To avoid confusion and misunderstandings, we adopt the terms symbolic, subsymbolic, and analogical with the following meaning. ``Symbolic'' stands for a representation system in which the atomic constituents of representations are, in their turn, representations. Such a representation system has a compositional syntax and semantics. The typical case of symbolic system is an interpreted logical theory.
We call a representation ``subsymbolic'' if it is made by constituent entities that are not representations in their turn, e.g., pixels, sound images as perceived by the ear, signal samples; subsymbolic units in neural networks can be considered particular cases of this category [317].
The term ``analogical'' refers to a representation in which the constituents and their relations are one-to-one with the represented reality. In this category we include mental models as defined by Johnson-Laird [151], mental imagery and diagrammatic representations. Analogical representations play a crucial role in multimedia knowledge representation. Note that, in this sense, analogical is not opposite to digital: in the hybrid model we are developing, analogical components are implemented by computer programs.
Esprit Project 8579/MIAMI (Schomaker et al., '95)