296 phonemes and chords. These data thus suggest that there are distinct cortical areas special-
ized in representing phonetic and musical sounds in both hemispheres. In other words,
the functional specialization does not only cover stimulus complexity (as shown by Alho
et al. comparing MMNm generator loci between single tones and chords)^30 but also the
informational content (phonetic vs musical).^29 However, this specialization is not present
prior to memory-related auditory processing as suggested by the dissociation between the
present P1m and MMNm data.
Additionally, the data reviewed above indicated that in the right hemisphere, the MMNm
was larger in amplitude for changes among chords than among phonemes. However, in the
left hemisphere, no corresponding dominance for phoneme changes was found when
compared with chord changes. This result was obviously discrepant with the previous
results.31,32To solve this discrepancy, a further study was conducted with PET technique
(see footnote b).
This PET study^24 used the phonetic and musical sounds developed for the MEG study
described above.^29 However, the phoneme duration was prolonged from originally used
200 ms until 400 ms, since in pilot studies, presence of deviant phonemes elicited no
changes in brain metabolism with phonemes of 200 ms duration. The subjects were con-
centrating on classifying the gender of visually presented words while they were presented
with sound sequences consisting of (1) both deviant and standard sounds or (2) standard
sounds only (phonemes and chords in separate sequences). The data, obtained after sub-
tracting the activity elicited by the standard sounds only from that elicited by both deviant
and standard sounds when intermixed, showed that the change from vowel /e/ to /o/ was
processed in the left auditory cortex, more specifically, in the middle and supratemporal
gyri. In a mirror-like manner, the change from A major to A minor chord was processed in
the right auditory cortex, in the supratemporal gyrus. These data thus indicate that hemi-
spheric specialization for phonetic vs musical processing, previously seen in dichotic-
listening studies and brain-imaging studies using an active task, may be present even
during the performance of a task unrelated to the sound stimulation. However, this phe-
nomenon is very vulnerable to changes in stimulus parameters as indexed by the impor-
tance of an adequate sound duration24,29(see also Refs 36–38 for recent evidence about the
effects of acoustic noise on lateralization of phonetic processing).
Sound complexity and perceptual accuracy
All natural speech and instrumental sounds consist of several parallel harmonic partials,
which, despite their wide frequency range, produce a percept of one single sound. The pres-
ence of harmonic partials is known to facilitate pitch naming in the possessors of absolute
pitch (for a review on absolute pitch, see Ref. 39) as well as the tuning of music instru-
ments.^40 Only recently it was systematically investigated whether harmonic partials facili-
tate pitch discrimination in ignore and attend conditions in nonmusicians.
bLocalizing the active neural population with MEG (and EEG) necessarily involves inverse modelling of the active
neural source. Accuracy of this operation depends from the adequacy of the head model available.^33 This source of
uncertainty can be avoided by using brain-imaging techniques such as PET and fMRI. Those methods detect directly
the locus of activation change in brain metabolism, presumably caused by increased neural activity.34,35