British Journal of Audiology ISSN: 0300-5364 (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/ijap20 Measurement of the Temporal-Modulation Transfer Function for a Single Listener with Cochlear Hearing Loss and Left-Hemisphere Damage F. Hescot, C. Lorenzi, X. Debruille & J-F. Camus To cite this article: F. Hescot, C. Lorenzi, X. Debruille & J-F. Camus (2000) Measurement of the Temporal-Modulation Transfer Function for a Single Listener with Cochlear Hearing Loss and Left-Hemisphere Damage, British Journal of Audiology, 34:6, 341-351, DOI: 10.3109/03005364000000150 To link to this article: http://dx.doi.org/10.3109/03005364000000150 Published online: 23 Mar 2011. Submit your article to this journal Article views: 40 View related articles Citing articles: 3 View citing articles Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ijap20 Download by: [University of California, San Diego] Date: 19 April 2016, At: 21:33 British Journal of Audiology, 2000,34,341-351 Measurement of the temporal-modulation transfer function for a single listener with cochlear hearing loss and left-hemisphere damage Downloaded by [University of California, San Diego] at 21:33 19 April 2016 F. Hescot,' C. Lorenzi,' X. Debmille2 and J-F. Camus' 'Institut de Psychologie, Universitk Rent Descartes, Paris V, France and 2Laboratoire Xavier Debruille, Reims, France (Received 10 October 1999; accepted 15 August2000) Abstract The modulation depth required for the detection of sinusoidal amplitude-modulation applied to a white noise carrier was measured as a function of modulation frequency, giving temporal modulation transfer functions (TMTFs). Five adult listeners with normal hearing (mean age 52 years), five elderly listeners with moderate cochlear hearing loss (mean age 66 years) and a single elderly listener (aged 73 years) with moderate cochlear hearing loss and left-hemisphere damage were tested in the right ear at 50 dB SL. The five elderly listeners were matched in audiogram with the brain-damaged listener. Modulation detection was systematically poorer than normal in the five elderly listeners with cochlear hearing loss. However, their TMTFs were lowpass in shape, as for the five normal-hearing adult listeners. Modulation detection was much poorer in the elderly listener with cochlear hearing loss and left-hemisphere damage compared to the five normal-hearing adults and the five elderly listeners with cochlear hearing loss. Moreover, modulation detection was poorer at 4,64 and 128 Hz than at 8,16 and 32 Hz in the brain-damaged listener, giving his TMTF a bandpass appearance. These results are in agreement with the hypothesis that the main factors limiting the ability to detect changes in the temporal-envelope of sounds are located at a central (retro-cochlear) level of the auditory system rather than at a peripheral (cochlear) level. They also suggest that the TMTF approach may prove useful in distinguishing peripheral and central hearing losses. Key words: amplitude modulation, auditory cortex, auditory temporal acuity, retro-cochlear hearing loss, temporal modulation transfer time (TMTF) Introduction Speech understanding involves supra-threshold auditory processing in the spectral and the temporal domains. Part of the reason some listeners may have difficulty understanding speech in quiet or in the presence of background noise, reverberation or competing talkers may be related to abnormal auditory temporal acuity, that is, poorer-than-normal ability to resolve changes in the temporal envelope of sounds (Festen and Plomp, 1990). Several methods have been developed to characterize auditory temporal acuity (see Viemeister and Plack, 1993; Moore 1995, 1997 for detailed reviews). The methods most commonly used in audiological and neurological studies are: Fusion threshold for a pair of successive clicks. Detection threshold for a temporal gap in a narrowband or broadband sound. Address for corresDondence: C. Lorenzi. Laboratoire de Psychologie Ex@rimentale, UMR CNRS 8581, Institut de These two methods generally give a consistent Psychologie, UniversitC RenC Descarta paris v, 71 A ~ , Edouard Vaillant. 92774 Boulopne-Billancourt CCdex. estimate Of temporal acuity O f about 1-3 ms in France. normal-hearing listeners. T h e same methods - 0300-5364/00/340341+I 0 $03.50/0 0 2000 British Society of Audiology Downloaded by [University of California, San Diego] at 21:33 19 April 2016 342 F . Hescot et a1 show that temporal acuity is similar in normalhearing listeners and listeners Nith cochlear lesions (Glasberg and Moore. '''92), providing that the stimuli used to measme tzmporal acuity are clearly audible and stimuli d,) not show audible intrinsic random fluctuations that could be confused with the temporal ekent to be detected in the temporal task. On the other hand, fusion or gap thresholds measured in patients with cortical or subcortical lesions (Albert and Bear, 1974; Chocholle et al., 1975; Auerbach et al., 1982; Efron et al., 1985; Tanaka et al., 1987; Yaqub et al., 1988; Divenyi and Robinson, 1989; Robin et al., 1990) are generally larger (sometimes considerably so) than normal (5-300 ms). Taken together, these data suggest that the main factors limiting the ability to detect changes in the temporal envelope of sounds are located at a central (retro-cochlear) level of the auditory system rather than at a peripheral (cochlear) level. The two methods described above generally give a consistent estimate of auditory temporal acuity in normal-hearing listeners (1-3 ms). However, fusion and gap thresholds cannot be regarded as measures of temporal acuity p e r se because the ability to distinguish two clicks from a single click depends also on the detection of spectral changes above 10 kHz (Leshowitz, 1971) and the ability to detect a silent interval in a tone or noise is also limited by the amplitude resolution of the auditory system (Buunen and van Valkenburg, 1979).* Consequently, the conclusions of previous studies concerning the effects of cortical and subcortical lesions upon temporal acuity and using fusion or gap thresholds measurements should be treated with caution. A general approach developed by Viemeister (1979) circumvents these methodological limitations. This approach involves measuring temporal modulation transfer functions (TMTFs), that is, the ability of listeners to detect sinusoidal amplitude-modulation applied to a broadband noise, as a function of modulation frequency. TMTFs show that, for normal-hearing listeners, modulation sensitivity is relatively independent of modulation frequency up to approximately 16 Hz and decreases progressively at higher modulation frequencies (Viemeister, 1979;Bacon and Viemeis- ter, 1985). In this modulation detection task, the use of spectral cues is fully precluded since the modulation of broadband noise does not affect its (flat) long-term power spectrum. For low modulation frequencies (below approximately 16 Hz), detection is limited by the amplitude resolution of the auditory system, rather than its temporal resolution. As the modulation frequency increases beyond approximately 16 Hz, temporal resolution starts to have an effect and modulation-detection thresholds increase. In the present experiment, the effects of a cortical lesion on temporal acuity were measured using clearly audible stimuli. The TMTF approach was used in order to avoid the methodological biases of fusion and gap thresholds measurements. TMTFs were measured for a white noise carrier using a single patient with lesions to the left auditory cortex. As this patient was relatively old and showed cochlear hearing loss prior to brain damage, the two following control groups were used: Five adult listeners with normal hearing. Five elderly listeners with matched cochlear hearing loss. The noise carrier was sufficiently wide to avoid audible fluctuations in amplitude. All subjects were tested at identical sensation levels (50 dB SL). Moreover, the five presbycusic listeners were chosen so as to be matched in audiogram with the brain-damaged patient, ensuring that differences between elderly listeners in terms of temporal acuity were not due to differences in the range of audible frequencies in the noise. Matching was based on the listeners' right ear audiometric thresholds, as modulation sensitivity was assessed using the right ear only (that is, the ear contralateral to the cortical lesion in the braindamaged patient). Psychoacoustical investigation of the brain-damaged patient was finally completed by measuring intensity discrimination thresholds, that is his ability to discriminate between two steady white noises presented at different intensities. This task was introduced as a control task of the modulation-detection task. It was performed to: *This argument has been developed by Viemeister and Plack (1993):a gap is equivalent to a decrement in the signal intensity. Buunen and van Valkenburg (1979) used a task that required listeners to detect brief decrements in the intensity of a broadband noise (rather than a period of complete silence, as in the gap detection task). Their results showed that the gap detection threshold was dependent on the depth of the decrement: the smaller the change in intensity, the larger the gap threshold. This illustratesthe important role played by intensity differences and therefore, intensity resolution, in gap detection. Downloaded by [University of California, San Diego] at 21:33 19 April 2016 Measurement of the temporal-modulation transfer function Evaluate the contribution of potentially confounding non-perceptual effects related to stimulus characteristics (e.g. stimulus duration) and the forced choice procedure (e.g. the sequential mode of presentation of auditory stimuli). Evaluate the contribution of amplitude resolution to the modulation-detection thresholds measured at low modulation frequency. Dissociate the ability to perceive steady amplitude changes from the ability to perceive dynamic amplitude changes. Method Listeners JPK is a 73-year-old, right-handed male. He was admitted to the neurological department of the Reims Hospital in February 1994 (at the age of 68 years) after suffering a cerebrovascular accident (CVA) with a sudden right hemiplegia and language disorders. A computed tomography (CT) scan revealed a left-sylvian ischemic lesion. Neurological assessment performed at this time confirmed a total right hemiplegia, and revealed both speech production and comprehension deficits and music perception deficits. JPK 343 underwent speech therapy for five months. In June 1994, JPK had recovered from his music perception deficit, with persistence of speech production and comprehension deficits. Neurological assessment performed at this time indicated a persistence of right leg and arm weaknesses. JPK was finally classified as a Wernicke aphasic. Figure 1shows an MRI scan (axial view) of JPK performed three years after the CVA. T,-weighted images reveal the presence of an extended necrosis of brain tissues in the left sylvian area (superficial and deep parts) involving the cortical and subcortical parts of the posterior temporal lobe including T1 (superior temporal) gyrus. Figure 2 shows a coronal MRI view of JPK, indicating that the left primary auditory area (Heschl’s gyrus) has been severely damaged due to the sylvian infarct. All psychophysical experiments were performed five years after the CVA, and JPK was medically stable during this period. JPK was also diagnosed with bilateral, symmetrical high-frequency cochlear hearing loss four months before his CVA, and was fitted with a behind-the-ear hearing aid on the right ear at this time. Audiometric thresholds measured before Fig. 1. MRIscan ofpatient JPK carried out three years after CVA. The Figure comprises a T,-weighted axial section revealing the presence of a n extended necrosis of brain tissues in the left sylvian area (superficial and deep parts) involving the cortical and subcortical parts of the posterior temporal lobe including T1 (superior temporal) gyrus. F . Hescot et al. Downloaded by [University of California, San Diego] at 21:33 19 April 2016 344 Fig. 2. Coronal M R I view ofpatient JPK (carried out three years after C V A as f o r Figure I ) , indicating that the leftprimary auditory area (Heschl’s gyrus) has been severely damaged due to the sylvian infarct. and after the CVA were identical, indicating that the hearing loss had a purely cochlear origin. Hearing loss was classified as moderate in the right ear according to JPK’s pure tone average (PTA) hearing loss at 250,500,1000,2000 and 4000 HZ(PTA = 43 dB HL). Table 1 gives JPK’s audiometric thresholds measured in the right ear three months before the psychophysical experiments. Two control groups of listeners without brain damage were also tested: five adult listeners (PH, FH, MCH, JEM, JOM; mean age 52 years; standard deviation (SD): 4 years) with normal hearing (pure tone thresholds >20 dB H L between 250 Hz and 4000 Hz) and no history of hearing disorders; five elderly listeners (GA, LG, MDE, BL, MDO; mean age 66 years; SD 3 years) with bilateral, symmetrical high-frequency cochlear hearing loss. As for JPK, the absence of conductive component to their hearing loss was indicated by normal tympanograms and air-bone differences in absolute threshold 4 . 5 dB; hearing Table 1. Characteristics of listeners: age (years) audiometric thresholds for the right ear (dB HL) and diagnosis for hearing-impairedlisteners Frequency (Hz) Listener Age 250 500 1000 2000 PH FH MCH JEM JOM GA LG MDE BL MDO JPK 51 48 58 53 51 67 70 65 67 61 73 20 10 20 10 15 35 25 20 5 35 30 10 10 20 10 10 25 25 30 15 30 15 15 10 20 15 10 40 25 35 35 50 25 15 15 10 10 10 65 50 45 65 60 65 4000 15 15 20 20 15 80 70 75 85 105 80 Diagnosis Normal Normal Normal Normal Normal Presbycusis Presbycusis Presbycusis Presbycusis + noise exposure Presbycusis Presbycusis + cortical lesions Downloaded by [University of California, San Diego] at 21:33 19 April 2016 Measurement of the temporal-modulation transfer function loss in the right ear was classified as moderate according to PTA hearing loss (39 I PTA I 56 dB HL). Table 1 gives ages and audiometric thresholds of these normal-hearing and hearingimpaired listeners in the right ear. Before experimentation, (unaided) speech intelligibility was assessed with the phonetic test of Lafon (Lafon, 1972; lists of 17 French words; each word including three phonemes and all lists being phonetically balanced) in listener JPK and the five elderly listeners with cochlear hearing loss. For each listener, speech intelligibility was assessed in the right ear in the presence of a contralateral white noise presented at 30 dB. All stimuli were delivered via Sennheiser H D 265 earphones. Figure 3 shows individual psychometric functions for these six listeners, that is, the percentage of phonemes correctly identified as a function of speech level. In the phonetic test of Lafon (1972), speech-identification threshold refers to the speech level needed for 50% of phonemes to be correctly identified. In agreement with their diagnosis of cochlear hearing 345 loss, the speech-identification thresholds for the five presbycusic control subjects ranged from 35 dB to 55 dB, and identification performance was nearly perfect at 80 dB. In comparison, JPK’s performance reached a maximum of 30% correct identification at 80 dB, supporting the diagnosis of central auditory disorder in addition to cochlear hearing loss. TMTFs Apparatus and stimuli All stimuli were generated by use of a 16-bit D/A converter at a sampling frequency of 44.1 kHz, and were delivered monaurally via a Sennheiser H D 265 earphone. The white noises carrier was independently generated for every stimulus. Listeners had to detect the presence of sinusoidal amplitude modulation applied to the white noise carrier. On each trial, a standard and a target stimulus were successively presented in random order to the listener. The standard was unmodulated white noise n(t). In the target, the white noise carrier was sinusoidally amplitude-modulated 100 90 80 70 60 50 40 30 20 10 0 0 20 40 60 80 100 120 Speech level (dB re: normal speech-identification threshold) Fig. 3. Unaided speech intelligibility assessed with the phonetic test of J. C. Lafon (Lafon, 1972) in listener JPK (black circles) and the five elderly listeners with cochlear hearing loss (grey symbols). Speech intelligibility was assessed in the right ear, in the presence of a contralateral masking white noise. For each listener, the percentage of phonemes correctly identified is plotted as a function of speech level (in d B re: speech-identification threshold f o r normal-hearing listeners). The performance level corresponding to 50% ofphonemes correctly identified is indicated by the dotted line. 346 F . Hescot et al. at a given modulation frequency. The expression describing the target was: s(t) = c [ l+m sin(2nfn,t ) ]n(t) (1) Downloaded by [University of California, San Diego] at 21:33 19 April 2016 where m is the modulation depth (0 2 m 2 l);fm is the modulation frequency Cf, was 4,8,16,32,64 or 128Hz). The term c is a multiplicative compensation term (Viemeister, 1979) set so that the overall power was the same in all intervals. The expression for c is: The stimuli were presented at a sensation level (SL) of SO dB. To further discourage the use of intensity cues, random variations of +1 dB in 1dB steps were applied independently to the standard and target intensities. Both the standard and target had a 2 s duration, including 25 ms rise/fall times, shaped using a raised-cosine function. The inter-stimulus interval was 1s. Stimuli were lowpass filtered at 20 kHz and delivered monaurally to the listeners’ right ears. Listeners were tested individually in a soundproof booth. Procedure Modulation-detection thresholds were obtained using an adaptive two-interval, two-alternative forced-choice (21,2AFC) procedure that estimates the modulation depth, m, necessary for 70.7% correct detection. The listener’s task was to identify the interval containing the modulation. Control subjects entered their response using a response box. Because of his aphasic disorder and right hemiplegia, JPK was unable to use the response box and to describe verbally what he perceived. He therefore responded by humming and gesturing what he perceived (stationary, fluctuating or buzzing sounds). Then, an experimenter entered JPK’s responses with the response box. Visual feedback about the correct interval was given after each trial. The step sizes and thresholds were based on the modulation depth in decibels (20 log m). The step size of m was initially 4 dB and was reduced to 2 dB after the first two reversals. The mean value of 20 log m at the last 10 reversals in a block of 14 reversals was taken as the threshold estimate for that block. Thresholds presented here are based upon three estimates for each control subject (i.e. normal-hearing and hearing-impaired listeners without brain damage), and upon two estimates for JPK (data collection was extremely slow with JPK (one threshold estimate/week) and JPK complained increasingly about the difficulty of the task). The worst threshold that could be measured is 0 dB, which corresponds to a modulation depth of 1 (100% modulated noise). The more negative the value of 20 log m , the better the detection threshold. Intensity discrimination thresholds Apparatus and stimuli Again, all stimuli were generated by use of a 16bit D/A converter at a sampling frequency of 44.1 kHz, and were delivered monaurally via a Sennheiser H D 26.5 earphone. The white noise carrier was independently generated for every stimulus. O n each trial, a standard and a target stimulus were successively presented in random order to JPK. The standard consisted of a white noise presented at a reference intensity (I) giving a SL of 50 dB. The target consisted of another white noise presented at a greater intensity (I + AI). Both the standard and target had a 2 s duration, including 25 ms rise/fall times, shaped using a raised-cosine function. The interstimulus interval was 1s. Stimuli were lowpass filtered at 20 kHz and delivered monaurally to JPK’s right ear. Procedure Intensity difference limens were obtained using a 21,2AFC procedure that estimates the intensity difference (AI) necessary for 70.7% correct detection. JPK’s task was to identify the loudest sound interval. As for the modulation-detection task, JPK responded by gesturing what he perceived. Then, an experimenter entered JPK’s responses with a response box. Visual feedback concerning the correct interval was given after each trial. The step sizes and thresholds (AL) were calculated in decibels (AL = 10 log ((I + AI)/I]). The step size of variation was initially 2 dB and was reduced to 0.25 dB after the first two reversals. The mean of the last 10 reversals in a block of 14 reversals was taken as the threshold estimate for that block. Two estimates could be collected on JPK’s right ear. Results Figure 4 (A) presents the individual TMTFs. Figure 4 (B) presents the mean TMTFs for normalhearing listeners, for hearing-impaired listeners and for patient JPK. The TMTFs of the five normal-hearing adults and five hearing-impaired Measurement of the temporal-modulation transferfunction 347 '+ ' U GA 4 0 LG -35 i ~ Q -30 h % v ~ -25 s -1 PH I3 FH M 0 A MCH 0 EM 0 N X -10 Downloaded by [University of California, San Diego] at 21:33 19 April 2016 BL x-MDO 0 - -20 -1 5 -5 A-MDE JOM +PK I - 1 I 1000 L U 100 10 f m (Hz) 4 0 -35 -30 ~ ~ 1 h -0 v E -20 M 2 0 N -15 1 -10 -5 0 1 u 1000 Figure 4. Temporal modulation transfer function (TMTFs) obtained with a gated white noise carrier. Thresholdsfor detecting sinusoidal amplitude modulation of the noise carrier are expressed as 20 log m, where m is the modulation depth, and plotted as a function of modulation frequency f,. Note that the y-axis is upside down (i.e. negative values of 20 log m are at the top). (A):Individual TMTFs for normal hearing adults (unfilled symbols with dotted lines); elderly listeners with cochlear hearing loss (grey symbols with continuous lines); and patient JPK (black circles with thick continuous lines). (B):Mean TMTFs for normal-hearing adults (unfilled squares with thin lines); and elderly listeners with cochlear hearing loss (grey squares with thin lines). Error bars indicate the across-listener standard error of the mean. Mean TMTFs are plotted along with the individual TMTFofpatientJPK (black circles with thick lines). elderly listeners without brain damage are similar to those reported by previous studies for normal-hearing adults (Robenburg, 1977; Viemeister, 1979; Takahashi and Bacon, 1992; Eddins, 1993; Akeroyd and Patterson, 1997), adults with moderate to severe sensorineural hearing loss (Bacon and Viemeister, 1985;Moore et al., 1992) and elderly people with mild sensorineural hearing loss (Takahashi and Bacon, 1992). Overall, modulation-detection thresholds Downloaded by [University of California, San Diego] at 21:33 19 April 2016 348 F . Hescot et al. are relatively constant up to approximately 16 Hz and increase beyond that frequency, giving the TMTFs their typical lowpass shape. In agreement with the data of Bacon and Viemeister (1985) and Takahashi and Bacon (1992), the five elderly listeners with cochlear hearing loss (mean age 66 years) have poorer modulation-detection thresholds than adult listeners with normal hearing (mean age 52 years), especially at high modulation frequencies (here, at 128 Hz). The mean difference in detection thresholds between normal-hearing adults and elderly listeners with cochlear hearing loss ranges from 3 dB to 6 dB for modulation frequencies between 4 Hz and 64 Hz and reaches a maximum of 10dB at 128 Hz. Figure 4 also reveals that the TMTF of patient JPK differed strongly from those for the two control groups. First, modulation detection is much poorer for JPK than for the five normal-hearing adults and the five elderly listeners with cochlear hearing loss. The deviation in sensitivity from the five elderly listeners with cochlear hearing loss reached a maximum of approximately 19 dB at 4 Hz. At this modulation frequency, JPK’s first and second estimates of modulation detection threshold are -10 dB and -5 dB, respectively. Second, JPK’s modulation thresholds were 6-12 dB poorer at 4,64 and 128 Hz than at 8,16 and 32 Hz, giving his TMTF a bandpass shape. Modulation detection was very poor at the lowest and highest modulation frequencies for JPK. Nevertheless, the relatively low modulation-detection threshold measured at 8 Hz (-20 dB) indicates that JPK was able to perform the task. An analysis of variance (ANOVA) was used to estimate the significance of the differences in modulation detection thresholds between normal hearing adults, elderly listeners with cochlear damage and JPK. This analysis, conducted with factors of group (three levels) and modulation frequency (six levels), showed significant main effects of group [F(2,8)=23.61; p<0.0005] and modulation frequency [F(5,40)=40.41;p