No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
... This research is a small part of research into the origins of intelligence and language using computer simulations (see e.g. [12,24]). In this respect it belongs to the branch of artificial intelligence that uses computers to increase the understanding of human intelligence, rather than to the branch of artificial intelligence that tries to build more intelligent computer programs. ...
... The emerging vowel systems are therefore truly the result of the interactions between the agents. The research is based on Steels' [22,23,24] ideas on the origins of language. Steels considers language as the result of a process of mainly cultural evolution, while the universal tendencies of language can be explained as the results of self-organisation under constraints of perception and production. ...
... Apparently the same mechanism can be used to learn an existing vowel system as well as to produce a sound system in a population where no sound system existed previously. This lends support to Steels' [23,24] hypothesis that the same mechanism that is responsible for the ability to learn language is responsible for the emergence of language in the first place. Computer simulations make it easy for the researcher to perform experiments like these, and thus provides an extra means to test and fine-tune linguistic theories. ...
This paper gives a concise overview of AI research in Belgium and The Netherlands. For each university, the main AI groups and research subjects are mentioned. In addition, a selection of research performed in Belgian and Dutch companies is presented.
... This research is a small part of research into the origins of intelligence and language using computer simulations (see e.g. [12,24]). In this respect it belongs to the branch of artificial intelligence that uses computers to increase the understanding of human intelligence, rather than to the branch of artificial intelligence that tries to build more intelligent computer programs. ...
... The emerging vowel systems are therefore truly the result of the interactions between the agents. The research is based on Steels' [22,23,24] ideas on the origins of language. Steels considers language as the result of a process of mainly cultural evolution, while the universal tendencies of language can be explained as the results of self-organisation under constraints of perception and production. ...
... Apparently the same mechanism can be used to learn an existing vowel system as well as to produce a sound system in a population where no sound system existed previously. This lends support to Steels' [23,24] hypothesis that the same mechanism that is responsible for the ability to learn language is responsible for the emergence of language in the first place. Computer simulations make it easy for the researcher to perform experiments like these, and thus provides an extra means to test and fine-tune linguistic theories. ...
this paper tries to explain the emergence and structure of systems of speech sounds. It investigates how a coherent system of speech sounds can emerge in a popu- lation of agents and how the constraints under which the system emerges impose structure through self-organisation. If self-organisation can explain structure, then innate and biologi- cally evolved mechanisms are not necessary. This effectively decreases the number of linguistic phenomena that have to be explained by biological evolution
... Opponents view human language as an emergent phenomenon, grounded in general cognitive abilities which do not necessarily involve "symbol manipulation"; no innate language organ needs to be postulated and neither do language genes. Transmission of language is cultural rather than genetic, language universals arise from universal contexts rather than a universal grammar and the transition from animal to human communication systems should be viewed gradual rather than principal (Steels, 1997a, MacWhinney, 1999, Elman et al., 1998. ...
... The most coherent of these alternative views, postulate cultural evolution as the main mechanism to generate the grammatical complexity of human languages (see figure 1). Several studies have shown that, if in a population of learners language elements are transmitted culturally, these elements themselves can form units of selection and more powerful language capabilities can emerge (Steels, 1997a, Kirby, 1999a. Interestingly, the view that language elements are the replicating entities and form a level of selection, is nicely in accordance with the view of major transitions and multi-level evolution (Hogeweg, 1989(Hogeweg, , 1998). ...
... An intriguing approach, however, has been added to the existing ones: mathematical and computer modeling of language origins, pioneered by Hurford (1989), Steels (1997a), Batali (1994), Hashimoto & Ikegami (1996). An effort is made in this line of research to understand the dynamics of language evolution by studying simple models ("minimal models") of communicating agents. ...
This paper reports on an internship at the department of theoretical biology at Utrecht University between October 1998 and February 1999. Aim of the project was to implement and study the computational model of language evolution reported by Hashimoto & Ikegami (1996) and to reproduce their results. The project was undertaken as a preparation for more original research for a master's thesis, and among other things to improve programming skills and to get more familiar with language evolution and modeling methodology. This report consists of six sections, discussing the origin of language debate, the Hashimoto - Ikegami model, the developed implementation, the main results with some hints on subsequent work, conclusions and a short self-evaluation. 1 Introduction 1.1 the origins of language The transition from short, finite communication systems found in many animal species, to the open ended language system of humans, is considered to be one of the major transitions in evolu...
... The emerging vowel systems are therefore truly the result of the interactions between the agents. The research is mostly based on Steels' (Steels 1996(Steels , 1997(Steels , 1998) ideas on the origins of language, but fits in the larger recent tradition of studying the origins of language using computer simulations of populations (see also Hurford, this volume and Kirby, this volume). Steels considers language as the result of a process of mainly cultural evolution, while the universal tendencies of language can be explained as the results of selforganisation under constraints of perception and production. ...
... Apparently the same mechanism can be used to learn an existing vowel system as well as to produce a sound system in a population where no sound system existed previously. This lends support to Steels' (Steels 1997(Steels , 1998 thesis that the same mechanism that is responsible for the ability to learn language is responsible for the emergence of language in the first place. The use of computer simulations makes it easy for the researcher to perform experiments like these, and thus provides an extra means to test and fine-tune linguistic theories. ...
Language has no counterpart in the animal world. Unique to Homo sapiens, it appears inseparable from human nature. But how, when and why did it emerge? The contributors to this volume - linguists, anthropologists, cognitive scientists, and others - adopt a modern Darwinian perspective which offers a bold synthesis of the human and natural sciences. As a feature of human social intelligence, language evolution is driven by biologically anomalous levels of social cooperation. Phonetic competence correspondingly reflects social pressures for vocal imitation, learning, and other forms of social transmission. Distinctively human social and cultural strategies gave rise to the complex syntactical structure of speech. This book, presenting language as a remarkable social adaptation, testifies to the growing influence of evolutionary thinking in contemporary linguistics. It will be welcomed by all those interested in human evolution, evolutionary psychology, linguistic anthropology, and general linguistics.
... animal communication, innateness and language universals, remains controversial and inconclusive. An intriguing alternative approach has emerged: mathematical and computational modeling of language origins [Hurford, 1989;Steels, 1997b;Hashimoto & Ikegami, 1996;Nowak & Krakauer, 1999]. In this line of research an effort is made to understand the dynamics of language evolution by studying simple models ("minimal models") of communicating agents. ...
... These models help to generate new hypotheses, to evaluate how generic certain properties are, to tackle the supposed self-evidence of arguments and to find a minimal set of assumptions sufficient to explain a phenomenon. Their main contribution so far is, that they have shown the plausibility of cultural evolution as a mechanism in the development of more complex languages [Steels, 1997b;De Jong, 1998;De Boer & Vogt, 1999;Batali, 1997;Kirby, 1999Kirby, , 2000. ...
... Mobile robots equipped with sensors such as cameras and sonars provide other means for perceptual grounding. This type of grounding underlies the approach to language development described by Steels (1998b). A perceptual grounding is also used in the robot soccer domain, for example to connect the concept "ball" in the robot's internal workings to the football in the playing field. ...
... Since the expressivity of agent communication has not come close to that of human language, ostensive teaching has proven useful to teach concepts in an agent communication language, e.g. (Steels, 1998b;Obitko and Marik, 2003;Tzitzikas and Meghini, 2003). Our use of ostensive definitions is described in Chapter 4 and 5. Some philosophical limitations of the method are described below. ...
Software agents sharing the same ontology can exchange their knowledge fluently as their knowledge representations are compatible with respect to the concepts regarded as relevant and with respect to the names given to these concepts. However, in open heterogeneous multi-agent systems, this scenario would be very unlikely, because it would require all involved system developers to reach consensus on which ontology to use. Furthermore, different agents may regard different concepts as relevant which causes their ontologies to differ in granularity and scope. In such an environment, the agents must possess the right conversational skills to effectively exchange information even when the speaker's ontology is only approximately translatable to the hearer's ontology. Furthermore, the agents must be able to autonomously establish an ontology translation by exchanging parts of their ontologies. In this thesis, we propose a layered communication protocol in which the agents gradually build towards a semantically integrated system by establishing minimal and effective shared ontologies. We will use the formal notions of sound and lossless communication to state the requirement that sufficient information should flow between the agents in a correct manner. The communication protocol detects when communication is ineffective and applies techniques for ontology exchange to build a shared ontology of minimal size. In this way, the agents exchange ontological information on an as-need basis. Agents first try to cope with the situation as it is; when communication fails to be effective, the agents seek a minimal solution which solves their communication problem at hand. The communication mechanism consists of three layers. The upper layer of the protocol is the Normal Communication Protocol (NCP) which deals with the kind of social interaction that agents normally exhibit when no ontology problems exist in the system. Every conversation starts in the NCP layer. If the agents fail to understand each other, the agents switch to the middle layer in the protocol which is the Concept Definition Protocol (CDP). In this layer, the agents explain the meaning of a concept to each other by exchanging concept definitions. The meaning of a concept is explained in terms of other concepts. If the communication difficulties are so severe that the agents do not even understand each other's concept definitions, the agents switch to the lowest layer in the protocol, i.e. the Concept Explication Protocol (CEP). In CEP, the agents exchange the meaning of a concept using non-symbolic communication, e.g. by pointing to examples of the concept. We tested our system, called ANEMONE, in three ways. Firstly, we provide a mathematical analysis. By formalizing the communication protocol, we can give solid proofs that it possesses the desirable properties. Secondly, we perform simulation experiments. By making the agents communicate in a simulation environment, we can analyze whether the agents indeed build a minimal communication vocabulary. Thirdly, we describe a case study with heterogeneous internet news agents. We show how these agents successfully exchange information on news articles, despite initial difficulties caused by heterogeneous ontologies.
... But as interactions become more complex, the cognition (Gershenson, 2004a) required by the elements should also be increased, since they need to process more information. New meanings can be learned (Steels, 1998;de Jong, 2000) to adapt to the changing conditions. These can be represented as "concepts" (Gärdenfors, 2000), or encoded, e.g. in the weights of a learning neural network (Rojas, 1996). ...
... Such self-organization has been shown in different simulations of the evolution of language (Hutchins and Hazelhurst, 1995;Steels, 1998Steels, , 2003de Boer, 1999;de Jong, 2000;Wiesman et al., 2002). Here, interacting software agents or robots try to develop a shared lexicon, so that they interpret the same expressions, symbols, or "words" in the same way. ...
Complex systems are usually difficult to design and control. There are several particular methods for coping with complexity, but there is no general approach to build complex systems. In this thesis I propose a methodology to aid engineers in the design and control of complex systems. This is based on the description of systems as self-organizing. Starting from the agent metaphor, the methodology proposes a conceptual framework and a series of steps to follow to find proper mechanisms that will promote elements to find solutions by actively interacting among themselves. The main premise of the methodology claims that reducing the “friction” of interactions between elements of a system will result in a higher “satisfaction” of the system, i.e. better performance. A general introduction to complex thinking is given, since designing self-organizing systems requires a non-classical thought, while practical notions of complexity and self-organization are put forward. To illustrate the methodology, I present three case studies. Self-organizing traffic light controllers are proposed and studied with multi-agent simulations, outperforming traditional methods. Methods for improving communication within self-organizing bureaucracies are advanced, introducing a simple computational model to illustrate the benefits of self-organization. In the last case study, requirements for self-organizing artifacts in an ambient intelligence scenario are discussed. Philosophical implications of the conceptual framework are also put forward.
... We use agent-based models to study the hypothesis that adult language contact caused morphological simplification, focusing on inflectional verb morphology. In our model, a population of agents play a language game (Steels, 1998), in which they try to communicate concepts (verb+person) using inflected verb forms. The model is initialized with phonetic representations of verb forms in Lewoingu Lamaholot (1) (cf. ...
... The task for agents is to successfully communicate about concepts in the world, roughly inspired by a Lewis signalling game [18] or naming game [29]. Every iteration, every agent in a population speaks: it picks a concept, produces a form based on that concept and sends it to the listener. ...
In this paper, we propose an outline for linguistic research on language change,
as observed in the languages of the world, using neural agent-based models of
emergent communication. We describe how such models could be used to study
morphological simplification, using a case study of language contact in Eastern
Indonesia. A neural architecture is used to represent hypothesized cognitive mechanisms of language change: a generalization mechanism, the procedural/declarative
model, and a phonological mechanism, the hyper & hypo articulation model, that
involves a theory of mind of the listener.
... Originally, the notion of self-organisation was introduced in the field of cybernetics [10,11]. These seminal ideas quickly propagated to almost all branches of science, including physics [1,12], biology [2,13], computer science [14,15], language analysis [16,17], network management [18,19], behavioral analysis [20,21] and neuroscience [22,23], to name a few. Despite this success, most working definitions of self-organisation still avoid formal definitions and rely on intuitions following an "I know when I see it" logic, which might eventually prevent further systematic developments [24]. ...
Self-organisation lies at the core of fundamental but still unresolved scientific questions, and holds the promise of de-centralised paradigms crucial for future technological developments. While self-organising processes have been traditionally explained by the tendency of dynamical systems to evolve towards specific configurations, or attractors, we see self-organisation as a consequence of the interdependencies that those attractors induce. Building on this intuition, in this work we develop a theoretical framework for understanding and quantifying self-organisation based on coupled dynamical systems and multivariate information theory. We propose a metric of global structural strength that identifies when self-organisation appears, and a multi-layered decomposition that explains the emergent structure in terms of redundant and synergistic interdependencies. We illustrate our framework on elementary cellular automata, showing how it can detect and characterise the emergence of complex structures.
... Steels' idea is that language can be looked at as a pattern of organization that is emerging in such multi-agent systems (Steels, 1998). (Furthermore he thinks the nature of intelligence in general can be described this way (Steels, 1996)). ...
... Although such direct feedback might be considered unrealistic, it is plausible to assume that speakers do obtain clues about the succesfullness of their utterance, e.g., from the failure to achieve a communicative goal. As such, the feedback in this model is a simplification of this process (compare the ''language games'' in Steels 1998 andde Boer 2001). ...
Language change has been described as an unintended eect of language in use(Keller 1994). In this view, change results from the way individuals use their language; the challenge is thus to explain change and its properties in terms of factors operating on the individual level, and population dynamics. An intriguing example of such a phenomenon is the finding that language change shows some highly regular tendencies. This has recently received considerable attention in the literature (Bybee et al. 1994; Heine and Kuteva 2002; Traugott and Dasher 2002; Hopper and Traugott 2003). In unrelated languages, similar words often change in similar ways, along similar trajectories of development. This phenomenon is called unidirectionality, and it is an important part of processes of grammaticalization, items changing from a lexical meaning to a grammatical function. It has been claimed that around 90-99% of all processes of grammaticalization are unidirectional (Haspelmath 1999). This article explores several mechanisms that may lead to language change, and examines whether they may be responsible for unidirectionality. We use a cultural evolutionary computational model with which the effects of individual behavior on the group level can be measured. By using this approach, regularities in semantic change can be explained in terms of very basic mechanisms and aspects of language use such as the frequency with which particular linguistic items are used. One example is that frequency dierences by themselves are a strong enough force for causing unidirectionality. We argue that adopting a cultural evolutionary approach may be useful in the study of language change.
... A particular proposal for the cognitive functions, ecological conditions and interactions patterns that are needed for language is operationalized and then used to simulate the emergence of language systems in populations of artificial agents. This approach started in the early nineties (see an early review in Steels, 1998) and has flourished considerably during the past decade (Lyon et al, 2007;Minett and Wang, 2005;Nolfi and Miroli, 2010). The language systems that emerge in these computational experiments are of course never equal to English or Hindi, given the historical contingencies that play a role in normal cultural language evolution, however, by using strategies reconstructed from human languages or by scaffolding the experiment with a vocabulary or partial grammar from an existing human language, the artificial languages are closer to a human source language, which makes the experiment more relevant and the evolution easier to follow. ...
This chapter introduces a new experimental paradigm for studying issues in the grounding of language and robots, and the integration of all aspects of intelligence into a single system. The paradigm is based on designing and implementing artificial agents so that they are able to play language games about situations they perceive and act upon in the real world. The agents are not pre-programmed with an existing language but with the necessary cognitive functions to self-organize communication systems from scratch, to learn them from human language users if there are sufficiently frequent interactions, and to participate in the on-going cultural evolution of language. © 2012 Springer Science+Business Media, LLC. All rights reserved.
... In a technical mathematical sense they are non-linear systems. As has already been observed by Steels (1998) language is a complex (non-linear) dynamic system. The behavior of such systems is not easy to predict or even to describe. ...
This paper describes the uses of computer models in studying the evolution of language. Language is a complex dynamic system
that can be studied at the level of the individual and at the level of the population. Much of the dynamics of language evolution
and language change occur because of the interaction of these two levels. It is argued that this interaction is too complicated
to study with pen-and-paper analysis alone and that computer models, therefore, provide a useful tool for understanding language
evolution. Different techniques are presented: direct optimization, genetic algorithms and agent-based models. Of each of
these techniques, an example is briefly presented. Also, the importance of correctly measuring and presenting the results
of computer simulations is stressed.
... This paper does not model the emergence of phonemic coding as such, as is done in for example [5], or the evolution of learning behavior [6], but focuses on the dynamics of the interaction between cultural and biological evolution. The influence of culture is important in the evolution of language e. g. [7,8], but it could be seen as a complicating factor in the transition from a holistic to a phonemic sound system: languages adapt to the abilities of the language users [9] and it can therefore be assumed that a population of holistic learners will shape the language towards holism. As it is assumed that holism is a better strategy for small systems, the system that exists in the population will at first be optimized for holistic learners. ...
This paper investigates the interaction between cultural evolution and biological evolution in the emergence of phonemic coding
in speech. It is observed that our nearest relatives, the primates, use holistic utterances, whereas humans use phonemic utterances.
It can therefore be argued that our last common ancestor used holistic utterances and that these must have evolved into phonemic
utterances. This involves co-evolution between a repertoire of speech sounds and adaptations to using phonemic speech. The
culturally transmitted system of speech sounds influences the fitness of the agents and could conceivably block the transition
from holistic to phonemic speech. This paper investigates this transition using a computer model in which agents that can
either use holistic or phonemic utterances co-evolve with a lexicon of words. The lexicon is adapted by the speakers to conform
to their preferences. It is shown that although the dynamics of the transition are changed, the population still ends up of
agents that use phonemic speech.
... [8,9,3,10]) and artificial agents, including very simple ones, (e.g. [5,11,12,6,13,14,15,16]). ...
Neuroscientists have suggested that the mirror-neurons in our primate ancestors may have provided a substrate for the emergence
of language in humans. Simulation studies of the emergence of language, using minimal implementations of proposed mechanisms,
are a way to assess their explanatory power for the emergence and evolution of communication. In this work, we study the emergence
and stability of linguistic labelling in a communities of agents with mirror-neuron mechanisms for associating deictic reference
with speech utterances. These minimal agents possessing a built-in mirror-neuron style temporal recurrent neural network architecture
are capable of perceiving and carrying out deixis (‘pointing’) to refer to others in their group, as well as producing and
perceiving utterances of another agent in their group. They are able to generate and learn temporally extended phonetic utterances
(‘names’) and associate these to deictic referents. Thus, the agents utter what they hear, and tend refer to the same entities
as another agent that they watch when it points. Previous work has shown the emergence and stability of arbitrary names generated
by the agents in certain fixed topologies of interaction. In this work, we systematically study the effects of different interaction
topologies on the dynamics of convergence to a common vocabulary in the population, and its stability over time. Results show
that certain topologies of interaction to be more conducive than others to the emergence of a stable vocabulary. Moreover,
some topologies of interaction (such as cycles) are seen to yield instability and to amplify feedback given the mirror-neuron
system. Linguistic convergence and change bear similarity to those of natural language. Homophony and multiple referents of
particular proto-words may also emerge. In light of results, we suggest that mechanisms for confirming joint-attention and
for suppression of mirroring could play an essential role in maintaining stability in the emergence of linguistic reference.
... In the talking heads (TH) and related experiments (see e.g. [Steels, 1998]) a population of robots develop a shared ontology and lexicon to communicate about differently shaped and colored objects by playing language games. Each game two agents are presented with a collection of objects called the context. ...
In this paper it is argued that the way the world is con- ceptualized for language is language dependent and the result of negotiation between language users. This is investigated in a computer experiment in which a popu- lation of artiflcial agents construct a shared language to talk about a world that can be conceptualized in multi- ple and possibly con∞icting ways. It is argued that the establishment of a successful communication system re- quires that feedback about the communicative success is propagated to the ontological level, and thus that lan- guage shapes the way we conceptualize the world for communication. Introduction and Research Question Language and communication involve many aspects of human cognition including the sensory-motor schema's needed to observe the world, the social abilities for es- tablishingjointattentionandcommunicativeintentand the mechanisms responsible for parsing and producing abstract grammatical expressions. A key issue here is how a population of distinct and onlylocallyinteractingagents(languageusers)canagree upon a global language. It is commonly accepted that at least part of the answer is self-organization: a con- sensus is reached through repeated peer-to-peer negoti- ations about how to expresssome meaning. Aprerequisiteforthis,whichisoftenneglected,isthat theagentsalreadyhavetoagreeuponthesetofexpress- ible meanings. It is implicitly assumed that all agents
... The importance of embodiment in word acquisition is featured in a variety of sources ( Plunkett et al. 1992, Steels 1997, Cohen et al. 2001, Roy and Pentland 2002, Yu et al. 2003, Weng et al. 2003) (also see Lungarella et al. (2004) for a good survey of developmental models). Among them, Plunkett et al. (1992) built a connectionist model of word learning in which a process termed auto-association maps preprocessed images with linguistic labels. ...
Language is about symbols, and those symbols must be grounded in the physical world. Children learn to associate language with sensorimotor experiences during their development. In light of this, we first provide a computational account of how words are mapped to their perceptually grounded meanings. Moreover, the main part of this work proposes and implements a computational model of how word learning influences the formation of object categories to which those words refer. This model simulates the bi-directional relationship between word and object category learning: (1) object categorization provides mental representations of meanings that are mapped to words to form lexical items; (2) linguistic labels help object categorization by providing additional teaching signals; and (3) these two learning processes interplay with each other and form a developmental feedback loop. Compared with the method that performs these two tasks separately, our model shows promising improvements in both word-to-world mapping and perceptual categorization, suggesting a unified view of lexical and category learning in an integrative framework. Most importantly, this work provides a cognitively plausible explanation of the mechanistic nature of early word learning and object learning from co-occurring multisensory data.
... We will show later that the co-occurrence assumption is not reliable and appropriate for modeling language acquisition. Luc Steels et al. [Steels andVogt, 1997, steels, 1997] reported the experiments in which autonomous visually grounded agents bootstrap meanings and language through adaptive language games. He argued that language is an autonomous evolving adaptive system maintained by a group of distributed agents without central control. ...
Language is about symbols and those symbols must be grounded in the physical environment during human development. Most recently, there has been an increased awareness of the essential role of inferences of speakersU referential intentions in grounding those symbols. Experiments have shown that these inferences as revealed in eye, head and hand movements serve as an important driving force in language learning at a relatively early age. The challenge ahead is to develop formal models of language acquisition that can shed light on the leverage provided by embodiment. We present an implemented computational model of embodied language acquisition that learns words from natural interactions with users. The system can be trained in unsupervised mode in which users perform everyday tasks while providing natural language descriptions of their behaviors. We collect acoustic signals in concert with user-centric multisensory information from nonspeech modalities, such as userUs perspective video, gaze positions, head directions and hand movements. A multimodal learning algorithm is developed that firstly spots words from continuous speech and then associates action verbs and object names with their grounded meanings. The central idea is to make use of non-speech contextual information to facilitate word spotting, and utilize userUs attention as deictic reference to discover temporal correlations of data from different modalities to build lexical items. We report the results of a series of experiments that demonstrate the effectiveness of our approach.
... (Savage-Rumbaugh & Brakke 1996, Herman & Austad 1996, Bickerton 1990) and artifacts (e.g. (Steels 1996, Steels 1997, Steels 1998b, Hurford et al. 1998, Billard & Hayes 1999, Billard & Dautenhahn 1999, Nehaniv 2000). The various degrees to which non-human animals exhibit various such capacities suggest that many linguistic and related phenomena can be explained in a manner parsimious also with human language capacties, as part of a general cognitive ethology of 'animal minds' (Griffin 1976, Griffin 1992) that does not require treating human abilities as somehow discontinous from the rest of nature. ...
In this paper we describe a series of interaction games in which an elementary `protolanguage ' is generated based on innate deictic ability in a community of asynchronously interacting connectionist agents. Deixis---or pointing (and its perception)---combined with interaction and `speech' generation ability support the emergence and drift of shared names within a family of `parent' and `child' agents. An agent may point at a `referent' agent and then associate any perceived speech with the referent. Perceived speech is either received as a response from an interlocutor or else the agent itself may generate speech, if it points but hears no `name' in response. Deixis toward a referent followed by perceived speech is associated (`learned') by the agent's recurrent time-delay connectionist artificial neural network establishing a basis for naming. Therefore subsequent deixis with the same referent triggers attempted reproduction of the associated ensuing speech. The interaction may also be `overheard' by other agents which then may associate the particular deitic acts perceived with the speech heard, resulting in a propagation of naming conventions. Interaction, via pointing and speaking, is asynchronously scheduled, and in this respect biologically plausible. Issues related to mirrorneurons, attention and deictic gaze, development of vocabularies, as well as synonomy, lexical drift and convergence are also discussed briefly. This illustrates the dynamics and development of deixis-grounded naming systems in (asynchronous) interaction and thus extends the work of Steels on the emergence of vocabularies in communities of agents as well as the work of Billard, Hayes, and Dautenhahn on the grounding and learning vocabularies in robots and agents using connectionist methods.
... We study this transition in a computational model of an evolving population of communicating agents. The main advantages of computational and mathematical models such as (Hurford, 1989;Steels, 1997;Hashimoto & Ikegami, 1996;Nowak & Krakauer, 1999), are that they are relatively precise and productive, in the sense that they generate new concepts and hypotheses. The main contribution so far is that they have shown the plausibility of 1 Present address: Sony CSL, 6, Rue Amyot, 75005, Paris, France; webpage: www-binf.bio.uu.nl/ jelle cultural evolution as a mechanism in the development of more complex languages (De Jong, 1998;De Boer & Vogt, 1999;Batali, 1997;Kirby, 2000). ...
We study the evolution of syntax in a simple multi-agent model. The fitness of agents in the model is not a fixed function of the individual languages, but depends on their communicative success in the group and thus on the composition of the population. This fact significantly alters the evolutionary dynamics, and can both facilitate and hinder the development of syntactic language. The results challenge the traditional picture of the transition towards syntactical language.
... But as interactions turn more complex, the cognition [Gershenson 2004a] required by the elements should also be increased. New meanings can be learned [Steels 1998;de Jong 2000] to adapt to the changing conditions. These can be represented as "concepts" [Gärdenfors 2000], or encoded, e.g., in the weights of a learning neural network [Rojas 1996]. ...
Our technologies complexify our environments. Thus, new technologies need to deal with more and more complexity. Several efforts have been made to deal with this complexity using the concept of self-organization. However, in order to promote its use and understanding, we must first have a pragmatic understanding of complexity and self-organization. This paper presents a conceptual framework for speaking about self-organizing systems. The aim is to provide a methodology useful for designing and controlling systems developed to solve complex problems. First, practical notions of complexity and self-organization are given. Then, starting from the agent metaphor, a conceptual framework is presented. This provides formal ways of speaking about "satisfaction" of elements and systems. The main premise of the methodology claims that reducing the "friction" or "interference" of interactions between elements of a system will result in a higher "satisfaction" of the system, i.e. better performance. The methodology discusses different ways in which this can be achieved. A case study on self-organizing traffic lights illustrates the ideas presented in the paper.
Self-organisation lies at the core of fundamental but still unresolved scientific questions, and holds the promise of de-centralised paradigms crucial for future technological developments. While self-organising processes have been traditionally explained by the tendency of dynamical systems to evolve towards specific configurations, or attractors, we see self-organisation as a consequence of the interdependencies that those attractors induce. Building on this intuition, in this work we develop a theoretical framework for understanding and quantifying self-organisation based on coupled dynamical systems and multivariate information theory. We propose a metric of global structural strength that identifies when self-organisation appears, and a multi-layered decomposition that explains the emergent structure in terms of redundant and synergistic interdependencies. We illustrate our framework on elementary cellular automata, showing how it can detect and characterise the emergence of complex structures.
This paper reviews the importance of the interaction between cultural evolution, biological evolution and individual cognition in understanding the cognitive nature of speech sound systems. Because of the effect of cultural evolution, typological properties of languages do not reflect individual cognitive mechanisms directly. In addition, the interaction between cultural evolution and biological evolution deeply influences what kind of cognitive adaptations to speech can evolve.Theoretical work and computer simulation have shown that at least two kinds of adaptations to speech and language can evolve. One consists of weak biases to discrete features of language (such as word order) that convey a functional advantage. The other consists of stronger adaptations involving continuous traits in which language and biology can co-evolve (the vocal tract being a possible example of such a co-evolved adaptation). Experimental work is underway to identify how exactly cultural and biological evolution interact in human speech and language, and what cognitive mechanisms (if any) may have undergone selective pressure related to speech and language.The paper reviews a number of studies that take the evolutionary perspective, focusing notably on agent-based computer simulations and on experimental work that simulates evolution in the laboratory or experimental work that investigates the interaction between individual learning behavior and cultural transmission directly. The paper argues that taking the evolutionary perspective (both cultural and biological, as well as their interaction) into account is necessary for a full understanding of the cognitive nature of language and speech.
To understand evolutionary aspects of communication is to understand the evolutionary development of the meaning relations between language and the world. Such meaning relations are established by the application of the interactive systems of semantic games. Subsumed under the evolutionary framework of repeated games, semantics in such games refers to the cases in which stable meanings survive populations of strategically interacting players. The viability of compositionality, common ground and salience in such evolutionary games is assessed. Foundationally, the discussion is rooted in Charles S. Peirce's pragmatist philosophy.
Language evolution might have preferred certain prior social configurations
over others. Experiments conducted with models of different social structures
(varying subgroup interactions and the role of a dominant interlocutor) suggest
that having isolated agent groups rather than an interconnected agent is more
advantageous for the emergence of a social communication system. Distinctive
groups that are closely connected by communication yield systems less like
natural language than fully isolated groups inhabiting the same world.
Furthermore, the addition of a dominant male who is asymmetrically favoured as
a hearer, and equally likely to be a speaker has no positive influence on the
disjoint groups.
This paper addresses the emergence of a common phonetic code in a society of communicating speech agents using evolutionary
techniques. Predictions for the large vowel systems of the world’s languages using the Maximum Use of Available distinctive
Features (MUAF) principle are discussed. Simulations of the use of supplementary phonetic features in large vowel systems
are presented. These experimental results show how simple local rules of interaction between speaking agents are sufficient
to explain some of the universal characteristics of the phonological structure of the world’s languages.
We examine the influence of inferring interlocutors' referential intentions from their body movements at the early stage of lexical acquisition. By testing human participants and comparing their performances in different learning conditions, we find that those embodied intentions facilitate both word discovery and word-meaning association. In light of empirical findings, the main part of this article presents a computational model that can identify the sound patterns of individual words from continuous speech, using nonlinguistic contextual information, and employ body movements as deictic references to discover word-meaning associations. To our knowledge, this work is the first model of word learning that not only learns lexical items from raw multisensory signals to closely resemble infant language development from natural environments, but also explores the computational role of social cognitive skills in lexical acquisition.
Speech is a perceptuo-motor system. A natural computational modeling framework is provided by cognitive robotics, or more precisely speech robotics, which is also based on embodiment, multimodality, development, and interaction. This paper describes the bases of a virtual baby robot which consists in an articulatory model that integrates the non-uniform growth of the vocal tract, a set of sensors, and a learning model. The articulatory model delivers sagittal contour, lip shape and acoustic formants from seven input parameters that characterize the configurations of the jaw, the tongue, the lips and the larynx. To simulate the growth of the vocal tract from birth to adulthood, a process modifies the longitudinal dimension of the vocal tract shape as a function of age. The auditory system of the robot comprises a “phasic” system for event detection over time, and a “tonic” system to track formants. The model of visual perception specifies the basic lips characteristics: height, width, area and protrusion. The orosensorial channel, which provides the tactile sensation on the lips, the tongue and the palate, is elaborated as a model for the prediction of tongue-palatal contacts from articulatory commands. Learning involves Bayesian programming, in which there are two phases: (i) specification of the variables, decomposition of the joint distribution and identification of the free parameters through exploration of a learning set, and (ii) utilization which relies on questions about the joint distribution.
Two studies were performed with this system. Each of them focused on one of the two basic mechanisms, which ought to be at work in the initial periods of speech acquisition, namely vocal exploration and vocal imitation. The first study attempted to assess infants’ motor skills before and at the beginning of canonical babbling. It used the model to infer the acoustic regions, the articulatory degrees of freedom and the vocal tract shapes that are the likeliest explored by actual infants according to their vocalizations. Subsequently, the aim was to simulate data reported in the literature on early vocal imitation, in order to test whether and how the robot was able to reproduce them and to gain some insights into the actual cognitive representations that might be involved in this behavior.
Speech modeling in a robotics framework should contribute to a computational approach of sensori-motor interactions in speech communication, which seems crucial for future progress in the study of speech and language ontogeny and phylogeny.
ResearchGate has not been able to resolve any references for this publication.