Catálogo de publicaciones - libros

In this paper, we discuss a viewpoint to regard individuals in a society as cognitive agents having internal dynamics, in order to study the dynamic nature of social structures. Internal dynamics is the autonomous changes of an agent’s internal states that govern his/her behavior. We first discuss the benefit of introducing internal dynamics into a model of humans and the dynamics of society. Then we propose a simple recurrent network with self-influential connection (SRN-SIC) as a model of an agent with internal dynamics. We report the results of our simulation in which the agents play a minority game. In the simulation, we observe the dynamics of the game as a macro structure itinerating among various dynamical states such as fixed points and periodic motions via aperiodic motions. This itinerant change of the macro structures is shown to be induced by the internal dynamics of the agents.

This paper proposes a novel interaction model of a human-robot cooperative task. The model employs CEA (Commands Embedded in Actions), which reduces a human cognitive load because it requires less explicit human-robot communication than direct commanding methods in conventional interaction models. We propose a guideline along which to design robots’ actions based on CEA, and apply it to a cooperative sweeping task by a human and a small mobile robot. CEA is experimentally shown to reduce the human cognitive load more than direct commanding methods do in this sweeping task.

In this paper, we discuss how to discover “skills” from motion data. Being able to understand how a skilled person moves enables beginners to make better use of their bodies and to become experts easier. However, only few attempts have so far been made for discovering skills from human motion data. To extract skills from motion data, we employ three approaches. As a first approach, we present association rule approach which extracts the dependency among the body parts to find the movement of the body parts performed by the experts. The second is an approach that extracts frequent patterns (motifs) from motion data. Recently, many researchers propose algorithms for discovering motifs. However, these algorithms require that users define the length of the motifs in advance. Our algorithm uses the MDL principle to overcome this problem so as to discover motifs with optimal length. Finally, we compare the motions of skilled tennis players and beginners, and discuss why skilled players can better serve.

EELC 2004, the First International Workshop on Emergence and Evolution of Linguistic Communication, was held in association with JSAI annual conference 2004, in Kanazawa, Japan.

We present a computational modeling approach to language based on an integrative view of the agent’s cognitive system. The emergence of linguistic abilities (both evolutionarily and developmentally) is strictly dependent on, and grounded in, other sensorimotor behaviors and cognitive abilities. Linguistic simulations imply the use of groups of autonomous agents that interact via language games to exchange information about the environment. The agents’ coordinated communication system is not externally imposed by the researcher, but emerges from the interaction between agents. We present a series studies on grounded simulation adaptive agent and on evolutionary and epigenetic robots.

The purpose of this paper is to develop a new formalism of language dynamics so that creole may emerge. Thus far, we modified the of the dynamics so as to change in accordance with the distribution of population of each language at each generation, and in addition, we introduced a new parameter called with which infants were exposed to other languages than mother tongues. Thus, we could observe creolization under limited conditions. In this paper, we revise the learning algorithm in our model, considering the amount of language input during the language acquisition period. Thus, the transition probability depends not only on the exposure rate but also on the amount of language input. With this model, we show that creolization occurs only when the influence of mother tongues and the socially dominant languages balance.

Recent studies on song birds reveal that their vocal communication has some common features with human language. In particular, a songbird called Bengalese finch has interesting vocal communication in which courtship song is arranged by finite-state grammar [6]. It has been hypothesized that the song grammar may have evolved as a result of sexual selection [10]. In order to explore the evolution of the song grammars, we model the co-evolution of male and female finches by asymmetric finite-state automata. In this paper, we demonstrate that song grammars could evolve by simple communication. We observe that a transition from lower complexity to higher complexity song grammars is driven by the changing of male birds’ courting strategy.

Human language may have evolved through a stage when words were combined into structured linear segments, before these segments were used as building blocks for a hierarchical grammar. This hypothesis is approached by examining the apparently ubiquitous prevalence of homophones. It suggests how, perhaps contrary to expectation, communicative capacity does not seem to be adversely affected by homophones, and how it is that they can be routinely used without confusion. These facts are principally explained by disambiguation through syntactic processing of short word sequences. Local sequential processing plays an underlying role in language production and perception, a hypothesis that is supported by evidence that small children engage in this process as soon as they acquire words. Experiments on a corpus of spoken English calculated the entropy for sequences of syntactically labelled words. They show there is a measurable advantage in decoding word strings when they are taken in short sequences, rather than as individual items. This suggests that grammatical fragments of speech could have been a stepping stone to a full grammar.

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.

We propose a role sharing model of language areas in which Broca’s area is for categorizing symbols used to represent rules stored and retrieved in other language areas. For example, at the syntactical level, the other language areas store rules represented with terminal symbols and also rules represented with non-terminal symbols, whereas Broca area invents non-terminals and forms abstract rules at language acquisition phase and converts a terminal symbol to a non-terminal corresponding to it to get an appropriate rule at performance phase. The model role of Broca’s area is supposed to be essential but minimal to support human language faculty. Under this assumption, the emergence of Broca’s area is hypothesized to have triggered the evolution of language supposing that the other mechanisms are fully evolved. The argument is based on the recent fMRI study of KE family members and related findings on a mutation of FOXP2 gene.