Language Evolution and Computation Bibliography

In this position paper we discuss how language influences the mind by comparing robots that have language with robots that do not have language. Robots with language respond more adaptively to objects belonging to different categories and requiring different behaviors compared to robots without language, and it is possible to show that categories of objects are represented differently in the neural network which controls the behavior of the two types of robots. By exposing the robots to sounds which co-vary systematically with specific aspects of their experience, the robots can distinguish nouns from verbs and can respond appropriately to simple noun-verb sentences. Robots can also be used to show that, while all animals develop a mental (neural) model of their environment which incorporates the co-variations among different aspects of their experiences, human beings develop a more analytical and modular model because specific sounds co-vary with different aspects of their experiences and this may explain why human beings have a more articulated and creative behavioral repertoire.

Cognitive Robotics can be defined as the study of cognitive phenomena by their modeling in physical artifacts such as robots. This is a very lively and fascinating field which has already given fundamental contributions to our understanding of natural cognition. Nonetheless, robotics has to date addressed mainly very basic, low-level cognitive phenomena like sensory-motor coordination, perception, and navigation, and it is not clear how the current approach might scale up to explain high-level human cognition. In this paper we argue that a promising way to do that is to merge current ideas and methods of embodied cognition with the Russian tradition of theoretical psychology which views language not only as a communication system but also as a cognitive tool, that is by developing a Vygotskyan cognitive robotics. We substantiate this idea by discussing several domains in which language can improve basic cognitive abilities and permit the development of high-level cognition: learning, categorization, abstraction, memory, voluntary control, and mental life.

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.

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.

The standard view of classical cognitive science stated that cognition consists in the manipulation of language-like structures according to formal rules. Since cognition is linguistic in itself, according to this view language is just a complex communication system and does not influence cognitive processes in any substantial way. This view has been criticized from several perspectives and a new framework (Embodied Cognition) has emerged that considers cognitive processes as non-symbolic and heavily dependent on the dynamical interactions between the cognitive system and its environment. But notwithstanding the successes of the embodied cognitive science in explaining low-level cognitive behaviors, it is still not clear whether and how it can scale up for explaining high-level cognition. In this paper we argue that this can be done by considering the role of language as a cognitive tool: i.e. how language transforms basic cognitive functions in the high-level functions that are characteristic of human cognition. In order to do that, we review some computational models that substantiate this view with respect to categorization and memory. Since these models are based on a very rudimentary form of non-syntactic language we argue that the use of language as a cognitive tool might have been an early discovery in hominid evolution, and might have played a substantial role in the evolution of language itself.

Like any other biological trait, communication can be studied from at least four perspectives: mechanistic, ontogenetic, functional, and phylogenetic. In this article, we focus on the following phylogenetic question: how can communication emerge, given that both signal-producing and signal-responding abilities seem to be adaptively neutral until the complementary ability is present in the population? We explore the problem of co-evolution of speakers and hearers with artificial life simulations: a population of artificial neural networks evolving a food call system. The core of the article is devoted to a careful analysis of the complex evolutionary dynamics demonstrated by our simple simulation. Our analyses reveal an important factor, which might solve the phylogenetic problem: the spontaneous production of good (meaningful) signals by speakers because of the need for organisms to categorize their experience in adaptively relevant ways. We discuss our results with respect both to previous simulative work and to the biological literature on the evolution of communication.

Selective pressures for the evolutionary emergence of human language tend to be interpreted as social in nature, i.e., for better social communication and coordination. Using a simple neural network model of language acquisition we demonstrate that even using language for oneself, i.e., as private or inner speech, improves an individual's categorization of the world and, therefore, makes the individual's behavior more adaptive. We conclude that language may have first emerged due to the advantages it confers on individual cognition, and not only for its social advantages.

The emergence of language in populations of primates that initially lacked language can be simulated with artificial organisms controlled by neural networks and living, evolving, and learning in artificial environment. Some simulations have already been done but most are a task for the future. We dis-cuss language evolution under two topics: language is learned from others on the basis of genetically inherited predispositions, and language has important influences on human cognition. We propose an evolutionary sequence accord-ing to which bipedalism and the emergence of the hands represent a selective pressure for developing an ability to predict the consequences of one's actions, this ability is the basis for learning by imitating other individuals, learning by imitating other individuals is applied to learning to imitate their communicative behaviour. The second topic include the consequences of language for various aspects of human cognition, especially when language is used to talk to oneself.

In this paper, we explore various adaptive factors that can influence the emergence of a communication system that benefits the receiver of signals (the hearer) but not the emitter (the speaker). Using computer simulations of a population of interacting agents whose behaviour is determined by a neural network, we show that a stable communication system does not emerge in groups of unrelated individuals because of its altruistic character. None the less, another set of simulations shows that the emergence of a language that confers an advantage only to hearers, not to speakers, is possible under at least three conditions: (1) if the hearer and the speaker tend to share the same genes, as predicted by kin selection theory; (2) if the population is `docile' and the communication system is culturally transmitted together with other adaptive behaviours, as predicted by Simon's docility theory; and (3) if the linguistic system is used not only for social communication, but also for talking to oneself, in particular as an aid to memory.

The paper describes a neural network model of early language acquisition with an emphasis on how language positively influences the categories with which the child categorizes reality. Language begins when the two separate networks that are responsible for nonlinguistic sensory-motor mappings and for recognizing and repeating linguistic sounds become connected together at 1 year of age. Language makes more similar the internal representations of different inputs that must be responded to with the same action and more different the internal representations of inputs that must be responded to with different actions.

The paper presents a computational model of language in which linguistic abilities evolve in organisms that interact with an environment. Each individual's behavior is controlled by a neural network and we study the consequences in the network's internal functional organization of learning to process different classes of words. Agents are selected for reproduction according to their ability to manipulate objects and to understand nouns (objects' names) and verbs (manipulation tasks). The weights of the agents' neural networks are evolved using a genetic algorithm. Synthetic brain imaging techniques are then used to examine the functional organization of the neural networks. Results show that nouns produce more integrated neural activity in the sensory-processing hidden layer, while verbs produce more integrated synaptic activity in the layer where sensory information is integrated with proprioceptive input. Such findings are qualitatively compared with human brain imaging data that indicate that nouns activate more the posterior areas of the brain related to sensory and associative processing, while verbs activate more the anterior motor areas.

This paper presents an Artificial Life and Neural Network (ALNN) model for the evolution of syntax. The simulation methodology provides a unifying approach for the study of the evolution of language and its interaction with other behavioral and neural factors. The model uses an object manipulation task to simulate the evolution of language based on a simple verb-noun rule. The analyses of results focus on the interaction between language and other non-linguistic abilities, and on the neural control of linguistic abilities. The model shows that the beneficial effects of language on non-linguistic behavior are explained by the emergence of distinct internal representation patterns for the processing of verbs and nouns.

The evolution of language implies the parallel evolution of an ability to respond appropriately to signals (language understanding) and an ability to produce the appropriate signals in the appropriate circumstances (language production). When linguistic signals are produced to inform other individuals, individuals that respond appropriately to these signals may increase their reproductive chances but it is less clear what the reproductive advantage is for the language producers. We present simulations in which populations of neural networks living in an environment evolve a simple language with an informative function. Signals are produced to help other individuals categorize edible and poisonous mushrooms, in order to decide whether to approach or avoid encountered mushrooms. Language production, while not under direct evolutionary pressure, evolves as a byproduct of the independently evolving perceptual ability to categorize mushrooms.

The aim of the paper is to show that an Artificial Life approach to language tends to change the research agenda on language which has been shared by both the symbolic paradigm and classical connectionism. Artificial Life Neural Networks (ALNNs) are different from classical connectionist networks because they interact with an independent physical environment; are subject to evolutionary, developmental, and cultural change, and not only to learning; and are part of organisms that have a physical body, have a life (are born, develop, and die), and are members of genetic and, sometimes, cultural populations. Using ALNNs to study language shifts the emphasis from research on linguistic forms and laboratory-like tasks to the investigation of the emergence and transmission of language, the use of language, its role in cognition, and language as a populational rather than as an individual phenomenon.