av Thomas_A » 2006-11-03 12:53
Experiment 1:
Ljud alstrat från högtalare med och utan HF-innehåll: Skillnad i EEG + en ljudvolymstest. (De fick sätta en behaglig ljudvolym som var signifikant högre när ljudet innehöll HF jämfört när det var filtrerat @ 22 kHz @ 80 dB/oktav)
Experiment 2:
Ljud alstrat från in-ear hörlurar. Ingen skillnad mellan EEG eller ljud mellan LF och LF+HF.
Experiment 3.
LF-innehåll från hörlurar och HF från högtalare som strålar mot kropp (HF själv var icke hörbart). Signifikant skillnad (men alltså bara tillsammans med LF-innehåll).
Experiment 4. Samma som 3 men både huvud och kropp var akustiskt isolerat från HF från högtalare. Ingen skillnad.
Mätningar från högtalare och hörlurar uppvisar att de ger respons upp till ca 50 kHz. Mätmik är 4939, Brüel & Kjær
Metoder:
The sound stimulus was a traditional gamelan composition, “Gambang Kuta,” of Bali Island, Indonesia, which contains a wealth of high frequencies with a conspicuously fluctuating structure and has been proven to induce the hypersonic effect. A bi-channel sound presentation system (Oohashi et al., 2000 and Yagi et al., 2002) was used to present the sound stimulus (Fig. 2). Using high-pass and low-pass filters (CF-6FH and CF-6FL; NF Corporation, Yokohama, Japan) with crossover frequency of 22 kHz, cut-off attenuation of 80 dB/octave, and ripple in the passband frequency of ±1 dB, we divided the source signals into audible LFC and inaudible HFC, and amplified them independently of each other. These signals were presented simultaneously or separately through speakers or earphones. In most audio systems conventionally used to present sound for determining sound quality, sounds containing HFC are presented as unfiltered source signals through an all-pass circuit, whereas sounds without HFC are produced by passing the source signals through a low-pass filter (Muraoka et al., 1978 and Plenge et al., 1979). Thus, audible LFC are presented through different pathways that might have different transmission characteristics, including frequency response and group delay. In addition, intermodulation distortion might differentially affect LFC. Therefore, any observed differences between the two sounds, those with and without HFC, might well result from differences in the audible LFC presented through the different pathways rather than from the existence of HFC. The bi-channel sound presentation system that we developed for a series of the studies on the hypersonic effect fundamentally overcomes such problems by presenting LFC through an identical pathway in the two sounds, those with and without HFC. Exact specifications of the sound source and the bi-channel sound presentation system (Authentic Signal Disc ARHS9002 and Authentic Hypersonic Sound System, Action Research Co., Ltd., Tokyo, Japan) have been described elsewhere (Yagi et al., 2002).
(37K)
Fig. 2. The experimental setup employed for this study. Using high-pass and low-pass filters with a crossover frequency of 22 kHz and a cut-off attenuation of 80 dB/octave, each of the source stereo signals was divided into LFC and HFC, and the two signals were independently amplified. They were presented either separately or simultaneously through speakers or earphones.
The speaker components of the system were placed approximately 2.0 m from the subjects' ears. Subjects used custom-made closed-air type insert earphones without ear pads. The portion inserted into the ear canal was made of injection-molded hard plastic approximately 2 mm thick. Both the right and left earphones contained two vibratory devices, one for LFC and the other for HFC. Fig. 3 shows the power spectra of the actual air vibration reproduced by the bi-channel sound presentation system and recorded with a microphone (4939, Brüel & Kjær, Nærum, Denmark) at the subject's position. The averaged power spectra of the entire 200-s piece of music used in the experiments were measured on a Fast Fourier Transform (FFT) analyzer (CF-5220, Ono Sokki, Yokohama, Japan) with 2048 sampling points, sampling frequency of 128 kHz, Hanning window, and averaging of approximately 25,000. The HFC alone was presented through speaker or earphone to each subject who, blind to the presentation, was asked to press a button whenever he or she could detect anything at all, not just auditory sensations, other than silence. None of the subjects could distinguish the presentation of HFC alone from silence. This is compatible with the notion that human beings cannot perceive elastic vibrations in the frequency range above 20 kHz as sound (Snow, 1931, Wegel, 1922 and Durrant and Lovrinc, 1977).
Measurement of EEG
In each experiment, there were two trials each for FRS and LFC alone with an intertrial interval of several minutes in an A–B–B–A fashion, in which FRS and LFC alone were assigned to A and B, or B and A, respectively, in a counterbalanced manner across subjects. The presentation of the sound in each trial lasted 400 s, consisting of two repetitions of the entire piece of music. Subjects were asked to keep their eyes naturally open. EEGs were recorded using a telemetric system (WEE-6124, Nihon Kohden, Tokyo, Japan) from 12 scalp sites (Fp1, Fp2, F7, Fz, F8, C3, C4, T5, Pz, T6, O1 and O2) according to the International 10–20 System using linked earlobe electrodes as a reference with a filter setting of 1–60 Hz (−3 dB). The power spectrum of the EEG at each electrode was calculated using FFT analysis for every 2-s epoch with an overlap of 1 s, with a sampling frequency of 256 Hz. The square root of the averaged power level in a frequency range of 10.0–13.0 Hz at each electrode position was calculated as the equivalent potential of EEGs in the alpha 2 band. To eliminate intersubject variability, the data were normalized with respect to the mean value across all time epochs and conditions for each subject. After excluding epochs contaminated by artifacts, the data obtained from 7 electrodes in the centro-parieto-occipital region (C3, C4, T5, Pz, T6, O1 and O2) were averaged across all the analysis epochs (alpha-EEG) and compared between the two conditions: FRS and LFC alone. This EEG index has been shown to correlate significantly with the net activity of the neuronal network in the deep-lying brain structure including the brainstem, thalamus, and hypothalamus, which are considered to be neuronal substrates of the hypersonic effect (Nakamura et al., 2004). Analysis of the whole alpha range (8–13 Hz) gave rise to the same results as that of the alpha2 band (10–13 Hz).
There was a considerable delay from the onset of sound presentation before a significant change in the EEG was observed (Oohashi et al., 2000), thus statistical evaluation (paired Student's t test between FRS and LFC alone) was performed for the latter 200-s half and the last 100-s period as well as for the entire 400-s period of the sound stimulus. The scalp distribution of the change in the alpha 2 component between FRS and LFC alone was also evaluated by constructing colored contour line maps using 2565 scalp grid points with linear interpolation and extrapolation (Duffy et al., 1979 and Ueno and Matsuoka, 1976) based on Z scores calculated from pair-wise comparisons of the strength of the alpha 2 component at each electrode.
CLL-metoden
The same four subexperiments were performed for CLL measurement as we did in the EEG experiments. In each experiment, two sessions each for FRS and LFC alone were performed with an intersession interval of several minutes in a counterbalanced manner across subjects. The experiment was performed under a double blind condition, namely one experimenter controlled the presentation of the experimental sound and another experimenter who did not know the experimental condition gave the instruction to the subjects and measured the listening level. The subjects were also blind as to the conditions of the sessions. One session consisted of 5 trials, each of which was a 200-s presentation of the same sound stimulus. Throughout a single session, either FRS or LFC alone was presented. The listening level was measured as equivalent continuous A-weighted sound pressure level (Laeq) using an integrated sound level meter (LA-5111; Ono Sokki, Yokohama, Japan). Note that a measured sound pressure level is not affected by the existence of HFC above 22 kHz because only the Laeq of sounds below 20 kHz was measured in this method. In fact, FRS and LFC alone showed equal levels of Laeq within ±0.1 dB at the same volume in the present experimental setting. In the first trial, subjects listened to the sound stimulus at a fixed level adjusted to 79.5 dB Laeq at the listening position when the sound stimulus was presented through speakers, and at a subjectively equivalent listening level when it was presented through earphones. During the next three trials, subjects were requested to freely adjust the listening level to what they considered to be comfortable using a remote controller with an up-down switch that controlled the motorized fader (PGFM3000; Penny and Giles, Gwent, UK) positioned between the player and the preamplifier. No visual or tactile information on the volume was given to subjects when they adjusted the listening level. Then, during the final trial, each subject listened to the sound fixed at the level that they had selected at the end of the preceding trial. The level measured in the final listening trial and adjusted in terms of Laeq was considered to be the CLL. Statistical evaluation was made using paired Student's t test between FRS and LFC alone.
