Language Evolution and Computation Bibliography

If artificial organisms are constructed with the goal to better understand the behaviour of real organisms, artificial organisms that resemble human beings should possess a communication system with the same properties of human language. This chapter tries to identify nine such properties and for each of them to describe what has been done and what has to be done. Human language: (1) is made up of signals which are arbitrarily connected to their meanings, (2) has syntax and, more generally, its signals are made up of smaller signals, (3) is culturally transmitted and culturally evolved, (4) is used to communicate with oneself and not only with others, (5) is particularly sophisticated for communicating information about the external environment, (6) uses displaced signals, (7) is intentional and requires recognition of intentions in others, (8) is the product of a complex nervous system, (9) influences human cognition. Communication presupposes a shared worldview which depends on the brain, body, and adaptive pattern of the organisms that want to communicate, and this represents a critical challenge also for communication between robots and us.

In this chapter I briefly summarize views on adaptation and language, some relevant neurobiological and genetic facts, the presence or absence of recursion in animals, the possible role of genetic assimilation in language evolution, the prerequisites of language and the nature of the human adaptive suite, and the relative merits of proposed evolutionary scenarios for the origin of natural language. I highlight the special difficulty of this last major transition and a possible integrative modelling approach to the problem. Finally, I give a summary showing that the transition from early hominine societies with protolanguage to modern society with language indeed qualifies as a major transition.

Rarely do human behavioral scientists and scholars study language, music, and other forms of communication as strategiesaa means to some end. Some even deny that communication is the primary function of these phenomena. Here we draw upon selections of our earlier work to briefly define the strategy concept and sketch how decision theory, developed to explain the behavior of rational actors, is applied to evolved agents. Communication can then be interpreted as a strategy that advances the afitness interestsa of such agents. When this perspective is applied to agents with conflicts of interest, deception emerges as an important aspect of communication. We briefly review costly signaling, one solution to the problem of honest communication among agents with conflicts of interest. We also explore the subversion of cooperative signals by parasites and by plants defending themselves against herbivores, and we touch on biases in human gossip. Experiments with artificial embodied and communicating agents confirm that when there are conflicts of interest among agents, deception readily evolves. Finally, we consider signaling among super-organisms and the possible implications for understanding human music and language.

Statistical physics has proven to be a very fruitful framework to describe phenomena outside the realm of traditional physics. In social phenomena, the basic constituents are not particles but humans and every individual interacts with a limited number of peers, usually negligible compared to the total number of people in the system. In spite of that, human societies are characterized by stunning global regularities that naturally call for a statistical physics approach to social behavior, i.e., the attempt to understand regularities at large scale as collective effects of the interaction among single individuals, considered as relatively simple entities. This is the paradigm of Complex Systems: an assembly of many interacting (and simple) units whose collective behavior is not trivially deducible from the knowledge of the rules governing their mutual interactions. In this chapter we review the main theoretical concepts and tools that physics can borrow to socially-motivated problems. Despite their apparent diversity, most research lines in social dynamics are actually closely connected from the point of view of both the methodologies employed and, more importantly, of the general phenomenological questions, e.g., what are the fundamental interaction mechanisms leading to the emergence of consensus on an issue, a shared culture, a common language or a collective motion?

The evolution of human language allowed the efficient propagation of nongenetic information, thus creating a new form of evolutionary change. Language development in children offers the opportunity of exploring the emergence of such complex communication system and provides a window to understanding the transition from protolanguage to language. Here we present the first analysis of the emergence of syntax in terms of complex networks. A previously unreported, sharp transition is shown to occur around two years of age from a (pre-syntactic) tree-like structure to a scale-free, small world syntax network. The observed combinatorial patterns provide valuable data to understand the nature of the cognitive processes involved in the acquisition of syntax, introducing a new ingredient to understand the possible biological endowment of human beings which results in the emergence of complex language. We explore this problem by using a minimal, data-driven model that is able to capture several statistical traits, but some key features related to the emergence of syntactic complexity display important divergences.

In this chapter we introduce the area of research that attempts to study the evolution of communication in embodied agents through adaptive techniques, such us artificial evolution. More specifically, we illustrate the theoretical assumptions behind this type of research, we present the methods that can be used to realize embodied and communicating artificial agents, and we discuss the main research challenges and the criteria for evaluating progresses in this field.

Communication plays a central role in the biology of most organisms, particularly social species. Although the neurophysiological processes of signal production and perception are well understood, the conditions conducive to the evolution of reliable systems of communication remain largely unknown. This is a particularly challenging problem because efficient communication requires tight coevolution between the signal emitted and the response elicited. We conducted experimental evolution with robots that could produce visual signals to provide information on food location. We found that communication readily evolves when colonies consist of genetically similar individuals and when selection acts at the colony level. We identified several distinct communication systems that differed in their efficiency. Once a given system of communication was well established, it constrained the evolution of more efficient communication systems. Under individual selection, the ability to produce visual signals resulted in the evolution of deceptive communication strategies in colonies of unrelated robots and a concomitant decrease in colony performance. This study generates predictions about the evolutionary conditions conducive to the emergence of communication and provides guidelines for designing artificial evolutionary systems displaying spontaneous communication.

The evolution of communication requires the co-evolution of two abilities: the ability to send useful signals and the ability to react appropriately to perceived signals. This fact poses two related but distinct problems, which are often mixed up: (1) the phylogenetic problem regarding how can communication evolve if the two traits that are necessary for its emergence are complementary and seem to require each other for providing reproductive advantages; (2) the adaptive problem regarding how can communication systems that do not advantage both signallers and receivers in the same way emerge, given their altruistic character. Here we clarify the distinction, and provide some insights on how these problems can be solved in both real and artificial systems by reporting experiments on the evolution of artificial agents that have to evolve a simple food-call communication system. Our experiments show that (1) the phylogenetic problem can be solved thanks to the presence of producer biases that make agents spontaneously produce useful signals, an idea that is complementary to the well-known areceiver biasa hypothesis found in the biological literature, and (2) the adaptive problem can be solved by having agents communicate preferentially among kin, as predicted by kin selection theory. We discuss these results with respect to both the scientific understanding of the evolution of communication and the design of embodied and communicating artificial agents.

We use Evolutionary Robotics to design robot controllers in which decision-making mechanisms to switch from solitary to social behavior are integrated with the mechanisms that underpin the sensory-motor repertoire of the robots. In particular, we study the evolution of behavioral and communicative skills in a categorization task. The individual decision-making structures are based on the integration over time of sensory information. The mechanisms for switching from solitary to social behavior and the ways in which the robots can affect each otheras behavior are not predetermined by the experimenter, but are aspects of our model designed by artificial evolution. Our results show that evolved robots manage to cooperate and collectively discriminate between different environments by developing a simple communication protocol based on sound signaling. Communication emerges in the absence of explicit selective pressure coded in the fitness function. The evolution of communication is neither trivial nor obvious; for a meaningful signaling system to evolve, evolution must produce both appropriate signals and appropriate reactions to signals. The use of communication proves to be adaptive for the group, even if, in principle, non-cooperating robots can be equally successful with cooperating robots.

This work investigates the conditions in which a population of embodied agents evolved for the ability to display coordinated/cooperative skills can develop an ability to communicate, whether and to what extent the evolved communication system can complexify during the course of the evolutionary process, and how the characteristics of such communication system varies evolutionarily. The analysis of the obtained results indicates that evolving robots develop a capacity to access/generate information which has a communicative value, an ability to produce different signals encoding useful regularities, and an ability to react appropriately to explicit and implicit signals. The analysis of the obtained results allows us to formulate detailed hypothesis on the evolution of communication for what concern aspects such us: (i) how communication can emerge from a population of initially non-communicating agents, (ii) how communication systems can complexify, (iii) how signals/meanings can originate and how they can be grounded in agentsa sensory-motor states.

In this chapter we summarize the progresses that have recently been made in the study of the emergence of communication in artificial embodied agents along different dimensions, including the understanding of the adaptive roles of communication, the expressive power and organization complexity of signalling systems, the stability, robustness, and evolvability of communication, and the knowledge gain obtained with such models. Finally, we briefly discuss what we think are the most important open challenges for future research in this area.

This chapter introduces the cultural approach towards the question how a symbolic communication system could form in a population of agents. This approach emphasises the role of communication, high level cognition, and social interaction. The chapter introduces briefly the main challenges for working out this approach and which methods could be used to address these challenges.

This chapter gives an overview of different experiments that have been performed to demonstrate how a symbolic communication system, including its underlying ontology, can arise in situated embodied interactions between autonomous agents. It gives some details of the Grounded Naming Game, which focuses on the formation of a system of proper names, the Spatial Language Game, which focuses on the formation of a lexicon for expressing spatial relations as well as perspective reversal, and an Event Description Game, which concerns the expression of the role of participants in events through an emergent case grammar. For each experiment, details are provided how the symbolic system emerges, how the interaction is grounded in the world through the embodiment of the agent and its sensori-motor processing, and how concepts are formed in tight interaction with the emerging language.

In this chapter we explore several language games of increasing complexity. We first consider the so-called Naming Game, possibly the simplest example of the complex processes leading progressively to the establishment of human-like languages. In this framework, a globally shared vocabulary emerges as a result of local adjustments of individual word-meaning association. The emergence of a common vocabulary only represents a first stage while it is interesting to investigate the emergence of higher forms of agreement, e.g., compositionality, categories, syntactic or grammatical structures. As an example in this direction we consider the so-called Category Game. Here one focuses on the process by which a population of individuals manages to categorize a single perceptually continuous channel. The problem of the emergence of a discrete shared set of categories out of a continuous perceptual channel is a notoriously difficult problem relevant for color categorization, vowels formation, etc. The central result here is the emergence of a hierarchical category structure made of two distinct levels: a basic layer, responsible for fine discrimination of the environment, and a shared linguistic layer that groups together perceptions to guarantee communicative success.

Computational and robotic research into symbolic communication systems requires sophisticated tools. This chapter introduces Babel, a tool framework that has been developed to engage in extensive repeatable multi-agent experiments including experiments with embodied robots. A brief example is presented of how experiments are configured in this framework.