SMART: Structured Memory for Abstract Reasoning and Thinking
An important skill considered part of intelligent behavior is Abstract Thinking and Decision Making which includes thinking about a problem and reaching a decision after reasoning about its different aspects. This aspect of intelligent behavior has not re
- PDF / 182,295 Bytes
- 5 Pages / 430 x 660 pts Page_size
- 75 Downloads / 172 Views
Abstract. An important skill considered part of intelligent behavior is Abstract Thinking and Decision Making which includes thinking about a problem and reaching a decision after reasoning about its different aspects. This aspect of intelligent behavior has not recieved much attention and most of the cognitive architectures present today either focus more on the perceptual-motoric aspects of human brain or delve into the psycological, common behavioral and common sense issues. Here we present a cognitive architecture which addresses the issue of Abstract Thinking and Decision Making by using a novel representation for knowledge in the memory of an agent. The memory of an agent consists of four components, Concept Net, Working Memory, Perceptions and Possessions. Concept Net is a multi-layered net representing various concepts and their relationships with one another. We contend that this way of knowledge representation supports the process of decision making. Keywords: Cognitive Architectures, Abstract Thinking, Decision Making, Knowledge Representation.
1
Introduction
Intelligence is an umbrella activity encompassing many different types of skills. Abstract Thinking and general purpose Decision Making is one such skill and covers thinking and reasoning about a problem and reaching a conclusion after considering all aspects and possibilities related to the problem. For example making a decision by a robotic house worker, “Shall I go to market for buying groceries”. To simulate intelligent behaviours, many different types of cognitive architectures have been proposed and different representations for knowledge have been used in different models. Generally, the two broad categories can be recognized as symbolic and connectionist. Although symbolic representations are very expressive in nature, encoding big domains using such kind of representation is very difficult since for every action and its preconditions and effects, the programmer has to code the details. Such representations are quite useful for addressing perceptual-motoric issues of human intelligence or for building such special purpose systems which naturaly M.A. Orgun and J. Thornton (Eds.): AI 2007, LNAI 4830, pp. 781–785, 2007. c Springer-Verlag Berlin Heidelberg 2007
782
U. Rafique and S.Y. Huang
create a search space for solution (solving puzzles, playing chess, etc) but for the task of abstract thinking and decision making, they are not very appropriate. In connectionist way of representing knowledge, the emphasize is more on “connecting” nodes in different ways. Different architectures use different type of nodes and connect them in different ways but generally one node represents one piece of knowledge and its connections to its neighbors determine how often these nodes occur together and the weight of the link determines the strength. In our view there are other possible ways of connecting different nodes as well and we discuss them in later sections. Most of the architectures proposed to date using this way of representation address issues of general h
Data Loading...
