68 |
4 Electrical and Magnetic Responses in the Central Auditory System |
the SVR and EEG studies of first reflection time t1, the left-hemisphere response is dominant. An almost direct relationship between individual scale values of subjective preference and the τ e values over the left hemisphere was found in each of the eight subjects. Results for each of the subjects are shown in Fig. 4.29. Remarkably, the correlation coefficient, r, was 0.94 over all subjects. However, as shown in Fig. 4.28b, there was only a weak correlation between the scale values of subjective preference and power in the α-band, Φ(0), for either hemisphere (r < 0.37).
The effective duration τ e in the alpha wave band reflects the persistence of alpha rhythms in time, so that under the preferred listening conditions, the brain repeats a similar rhythm of alpha activity for a longer period of time. This tendency for a longer effective duration τ e of the alpha rhythm in the preferred condition is much more significant than the aforementioned results in Section 4.3 that were obtained through similar analyses of EEG signals.
4.4.2 Preferences and the Spatial Extent of Alpha Rhythms
Magnetic responses were also analyzed using crosscorrelation functions (CCF) between 36 reference channels and 35 test channels. In MEG measurements using the word, “piano” as the source signal, combinations of a reference stimulus ( t1 = 0 ms) and test stimuli ( t1 = 0, 5, 20, 60, and 100 ms) were presented 50 times alternately at a constant 1-s interstimulus interval (Soeta et al., 2003). Eight 23to 25-year-old subjects participated in the experiment. The scale value of the subjective preference of each subject was obtained by paired comparison (PCT) also.
Results from this experiment showed that (1) the maximum amplitude of the crosscorrelation function CCF, |φ(t)|max, between alpha band signals (8–13 Hz) recorded at two different channels increases with increasing subjective preference, and (2) the maximum amplitudes of channel crosscorrelations decrease with increasing channel distance. These imply that when listeners are stimulated using preferred sound fields, alpha rhythms persist over wider cortical territories, and that there is a higher degree of alpha rhythm coherence over these larger areas.
MEG experiments also reconfirmed, using the same speech signal with changing interaural correlation magnitudes (IACC = 0.27, 0.61, and 0.90), that effective durations τ e and maximum MEG channel crosscorrelation amplitudes increased when the IACC decreased (Soeta et al., 2005).
4.4.3 Alpha Rhythm Correlates of Annoyance
In addition to the neural correlates of preferred listening conditions, one can also study distinctly non-preferred, annoying sounds and listening conditions. We undertook a series of MEG experiments similar to those described in the last section to measure neural responses to annoying stimuli: pure tones and band-pass noises with center frequencies of 1,000 Hz (Soeta et al., 2004).
In order to control the ACF of the source signal, the bandwidth of the noise, centered on 1,000 Hz, was varied at five levels (0, 40, 80, 160, and 320 Hz) using
4.4 Magnetoencephalographic (MEG) Correlates of Preference and Annoyance |
69 |
Fig. 4.29 Correlations between scale values for first reflection time t1 preference and effective durations τ e of MEG alpha rhythms values for eight individual subjects. The individual preference curves (dashed lines and open circles) and the effective durations τ e of MEG alpha rhythms from the left hemisphere of the same individual subject (solid lines, filled circles) are superimposed. The averaged effective duration τ e value and the scale value was the highest correlation over the eight channels. Error bars show standard errors. Correlations between preferences and alpha rhythm effective durations are shown for each individual
70 |
4 Electrical and Magnetic Responses in the Central Auditory System |
an extremely sharp filters with slopes of 2,000 dB/octave (Sato et al., 2002). We employed such filters in consideration of the sharpening effect previously observed at the level of the inferior colliculus for higher frequencies (Katsuki et al., 1958). It is worth noting that a filter with a slope of 60 dB/octave is too small to apply to any acoustic measurement. The “0 Hz” bandwidth signal was produced with an actual filter, for which the cutoff frequencies of a high-pass filter and a low-pass filter were both set at 1,000 Hz, so that only the slope component remains. The sound pressure level for all source signals was fixed at 74 dBA by measurement of the ACF, Φ(0). Source signals were characterized by the ACF temporal factors: τ 1, ø1, and τ e. The dependent factors, ø1 and τ e, can be controlled by the bandwidth of the source signal.
Seven 22to 28-year-old subjects participated in this experiment. Paired comparisons were performed for all combinations of 15 pairs in a single session, and subjects were run in a total of ten sessions. The signal duration was 2 s with rise/fall times of 10 ms. Subjects were asked to judge which of the two sound signals was more annoying, and thus the scale value of annoyance for each individual subject was obtained (Ando, 1998). The same subjects who participated in annoyance tests also participated in parallel MEG recording experiments in which the paired stimuli which were presented in exactly the same way. Combinations of the reference puretone stimulus and a test noise stimulus were presented alternatively 30 times and their MEG responses were recorded. Eighteen channels located around the temporal area in each hemisphere were selected for the autoand crosscorrelation analysis of alpha wave activity in the 8–13 Hz range. Examples of recorded MEG signals are shown in Fig. 4.30.
Fig. 4.30 Examples of recorded MEG signals in response band-pass noise with center frequency of 1,000 Hz and an extremely narrow bandwidth (Soeta et al., 2004). Eighteen channels located around the left and right temporal regions (enclosed areas) were selected for alpha-band autocorrelation and crosscorrelation analysis
4.4 Magnetoencephalographic (MEG) Correlates of Preference and Annoyance |
71 |
A two-way ANOVA showed significant effects of stimulus (p < 0.01) and subject (p < 0.01). Thus, we first discuss the effects of stimulus and individual difference. A remarkable finding, shown in Fig. 4.31, is that relative degree of annoyance (difference in scale value) was inversely correlated with the effective duration τ e of the band-pass noise relative to that of the tone (the ratio of these effective durations). The correlation coefficient of this relation was r = −0.83. Not unexpectedly, annoyance behaves in a manner opposite that of subjective preference, because effective duration τ e increases with increasing subjective preference. Thus, effective durations τ e were shorter for annoying stimuli. Also, it was found that the MEG channel crosscorrelation magnitudes |φ(τ )|max in the alpha band decreased with increasing annoyance. Thus, when an annoying stimulus is presented, the brain is not relaxed either temporally or spatially. Previous studies of EEG and MEG responses showed that effective duration τ e increased significantly with increasing preference. This signifies that the brain is repeating a similar (alpha) rhythm over a wider area under the preferred conditions. It is remarkable that the sites that signify the highest correlation between the scale values of annoyance and the values of τ e were observed in the right hemisphere for all of subjects who participated. This implies a right hemispheric dominance for responses to noise and other non-verbal stimuli (Chon, 1970).
Fig. 4.31 Relation of alpha rhythm persistence and annoyance. Differences in annoyance ratings of band-pass noise and pure tones are plotted as a function of the ratios of the effective durations of evoked MEG alpha rhythms for the two stimuli. Taking the difference of scale values of annoyance [SV(band-pass noise)–SV(pure tone)] and ratios of effective duration alpha rhythm durations normalizes these responses so that their relationship can be compared across subjects. Different symbols signify results of different subjects. A strong negative correlation (r = −0.83) between the scale value of annoyance and the effective durations of MEG alpha rhythms was found. The shorter period of time the alpha rhythm persists, the greater the annoyance rating
72 |
4 Electrical and Magnetic Responses in the Central Auditory System |
Fig. 4.32 The scale value of annoyance as a function of the bandwidth of band-pass noise of individual subjects. Different symbols of signify results of different subjects
Although there is a deep relationship between annoyance and effective durations τ e of the MEG alpha band responses in each subject, annoyance itself differs between individuals. As shown in Fig. 4.32, large individual differences in the scale values of annoyance resulted as a function of bandwidth within the critical band. In this context of critical bands and loudness summation, it is also worth noting that evoked magnetic responses show N1m amplitudes that correspond well to subjective loudness judgments in the frequency range between 250 and 2,000 Hz (Soeta et al., 2006).