|
|
|
|
______________________________________________________________________ The "Zeitschrift für
Semiotik": Abstracts
_________________________________ René J. Jorna Thomas Goschke William E. Smythe Cees van Leeuwen, Lieselotte van Leeuwen Henk Visser Klaus Mainzer Boris Kaltschew Theo Herrmann Knowledge representation in artifical intelligence. Sign theory and cognition research René J. Jorna, University of Groningen Summary. This article discusses what cognitive science and artificial intelligence have
to do with semiotics. For this purpose, the definition of semiotics (Zeitschrift für
Semiotik 1, 1979) is extended, and the central notions of functionalism, symbol
system, and representation, which constitute the conceptual foundations of cognition and
artificial intelligence, are analyzed. The analysis reveals a variety of applications of
these notions, and it is argued that the fields of cognitive science,
artificial-intelligence research and semiotics will reach more conceptual clarity if they
learn from the strong points of each other.
Knowledge without symbols? The program of "new connectionism" Thomas Goschke, University of Osnabrück Summary. Until recently, mainstream cognitive science was based on the assumption that
cognitive processes can be modelled as rule-governed manipulation of formal symbols. With
the advent of the "New Connectionism" in the last few years, an alternative
research program has been established, which conceptualizes within the framework of
abstract neural networks. In this paper it is argued that a core difference between the
two paradigms concerns the status they give to syntactically structured symbols. Whereas
symbol-oriented models presuppose context-free constituents, connectionist representations
are highly context-sensitive. Psycholinguistic evidence for the context-sensitivity of
meaning in natural language is discussed. Finally, implications of the context-sensitivity
of connectionist representations for the relation between symbolic and connectionist
models are outlined. In particular, arguments are advanced against the attempt to view
connectionist models as mere implementations of symbol-oriented algorithmic models. How can symbols be interpreted? Representations in Frege and Peirce William E. Smythe, University of Alberta Summary. Contemporary psychology has yet to come to grips with the basic function of
symbolic representation, despite the widerspread application of the notion in cognitive
theory. It is argued that the semiotic systems of Gottlob Frege and Charles S. Peirce can
shed considerable light on the fundamental nature of symbolic representation.
Peirce’s analysis in particular reveals some important functional ties between
symbolic representation and interpretation. The present paper attempts to motivate and
develop Peirce’s interpretation-based conception of symbolic representation and to
point to some of its implications for cognitive theory.
How are symbols perceived? The approaches of Helmholtz and Koffka Cees van Leeuwen, University of Amsterdam Summary. It is argued that the perception of symbolic aspects of the environment
requires a dynamic, interactive account of perceptual processing. Two alternative programs
for a theory of perception are evaluated with respect to this requierement, viz. the
approach of Hermann von Helmholtz and the Gestalt-theoretic approach of Kurt Koffka. The
Gestalt-theoretic approach seems better suited to account for symbol perception than the
Helmholtz program.
Mathematic intelligence: natural and artifical
Summary. This article discusses mathematical versus propositional representation. Studies in artificial intelligence and in cognitive modeling suggest that toy problems are solvable with rigid mathematical representations, whereas world problems require flexible propositional representations. In this paper it is argued that world problems, too, should be handled with mathematical representations, although of a special kind. The evolution of intelligent systems Klaus Mainzer, University of Augsburg Summary. The development of human knowledge and knowledge-technology is sometimes
described as "evolution" driven by technical innovations on given levels of
"emergence". The first level was realized by simple tools like hammer, lever
etc. On the next level machines of force and energy were invented. Today
program-controlled computers and information-processing automata have become tools of
everyday-life. Computer scientists distinguish several generations of hardware and
software in the historical development of their machines. In artificial intelligence
research one speaks of the "second computer age", meaning the transition from
number-processing machines to knowledge-processing systems such as expert systems, which
are said to simulate human experts, at least partially. Even biological and neurological
computers are being projected. They are self-organizing complex systems, built according
to structural principles of neurophysiology. In this situation the critical question
arises whether the technological development of artifical intelligence can be understood
as part of the biological evolution of intelligent systems. The article analyzes the
linguistic, logical, mathematical, physical, and biological paradigms which led to the
development of knowledge-based systems.
The sensory realm. Thoughts about image, concept and
language Summary. On the basis of an example taken from chess it is demonstrated that concepts
are of a sensory-mental nature. It follows that they cannot be regarded as part of
language. Isolated from aspects alien to it, language can be defined rigidly.
The partner-related localization of objects in communication. A new research topic between langauge psychology and linguistics Theo Herrmann, University of Mannheim Summary. Up to the present, linguistics as well as psychology of language have largely been neglecting the fact that speech production and other semiotic processes are codetermined by situational factors, especially those pertaining to the listener. The partner-related localization of objects is still a widely unresearched phenomenon. Speakers cause their partner to identify an object by verbally establishing a spatial relationship between the object in question and a second: "Will you please hand me the pair of tongs lying to the right of the box." Localizing objects verbally is often performed egocentrically, i.e. the speaker’s utterance may be supplemented by a phrase like "seen from where I am". If, however, the speaker’s localizing utterance can be supplemented by a phrase like "seen from where you are" we are dealing with a partner-related localization. When will a speaker use the former and when will he use the latter kind of localization? Why does he localize objects in a partner-related manner? What is the nature of the cognitive processes that occur during partner-related localizing utterances? What may either facilitate or impede such processes? Are there any alternatives to partner-related localization by which a speaker may reach his goal as well or even better? The following report presents first considerations and results concerning the partner-related localization of objects. Furthermore, possible ways of investigating the partner-related localization of objects by means of psychological experiments are demonstrated. |
||||||||||||||||||||||||
© 1999-2001, Webmaster Research Center for Semiotics, Institute for Linguistics, Fac. 1, Technical University Berlin, Berlin, Germany |
|||||||||||||||||||||||||
