Information Technology Reference
In-Depth Information
Moreover, the AI, which we have already discussed, is historically one of
the cognitive science fields, after the attempts with the first cybernetics
(general study of the mind's operation, involving as much psychologists and
anthropologists as mathematicians and logicians), the second cybernetics
(study of the cognitive systems and self-organization evolution), then the
system theory that emerged with robotics and expert systems. Its goal is to
develop artificial systems, and it approaches language through
computer-based modeling and simulation [GAR 11]. Even though we now
speak of ambient intelligence much more, this is of course the path of MMD.
Guyomard et al. [GUY 93-06] show how AI provides an additional point of
view to linguistic approaches: while they suppose that the structure of the
dialogue focuses on the utterances themselves, their shape and linguistic
content, the AI, by associating itself with logic and language philosophy,
provides notions of planning, representation, reasoning and speech act to
explain the dialogue structure. It also contributed in defining a plan-by-plan
approach of MMD, also called differential approach. In this section, we will
explore the cognitive science points of view and their potential applications in
MMD around the abilities of perception, representation and learning.
2.1.1. Perception, attention and memory
The human cognitive system has perceptive mechanisms, which allow it to
distinguish, and thus extract, objects from their environment. These
mechanisms are fast, automatic and reliable. They are characterized by
certain remarkable aspects such as invariance: in visual perception, an object
is always recognized, even if it is rotated or if the scale is changed [GAO 06].
Audio perception and spoken language perception also involve specific
mechanisms, which are worth noting from a cognitive point of view, such as
the ability to select a sound source in a noisy environment, for example.
Another example, the essential characteristics of the perception of spoken
language are intensity (loud or soft voice), pitch (high or low voice) and
timbre as a set of aspects such as texture that lead to the speaker being
identified. Visual perception and audio perception have been the subject of an
impressive number of experiments and models in cognitive sciences. The
Gestalt theory, the Adaptive Control of Thought (ACT) model by J. Anderson
and those derived from it, or even more specialized models such as those
focusing on the perception of speech and lexical access (Forster, Morten,
Marslen-Wilson models), were repeatedly presented in literature, see
Search WWH ::




Custom Search