You are here


A Framework for Mixed-Level Social Simulation with Socio-cognitive Agents
Jan 2014 - Dec 2016
Project description

The aim of the project is to contribute to the development of conceptual and technological foundations for agent-based simulation of socio-cognitive systems where the interplay between the behaviour of the individuals and their social context is the key feature. We will formalize the conceptual elements necessary to model socio-cognitive systems and, based on those elements, propose the technological artefacts needed to implement the simulations.

Social simulation has been used extensively as a mechanism to study many aspects of human sociology like organizations, social relations, trust and reputation or normative systems among others and also as a tool for policy-making and governance. Computer social simulation is in fact the keystone of Analytical Sociology, a subfield of sociology that tries to explain macro sociological facts not merely by relating them to other macro-level facts, but by detailing in clear and precise ways the micro mechanisms through which they were brought about [Hedström].

The main characteristic of social simulation is that the simulated elements are not simple entities whose aggregated behaviour can be described using mathematical equations. Every individual is unique, needs to show an autonomous behaviour and interacts with the other individuals. This focus of attention is what makes the multiagent systems paradigm predominant in social simulation. On the other hand, in a socio-cognitive system there are social structures acting at a macro-meso-micro level that influence the behaviour of the individuals. These social elements are not independent of the individuals behaviour. The individual is influenced/constrained by the social elements but at the same time, by using his/her cognitive capabilities, can act/modify/ignore them. As N. Gilbert points out Humans differ from atoms and ants in that they are able to comprehend macro patterns (what sociologists call institutions) and these institutions can change individual behaviour.

We need to understand the dynamics of the interaction between the two levels (cognitive and social). Taking this dependency among levels into account is what is called the mixed-level approach [Sun et al., 2005]. The study of these dynamics is still in its infancy and, given its importance, there are many efforts from the social sciences in that direction. Our approach builds upon the conceptual and technological state of the art on computer mediated socio-cognitive systems: (i) social constructs like trust, reputation and punishment, (ii) mechanisms for social coordination like organizations and normative systems and (iii) models of cognition for social behaviour.

The general objectives of the project are to provide conceptual foundations for modelling social phenomena from a mixed-level approach and to propose means to model and implement mixed-level simulations of social phenomena.