19
MUSICAL SOUND
PROCESSING: EEG
AND MEG EVIDENCE
Abstract
Even during the performance of a simultaneous task unrelated to the sounds, the human auditory
cortex is able to model precisely the invariances of the acoustic environment. Data acquired in mis-
match negativity (MMN) paradigm have shown that temporally and spectrally complex sounds as
well as their relations are automatically represented in the human auditory cortex. Furthermore, the
MMN data indicate that these neural sound representations are spatially distinct between phonetic
and musical sounds within and between the cerebral hemispheres. Majority of the MMN studies were
conducted in pitch dimension but also temporal aspects of sound processing are under increasing
experimentation. Up to some extent, also musical expertise is reflected in sound representation accu-
racy as indexed by the MMN.
Introduction
Sound perception and cognition can be conceptualized as a process, in which the first
stages are the least and the latest stages the most dependent on the attentional efforts of the
listener.1–3For a long time, due to micro- and macro-level complexity of the acoustic sig-
nal, music sound encoding was assumed to require attentional effort of the listener. In the
present review, some of these assumptions will be challenged.
In the majority of the studies to be reviewed below, the mismatch negativity (MMN)
paradigm was utilized.^2 The MMN is evoked by an infrequently presented sound
(‘deviant’) differing from the frequently occurring stimuli (‘standard’) in one or several
parameters. Its presence implies that the invariant parameters of the standard sound were
neurally encoded, thus producing a neural mismatch with those of the deviant sound. In
other words, the MMN is an index for the neural traces of short-term auditory memory.
The MMN can be recorded even when the subject is performing a task unrelated to the
stimulation under interest such as reading a book or playing a computer game. Thus it
offers a direct measure for the similarity of neural sound representations, without being
contaminated by differences, for instance, in attentional or motivational involvement of the
subject during an experimental session or between the subject groups.
Yet, several data indicated that the MMN parameters closely correlate to the indicators
of the subjects’ perceptual accuracy as determined in a separate behavioural session. For
instance, the MMN amplitude and latency reflect perceptual accuracy, as determined by