Functional neuroanatomy of the comprehension of
harmony, melody, and rhythm
We pursued these results in our second functional neuroimaging study, in which we
attempted to map the brain basis of the comprehension of individual principal compon-
ents of music per se.^46 Using the same PET methodology as in our first study, my col-
leagues and I imaged eight right-handed, male Ph.D. faculty musicians (none of whom
participated in our first study). These subjects are experts at monitoring musical scores
while listening for accurate performance. Each subject was competent on several instru-
ments, with primary specializations in either winds, strings, voice, or keyboard.
During the experiment, these subjects detected errors in a computer-generated instru-
mental performance of unfamiliar Bach chorales as they read the corresponding score. The
subjects made no overt response during a trial; rather they acknowledged covertly the loca-
tion and presence of specific errors. The errors, which we implanted on each trial, were
exclusively either rhythmic, harmonic, or melodic. The errors occurred unpredictably
every two to three beats. On rhythm trials, errors occurred only in the performed rhythm
of a note in one of the four voices but were otherwise correct. The rhythm errors changed
either the duration or onset of a written note only (leaving unchanged the other aspects of
rhythm such as meter and tempo). On melody trials, errors perturbed a written note in the
highest voice by raising or lowering it a chromatic half step; otherwise they were accurate.
On harmony trials, the errors could occur in any one of the four voices but were otherwise
correct. The harmony errors perturbed the written note by raising or lowering it a chro-
matic half step. Pilot studies were used to develop stimuli that equated for difficulty across
the three conditions. In prescan and postscan performance, subjects were accurately detect-
ing approximately 80 per cent of the errors in each task.
After practicing the tasks, each subject was scanned performing each of the following
five trials twice in pseudorandom order. On harmony, rhythm, and melody trials, subjects
251Figure 17.3 Correlated blood flow activations in right anterior auditory cortex and left cerebellum (see arrows)
as subjects played the Bach score.^21 These are group-averaged PET images for Bach (contrasted with rest) over-
laid on anatomical MRIs. PET data are z-scores displayed on a colour scale ranging from 1.65 (yellow;p0.1) to
4.0 (red;p0.0001). (See Plate 7 in colour section.)