In Chapter 7, the spatial sensations of the sound field are described in terms of these factors extracted from the IACF. These spatial sensations become immediately apparent when we enter a sound field, because our binaural system evaluates the IACF within a relatively short temporal window, as discussed in the next section. This situation is quite different from the adaptive temporal window for the sound signals themselves, which varies according to the effective duration of the ACF of the sound-source signal.

5.5 Auditory Temporal Window for Binaural Processing

The interaural correlation function IACF contains the information needed for localizing a sound in the horizontal plane. When a sound signal is moving such that it changes direction in the horizontal plane, we must identify a suitable “temporal window” 2T for analyzing the running IACF. This binaural temporal integration window corresponds to our ability to track the moving image of the localized sound. The range of τIACC values extracted from the IACF determines the range of conditions under which such a moving image can be followed. The binaural temporal integration window (2T) must be shorter than the period of movement, or the image will be smeared and no movement will be detected. It should also be longer than the interaural delays used for binaural localization because its value fluctuates greatly when 2T is shorter than 1 ms, which is the largest value of τIACC that is possible within the natural physiological range. For a sound source moving sinusoidally in the horizontal plane with a cycling rate of less than 0.2 Hz, 2T may be selected in a range from 30 to 1,000 ms. When a sound source is moving with a cycling rate of 4.0 Hz or less, a binaural integration window of 30–100 ms is acceptable (Mouri, Fujii, Shimokura and Ando, unpublished). In order to obtain reliable results, the recommended temporal window for the IACF that covers a wide range of movement velocities in the horizontal localization is around 30 ms. For measurement of spatial factors at each audience seat location in a concert hall using a sound source that is fixed on the stage, the value of the binaural integration window may be selected to be longer than 1.0 s.

5.6Hemispheric Specialization for Spatial Attributes of Sound Fields

Characterization of the independent influences of the aforementioned temporal and spatial factors on subjective preference judgments has been achieved (Ando, 1983, 1985, Chapter 4).

Recordings over the left and right hemispheres of SVR, EEG, and MEG have revealed the following evidence for functional specialization with respect to these perceptual attributes (Table 5.1).

1.The leftand right-relative peak amplitudes of the first major SVR waves, A(P1– N1), indicate lateralization of responses associated with the temporal factor ( t1 in the left hemisphere) and spatial factors (LL and IACC in the right hemisphere). Correlates of loudness, the sensation level SL and the binaural listening level LL, were classified as temporal-monaural factors from a physical viewpoint. However, the neural correlates of these parameters were observed in slow vertex responses (SVRs) predominantly at electrode locations over the right hemisphere. Thus, in terms of central processing mechanisms, SL and LL should be reclassified as spatial factors.

2.Both the left and right latencies of N2 covary with the IACC.

3.Results of EEG measurements that showed hemispheric lateralization of the temporal factors, i.e., t1 and Tsub reconfirmed left hemisphere dominance for these factors. Similarly, spatial factors associated with interaural magnitude IACC showed right hemispheric dominance. Thus, there appears to be a high degree of independence between the factors that predominant in the neural responses in the left and right hemispheres.

4.Scale values of subjective preferences are well predicted from the values of the effective durations τ e of EEG α-band activity recorded over the left and right hemispheres. Neural correlates of preferences related to changing temporal factors dominate in left hemisphere EEG signals, while those associated with changing spatial factors dominate in the right hemisphere responses.

5.Recorded MEG amplitudes reconfirmed the left hemisphere specialization for first reflection time t1.

6.Scale values of individual subjective preferences relate directly to the effective duration τe of MEG α-band activity. Note that it is the effective durations of the α-rhythms in EEG and MEG recordings, and not the absolute amplitudes of these waves, that correspond to subjective preferences.

7.A right hemisphere specialization was reconfirmed using MEG recordings for the IACC and tIACC (Soeta and Nakagawa, 2006).

In addition to the above-mentioned temporal response patterns, spatial patterns of cortical neural response were analyzed by examining crosscorrelations between α- band activity at different scalp locations in the two hemispheres using EEG (see Fig. 4.24 and Section 15.1.2 in this volume; Sato et al., 2003; Okamoto et al., 2003) and MEG signals (Soeta et al., 2003). The results show that larger areas of the cerebral cortex show α-rhythm activity when preferred sound fields are presented. These observations imply that the brain repeats a similar temporal rhythm in the α-frequency band over greater numbers of neurons when preferences are better satisfied.

It has also been reported that the left hemisphere is mainly associated with identification of temporal sequences (Zatorre and Belin, 2001, Wong, 2002), while the right hemisphere is fundamentally concerned with spatial identifications. It is worth noting that the spatial factor WIACC extracted from the IACF is closely related to spectral features, and thus might be expected to show right hemisphere dominance. Left-hemispheric specialization for speech signals has been reported by a number

of authors using EEG and MEG recordings (e.g., Eulitz et al., 1995, Näätänen et al., 1997, Alho et al., 1998). Because speech signals can be characterized by temporal features of autocorrelation functions (such as τ 1, φ1, τ e, and patterns of local peaks τ 2, φ2, τ 3, φ3) neural responses to speech sounds might be expected to be associated with the left hemisphere under monaural conditions such as telephone listening. A left hemisphere specialization for speech-coded information and a right hemisphere specialization for nonverbal information (Opitz et al., 2000) have been shown using functional magnetic resonance imaging (fMRI). This reinforces the hypothesis that the left hemisphere is mainly associated with speech and identification of temporal sequences, while the right is more concerned with nonverbal and spatial identification (Kimura, 1973; Sperry, 1974).

These observations notwithstanding, when the IACC was changed for speech and music signals that would be expected to produce left hemisphere-dominant responses, the response to this change in a spatial factor was predominantly from the right hemisphere (Table 5.1). Therefore, rather than having an absolute response bias for particular kinds of signals, hemispheric response dominance is relative and depends on the nature of the factor being changed in the comparison pair.