Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 3.
Published in final edited form as: Ear Hear. 2020 Dec 3;42(3):691–699. doi: 10.1097/AUD.0000000000000975

Binaural Frequency Modulation Detection in School-age Children, Young Adults, and Older Adults: Effects of Interaural Modulator Phase

Stacey G Kane 1, Emily Buss 1, John H Grose 1
PMCID: PMC8087618  NIHMSID: NIHMS1634107  PMID: 33306546

Abstract

Objectives.

The purpose of this study was to measure low-rate binaural frequency modulation (FM) detection across the lifespan as a gauge of temporal fine structure processing. Children and older adults were expected to perform more poorly than young adults but for different reasons.

Design.

Detection of 2-Hz FM carried by a 500-Hz pure tone was measured for modulators that were either in-phase or out-of-phase across ears. Thresholds were measured in quiet and in noise. Participants were school-age children (n = 44), young adults (n = 11), and older adults (n = 17) with normal or near-normal hearing.

Results.

Thresholds were lower for out-of-phase than in-phase modulators among all listening groups. Detection thresholds improved with child age, with larger effects of age for dichotic than diotic FM. Introduction of masking noise tended to elevate thresholds; this effect was larger for the dichotic condition than the diotic condition, and larger for older adults than young adults. In noise, young adults received the greatest dichotic benefit, followed by older adults, then young children. The relative effects of noise on dichotic benefit did not differ for young adults compared to young children and older adults; however, young children saw greater reduction in benefit due to noise than older adults.

Conclusion.

The difference in dichotic benefit between children and young adults is consistent with maturation of central auditory processing. Differences in the effect of noise on dichotic benefit in young children and older adults support the idea that different factors or combinations of factors limit performance in these two groups. Although dichotic FM detection appears to be more sensitive to the effects of development and aging than diotic FM detection, the positive correlation between diotic and dichotic FM detection thresholds for all listeners suggests contribution of one or more factors common to both conditions.

Introduction

Young adults perform better on many auditory tasks compared to young children and older adults. For example, masked speech recognition thresholds are often lower for young adults than school-age children (Buss et al. 2018; Corbin et al. 2016; Wightman et al. 2005) and older adults (Goossens et al. 2017; Helfer et al. 2014; Rajan et al. 2008; Tun et al. 2002). Likewise, age effects are evident in non-linguistic paradigms like temporal masking (Buss et al. 2013; Grose et al. 2016), gap detection (Buss et al. 2017; Gallun et al. 2014; Grose et al. 2006), frequency discrimination (Buss et al. 2014; Grose et al. 2012), and the binaural masking level difference (BMLD; Anderson et al. 2018; Eddins et al. 2018; Grose et al. 1997). The majority of these studies have focused on either development or on senescence, testing exclusively children or older adults, respectively. The purpose of this study was to include both ends of the age spectrum in a single investigation examining temporal processing, probed with a binaural task that relies on binaural processing of temporal fine structure (TFS). The rationale for the lifespan approach was to gain insight into the presumably different mechanisms that contribute to poorer performance in normal-hearing children and older adults on this task.

Temporal cues are typically subdivided into two categories: envelope cues and TFS cues. Envelope cues are those associated with the relatively slow changes in a signal’s amplitude over time, and they play an important role in speech perception (Shannon et al. 1995). Using gap detection as a measure of envelope processing, Buss et al. (2017) found that monaural gap detection thresholds are higher in school-age children than in young adults for both wideband and narrowband noise. Similarly, older adults produce higher monaural and binaural gap detection thresholds than young adults when tested with broadband and pure-tone stimuli (Gallun et al. 2014; Grose et al. 2006). Studies of masking period patterns have also shown that both children (Buss et al. 2013) and older adults (Grose et al. 2016) experience more forward and backward masking than young adults.

The envelope is carried by the TFS, which refers to the cycle-by-cycle pressure changes associated with the carrier stimulus. In order to make use of TFS cues, these rapid oscillations in stimulus pressure must be encoded with high fidelity in the peripheral auditory system. This is thought to be accomplished via the phase locking capacity of the auditory nerve. Evidence suggests that encoding of TFS cues is important to speech perception (Sheft et al. 2008), particularly for recognition of speech in noise (Hopkins et al. 2009), and in spatial release from masking (Swaminathan et al. 2016). The ability to use TFS cues by listeners of different ages has been tested using both monaural and binaural psychophysical tasks, although there is debate regarding the extent to which monaural tasks characterize a listener’s access to TFS cues (Verschooten et al. 2019).

Of the several monaural tasks that have been proposed to study TFS processing (e.g., Hopkins et al. 2007, 2011; Moore et al. 2009), the one relevant here is low-rate frequency modulation (FM) detection for low-frequency carriers. Whereas FM detection at high carrier frequencies is thought to rely on a change in place of transduction, performance at low carrier frequencies and low rates of modulation is sometimes argued to reflect phase locking in the auditory nerve (Moore et al. 1996). Thresholds are quantified as the minimum modulation depth necessary to distinguish between a pure tone and a FM tone with the same center frequency. Children’s low-rate FM detection is not adult-like until 9–13 years of age (Buss et al. 2014). Similarly, older adults are poorer at FM detection than young adult listeners (Grose & Mamo 2012).

In contrast to monaural tasks, binaural tasks relying on TFS processing require detection of differences in signals presented to the two ears and are thought to reflect the combined strength of phase locking to stimuli presented to each ear. Batra et al. (1997) demonstrated that the interaural synchrony index (a measure of phase-locking strength that varies from 0 – 1) was approximately the product of the two monaural indices. This suggests effects of degraded TFS coding should be more evident in binaural than monaural tasks. This prediction assumes that the fidelity of TFS coding is independent across ears and no additional temporal jitter is introduced by the binaural system. Examples of binaural TFS tasks include the BMLD, detection of interaural phase differences (IPDs), and binaural FM detection. Of these, only the BMLD has been examined in both children and older adults.

The BMLD typically refers to the reduction in thresholds for a binaural signal that is out-of-phase across ears relative to one that is in-phase, with the masker being diotic. This benefit is due in part to IPD cues, and is larger at low than high frequencies, which has been interpreted in terms of greater reliance on TFS cues for detection at low frequencies and envelope cues at high frequencies (Buss et al. 2007; Eddins et al. 1998). The limited ability of human listeners to use high-frequency TFS cues may indicate limits to temporal resolution for binaural hearing (Bernstein et al. 1996; Verschooten et al. 2019). The BMLD is adult-like for school-age children tested with a wideband noise masker, but more prolonged developmental effects are observed when the masker is narrowband noise (Grose et al. 1997). Hall et al. (2004) suggested the narrowband BMLD paradigm may be more sensitive to immature temporal resolution than the wideband BMLD. Similarly, the BMLD is smaller for older adults than young adults for both wide and narrowband maskers (Anderson et al. 2018; Eddins & Eddins 2018); these effects are observed for speech and pure-tone signals (Grose et al. 1994). Taken together these results suggest reduced benefit from binaural TFS cues in children and older adults compared to young adults.

Whereas the BMLD paradigm typically provides both envelope cues and TFS cues, IPD detection is restricted to TFS cues. In a study of older adults, Grose and Mamo (2010) measured IPD detection thresholds using 5-Hz sinusoidally amplitude modulated (SAM) tones. In the standard intervals, the carrier frequency remained interaurally in-phase for all modulation cycles, whereas in the target intervals the carrier alternated between being in-phase and out-of-phase across sequential modulation cycles. Older adults performed more poorly than young adults on this task. In another study using a similar test paradigm designed for clinical use, Fullgrabe and colleagues (2017) found a similar pattern of results.

It is also possible to obtain a binaural benefit from dynamic interaural phase discrepancies (Grose & Mamo 2012; Witton et al. 2000). For example, Grose and Mamo (2012) measured 2-Hz FM detection carried by a 500-Hz tone, with and without a 500-Hz contralateral stimulus that was either a pure tone or an FM tone with the modulator out-of-phase relative to the target. Performance improved significantly with the introduction of the contralateral stimulus, but the magnitude of the benefit differed across age groups. Young adults received the most benefit, followed by middle-aged adults, and older adults received the smallest benefit from a contralateral stimulus. This suggests that binaural tests for TFS, such as FM detection for interaurally out-ot-phase modulators, may be more sensitive than monaural tests to differences in TFS fidelity. This paradigm, however, has not been tested with children, and that forms one of the goals of this study.

In summary, although young adults tend to perform better than young children or older adults on tasks thought to rely on TFS, even in the presence of normal audiometric hearing at the test frequency, there has not been a systematic comparison of performance across the lifespan. This gap in the literature is important because different mechanisms have been proposed to account for effects of development and aging. In the case of young children, the assumption is that the peripheral auditory system represents TFS information with high fidelity (Moore & Linthicum, 2007), but that more central processing is immature. In contrast, limitations in temporal encoding alone or in combination with central effects of aging are often invoked to explain TFS performance in older adults. The goal of this study was to determine whether performance on a binaural TFS task reveals characteristics of performance that differentiate mechanisms of aging and childhood development on sensitivity to TFS.

The present study evaluated binaural FM detection for in-phase and out-of-phase modulators in school-age children, young adults, and older adults with normal or near-normal hearing. Based on prior data, we expected better in-phase and out-of-phase FM detection thresholds for young adults than for children or older adults. If neural synchrony is compromised in older adults, then thresholds in the out-of-phase condition should be particularly elevated, due to the binaural combination of synchronization error in the encoding of sound from each ear.

In a further novel extension of this paradigm, FM detection thresholds were measured in both quiet and in noise. The addition of noise weakens neural synchrony at the monaural level by undermining the entrainment of neural responses to the periodicity of the signal and ‘capturing’ some of the spike activity (e.g., Henry et al. 2013). Compromising the monaural synchrony index in this way should result in a proportionally larger reduction of the binaural index (Batra et al. 1997). The use of background noise to undermine psychoacoustic measures of TFS coding has been used previously (Strelcyk & Dau 2009). We hypothesized that masking noise would elevate thresholds to a similar degree for children and young adults, two groups presumed to have intact peripheral encoding. In older adults, for whom we suspected poorer encoding at the input stages of auditory processing, we considered two potential outcomes. An aging auditory system with compromised neural synchrony could be less susceptible to the addition of low-level masking noise than a system with an intact periphery. This would occur if the magnitude of inherent neural jitter in older adults was sufficient to partially offset effects of masking noise. Alternatively, poor representation of temporal information in the peripheral auditory system of older adults could decrease tolerance for further cue degradation, such as that associated with the addition of masking noise. Based on these two scenarios, we predicted different effects of noise for older adults compared to children and young adults, without a clear prediction regarding the direction of that difference.

Materials and Method

Participants

A total of 72 listeners participated in three age groups: 44 children (5.2 –12.9 yrs, 24 female), 11 young adults (20.8 – 35.4 yrs, 7 female), and 17 older adults (62.5 – 76.4 yrs, 13 female). All children and young adults had normal audiometric hearing, with pure-tone detection thresholds ≤ 20 dB HL from 250 – 8000 Hz in both ears (ANSI 2010). Older adults had hearing thresholds ≤ 25 dB HL from 250 – 4000 Hz in both ears. All participants or parents/caregivers denied a significant history of middle ear disease. Adult participants gave written consent to participate; child participants ≥ 7 years gave written assent, with their parents/caregivers providing written consent; child participants < 7 years provided verbal assent, with written consent from parents/caregivers. All participants were reimbursed for their participation. This study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill.

Stimuli

Stimuli were 500-Hz pure tones and FM tones, modeled after those used by Grose and Mamo (2012). The sinusoidally FM tones were computed according to the equation:

X(t)=Asin[2πfct+βsin(2πfmt+θ)]

where t is time in sec, ƒc is the carrier frequency (500 Hz), and ƒm is the modulator frequency (2 Hz). Modulation depth was controlled by the modulation index, ß, which is the change in frequency (Δƒ) divided by ƒm. For the in-phase condition (diotic), the starting phase of FM (θ) was 0 radians in both ears. For the out-of-phase condition (dichotic), the starting phase of FM was 0 radians in the left ear and π radians in the right ear. Tones were 1250 ms in duration, including 25-ms rise/fall ramps, and they were presented at 65 dB SPL.

Masking noise, when present, was a one-octave wide continuous Gaussian noise centered at 500 Hz (356 – 707 Hz) and presented at an overall level of 61.5 dB SPL. This level was selected so that the 79-Hz wide sub-band of masking noise constituting the normal equivalent rectangular bandwidth (ERB, Moore et al. 1983) centered at 500 Hz had a level of 55 dB SPL (i.e., −10 dB re the signal level). Independent samples of noise were presented to the left and right ear. Presentation level for the noise was selected based on the masked lateralization study of Strelcyk and Dau (2009). That study showed a decrement in binaural performance with addition of a noise that was 11-dB-down from the signal in the ERB centered on the signal. The presence of masking noise is known to have a modest effect on detection of in-phase FM (Wier et al. 1976). Stimuli were generated at a sampling rate of 25 kHz and presented binaurally through Sennheiser HD 251-II supra-aural headphones.

Procedure

The FM detection threshold was estimated with a three interval, three-alternative forced-choice (3AFC) procedure. Two intervals contained a 500-Hz pure tone, and one interval, chosen at random, contained the FM tone. The inter-stimulus interval was 500 ms. An adaptive 2-down, 1-up stepping procedure estimated 71% correct performance. Each threshold estimation track continued for eight reversals. The FM depth was adjusted by changing Δƒ in factorial steps; the initial step size was a factor of 2, and this factor was adjusted by an exponent of 0.5 after both the first and second reversals in FM depth direction. The final threshold estimate was the geometric mean of the last six1 reversals. Individual trial-by-trial data were recorded for all participants.

All testing was completed in a double-walled, sound-treated booth. For each trial, participants were presented with a graphic of three frogs programmed to open/close their mouths in synchrony with the onset/offset of each listening interval. After listening to the three intervals, participants were instructed to select the frog that made a different sound using the touchscreen or a mouse. The target sound was described as the one that moved or wobbled. As feedback to the listener, the frog corresponding to the target interval caught a fly. All listeners were provided with as many breaks as needed; in some cases, testing was completed in two separate sessions. Young children often completed testing with an examiner present in the booth for encouragement.

Presentation order was counterbalanced for all participants. Adult participants listened in all four conditions (diotic/dichotic FM and quiet/masking noise). Child participants listened in either quiet or noise, with one exception; one child (12.9 years) listened in all four conditions. Adults completed three to five estimates per condition, whereas children completed between four and eight estimates per condition. The rationale for obtaining more estimates for children than adults was the expectation that child listeners might require more practice to reach asymptotic performance than adults. Throughout testing for all participants, research assistants monitored the range of threshold estimates for each condition. If an estimate was outside a range of 1.5 times the lowest threshold, an additional track was collected for that condition. Evaluation of child thresholds did not reveal reliable evidence of a practice effect.

Analysis

Thresholds for detecting FM were expressed in log10 units. To evaluate the dataset for outliers, the standard deviations of the threshold estimates for each child listener and condition were calculated and the overall child group mean standard deviation (group STD) was computed. Threshold estimates for any participant that were 3 or more group STDs from the mean threshold for a participant and condition were considered outliers and discarded. This resulted in nine of the 971 threshold estimates being removed from further analysis across both child and adult participants.

Linear mixed model analyses were used to evaluate the effects of sex, age, modulator phase, and masker on FM detection thresholds in all four listening conditions; group-level analyses were sum coded. For the children, age was log10-transformed to accommodate the decelerating effects of child age (e.g., Buss et al. 2014) and mean centered (log10 of 8.3 yrs). Although a difference in performance based on sex was not expected, current National Institutes for Health guidelines on relevant biologic variables prompted the consideration of this factor here. Pearson’s Product Moment correlations were calculated between diotic and dichotic FM thresholds in quiet to evaluate the association between sensitivity presumed to reflect monaural and binaural TFS cues.

Results

Figure 1 displays mean FM detection thresholds for diotic and dichotic FM in both quiet (left panel) and in noise (right panel) as a function of participant age. For child listeners, there was a trend for thresholds to fall with increasing age. Thresholds tended to be lower for dichotic FM (circles) than diotic FM (asterisks), and this effect appeared to be larger in quiet than in noise.

Fig. 1.

Fig. 1

Open in a new tab

FM detection threshold (Hz) plotted as a function of participant age (yrs). Results obtained in quiet are shown in the left panel, and those obtained in noise are shown in the right panel. Symbol shape reflects signal FM modulator phase, as defined in the legend.

Dealing first with the child data, Table 1 displays results from the linear mixed model analysis of children’s thresholds as a function of masker (quiet, noise), FM modulator phase (in-phase, out-of-phase), log age, and participant sex; this model included two- and three-way interactions between age, masker, and phase. Because children completed testing in either quiet or in noise, degrees of freedom in the model did not allow for additional effects; there was a random intercept for each participant. There were significant main effects of masker (p = 0.001), age (p < 0.001), and phase (p < 0.001), reflecting lower thresholds for older children, thresholds in quiet, and the out-of-phase condition. There was no significant effect of sex (p = 0.295). There were significant interactions between masker and phase (p < 0.001), and between phase and age (p = 0.042), but neither of the other interactions reached significance. The masker-by-phase interaction reflects greater effects of noise in the out-of-phase condition than in the in-phase condition. Paired contrasts revealed a significant effect of masking noise for the out-of-phase condition (β = −0.472, t = −5.43, p < 0.001) but not for the in-phase condition (β = −0.062, t = −0.72, p = 0.475). The age-by-phase interaction reflects a larger age effect for the out-of-phase condition than the in-phase condition. Paired contrasts indicate significant effects of age for both in-phase (β = −0.834, t = −2.18, p = 0.033) and out-of-phase (β = −1.681, t = −4.40, p < 0.001). The three-way interaction did not approach significance.

Table 1.

Linear mixed model analysis of FM detection in children as a function of sex, masker, FM modulator phase, and age.

Value Std. Error df t-value p-value
(Intercept) 0.365 0.037 38.83 9.94 < 0.001
Sex 0.039 0.037 36.62 1.06 0.295
Masker 0.134 0.037 44.35 3.64 0.001
Phase −0.249 0.023 38.90 −10.68 < 0.001
Age −1.257 0.325 36.79 −3.87 < 0.001
Masker: Phase 0.102 0.023 38.90 4.39 < 0.001
Masker: Age 0.028 0.294 62.73 0.10 .924
Phase: Age −0.424 0.202 38.90 −2.10 .042
Masker: Phase: Age −0.081 0.202 38.90 −0.40 .690
Open in a new tab

Attention now turns to the adult data. Table 2 displays results from the linear mixed model analysis of data from adults, evaluating threshold as a function of masker (quiet, noise), FM modulator phase (in-phase, out-of-phase), age group (young adults, older adults), and sex. Interactions were included to evaluate the prior expectation that age effects would be most evident for dichotic FM presented in the noise masker. This model initially included a random intercept and random effects of phase and masker for each participant. Error variance for the random effect of masker was found to be greater than individual variance, so it was removed from the model (see West et al. 2015), but the random effect of phase was maintained. An independent covariance structure was used for this model.

Table 2.

Linear mixed model analysis of FM detection in adults as a function of sex, masker, FM modulator phase, and age group. Reference conditions were quiet (masker), in-phase FM (phase), young adult (group), and female (sex).

Value Std. Error df t-value p-value
(Intercept) 0.107 0.046 25 2.35 0.027
Sex −0.010 0.047 25 −0.22 0.831
Masker 0.168 0.010 52 16.41 <0.001
Phase −0.369 0.019 26 −19.86 <0.001
Group 0.070 0.041 25 1.70 0.101
Masker: Phase 0.054 0.010 52 5.26 <0.001
Masker: Group 0.023 0.010 52 2.24 0.030
Phase: Group 0.073 0.019 26 3.92 0.001
Masker: Phase: Group −0.016 0.010 52 −1.55 0.128
Open in a new tab

There were significant main effects of masker and modulator phase (p < 0.001), but not group (p = 0.101) or sex (p = 0.831). All three of the two-way interactions were significant (p ≤ 0.030), but the three-way interaction was not (p = 0.128). The masker-by-phase interaction reflects more modest increases in FM detection threshold with the addition of masking noise for the in-phase (β = −0.229, t = −7.88, p < 0.001) than the out-of-phase (β = −0.444, t = −15.32, p < 0.001) conditions. The masker-by-group interaction reflects the fact that older adults were more detrimentally affected by noise (β = −0.382, t = −14.87, p < 0.001) than young adults (β = −0.291, t = −9.09, p < 0.001). The phase-by-group interaction reflects the fact that older listeners had significantly higher thresholds than younger listeners in the out-of-phase condition (β = 0.286, t = 3.16, p < 0.003) but not the in-phase condition (β = −0.005, t = −.06, p = .954). In other words, older adults received less benefit from out-of-phase FM than young adults (β = −0.291, t = −3.92, p = 0.001).

A third analysis compared performance of children and adults. Due to the effect of child age, this analysis only evaluated the data of children younger than 8 years old, with the rationale that data from the youngest children represent the most prominent effects of development when compared to adults. Figure 2 shows the benefit of out-of-phase FM for young children, young adults, and older adults in quiet and noise. Benefit was calculated as the difference in log10-transformed thresholds for in-phase and out-of-phase FM, such that larger numbers represent greater benefit of out-of-phase FM. Boxplots show the distribution of values for each age group, with the parameter of masking condition indicated by differential shading (see inset). The benefit of dichotic FM tended to be positive, with the notable exception of children tested in masking noise; for this condition half of the eight child listeners had a negative value of benefit, indicating poorer thresholds for the dichotic than the diotic FM.

Fig. 2.

Fig. 2

Open in a new tab

Benefit of dichotic FM plotted as function of age group in quiet and noise. The benefit of dichotic FM is the difference between thresholds for detecting in-phase and out-of-phase FM, with a log10 transformation applied prior to subtraction. Results obtained in quiet are shown with unfilled boxes, and those in noise with grey boxes. Horizontal lines indicate the median, boxes span the 25th to 75th percentiles, and vertical lines span the 10th-90th percentiles. Data for individual listeners are shown with circles. The dashed line at 0 indicates no advantage for dichotic FM presentation.

Table 3 reports the results of a linear mixed model evaluating the effect of listener age group on the benefit associated with dichotic FM relative to diotic FM. There were two levels of masker (quiet, noise) and three levels of age group (young child, young adult, older adult). Because each child listened in either quiet or noise, and masker did not account for substantial individual variance among adult listeners, this model did not include a random effect of masker; there were random intercepts for each participant. Note, because factors in this analysis were sum coded, coefficients reflect group differences from the overall mean. Therefore, factors with more than two levels (group) are not uniquely defined. Using model estimated means for each group, paired comparisons were conducted to further evaluate results of this model.

Table 3:

Results of a linear mixed model predicting benefit of dichotic FM with factors age group and masker.

Value Std. Error df t-value p-value
(Intercept) 0.614 0.033 39.46 18.39 < 0.001
Masker −0.154 0.025 63.25 −6.27 < 0.001
Group (young children) −0.248 0.046 48.44 −5.34 < 0.001
Group (older adults) −0.022 0.045 36.12 −0.49 0.631
Masker: Group (young children) −0.093 0.041 69.25 −2.29 0.025
Masker: Group (older adults) 0.078 0.030 47.17 2.59 0.013
Open in a new tab

Overall, masking reduced dichotic benefit for children (β = 0.495, t = 4.427, p< 0.001), younger adults (β = 0.279, t = 3.706, p = 0.001), and older adults (β = 0.152, t = 2.514, p = 0.019). Post hoc analysis compared differences in dichotic benefit between quiet and noise among all three age groups. There was an overall effect of group (F (2, 39.39) = 18.28, p < 0.001) and masker by group interaction (F (2, 36.17) = 3.76, p = 0.033). Paired comparisons revealed no significant differences in relative benefit for older and younger adults (β = −0.127, t = −1.31, p = 0.202), or for young children and younger adults (β = 0.215, t = 1.60, p = 0.115). The masking noise reduced the dichotic benefit significantly more for young children than older adults (β = 0.342, t = 2.69, p = 0.009).

Finally, we evaluated the differences in benefit for each age group in quiet (F(2, 58.79) = 9.51, p< 0.001) and noise (F(2, 58.79) = 14.72, p< 0.001). In quiet, there was greater dichotic benefit for younger adults compared to both older adults (β = − 0.355, t = −3.69, p < 0.001) and young children (β = −0.410, t = −4.02, p < 0.001); there was no difference in benefit in quiet for children and older adults (β = − 0.055, t = −0.60, p = 0.552). In noise, there were differences in performance across all comparisons; younger adults experienced greater benefit than both older adults (β = − 0.228, t = −2.37, p = 0.021) and children (β = −0.625, t = −5.41, p < 0.001), and older adults experienced greater benefit than young children (β = −0.397, t = −3.73, p < 0.001).

A final analysis evaluated Pearson Product correlations between diotic and dichotic FM detection thresholds in quiet. Table 4 displays the results for all participants and for each age group evaluated separately. Results suggest a significant positive relationship between diotic and dichotic FM detection when evaluating data from all participants (p < 0.001) and for each age group (children: p = 0.001; young adults: p = 0.044; older adults: p = 0.004).

Table 4.

Pearson’s Product-Moment Correlation analyses for diotic and dichotic FM detection thresholds in quiet for all age groups.

r 95% CI df t-value p-value
All Participants Quiet 0.66 0.47–0.80 47 6.06 < 0.001
Children Quiet 0.65 0.31–0.85 19 3.78 0.001
Young Adults Quiet 0.62 0.03–0.89 9 2.35 0.044
Older Adults Quiet 0.66 0.26–0.86 15 3.38 0.004
Open in a new tab

Discussion

Thresholds obtained here are largely consistent with FM detection thresholds reported in the literature. For example, Buss et al. (2014) evaluated detection of 2-Hz FM with a monaural 500-Hz carrier presented in quiet; mean FM detection thresholds for 5–13 year-olds were approximately 5–10 Hz in that study. That result can be compared to a mean threshold of 4.1 Hz for in-phase FM detection in quiet for children tested in the present study. Likewise, Grose and Mamo (2012) reported FM detection in quiet for older and younger adults, using stimuli like those of the present study. Mean thresholds for in-phase stimuli in that study were 3.5 Hz (older adults) and 2.0 Hz (younger adults); these values can be compared to mean thresholds for in-phase FM measured in quiet of 2.1 and 2.5 Hz, respectively, in the present study. Mean thresholds for out-of-phase stimuli in Grose and Mamo (2012) were 1.8 Hz (older adults) and 0.4 Hz (younger adults); these values can be compared to mean thresholds of 0.5 Hz and 0.2 Hz, respectively. While the data patterns for younger and older adults observed here are generally similar to the published data, differences between groups were generally smaller than seen previously, with no indication of poorer performance in older adults for in-phase FM detection.

The purpose of this study was to compare binaural 2-Hz FM detection thresholds for in-phase and out-of-phase modulators in individuals with normal hearing who varied in age across the lifespan; the FM signal was carried by a 500-Hz pure tone and was presented in quiet or in masking noise. For these stimulus parameters, FM detection is thought to reflect use of TFS cues (Moore & Sek 1996). The motivation for comparing thresholds across age groups was to evaluate possible mechanisms affecting performance in young children and older adults relative to young adult listeners. For instance, when compared to young adults, poorer performance in older adults on TFS tasks has been attributed to poor peripheral encoding (e.g., reduced neural synchrony). This would not be a reasonable explanation for poorer performance in children considering the expectation of a fully developed, pristine peripheral auditory system (Moore & Linthicum, 2007), and previous results showing a similar child/adult difference for auditory tasks relying to different degrees on TFS (e.g., FM detection at different rates and carrier frequencies, Buss et al. 2014). A more likely explanation for poorer performance in children might be the effects of developing cognitive skills, including selective attention and auditory memory. Because of these age differences in potential mechanisms, we sought to evaluate the pattern of FM detection thresholds for in-phase and out-of-phase modulators, with the goal of differentiating between the effects of development and aging on estimates of sensitivity to changes in TFS.

The results indicated that children, young adults, and older adults all received benefit from the dichotic FM cue, and that thresholds increased when stimuli were presented in noise. There were significant improvements in both diotic and dichotic FM detection with increasing child age, consistent with previous data on monaural FM detection (Buss et al. 2014). Relative to the young adults, older adults were more affected by the noise masker and received less overall benefit from dichotic FM.

These findings highlight some important characteristics of FM detection in children. First, even young children can benefit from dichotic FM cues, demonstrating their ability to make use of binaural TFS cues. Second, consistent with previous FM detection studies, this dataset shows a significant effect of development. This improvement in performance with child age has also been demonstrated in other psychophysical tasks for which differences in peripheral encoding of auditory information based on age would not be expected. For example, children show significant improvements with age in gap detection, a task that relies on temporal envelope cues (Buss et al. 2017). Because this maturational trend can be seen across paradigms and stimuli, relying on different auditory cues, the most parsimonious explanation is development at a central level of auditory processing (e.g., attention to the task).

The finding that dichotic FM detection is poorer in older adults than younger adults is consistent with results from Grose and Mamo (2012) and provides additional support for the use of dichotic FM stimuli in differentiating TFS fidelity among adult listeners. Moreover, this indicates that TFS cues may not be as readily available to older listeners. One difference between the Grose and Mamo (2012) study and the present experiment is the application of rove to the stimulus carrier frequency. In the published study, the carrier frequency varied between 460–540 Hz on a trial-by-trial basis, to discourage reliance on place cues. Rove was not applied in the present study due to concern that rove would increase task difficulty for young children.

Results showed that introduction of a noise masker degrades the ability to benefit from dichotic FM cues in all age groups, but to differing extents. While absolute FM detection thresholds were significantly higher for older adults than young adults in noise, data did not indicate a greater difference in dichotic benefit between quiet and noise for older versus younger adult listeners. Likewise, dichotic benefit was similarly affected by noise in young adults and young children. However, young children experienced a significantly larger reduction of dichotic benefit with the addition of noise than older adults.

If neural synchrony is already fragile in the senescent auditory system, this could result in two different patterns of performance in older adults relative to young adults and young children. First, an older adult listener may be more susceptible to disruption from masking noise than children and young adults, due to tenuous encoding of temporal information. In this instance, we would expect our data to reveal the greatest detrimental effects of noise in older adults accompanied by more modest effects of noise in children and young adults. Alternatively, inherent neural jitter in an aging auditory system could offset any additional stimulus variability introduced via masking. In this instance, young children and young adults might be more affected by the addition of masking than older adults. Results of the present study do not fully support either of these hypotheses.

Compared to young adults, older adults’ FM detection thresholds were higher and their dichotic benefit in noise was poorer. However, the dichotic benefit was overall smaller in young children compared to both young and older adults, and young children experienced greater reductions in benefit due to noise compared to older adults. To the extent that low-rate FM detection reflects TFS processing, the difference between susceptibility to masking in young children and older adults supports the idea that an auditory system with inherent neural jitter is less affected by noise than one without neural jitter. The finding of greater reduction in the dichotic benefit with the addition of masking noise for children than for older adults could reflect the greater fidelity of neural encoding in children, such that variability introduced by the noise affects performance. However, it is not clear why noise would entirely eliminate dichotic benefit for children in this scenario, or why there is no significant difference in the effect of noise on dichotic benefit for young adults and older adults.

An alternative explanation for children’s poorer FM detection in the presence of masking noise is related to perceptual similarity between the masker and the FM detection cue. In quiet, one cue that listeners could use to detect dichotic FM is deviation in perceived intracranial position of the sound from the midline; i.e., whereas the diotic standard stimulus generates a steady centralized auditory image, the 2-Hz dichotic FM signal sounds as if it is moving from side to side. Recall that the masker was composed of interaurally independent noise samples, which itself results in a diffuse spatial percept. Children may have more difficulty focusing their attention on the spatial cue associated with dichotic FM in the presence of the spatially diffuse masker than in quiet. This possibility receives indirect support from the observation that young school-age children perform more poorly than adults on a wide range of tasks with perceptually complex stimuli (for review, see Buss et al. 2012; Leibold 2012).

Finally, the correlational analysis in this study demonstrated a reliable relationship between performance in the diotic and dichotic FM conditions in all age groups. This supports an association between monaural and binaural TFS processing, insofar as diotic FM detection reflects monaural TFS processing. This association is relevant in light of previous suggestions that monaural and binaural TFS tests might assess different facets of TFS processing (Fullgrabe et al. 2014; Moore et al. 2012). Further study is required to better understand the factors affecting the use of TFS cues under different stimulus conditions, as well as possible effects of central auditory processing (e.g., attention and working memory) that may affect in-phase and out-of-phase FM detection.

In summary, this study examined age-related effects in TFS processing using the sensitive test of binaural FM detection as a function of interaural modulator phase. As hypothesized, children and older adults with normal or near-normal audiometric hearing performed more poorly than young adults. Whereas this pattern of results was consistent with different underlying mechanisms contributing to the deficits observed in both children and older adults, further study of underlying mechanisms as a function of age is warranted. Additionally, much remains to be understood about how TFS cue usage as measured here relates to speech perception and spatial hearing abilities across the lifespan. For example, a strong relationship between binaural TFS processing and spatial release from masking (SRM) has been found in adults of different ages (Papesh et al. 2017), but it is unclear whether such an association generalizes to children. Current work in our laboratory is pursuing these questions.

Acknowledgements

Data collection was carried out with the help of Kathryn Sobon and Megan Bartoshuk. We acknowledge the editorial assistance of the NC Translational and Clinical Sciences (NC TraCS) Institute, which is supported by the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health, through Grant Award Number UL1TR002489. Advice regarding statistics was provided by Dr. Chris Wiessen, UNC Odum Institute for Research in Social Science.

Conflicts of Interest and Source of Funding:

This study was funded by NIH NIDCD (R01 DC001507[JHG] & R01 DC000397[EB]). There are no conflicts of interest, financial or otherwise.

Footnotes

1

For three child participants, thresholds were initially based on five reversals, and no trial-by-trial data were available. In these cases, threshold was estimated using the following procedure. Thresholds based on five and six reversals were estimated for the remaining 69 listeners. We found thresholds to be approximately 3% lower with six than five reversals. This correction factor was applied to the data for the three participants with five reversal thresholds.

References

  1. Anderson S, Ellis R, Mehta J et al. (2018). Age-related differences in binaural masking level differences: behavioral and electrophysiological evidence. J Neurophysiol, 120, 2939–2952. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. ANSI. (2010). Specification for Audiometers. In American National Standards Institute/Acoustical Society of America S3.6–2010. New York: American National Standards Institute. [Google Scholar]
  3. Batra R, Kuwada S, Fitzpatrick DC (1997). Sensitivity to interaural temporal disparities of low- and high-frequency neurons in the superior olivary complex. II. Coincidence detection. J Neurophysiol, 78, 1237–1247. [DOI] [PubMed] [Google Scholar]
  4. Bernstein LR, Trahiotis C (1996). The normalized correlation: accounting for binaural detection across center frequency. The Journal of the Acoustical Society of America, 100, 3774–3784. [DOI] [PubMed] [Google Scholar]
  5. Buss E, Hall JW III, Grose JH (2012). Development of auditory coding as reflected in psychophysical performance. In Werner L, Fay RR, Popper AN (Eds.), Human Auditory Development (pp. 107–136). New York: Springer. [Google Scholar]
  6. Buss E, Hall JW 3rd, Grose JH (2007). Individual differences in the masking level difference with a narrowband masker at 500 or 2000 Hz. J Acoust Soc Am, 121, 411–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Buss E, He S, Grose JH et al. (2013). The monaural temporal window based on masking period pattern data in school-aged children and adults. J Acoust Soc Am, 133, 1586–1597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Buss E, Hodge SE, Calandruccio L et al. (2018). Masked Sentence Recognition in Children, Young Adults, and Older Adults: Age-Dependent Effects of Semantic Context and Masker Type. Ear Hear. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Buss E, Porter HL, Hall JW 3rd et al. (2017). Gap Detection in School-Age Children and Adults: Center Frequency and Ramp Duration. J Speech Lang Hear Res, 60, 172–181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Buss E, Taylor CN, Leibold LJ (2014). Factors affecting sensitivity to frequency change in school-age children and adults. J Speech Lang Hear Res, 57, 1972–1982. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Corbin NE, Bonino AY, Buss E et al. (2016). Development of Open-Set Word Recognition in Children: Speech-Shaped Noise and Two-Talker Speech Maskers. Ear Hear, 37, 55–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Eddins AC, Eddins DA (2018). Cortical Correlates of Binaural Temporal Processing Deficits in Older Adults. Ear Hear, 39, 594–604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Eddins DA, Barber LE (1998). The influence of stimulus envelope and fine structure on the binaural masking level difference. J Acoust Soc Am, 103, 2578–2589. [DOI] [PubMed] [Google Scholar]
  14. Fullgrabe C, Harland AJ, Sek AP et al. (2017). Development of a method for determining binaural sensitivity to temporal fine structure. Int J Audiol, 56, 926–935. [DOI] [PubMed] [Google Scholar]
  15. Fullgrabe C, Moore BC, Stone MA (2014). Age-group differences in speech identification despite matched audiometrically normal hearing: contributions from auditory temporal processing and cognition. Front Aging Neurosci, 6, 347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gallun FJ, McMillan GP, Molis MR et al. (2014). Relating age and hearing loss to monaural, bilateral, and binaural temporal sensitivity. Front Neurosci, 8, 172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Goossens T, Vercammen C, Wouters J et al. (2017). Masked speech perception across the adult lifespan: Impact of age and hearing impairment. Hear Res, 344, 109–124. [DOI] [PubMed] [Google Scholar]
  18. Grose J, Hall J, Dev M (1997). The MLD in children: Effects of signal and masker bandwidths. Journal of Speech and Hearing Research, 40, 955–959. [DOI] [PubMed] [Google Scholar]
  19. Grose JH, Hall JW 3rd, Buss E (2006). Temporal processing deficits in the pre-senescent auditory system. J Acoust Soc Am, 119, 2305–2315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Grose JH, Mamo SK (2010). Processing of temporal fine structure as a function of age. Ear Hear, 31, 755–760. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Grose JH, Mamo SK (2012). Frequency modulation detection as a measure of temporal processing: age-related monaural and binaural effects. Hear Res, 294, 49–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Grose JH, Menezes DC, Porter HL et al. (2016). Masking Period Patterns and Forward Masking for Speech-Shaped Noise: Age-Related Effects. Ear Hear, 37, 48–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Grose JH, Poth EA, Peters RW (1994). Masking level differences for tones and speech in elderly listeners with relatively normal audiograms. J Speech Hear Res, 37, 422–428. [DOI] [PubMed] [Google Scholar]
  24. Hall JW, Buss E, Grose JH et al. (2004). Developmental effects in the masking-level difference. J Speech Lang Hear Res, 47, 13–20. [DOI] [PubMed] [Google Scholar]
  25. Helfer KS, Freyman RL (2014). Stimulus and listener factors affecting age-related changes in competing speech perception. J Acoust Soc Am, 136, 748–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Henry KS, Heinz MG (2013). Effects of sensorineural hearing loss on temporal coding of narrowband and broadband signals in the auditory periphery. Hear Res, 303, 39–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Hopkins K, Moore BC (2007). Moderate cochlear hearing loss leads to a reduced ability to use temporal fine structure information. J Acoust Soc Am, 122, 1055–1068. [DOI] [PubMed] [Google Scholar]
  28. Hopkins K, Moore BC (2009). The contribution of temporal fine structure to the intelligibility of speech in steady and modulated noise. J Acoust Soc Am, 125, 442–446. [DOI] [PubMed] [Google Scholar]
  29. Hopkins K, Moore BC (2011). The effects of age and cochlear hearing loss on temporal fine structure sensitivity, frequency selectivity, and speech reception in noise. The Journal of the Acoustical Society of America, 130, 334–349. [DOI] [PubMed] [Google Scholar]
  30. Leibold L (2012). Development of auditory scene analysis and auditory attention. In Werner L, Fay RR, Popper AN (Eds.), Human Auditory Development (pp. 137–162). New York: Springer. [Google Scholar]
  31. Moore BC, Sek A (1996). Detection of frequency modulation at low modulation rates: evidence for a mechanism based on phase locking. J Acoust Soc Am, 100, 2320–2331. [DOI] [PubMed] [Google Scholar]
  32. Moore BC, Sek A (2009). Development of a fast method for determining sensitivity to temporal fine structure. Int J Audiol, 48, 161–171. [DOI] [PubMed] [Google Scholar]
  33. Moore BC, Vickers DA, Mehta A (2012). The effects of age on temporal fine structure sensitivity in monaural and binaural conditions. Int J Audiol, 51, 715–721. [DOI] [PubMed] [Google Scholar]
  34. Moore BCJ, Glasberg BR (1983). Suggested formulae for calculating auditory filter bandwidths and excitation patterns. Journal of the Acoustical Society of America, 74, 750–753. [DOI] [PubMed] [Google Scholar]
  35. Moore JK, & Linthicum FH Jr. (2007). The human auditory system: A timeline of development. International Journal of Audiology 46: 460–478. [DOI] [PubMed] [Google Scholar]
  36. Papesh MA, Folmer RL, Gallun FJ (2017). Cortical Measures of Binaural Processing Predict Spatial Release from Masking Performance. Front Hum Neurosci, 11, 124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rajan R, Cainer KE (2008). Ageing without hearing loss or cognitive impairment causes a decrease in speech intelligibility only in informational maskers. Neuroscience, 154, 784–795. [DOI] [PubMed] [Google Scholar]
  38. Shannon RV, Zeng FG, Kamath V et al. (1995). Speech recognition with primarily temporal cues. Science, 270, 303–304. [DOI] [PubMed] [Google Scholar]
  39. Sheft S, Ardoint M, Lorenzi C (2008). Speech identification based on temporal fine structure cues. J Acoust Soc Am, 124, 562–575. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Strelcyk O, Dau T (2009). Relations between frequency selectivity, temporal fine-structure processing, and speech reception in impaired hearing. J Acoust Soc Am, 125, 3328–3345. [DOI] [PubMed] [Google Scholar]
  41. Swaminathan J, Mason CR, Streeter TM et al. (2016). Role of Binaural Temporal Fine Structure and Envelope Cues in Cocktail-Party Listening. J Neurosci, 36, 8250–8257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Tun PA, O’Kane G, Wingfield A (2002). Distraction by competing speech in young and older adult listeners. Psychol Aging, 17, 453–467. [DOI] [PubMed] [Google Scholar]
  43. Verschooten E, Shamma S, Oxenham AJ et al. (2019). The upper frequency limit for the use of phase locking to code temporal fine structure in humans: A compilation of viewpoints. Hear Res, 377, 109–121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. West B, Welch K, Galecki A (2015). Linear Mixed Models. New York: Chapman and Hall/CRC. [Google Scholar]
  45. Wier C, Jesteadt W, & Green D (1976). Frequency discrimination as a function of frequency and sensation level. J Acoust Soc Am, 61, 178–184. [DOI] [PubMed] [Google Scholar]
  46. Wightman FL, Kistler DJ (2005). Informational masking of speech in children: effects of ipsilateral and contralateral distracters. J Acoust Soc Am, 118, 3164–3176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Witton C, Green GG, Rees A et al. (2000). Monaural and binaural detection of sinusoidal phase modulation of a 500-Hz tone. J Acoust Soc Am, 108, 1826–1833. [DOI] [PubMed] [Google Scholar]

RESOURCES