DocumentCode
3241697
Title
The creativity mechanisms in embodied agents: An explanatory model
Author
Wiedermann, Jiri
Author_Institution
Inst. of Comput. Sci., Prague, Czech Republic
fYear
2013
fDate
16-19 April 2013
Firstpage
41
Lastpage
47
Abstract
In his recent writings, David Deutsch has argued that no progress in AGI is possible unless a breakthrough in philosophy will occur - a new epistemological theory explaining how brains create explanatory knowledge will be found. The purpose of the present paper is to present first ideas leading to such a theory within the framework of a high-level model of an embodied cognitive agent. In order to support creativity an agent must have the ability to draw and understand analogies which may be used in its otherwise standard reasoning processes in place of facts that served as the basis of the analogy. The use of analogies may then lead to the derivation of new explanatory knowledge in form of conjectures that can further become a subject of criticism, testing and adjustment or rejection. In the respective creative processes embodiment plays an indispensable role. We show that the existing model of an embodied cognitive agent designed recently by the author can serve as an explanatory model elucidating the mechanisms of creativity. Technically, the process of drawing analogies and their use in derivations of explanations can be seen as a means of mining new knowledge originally rooted in the environment. This can boost the computational power of cognitive systems above that of the classical Turing systems.
Keywords
cognitive systems; data mining; inference mechanisms; multi-agent systems; AGI; classical Turing systems; cognitive systems; computational power; creativity mechanisms; drawing analogy; embodied agents; embodied cognitive agent; epistemological theory; explanatory knowledge; explanatory model; high-level model; knowledge mining; reasoning processes; Abstracts; Computational modeling; Mirrors; Oceans; Pragmatics; Semantics; Syntactics;
fLanguage
English
Publisher
ieee
Conference_Titel
Computational Intelligence for Human-like Intelligence (CIHLI), 2013 IEEE Symposium on
Conference_Location
Singapore
Type
conf
DOI
10.1109/CIHLI.2013.6613263
Filename
6613263
Link To Document