The Machine in the Ghost: The Syntax of Mind

Abstract

Experience with static, fact-based Knowledge Representation (KR) in past decades has revealed its limitations: it is inexible (for adjustments) and non-portable (knowledge in one domain cannot be directly used in another domain). We believe that dynamical, process-based KR oers better perspectives. Below we start our presentation with a philosophically informed underpinning of a process view of KR. The most important conclusion of this enterprise is that the processes of perception and cognition can be modeled in the same way. This part is followed by the introduction of a cognitively based, semiotically inspired model of KR that complies with our philosophical considerations. As computer science in its core meaning is necessarily fact oriented, due to the limitations of current computers, the proposed theory of KR is based in logic, but this logic is less articulate than predicate calculus (regarded as a language for the specication of recursively computable functions). Note that it is a limitation of syntax, not of expressive power. In knowledge representation, in the broader sense, including problem elicitation and specication, the restricted syntax may turn out to be more practical than the whole (i.e. predicate calculus).