Computational approach to social knowledge

Abstract

Social and cognitive sciences have always faced the choice: either to meet the methodologi- cal standards given by successful natural sciences or to rely on their own. Talking about the conversion of knowledge into technology, the second way did not bring great success. The first way implies two alternative opportunities: reductionism or discovery of proprietary general laws. None of these chances have been realized with any satisfactory results, too. Methodological analysis shows that, to achieve significant progress in social sciences, what is missed there is not new facts or definitions but new conceptual schemes. The reason, as the author supposes, is the nomothetic approach being applied to systems with high degree of complexity and hierarchy. If we assume that social structures and processes are built upon cognitive psychological structures and processes, the former inherit the distributed computational architecture of the latter. The paper analyzes various conceptions of computations in order to determine their relevance to the task of building computational social science. The author offers a “generic” definition of computations as a process carried out by a computational system if the latter is understood as a mechanism of some representation. According to the author, the computationalization of social science implies “naturalization” of computations. This requires a theory that would explain the mechanism of growing complexity and hierarchy of natural (in particular, social) computational systems. As a method for constructing such a science, a kind of reverse engineering is proposed, which is recreation of a computational algorithmic scheme of social tissue by the determination and recombination of “social primitives” – elementary operations of social interaction.