No Effect of Musical Training on Frequency Selectivity Estimated Using Three Methods

Abstract

It is widely believed that the frequency selectivity of the auditory system is largely determined by processes occurring in the cochlea. If so, musical training would not be expected to influence frequency selectivity. Consistent with this, auditory filter shapes for low center frequencies do not differ for musicians and nonmusicians. However, it has been reported that psychophysical tuning curves (PTCs) at 4000 Hz were sharper for musicians than for nonmusicians. This study explored the origin of the discrepancy across studies. Frequency selectivity was estimated for musicians and nonmusicians using three methods: fast PTCs with a masker that swept in frequency, “traditional” PTCs obtained using several fixed masker center frequencies, and the notched-noise method. The signal frequency was 4000 Hz. The data were fitted assuming that each side of the auditory filter had the shape of a rounded-exponential function. The sharpness of the auditory filters, estimated as the Q₁₀ values, did not differ significantly between musicians and nonmusicians for any of the methods, but detection efficiency tended to be higher for the musicians. This is consistent with the idea that musicianship influences auditory proficiency but does not influence the peripheral processes that determine the frequency selectivity of the auditory system.

Keywords

frequency selectivity musicianship auditory filter psychophysical tuning curve notched-noise method

Introduction

It is commonly assumed that the frequency selectivity of the auditory system is primarily determined by the filtering that takes place in the cochlea. Consistent with this, behavioral estimates of frequency selectivity obtained in masking experiments are similar to estimates obtained from the auditory nerve or via otoacoustic emissions (Evans, 2001; Evans, Pratt, & Cooper, 1989; Sumner et al., 2018). If frequency selectivity is determined in the cochlea, one would not expect musical training to influence frequency selectivity, as it seems unlikely that such training would affect the biophysical processes that determine tuning in the cochlea. However, there is controversy about whether or not musicians have sharper auditory filters than nonmusicians.

Soderquist (1970) showed that musicians were better than nonmusicians at hearing out individual components in a complex sound with many sinusoidal components. At first sight, this might be taken as indicating superior frequency selectivity for musicians. However, it might also be a consequence of the generally greater skill of musicians in using a limited amount of sensory evidence, especially in pitch-related tasks (Micheyl, Delhommeau, Perrot, & Oxenham, 2006). Fine and Moore (1993) confirmed the finding of Soderquist (1970) that musicians were better than nonmusicians at hearing out individual components in a complex sound. However, they also used the notched-noise method (Patterson, 1976) to measure auditory filter shapes at center frequency of 460, 1000, and 2050 Hz and found no difference in auditory filter bandwidth between musicians and nonmusicians. Consistent with this, Oxenham, Fligor, Mason, and Kidd (2003) also used the notched-noise method to measure frequency selectivity at 1000 Hz and found no difference in auditory filter bandwidth between musicians and nonmusicians. Fine and Moore concluded that the superior ability of musicians to hear out components in a complex sound was related to better “listening proficiency” rather than better frequency selectivity.

More recently, it has been reported that psychophysical tuning curves (PTCs) estimated using a fast method with a sweeping masker (Sek & Moore, 2011) at a center frequency of 4000 Hz were sharper for musicians than for nonmusicians (Bidelman, Nelms, & Bhagat, 2016; Bidelman, Schug, Jennings, & Bhagat, 2014). If this truly reflects a difference in frequency selectivity, it implies either that frequency selectivity measured behaviorally is not determined entirely in the cochlea or that the tuning of the cochlea can be modified as a result of musical training, perhaps via efferent feedback to the cochlea through the medial olivocochlear (MOC) reflex, which regulates the operation of the outer hair cells and can affect cochlear tuning (Bidelman et al., 2016; Guinan, 2006). Consistent with the latter interpretation, Bidelman et al. (2016) reported that cochlear tuning estimated from otoacoustic emissions was sharper for musicians than for nonmusicians.

There are several differences between the research of Fine and Moore (1993) and Oxenham et al. (2003) and that of Bidelman et al. (2014, 2016). One is the signal frequency, which was 2050 Hz or below for the former and 4000 Hz for the latter. Indeed, Bidelman and coworkers did not find a difference between musicians and nonmusicians in the sharpness of fast PTCs for a center frequency of 1000 Hz, supporting a role of signal frequency. However, it is unclear why musicianship should affect frequency selectivity at 4000 Hz but not at lower frequencies.

Another difference between studies is in the methods used: notched noise versus fast PTCs. PTCs obtained in simultaneous masking can be influenced by the detection of beats between the signal and masker (Alcántara, Moore & Vickers, 2000; Kluk & Moore, 2004; Moore, Alcántara, & Dau, 1998). Beats provide a salient detection cue for masker frequencies close to the signal frequency but not for large masker–signal frequency separations (Kohlrausch, Fassel, & Dau, 2000), and this can artificially sharpen the tips of the PTCs, even when a narrowband noise masker is used (Kluk & Moore, 2004). Musicians may be more adept at detecting beats than nonmusicians, and this may account for the sharper PTCs of the former. Beats are not audible for a tone presented in notched noise, and this could explain why Fine and Moore (1993) did not find a difference between musicians and nonmusicians in the bandwidths of auditory filters estimated using the notched-noise method. However, beat detection cannot account for the finding of Bidelman et al. (2014) that PTCs obtained using forward masking were sharper for musicians than for nonmusicians for a center frequency of 4000 Hz (but not 1000 Hz).

PTCs can also be influenced by “off-frequency listening,” which refers to the use of auditory filters centered away from the signal frequency in order to detect the signal. It is assumed that, when detecting a signal in noise, the subject makes use of the output of the auditory filter that has the highest signal-to-masker ratio (Fletcher, 1940; Patterson & Moore, 1986), and this filter is not necessarily centered at the signal frequency (Patterson & Nimmo-Smith, 1980). Off-frequency listening leads to sharper PTCs than would occur if subjects always made use of the output of the auditory filter centered at the signal frequency (Johnson-Davies & Patterson, 1979; O’Loughlin & Moore, 1981; Patterson & Moore, 1986). Musicians may be more adept at off-frequency listening than nonmusicians; this is a second factor that may account for the sharper PTCs of the former. In contrast, the notched-noise method limits off-frequency listening (because the “best” auditory filter to use is always centered close to the signal frequency), and off-frequency listening can be taken into account when analyzing the results (Glasberg & Moore, 1990; Patterson & Nimmo-Smith, 1980; Patterson, Nimmo-Smith, Weber, & Milroy, 1982).

Finally, during the measurement of PTCs, especially using the fast method, musicians may be better at selecting the optimal detection cues and auditory filter center frequencies for any specific combination of masker and signal frequency, and this could affect the shapes of the PTCs, especially around their tips, where the detection cues and the optimal auditory filter center frequencies change rapidly with masker frequency.

The objective of this study was to provide a more controlled and comprehensive comparison of the frequency selectivity of musicians and nonmusicians using a method for which beat detection and off-frequency listening play no role or a minor role (notched noise) and using two methods (traditional PTCs and fast PTCs, both determined in simultaneous masking) for which beat detection and off-frequency listening probably play some role. Traditional PTCs are defined here as PTCs determined using several fixed masker frequencies (Chistovich, 1957; Moore, 1978; Small, 1959). These were included as the detection cues remain stable for any single threshold run (a specific combination of signal frequency and masker frequency), so there is no need to rapidly select the optimal detection cues. A signal frequency of 4000 Hz was used to maximize the chance of finding a difference between musicians and nonmusicians based on the work of Bidelman et al. (2014, 2016).

Method

Subjects

Thirty five young adults participated in the experiments, 20 females and 15 males, although not all subjects were tested with all three methods. Their ages ranged between 18 and 30 years with an average of 22 years (standard deviation, SD = 4 years). There were 17 musicians (average age of 22 years) and 16 nonmusicians (average age of 22 years). Musicians had received more than 5 years of continuous instruction on an instrument or voice, began instruction prior to the age of 12 years, and were currently active in music practice. Nonmusicians had no more than 2 years of self-directed musical training and had no instruction or teaching in music in the last 5 years. They did not regularly perform music at the time of participation in the experiment. All subjects had no known hearing problems. The criteria for selecting the two groups were similar to those used by Bidelman et al. (2014, 2016)

Subjects were paid for taking part and signed a consent form before the experiment commenced. The study was approved by the Research Ethics Committee of the Department of Psychology of the University of Cambridge.

General Procedure

All testing took place in a double-walled sound-attenuating chamber. Each subject was screened to ensure normal hearing in both ears using a Grason-Stadler audiometer (GSI-61) equipped with Telephonics headphones (TDH 50P) and a Radioear bone vibrator (B-71). All had audiometric thresholds below or equal to 20 dB hearing level for octave frequencies between 125 and 8000 Hz with no air-bone gaps greater than 10 dB. The audiometric thresholds did not differ significantly for the musicians and nonmusicians at any test frequency. Absolute thresholds in dB sound pressure level (SPL) at the test frequency of 4000 Hz used in the main experiment were also measured using the fast-PTC software (Sek & Moore, 2011). Only the left ear of each participant was tested.

After measurement of absolute thresholds, the shapes of the auditory filters were estimated using the three methods, which were tested in an order that was selected randomly for each subject. The methods were fast PTCs, traditional PTCs determined in simultaneous masking, and the notched-noise method. Subjects faced a screen while being presented with stimuli via Sennheiser HD580 headphones. These have a frequency response at the eardrum that approximately matches the diffuse-field-to-eardrum transfer function (Killion, Berger, & Nuss, 1987; Moore, 2012). Response choices were made with a keyboard and mouse. Stimuli were generated digitally via an external 16-bit resolution soundcard (M-Audio Delta) and fed to the headphones via a Hatfield manual attenuator Type 2125.

For all measurements involving forced-choice procedures (absolute thresholds, notched noise, and traditional PTCs), the observation intervals lasted 500 ms and were separated by 300 ms. The signal duration was 300 ms, including 20-ms raised-cosine ramps. The correct interval was identified after each trial by flashing a box on the screen. A three-down, one-up decision rule was used, which tracks the 79.4% correct point on the psychometric function (Levitt, 1971). The step size was 4 dB until two turnpoints had occurred and was 2 dB thereafter. Ten turnpoints were obtained, and the threshold was estimated as the average level at the last eight. Two estimates were obtained, and the final estimate was obtained by averaging these two.

Determination of Absolute Threshold

The absolute threshold at 4000 Hz was determined initially from the audiogram and then measured more precisely using the fast-PTC software package (Sek & Moore, 2011). A three-alternative forced-choice procedure was used, with intervals when the signal might occur indicated on the screen.

Notched-Noise Method

The abbreviated notched-noise method described by Stone, Glasberg, and Moore (1992) was used here. The noise masker consisted of two bands with a spectral notch between them, and the signal frequency f_s (4000 Hz) fell in the notch. The overall level of each band was fixed at 67 dB SPL, giving a level of 70 dB SPL in total, and the signal level was varied to estimate the detection threshold. All noise bands had a width of 0.4f_s, and there were eight notch widths, four symmetrical and four asymmetrical about f_s. Notch widths are expressed as the deviation of the lower and upper edges of the notch from f_s. The notch widths were 0.0f_s and 0.0f_s, 0.1f_s and 0.1f_s, 0.2f_s and 0.2f_s, 0.4f_s and 0.4f_s, 0.1f_s and 0.3f_s, 0.3f_s and 0.1f_s, 0.2f_s and 0.4f_s, and 0.4f_s and 0.2f_s, and they were tested in that order. Practice was given using condition 0.0f_s and 0.0f_s and condition 0.4f_s and 0.4f_s before the formal experiment. A two-alternative forced-choice procedure was used.

Traditional PTCs

Traditional PTCs were measured using a 4000-Hz signal whose level was fixed at 15 dB sensation level (SL). The masker level was varied to determine the detection threshold for each masker center frequency. The masker bandwidth was 300 Hz, as recommended by Kluk and Moore (2004). This reduces the salience of beats as a cue but probably does not eliminate the influence of beat detection. Seven fixed center frequencies of the noise were used, one at a time: 3005, 3306, 3636, 4000, 4400, 4840, and 5320 Hz. Subjects were tested at 3005 Hz and 3306 Hz once for training, and the formal experiment was done following a sequence from lower frequencies to higher frequencies. A two-alternative forced-choice procedure was used.

Fast PTCs

The level of the 4000-Hz signal was set to 15 dB SL. The signal was pulsed on and off in a regular sequence (500-ms on, 300-ms off), while the noise was continuous and was swept in center frequency from 3005 to 5320 Hz (upward sweep) or 5320 to 3005 Hz (downward sweep). Subjects were asked to press and hold the space bar when they heard the signal and to release it when they did not. The masker level was increased by 2 dB/s while the space bar was pressed, and the masker level was decreased by 2 dB/s while the space bar was released. At the start of each run, the signal was played for 3 s without any masker so that the subject “knew what to listen for.” The initial masker level was set 30 to 40 dB above the signal level. Each sweep took 4 min. Participants were trained using one upward and one downward sweep. If the PTC appeared very erratic, up to two additional runs were used. However, even after training, the PTCs appeared erratic for some subjects. In the main experiment, an upward and a downward sweep were used in random order.

Data Analysis

The spectra of all stimuli were adjusted to allow for the frequency response of the Sennheiser HD580 headphone and the response of the middle ear so that levels were expressed as the effective level at the cochlea. This allows the shape of the auditory filter to be estimated more accurately (Glasberg & Moore, 1990; Moore, Glasberg, & Baer, 1997). The frequency response of the headphone was estimated using an acoustic manikin, Knowles Electronic Manikin for Acoustic Research (Burkhard & Sachs, 1975), using the average of responses measured with the large and small pinnae. The assumed transfer function of the middle ear was the same as described by Glasberg and Moore (2006).

The data obtained using all three methods were analyzed assuming that each side of the auditory filter took the form of a rounded-exponential function (Patterson et al., 1982). The lower side of the filter was characterized as

W (g) = (1 - w) (1 + p_{l} g) exp (- p_{l} g) + w (1 + t_{l} g) exp (- t_{l} g)

(1)

and the upper side was characterized as

W (g) = (1 + p_{u} g) exp (- p_{u} g)

(2)

where g is the deviation from the center frequency of the filter divided by the center frequency, p_l and p_u are parameters characterizing the lower and upper slopes of the main passband of the filter, respectively, t_l characterizes the slope of the shallower “tail” of the auditory filter on the low-frequency side, and w determines the transition between the main passband and the tail.

For the notched-noise data, the best-fitting values of p_l, p_u, t_l, and w for each subject were determined using the method described by Glasberg and Moore (1990). Starting values of the parameters were chosen, and the filter shape defined by Equations 1 and 2 was used to predict the threshold for each notch width based on the assumptions of the power-spectrum model of masking (Patterson & Moore, 1986). The parameter values were then iteratively adjusted so as to minimize the root-mean-square difference between the obtained and predicted values. Off-frequency listening was taken into account by allowing the center frequency of the auditory filter to shift over the range 0.85f_s to 1.15f_s to find the center frequency at which the signal-to-noise ratio at the output of the filter was highest for each notch width. To ensure “reasonable” results, the passband slope parameters were restricted in value (0 ≤ p_l, p_u ≤ 50) and asymmetry (0.1 ≤ p_u/p_l ≤ 10). The analysis yielded estimates of the values of the four parameters characterizing the auditory filter and also an estimate of parameter k, the signal-to-masker ratio at the output of the auditory filter required to reach the threshold value. This is a measure of detection efficiency (Patterson & Moore, 1986). Detection efficiency refers to the ability to make use of a given amount of sensory evidence. For example, the signal may be detected as an increase in the level of the output of the auditory filter. Some subjects require a smaller increase in level than others, and this results in a smaller value of k, that is, greater detection efficiency. We hypothesized that the values of k would be lower (better) for musicians than for nonmusicians because of the generally greater auditory proficiency of the former.

The data obtained using the fast-PTC method were jagged in form, reflecting the continually changing masker level (Sek & Moore, 2011). These data were initially smoothed using a two-point moving average. Following this, the data for the fast PTCs and the traditional PTCs were analyzed in the same way. The method was similar to that used to analyze the notched-noise data, except that the rounded-exponential filter model was used to predict the masker level required just to mask the signal for each masker center frequency. For the PTC data, off-frequency listening was not taken into account, as the extent of off-frequency listening depends on the exact form of the absolute threshold as a function of frequency in the region around f_s, and this was not known. Also, we wanted our analyses to be comparable to those of Bidelman et al. (2014, 2016), and they did not take off-frequency listening into account.

The overall sharpness of the auditory filters was characterized by the Q₁₀ value, which is the center frequency divided by the bandwidth measured 10 dB away from the tip. For the fast PTCs, Q₁₀ was estimated separately for the upward masker sweep and the downward masker sweep for each subject, and then the Q₁₀ values were averaged. Bidelman et al. (2014, 2016) also used Q₁₀ to quantify the sharpness of the PTCs. For all three methods, the values of w were small. This allowed the Q₁₀ values to be calculated approximately (Hartmann, 1997; Patterson et al., 1982) as follows:

Q_{10} = p_{l} \times p_{u} / [3.9 \times (p_{l} + p_{u})]

(3)

Results

Fast PTCs

Some subjects did not produce consistent fast PTCs or did not produce PTCs with a well-defined tip. In a few cases, this was due to a programming error that led to the signal being set to a level lower than 15 dB SL. In other cases, the subjects simply gave erratic results, possibly because they sometimes “forgot what to listen for.” Consistent PTCs, with similar shapes for upward and downward frequency sweeps, were obtained for 11 nonmusicians and 12 musicians, and the data were analyzed only for these subjects. Figure 1 shows histograms of the Q₁₀ values for the two groups. The histograms were very similar for the two groups. Mean Q₁₀ values were 5.3 (SD = 1.0) for the nonmusicians and 5.5 (SD = 1.1) for the musicians. As the Q₁₀ values were not normally distributed, the two groups were compared using the nonparametric Mann–Whitney U test. The difference between the groups was not significant (U = 56.5, z = .55, p = .58).

Figure 1.

Histogram of Q₁₀ values estimated from the fast PTCs for the nonmusicians (top) and the musicians (bottom). PTC = psychophysical tuning curve.

The Q₁₀ values are similar in overall magnitude to those for the fast PTCs obtained by Bidelman et al. (2014), except that they found a slightly lower mean Q₁₀ for 9 nonmusicians (about 4.5) and a slightly higher mean value for 10 musicians (about 6). However, Bidelman et al. (2016) obtained markedly higher Q₁₀ values for 14 musicians (mean ≈ 10), while the Q₁₀ values for 13 nonmusicians (mean ≈ 6) were similar to those found here.

Traditional PTCs

Due to limited time availability, two nonmusicians and one musician did not complete testing for the traditional PTCs. Hence, data were available for 15 nonmusicians and 17 musicians. Figure 2 shows histograms of the Q₁₀ values for the two groups. The histograms were very similar for the two groups. Mean Q₁₀ values were 5.9 (SD = 0.86) for the nonmusicians and 5.6 (SD = 1.15) for the musicians. Based on a Mann–Whitney test, the Q₁₀ values for the two groups did not differ significantly (U = 113.5, z = −.51, p = .61).

Figure 2.

As Figure 1 but for Q₁₀ values estimated from the traditional PTCs. PTC = psychophysical tuning curve.

Notched-Noise Method

Notched-noise data were available for all subjects. Figure 3 shows histograms of the Q₁₀ values for the two groups. The histograms were very similar for the two groups. Mean Q₁₀ values were 4.6 (SD = 0.78) for the nonmusicians and 4.5 (SD = 0.76) for the musicians. Based on a Mann–Whitney test, the Q₁₀ values for the two groups did not differ significantly (U = 138.5, z = −.46, p = .65).

Figure 3.

As Figure 1 but for Q₁₀ values estimated from the notched-noise data.

Figure 4 shows histograms of k values for the two groups. Although there was considerable overlap of k values for the two groups, the mean value was lower (better) for the musicians (mean = 1.2 dB, SD = 3.0 dB) than for the nonmusicians (mean = 3 dB, SD = 3.9 dB). The k values were approximately normally distributed. The significance of the difference between groups was assessed using a one-tailed nonmatched samples t test, as we hypothesized that the musicians would have lower k values than the nonmusicians. The outcome was significant: t = −1.49, p = .037.

Figure 4.

Histogram of k values (a measure of processing efficiency) estimated from the notched-noise data for the nonmusicians (top) and the musicians (bottom).

Discussion

The results for the fast PTCs differ from those obtained by Bidelman et al. (2014, 2016), despite the fact that the methods used were very similar. One difference across studies is that Bidelman et al. used only upward frequency sweeps, while we used both upward and downward sweeps. When the masker center frequency is swept upward, the masker level required for threshold usually increases rapidly when the masker frequency is just above the signal frequency. In this region, subjects need to press and release the space bar rapidly in order to “track” the threshold correctly. Musicians may be more adept at this than nonmusicians. When we analyzed the Q₁₀ values only for the upward sweeps, the results showed a similar trend to those of Bidelman et al. (2014, 2016). The mean Q₁₀ values were 5.9 for the musicians and 5.3 for the nonmusicians, compared with values of about 6 and 4.5, respectively, obtained by Bidelman et al. (2014), and 10 and 6, respectively, obtained by Bidelman et al. (2016). However, the difference in our data was not significant. It is noteworthy that the Q₁₀ values for our fast PTCs never exceeded 7 for either group.

Another difference across studies is that we fitted the data objectively using the rounded-exponential filter model. Bidelman et al. (2014, 2016) did not specify exactly how Q₁₀ values were estimated from their fast PTCs, but presumably this was done “by eye.” Finally, we tested only the left ear of each subject, while Bidelman et al. tested only the right ear. However, we are not aware of any data showing differences in frequency selectivity across ears for young normal-hearing subjects.

Bidelman et al. (2014) also estimated PTCs in forward masking using a signal frequency of 4000 Hz. The PTCs obtained in this way were significantly sharper for musicians than for nonmusicians. The shapes of PTCs in forward masking depend on the ability to use the available detection cues. When the signal frequency is equal to the masker center frequency, the signal may be confused with the masker, as it sounds like a continuation of the masker (Moore, 1980; Moore & Glasberg, 1985; Neff, 1986; Terry & Moore, 1977). When the signal frequency differs from the masker center frequency, subjects report listening for a small change in pitch at the end of the masker (Moore, 1981). Musicians may be more sensitive to this pitch cue than nonmusicians, and this could account for the sharper PTCs of the musicians found by Bidelman et al. (2014). Thus, the difference across groups may not reflect a difference in frequency selectivity but instead may reflect a greater ability of musicians to use pitch cues. A weakness of this explanation is that it does not account for the fact that Bidelman et al. found sharper forward-masking PTCs for musicians than for nonmusicians only at 4000 Hz and not at 1000 Hz. If the sharper PTCs for musicians were a result of a better ability to use pitch cues, one would expect a difference at both center frequencies.

Bidelman et al. (2016) estimated physiological tuning curves by measuring the suppression of stimulus-frequency otoacoustic emissions, which are assumed to reflect the tuning characteristics of the cochlea (Charaziak, Souza, & Siegel, 2013). For a frequency of 4000 Hz, these also showed significantly sharper tuning for musicians than for nonmusicians. Clearly, this finding cannot be explained in terms of group differences in the ability to use the available detection cues. Bidelman et al. (2016) suggested that the effect of musical training might be a consequence of changes in cochlear mechanics, such as changes in the regulation of the motor protein prestin (Liberman et al., 2002). They also suggested that “sharper cochlear tuning could originate from training-related efferent feedback through the medial olivocochlear (MOC) fibers.” Consistent with this, it has been shown that musical training is associated with strengthening of the ipsilateral and contralateral MOC efferent system (Bidelman, Schneider, Heitzmann, & Bhagat, 2017). However, a problem with this explanation is that activation of the efferent system by stimulation of the ear opposite to the test ear tends to reduce the sharpness of tuning measured using PTCs for high center frequencies, rather than to increase it (Vinay & Moore, 2008; Wicher & Moore, 2014).

None of the three methods used to estimate frequency selectivity in this study showed a significant difference in Q₁₀ values between musicians and nonmusicians. The results obtained using the notched-noise method are consistent with those of Fine and Moore (1993) who found no significant difference between musicians and nonmusicians in the sharpness of the auditory filters for center frequencies of 460, 1000, and 2050 Hz and of Oxenham et al. (2003) who found no difference across groups at 1000 Hz. We consider the notched-noise method to be the most valid of the methods used here to estimate frequency selectivity, as the results are minimally if at all affected by the detection of beats between the signal and masker and off-frequency listening is taken into account in the analysis of the results. Furthermore, the detection cues used and the “optimum” auditory filter center frequency to use remain stable throughout a threshold run for a given notch width. Hence, the results obtained using the notched-noise method lead us to conclude that frequency selectivity as measured in simultaneous masking does not differ for musicians and nonmusicians.

The Q₁₀ values obtained using the notched-noise method were somewhat smaller, indicating less sharp tuning, than obtained for the PTCs. This is probably a consequence of the results for the fast and traditional PTCs being influenced by beat detection and off-frequency listening both of which tend to lead to sharper PTCs (Johnson-Davies & Patterson, 1979; Kluk & Moore, 2004; O’Loughlin & Moore, 1981). As noted earlier, the results for the notched-noise method are probably minimally influenced by beat detection, and off-frequency listening was taken into account in the analysis of the results.

Our results showed that detection efficiency, as quantified by the parameter k estimated from the notched-noise data, was significantly better for musicians than for nonmusicians. This may reflect a greater ability of musicians to make use of the available detection cues. When detecting a tone in notched noise, subjects usually report that they listen out for a pitch corresponding to the signal frequency. This pitch might be based on a local peak in the excitation pattern evoked by the tone-plus-noise, increased neural synchrony to the signal frequency, or by reduced envelope fluctuations at the outputs of auditory filters centered close to the signal frequency (Mao, Vosoughi, & Carney, 2013; Moore, 1975). Whatever the neural code involved, it may be the case that musicians are more sensitive than nonmusicians to weak pitch cues, and that this underlies the greater detection efficiency found for musicians. Musicians may also be better than nonmusicians at remembering the weak pitch that they are listening for.

Summary and Conclusions

Frequency selectivity was compared for musicians and nonmusicians using a signal frequency of 4000 Hz for three methods: fast PTCs, traditional PTCs, and the notched-noise method. The methods differed in the extent to which the results might be influenced by factors such as the detection of beats and off-frequency listening. Frequency selectivity, quantified by the measure Q₁₀, did not differ significantly across musicians and nonmusicians for any of the three methods, although detection efficiency measured using the notched-noise method was better for musicians than for nonmusicians. The results for the fast PTCs are not consistent with those reported by Bidelman et al. (2014, 2016). Overall, the present results indicate that frequency selectivity does not differ for musicians and nonmusicians.

Footnotes

Acknowledgments

The authors thank Andrew Oxenham and two reviewers for helpful comments on an earlier version of this article.

Declaration of Conflicting Interests

The authors declared no potential conflict of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Engineering and Physical Sciences Research Council (UK, grant number RG78536).

References

Alcántara

J. I.

Moore

B. C. J.

Vickers

D. A.

(2000) The relative role of beats and combination tones in determining the shapes of masking patterns at 2 kHz: I. Normal-hearing listeners. Hearing Research 148: 63–73. doi:10.1016/S0378-5955(00)00114-3.

Bidelman

G. M.

Nelms

Bhagat

S. P.

(2016) Musical experience sharpens human cochlear tuning. Hearing Research 335: 40–46. doi:10.1016/j.heares.2016.02.012.

Bidelman

G. M.

Schneider

A. D.

Heitzmann

V. R.

Bhagat

S. P.

(2017) Musicianship enhances ipsilateral and contralateral efferent gain control to the cochlea. Hearing Research 344: 275–283. doi:10.1016/j.heares.2016.12.001.

Bidelman

G. M.

Schug

J. M.

Jennings

S. G.

Bhagat

S. P.

(2014) Psychophysical auditory filter estimates reveal sharper cochlear tuning in musicians. Journal of the Acoustical Society of America 136: EL33–EL39. doi:10.1121/1.4885484.

Burkhard

M. D.

Sachs

R. M.

(1975) Anthropometric manikin for acoustic research. Journal of the Acoustical Society of America 58: 214–222. doi:10.1121/1.380648.

Charaziak

K. K.

Souza

Siegel

J. H.

(2013) Stimulus-frequency otoacoustic emission suppression tuning in humans: Comparison to behavioral tuning. Journal of the Association for Research in Otolaryngology 14: 843–862. doi:10.1007/s10162-013-0412-1.

Chistovich

L. A.

(1957) Frequency characteristics of masking effect. Biofizika 2: 743–755.

Evans

E. F.

(2001) Latest comparisons between physiological and behavioural frequency selectivity. In: Breebaart

D. J.

Houtsma

A. J. M.

Kohlrausch

Prijs

V. F.

Schoonhoven

(eds) Physiological and psychophysical bases of auditory function, Maastricht, the Netherlands: Shaker, pp. 382–387.

Evans

E. F.

Pratt

S. R.

Cooper

N. P.

(1989) Correspondence between behavioural and physiological frequency selectivity in the guinea pig. British Journal of Audiology 23: 151–152.

10.

Fine

P. A.

Moore

B. C. J.

(1993) Frequency analysis and musical ability. Music Perception 11: 39–53. doi:10.2307/40285598.

11.

Fletcher

(1940) Auditory patterns. Reviews of Modern Physics 12: 47–65.

12.

Glasberg

B. R.

Moore

B. C. J.

(1990) Derivation of auditory filter shapes from notched-noise data. Hearing Research 47: 103–138. doi:10.1016/0378-5955(90)90170-T.

13.

Glasberg

B. R.

Moore

B. C. J.

(2006) Prediction of absolute thresholds and equal-loudness contours using a modified loudness model. Journal of the Acoustical Society of America 120: 585–588. doi:10.1121/1.2214151.

14.

Guinan

J. J.

Jr. (2006) Olivocochlear efferents: Anatomy, physiology, function, and the measurement of efferent effects in humans. Ear and Hearing 27: 589–607. doi:10.1097/01.aud.0000240507.83072.e7.

15.

Hartmann

W. M.

(1997) Signals, sound, and sensation, Woodbury, NY: AIP Press.

16.

Johnson-Davies

Patterson

R. D.

(1979) Psychophysical tuning curves: Restricting the listening band to the signal region. Journal of the Acoustical Society of America 65: 765–770. doi:10.1121/1.382490.

17.

Killion

M. C.

Berger

E. H.

Nuss

R. A.

(1987) Diffuse field response of the ear. Journal of the Acoustical Society of America 81: S75 doi:10.1121/1.2024388.

18.

Kluk

Moore

B. C. J.

(2004) Factors affecting psychophysical tuning curves for normally hearing subjects. Hearing Research 194: 118–134. doi:10.1016/j.heares.2004.04.012.

19.

Kohlrausch

Fassel

Dau

(2000) The influence of carrier level and frequency on modulation and beat-detection thresholds for sinusoidal carriers. Journal of the Acoustical Society of America 108: 723–734. doi:10.1121/1.429605.

20.

Levitt

(1971) Transformed up-down methods in psychoacoustics. Journal of the Acoustical Society of America 49: 467–477. doi:10.1121/1.1912375.

21.

Liberman

M. C.

Gao

D. Z.

Jia

Zuo

(2002) Prestin is required for electromotility of the outer hair cell and for the cochlear amplifier. Nature 419: 300–304. doi:10.1038/nature01059.

22.

Mao

Vosoughi

Carney

L. H.

(2013) Predictions of diotic tone-in-noise detection based on a nonlinear optimal combination of energy, envelope, and fine-structure cues. Journal of the Acoustical Society of America 134: 396–406. doi:10.1121/1.4807815.

23.

Micheyl

Delhommeau

Perrot

Oxenham

A. J.

(2006) Influence of musical and psychoacoustical training on pitch discrimination. Hearing Research 219: 36–47. doi:10.1016/j.heares.2006.05.004.

24.

Moore

B. C. J.

(1975) Mechanisms of masking. Journal of the Acoustical Society of America 57: 391–399. doi:10.1121/1.380454.

25.

Moore

B. C. J.

(1978) Psychophysical tuning curves measured in simultaneous and forward masking. Journal of the Acoustical Society of America 63: 524–532. doi:10.1121/1.381752.

26.

Moore

B. C. J.

(1980) Detection cues in forward masking. In: van den Brink

Bilson

F. A.

(eds) Psychophysical, Physiological and Behavioural Studies in Hearing, Delft, the Netherlands: Delft University Press, pp. 222–229.

27.

Moore

B. C. J.

(1981) Relation between pitch shifts and MMF shifts in forward masking. Journal of the Acoustical Society of America 69: 594–596. doi:10.1121/1.385435.

28.

Moore

B. C. J.

(2012) An introduction to the psychology of hearing, 6th ed. Leiden, the Netherlands: Brill.

29.

Moore

B. C. J.

Alcántara

J. I.

Dau

(1998) Masking patterns for sinusoidal and narrowband noise maskers. Journal of the Acoustical Society of America 104: 1023–1038. doi:10.1121/1.423321.

30.

Moore

B. C. J.

Glasberg

B. R.

(1985) The danger of using narrowband noise maskers to measure suppression. Journal of the Acoustical Society of America 77: 2137–2141. doi:10.1121/1.391738.

31.

Moore

B. C. J.

Glasberg

B. R.

Baer

(1997) A model for the prediction of thresholds, loudness and partial loudness. Journal of the Audio Engineering Society 45: 224–240.

32.

Neff

D. L.

(1986) Confusion effects with sinusoidal and narrowband-noise forward maskers. Journal of the Acoustical Society of America 79: 1519–1529. doi:10.1121/1.393678.

33.

O’Loughlin

B. J.

Moore

B. C. J.

(1981) Off-frequency listening: Effects on psychoacoustical tuning curves obtained in simultaneous and forward masking. Journal of the Acoustical Society of America 69: 1119–1125. doi:10.1121/1.385691.

34.

Oxenham

A. J.

Fligor

B. J.

Mason

C. R.

Kidd

Jr. (2003) Informational masking and musical training. Journal of the Acoustical Society of America 114: 1543–1549. doi:10.1121/1.1598197.

35.

Patterson

R. D.

(1976) Auditory filter shapes derived with noise stimuli. Journal of the Acoustical Society of America 59: 640–654. doi:10.1121/1.380914.

36.

Patterson

R. D.

Moore

B. C. J.

(1986) Auditory filters and excitation patterns as representations of frequency resolution. In: Moore

B. C. J.

(eds) Frequency Selectivity in Hearing, London, England: Academic, pp. 123–177.

37.

Patterson

R. D.

Nimmo-Smith

(1980) Off-frequency listening and auditory filter asymmetry. Journal of the Acoustical Society of America 67: 229–245. doi:10.1121/1.383732.

38.

Patterson

R. D.

Nimmo-Smith

Weber

D. L.

Milroy

(1982) The deterioration of hearing with age: Frequency selectivity, the critical ratio, the audiogram, and speech threshold. Journal of the Acoustical Society of America 72: 1788–1803. doi:10.1121/1.388652.

39.

Sek

Moore

B. C. J.

(2011) Implementation of a fast method for measuring psychophysical tuning curves. International Journal of Audiology 50: 237–242. doi:10.3109/14992027.2010.550636.

40.

Small

A. M.

(1959) Pure-tone masking. Journal of the Acoustical Society of America 31: 1619–1625. doi:10.1121/1.1936092.

41.

Soderquist

D. R.

(1970) Frequency analysis and the critical band. Psychonomic Science 21: 117–119. doi:10.3758/BF03335794.

42.

Stone

M. A.

Glasberg

B. R.

Moore

B. C. J.

(1992) Simplified measurement of impaired auditory filter shapes using the notched-noise method. British Journal of Audiology 26: 329–334. doi:10.3109/03005369209076655.

43.

Sumner

C. J.

Wells

T. T.

Bergevin

Sollini

Kreft

H. A.

Palmer

A. R.

Shera

C. A.

(2018) Mammalian behavior and physiology converge to confirm sharper cochlear tuning in humans. Proceedings of the National Academy of Sciences 115: 11322–11326. doi:10.1073/pnas.1810766115.

44.

Terry

Moore

B. C. J.

(1977) ‘Suppression’ effects in forward masking. Journal of the Acoustical Society of America 62: 781–784. doi:10.1121/1.381553.

45.

Vinay Moore

B. C. J.

(2008) Effects of activation of the efferent system on psychophysical tuning curves as a function of signal frequency. Hearing Research 240: 93–101. doi:10.1016/j.heares.2008.03.002.

46.

Wicher

Moore

B. C. J.

(2014) Effect of broadband and narrowband contralateral noise on psychophysical tuning curves and otoacoustic emissions. Journal of the Acoustical Society of America 135: 2931–2941. doi:10.1121/1.4871358.