Academic
Publications
Talking Nets: exploring a multi-agent, connectionist model of distributed communication and cognition

Talking Nets: exploring a multi-agent, connectionist model of distributed communication and cognition,Francis Heylighen,Frank Van Overwalle

Talking Nets: exploring a multi-agent, connectionist model of distributed communication and cognition  
BibTex | RIS | RefWorks Download
Summary This research project is proposed by a multidisciplinary team led by F. Heylighen and F. Van Overwalle. Its members have expertise in cognitive science, psychology, AI, philosophy, economics, political science, and linguistics, and advanced research experience in connectionist simulation, multi-agent systems, computer-mediated communication and group experiments. The project aims to further test and develop our new "Talking Nets" simulation model of distributed cognition published in Psychological Review (Van Overwalle & Heylighen, 2006). Distributed cognition concerns the information processing and learning that occurs on the social level, by the propagation of information from agent to agent across a social network. Our simulation has accurately replicated the results of half a dozen key experiments in social psychology, illustrating such effects like persuasion, polarization and the reinforcement of stereotypes during group communication. The simulation models individuals as recurrent connectionist networks who are themselves connected by a social network of trust-based links. Both individual and social networks learn from experience, by strenghtening links whose expectations are confirmed. While the model is relatively simple and apparently powerful in predicting known effects, its assumptions and implications are broad, abstract and complex, requiring extensive further testing and exploration. We propose to investigate the remaining issues using four complementary methodologies: 1) multi-agent simulations, 2) individual psychological experiments, 3) group experiments, 4) development and monitoring of Internet-based social software to support distributed cognition. These methods should clarify at least the following issues: What is the precise role of trust in communication? Which social structures emerge from trust- based links? Which information is preferentially propagated across such network? Under which circumstances is the eventual consensual belief better ("collective intelligence") or worse ("groupthink") than the individual beliefs?
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.