Artificial Life VII:Proceedings of the Seventh International Conference on Artificial Life

This chapter contains sections titled: Half Title, Title, Copyright, Contents, Preface View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

Borrowing inspiration from von Neumann's dream and the idea of a true cellular automaton, we describe a multicellular universal Turing machine implementation with self-replication and self-repair capabilities. This implerrientation was made possible thanks to a new “multicellular” automaton developed as part of the Embryonics (embryonic electronics) project. This new automaton, in which every artificial cell contains a complete copy of the geriome, is endowed with self-replication and selfrepair capabilities. With these properties arid by using a modified version of the W-machine, it was possible to realize the mapping of the universal Turing machine onto our multicellular array View full abstract»

A physically-based system of Interacting polyhedral objects is used to model physical and chemical processes characteristic of living organisms. These processes include auto-catalysis, cross-catalysis and the self-assembly and spontaneous organization of complex, dynamic structures constituting virtual organisms. The polyhedra in the simulation are surfaced with bonding sites in states akin to those of cellular automata. These bonding sites interact with sites on neighbouring polyhedra to apply forces of attraction and repulsion between bodies and to trigger transitions in their states. Locally controlled assembly of this kind acts without the guidance of an external agent or central control. Such mechanisms, perhaps a defining property of biological construction, are seldom employed to create complex artificial structures. This paper therefore presents a novel model for the construction of complex virtual structures using multiple reactive, virtual elements, acting independently under virtual physical and chemical laws View full abstract»

A new self-replicating cellular automata (CA) model is proposed as a latest effort toward the realization of an artificial evolutionary system on CA where structural complexity of self-replicators can increase in some cases. I utilize the idea of ‘shape encoding’ proposed by Morita and Imai (Morita & Imai 1996b) and make the state-transition rules of the model allow organisms to transmit genetic information to others when colliding against each other. Simulations with random initial configuration demonstrate that it is possible that the average length of organisms and the average frequency of brancing per organism both increase, with decreasing self-replication fidelity, and saturate at some constant level. The saturation is caused in part by the fixation of place and shape of organisms onto particular sites. This implies the necessity of introducing some fluidity of site arrangements into the model for further development of evolutionary models using CA-like artificial media View full abstract»

This paper reports on a model situated between the biological and the artificial scenario using a second generation, massively parallel reconfigurable computer — POLYP — based on the Xilinx micro-reconfigurable field programmable gate arrays (FPGAs, 6200 series) with additional distributed SRAM memory circuits under local control and broad-hand dynamically reroutable optical interconnect technology. It is shown that self-organisation of clocked Boolean networks indeed is possible where logical description and execution evolve concomittently. SeIf-replicators emerge, evolution over long timescales show the rise and fall of certain species and an inherent mutation control is established. View full abstract»

The work presented in this paper continues the work presented in the previous Alife (Bersini, 1999) conference by improving the Connections between the computerized object-oriented chemistry and the real natural chemistry. This improvement is the result of a clearer definition of the type of computational structure which codes for a molecule, of the different reaction mechanisms between two molecules to obtain one or several new ones, and of the way the dynamics (i.e. the molecular concentration change in time) and the metadynamics (i.e. the appearance of new molecules) simultaneously make the whole chemical system to evolve in time. What makes a molecule computationally unique. in terms of a strictly ordered tree, will be deeply described. Simulation results are presented for a simple chemical reactor composed of four atoms with different valence, and allowing molecules to deterministically or randomly interact according to two mechanisms: the chemical single-link-crossover and the open-bond reaction. Flow the chemical kinetics is called into play will be shown for very elementary reactions View full abstract»

We start with AlChemy system (Fontana, 19921, and with four further modification we move into an Artificial Chemistry system where for each element is present an atomic structure, elements follow the conservation laws on the number of atoms, and the operation permits to two elements to generate more than one element. The resulting system still generates Organisations (self-sustaining, closed sets), but a wider vanety of them are easily reachable without imposing any filter rules on the results of the operation [Fontana, and Buss, 19931. The resulting Organisations are nearly equally divided into two classes (A and B), one (A) which contains Organisations that metabolise external elements to keep themselves ‘alive’, while the other (B) tends to expand until no place is left in reactor, without any need on external elements View full abstract»

We develop an Artificial Cell System (ACS) based on an abstract chemical system. ACS consists of a multiset of symbols, a set of rewriting rules (reaction rules) and membranes. Throughout simulations, we find that chemical evolution like behavior emerges and cells evolve to a structure consisting of several cell-like membranes. We investigate the correlation among the type of reaction rules (rewriting rules), characteristic of a membrane and the evolution of cells, and then find the characteristics of a membrane effects on its evolution and obtains a parameter to describe the correlation. Furthermore, we introduce a genetic method to the system, and we attempt to apply it to Genetic Programming View full abstract»

Populations of strings which interact in ways defined by an artificial chemistry can self-organise spontaneously into an autocatalytic set. This paper considers populations of binary strings with fixed length and a reaction scheme that uses strings as both data (or tape) and machine. Here the machine is a boolean network where some parameters are determined by a string from the population. The input to the machine is given by a second string drawn from the population. In the artificial chemistry based on boolean networks, simulations have revealed a high sensitivity on a probabilistic rate that filters out trivial patterns. By variation of the rate parameter, multiple stable sets have been found. Short string lengths are used here, in order not to rely only on simulations, but also to keep the reaction graph small enough to be able to search for possible autocatalytic sets. A search method has been developed that finds all closed subgraphs of the reaction network, which indicate to a high degree what autocatalytic sets are possible. While simulations most often give only one as a result, the search saves many simulation runs, because it is independent of the initial populations. The resulting number and size of autocatalytic Sets gives information about any freedom of the system to adapt, e.g. when coupling such a system to an environment that can impose constraints. So this description of the behaviour of artificial chemistries appears useful for further artificial studies of molecular evolution and the origin of life View full abstract»

Many interesting properties of Boolean networks, cellular automata, and other models of complex systems rely heavily on the use of synchronous updating of the individual elements. This has motivated some researchers to claim that, if the natural systems being modelled lack any clear evidence of synchronously driven elements, then asynchronous rules should be used by default. Given that standard asynchronous updating precludes the possibility of strictly cyclic attractors, does this mean that asynchronous Boolean networks, cellular automata, etc., are inherently bad choices at the time of modelling rhythmic phenomena? In this paper we focus on this subsidiary issue for the case of Asynchronous Random Boolean Networks (ARBNs). We find that it is rather simple to define measures of pseudo-periodicity by using correlations between states and sufficiently relaxed statistical constraints. These measures can be used to guide an evolutionary search process to find appropriate examples. Success in this search for a number of cases, and subsequent statistical studies lead to the conclusion that ARBNs can indeed be used as models of coordinated rhythmic phenomena, which may be stronger precisely because of their built-in asynchrony. The methodology is flexible, and allows for more demanding statistical conditions for defining pseudo-periodicity, and constraining the evolutionary search View full abstract»

This paper addresses the use of particular encoding schemes in evolutionary systems. We define three paradigms of DNA encodings: non-compartmentalized, partially compartmentalized, and fully compartmentalized. We demonstrate that there is a significant and increasing advantage to the use of partially and fully compartmentalized models as the complexity of a structure increases. Implications for the design of evolutionary systems including biological systems are discussed View full abstract»

Gaia theory, the view that the biota can both affect their environment and do so in a manner that benefits life in general, is an extremely controversial interpretation of the complex relationships between the biota and biosphere. Since individual Gaian phenomena can span spatial scales from cellular to planetary, they evade thorough analysis and empirical validation. Consequently, a good deal of Gaian thinking revolves around an abstract computer model, Daisyworld [24]. However, this model fails to properly account for natural selection's role in Gaian emergence. Although we propose an alternate scheme that offers some improvement - one based on evolutionary computation and individual-based simulation - the field remains wide open for investigations from the alife perspective. This paper reviews both models along with a few natural Gaian phenomena before generalizing a set of common primitive features and emergent properties from the real and artificial examples. These shared characteristics will hopefully provide a backbone for a muchdesired “Gaia-logic” and assist other alife researchers in the search for additional Gaian models View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

By introducing a dynamical system model of a multicellular system, it is shown that an organism with a variety of differentiated cell types and a complex pattern emerges through cell-cell interactions even without postulating any elaborate control mechanism. Such organism is found to maintain a larger growth speed as an ensemble by achieving a cooperative use of resources, than simple homogeneous cells which behave ‘selfishly’. This suggests that the emergence of multicellular organisms with complex organization is a necessity in evolution. According to our theoretical model, there appear multipotent stem cells initially, which exhibit stochastic differentiation to other cell types. With the development and differentiation, both the chemical diversity and complexity of intra-cellular dynamics are decreased, as a general consequence of our system. Robustness of the developmental process is also confirmed View full abstract»

A novel viewpoint for evolution is presented, by taking seriously into account the relationship between genotype and phenotype. First, as a consequence of dynamical systems theory, phenotypes of organisms can be differentiated into distinct types through the interaction, even though they have identical genotypes. Then, with the mutation in genotype, it is shown that the genotype also differentiates into discrete types, while maintaining the ‘symbiotic’ relationship between the types. This process is robust against, sexual recombination, because offspring with intermediate genotypes are less fit than their parents. Accordingly, a plausible scenario for sympatric speciation is presented. Relevance of our scenario to the historical evolution as well as to artificial evolution is discussed View full abstract»

Evolution of genetic code is studied as the change in the choice of enzymes that are used to synthesize amino acids from the genetic information of nucleic acids. We propose the following scenario: the differentiation of physiological states of a cell allows for the different choice of enzymes, and this choice is later fixed genetically through evolution. To demonstrate this scenario, a dynamical systems model consisting of the concentrations of metabolites, enzymes, amino acvl tRNA synthetase, and tRNA-amino acid complex in a cell is introduced and numerically studied. It is shown that the biochemical states of cells are differentiated by cell-cell interaction, and each differentiated type takes to use different synthetase. Through the mutation of genes, this difference in the genetic code is amplified and stabilized. Relevance of this scenario to the evolution of non-universal genetic code in mitochondria is suggested View full abstract»

Examination of code execution patterns in different threads of multithreaded digital organisms reveals the clustering of threads into tissues which differ in patterns of code execution. Evolution in a network environinent has led to the differentiation of new tissues, which through a division of labor, work together to accomplish the sensory function of a single tissue type of the ancestral organism View full abstract»

We present a simu!ation of morphogenesis and cell interaction which allows us to address embryogenesis as an engineering problem. A space of cell control functions is defined using a cybernetics approach by identifying the controlling variables and the degrees of freedom of each individual cell. These functions receive inputs which are dependent on a cell's context, and determine its responses to that context. Multiple cells are coupled through a simulation of a 3d reaction-diffusion fluid matrix in order io generate interactive behaviour. A fitness function is designed which characterises the desired behaviour of the cellular collective. We tune ihe parameterised control funciion for the individual cells using a genetic algorithm to maximise the fitness of their collective behaviour View full abstract»

Designing artificial systems with ever more biologically-plausible ‘brains’ continues apace and permits investigations into the computational capabilities of engineered systems. Creating artificial neurons with biologically-realistic morphologies is however a non-trivial problem. This paper addresses growing neurons to order, neurons with morphologies exhibiting strong biological traits. A biologically-inspired simulator of neural development is coupled with a genetic algorithm to evolve 3-dimensional neuron morphologies. The morphology of a biological neuron provides the exemplar target against which the developmental evolution process is gauged View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

Here we show that connectivity and catastrophe play a key role in driving species evolution within a landscape. They also form a special case of a more general process, which occurs widely in natural and artificial systems. In this process, catastrophes cause a temporary phase change in the connectivity of a system. Different mechanisms (selection and variation) predominate in each phase. The system passes through the critical point without being poised there. The “chaotic edge” associated with the phase change may be an important source of variety in biological and other systems. In species evolution (and landscape ecology) the process is mediated by cataclysmic events, which fragment widely distributed species and trigger population explosions of new species. The proposed mechanism explains a number of biological and physical phenomena, as well as certain algorithms used for optimisation and evolutionary programming View full abstract»

Natural evolutionary systems exhibit a complex mapping from genotype to phenotype. One property of these mappings is neutrality, where many mutations do not have an appreciable effect on the phenotype. In this case the mapping from genotype to phenotype contains redundancy such that a phenotype is represented by many genotypes. Studies of RNA and protein molecules, the fundamental building blocks of life, reveal that this can result in neutral networks - sets of genotypes connected by single point mutations that map into the same phenotype. This allows genetic changes to be made while maintaining the current phenotype and thus may reduce the chance of becoming trapped in sub-optimal regions of genotype space. In this paper we present several redundant mappings and explore their properties by performing random walks on the neutral networks in their genotype spaces. We investigate whether the properties found in nature's search space can he engineered into our artificial evolutionary systems. A mapping based on a random hoolean network was found to give particularly promising results View full abstract»

Our investigations concern the role of symbiosis as an enabling mechanism in evolutionary adaptation. Previous work has illustrated how the formation of mutualist groups can guide genetic variation so as to enable the evolution of ultimately independent organisms that would otherwise be unobtainable. The new experiments reported here show that this effect applies not just in genetically related organisms but may also occur from symbiosis between distinct species. In addition, a new detail is revealed: when the symbiotic group members are drawn from two separate species only one of these species achieves eventual independence and the other remains parasitic. lt is nonetheless the case that this second species, formerly mutualistic, was critical in enabling the independence of the first. We offer a biological example that is suggestive of the effect and discuss the implications for evolving complex organisms, natural and artificial View full abstract»

We show that the mutual coevolution of cooperating traits amongst interacting populations permit the solution of a matching problem. This solution, within a highly uncorrelated fitness landscape, is difficult in the absence of coevolution or other powerful agencies. We start with the GA environment of Hinton and Nowlan (1987) who originally showed that in the absence of individual learning evolving agents are not able to solve a related problem. While a number of researchers have demonstrated that the coevolution of cooperating and (in particular) competing populations of agents can improve simulated evolution, we argue that coevol ved mutualists can help evolution find a solution it otherwise could not solve (namely, selection of some particular single bit string in an uncorrelated landscape). We posit that coevolved mutualists succeed at this problem because they are able to benefit from genetically stored solutions to sub-problems. This result suggests that perhaps natural problems such as wasp/fig tree signaling, or gene-culture coevoludon of vocal learning in songbirds or human natural language may be guided by coevolutionary mutualism View full abstract»

Social dilemma, problems in the formation and maintenance of cooperation among selfish individuals, are of fundamental importance for the biological and social sciences. To consider the spontaneous formation of cooperation in society under social dilemma, dynamical systems game is adopted. To be specific, the ‘Lumberjacks’ Dilemma (LD) game' is studied, which includes the concrete description of the dynamics of resources depending on players' actions. With the numerical study of the evolution of the strategies in the LD game, it is shown that the cooperation is formed and maintained, through the formation of articulation of strategies based on bifurcation in dynamical systems. Evolution of the cooperative society is found to occur successively with the dynamical change of game structure and of norms adopted in the society. In contrast with the models of the traditional game theory, the cooperation is shown to be sustained even if the number of the members sharing the social dilemma is increased View full abstract»

The Baldwin effect is known as interactions between learning and evolution, which suggests that individual lifetime learning can influence the course of evolution without the Lamarckian mechanism. Our concern is to consider the Baldwin effect in dynamic environments, especially when there is no explicit optimal solution through generations and it depends only on interactions among agents. We adopted the iterated Prisoner's Dilemma as a dynamic environment, introduced phenotypic plasticity into strategies, and conducted the computational experiments, in which phenotypic plasticity is allowed to evolve. The Baldwin effect was observed in the experiments as follows: First, strategies with enough plasticity spread, which caused a shift from defect-oriented population to cooperative population. Second, these strategies were replaced by a strategy with a modest amount of plasticity generated by interactions between learning and evolution. By making three kinds of analysis, we have shown that this strategy provides the outstanding performance. Further experiments towards open-ended evolution have also been conducted so as to generalize our results View full abstract»

A variable-length protein genotype space (the whole amino-acid sequence space) is mathematically analyzed and a lower threshold density for adequate connectivity of functional (viable) genotypes is estimated. Functional genotypes are assumed to distribute as a ‘hyperblob’ which means a cluster or an island, and connectivity between hyperblobs is estimated using the theory of regular languages and the random graph theory. It is shown that the logarithmic value of the threshold density approximately decreases with an increase in the genotype length View full abstract»

We evolved multiple clones of populations of Digitaiza, a type of digital organism, to study the effects of chance, history, and adaptation in evolution. We show that clones adapted to a specific environment can adapt to new environments quickly and efficiently, although their history remains a significant factor in their fitness. Adaptation is most significant (and the effects of history less so) if the old and new environments are dissimilar. For more similar environments, adaptation is slower while history is more prominent. For both similar and dissimilar transfer environments, populations quickly lose the capability to perform computations (the analog of beneficial chemical reactions) that are no longer rewarded in the new environment. Populations that developed few computational “genes” in their original environment were unable to acquire them in the new environment View full abstract»

A number of models of macroevolutionary extinction dynamics have been proposed during the past decade or so. Many of these models produce results that are in good accord with empirical data, as drawn from the fossil record. However, they all still suffer from significant shortcomings. A new simulation based model is presented here that attempts to address some of the weaknesses and limitations of these existing models. The simulation is driven by ecology level interactions amongst dynamically changing populations of species in a theoretical food web, coupled with the effects of external environmental stresses. The results of a number of extended runs of the simulation are presented and discussed. It is observed that it is the interactions between intrinsic ecological factors, and external environmental factors, that determine the specific extinction dynamics generated. Ecological factors appear key in defining the large-scale statistical trends of the system. Environniental factors appear to act as a sort of ‘tempo keeper’, determining the precise timing of extinction events within the large-scale framework. Overall, the results from the simulation suggest that macroevolutionary patterns of extinctions are primarily generated intrinsically by an ecosystem. Environmental factors are not so much a direct cause of extinction events, as a determinant of the precise timing of events that will, in any case, inevitably ensue View full abstract»

This paper describes a new standpoint to further understanding the evolvability of A-life systems, that is, quantification of the dependency of system constituents. If a system exhibits evolutionary activity, new events, including the development or extinction of systems constituents, may emerge one after the other. The occurrence of new events would involve a changing complexion in the course of the future direction of system behaviors. In this respect, there have been a considerable number of studies on captunng longterm dynamics in A-life systems in terms of, for example, population dynamics. On the other hand, we focus on the quantification of relational concepts between constituents to clarify the occurrence of fundamental events affecting longterm course of evolution. We apply a quantification method to our A-life simulation model. Sequential analysis of the obtained resulLs enables us to visualize its long-term evolutionary dynamics. We discuss the effectiveness of our approach and the quantified evolutionary dynamics View full abstract»

At Alife VI, Mark Bedau proposed some evolutionary statistics as a means of classifying different evolutionary systems. Ecolab, whilst not an artificial life system, is a model of an evolving ecology that has advantages of mathematical tractability and computational simplicity. The Bedau statistics are well defined for Ecolab, and this paper reports statistics measured for typical Ecolab runs, as a function of mutation rate. The behaviour ranges from class 1 (when mutation is switched off), through class 3 at intermediate mutation rates (corresponding to scale free dynamics) to class 2 at high mutation rates. The class 3/class 2 transition corresponds to an error threshold. Class 4 behaviour, which is typified by the Biosphere, is characterised by unbounded growth in diversity. It turns out that Ecolab is governed by an inverse relationship between diversity and connectivity, which also seems likely of the Biosphere. In Ecolab, the mutation operator is conservative with respect to connectivity, which explains the boundedness of diversity. The only way to get class 4 behaviour in Ecolab is to develop an evolutionary dynamics that reduces connectivity of time View full abstract»

Focusing on situations in which sexual preferences inspect several traits, this paper explores the question of how complex the process of assessing mate attractiveness might get, through self-organization, over the course of evolution: do sexual preferences evaluate traits in a simple, linear way or in a complex, nonlinear fashion? Arguments and simulation results (individual-based model in which preferences with different degrees of evaluation-complexity “compete” in a population) are presented, suggesting that a bias should exist, and that its direction should depend on levels of different types of noise (in perception and in the mate-evaluation process) and on mutation rates: while noise in perception and mutation favor simple preferences, noise in evaluation favors complex ones; for possibly the most plausible parameters, complex preferences are favored. Possible implications on the speciation process are mentioned. This novel way of looking at mate choice could be a rich source of new insights in this intrigu]ng process View full abstract»

Stephen Gould has introduced the hypothesis that progress in evolutionary history is due to contingency. Daniel Dennett has further suggested that this hypothesis could be confirmed or denied by artificial life models, in which multiple evolutionary histories can be produced and the role of contingency thus evaluated. While existing models do not allow for this evaluation, Calabretta et al. (1998) suggest a general form to be used in developing artificial life models to answer questions about biology. Using this form, a model is outlined which may be able to test this hypothesis View full abstract»

This paper describes some results of our computer simulation concerning ecological competition on the target area of seed dispersal. It is better for any kinds of plants to disperse their seeds as far as possible because it might spread in new frontier earlier than the others. But, it would be better to put the seeds down just at the neighbor position when the environmental condition is stable. From drawing a fitness landscape for distance and area of dispersal through a computer simulation, it was revealed that both of these strategies are locally optimal to gain more reproductive success, and neighboring strategy is the best when the environment is unchanged and uniform. We examined a type of evolutionary process to investigate the effects of three kinds of environmental parameters, scale of disturbance, death-sprout ratio, and geographical granularity of fertility. The population initialized by random parameters converged into either or both of two types of species, far and broad dispersal and neighboring reproduction, For all of three parameters, the expenmental results showed that the probability to converge into dispersal of longer distance becomes greater corresponding to the degree of environmental change in time and space View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

We have developed a hybrid neuro-robotic system based on a two-way communication between the brain of a lamprey and a small mobile robot. Time purpose of this system is to offer a new paradigm for investigating the behavioral, computational and neurobiological mechanisms of sensory motor learning in a unified context. The mobile robot acts as an artificial body that delivers sensory information to the neural tissue and receives command signals from it. The sensory information encodes the intensity of light generated by a fixed source. The closed-loop interaction between brain and robot generates autonomous behaviors whose features are strictly related to the structure and operation of the neural preparation. In this paper we provide a detailed description of the hybrid system and we present experimental findings on its performance. In particular, we found (a) that the hybrid system generates stable behaviors; (b) that different preparation display different but systematic responses to the presentation of an optical stimulus and (c) that alteration of the sensory input lead to short and long term adaptive changes in the robot responses. The comparison of the behaviors generated by the lamprey's brainstem with the behaviors generated by network models of the same neural system provides us with a new tool for investigating the computational properties of synaptic plasticity View full abstract»

Most work in evolutionary robotics has focussed on evolving neural network based control systems, rather than high level control programs using genetic programming. One reason for this is that the fitness landscapes generated by a genetic programming framework may be poorly suited to the evolutionary techniques employed. The aim of this paper is to investigate this claim. The first part of the paper demonstrates two simulations of evolving populations of simple robots, one based on genetic programming, the other based on evolved neural nets. The latter part of the paper then introduces a technique for ‘fitness landscaping’, that is for constructing 3-D visualisations of the sorts of complex, high dimensionality fitness landscapes involved. This technique is applied to the results from the two siniulations, and the resulting representations are discussed View full abstract»

One of the prevailing characteristics of natural life is autonomy. The field of Artificial Life has so far addressed the notion of autonomy mostly in terms of power and behavior. In this paper we attempt to extend the notion of autonomy to include also design and fabrication: We claim that not only should artificial creatures be able to operate untethered and without external guidance, but they should also he, like living systems, autonomously designed and fabricated without external intervention. Only then can we expect synthetic creatures to bootstrap and sustain their own evolution. In this work we demonstrate for the first time a path that allows transfer of virtual diversity into reality, and so reduce this key principle of Artificial Life into practice. Our approach is the use of only elementary building blocks in both the design and embodiment. We describe a set of preliminary experiments evolving electromechanical systems composed of thermoplastic, linear actuators and neurons for the task of locomotion, first in simulation then in reality. Using 3D solid printing, these creatures then replicate automatically into reality where they faithfully reproduce the performance of their virtual ancestors View full abstract»

This article presents a model for the evolution of locomotion behavior in a simulated insect. In our model, locomotion is defined over a discrete state space using non-uniform cellular automata. The architecture of the model is inspired from the distributed model for leg coordination proposed by Cruse. We apply a genetic algorithm to a population of non-uniform cellular automata to evolve locomotion behaviors. We demostrate that this model can be used to evolve several commonly observed gaits of insects. Additionally, we show that the evolutionary process yielded periodic attractors which are invariant from the initial conditions View full abstract»

The problem of achieving open-ended evolution in complex systems is studied in this paper. We propose various techniques that support it. and present a gradual approach. These techniques are used and tested in the Framsticks system. which is a realistic, three-dimensional artificial life simulator ith rich capabilities. Specifically, as there are no strong constraints imposed on the structure (body) and control (neural network brain), this system is suitable for testing open-endedness. In Framsticks. energetic requirements form the basis for competition. while interactions occurring on various levels act as a source of complexity and variation. The way towards open-ended evolution is discussed, developmental genotype encodings are proposed, arid the results of so-far experiments are presented View full abstract»

We propose the artificial neural development method that generates the three-dimensional multi-regional pulsed neural network arranged in three layers of the nerve area layer, the nerve sub-area layer, and the cell layer. In this method, the neural development process consists of the lirst genomecontrolled spatiotemporal generation of a neural network structure and the latter activity-dependent regulation of it. In the first process, by decoding a genome, 1)a nerve sub-arca is generated in each nerve area and neurons are produced in it, 2)axonal outgrowth target sub-areas are recognized according to the attraction and repulsion rule, and 3)synapse formation is controlled under the topology preservation projection rule between origin cells and target cells. In the latter process, 4)programmed cell death occurs under control of spiking activity and a neurotrophic factor, then 5)synaptic efficacy is regulated according to the spike-based hebbian rule and weakened synapses are eliminated as a result of competition of spiking activity. For design of genomes, the steady state genetic algorithm is introduced and it is applied to initial genomes partially designed manually. To evaluate our artificial neural development method, simulation experiments are conducted to generate a pulsed neural network of an animal-like robot (animat) which moves in an environment. We evolve and develop an animat's place recognition circuit that contains the place cell area The place recognition performance is evaluated in an environment where an animat comes into existence and in another environment where the animat enters after development. Through these experiments, we show our artificial neural development method is useful for generating a biologically realistic pulsed neural network of the animat View full abstract»

This paper proposes a method for controlling a flexible SMA (Shape Memory Alloy) - Net robot by using a coupled oscillator system. The control method is inspired by the emergent behavior control mechanism of slime mold amoebae and using highly distributed information processing system. The control method enables basic behaviors of movement to attractive stimulus and escape from unattractive stimulus View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

A developmental view of evolving systems (ecological, social, economical, organizational) is examined to clarify 1) the role of selection processes versus collective, non-selective processes, 2) the origins of diversity and its role in system performance and robustness 3) the origin of explicit subsystem interactions (cooperation/symbiosis that enhance individual and system performance, 4) the preconditions necessary for further evolutionary development, and 5) the effect of environmental timescales with adaptation timescales. Three sequential stages of evolving systems (based on the work of Salthe) are proposed: a Immature stage dominated by highly decentralized, selective processes with chaotic local and global dynamics, a Mature stage dominated by non-selective, self-organizing processes with global robustness but locally chaotic dynamics, and a Senescent stage dominated by rigid interactions with global fragility. A simple model problem with many optimal and non-optimal solutions - an agent solution to a maze - illustrates the entire developmental history. Within the model, the agents evolve their capability from a random approach to an optimized performance by natural selection. As the agents develop improved capability, natural selection becomes rare, and an emergent collective solution is observed that is better than the performance of an average agent. As the collective, self-organizing structures are incorporated into individual capability within a stable environment, constraints arise in the agent's interactions, and the system loses diversity. The resulting Senescent system exhibits reduced randomness due to the rigid structures and ultimately becomes fragile. Depending on the degree environmental change, the Senescent system will either “die,” or collapse under environmental stress to the Mature or Immature stage, or incorporate the constrain ts system-wide into a new hierarchical system. The current study adds to the literature on developmental systems by finding: Transitions between stages are dependent on the degree of sustained environmental stability and how exclusive cooperation (e.g., symbiosis) in a subsystem can originate, and how it results in a decline in diversity View full abstract»

In this paper, the correlation between behavioural heterogeneity and behavioural complexity within groups of cooperating agents is investigated. This investigation is accomplished using the Legion system, a type of evolutionary algorithm for evolving group behaviours in which behavioural differences among agents in the group is subject to selection pressure. Two collective task domains are studied, and two types of control architecture for the agents are used. From the experiments reported here it is concluded that increased behavioural heterogeneity within a group leads to reduced average control complexity, and also that reducing the maximum size of control architectures results in the evolution of increased behavioural heterogeneity. It is argued that this correlation helps to clarify the relationship between robustness, division of labour and variation within cooperating agent populations, and also that heterogeneity can be a useful tool for robot group design View full abstract»

Mathematical treatment of attention dynamics of a predator chasing prey is studied. In our model, prey and a predator move around a two-dimensional surface. Real numbers that correspond to individuals of prey are used. A predator selects prey of the highest number. If prey's motion is ordered, the predator selects the nearest individual. If prey's motion is disordered, the predator tends to select the individual which has split away from a collective, ie. alone. It is emphasized that partially disordered motion, is rather difficult for the predator to select an individual View full abstract»

Our paper studies the emergence of social norms and their subsequent influence in a simulated society of artificial agents. These norms and their propagation in a society is put in close relation with the meme concept of Dawkins. In particular, the norms studied in this paper are concerned with the possession of goods. Here a global norm regarding possession of goods may result from the collective dynamics of the society and arise from purely local agent interactions. The role of sanctions and costs of enforcing sanctions and their relationship to the establishment of a possession norm are studied View full abstract»

This paper presents a viscous population multi-agent system, which is claimed to provide scope for the emergence of cooperation both through iterated interaction and through kin selection. Theoretical examinations of iterated Interaction and kin selection within the model are conducted and compared with empirical results. It is concluded that the model does allow for the operation both of iterated interaction and kin selection. The methods presented in the paper allow the operation of the two mechanisms to be distinguished in any instance of the model View full abstract»

In this paper I discuss the notion of multiple levels of emergence, and motivate its usefulness as a conceptual tool. I argue that a multi agent system containing emergent properties can be divided into i) Axiomatic behaviour ii) Emergent behaviour and iii) World rules. I show how to use this division as an informal design tool, when designing and analysing emergent properties. Specifically I discuss where and why to place the axiomatic level, given the goal of the model. Further I present an ant model, called Aintz. I show that the very simple basic behavior of the individual agents cause complicated emergent behaviors of multiple levels. I use the Aintz model to exemplify the notion of multiple levels of emergence View full abstract»

It was shown that the evolution of a world consisting of agents that play Iterated Prisoner Dilemma (IPD) games each other is open-ended by Lindgren. The behavior of the world is very sensitive to the values of the payoff matrix used in IPD games, because the values have great influence on the population dynamics of the world. In general, the values are fixed throughout the simulations. In the real world, however, morals and the behaviors of individuals that follow the morals have been evolved influencing mutually. In this paper, we propose a co-evolution model of agents and scores of IPD games toward the understanding of emergence of social morals. The coevolution model consists of two layers. In the first layer, scores for IPD games are evolved using a genetic algorithm. Scores vary within the range of dilemma games, and scores that attract more agents in the second layer will gradually increase. In the second layer, agents play IPD games with all other agents following the scores that they believe and are evolved using Lindgren's model. Simulation results showed that the values of the scores evolve toward the score which gives more payoffs for cooperative strategies and less payoffs for defective strategies as the strategies of the IPD agents are evolved. The results also showed that small colonies of defective strategies repeatedly appear and disappear throughout the simulations. View full abstract»

Nowadays, an increasing number of web sites allow people to directly trade stocks and several other types of investments in financial markets through the Internet. These sites offer many conveniences to the customers using the network infrastructure, sucht as “stock alerts”, winch are mechanisms that. promptly notify the customer when a certain stock reaches a threshold price value, set in advance by himself. The expected action front the stock alert user, upon receiving this notification, is to place an order to buy or sell stock shares, as the threshold price is supposedly associated with the profit the user is aiming for. The question addressed in this paper is how features like “stock alerts” affect tue overall i,eliavior of a financial market, as the number of individual stockholders grows due to the increased access to the market allowed by the Internet. Initial simulation results with an agent-based model of a financial market are presented View full abstract»

Words, like genes, are replicators in competition to colonize our brains. Some, by luck or thanks to their intrinsic qualities, manage to spread in entire populations. In this paper we take the approach of cultural selectionism to study the emergence of communication systems in a population of agents. By studying simple models of word competition in noisy environments, we define the basic dynamics of such systems. We then argue for their generality and introduce the notion of semiotic schemata, generic replicators that account for the different competitions that are going on during lexicon formation. Eventually, we present a synthesis of the dynamics using this new formalism View full abstract»

The paper reports on an experiment in which a group of autonomous agents self-organises through cultural evolution constraints on the combination of the individual sounds (phonemes) in their repertoires. We use a selectionist approach whereby a repertoire evolves by mutations of patterns, constrained by functional pressures from perception and production and the need to conform to the group View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

The transformation of normal cells into cancerous cells is an evolutionary process. Populations of precancerous cells reproduce, mutate, and compete for resources. Some of these mutations eventually lead to cancer. We calculate the probability of developing cancer under a set of simplifying assumptions and then elaborate these calculations, culminating in a simple simulation of the cell dynamics. The agent-based model allows us to examine the interactions of neutral and selective mutations, as well as mutations that raise the mutation rate for the entire cell. The simulations suggest that there must be at least two selectively neutral mutations necessary for the development of cancer and that preventive treatments will be most effective when they increase this number View full abstract»

An assorted range of approaches have contributed to our understanding of the oscillatory behavior of population sizes in predation models. Among these are Mathematical Biology, Statistics and Artificial Life (ALife). In this paper, I will give a review of these different approaches. In addition, another approach, based on EvolutioHary Game Theory, is proposed and discussed. This paper also suggests that a complementary study of both the Mathematical, Artificial Life and Game Theory approach is needed to explain some of the mysticism surrounding the global emergent behavior of local predator-prey relationships. View full abstract»

The potentials and tools that are offered by Alife for biology in modeling the nervous system and animal behavior are mainly unexploited. There is no consistent Alife model of the biological evolution of the nervous system as yet. whereas the modeling tools are at hand and their application for this purpose seems evident. In a biologically grounded model we have to make every possible effort to use principles known from biology, and to minimize the arbitrarily employed organizing rules. The aim of our work is to create a biologically accurate Alife model of the formation and evolution of the nervous system in connection with the adaptive behavior. In this article we concentrate on the structure of the modeled genome, which is the basis of playing a double biological role: to ensure an open-ended evolutionary process, as well as to direct the ontogenesis. The main questions we examined are: what are the basic rules of construction that are sufficient to create a workable nervous system and how can we model them in a biologically realistic way? View full abstract»

We present a market-based approach to call routing in telecommunications networks. A system architecture is described that allows self-interested agents, representing various network resources, potentially owned by different real world enterprises, to co-ordinate their resource allocation decisions without assuming a priori co-operation. lt is argued that such an architecture has the potential to provide a distnbuted, robust and efficient means of traffic management. In particular, our architecture uses an adaptive pncing and inventory setting strategy, based on real bidding, to reduce call blocking in a simulated telecommunications network View full abstract»

Various studies in the Context of one-dimensional cellular automata (CA) have been done on defining parameters directly obtained from their transition rule, which might be able to help forecast their dynamic behaviour. Out of a critique of the most important parameters available for (his end, and out of the definition of a set of guidelines (hat should be followed when defining that kind of parameter, we took two parameters from the literature and defined three new ones, which, jointly provide a good forecasting Set. We then used them to define an evolutionary search heuristic to evolve CA that perform a predefined computational task; here the well-known Density Classification Task is used as reference. The results obtained show that the parameters are effective n helping forecast the dynamic behaviour of onedimensional CA, and can effectively help a genetic algorithm in searching for CA of a predefined kind View full abstract»

This paper describes the application of an Artificial Life cellular automata (CA) microsimulation to model the emergent collective behavior of bi-directional pedestrian flows. Since pedestrian flow is inherently complex, even more so than vehicular flow, previous CA models developed for vehicle flow are not directly applicable. Itis shown that a relatively small rule Set is capable of effectively capturing the collective behaviors of pedestrians who are autonomous at the micro-level. The model provides for simulating three modes of bi-directional pedestrian flow: (a) flows in directionally separated lanes, (b) interspersed flow, and (c) dynamic multi-lane flow. The emergent behavior that arises from the model is consistent with well-established fundamental properties of pedestrian flows View full abstract»

This paper describes a new method for the analysis of spatial data that can be used to solve the NP-hard problem of point pattern analysis in geographic, high-dimensional attribute data. The method builds on an established, highly developed and extensively tested methodology (GAM) and extends and combines it with concepts and methodologies found in the field of Artificial Life (Flocking and Agents). The new methodology is smart in that it is able to adapt to the characteristics of various data sources and makes intelligent use of available computational resources when the problem space becomes so large that brute-force techniques would quickly exhaust all available computer power. The system is also, by nature, comprised of multiple, discrete computational units that lend themselves to easy parallelization, thus facilitating the use of parallel, or even distributed, architectures. The inspiration for many of the intelligent aspects of the methodology came from existing research into Artificial Intelligence, Artificial Life and Multi-Agent technologies. View full abstract»

We report on a software toolbox that is part of an architecture design process we have named “Emergent Design” The toolbox incorporates concepts of artificial life that allow architects to realize conceptual experiments in which the elements of an architectural scenario are endowed with agency and dynamic, spatial interaction. Elements of the scenario combine and interact spatially over time to result in an emergent design. We believe this use of ALife concepts in the design process of architecture to be both novel and powerful View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»

In the late 1940's John von Neumann began to work on what he intended as a comprehensive “theory of [complex] automata”. He started to develop a book length manuscript on the subject in 1952. However, he put this aside in 1953, apparently due to pressure of other work. Due to his tragically early death in 1957, he was never to return to it. The draft manuscript was eventually edited, and combined for publication with some related lecture transcripts, by Burks (1966). It is clear from the time and effort which von Neumann invested in it that he considered this to be a very significant and substantial piece of work, However: subsequent commentators (beginning even with Burks) have found it surprisingly difficult to articulate this substance. Indeed, it has since been suggested that von Neumann's results in this area are either trivial, or, at the very least, could have been achieved by much simpler means. It is an enigma. In this paper I review the history of this debate (briefly) and then present my own attempt at resolving the issue by focusing on an analysis of von Neumann's problem situation (Popper, 1976). I claim that this reveals the true depth of von Neumann's achievement and influence on the subsequent deveopment of this field; and, further, that it generates a whole family of new consequent problems which can still serve to inform—if not actually define—the field of Artificial Life for many years to come View full abstract»

A simple computer program dating from the first half of the nineteenth century is presented as the earliest known example of an evolutionary simulation model. The model is described in detail and its status as an evolutionary simulation model is discussed. Three broad issues raised by the model are presented and their significance for modern evolutionary simulation modelling is explored: first, the utility of attending to the character of a system's entire dynamics rather than focusing on the equilibrium states that it admite of; second, the worth of adopting an evolutionary perspective on adaptive systems beyond those addressed by evolutionary biological research; third, the potential for the non-linear character of complex dynamical systems to be explored through an individual-based simulation modelling approach View full abstract»

Artificial life research typically employs digital computers to implement models of living systems. However, there is now a growing if pro-theoretical feeling that computers, or perhaps the software running on them, are themselves some kind of living systems. Such a possibility can impact artificial life research in at least two ways By highlighting that computers and comnnunications networks can be subjects, as well as tools, for artificial life muodelling, and by highlighting that insights, tools, and models from the life sciences can have explanatory, predictive, and design consequences for the construction of future computation and communications systems. This paper seeks perspective on such ‘real artificial life’, looking backwards and forwards at the rise of living systems in manufactured computer and communications systems View full abstract»

We review and critique a range of perspectives on the scientific role of individual-based evolutionary sinlulation models as they are used within artificial life. We find that such models have the potential to enrich existing modelling enterprises through their strength in modelling systems of interacting entities. Furthermore, simulation techniques promise to provide theoreticians in vanous fields with entirely new conceptual, as well as methodological, approaches. However, the precise manner in which simulations can he used as models is not clear. We present two apparently opposed perspectives on this issue: simulation models as “emergent computational thought experiments” and simulation models as realistic simulacra. Through analysing the role that armchair thought experiments play in science, we develop a role for simulation models as opaque thought experiments, that is, thought experiments in which the consequences follow froni the premises, hut in a non-obvious manner which must be revealed through systematic enquiry. Like their better-known transparent cousins, opaque thought experiments, when understood, result in new insights and conceptual reorganisations. These may stress the current theoretical position of the thought. experimenter and engender empincal predictions which must be tested in reality. As such, simulation models, like all thought experiments, are tools with which to explore the consequences of a theoretical position View full abstract»

Artificial Life is developing into a peculiar type of discipline, claiming the principle of computational construction a ihe main avenue to explore and produce a new science of life “as it could be”. In this research program, the generation of complex virtual systems may become the actual object of the theories, somehow substituting the usual empirical domain. This brings along not only a deep change in the traditional relationship between the ontological, epistemological and methodological levels, but also the appearance of a new relationship between the theoretical and technical realms, in what constitutes a relevant epistemic problem. Such a state of affairs forces us to reconsider the solid differences apparently established between science and philosophy. Even if the frontiers between these two kinds of knowledge do not completely disappear, new, very dynamic and complex, technologically mediated ways of interaction are being developed between them View full abstract»

Recently, there has been an attempt among some philosophers to provide cultural evolutionary grounds for certain norms of distributive justice. The most noteworthy attempt (Skyrms, 1996) uses a simple evolutionary model based upon the replicator dynamics. I argue that the replicator dynamics is not the most appropriate model of interactive human behavior in societies, and present an alternative agent-based model. I demonstrate how the conditions under which norms of distributive justice arise depend on how we construe the underlying evolutionary dynamics View full abstract»

We examine the eventual role of surprise in three domains of human endeavor: classical engineering, what we call “emergent engineering,” and the general unrestricted field of artificial life. Our study takes place within the formal framework of the recently proposed “emergence test.” We argue that the element of surprise, central in the test, serves to illuminate the fundamental differences between these three fields. This we achieve by distinguishing between three different forms of surprise: unsurprise, unsurprising surprise, and surprising surprise View full abstract»

Looking backwards, we recall art of Sommerer and Mignonneau, Sims, and Latham that was inspired by artificial life principles. Assessing current artificial life inspired art, we examine the methods of fitness by aesthetics and user-guided evolution of evolving expressions as practiced by Rooke, Ibrahim, Musgrave, Unemi and the author. Looking forwards, we consider autonomously evolved artistic works using algorithmic aesthetics. We survey what little is known about this topic and proceed to describe our new coevolutionary approach based on hosts and parasites View full abstract»

This paper deals with a software environment based on cellular automata devoted to musical experimentation, realised through a methodology by which, mathematical structures, produced by AL models, the general theory of signs, as proposed by Charles Peirce and music, which Consists of acoustic and perceptual relationships are connected. The main features of this environment are the following: 1. semiotics and musical language as tools for reading and interpreting mathematical configurations generated by cellular automata and other AL models; 2. musical expression as creative artefacts; 3. artificial universes as contexts in which to detect perceptual patterns and the correlated emotions music produces; 4. experimentations in aural perception in humans as a method for evolving musical artefacts. We can know the real world which is near us and the artificial world which is in the computer only by means of thought. Artificial worlds can be equivalent to the phenomenological world and both could be manipulated and organised by thought. We have to apply to artificial worlds the same method humans have developed in organising and giving meanings to the physical world. It'll be necessary to detect patterns generated by Artificial Life machines and to give them meanings. We are trying to apply this methodology through mathematics and music, combining them in a semiotic approach. View full abstract»

The origin of this paper lies in the fundamental question of how complexity arose in the development of life and how one could construct an artistic interactive system that can model and simulate this emergence of complexity. Based on the idea that interaction and communication between entities of a system are the driving forces for the emergence of higher and more complex structures than its mere parts, we propose to apply principles of Complex System Theory to the creation of an interactive, computer generated and audience participatory artwork on the Internet and to test whether complexity within the system can emerge View full abstract»

In this paper the aims and goals of artificial evolution are discussed in relation to two of the founding features of Alife: how to characterize the domain of the possible and the criterion of lifelikeness. lt is argued that artificial evolution should aim to understand the evolution of organizations and that this will bring about a better understanding of possible evolutions View full abstract»

The term "artificial life" describes research into synthetic systems that possess some of the essential properties of life. This interdisciplinary field includes biologists, computer scientists, physicists, chemists, geneticists, and others. Artificial life may be viewed as an attempt to understand high-level behavior from low-level rules -- for example, how the simple interactions between ants and their environment lead to complex trail-following behavior. An understanding of such relationships in particular systems can suggest novel solutions to complex real-world problems such as disease prevention, stock-market prediction, and data mining on the Internet.Since their inception in 1987, the Artificial Life meetings have grown from small workshops to truly international conferences, reflecting the field's increasing appeal to researchers in all areas of science. View full abstract»