Brain, Vol. 125, No. 3, 524-537,
February 2002
© 2002 Guarantors of Brain
Local improvement in auditory frequency discrimination is associated with hearing-loss slope in subjects with cochlear damage
1 Service d’Explorations Fonctionnelles ORL et Audiophonologiques, Hôpital Edouard Herriot and 2 Unité CNRS UMR 5020 Laboratoire ‘Neurosciences et Systèmes Sensoriels’, CNRS GDR 2213 ‘Prothèses auditives’, Université Claude Bernard Lyon I, Lyon, France
Correspondence to: Dr Hung Thai-Van, Service d’Explorations Fonctionnelles ORL et Audiophonologiques, Pavillon U-CHU Edouard Herriot, Place d’Arsonval, 69437 Lyon Cedex 03, France E-mail: hung.thaivan{at}chu-lyon.fr
Received July 2, 2001. Revised October 1, 2001. Accepted October 5, 2001.
 |
Summary |
|---|
 Top Summary IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Earlier data in the literature have shown local improvements in frequency discrimination performance near the cut-off frequency of steeply sloping, high-frequency hearing loss in subjects with cochlear damage. The general objective of the present study was to characterize further the relationships between this effect and various audiometric variables: namely, the slope, extent and shape of the hearing loss. In particular, we were interested in determining whether the effect was present in subjects with more moderately sloping hearing loss and/or other patterns of loss. Frequency difference limens (DLFs) were measured in 20 subjects (eight female, 12 male, median age 55.5 years) with high-frequency hearing loss. At least 12 frequencies were tested at intervals of 1/8 octave over a range of 1.5 octaves around the cut-off frequency for hearing loss (Fc). The Fc corresponded to the audiogram edge frequency and was defined as the highest test frequency, at the beginning of the slope, with a hearing threshold of no more than 5 dB HL above that of the best hearing frequency. The level of the test tones was randomized over a range of 6 dB around a nominal level, following an equal-loudness contour curve measured at 1/2-octave intervals. Results showed that DLFs were significantly smaller in a frequency band 1/4 octave wide centred on Fc than in the other bands. Furthermore, the average DLF measured in this band proved to be negatively correlated with the slope of hearing loss. No such significant relationship was found with the other audiometric indices considered, namely, the extent and maximum amount of hearing loss and the log-transformed cut-off frequency. The 20 subjects were divided into three groups according to the slope of their hearing loss relative to Fc (steep, >25 dB/1/2 octave; medium, between 12 and 25 dB/1/2 octave; and shallow, <12 dB/1/2 octave). A local improvement in DLF around Fc was observed in the steep- and medium-slope groups and was confirmed statistically in the steep-slope group. Similar measurements in subjects with low-frequency or notched hearing loss allowed us to establish the presence of similar local improvements in DLFs around audiogram edges. These results, which suggest the slope of the hearing loss to be the most important factor for the occurrence of local DLF improvements, are consistent with both an interpretation in terms of peripheral mechanisms and one in terms of central mechanisms, i.e. injury-induced neural reorganization.
Keywords: hearing loss; frequency discrimination; audiogram slope; neural plasticity
Abbreviations: ANOVA = analysis of variance; CF = characteristic frequency; DLF = frequency difference limen; Fc = cut-off frequency
|
Introduction |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Massive reorganization of sensory maps has been shown to occur in the primary somatosensory cortex of adult animals and humans following limb amputation. Following long-term upper limb denervation in monkeys, neurones that initially responded specifically to this limb could later be activated by stimulation applied to the face (Jones and Pons, 1998
). In human hand/arm amputees, it is well known that touching the face or the amputation stump can give rise to sensations referred to the amputated ‘phantom’ limb (Ramachandran et al., 1992). Since the face and upper arm are represented adjacent to the representation of the lower arm and hand in Penfield’s homunculus, this perceptual phenomena is thought to reflect the invasion of the deprived cortical region of the hand/lower arm representation by the adjacent face and upper arm representations (Ramachandran, 1993). Analogous neural reorganization phenomena have been reported in the visual modality after focal retinal lesions: primary visual cortex units deprived of peripheral input start responding to stimulation of neighbouring regions of the visual field (Kaas et al., 1990; Kaas, 1991; Gilbert and Wiesel, 1992).
In the auditory modality, restricted cochlear lesions have been shown to result in modified neural representation of the receptor surface in the primary auditory cortex (Robertson and Irvine, 1989). When hearing thresholds for a given frequency region have become abnormally elevated due to a peripheral lesion, neurones with initial characteristic frequencies (CFs) falling in such a region will develop low-threshold responses to frequencies whose cochlear place was at the edge of the lesion (lesion-edge frequencies). The net effect of these receptive field modifications is to expand the cortical representation of the lesion-edge frequencies (Rajan et al., 1993; Schwaber et al., 1993; Irvine and Rajan, 1995).
In parallel with these studies, neurophysiological investigations have shown relationships between the cortical representation of receptor surfaces and perceptual or motor performance in the corresponding sensory or motor modality. Notably, Recanzone et al. (1992a, b) have shown enlarged cortical representation of digits following selective tactile discrimination training in monkeys. Human studies using imaging techniques have shown enhanced representation of the fingers used for reading Braille in blind subjects (Pascual-Leone and Torres, 1993; Sterr et al., 1998a, b) and of left-hand fingers in string players (Elbert et al., 1995). In the auditory modality, Recanzone et al. (1993) have demonstrated correlated changes in frequency discrimination limens (DLFs) and cortical representations after frequency-discrimination training in rhesus monkeys.
The observation that use- and injury-related plasticities result in broadly similar changes in cortical sensory maps, with over-representation of certain regions of the receptor surface, prompts the hypothesis that cochlear lesions might, like perceptual training, induce frequency-specific improvements in auditory discrimination performance. One study was carried out by Buss et al. (1998) in order to test this hypothesis. The authors measured performance in a range of basic perceptual auditory tasks, including frequency-modulation detection, intensity discrimination, gap detection and gap-duration discrimination, in normal-hearing listeners and in subjects with steeply sloping high-frequency hearing loss. They found no significant difference between the two groups of listeners in performance on these various tasks that could not be explained simply in terms of a difference in peripheral auditory function. A possible interpretation of this result is that injury-induced plasticity in the auditory centres does not necessarily have significant perceptual consequences, at least in humans. However, McDermott et al. (1998) demonstrated a local improvement in frequency-discrimination performance around the hearing-loss cut-off frequency in subjects with steeply sloping high-frequency hearing loss. As pointed out by the authors, this finding is hard to reconcile with earlier results from the psychoacoustic literature. In particular, existing data have generally shown cochlear hearing loss to have a detrimental effect on DLF (Turner and Nelson, 1982; Moore and Peters, 1992). In line with existing theories that frequency can be encoded in place (Zwicker, 1970) or temporal (Rose et al., 1968; Moore, 1973, 1989) responses of the peripheral auditory system, such a detrimental influence of hearing loss on frequency discrimination performance could be mediated either by reduced frequency selectivity or by a decrease in the phase-locking of afferent auditory neurones following cochlear damage. Both phenomena would lead to a reduction in the amount or accuracy of frequency-related information at the peripheral level. This view is consistent with the finding that low-pass, high-pass and notched masking noises, which prevent subjects using information from frequencies located below and/or above the stimulus frequency, lead to a decrease in frequency discrimination performance (Emmerich et al., 1986; Moore and Glasberg, 1989). Thus, one would expect DLFs to be increased, not decreased, near the hearing loss cut-off in subjects with high-frequency hearing loss who have no or limited access to high-frequency information. In order to resolve the apparent contradiction between these data and their own finding (of locally enhanced discrimination performance near the hearing loss cut-off in subjects with steeply sloping hearing loss), McDermott et al. (1998) considered two interpretations. The first was that the better DLFs near the audiogram cut-off frequency resulted from the fact that subjects had received more training around that frequency than at more distant test frequencies. This interpretation was rejected by the authors because of the fact that they observed no significant improvement in DLFs across test runs. The second main interpretation, which the authors considered to explain their finding, was in terms of injury-induced plasticity in the central auditory system. Referring to the above-mentioned finding of enhanced spatial representation of both lesion-edge and trained frequencies in the auditory cortex of adult mammals (Rajan et al., 1993; Recanzone et al., 1993), they suggested that a lesion-induced increase in the number of neurones responding to a narrow range of frequencies near the hearing loss cut-off might explain the local enhancement in discrimination performance that they observed.
The present study sought to investigate further the characteristics of the hearing loss that are associated with frequency discrimination enhancement effects near the audiogram cut-off frequency. A first potentially important audiometric variable in relationship to DLF enhancement effects consists of the slope of the hearing loss. In most of the cases in which injury-induced reorganization has been observed in animals, the hearing loss, whether progressive (as with normal ageing, i.e. presbyacusis) or sudden (like that induced experimentally by mechanical lesions), was profound and steeply sloping (Willott et al., 1993; Rajan and Irvine, 1996, 1998). Moreoever, data in the literature suggest that central reorganization is unlikely to occur for moderate or shallow-sloping hearing losses (Rajan and Irvine, 1996; Rajan, 1998). Previous psychoacoustical studies in humans, which have involved exclusively subjects with steeply sloping (so-called ‘ski slope’) hearing loss (Buss et al., 1998; McDermott et al., 1998), did not determine whether the DLF enhancement effect was absent in subjects with moderate or shallow-sloping hearing losses. In the present study, we examined the influence of the slope of hearing loss by measuring and comparing DLFs in a relatively large sample of subjects exhibiting hearing loss with widely different slopes, ranging from very steep to very shallow.
Another potentially important factor of the DLF enhancement effect relates to the shape of the hearing loss. Is the effect associated only with high-frequency hearing loss or is it also present with low-frequency and notched hearing losses? Rajan and Irvine (1998) did not find any cortical reorganization in a cat with a low-frequency hearing loss. Reorganization of cortical frequency maps was reported in guinea-pigs with notched hearing loss (Robertson and Irvine, 1989). In the present study, subjects with low-frequency or notched hearing loss were included along with subjects with high-frequency hearing loss.
Finally, we also tried to gather some preliminary data on the potential influence of differences in the shape of the hearing loss across ears on the local DLF enhancement effect. Rajan et al. (1993) have shown that, in animals with hearing loss in one ear and normal hearing in the other, changes in the frequency tuning of binaurally influenced primary auditory cortex neurone clusters were observed in the cerebral hemisphere contralateral to the lesioned ear when that ear was stimulated, but not when the healthy ipsilateral ear was stimulated. Consequently, one may speculate that, in humans with unilateral hearing loss, perceptual consequences are unlikely to occur in the normally hearing ear. However, no clear prediction can be made regarding what will happen in subjects with asymmetrical hearing loss. Are the reorganization-related perceptual changes observable in both ears? In order to address this issue, we compared DLFs in the left and right ears in a few subjects with symmetrical or asymmetrical hearing loss.
|
Material and methods |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Subjects
Twenty-four subjects took part in the study, which was approved by the local ethics committee (Hospices Civils de Lyon, France). Written consent was obtained in accordance with the Declaration of Helsinki. Twenty subjects (eight female, 12 male, aged 30–66 years, median 55.5 years) had high-frequency hearing loss. Table 1 summarizes the demographic data and hearing-loss characteristics. Three subjects (males aged 48, 50 and 55 years) had notched hearing loss. The remaining subject (female aged 54 years) had non-fluctuating low-frequency hearing loss. Except for three subjects who were tested in both ears, subjects were tested in the ear in which the hearing-loss slope was steeper, even if only slightly so. Accordingly, 11 subjects were tested in their right ear and nine in their left ear.
|
Experimental design
Subjects were tested individually in a double-walled sound-attenuating booth. All stimuli were delivered monaurally via Sennheiser HD 545 headphones and consisted of pulsed pure tones with a 20 ms rise/fall time and a 350 ms plateau. Customized software, running on an IBM-compatible computer, was used to control the stimuli and to collect each subject’s responses. We used a Turtle Beach Systems Pro Series Digital Audio sound card (version 1.0.10), with a sampling rate of 44.1 kHz. Subjects were tested during two sessions, each with an overall duration of
4 h including rest periods. Both sessions took place within the same week. The first phase of the experiment involved measurement of absolute hearing thresholds and equal-loudness contours. The stimulus frequencies were distributed between 250 and 8000 Hz at intervals of 1/2 octave. For the measurement of absolute hearing threshold, frequency intervals were reduced to 1/8 octave near the beginning of the audiogram slope, so that the cut-off frequency could be estimated more precisely. The cut-off frequency (Fc) was defined as the test frequency that corresponded to the audiogram edge and as the highest with a hearing threshold of no more than 5 dB HL above that of the best hearing frequency. A total of 13 frequencies, spaced at intervals of 1/8 octave, were tested in the vicinity of the cut-off frequency, the lowest being set 1/2 octave below cut-off and the highest 1 octave above. In patients with notched hearing loss, a wider range of frequencies, spaced at the same interval, was tested, so as to include the two edges of the notch in the audiogram. In the patient with low-frequency hearing loss, 17 frequencies, spaced 1/8 octave apart, were tested, from 1 octave below the cut-off frequency to 1 octave above.
In subjects with high-frequency hearing loss, the equal-loudness contour was measured using as a reference a 500 Hz tone 30 dB above the hearing threshold of the subject (30 dB SL), i.e. at a comfortable level. For the subject with low-frequency hearing loss, a 2000 Hz tone, also delivered at a level of 30 dB SL, was chosen as the reference. Each subject had to determine the intensity level of various comparison tones that created the same loudness sensation as the reference tone. Comparison tones were distributed between 500 and 8000 Hz and spaced at intervals of 1/2 octave.
DLFs for pure tones were then measured. All frequencies for which the hearing threshold had been determined were tested. Methodological precautions were taken, so that subjects could not base their judgements in the DLF task on frequency-related changes in loudness rather than on changes in pitch. First, stimuli were delivered at a constant loudness level, using each subject’s previously obtained equal-loudness contour. Levels for frequencies between those tested in the loudness matching procedure were calculated by linear interpolation. Secondly, to prevent subjects from exploiting any residual loudness differences as the frequency changed in the DLF procedure, the level of each stimulus was varied in each trial by a random value selected from a uniform distribution ranging between –3 and +3 dB. Each subject performed pretest DLF training with a small number of runs, with feedback provided at the end of each trial in the form of a green light that was switched on after each correct response and a red light after each incorrect response; responses were not collected from these training runs.
Psychophysical procedures
Absolute hearing threshold measurement
Absolute thresholds were measured using a one-interval, yes/no procedure in which the subject’s task was to indicate whether he or she had perceived a tone. A one-down, one-up adaptive tracking rule was followed: the stimulus level, initially set at 50 dB HL, was decreased after a ‘yes’ response and increased after a ‘no’ response. With this rule, thresholds correspond to the 50% correct response point on the psychometric function (Levitt, 1971). The level was varied in 6 dB steps until the fourth reversal in level variation and in 2 dB steps thereafter. The procedure stopped after a total of eight reversals. The absolute threshold was computed as the arithmetic mean of the levels at the last four reversals. Two threshold estimates were obtained for each frequency; the absolute thresholds shown in the Results section correspond to the average of these two estimates. The various frequencies were tested in random order. It took 1.5 h to measure the absolute thresholds for 20 frequencies.
Loudness matching procedure
Loudness matches were performed using an adaptive two-interval, two-alternative procedure with a two-down, two-up tracking rule. On each trial, two successive observation intervals were presented, separated by a 300 ms silent gap. One interval contained a reference tone with a fixed 30 dB SL level and the other contained a comparison tone of variable level. The two observation intervals were presented at random, with an equal probability of either one coming first. The subject was asked to indicate the interval in which the tone was louder. The level of the comparison tone, which was initially set at random 10 dB below or 10 dB above that of the reference tone, was decreased when it was judged louder than the reference on two consecutive trials, and increased when it was judged less loud on two consecutive trials. Steps of 5 dB were used until the second level reversal and 1 dB steps thereafter. The arithmetic mean of the levels at the last four reversals was computed. The order of testing was determined by random selection from among the set of frequencies spaced 1/2 octave apart in the audiogram. For each tested frequency, the result was the average of two runs. The overall procedure took about 1.5 h.
Frequency discrimination procedure
The DLFs were measured using a three-interval, two-alternative forced-choice procedure with a two-down, one-up decision rule. On each trial, two of three successive observation intervals contained pure tones at a reference frequency; the remaining interval, which had a 50% chance of coming second and a 50% chance of coming third, contained a pure tone of variable frequency which was always higher than the reference. The intervals were separated by silent gaps of 300 ms. The subject’s task was to identify the interval that contained the higher tone. The difference between the reference and the variable frequency was initially set to 20% of the reference. After two consecutive correct responses, the difference in frequency was decreased; it was increased after each incorrect response. Ten reversals were obtained in each run. The frequency difference was varied by a factor of 2 until the fourth reversal occurred and by a factor of 1.41 thereafter. After each run, the geometric mean for the last six reversals was computed. The order of testing was determined by random selection from among the set of frequencies spaced 1/8 octave apart, for which the absolute hearing threshold had been previously measured. Three runs were collected at each of these test frequencies. The overall DLF task lasted 4 h.
|
Results |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
DLFs in subjects with high-frequency hearing loss
Inter-frequency and frequency band comparisons
Figure 1A shows the average hearing thresholds and DLFs measured at and around Fc in the 20 subjects with high-frequency hearing loss. The DLF data were first subjected to repeated-measures analysis of variance (ANOVA) with frequency and run number as within-subject factors. The results indicated significant differences in DLF across frequencies [F(11,187) = 6.04, P < 0.001] but not across runs [F(2,34) = 0.91]. Planned comparisons were performed between the DLFs measured at each test frequency (except the lowest and the highest) and the DLFs measured at the two surrounding test frequencies, i.e. 1/8 octave below and above. Significant differences were found between Fc versus Fc – 1/8 octave and Fc + 1/8 octave [F(3,17) = 3.97, P < 0.05] and between Fc + 1/8 octave versus Fc and Fc + 1/4 octave [F(3,17) = 3.74, P < 0.05]. No significant difference was found for the other frequencies. Further statistical analysis of the DLF data was performed by dividing the frequency range from Fc – 1/2 octave to Fc + 7/8 octave into four frequency bands of 1/4 octave. As a result, the second frequency band contained DLFs measured at Fc, Fc – 1/8 octave and Fc + 1/8 octave. For each band, the nine DLF estimates (three DLF values per frequency) were pooled. A one-way repeated-measures ANOVA was then performed on the log-transformed DLFs with frequency band as factor. The results revealed a significant main effect of band factor [F(3,57) = 14.08, P < 0.001]. Figure B shows that DLFs in the 1/4 octave-wide frequency band centred on Fc were, on average, smaller than those measured in the other three bands. A contrast analysis confirmed that DLFs measured in the frequency band centred on Fc were significantly different from those measured in the band just above [F(1,19) = 14.42, P < 0.05] and in the highest band [F(1,19) = 16.48, P < 0.005], but also from those measured in the lower band [F(1,19) = 6.48, P < 0.05]. As pointed out by McDermott et al. (1998
), the comparison of the DLFs measured at and around Fc with those measured at lower frequencies is probably more important than the comparison with higher frequencies, as poorer discrimination is expected at frequencies at which greater hearing loss is present. The contrast analysis further indicated a significant difference in the lowest and highest bands [F(1,19) = 13.93, P < 0.005]. The comparison between the two frequency bands surrounding that centred on Fc just failed to reach the threshold for statistical significance [F(1,19) = 4.22, P = 0.054].
|
Correlation between DLF data and hearing-loss characteristics
Pearson’s correlation was used to assess the relationship between the mean DLF in the 1/4 octave-wide frequency band centred on Fc (DLFFc±1/8) and several audiometric indices. These indices were as follows: slope of the hearing-loss relative to Fc, measured as the difference between hearing thresholds measured at Fc and 1/2 octave above Fc; extent of hearing loss, measured as the number of frequencies at which hearing thresholds exceeded 45 dB HL; maximum amount of loss, corresponding simply to the largest hearing threshold measured in that subject; and log-transformed cut-off frequency. We also computed the difference between DLFFc±1/8 and the mean DLFs measured in the two adjacent 1/4-octave-wide frequency bands, and tested for correlation between this difference (
DLF) and the same audiometric indices.
Testing for correlation between DLF data and hearing-loss characteristics revealed a significant negative correlation between DLFFc±1/8 and hearing-loss slope (r = –0.56, P < 0.05). DLFFc±1/8 was itself negatively correlated to DLF (r = –0.55, P < 0.05). The correlation between (DLF and the hearing-loss slope just failed to reach the threshold for statistical significance (r = 0.43, P = 0.059). Neither DLFFc±1/8 nor DLF was significantly correlated with the other audiometric indices considered.
Analysis of DLF data according to hearing-loss slope
In order to investigate further the influence of hearing-loss slope, the 20 subjects with high-frequency hearing loss were divided into three groups according to the slope of their hearing loss relative to Fc. For purposes of comparison with the results obtained by McDermott et al. (1998), whose subjects had steep slopes (
50 dB/octave), one group was formed containing all subjects with relative hearing loss >25 dB/1/2 octave (n = 9). A second, ‘medium-slope’ group was formed of six subjects with relative hearing loss between 12 and 25 dB/1/2 octave. The five remaining subjects, in whom relative hearing loss was <12 dB, formed a ‘shallow-slope’ group.
Figure 2 shows the average hearing thresholds and DLFs measured at and around Fc in the three slope groups. No significant differences between the three groups were noted regarding age [one-way ANOVA, F(2,17) = 1.17], sex (
2 likelihood ratio = 4.74), ear tested (2 likelihood ratio = 0.11) or Fc value [one-way ANOVA, F(2,17) = 2.79]. The three groups’ DLF data were subjected to repeated-measures ANOVA with frequency and run number as intrasubject factors and slope group as an intersubject factor. In addition to the previously observed main effect of the frequency factor and the lack of effect of the run factor, the results indicated that the DLF values obtained in the three slope groups did not differ significantly overall [F(2,17) = 2.22]. However, a significant interaction between the group and frequency factors was observed [F(22,187) = 3.47, P < 0.005], indicating that DLF variation across frequencies was not the same in the three slope groups. In order to get further insight into this result, we performed, for each group, contrast analyses comparing the DLFs obtained at each frequency with those for the two neighbouring (upper and lower) frequencies. No statistically significant contrast was found in the shallow- and medium-slope groups. In the steep-slope group, DLFs were found to be significantly smaller at Fc + 1/8 octave than at the neighbouring frequencies [F(1,8) = 6.33, P < 0.05].
|
Inter-ear comparison
Figure 3 shows the absolute thresholds and DLFs measured in the left and right ears of three subjects whose audiometric characteristics are indicated in Table 2. As indicated by the values in this table, two of these subjects had quasi-symmetrical hearing loss with similar cut-off frequencies and slopes in both ears. The third subject had similar cut-offs but widely different slopes in the two ears.
|
|
Since individual DLF curves are often jagged, it is difficult to determine whether local improvements in DLF have occurred in a given subject. A possible approach in this case is to look for absolute rather than relative minima in the DLF curves. In the two subjects with similar hearing-loss slopes, DLF was found to be the best, over the whole Fc – 1/2 octave to Fc + 1 octave range, near Fc, namely, at Fc – 1/8 octave, Fc, Fc + 1/8 octave or Fc + 1/4 octave, in both ears. In the subject with asymmetrical hearing loss, the best DLF occurred at Fc in one ear—the ear with the steeper slope; in the other ear, in which the slope was smaller than 12 dB/1/2 octave, the best DLF was found far away from the Fc, namely at Fc – 1/2 octave.
Comparison with low-frequency hearing loss
Figure 4 shows the absolute thresholds and DLFs measured in the left ear of the subject with low-frequency hearing loss. The hearing loss cut-off frequency (defined here as the test frequency that corresponded to the audiogram edge and as the lowest with a hearing threshold of no more than 5 dB HL above that of the best hearing frequency) was found to be at 1414 Hz. The hearing loss relative to Fc, measured between Fc and the point 1/2 octave below it, was 16 dB, and that measured between Fc and the point 1 octave below was 33 dB. The best DLF was found to occur precisely at the estimated hearing-loss cut-off frequency. Comparison between frequency bands was performed by dividing the frequency range from Fc – 7/8 octave to Fc + 7/8 octave into five frequency bands 1/4 octave wide. For each band, the nine DLF estimates (three DLF values per frequency) were pooled. Across-band comparisons revealed that DLFs in the band centred on Fc were, on average, smaller than those measured in other frequency bands. However, in this single case this could not be confirmed statistically by one-way repeated-measures ANOVA performed on the log-transformed DLFs with frequency band as factor [F(4,32) = 2.29, P = 0.08].
|
Comparison with notched hearing loss
Figure 5 shows the absolute thresholds and DLFs measured in three subjects with notched hearing loss. For these subjects, one-way ANOVA and Student–Newman–Keuls pairwise comparisons were used to test for the significance of local minima in the DLF curve that could be observed across frequencies. Table 3 shows the estimated low (Fclow) and high (Fchigh) cut-off frequencies and their relative hearing losses measured, respectively, between Fclow and the point 1/2 octave above it and between Fchigh and the point 1/2 octave below it. In the first subject, the best DLF was found to occur at Fclow – 1/8 octave and local minima in the DLF curve were observed at Fclow – 1/8 octave, Fclow + 3/8 octave and Fchigh – 5/8 octave; however, no significant difference between mean DLFs obtained at each of these frequencies and those obtained at their surrounding test frequencies could be demonstrated. In the second subject, the best DLF was found to occur at Fclow + 1/8 octave and a local minimum in the DLF curve was also observed at Fchigh. Because of the very jagged DLF curve in this subject, too many local minima were identified within the intermediate frequency range to be related reliably to either the low or the high cut-off frequencies. ANOVA and Student–Newman–Keuls pairwise comparisons showed that the mean DLF was significantly smaller at Fclow + 1/8 octave than at the adjacent lower (P < 0.05) and higher test frequencies (P < 0.05). The mean DLF was also significantly smaller at Fchigh than at the lower neighbouring test frequency (P < 0.01); however, it was not significantly smaller than that at the higher neighbouring frequency (P = 0.09). No significant differences were found for mean DLFs at test frequencies between Fclow + 1/8 octave and Fchigh. In the third subject, the best DLF was found to occur at Fclow + 1/8 octave. Local minima were observed at Fclow + 1/8 octave and Fchigh. The mean DLFs at Fclow + 1/8 octave and at Fchigh were found to be significantly improved compared with those at both lower (P < 0.001) and higher neighbouring test frequencies (P < 0.001).
|
|
|
Discussion |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The occurrence of local DLF enhancement effects depends essentially on hearing-loss slope
In addition to confirming the recent finding by McDermott et al. (1998
) of improved DLFs near the hearing loss cut-off frequency in subjects with steeply sloping (>50 dB/octave) high-frequency hearing loss, the present results shed further light on the factors underlying this local perceptual enhancement phenomenon. First of all, they indicate that the effect depends primarily on the slope of the hearing loss: basically, the steeper the slope, the greater the difference between DLFs measured around Fc and in adjacent bands. Furthermore, when subjects are grouped according to hearing-loss slope, the effect is generally found to be non-significant in groups with a slope shallower than 25 dB/1/2 octave (i.e. 50 dB/octave). One may argue that the complete absence of local DLF improvement in the shallow-slope group might be explained by the limited number of subjects (n = 5) in this group. However, this group contained at least as many subjects as in the study of McDermott et al. (1998), in which statistical evidence for DLF enhancement was found.
An important question regards the locus of the DLF improvement relative to the cut-off frequency of the hearing loss. In the present study, DLFs were found to be improved locally in a frequency range of ±1/8 octave around the Fc. However, the exact locus varied between subjects: it was at the Fc itself in some subjects, at Fc – 1/8 octave in others and at Fc + 1/8 octave in yet other subjects. Rajan and Irvine (1996) observed, in cochlear-damaged cats with a cut-off frequency of 18 kHz, an expanded cortical representation of frequencies within 1–2 kHz of the lesion edge. This width corresponds to 1/12 to 1/6 octave around the cut-off frequency. The frequency range in which we observed locally decreased DLFs in the present study, namely 1/8 octave, falls within this estimated range. Another important question concerns the factors determining the locus of the DLF improvement. Interestingly, it was found that the average improvement locus differed between the two groups in which an improvement effect was observed: in the medium-slope group, DLFs were generally found to be smallest at Fc, whereas in the steep-slope group the smallest DLFs were generally obtained at Fc + 1/8 octave. At first sight, this observation appears to argue against the suggestion that the narrow range of frequencies that are over-represented in the central auditory system corresponds simply to those just below that for which CF representation in cortical maps has been lost due to hearing thresholds reaching a certain critical value (Rajan and Irvine, 1998). Indeed, because this critical amount of loss should be reached at lower frequencies in subjects with a steep slope, a lower breakpoint should be observed in these subjects than in subjects with shallower slopes. However, it should be noted that, over all frequencies up to the cut-off and even a bit further, hearing thresholds were in fact 10 dB better in the steep- than in the medium-slope group. Consequently, in the steep-slope group, hearing thresholds approximated 10 dB at a higher frequency than in the medium-slope group. This might resolve the apparent contradiction between the present results and the hypothesis, inspired by the above-mentioned neurophysiological data, that reorganization affects those frequencies just below that at which there has been a significant amount of loss. Differences in low-frequency hearing threshold between the three slope groups may also account for the generally better low-frequency DLFs found in the steep-slope group. At present, we have no satisfactory explanation for the fact that hearing thresholds were generally better at low frequencies in subjects with steeply sloping hearing loss than in subjects with shallower slopes. In the light of literature data indicating age-related decreases in low-frequency hearing sensitivity (Brant and Fozard, 1990; Pearson et al., 1995), this difference might be related to the fact that subjects in the steep-slope group were, on average, 10 years younger than those in the medium-slope group and 5 years younger than those in the shallow-slope group.
Is there an effect of the extent of hearing loss?
While the audiograms of our steep-slope subjects are comparable to those of the subjects in the study of McDermott and colleagues with respect to slope, they differ substantially in cut-off frequency. Hearing-loss cut-off frequencies varied between 1414 and 2594 Hz (mean 1981 Hz) in our steeply sloping group versus between 480 and 850 Hz (mean 674 Hz) in the study of McDermott and colleagues. This difference has some implications for our understanding of the conditions necessary for cortical reorganization to occur. One widespread interpretation of the DLF improvement found in a narrow range around Fc draws upon the neurophysiological finding in animals (Irvine et al., 2000) that this neighbouring hearing-loss cut-off with a narrow frequency range is over-represented on the primary auditory cortex’s tonotopic map; more neurones are thus available for encoding frequencies falling in that range, and discrimination performance is correspondingly better. If this interpretation is correct, the enhancement effect should increase as the cut-off frequency gets lower, because the lower the cut-off frequency the larger is the cortical area that contains units having characteristic frequencies that are either above the edge or are the best frequencies at the edge. Thus, lower cut-offs should be statistically associated with greater over-representation of the narrow edge frequency range and greater DLF enhancement. Conversely, in subjects with relatively high cut-offs, improvements in frequency discrimination performance near the hearing-loss cut-off should be smaller. These predictions were not confirmed by the present results. First, DLF enhancement effects were observed even in subjects with high cut-off frequencies, and these effects were similar to those obtained in the subjects of McDermott et al. (1998). Secondly, the amount of DLF improvement was not found to be correlated either to the hearing-loss cut-off frequency or to the frequency range over which hearing thresholds exceeded 45 dB HL. The lack of relationship between DLF improvement and the frequency range of hearing loss may be due to the critical loss value chosen here (45 dB HL) being too small for neurones with CFs within the loss range to be completely deprived of input. However, when critical loss values >45 dB HL were tried, the estimated hearing-loss range reached zero in an increasing number of subjects as the critical loss value was increased; thus, more and more scatterplot points fell on a vertical line, preventing observation of any significant correlation between the two variables. It could be argued that, if there had been more subjects with large amounts of hearing loss spanning frequency ranges of different widths, significant relationships between our ‘extent of loss’ index and DLF improvement might have been observed.
Is there an effect of the amount of hearing loss?
Neurophysiological data suggest that, for injury-induced primary auditory cortex reorganization to occur, there must be ‘a cochlear region in which damage is so comprehensive that there is no neural outflow from this region, (Rajan and Irvine, 1998). Psychoacoustic data in humans suggest that, at least in the 250–8000 Hz range, hearing losses larger than 85 dB HL are likely to be associated with ‘dead regions’ in the cochlea (Moore et al., 2000). In most of our subjects (three out of six in the medium-slope group and five out of nine in the steep-slope group) the maximum hearing loss measured in the 250–8000 Hz range was 
75 dB. Nonetheless, the results of Moore and Alcantara (2001) leave open the possibility that, in some cases, dead cochlear regions are associated with even smaller hearing losses. Furthermore, the degree of correlation between the dead regions identified by psychoacoustic means and actual dead regions in the cochlea has not been investigated. Thus, the present results do not allow a clear conclusion regarding whether the presence of dead regions in the cochlea is a prerequisite for the occurrence of cortical reorganization.
In fact, the present results are equivocal regarding the influence of the amount of hearing loss on the occurrence of injury-induced neural reorganization, as estimated from local DLF enhancement. On the one hand, they provide some evidence for the notion that considerable hearing loss is required for local DLF enhancement effects near Fc to occur. In the shallow-slope group, in which no significant perceptual enhancement effect was generally observed, the maximum hearing loss measured within a 2-octave range around Fc was 54 dB HL (versus 65 dB HL in the medium-slope and 76 dB HL in the steep-slope groups, in which local DLF enhancement effects near Fc were observed). On the other hand, the present results provide some evidence against the notion that hearing thresholds have to be greatly elevated over a wide frequency range for a local DLF enhancement near Fc to appear. In subjects with notched audiograms, hearing loss was limited both in frequency and in intensity (50 dB HL over the frequency range tested), indicative of local DLF improvement effects, and the present data suggest that hearing loss does not have to be very large and wide for neural injury-induced reorganization to occur. In view of these contradictory findings regarding the influence of the amount of hearing loss on the occurrence of local DLF enhancement effects near the hearing-loss cut-off, one may conclude that the overall amount of hearing loss is not a critical factor in the effect. Rather, it appears once again that the hearing-loss slope is the predominant factor, since the notched audiogram patients, in whom DLFs appeared to be locally improved near cut-off while hearing loss was limited, had large hearing-loss slopes (with a loss relative to Fc of 33 dB/1/2 octave, on average).
Is there an effect of the pattern of hearing loss?
Our observation of minimum DLFs near cut-off frequencies in both notched and low-frequency hearing loss suggests that high-frequency hearing loss is not a necessary condition for injury-induced auditory reorganization to occur. Rajan and Irvine (1998) have reported the absence of reorganization in the primary auditory cortex of a cat with a low-frequency hearing loss. This may suggest that central reorganization does not occur in this type of loss. Our observation of a minimum DLF at hearing-loss cut-off frequency in one case of low-frequency hearing loss brings this view into question. However, it should be noted that the hearing loss in Rajan and Irvine’s study was conductive and very limited (maximum 30 dB). One may hypothesize that the hearing-loss slope in our case was sufficiently steep to result in cortical reorganization with perceptual consequences. The observation of a local DLF improvement effect in patients with notched audiograms is consistent with previous results from Robertson and Irvine (1989) in guinea-pigs with induced notched hearing loss, in which there was over-representation of both edge frequencies. The three subjects with such a hearing-loss pattern tested in this study appeared to show local DLF improvement near or at the lower edge of their hearing loss. In addition, two of them also appeared to show local improvement at the upper edge.
Do local DLF improvements near the audiogram cut-off reflect central plasticity?
In the above discussion, following McDermott et al. (1998), we have interpreted the finding of improved DLFs near the audiogram cut-off in hearing-impaired subjects in terms of injury-induced central plasticity. The results of the present study are generally consistent with this interpretation. In particular, the fact that the effect is observed only in subjects whose hearing-loss slope is relatively steep is in agreement with earlier neurophysiological results reporting injury-induced central plasticity effects in animals with steep but not shallow or moderate hearing loss slopes (Rajan and Irvine, 1996; Rajan, 1998). Furthermore, the range of frequencies over which DLFs were found to be improved corresponds to that which neurophysiological data in animals indicate to be over-represented following cochlear damage (Rajan and Irvine, 1998). Nevertheless, it is worth considering other possible interpretations. A first alternative interpretation, as pointed out by McDermott et al. (1998), is that the observed improvements in DLFs might reflect perceptual training. Indeed, in their study, as in the present, DLFs were measured around the hearing loss cut-off frequency. Assuming that frequency discrimination learning is to some extent frequency-specific, which remains uncertain for frequencies below 4–5 kHz (Demany, 1985; Irvine et al., 2000), the largest cumulated training effects occurred at or near the audiogram cut-off frequency in both studies. This interpretation was ruled out by McDermott et al. (1998) because no significant learning was observed across test runs in the frequency discrimination task. It is worth noting that, similarly, the analysis of our DLF measurements did not reveal any run effect.
Another interpretation of the finding of locally improved DLFs near the hearing loss cut-off relates to the possibility that subjects’ judgements in the frequency-discrimination task were based not on pitch differences but on loudness differences between the tones. When tones are presented at a constant physical level, systematic changes in loudness are likely to occur with changes in frequency in subjects whose hearing thresholds and/or loudness function slopes differ across frequencies. In the present study, as in the work of McDermott et al. (1998), such systematic loudness changes were probably weakened by the use of randomized level stimulation together with mean stimulus levels falling along an equal-loudness contour in each subject. However, because the contours were determined only at intervals of 1/2 octave, they may not have sufficiently taken into account the rapid changes in loudness with frequency that occur around the cut-off frequency. Furthermore, the 6 dB range of level randomization used in the present study, as well as that used by McDermott and colleagues (McDermott et al., 1998), may not have been enough to eliminate completely the loudness differences across frequencies, on average (Green, 1988). Thus, it could be that loudness cues were still present in frequency regions where the slope of the audiogram changed rapidly. The finding of local improvements in DLFs in subjects with steeply sloping audiograms but not in those with shallow-sloping audiograms is consistent with this interpretation. The fact that thresholds were improved very locally near the cut-off frequency rather than over the whole range of high frequencies over which hearing thresholds fell rapidly may be explained by a trade-off between the steepness of the audiogram slope and the extent of the hearing loss. Whilst the former promoted an improvement in DLFs, the latter probably had a detrimental influence, as indicated by the fact that DLFs are in general abnormally elevated in hearing-impaired subjects.
Conclusion
The results obtained in this study confirm the notion that frequency-discrimination performance is locally enhanced near the audiogram edge in subjects with high-frequency hearing loss. They suggest that this effect can be clearly demonstrated only in subjects with steeply sloping audiograms. On the other hand, the effect does not appear to depend critically on the extent or shape of the hearing loss, as it can be observed in subjects with low-frequency as well as notched hearing losses. These findings, pointing to the steepness of the hearing loss as the most important factor of the local DLF enhancement effect, are consistent with an interpretation in terms of injury-induced central auditory system plasticity. However, they also provide further support for an interpretation in terms of peripheral mechanisms, i.e. loudness cues. Further study is required before these two interpretations can be teased apart.
Note
In the context of this article, the expression dB HL refers to stimulus levels expressed relative to hearing thresholds measured in a population of 20 young subjects who had been found beforehand to have normal hearing (i.e. hearing thresholds between –10 and 10 dB HL) using standard (Telephonics TDH 49) headphones and a common (InterAcoustics AC30) audiometer. These reference hearing thresholds were measured using exactly the same testing material and headphones as used with the subjects of the experiments. Test frequencies ranged between 125 and 8000 Hz in half-octave intervals; the reference values for intermediate frequencies were derived by linear interpolation.
|
Acknowledgements |
|---|
The authors wish to thank Dexter Irvine, Hugh McDermott and Brian Moore for constructive comments on an earlier version of this manuscript. This work was supported by research grants from CNRS, CCA Groupe, Entendre, Oticon, Phonak and Siemens Audiologie.
|
References |
|---|
|
TopSummaryIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Brant LJ, Fozard JL. Age changes in pure-tone hearing thresholds in a longitudinal study of normal human aging. J Acoust Soc Am 1990; 88: 813–20.[ISI][Medline]
Buss E, Hall JW 3rd, Grose JH, Hatch DR. Perceptual consequences of peripheral hearing loss: do edge effects exist for abrupt cochlear lesions? Hear Res 1998; 125: 98–108.[ISI][Medline]
Demany L. Perceptual training in frequency discrimination. J Acoust Soc Am 1985; 78: 1118–20.[ISI][Medline]
Elbert T, Pantev C, Wienbruch C, Rockstroh B, Taub E. Increased cortical representation of the fingers of the left hand in string players. Science 1995; 270: 305–7.
Emmerich DS, Fantini DA, Brown WS. Frequency discrimination of tones presented in filtered noise. J Acoust Soc Am 1986; 80: 1668–72.[ISI][Medline]
Gilbert CD, Wiesel TN. Receptive field dynamics in adult primary visual cortex. Nature 1992; 356: 150–2.[Medline]
Green DM. Profile analysis. New York: Oxford University Press; 1988.
Irvine DRF, Rajan R. Plasticity in the mature auditory system. In: Manley GA, Klump GM, Köppl C, Fastl C, Oeckinghaus H, editors. Advances in hearing research. Singapore: World Scientific; 1995. p. 3–23.
Irvine DR, Martin RL, Klimkeit E, Smith R. Specificity of perceptual learning in a frequency discrimination task. J Acoust Soc Am 2000a; 108: 2964–8.[ISI][Medline]
Irvine DR, Rajan R, McDermott HJ. Injury-induced reorganization in adult auditory cortex and its perceptual consequences. [Review]. Hear Res 2000b; 147: 188–99.
Jones EG, Pons TP. Thalamic and brainstem contributions to large-scale plasticity of primate somatosensory cortex. Science 1998; 282: 1121–5.
Kaas JH. Plasticity of sensory and motor maps in adult mammals. [Review]. Annu Rev Neurosci 1991; 14: 137–67.[ISI][Medline]
Kaas JH, Krubitzer LA, Chino YM, Langston AL, Polley EH, Blair N. Reorganization of retinotopic cortical maps in adult mammals after lesions of the retina. Science 1990; 248: 229–31.
Levitt H. Transformed up-down methods in psychoacoustics. J Acoust Soc Am 1971; 49: 467–77.
McDermott HJ, Lech M, Kornblum MS, Irvine DR. Loudness perception and frequency discrimination in subjects with steeply sloping hearing loss: possible correlates of neural plasticity. J Acoust Soc Am 1998; 104: 2314–25.[ISI][Medline]
Moore BC. Frequency difference limens for short-duration tones. J Acoust Soc Am 1973; 54: 610–19.[ISI][Medline]
Moore BCJ. An introduction to the psychology of hearing. 3rd ed. London: Academic Press; 1989.
Moore BCJ, Alcantara JI. The use of psychophysical tuning curves to explore dead regions in the cochlea. Ear Hear 2001; 22: 268–78.[ISI][Medline]
Moore BCJ, Glasberg BR. Mechanisms underlying the frequency discrimination of pulsed tones and the detection of frequency modulation. J Acoust Soc Am 1989; 86: 1722–32.
Moore BC, Peters RW. Pitch discrimination and phase sensitivity in young and elderly subjects and its relationship to frequency selectivity. J Acoust Soc Am 1992; 91: 2881–93.[ISI][Medline]
Moore BC, Huss M, Vickers DA, Glasberg BR, Alcantara JI. A test for the diagnosis of dead regions in the cochlea. Br J Audiol 2000; 34: 205–24.[ISI][Medline]
Pascual-Leone A, Torres F. Plasticity of the sensorimotor cortex representation of the reading finger in Braille readers. Brain 1993; 116: 39–52.
Pearson JD, Morrell CH, Gordon-Salant S, Brant LJ, Metter EJ, Klein LL, et al. Gender differences in a longitudinal study of age-associated hearing loss. J Acoust Soc Am 1995; 97: 1196–205.[ISI][Medline]
Rajan R. Receptor organ damage causes loss of cortical surround inhibition without topographic map plasticity. Nat Neurosci 1998; 1: 138–43.[ISI][Medline]
Rajan R, Irvine DRF. Features of, and boundary conditions for, lesion-induced reorganization of adult auditory cortical maps. In: Salvi RJ, Henderson D, Fiorino F, Colletti V, editors. Auditory system plasticity and regeneration. New York: Thieme Medical; 1996. p. 224–37.
Rajan R, Irvine DR. Neuronal responses across cortical field A1 in plasticity induced by peripheral auditory organ damage. Audiol Neurootol 1998; 3: 123–44.[Medline]
Rajan R, Irvine DR, Wise LZ, Heil P. Effect of unilateral partial cochlear lesions in adult cats on the representation of lesioned and unlesioned cochleas in primary auditory cortex. J Comp Neurol 1993; 338: 17–49.[ISI][Medline]
Ramachandran VS. Behavioral and magnetoencephalographic correlates of plasticity in the adult human brain. [Review]. Proc Natl Acad Sci USA 1993; 90: 10413–20.
Ramachandran VS, Rogers-Ramachandran D, Stewart M. Perceptual correlates of massive cortical reorganization. [Review]. Science 1992; 258: 1159–60.
Recanzone GH, Jenkins WM, Hradek GT, Merzenich MM. Progressive improvement in discriminative abilities in adult owl monkeys performing a tactile frequency discrimination task. J Neurophysiol 1992a; 67: 1015–30.
Recanzone GH, Merzenich MM, Jenkins WM, Grajski KA, Dinse HR. Topographic reorganization of the hand representation in cortical area 3b of owl monkeys trained in a frequency-discrimination task. J Neurophysiol 1992b; 67: 1031–56.
Recanzone GH, Schreiner CE, Merzenich MM. Plasticity in the frequency representation of primary auditory cortex following discrimination training in adult owl monkeys. J Neurosci 1993; 13: 87–103.[Abstract]
Robertson D, Irvine DR. Plasticity of frequency organization in auditory cortex of guinea pigs with partial unilateral deafness. J Comp Neurol 1989; 282: 456–71.[ISI][Medline]
Rose JE, Brugge JF, Anderson DJ, Hind JE. Patterns of activity in single auditory nerve fibers of the squirrel monkey. In: Reuck AVS, Knight J, editors. Hearing mechanisms in vertebrates. London: Churchill; 1968. p. 144–57.
Schwaber MK, Garraghty PE, Kaas JH. Neuroplasticity of the adult primate auditory cortex following cochlear hearing loss. Am J Otol 1993; 14: 252–8.[ISI][Medline]
Sterr A, Müller MM, Elbert T, Rockstroh B, Pantev C, Taub E. Changed perceptions in Braille readers. Nature 1998a; 391: 134–5.[Medline]
Sterr A, Müller MM, Elbert T, Rockstroh B, Pantev C, Taub E. Perceptual correlates of changes in cortical representation of fingers in blind multifinger Braille readers. J Neurosci 1998b; 18: 4417–23.
Turner CW, Nelson DA. Frequency discrimination in regions of normal and impaired sensitivity. J Speech Hear Res 1982; 25: 34–41.
Willott JF, Aitkin LM, McFadden SL. Plasticity of auditory cortex associated with sensorineural hearing loss in adult C57BL/6J mice. J Comp Neurol 1993; 329: 402–11.[ISI][Medline]
Zwicker E. Masking and psychological excitation as consequences of the ear’s frequency analysis. In: Plomp R, Smoorenburg GF, editors. Frequency analysis and periodicity detection in hearing. Leiden: Sijthoff; 1970. p. 376–94.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Guiraud, J. Besle, L. Arnold, P. Boyle, M.-H. Giard, O. Bertrand, A. Norena, E. Truy, and L. Collet Evidence of a Tonotopic Organization of the Auditory Cortex in Cochlear Implant Users J. Neurosci., July 18, 2007; 27(29): 7838 - 7846. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Weisz, C. Wienbruch, K. Dohrmann, and T. Elbert Neuromagnetic indicators of auditory cortical reorganization of tinnitus Brain, November 1, 2005; 128(11): 2722 - 2731. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
|