and then introduce a new technique for measuring cortical activity during the perception
of melodies.^44
Existing approaches to the neuroscience of melody
The study of melody and the brain has relied primarily on three approaches. The first is the
neuropsychological approach. This approach focuses on melodic perception deficits in
individuals with localized brain damage (e.g. from stroke or surgery), in order to determine
if the affected brain regions are involved in aspects of melody perception.45,46For example,
such studies have suggested that left superior temporal cortex is involved in the processing
of pitch interval information, while analogous regions in the right hemisphere have similar
capacities but also play a special role in the perception of melodic contour.19,47
The second method for studying the neuroscience of melody is the evoked potential or
ERP (event-related potential) approach. This approach is based on extracting stimulus-
locked population-level neural activity from the EEG or electroencephalogram.^48 –^50 ERPs
provide excellent time resolution—on the order of 10s to 100s of milliseconds—but due to
the spatial spreading of bioelectric currents by the skull and scalp it is extremely difficult to
localize the brain sources of ERPs. Thus ERPs are not used for localization, but to examine
the neural response to individual tones in melodies in fine temporal detail. Initially it might
seem that ERPs are ideally suited to studies of melodic processing, since they can resolve
the details of neural responses to single tones. However, the low signal-to-noise ratio of
evoked potentials means that they must be extracted from the EEG by averaging, that is,
by repeating a similar stimulus many (e.g. 30–60) times and averaging the brain’s response
time-locked to stimulus onset. Thus ERP studies have not been used to study the brain’s
response to successive tones in a single melody, but have focused instead on responses to a
particular tone in multiple repetitions of similar melodies (e.g. an out-of-key note at the
end of a melody).iDespite this limitation, ERPs have proved a valuable tool: for example,
they have demonstrated the influence of musical training on the brain’s processing of
tonality relations.^51
The third approach to melody and the brain is the haemodynamic approach, based on
techniques such as positron emission tomography (PET) and functional magnetic reson-
ance imaging (fMRI). These methods provide indirect measures of neural activity via the
measurement of metabolism and/or blood flow in different regions of the brain. In con-
trast to the ERP method, these techniques provide excellent spatial resolution, and can
identify specific brain regions whose activity is modulated by particular types of perceptual
processes. For example, research by Zatorre and colleagues points to regions in the right
inferior frontal gyrus and the right superior temporal gyrus, which are involved in the
maintenance of a tone in memory during the perception of a melodic sequence.^52
Unfortunately, both PET and fMRI have relatively poor temporal resolution since blood
333iTheoretically, there is no reason why the responses to each tone in a repeated melody could not be examined.
However, this would require a listener to hear the same melody dozens of times in an experimental session. This
could lead to perceptual/neural responses which are quite different from those occurring in natural listening
situations.