Information Technology Reference
In-Depth Information
program which executes these algorithms, we are therefore exploring the range of
works within this class, which can enhance our understanding of their properties.
Besides the formal benefits offered by describing a style in an algorithmic form,
this also serves to reveal selective bias within the application of these procedures.
It is distinctly possible that artists fail to follow one pathway in some creative ter-
rain due to their tendency to automatically follow a more normative path, as trodden
by previous artists or by themselves on previous occasions. Like many tools, algo-
rithmic descriptions of music are likely to emphasise existing tendencies, some of
which the composer may previously been unaware of; conversely, there are many
examples in the field of empirical musicology (e.g. Huron,
2006
) in which algorith-
mic processes reveal novel patterns.
We might also create conjectural models based on emergent cognitive properties
of music perception, such as those of Narmour (
1990
), Temperley (
2007
) and Wool-
house (
2009
). Rather than construct a descriptive system through stylistic analysis,
this approach incorporates sensory capabilities such as patterns of auditory percep-
tion that exist
behind
traditional systems of musical composition—the systems be-
neath the systems. Such models allow us to reflect on the meta-reasoning behind
whole classes of compositional style, such as the Western diatonic tradition.
We can likewise develop our insight into wider cognitive processes through com-
putational simulation. Tresset and Leymarie's
Aikon-II
14
creates facial sketches by
observing the subject's salient features and drawing with a mechanical plotter on
paper, visually perceiving the sketch as it draws. The project aims towards gaining
an understanding of our own observational mechanisms by computationally imple-
menting them, and in doing so illuminating any irregularities in the theory that may
not exposed by contemplation.
The above approaches can be viewed as applied forms of cultural study, serving
to illuminate historical and social tendencies on a broad scale. Following Boden's
(
2004
) distinction between H-creativity (historical creativity, novel to an entire his-
torical frame) and P-creativity (personal creativity, novel only to its creator), we de-
scribe this pursuit of understanding through cultural modelling as
H-introspection
.
Its counterpart is
P-introspection
, which applies to tools used to reflect and un-
derstand the user's personal creative acts. An example of P-introspection is Pachet's
Continuator
(Pachet
2003
), which uses a Markov model to reflect a player's perfor-
mance style through its statistical properties. The approach taken by the Continua-
tor is what Spiegel (
1981
) describes as “extrapolation”: the “extension beyond that
which already exists in such a way as to preserve continuity with it, to project from
it...”. The high-level characteristics of a style are maintained, whilst creating new
works “projecting” from the original.
By mirroring certain properties in such a way, the player may become attuned to
features that they were not aware they exhibited, leading towards a more insightful
mode of creative development.
