Adaptive Behavior, 7 (3/4)

We present computer simulations and analysis for the global behavior arising from a population of heterogeneous social agents with bounded rationality. The particular model studied, termed the "bar-attendance" model, offers a simple paradigm for such complex adaptive systems involving competitive agents. The model considers p adaptive agents, each possessing n prediction rules chosen randomly from a pool of N, who attempt to attend a bar where the seating capacity is s. The global attendance time-series x(t) has a mean near, but not equal to, s. Suprisingly, the standard deviation or "volatility" of x(t) can show a minimum with increasing adaptability of the individual agents. Various arguments based on random walk models are discussed. It is shown that effects of crowding have to be included in order to understand the volatility in this system.

Pages 255-268

The Process of Crescive Legitimation: Theory, Simulation Model, and Three Empirical Tests

By Chanoch Jacobson

Abstract

It is argued that the source of crescive legitimation lies in social conditions that permit nonconformities to normative expectations, so that unsanctioned violations increase and spread. This, together with the paucity of palpable reactions create the positive feedback that propels legitimation. Crescive deligitimation, the reverse process, orginates in open challenges to legitimate objects when they appear to conflict with established norms or beliefs. It is driven by a positive feedback in the opposite direction. This theoretical argument was transformed into a System Dynamics simulation model and tested with time series data on unmarried cohabitation (U.S.A., 1960-1994), reform marriages in Israel (1990-1996), and homosexuality (U.S.A., 1973-1996). The plotted output matched the data trends in each set, and the model reproduced over 85% of the variance of the data.

Pages 269-288

Minds and Cultures: Particle Swarm Implications For Beings in Sociocognitive Space

By James Kennedy

Abstract

Particle swarm theory suggests that both minds and cultures are effects of local social interaction. This paper proposes a social-psychological view of intelligence as immerging from culture, which emerges from social interaction. A framework for the depiction of mental states is presented, and the optimizing effect of social interaction is demonstrated. Simulated beings called eleMentals are shown to be able to find optimal regions in their NK-landscape minds through an algorithm comprising two terms: a ``Law of Effect'' term that represents reinforcement learning, or learning from experience, and a social influence term. The model is consistent with social-psychological data and theory, and the results support a hypothesis that human sociality may be in part responsible for human intelligence.

Key Words

social psychology; social intelligence; optimization

Pages 289-306

How Culture Can Guide Evolution: An Inquiry into Gene/Meme Enhancement and Opposition

By Michael L. Best

Abstract

We study the relationship between genetic evolution, learning, and culture. We start with the simulation environment of Hinton and Nowlan in which individual learning was shown to guide genetic evolution towards a difficult adaptive goal. We then consider, in lieu of individual learning, culture in the form of social learning by imitation. Our results demonstrate that when genes and culture cooperate, or enhance one another, culture too is able to guide genetic evolution towards an adaptive goal. Further, we show that social learning is superior to individual learning insofar as it with genetic evolution converges more quickly to the goal. However, the social learning algorithm results in slower genetic assimilation of adaptive alleles than with individual learning. It is as if, we argue, the adaptive values are stored in the culture rather than in the genes. Finally, we consider what happens when culture and genes pursue diametrically opposed goals. Here we show that culture, in the form of social learning, is no real match when opposed to genetic evolution with individual learning. In fact, only the most herculean of social learning algorithms is able to keep a neutralizing toe-hold against the slow plodding force of genetic evolution. Finally, our results suggest that in both cases, opposition and enhancement, transmission forces such as the ratio of teacher to learner are central to the success of social learning.

Key Words

cultural evolution; gene/culture co-evolution; social learning; genetic algorithm

Pages 307-322

Reproduction of Normative Agents: A Simulation Study

By Mario Paolucci and Rosaria Conte

Abstract

The paper presents experiments of a simulation model of aggression control in populations of agents endowed with knowledge about others' reputation (compliant vs. cheaters). The reproductive advantages of different strategies of aggression control, i.e., normative (e.g., "do not attack a food-owner") vs. non-normative ("do not attack a stronger agent"), are compared under given cirmumstances (e.g., exchange of information about others' reputation).

The objectives of the model are: (a) investigate the role of social agents' internal variables (representations, rules, etc.) in the interaction between social processes and agents; while the direction from agents to social processes is usually a focus of attention in simulation models, the reverse one (from social processes to agents) is not; our findings seem to show that some normative disposition of agents is an `"emergent" effect of social processes; (b) question the widely shared assumption that agents are egoist and all which is good in society is either an emergent effect of social life or an institutional task. Our findings seem to show that there may be an adaptive advantage in acting pro-socially.

Pages 323-348

Capturing Social Embeddedness: A Constructivist Approach

By Bruce Edmonds

Abstract

A constructivist approach is applied to characterizing social embeddedness. Social embeddedness is intended as a strong type of social situatedness. It is defined as the extent to which modeling the behavior of an agent requires the inclusion of other agents as individuals rather than as an undifferentiated whole. Possible consequences of the presence of social embedding and ways to check for it are discussed. A model of co-developing agents is exhibited which demonstrates the possibility of social embedding. This is an extension of Brian Arthur's "El Farol Bar" model, with added learning and communication. Some indicators of social embedding are analyzed and some possible causes of social embedding are discussed. It is suggested that social embeddedness may be an explanation of the causal link between the social situatedness of the agent and it employing a constructivist strategy in its modeling.

Pages 349-370

Cooperation, Conflict and the Evolution of Communication

By Jason Noble

Abstract

This paper presents a general model that covers signaling with and without conflicts of interest between signalers and receivers. Krebs and Dawkins (1984) argued that a conflict of interests will lead to an evolutionary arms race between manipulative signalers and sceptical receivers, resulting in ever more costly signals; whereas common interests will lead to cheap signals or "conspiratioral whispers." Previous simulation models of the evolution of communication have usually assumed either cooperative or competitive contexts. Simple game-theoretic and evolutionary simulation models are presented; they suggest that signialing will evolve only if it is in the interests of both parties. In a model where signalers may inform receivers as to the value of a binary random variable, if signaling is favored at all, then signalers will always use the cheapest and the second cheapest signal available. Costly signaling arms races do not get started. A more complex evolutionary simulation is described, featuring continuously variable signal strengths and reception thresholds. As the congruence of interests between the parties become more clear-cut, successively cheaper signals are observed. The findings support a modified version of Krebs and Dawkins's argument. Several variations on the continuous-signaling model are explored.

Key Words

behavioral ecology; communication; competition; coevolution; cooperation; signaling

Pages 371-384

The Learning Barrier: Moving from Innate to Learned Systems of Communication

By Michael Oliphant

Abstract

Human language is a unique ability. It sits apart from other systems of communication in two striking ways: it is syntactic, and it is learned. While most approaches to the evolution of language have focused on the evolution of syntax, this paper explores the computational issues that arise in shifting from a simple innate communication system to an equally simple one that is learned. Associative network learning within an observational learning paradigm is used to explore the computational difficulties involved in establishing and maintaining a simple learned communication system. Because Hebbian learning is found to be sufficient for this task, it is proposed that the basic computational demands of learning are unlikely to account for the rarity of even simple learned communication systems. Instead, it is the problem of "observing" that is likely to be central-in particular the problem of determining what meaning a signal is intended to convey.

Pages 385-414

Animat Market - Trading Interactions as Collective Social Adaptive Behavior

By Dave Cliff and Janet Bruten

Abstract

We argue that human economic interactions, particularly bargaining and trading in market environments, can be considered as collective social adaptive behaviors. Such interactions are social in the sense that they depend on socially-agreed marked regulations and communication protocols, and are collective in the sense that global market dynamics depend on the interations of groups of traders. Moreover, the tools and techniques of adaptive behavior research could be profitably employed to build predictive models of existing or planned marked systems. Despite these potential applications, we note that there is a near-total absence of papers in the adaptive behavior literature that deal with autonomous agents capable of exhibiting trading behaviors. We summarize work in experimental economics where human trading behavior is studied under laboratory conditions. We propose that such experiments could and should be used as `benchmarks' for evaluating and comparing different architectures and strategies for trading animats. We present results from experiments where an elementary machine learning technique endows simple autonomous-software agents with the capability to adapt while interacting via price-bargaining in market environments. The environments are based on artificial retail markets used in experimental economics research. We demonstrate that groups of simple agents can exhibit human-like collective market behaviors. These results invite a Braitenberg-style eliminative materialism perspective on the dynamics of experimental retail markets.

Key Words

microeconomics; markets; auctions; trader-animats; Widrow-Hoff learning

Pages 415-438

Experiments in Learning by Imitation - Grounding and Use of Communication in Robotic Agents

By Aude Billard and Kerstin Dautenhahn

Abstract

Social behaviour and in particular social learning are key mechanisms for the cohesion and evolution of primate societies. Similarly, social skills might be desirable for artificial agents who are expected to interact with other natural or artificial agents. We view learning, communication and imitation as important capabilities to possess by social artificial agents and study how these skills can be designed and used by physically embodied autonomous robots. We study grounding and use of communication among heterogeneous agents. In particular, we investigate the role of social interactions for sharing of context and building of joint attention among communicative agents. Grounding and use of communication is investigated through simulations within a group of autonomous agents. Results show that social behaviour benefit the agents in two circumstances: (1) agents capable of following one another, and in this way imitating each other's movements, develop faster and better a common understanding of the language; (2) furthermore, the agents' capability of communicating with one another via a common vocabulary benefits to the group and to each agent individually as it speeds up the transmission of information. We use a connectionist model, based on Hebbian associative learning, for the learning of the word-signal pairs. This work follows robotic experiments [6, 5, 7] in which a physical autonomous robot was taught a vocabulary to describe its perceptions of objects, movement, inclination and orientation. The robot was taught either by a human instructor or by another robot. The teacher-learner robot experiments were based on an imitative strategy whereby the learner robot followed the teacher robot. The work of this paper demonstrates scaling up of this movement imitative strategy for transmitting a vocabulary across a group of robotic agents, i.e. from a teacher agent to several learner agents. In particular, it shows that imitative behaviour is necessary for the grounding of the agents' proprioceptions and speeds up the grounding of exteroceptions. These studies stress the importance of behavioural social mechanisms in addition to general cognitive abilities of associativity for grounding communication in embodied agents. In particular, it shows that a simple movement imitation strategy is an interesting scenario for the transmission of a language, as it is an easy means of getting the agents to share a common context of perceptions, which is a prerequisite for a common understanding of the language to develop. It is thus suggested that a behaviour-oriented approach might be more appropriate than a pure cognitivist one which is dominating in related studies of the mechanisms involved in grounding communication.

Key Words

social learning and behavior; grounding communication; embodied and situated agents

back to TOC, back to top

Adaptive Behavior, 7 (3/4)

Adaptive Behavior

Volume 7, Number 3-4

Winter 1999

Table of Contents

Kerstin Dautenhahn

Introduction to the Special Issue: Simulation Models of Social Agents

P. M. Hui, Y. R. Kwong, Ping Cheung and N. F. Johnson