As an aside, it should be noted that mapping neural mechanisms and systems will be
advanced more efficiently if, whenever possible, the coordinate-based analyses just
described replace analyses based only on lower resolution measures such as hemisphere,
lobe, sulcus, gyrus, and, eventually, Brodmann area. Interestingly, close analysis suggests
that much of the seemingly large variability in structure-function correlations remarked
upon by some researchers using surface-based anatomical landmarks may vanish when the
frame of analysis shifts to stereotactic coordinate-based analysis.^88
Be that as it may, hypothesizing an operation (or computation) for an area requires a
principled analysis of cognitive and behavioural processes and tasks. In spite of efforts in
the cognitive sciences, there is no formal language in which to capture component opera-
tions that fractionate a task, including musical ones. A useful first step in that development
would be for circumscribed communities of researchers studying a particular operation,
area, or paradigm(s) in music processing to settle on an effective common vocabulary, with
an eye towards establishing a working standard vocabulary with researchers studying other
specific operations, areas, or paradigms. At the present time, this is rarely done, and appar-
ently never in a concerted fashion. Standardized component models will be of great bene-
fit for the process of mapping operations down to brain areas and up to system-level tasks
in order to compare music and other cognitive processes.
Another significant point with regard to these goals is that whenever possible the opera-
tion hypothesized for a brain area should be consistent with its underlying microcircuitry.
This is also rarely done in cognitive neuroscience. Take, for example, the case of hypothesiz-
ing operations for the cerebellum. The cerebellum has a uniform microcircuitry through-
out, so until proved otherwise, we should assume that it performs a uniform operation, even
if that operation may be performed on different kinds of information (as consistent with
connectivity). Of course, even the most widely accepted facts about microcircuitry in an area
can change from time to time. In the case of the cerebellum, David Marr’s model of cere-
bellar function^94 was shown to be inconsistent with the underlying microcircuitry some
time after its publication (although it continued to be influential).
Finally, there is a misguided tendency when interpreting neuroimaging data to assume
that every activated area is directly involved in the cognitive or behavioural task eliciting it.
A parallel tendency leads to the assumption that every cognitive or behavioural process
affected by damage to a particular brain area necessarily implicates direct involvement of
that area in those processes. As our sophistication in the computational and neural sciences
grows, we will likely recognize that there are a range of indirect roles a mechanism might
assume.
Consider again the cerebellum. The cerebellum is thought by various researchers to be
directly involved in a variety of cognitive and behavioural processes. However, the conse-
quences of injury to, or absence of, the cerebellum are often not fully appreciated. Although
70 – 80 per cent of the neurons in the brain are packed tightly in the cerebellum, the symp-
toms are mild and transient in cases when nearly all of it is removed. Likewise, persons
born without a cerebellum show few impairments. These observations imply that the func-
tion of the cerebellum is subtly supportive and not necessary. On the hypothesis of cere-
bellar function that my colleagues and I are testing,37,73a brain with a healthy cerebellum
is assumed to be optimized for operations necessary for fine sensory acquisition.
263