Social Embeddedness –
Origins, Occurrence and Opportunities

A picture of two chimps Looking under a leaf Image of robots
a half-day tutoprial to be held at SAB 2002
University of Edingurgh, the morning of the 10th August 2002

Introduction to topic

It is now well established that the embedding of creatures or animats (software or robotic) in their physical environment is crucial to understanding and implementing adaptive behaviour.For social animals the social environment can be as important as the physical environment, serving many important functions, e.g.: it is an important “computational resource” which can be exploited by the individual; it can generate overwhelming complexity and uncertainty beyond the capability of the individual; and it can be the vehicle for trans-individual adaptivity.There are many parallels between physical embedding and social embedding, for example: the exploitation of social environmental phenomena as a complement to individual ability (e.g. learning from and imitating others in addition to individual learning capabilities); the use of rapid sampling of the environment instead of attempting to maintain predictive maps (e.g. gossip); and the development of regulatory systems that extend beyond the individual.However there are also important differences, including issues of internationality, agencyand minds (treating others as “objects with a viewpoint/opinion/mind”).

Aim

The aim of the half-day tutorial is to introduce participants to some of the major themes and issues to do with Social intelligence in general and social embeddedness(SI) in particular, and social systems as it impinges on the understanding and implementation on adaptive behaviour in creatures and animats.The tutorial will continually contrast the agent and society viewpoints in an effort to get a handle on the combined individual-societal system and some of its processes.

Style

Socially Intelligent Agents (SIA) is not yet a fully established field - it exists at the intersection of many other disciplines including: robotics, cognitive science, artificial intelligence, ethology, sociology, and computer science. As a result the tutorial will not be just a sequence of presentations but include question and discussion sessions that involve the audience.

Outline

1.    Introduction

2.    Social Embedding - The Societal Viewpoint

2.1.    The nature of social embedding (SE)
Analogy with embodiment and situatedness; interaction with environment; beyond one-shot & off-line interaction; web of social interaction; modelling stance towards characterising SE.
2.2.    Some of the causes of SE
Co-evolution of social entities & abilities; parallel evolutionary processes (biological, cultural and neural); cognitive arms races; cultural adaptation to fit biological niches; development of expoitable computational and informational resources in the society.
2.3.    Some of the consequences of SE
Impossible modelling burden for individual; importance of naming; importance of local communicative mechanisms; complexity of society and the individual; simple coping strategies (imitation, rapid sampling, games, use of proxies);emergence of new context and niches; emergence of heterogeneity.
2.4.    Some different ways understanding SE systems
A prior vs. descriptive; bottom-up vs. top-down; different sources (philosophical, economic, ethology, ethnology, biology); different focus processes (biological processes, cognitive processes, 1-1 social interaction, social institutions and processes); different styles of model (descriptive, mathematical/logical, computational, philosophical); trans-individual entities and processes.
2.5.    Existing modelling approaches
Economic; game theory; population dynamics; sociological theory; memetics; Alife; social robotics; social simulation; biological “models”, models from physics (e.g. self-organised criticaility).
2.6.    Example:a stock market
Imitation, arms-races, gossip and signalling, deception games, proxies, market “moods”, statistical models, chaos models, agent-based models, unpredictability, emergence of unpredictability and heterogeneity, limitations of design stance, learning, fashion.
2.7.    SE in existing social societies
Ants; song birds, primates, humans; agents; robots; mixed societies.
2.8.    Discussion

3.    Break

4.    Social Embedding - Implications for the Individual and its Interactions

4.1.    Phylogenetic and ontogenetic origins of social intelligence (SI)
The Social/Machiavellian Intelligence Hypothesis, social situatedness and social embeddedness, origins of human societies, the role of SI in the evolution of human intelligence
4.2.    Examples of SI in humans, other primates, and other animals
Primate politics, alliances, communication and cooperation, language in non-human animals (e.g. bonobos, parrots)
4.3.    Social learning and imitation in animals
Social learning mechanisms, conspecifics as social tools, definitions of imitation, agent-based perspective to imitation, imitation research in biology and psychology
4.4.    Examples of imitation and social learning in robots and software
Programming by example as a new human-computer programming paradigm, imitation research in robotics, open research challenges
4.5.    The relationship of SI and “theories of mind”
Mindreading, simulation-theory versus theory-theory, empathic understanding, attribution of intentionality and agency, folk psychology, role of anthropomorphism in designing SIA's
4.6.     SI and the origins of culture
Culture and imitation, examples of non-human culture (chimpanzees, cetaceans), implications for agent culture, culturally adaptive agents, agents that support cultural diversity
4.7.    Discussion

5.    Conclusion

The Handout and PowerPoint Slides (will be posted after the tutorial)

The Tutors

Kerstin Dautenhahn,
Adaptive Systems Research Group
University of Hertfordshire
Department of Computer Science
Whiteknights, PO Box 225
Reading, RG6 6AY. UK.
E-mail: K.Dautenhahn@herts.ac.uk
Fax: +44 (0) 1707-284-303
Tel: +44 (0) 1707-284-333		 Bruce Edmonds,
Centre for Policy Modelling,
Manchester Metropolitan University,
Aytoun Building, Aytoun St.,
Manchester M1 3GH. UK.
E-mail: bruce@cfpm.org
Fax: +44 (0) 161-247 6802 
Tel: +44 (0) 161-247 6479

Kerstin Dautenhahn is Reader in Artificial Intelligence in the Department of Computer Science at University of Hertfordshire. She is editor of “Human Cognition and Social Agents Technology” (John Benjamins Publishing Company, 2000) and “Imitation in Animals and Artefacts” (MIT Press, to appear in April 2002.) Previously, she gave a half-day tutorial "On Minds and Agents: Social Intelligence in Animals and Artifacts" at Autonomous Agents 2000, Fourth International Conference on AUTONOMOUS AGENTS (Agents 2000), Barcelona, Spain, June 2000, and a two-hour tutorial “Socially Intelligent Agents - From Animals to Animats” within the Joint Tutorial Programme of SAB2000: 6th Int'l Conf. on the Simulation of Adaptive Behavior and PPSN2000: 6th Int'l Conf. on Parallel Problem Solving From Nature, Paris, France, Sunday 17th September 2000.

Bruce Edmonds is Senior Research Fellow at the Centre for Policy Modelling, a research unit that specialises on social simulation.His areas of research include: the methodology of social simulation, evolutionary computation, and context-dependency. He is a co-editor of the Journal of Memetics - Evolutionary Models of Information Transmission.He was part of the team that developed SDML a development environment for agent-based social simulation.He has recently co-edited special issues on the topics of "Computational Memetics" and "Context in Context".

| Top | SAB'02 | Other SAB'02 Tutorials | KD Home | BE Home |