of research supporting the use of musicians as a model of plasticity has focused on child or young adult populations (for review see: Münte et al., 2002; Zatorre and McGill, 2005; Habib and Besson, 2009; Kraus and Chandrasekaran, 2010). While this work has increased our understanding of the effects of music on the nervous system, the role of musical training in the older normal hearing adult population remains largely unexplored. Given that musical training strengthens those skills that decline with age, we argue that the musician's brain provides an optimal model for studying the effects of age on the nervous system. Aging declines are thought to start as early as middle age (Salthouse et al., 1996; Helfer and Vargo, 2009; Ruggles et al., 2011, 2012; Parbery-Clark et al., 2012) and are accompanied by a decrease in central nervous system function, which holds important implications for perceptual and cognitive skills (Craik and Salthouse, 2000). Given that aging musicians maintain an advantage over nonmusicians in terms of neural processing (Parbery-Clark et al., 2012), auditory perception (Parbery-Clark et al., 2011a; Zendel and Alain, 2011) and cognitive abilities (Hanna-Pladdy and MacKay, 2011; Parbery-Clark et al., 2011a; Hanna-Pladdy and Gajewski, 2012), older musicians may provide a means to better understand what contributes to successful aging.

The application of musical experience to the study of aging requires knowledge of the effects of aging on the nervous system. One of the major neurophysiological hallmarks of aging is delayed neural timing and decreased temporal processing (Walton et al., 1998; Burkard and Sims, 2001; Frisina, 2001; Finlayson, 2002; Tremblay et al., 2003; Frisina and Walton, 2006; Lister et al., 2011; Parthasarathy and Bartlett, 2011; Recanzone et al., 2011; Vander Werff and Burns, 2011; Wang et al., 2011; Anderson et al., 2012; Konrad-Martin et al., 2012; Parbery-Clark et al., 2012). These age-related deficits are caused, at least in part, by a decrease in inhibitory mechanisms. With aging, the inhibitory neurotransmitters that facilitate the accurate neural encoding of temporally dynamic and complex sounds (Walton et al., 1998; Caspary et al., 2002, 2008) as well as response consistency (Pichora-Fuller and Schneider, 1992) are reduced throughout the auditory pathway (Caspary et al., 1995, 2005; Wang et al., 2009; de Villers-Sidani et al., 2010; Hughes et al., 2010; Juarez-Salinas et al., 2010). Because the ABR requires a high-degree of neural synchronicity (Kraus et al., 2000), decreased neural consistency such as that caused by temporal jitter (Pichora-Fuller et al., 2007) or neural response variability (Turner et al., 2005; Yang et al., 2009) associated with aging can also contribute to delayed neural response timing and reduced spectral encoding (Anderson et al., 2012). Here we present musician advantages for neural response timing, spectral encoding, and neural response consistency—all factors known to decline with age. For these reasons, we propose that the study of the older musician may be beneficial in elucidating the specific neural components that are enhanced relative to nonmusicians or impervious to age-related declines—highlighting which aspects may be amenable to rehabilititation.

We document enhanced neural encoding in a normal hearing, middle-aged adult musician population. Because aging also results in a higher prevalence of hearing loss, it will be important to define how musical experience interacts in an older adult population with sensory hearing loss. Additionally, our earlier work demonstrated that young adult musicians (19–30 years) had minimal neural differences in quiet (Parbery-Clark et al., 2009a), yet the present results show striking group differences in a middle-aged group (45–65 years) for the same condition. Determining the time course of the neural changes that occur between these two age groups (i.e., young and middle-aged adults) will further our understanding of the effects of aging on the nervous system, as well as the role musicianship plays to offset these declines.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.