Information Technology Reference
In-Depth Information
4.7
Conclusion
To date, the design of a control strategy of sound synthesis processes that uses a
BCI is still a challenging perspective. As discussed in (V
e et al.
2013
), a
synthesis control of sounds directly from the brain through the measurement of its
cerebral activity is still in its early stages. In particular, the mapping between
electrophysiological signal features and synthesis parameters is generally validated
on the basis on different metrics depending on applications. However, the de
ä
ljam
ä
nition
of such metrics implies a given conception on the way we interact with the sur-
rounding world.
To broach this issue, we introduced two conceptual approaches inspired from the
representational-computational and the enactive paradigms from cognitive neuro-
science. In light of these paradigms, we revisited the existing main approaches for
synthesis and control of sounds. In fact, the viewpoints adopted to synthesize
sounds are intricately underpinned by paradigms that differ in the epistemological
positions of the observer (from a third or a
first-person position) and have a sub-
stantial consequence on the design of a control strategy (cf. Figure
4.2
). On one
hand, synthesis processes based on the modelling of physical sources (from either
the mechanical behaviour or the resulting vibration) are controlled by physical or
signal parameters. This approach is based on the existence of a correct represen-
tation of the physical world and introduces the notion of an error function between
the model and the physical reality as a quality criterion. Therefore, it requires a
certain expertise from the end-user. On the other hand, synthesis processes based on
the modelling of perceptual effects involve the identi
cation of invariant sound
morphologies speci
c to given perceptual attributes of the sound source. This
approach assumes the emergence of an embodied auditory world from an enactive
process. The perceptual judgments are considered as a quality criterion for the
model, leading to the design of a more intuitive control.
By associating these conceptual and pragmatic considerations, we proposed a
prospective view on the methodology to be used to design a BCI control. For the
sake of illustration, we treated limited aspects of BCIs by addressing explicit BCI
from the representational-computational point of view and implicit BCI from the
enactive point of view. Actually, we are aware that the frontier between explicit and
implicit BCI might be dif
cult to establish and less didactic than what this article
presents. Indeed, the implicit communication channel might sometimes be used in
an explicit way (George and L
cuyer
2010
), and inversely brain plasticity can
enable the participant to make use of the training experienced from the explicit BCI
to generate implicit recurrent sensorimotor patterns (Bach-y-Rita and Kercel
2003
).
With current apparatus performances, the rate of transfer information between the
BCI and the device is quite limited and the final task has to be defined accordingly.
While this technique may represent a restricted interest for healthy users (in some
cases, it would be easier to directly control the device manually), it constitutes a
relevant medium for medical applications and can be used as a substitutional device
for diseases. In the implicit BCI, the control is included in an optimization system in
é
