HOME NEW PERSONAL STUDY LECTURE FAVORITE LINK
Dagstuhl Seminar
Proceedings 09121
Normative
Multi-Agent Systems ,
G. Boella, P. Noriega, G. Pigozzi, H. Verhagen (Eds.)
Abstract
Corporate Social Responsibility (CSR), such as pro-environmental
behaviour and fair trade, is a kind of normative behaviour by private
companies to provide a quasi-public good.We study dissemination
mechanism of CSR with a multi-agent model in which corporation agents
and consumer agents interact with each other. We show that the mechanism
to disseminate CSR is a positive feedback between the corporations'
popularity seeking behaviour and the consumers' social learning in which
CSR-seeking preference is evaluated according to both the local average
of the preferences of surrounding consumers and the global average of
the investment in CSR by all corporations. We also discuss an
institutional design to establish CSR from an objectionable social
state.
Keywords
Keywords. CSR (corporate social responsibility), Quasi-public good,
Institutional design, Positive Feedback, Multi-agent simulation
JAIST知識科学研究科リサーチレポート, KS-RR-2009-003, 2009, 1-11
Abstract
株価急変時に取引を一時的に停止させることで株式市場の安定化を図る制度であ
るサーキットブレーカーには，逆に株価の下落を加速させる磁石効果があるとい
われている．しかし，サーキットブレーカーのような稀にしか適用されない制度
では，統計的データの?なさに加え，取引参加者がサーキットブレーカーの発動を
意識した上で売買行動するという前提が成り立っているのかという疑問がある．
われわれは，サーキットブレーカーの発動基準について詳細にまとめているサイ
トの閲覧行動に着目し，金融機関と非金融機関によるアクセス数とアクセス時間
の差異という視点から磁石効果の有無について考察した．結果から，サーキット
ブレーカー発動直後は非金融機関よりも金融機関からのアクセスが多く， 時間と
共に非金融機関からのアクセスが上回ることがわかった．この特徴から， 金融機
関でもサーキットブレーカーの発動基準について把握していない人たちが多数お
り，必ずしも磁石効果を生み出すような一般投資家に先んじる取引行動を起こし
ていないことが示唆された．
情報処理学会 研究報告 数理モデル化と問題解決（MPS）, 2009(19), 113-116
Abstract
恒常的可塑性は生物が動的な環境で生存するために不可欠な性質である．視覚反
転に適応する走光性をもつロボットを拡張し, 複数回の視覚反転というより動的
な状況に適応できる高度な恒常的可塑性をもつ方策を考察する．本研究では光の
時間変化(差分)を知覚できる能力と光の強さを記憶できる能力を仮定する．シミュ
レーションを行った結果, モデルの違いによらず, 視覚反転, 複数回の視覚反転
に適応できる個体はモータの推進力 - 舵分担と後退移動を獲得していることがわ
かった．このような方策が高度な恒常的可塑性に対してもつ意義を議論する．
情報処理学会 研究報告 数理モデル化と問題解決（MPS）, 2009(19), 117-120
Abstract
マルチエージェントシミュレーションを用いてCSR が普及する条件とメカニズム
を解析し,CSR が普及するには消費者が自らを評価する際の基準として，企業の
CSR への投入割合の平均のようなグローバル情報を考慮に入れる必要があること
を示す．CSR 普及のメカニズムは，消費者のCSR を重視する選好の社会学習と企
業のシェア追及の間のポジティブフィードバックである．また，CSR が普及しに
くい社会状態においても，CSR を普及させる制度設計を議論する．
情報処理学会 研究報告 数理モデル化と問題解決（MPS）, 2009(19), 101-104
Abstract
本論では，非論理的推論，多義的発話，意味変化の関係を，文化進化を通じた文
法形成の計算機実験を用いて考察し，次のことを示す．1. 非論理的推論によって
表現可能な意味空間が拡大されると共に多義的発話が生じる．2. 多義的発話は意
味変化を間接的に増加させる働きを持つ．3. 多義的発話を抑制する仕組みを加え
ても合成的な言語が形成される．この結果から，ヒトの言語学習には非論理的な
推論を適切に制御する仕組みがあると考えられることを示唆する．
日本言語学会 第137回大会予稿集, 2008年11月30日，金沢大学，pp.36-41
Abstract
言語変化の一種に内容語が文法的機能を帯びるように意味変化する文法化という
現象がある．その特徴は変化の一方向性，および，変化の型の普遍性であり，そ
れは，人間の認知構造の反映と解釈可能である．われわれは，言語学習過程に着
目した抽象的認知モデルの構築とシミュレーションによる解析を通じて，文法化，
とくに一方向的意味変化が生じるメカニズムを検討した．その結果，意味変化が
起きるためには，獲得した言語的ルールをほかの言語知識に拡大適用できる「言
語的類推能力」が重要であることが示唆された．さらに，一方向性を持った変化
を実現するには，ある形式を関連する別の意味を表すことに流用する「語用論的
拡張」，および，特定の意味の随伴性・関連性を認識しやすい「共起」という認
知バイアスが有効であることを示す．
Center for Experimental Research in Social Sciences
Working Paper Series,
85, October 31, 2008
Abstract
制度は主体の行動・認識（ミクロ）および社会的帰結（マクロ）との間（メゾ）
で二重円環的に相互作用することで生成，維持，変化，消滅する．本稿では，貨
幣制度の下で自己の行動を決定するための価値基準である“貨幣意識”と“ミク
ロ・メゾ・マクロ・ループ”の関係を明らかにするため，社会活動によって貨幣
意識に差異があるという仮説を提唱し，その検証を行った．社会集団として対照
的な存在と考えられる地域通貨関係者と金融関係者を対象に社会活動の違いと貨
幣意識の違いとの相関関係を調べた．そのために，各集団に対して 5 つのメタルー
ル（効率・自由・平等・安全・友愛）を導入した貨幣意識アンケートを実施した．
結果から，貨幣意識のメタルールとして「多様性」「公正」「利益志向」の３つ
があることがわかった．そのうち地域通貨関係者は，金融関係者に比べて「多様
性」と「公正性」を重視する傾向にあり，「利益志向」は両者で有意な差がない
ことがわかった．我々が導入したメタルールにおいても，地域通貨関係者は「平
等」「安全」「友愛」といった公共性に関するメタルールを重視する傾向にある
が，逆に金融関係者はこれらのメタルールを重視しない傾向にあった．以上のこ
とから，地域通貨関係者と金融関係者では貨幣意識に差異があり，貨幣発行の
「多様性」と政府の積極的な役割を求める「公正」に違いがあると結論付ける．
Keywords
貨幣意識，ミクロ・メゾ・マクロ・ループ，制度生態系，メタルール
Artificial Life and Robotics,
13(1), 194-198, 2008
Abstract
We consider how permselectivity as a function of the
membrane is related to the cell evolution with an abstract
proto-cell model. We construct an autopoietic proto-cell
model having primitive auto-catalytic reaction cycle
inside. In this model, several primitive membrane substances
are assumed to be produced from the reaction cycle and the
membrane is assumed to have a permeability to specific
chemical species. We show that the permselectivity of the
membrane induces the diversi?cation of the cell volume. We
discuss how the diversification of the cells is related to
the Darwinian evolution.
Keywords
Cell evolution, Proto-cell model, Membrane permselectivity,
Auto-catalytic reaction, Diversi?cation.
Artificial Life and Robotics,
13(2), 500-503, 2008
Abstract
We propose a kind of self-amendment game, Minimum Nomic, as
a model to study rule dynamics. Nomic is a game in which
changing the rule of game is a move. Minimum Nomic is a
reduced version of the original Nomic to keep the essence
but promote evolvability of the self-amendment game. We
discuss the characteristics of Minimum Nomic from the
viewpoint that how changeability of the rules and durability
of the games change with the progress of the game. By
analyzing the dynamics of purpose and goals and the
self-referential property in observations of the games
played, we claim that Minimum Nomic is an interesting tool
for study rule dynamics.
Keywords
Rule dynamics, Self-amendment, Self-reference, Nomic, Evolvability
Proceedings of
The Tenth Pacific
Rim International Conference on Artificial Intelligence
(PRICAI-08)
,
Tu Bao Ho and Zhi-Hua Zhou (Eds.)
Springer, pp. 152-163,
Abstract
This paper studies the characteristics of interaction among
genetic evolution, individual learning and social learning
using an evolutionary computation system with NK Fitness
landscape, both under static and dynamic environments. We
show conditions for effective social learning: at least 1.5
times lighter cost of social learning than that of
individual learning, beneficial teaching action, low
epistasis and dynamic environment.
Keywords
Key words: Evolutionary computation, Genetic evolution, Individual
learning, Social learning, NK ?tness landscape
In: Proceedings of Second IEEE
International Conference on Self-Adaptive and Self-Organizing
Systems
(SASO2008)
Sven Brueckner, Paul Robertson, and Umesh Bellur (Eds.), pp.20-24,
2008
Abstract
Creolization is a self-organization process of new language
community. Thus far, a simulation study of the emergence of
creoles has been reported in the mathematical framework. In
this paper we introduce a spatial structure to the
framework. We show that local creole communities are
organized, and creolization may occur when language learners
learn often from non-parental language speakers, in contrast
to the non-spatial model.
Proceedings of the 6th International Conference of
Cognitive Science, The Korean Society for Cognitive
Science (Eds.), 115-118, 2008
Abstract
In this study, we investigate how dynamic features of
communication are realized in simulated processes. We
hypothesize that ambiguity and context play important roles
for maintaining the dynamics of success and failure in
communication. We propose a game model of communication by
adding ambiguity and context to Steels’ Language-Game. As a
result of simulation experiments of two types of the game
model, it was shown that ambiguous words are disambiguated
at the situation of communication by utilizing contextual
information. We claim that the disaccord of interpretation
arose by the ambiguity is one of the integrant parts
bringing the dynamics of communication, such as superficial
communicative success, failure and re-establishment of
communication.
Recent Advances in Modelling and Simulation, Giuseppe Petrone and Giuliano Cammarata (Eds.), pp. 111-136, I-Tech Books, 2008
Abstract
In this chapter, we introduce a scientific methodology
called "evolutionary constructive approach". That is
suitable for studying dynamic features of complex
systems. At first, we identify two types of simulation
methodologies, realistic and constructive. The latter is
especially needed to clarify complex systems, since the
complex systems have distinct characteristics from objects
of conventional scientific studies. We characterize such
characteristics as "undecomposabilities". We show three
example simulation studies using the evolutionary
constructive approach. They are about 1) dynamic change of
social structures, 2) language change and displacement, and
3) dynamics of communication.
統計数理研究所
共同研究リポートNo.217，「動的システムの情報論(7) 自然言語の
ダイナミズム」，pp. 13-21, 2008
Abstract
言語の起源と進化の研究において，文法の複雑化・構造化を担う可
能性がある文法化（grammaticalization）という現象が注目されて
いる．文法化とは，動詞や名詞といった内容を表す語が，助動詞や
代名詞のような文法的機能を帯びるようになる意味変化である．こ
の現象は多くの言語で同型のものが見つかっており，人間の認知構
造が反映した言語（変化）現象と考えられる．本研究では，言語使
用者の言語学習過程に着目した文法化が生じるプロセスのモデルを
構築する．このモデルの構築と解析により，文法化を含む意味変化
が起きるためには，新たに発見した言語的ルールを拡大適用する能
力（言語的類推能力）が重要であることを示す．また，内容語から
機能語への一方向性を持つ意味変化が起きるためには，メタファー
的推論とメトニミー的推論の組み合わせが重要であることを示唆す
る．最後に，文法化を実現する言語能力に関する考察を通じて，言
語の起源と進化に関する仮説的シナリオを提示する．
Abstract
It has been pointed out that gaze alternation by infants,
which is behavior alternately gazing at a caregiver and
particular objects, is related to the process of the
development of intentional agency. Intentional agency is
de?ned as an act with a desired goal and means. In this
paper, we adopt a theoretical hypothesis that the infants
understand others' intentions based on intentional agency,
and we consider how intentional agency can be constructed
with any computational mechanism. We designed a model of
infant agent which acquires gaze alternation through
interaction with caregivers based on a reflex behavior and
an emotional behavior. First, the agent learns the visual
orientation of gazing at a target in the center of the
visual field as the reflex behavior. Based on the visual
orientation, the agent learns regularities in its
experiences to gaze at the caregiver's eye directions
followed by gazing at the particular objects. The learning
is implemented with an association module which is serially
connected with the visual orientation module. In the model,
the agent associates the caregiver's eye direction with an
object and orients the agent's eye direction to gaze at the
object. This behavior is to use the visual orientation as a
means appropriately in order to attain the agent's goal to
gaze at the associated object as a visual target. The
internal states composed of goals and means are considered
as intentional agency. Second, we add two emotional states,
ease and anxiety, to realize an emotional behavior to the
serial architecture acquiring intentional agency. The agent
looks back the caregiver when the agent comes to the anxiety
state. The emotional behavior provides opportunities of
interaction with caregivers to the infant agent. Finally,
we discuss how the intentional agency functions as a basis
of understanding others' intentions. Through the discussion,
we argue that a nested structure of intentional agency
between self and other is a primitive mechanism of
understanding others' intentions and shared intentionality.
Keywords
gaze alternation (交互凝視), intentional agency (意図的主体性),
understanding others' intentions (他者の意図理解),
shared intentionality (意図性の共有), constructive approach
(構成論的アプローチ)
Abstract
進化と学習の相互作用、特に遺伝的進化・個体学習・社会学習の 3
者の相互作用に注目したマルチエージェントシステムを開発し、NK
適応度地形上でその特性を研究する。本論では、学習のコスト、エ
ピスタシスの程度、環境の動的性を変化させ、各要素が個体集団に
与える影響や学習回数の遷移を調べた。その結果、社会学習が有効
になる条件として、10 倍以上の個体学習コストと社会学習コストの
差異、教えることがベネフィットとなる、低エピスタシス環境、動
的環境が考えられる。
Abstract
コミュニケーションが持つ動的な特徴は，以心伝心的なコミュニケー
ションにおいて顕著に確認される．本論文では，Steelsの
Language-Gameに語の多義性と文脈を追加し，以心伝心的コミュニケー
ションを再現するゲームモデルを提案する．このゲームのシミュレー
ション実験において，以心伝心的なコミュニケーションでは，文脈
情報を利用することで多義語がコミュニケーションの場で一義化さ
れるという過程が示された．本論文では，多義性が引き起こす意味
解釈の齟齬が，見かけ上のコミュニケーション，コミュニケーショ
ンの失敗や再確立といったコミュニケーションのダイナミクスを生
み出す要素の一つである可能性を主張する．
言語処理学会 第14回年次大会（NLP2008） 発表論文集CD-ROM, 989-992
Abstract
本研究では、理想的な状態を表す類義語「完全」「完璧」と、対応
する中国語・韓国語の類義語を取り上げ、認知意味論のアプローチ
で考察を行う。認知意味論では、認識主体の解釈に基づいた経験が
言葉の意味に反映していると考えられている。母語話者は類義語を
何気なく使い分けているが、その差異を記述することは難しい。こ
の類義語を多言語間で認知意味論的に分析することで、言語使用の
背後にある日中韓３ヶ国語話者の認知傾向やものごとの捉え方を考
察できる。「完全」「完璧」の比較分析よって２語の使用の背後に
量と質という視点の違いがあることが分かった。また、比較分析及
びインフォーマントテストの結果、対応する単語の意味に各言語間
で相違点があることが見出された。これらの結果は３ヶ国語話者で
理想状態の捉え方に若干の違いが存在する可能性を示唆している。
In:
Proceedings of The 13th International Symposium on
Artificial Life and Robotics (AROB08),
M. Sugisaka and H. Tanaka (Eds.), (2008)
Abstract
We consider how permselectivity as a function of the
membrane is related to the cell evolution with an abstract
proto-cell model. We construct an autopoietic proto-cell
model having primitive auto-catalytic reaction cycle inside.
In this model, several primitive membrane substances are
assumed to be produced from the reaction cycle and the
membrane is assumed to have a permeability to speci?c
chemical species. We show that the permselectivity of the
membrane induces the diversification of the cell volume. We
discuss how the diversification of the cells is related to
the Darwinian evolution.
In:
Proceedings of The 13th International Symposium on
Artificial Life and Robotics (AROB08),
M. Sugisaka and H. Tanaka (Eds.), (2008)
Abstract
We propose a kind of self-amendment game, Min imum Nomic, as
a model to study rule dynamics. Nomic is a game in which
changing the rule of game is a move. Minimum Nomic is a
reduced version of the original Nomic to keep the essence
but promote evolvability of the self-amendment game. We
discuss the characteristics of Minimum Nomic from the
viewpoint that how changeability of rules and durability of
games change. By analyzing the dynamics of purpose and goals
and the self-referential property, we claim that Minimum
Nomic is an interesting tool for study rule dynamics.
計測自動制御学会 システム・情報部門学術講演会2007, 205-210
Abstract
The symbol system of human language has a remarkable feature
called "displacement" which denotes the humas'
characteristics to be able to represent an event that is
distant from "now, here, and I". Understanding this
feature is indespensable to comprehend the emergence and
evolution of symbol systems. This paper considers
"displacement" through constructive modelling of
grammaticalization, a kind of meaning change process from
content words to functional words. It is shown that the
ability of linguistic analogy, extensive application of
acquired linguistic rules, may play a critical role for
displacement, the metaphoric and metonimic inferences work
to induce unidirectional change, which is a characteristic
of grammaticalization.
Abstract
市場流動性を高めるとされるランダムに取引をするエージェントが，
株価の急変動で崩れた需給バランスをどのように是正し，市場流動
性を高めるかを人工市場シミュレーションにて明らかにする．実験
結果から，ランダムエージェントの割合が大きいものほど約定率が
高くなることがわかった．このことから，ランダムエージェントが
市場流動性に寄与していることが示唆される．また売買行動分析か
ら，急激な価格変動が発生して市場の需給が一方に傾いたとしても，
市場状況を加味しないランダムエージェントによる注文が需給のバ
ランスをある程度是正するとことが判明した．こうしたランダムエー
ジェントの特徴は，現実のマーケットメーカーの制度設計を考える
上で重要な知見を与えるものである．
Keywords
市場流動性，ランダムエージェント，人工市場，制度設計，マーケットメーカー
Abstract
欺き行動は社会で他個体より有利に立つことで利益を得る方策とし
て，ヒトの社会的知能の進化を説明するマキャベリ的知性仮説で着
目されている．また，欺きの能力は他者の内部状態を推定する能力
と関連していることが実験などで示されている．本論文では，欺き
能力と内部状態推定能力の進化との関係を研究する基礎として，欺
き行動を形式的にモデル化することを試みる．ここでは，欺き行動
を，ランナーがキャッチャーから逃れようとする「逃走・捕獲ゲー
ム」における，ランナーによるフェイント行動としてモデル化する．
ここでは，ランナーはキャッチャーの次の行動が読めるとし，その
情報を利用してフェイントの仕方を決める．また，キャッチャーは
ランナーの過去の行動から次の行動を予測する．このモデルの振る
まいをエージェントシミュレーションで調べたところ，キャッチャー
が予測能力を持つ場合にのみ，フェイントが十分有効であることが
わかった．この結果から，相手に誤った情報をわざと与えて誤導す
ることで利益を得るという，欺き行動の本質的部分がモデル化でき
たと考えられる．
Keywords
欺き， 逃走・捕獲ゲーム， エージェントモデル，マキャベリ的知
性仮説
日本認知科学会 第24回大会発表論文集, 80-81
Abstract
乳幼児が内部に意図性を形成する過程で，情動が不安定化すること
によって引き起こされる参照視がどのような役割を担い得るのかを，
計算モデルの構築とそのコンピュータシミュレーションへの実装に
より検討する。本稿では，乳幼児が意図性を形成する過程で，情動
的な行動が，社会的コミュニケーション能力の形成を助けているの
ではないかということを論じる。
Keywords
乳幼児の発達過程，参照視，社会的参照，交互凝視，
意図性，計算機シミュレーション
認知言語学会論文集 第７集, 33-43, 2007
Abstract
Grammaticalization is a notable topic in not only cognitive linguistics
but also evolutionary linguistics. In this paper, we investigate
cognitive abilities required for grammaticalization with the
constructive approach and discuss a process and a mechanism of
grammaticalization observed in computer simulations from the viewpoint
of the origin and the evolution of language. The constructive approach
is a method to understand certain phenomena through constructing and
operating a model of the phenomena. Concretely, we constructed an agent
model that showed a process of grammaticalization, in particular, a
meaning change from a verb to a tense auxiliary verb.
We adopted a hypothesis about grammaticalization, proposed by Hopper and
Traugott, that reanalysis and analogy are indispensable for
grammaticalization. Considering reanalysis and analogy from the
cognitive viewpoint, we define the following three cognitive abilities
as the basic characteristics of agent:
- Cognitive analogy: an ability to find similarity between
situations and between forms,
- Reanalysis: an ability to punctuate sentences based on
the similarities,
- Linguistic analogy: an ability to apply acquired
grammatical rules extensively to other linguistic knowledge.
The framework of computational model we used is called ``iterated
learning model'', proposed by Hurford and Kirby, in which two types of
agents, adult and child, conversationally interact and linguistic
knowledge is transfered over generations. In this model, the agent has
three learning mechanisms, chunk, merge and replace. We analyzed the
relationships between the ability defined above and these three
mechanisms, and found that reanalysis and linguistic analogy are borne
mainly by chunk and replace, respectively. In addition to the original
model, we introduced functional meanings and agents' tendencies to
understand meanings.
We showed the followings by analyzing the process of meaning changes
including grammaticalization phenomena with computer simulations of the
model. The above three cognitive abilities are effective to occur
meaning changes, but not enough for unidirectional change as found in
actual grammaticalization processes. The meaning change is promoted by
a tendency to understand meanings, i.e.,
- the agent recognizes overlaps of the meaning regions between the
source of the change and other related meanings.
Further, in order for the meaning change to have unidirectionality,
another tendency is important, i.e.,
- the agent recognizes a relevancy of the meanings between the source
and the target of the change.
In the perspective of the origin and the evolution of language, we claim
that, among other abilities, the linguistic analogy is significant as a
cognitive ability for human language, since the linguistic analogy can
contribute to acquire ``displacement'' that is a characteristic uniquely
found in human languages.
Proceedings of ESSA'07 the 4th Conference of the European Social Simulation Association, pp.315-317
Abstract
Securing liquidity in a market is indispensable for the
market mechanism to work efficiently. While there is a
discussion that random agents have the effect to ensure the
liquidity, the mechanism of this effect has not been
clarified. We show that, using an artificial market system
U-Mart, the random agents can improve the liquidity of a
market by maintaining the balance of supply and demand. We
also suggest that it is necessary for the random agents to
manage their position in order to work as a market maker.
Keywords
Market Liquidity, Random Agent, Artificial Market, Institutional
Design, Market Maker
In:
Proceedings of The Twelvth International Symposium on
Artificial Life and Robotics (AROB07),
M. Sugisaka and H. Tanaka (Eds.), (2007)
(held at Beppu, Japan, January 25-27, 2007)
Abstract
It has been pointed out that gaze alternation by infants,
which is a basis for social communication, is related to the
process of the development of intentional agency. Intentional
agency is defined as an act with a desired goal and
a means. It has been pointed out that infants understand
others’ intentions based on intentional agency. In our recent
work, we constructed a computational model which
acquired gaze alternation behavior with elementary intentional
agency. In this paper, through the analysis of the
behavior and the internal states of the constructed model,
we confirm that two mechanisms, discrimination and accumulation,
play an important role in developing elementary
intentional agency. The former discriminates between
a caregiver and objects that are producing stimuli. The
latter associates the caregiver with the objects by accumulating
relationships between sensory states. We also
argue that a nested structure, in which gaze alternation
composed of a goal and a means is utilized as a means to
achieve another goal, is an important feature in the development
of intentional agency.
Keywords
Gaze Alternation, Communicative Eye Gaze,
Understanding Others' Intention,
IntentionalAgency, Constructive Approach.
In:
Proceedings of The Twelvth International Symposium on
Artificial Life and Robotics (AROB07),
M. Sugisaka and H. Tanaka (Eds.), (2007)
(held at Beppu, Japan, January 25-27, 2007)
Abstract
In various stock markets, there is a system called "circuit breakers"
that interrupts dealing of stocks for a certain period when stock price
changes greatly. In this paper, we consider the influence of the
circuit breakers on a stock market using an agent-based artificial
market simulator called "U-Mart", by controlling the period of
interruption and the criterion to invoke the circuit breakers. From
experiments, we found the following: circuit breakers play an important
role in the control of price fluctuations and the stabilization of the
settlement system, while the circuit breakers reduce the stock trading
volume. We also suggest that the period of interruption is important
parameter in an institutional design of the circuit breakers, since the
stock trading volume and the volatility are sensitive to the period.
In:
Emergence of Communication and Language ,
Caroline Lyon, Angelo Cangelosi and Cristopher
L. Nehaniv (Eds.), pp.223-236 (2007)
Abstract
We aim at constructing a model for an agent showing symbolic
activities. Based on a discussion of the concept of a symbolic system
we propose correspondences between symbolic activities and the
behaviour of dynamical systems in which attractors develop in response
to inputs. We examine correspondences such as, symbols to attractors,
symbol manipulations to transitions among attractors, and regularities
in symbol manipulation to the order of the transitions. By using a
dynamical system showing transitions among attractors, we may be able
to represent symbolic activities, in part. We try to construct such a
system with a kind of chaotic neural networks, derived from the
Hopfield model. We confirm that this system has multi-attractors and
shows transitions among attractors in some parameter regions. Further,
we investigate the response of the system to certain external inputs.
Examining these simulation results suggests that this system learnt in
a certain environment can be used as a model of a symbolic agent for
the study of the emergence of linguistic communication.
In:
Emergence of Communication and Language ,
Caroline Lyon, Angelo Cangelosi and Cristopher
L. Nehaniv (Eds.), pp.237-252 (2007)
Abstract
Language change is considered as a transition of a user population among
languages. Language dynamics equations represent such a transition of
population. Our purpose in this paper is to develop a new formalism of
language dynamics based on a realistic situation of multiple language
contact. We assume a situation where
memoryless learners are exposed to a number of languages. Our
experiments show that contact with other language speakers during the
acquisition of a first language reduces learning accuracy and prevents
the emergence of a dominant language. We suppose there is a special communicative language which has a
higher similarity to some languages than others; when learners are
frequently exposed to a variety of languages, these similar languages
attract a relatively higher proportion of the population. We discuss
the simulation results from the viewpoint of the language bioprogram
hypothesis.
Evolutionary and Institutional Economics Review
,
3(1), 27-46, (2006)
Abstract
Evolutionary linguistics is going to clarify the evolutionary process of
language and evolutionary economics is clarifying the dynamics of
economic phenomena. In this paper, we consider the relevance between
these two fields. In evolutionary linguistics, the origin of language is
thought of as the biological evolutionary process of language capacity
of humans and the evolution of language the processes of
complexification and structuralization of language structure. The
evolutionary process of language forms a double loop dynamics among
biological evolution, learning and cultural evolution. This double loop
among these three adaptive dynamics ranging in large spatio-temporal
scales characterizes language evolution as complex systems. The double
loop dynamics can also be found in the origin and the evolution of
economics. Thus, evolutionary linguistics and evolutionary economics can
collaborate in understanding of complex evolutionary phenomena. The
constructive approach is a useful tool to study such complex
systems. Communicating the development of methodology and findings to
understand such complex systems as language evolution and economic
evolution may lead these two fields to fruitful synergism.
Keywords
Evolutionary linguistics, Evolution and origin of language, Double loop
dynamics, Rule dynamics, Constructive approach
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2006(10), 195-202, (2006)
Abstract
乳幼児が見せる原型的な社会的コミュニケーション行動の1 つとし
て注目される交互凝視行動には，意図的主体性の形成過程があると
指摘されている。この意図的主体性は目的と手段から構成される行
動と定義され，これが他者の意図を理解する足掛かりになると言わ
れている。本論では，この意図的主体性の萌芽的な形成過程を調査
するため，試行錯誤学習によって視覚定位という手段を獲得し，経
験の蓄積によって目的を形成することで交互凝視行動を表出する計
算モデルを構成した。この計算モデルにおいて，段階的な能力や機
能の獲得過程を発達的に構成することで，受け取る感覚刺激のカテ
ゴリ識別と，その刺激経験の蓄積規則が，意図的主体性の形成に重
要な役割を担う可能性があることを示す。また，乳幼児の見せる意
図的主体性との接点を考察することにより，目的と手段によって構
成される交互凝視行動が，また別の目的の手段として選択／活用さ
れるような入れ子の構造を持つことが，意図的主体性の形成におけ
る重要な性質のひとつであることを示唆する。
Keywords
乳幼児の発達過程，交互凝視，意図的主体性，目的と手段の分離，
構成論的アプローチ
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2006(10), 145-154, (2006)
Abstract
本研究では群行動の設計と評価を行う．群れは，集団の性能が評価
対象であるが設計対象は個体の行動である．このギャップが当該研
究を困難なものにしている．そこで本研究では、動物行動学におけ
る「環世界」という概念を用いて群れの振る舞いを設計した．この
概念を用いる利点は、個体の目的から群れの振舞いに必要な全ての
行動を導出可能な点である。この概念を用いることにより群れの振
る舞いを個体の行動アルゴリズムに組み込むことができる．そして、
この群行動アルゴリズムを実装する．計算機実験では，各種パラメー
ターには現実的な値を入力する．解析では，群れがタスクを解決す
るときの「タスク解決時間」に着目し，群れの並列性・作業分担に
ついて解析を行った．解析の結果，本研究で構築した群行動アルゴ
リズムにおいて，群れの特徴である並列性が有効に働いていること
がわかった．また，協調しなければ運搬できないような物体を運搬
する際，群れは自己の内部状態を効果的に遷移させて作業分担して
いることが判明した．
Keywords
環世界, 群行動アルゴリズム, 仮想フェロモントレイル, 自己組立, 作業分担
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2006(10), 29-36, (2006)
Abstract
本研究では，株式市場の緊急取引停止措置であるサーキットブレー
カーを導入した人工市場シミュレーションによって，サーキットブ
レーカーの発動期間を変えたときの市場の振る舞い，および，サー
キットブレーカーの有効性について考察を行った．実験結果から，
サーキットブレーカーが約定数量の減少との引き換えに価格変動の
抑制，市場決済システムの安定化のために重要な役割を果たすこと
が示唆され，さらにサーキットブレーカーの発動期間の変化がサー
キットブレーカーの設計において重要なパラメータであることも示
された．本研究では現実には観察しにくい事象に対して制度的考察
を行っており，有効なサーキットブレーカーの発動期間を定量的に
示す可能性を大いに含んでいる．
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2006(10), 217-224, (2006)
Abstract
カオス的遍歴(Chaotic Itinerancy, CI) の動的な振る舞いは，神経
系における情報処理との関連や，大自由度力学系に特有の現象とい
う観点などから注目されている．CI が生じる力学系としては，大域
結合写像系(Globally Coupled Map, GCM) が広く知られている．ま
た，CI の発生と，摂動に対する強い不安定性を持つミルナーアトラ
クタの存在との関係が指摘されている．そこで本研究では，GCM に
おけるミルナーアトラクタの性質をより詳しく理解するため，その
ベイスン構造を調べた．その結果，ミルナーアトラクタのベイスン
はリドルドベイスンを形成していることが示唆された．このことは、
共存するアトラクタのベイスン同士が微小なスケールに至るまで混
合していることを意味し，ミルナーアトラクタの持つ不安定性を反
映した構造だと言える．
Proceedings of the Fifth International Conference on Development and
Learning ICDL 2006 ,
(CD-ROM)
,
(held at Indiana University, Bloomington, IN
May 31 - June 3rd, 2006)
Abstract
In the communicative eye gaze such as joint visual attention and gaze
alternation, the importance of intentional agency is pointed
out. Namely, infants understand others' intentions and also
intentionally gaze at others' gazing points. In this paper, for
investigating the development of the intentional agency that the infant
separates desired goals from the actions, we constructed a computational
model that is able to acquire the ability of the gaze alternation to an
object outside the visual field based on the intentional agency. In the
construction and operation of the model, we confirmed that two functions
play an important role in the visual joint attention. The one function
performs discrimination between the caregiver and the object, the other
does association evoked by visual stimuli by storing the relations
between sensory states during looking at the caregiver's face and that
during looking at the objects as a frequency distribution. We discussed
the process of the joint visual attention from the viewpoint of the
intentional agency. Based on the discussion, we claims that the
association can be thought of as a goal that is possessed as the
agent's internal state and the movement of the gaze point can be
thought of as an action to achieve the goal.
Keywords
Gaze Alternation, Joint Visual Attention, Communicative
Eye Gaze, Intentional Agency, Constructive Approach.
ニューロコンピューティング(NC)研究会
「企画セッション：進化・発達の現象とモデル」
Abstract
コミュニケーションの原型的能力として注目される乳幼児の行動の 1つに交互
凝視行動がある．この行動の捉え方として，乳幼児が他者の意図を理解し，対
象を主体的に見ることが重要であるという指摘がある．そこで本論では，交互
凝視行動の主体的側面の理解を目的として，構成論的アプローチを用いた数理
モデルの構成を試みる．このモデルではまず，エージェントとして表現する乳
幼児が，親の顔やオブジェクトを注視する行動を強化学習によって獲得する．
そして，その過程で親やオブジェクトを見ているときに得る感覚情報(視界内
での親やオブジェクトの配置情報，視線方向と形状の特徴情報，筋肉状態の自
己受容感覚)の遷移過程を頻度分布に蓄積し，この頻度分布を用いて親を見て
いるときに得る感覚情報からオブジェクトを見たときに得た感覚情報を連想す
る．このモデルをコンピュータシミュレーションに実装した結果，親の顔を見
ているときの感覚情報からオブジェクトの感覚情報を連想することで，親の視
線が指し示す方向にある視界外のオブジェクトを注視できるようになることが
確認された．また，反対にオブジェクトを見ているときの感覚情報から親の感
覚情報を連想することで，自己受容感覚が指し示す方向にいる視界外の親の顔
を注視できるようになることが確認された．
キーワード
交互凝視，共同注視，他者の意図理解，乳幼児の能動的主体性，構成論的アプローチ
MPSシンポジウム2005「計算科学シンポジウム」論文集, pp.379-384
Abstract
株式市場には相場が大きく変動した時に取引所が市場での取引を一時中断させ
るサーキットブレーカーという制度がある．この措置の目的は，株価の急激な
変動を直接抑制し，市場の混乱を回避することにある．しかしサーキットブレー
カーの発動要因となる株価の大規模変動が極端に少ないことから，これまでサー
キットブレーカーに対する実証的な研究は十分に行われておらず，実効性に関
しては議論の分かれるところである．そこで本研究は，エージェントベースシ
ミュレーションの枠組みである人工先物市場U-Martを用いて，サーキットブレー
カーの発動期間と発動基準を操作し，サーキットブレーカーが市場全体に与え
る影響を考察する．実験結果から，価格の下落幅が小さい場合には，サーキッ
トブレーカーの発動期間を変えても価格変動の抑制に効果がないことが示唆さ
れ，逆に価格の下落幅が大きい場合には，サーキットブレーカーの発動期間に
よっては価格形成がスムーズに働かない可能性があることを示した．
キーワード
人工市場,サーキットブレーカー,U-Mart,マルチエージェントシミュレーシ
MPSシンポジウム2005「計算科学シンポジウム」論文集, pp.179-186
Abstract
コミュニケーションの原型的能力として注目される乳幼児の行動の 1つに共
同注視行動がある。この行動の捉え方として，乳幼児が他者の意図を理解し，
対象を主体的に見ることが重要であるという指摘がある。そこで本論では，
共同注視行動の主体的側面の理解を目的として，構成論的アプローチを用い
た計算論的なモデルの構成を試みる。このモデルは，親の顔やオブジェクト
を注視する行動を強化学習によって獲得する。その獲得過程で，体験する親
の視線方向とオブジェクトの配置方向を頻度分布に蓄積し，この頻度分布を
用いて親の視線方向からオブジェクトの配置方向を連想する。このモデルを
コンピュータシミュレーションに実装した結果，親の視線方向からオブジェ
クトを連想することで，親の視線が指し示す方向にある視界外のオブジェク
トを注視できるようになることが確認された。
キーワード
共同注視，他者の意図理解，乳幼児の主体性，構成論的アプローチ
情報処理学会：数理モデル化と応用(
Transactions on Mathematical Modeling and its Applications (TOM)
),
46, 81-92, 2005
Abstract
社会を動的に変化するものととらえる動的社会観の立場から，マルチ
エージェント・システムを用いて社会構造のダイナミクスが生じる
条件を考察する．
はじめに，内部ダイナミクスを持つエージェントを用いたマルチ
エージェント・シミュレーションの実験結果を紹介する．
この実験では，マクロレベルにおいて様々な種類の秩序的パターンを
非周期的変化を経て繰り返し遍歴するという社会構造のダイナミクスが
示されることがある．
本論では，社会構造のダイナミクスの発生要因の1つと考えられる
ミクロマクロ・ループの効果について調べた実験の結果を報告する．
これらの結果の分析より，内生的な社会構造の
ダイナミクスの形成および維持にはエージェントの内部
ダイナミクスとミクロマクロ・ループが必要であることを論じる．
Advances in Artificial Life
(Proceedings of 8th European Conference on Artificial Life)
(ECAL2005),
,
M.S. Capcarrere, et al. (Eds.), Springer, Berlin,
pp.675-684, (2005)
(held at University of Kent, Canterbury, Kent (UK), 5-9,
September,
2005)
Abstract
In human societies, facts are constructed through social
consensus. Here, the formation of social institutions in such a
society is studied using a multi-agent-based simulation.
Institutions are formed through communications among members, and
the effects of errors in communication on the formation of
institutions are investigated. Our results show that the
institution is established when information suppliers frequently
make errors in their information interpretation. We propose here
that there is a phase transition in the error rate of the
information suppliers in the formation of institutions.
In:
Proceedings of The Tenth International Symposium on
Artificial Life and Robotics (AROB05),
M. Sugisaka and H. Tanaka (Eds.), (2005)
(held at Beppu, Japan, February 4-6, 2005)
Abstract
In order to understand dynamic nature of our world and ourselves,
we propose three models of individuals that can be used in
multi-agent systems for dynamic linguistic or social simulations.
They are developing word-web, simple recurrent network with
self-influential connection (SRN-SIC), and coupled NZ maps. We
briefly summarize the architecture of these models and show some
dynamic behavior in multi-agent simulations composed by these
models.
Keywords
Developing Word-Web, Simple Recurrent Network with
Self-Influential Connection (SRN-SIC), Coupled NZ Maps,
Constructive Approach, Dynamic View of Language, Dynamic View of
Society
In:
Proceedings of The Tenth International Symposium on
Artificial Life and Robotics (AROB05),
M. Sugisaka and H. Tanaka (Eds.), (2005)
(held at Beppu, Japan, February 4-6, 2005)
Abstract
In this paper, we address a new concept to view a social system as
consisting of diversified institutions interacting with each other and
dynamically changing through actions of individuals in a society. A
mathematical framework for the concept is formalized. The framework,
called Rule Ecology Dynamics (RED), is an extension of multi game
dynamics, in which players play plural games simultaneously, by
introducing time dependent weights for the plural games and a
``meta-rule''. The meta-rule is a rule to determine the change of
rules. We show simulation results of two kind of meta-rules, average
payoff type and inverse standard deviation type. We discuss that this
framework is a realization of rule dynamics and it has certain
relevance to describe real phenomena of institutional dynamics.
Keywords
Rule Ecology Dynamics, Formation and Change of
Institutions, Replicator Equation, Multi-Game Dynamics,
Evolutionary Game Theory, Rule Dynamics
In:
Proceedings of The Tenth International Symposium on
Artificial Life and Robotics (AROB05),
M. Sugisaka and H. Tanaka (Eds.), (2005)
(held at Beppu, Japan, February 4-6, 2005)
Abstract
A dynamical systems game model of market, called Outwit Game, is
proposed. In the game, individual behavior in a market is
abstracted as orienting majority and orienting minority. We
suppose that individuals' profit changes with a course of these
two actions. Through computer simulations, we show that an index
representing both micro and macro dynamics show a power law
distribution.
Keywords
Outwit Game, Dynamical Systems Game, Market Dynamics,
Micro-Macro Loop, Power Law Distribution
New Frontiers in Artificial Intelligence: Joint Proceeding of the 17th and 18th Annual Conferences of the Japanese Society for Artificial Intelligence , Akito Sakurai, Koiti Hashida, and Katsumi Nitta (Eds.) pp.237--251, Springer (2007)
Abstract
In this paper, we discuss a viewpoint to regard individuals in a
society as cognitive agents having internal dynamics, in order to
study the dynamical nature of social structures. Internal dynamics is
the autonomous changes of internal states upon which an agent's
behavior depends. We first discuss the benefit of introducing
internal dynamics into the 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 structure is
shown to be induced by the internal dynamics of the agents.
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 403-406, (2004)
Abstract
本研究は，記号生成の数理モデルを構築することを目的としている．
本論ではまず，力学系の枠組みに基づいて記号の性質を議論する．
力学系の観点からは，アトラクタを記号として，外部入力に応じた
アトラクタ間の規則的な遷移を記号の操作と捉えられる．よってア
トラクタ間の遷移が見られる力学系を用いれば，記号の働きを表現
できる可能性がある．我々は，カオス的な振る舞いを見せる写像で
あるNZ マップによって構成されたカオスニューラルネットを用いて
そのような系の構築を試みる．本論では，このシステムがマルチア
トラクタ系であり，アトラクタ間の自律的な遷移が見られることを
確認した．さらに系に外部入力を加えることにより，外部入力に応
じたアトラクタ間の遷移が生じ得ることを数値計算によって示した．
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 367-372, (2004)
Abstract
人間の認知過程の解明に迫るため、ヴァレラは一人称的視点による
生き生きとした経験の記述を重視し、これと従来の科学的手法の統
合を提案した。本研究では、左右反転視実験に内観法を適用し、身
体的経験を通じた諸知覚の統合過程を調べた。この統合過程で中心
的に機能するのが、自己身体の姿勢・位置・動きについての内的表
象であるボディイメージである。実験では、このボディイメージの
獲得・変容過程を、視覚と聴覚または自己受容感覚の統合過程や、
知覚−運動協応の変化を見る課題から探るとともに、一人称的手法
の有効性について検討した。実験結果から、新たなボディイメージ
の形成過程において諸知覚が視覚的に統合されることが示された。
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 263-270, (2004)
Abstract
社会を動的に変化するものと捉える動的社会観の立場から、マルチ
エージェント・システムを用いて社会構造のダイナミクスが生じる
条件を考察する。はじめに、内部ダイナミクスを持つエージェント
を用いたマルチエージェント・シミュレーションの実験結果を紹介
する。この実験では、マクロレベルにおいて様々な秩序状態を無秩
序状態を経て遍歴するという社会構造ダイナミクスが示されること
がある。本論では、社会構造ダイナミクスの発生要因の一つと考え
られるミクロマクロ・ループの効果について調べた実験の結果を報
告する。これらの結果の分析より、内生的な社会構造ダイナミクス
の形成および維持にはエージェントの内部ダイナミクスとミクロマ
クロ・ループが必要であることを論じる。
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 247-254, (2004)
Abstract
本論では、市場における個人の振舞いに注目し、市場のダイナミク
スを定性的に捉えるモデルとしてOutwit Game を提案する。市場の
モデル化に際し、完全競争市場と合理的経済主体を仮定せず、市場
が動的に変化していることを重視する。よってゲームの枠組みとし
て力学系ゲームを採用する。ここで提案するOutwit Game では、市
場における個人の行動を多数派行動と少数派行動に抽象化し、その
履歴により個人の損益が変化する。Outwit Game のシミュレーショ
ンより、個人の行動と市場状態のダイナミクスに対応する指標がベ
キ分布を示すことがわかった。
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 121-128, (2004)
Abstract
言語進化の分野において，クレオールといわれる言語現象が取り扱
われている．本研究の目的は，クレオールが発現するための言語能
力と社会，言語環境についての条件を数理科学的な側面から一般化
された形で定式化することである．我々はこれまでにM.Nowak らが
提唱する言語動力学を修正し，他言語話者との接触に関するパラメー
タを追加した．これにより，言語動力学上でクレオールの発現を観
察し，そのために必要とされる言語間の類似性に関する条件を導き
出している．本稿では言語学習者に焦点を当て，クレオール化の学
習環境への依存性を検証する．実験により言語獲得期における言語
入力が少ないとクレオールが発現しやすいという結果が得られた．
情報処理学会シンポジウムシリーズ
（数理モデル化と問題解決シンポジウム論文集），
2004(12), 107-112, (2004)
Abstract
人間言語の語彙は、なんらかの対象を表象する単語のリストのよう
な単純な構造を持つのではなく、単語間に関係があり全体として構
造化されたカテゴリー構造を持つ。本稿では、このようなカテゴリー
構造を発展させる能力がなにかという言語進化に関する問いを、計
算モデルを用いて考察する。ここではその能力として、記号と対象
を結びつける能力と、記号間の関係を推論する能力を仮定する。こ
の能力をモデル化したエージェントを動的単語網モデルを用いて構
築する。マルチエージェントの会話のシミュレーションにより、様々
な語彙のダイナミクスと語彙構造の組織化を通じて、プロトタイプ
的なカテゴリー構造が発達しえることを示す。
日本認知科学会第21回大会 (2004年7月31日-8月1日)
Abstract
First-Person Approach (FPA) was suggested by Varela as a method to
describe subj ective aspects of lived experiences. He proposed to
combine FPA with Third-Perso n Approach which is an objective and a
usual scientific method of psychological studies. We conduct an
experiment of left-right reversed vision to examine the e ffectiveness
of FPA for considering the subjective aspect of embodied cognition. A
subject wearing the left-right reversed glasses is assigned several
tasks abo ut vision and body movements and is requested
introspections. By discussing the results of objective observation of
the tasks and the introspective reports, we conclude that reality of the
subjective aspects of tactual-kinesthesis perform a n important function
in developing new relationship between vision and body whic h process
correlate closely with human cognition. It is suggested that FPA works
well to some extent for describing this integration process.
Abstract
The evolution of a primitive cell is studied with computer simulations,
by focus ing on the permeability of cell membrane. We make a proto-cell
model which includes internal chemical dynamics and interac tion with
external environment. We assume two types of membrane transportation:
simple diffusion and facilitated diffusion. Under the former, the
transportation depends only on the difference of concentration between
inside and outside of a cell. Under the later, in addition, the
transportation depends on activity of internal chemicals. In the
computer simulations, we observed following results. The facilitated
diffusion cell is more advantageous to survival than a simple diffusion
cell when permiability is low. The evolution of permiability from low
range promotes diversity of permiability per each chemical and per
diversity of cells.
Abstract
We examine the effectiveness of First-person approach (FPA) for
investiga ting the subjective aspect of embodied cognition. FPA was
suggested by Varela as method to describe subjective aspect of lived
experiences. He proposed to combine FPA with Third-person approach
which is objective and usual scienti fic method of psychological
studies. We conduct an experiment of left-right reversed vision. In
this paper, three tests are reported: the posture repre sentation at
head-body coordination, the orientation of sound source and the sh ape
drawing with touch. From the results of these tests, it is suggested
that F PA works well to some extent for describing how subjective body
image and experi ences are integrated.
Abstract
In this paper, we discuss a viewpoint to regard individuals in a society
as cogn itive agents having internal dynamics, in order to study the
dynamical nature of social structures. Internal dynamics is the
autonomous changes of internal states upon which an agentfs behavior
depends. We first disc uss what can be gained by taking internal
dynamics into consideration when thinking about humans and the dynamics
of society. Then we propose an extended simple recurrent network
(extended-SRN) as a model of an agent havin g internal dynamics. We
report the results of our simulation, in which the agents play the
minority game as a s ocial interaction. In the simulation, we observed
the dynamics of game as a macro structure itinerated among various d
ynamical states such as fixed points and periodic motions via aperiodic
motions. This itinerant change of the macro structure is shown to have
been induced by internal dynamics of the agents.
Abstract
To study phenomena in which functions governing change of states should
be regar ded as variable, we investigate a shift dynamics called a
functional shift which can describe dynamic change of functions. The
framework of functional shifts has an advantage that is able to analyze
both dynamical and computational characteristics, because this framework
is an exten sion of symbolic dynamics. To analyze the property of
functional shifts, we measure them in terms of topolo gical entropy and
formal languages. Through this study, we argue that considering
functional shifts from the viewpoi nts of both dynamics and computation
give us opposite results about the complexi ty of systems.
第18回 日本人工知能学会 AIレクチャ(2004年6月2日)
Abstract
Researches on the evolution of language are reviewed. At first,
we address the research questions of the origin and the evolution
of language. The development of the researches is summarized by
focusing on the deference between human and non human
communication systems. After mentioning a constructive approach,
we exemplify some constructive studies on the evolution of
language. Finally, we discuss the characteristics of evolutionary
linguistics.
第16回 自立分散システム・シンポジウム(2004年1月26,27日), pp. 125-130
Abstract
Dynamics of language including the formation of symbols are
studied with a constructive modelling. We discuss the dynamic view
of language, which is the basic concept of this study, and the
constructive approach. We propose a model, called developing
word-web model, that is an implementation of the basic concept by
a computational model. Simulation results of the developing
word-web model show that the model realizes the coexistence of
stability and adaptability and that of commonality with
individuality. These features are summarization of the
necessities for an established language system that is an
essentially dynamic phenomenon.
Keywords
Dynamic View of Language, Constructive Approach, Developing Word-web Model
Abstract
Creole is one of the main topics in the fields concerning the language
change and evolution, such as sociolinguistics, the developmental
psychology of language and so on. Our purpose in this paper is to
develop an evolutionary theory of language to study the emergence of
creole. We discuss how the emergence of creole is dealt with regard to
population dynamics. We modify the language dynamics equations by
Komarova et al., so as to include the generation parameter et 't
'. From the viewpoint of the population dynamics, we give the
definition of creole as a language, which is predefined by the universal
grammar together with pre-existing languages. We show experimental
results, in which we could observe the emergence of creole. Furthermore,
we analyze the condition of creolization in terms of similarity among
languages.
Keywords
Creole, Population Dynamics, Language Dynamics Equations
Language evolution has been dealt with in the framework of population dynamics for recent years in order to establish a mathematical theory of language. The representative models include the language dynamics proposed by Komarova~et~al.[1], and the results in population dynamics are forthcoming. We applied the language dynamics to the research of the emergence of creole by modifying it to be more adaptable to the environment of the actual creole. Our prime revision is that the transition rate in languages is sensitive to the distribution of population of each generation. In addition, we introduce the exposure probability that is the measure of influence from other languages. In the series of our study from the perspective of the population dynamics, we have already shown the process of the emergence of creole[2] and the condition of the similarities among languages for creole to emerge and to be dominant[3].
In the theme of the present research, we aim to overhaul the term of fitness corresponding to the communicability of language. In our model so far, language speakers bear offspring in proportion to the fitness. However, creole emerges not because creole speakers have more offspring than other pre-existing languages but because the influence of infants' learning biases during the repeated cultural transmission of the creole forces it to take on certain structures. Therefore, we need to find out how the fitness works in the process of the emergence of creole. Because the behavior of the dynamics without fitness can easily be compared to that of with fitness, we start to delete the term of the fitness from the dynamics equations.
A major difference is that the range of the exposure probabiity in which the creole emerges is remarkably larger in the model without fitness than one with fitness\footnote{ Although we have found other differences, there is no space to express here. They will be discussed at the conference.}. For the equation without fitness term, the creole exists in the whole range. In contrast, for one with fitness, there is a critical value in the probability only above which creole emerges. While the value depends on the similarities among languages, it is larger than 0.7 for any similarity settings. Below the critical probability, the creole once grows but lose its population eventually. We coclude from these numerical analysis that the fitness works as a suppressor for the creole to emerge when children do not talk with people other than their parents so often.
Abstract
We introduce a new type of shift dynamics as an extended model of
symbolic dynam ics, and investigate the characteristics of shift spaces
from the viewpoints of both dynamics and computation. This shift
dynamics is called a functional shift that is defined by a set of
bi-infinite sequences of some functions on a set of symbols. To analyze
the complexity of functional shifts, we measure them in terms of topo
logical entropy, and locate their languages in the Chomsky hierarchy.
%Through this study, we argue that complexity of dynamics does not
correspond to that of computation. Through this study, we argue that
considering functional shifts from the viewpoi nts of both dynamics and
computation give us opposite results about the complexi ty of systems.
We also describe a new class of shift spaces whose languages are not
recursively enumerable.
Keywords
Rule Dynamics, Functional Shifts, Topological Entropy,
Self-referential Shifts
Abstract
Creole is one of the main topics in various fields concerning the
language origin and the language change, such as sociolinguistics, the
developmental psychology of language, paleoanthropology and so on. Our
purpose in this paper is to develop an evolutionary theory of language
to study the emergence of creole. We discuss how the emergence of
creole is dealt with in the perspective of population dynamics. The
proposal of evolutionary equations is a modification of the language
dynamics equations by Komarova et al. We show experimental results, in
which we could observe the emergence of creole. Furthermore, we analyze
the condition of creolization in terms of similarity among languages.
We conclude that a creole becomes dominant when pre-existing languages
are not similar to each other and rather similar to the newly appeared
language (would-be-creole); however the new language must not be too
similar, in which case pre-existing languages remain and coexist.
Keywords
Population Dynamics, Creole, Similarity among Languages,
Language Dynamics Equations
Abstract
Children will develop their parental languages correctly, since language
learners come to obtain the one which they contact most in the
community. At the same time, children would be affected by other
languages, the influence of which is proportional to the population of
those languages. In this paper, we revise the foregoing evolutionary
theory of language, that is differential equations of the population
dynamics. We propose that the transition rate in languages is sensitive
to the distribution of population of each generation. In addition, we
introduce the exposure probability that is the measure of influence from
other languages. We show experimental results, in which we could
observe the emergence of creole. Furthermore, we analysed which language
would be dominant, dependent on the initial distribution of population,
together with the exposure probability.
Keywords
language acquisition, population dynamics, creole
Abstract
This paper proposes an evolutionary-game-theory model, called
meta-evolutionary game dynamics, for studying the dynamics of
rules and individual behaviour. Although there are two game
theoretical views of rules, i.e., seeing rules as game forms and
as equilibria of games, endogenous changes of rules are not
modelled effectively by either of these two views. We introduce a
model for integrating the two views, in which the interaction
rules of replicator equations change dynamically. Computer
simulations of an example of the model that include mutation and
extinction of both strategies and games show (1) an intermittent
change of strategy distributions, (2) a continual transition from
a dominant strategy to another, and (3) metastable states distinct
from the Nash equilibria. We discuss the notion of evolutionary
stability of games and some natural examples showing rule
dynamics. We conclude that meta-evolutionary game dynamics
enables the study of the endogenous dynamic of rules. Our model
contributes, therefore, the development of game theory approach to
the dynamics of rules.
Keywords
Dynamics of rules; Mathematical modelling; Meta-evolutionary game
dynamics; Replicator equations; Evolutionary stable games
Abstract
A framework to study evolution of rules of dynamical systems is
proposed with a shift-like dynamics called a functional shift. The
functional shift is determined by a shift map on a set of bi-infinite
sequences of some functio ns having the same domain and codomain.
Considering the bi-infinite sequence of functions corresponding to an
iterated function of a dynami cal system, the functional shift allows
us to analyze the dynamics of a function governing the stat e change
of a dynamical system. Here the function is referred to as a `rule'.
We study the relevance of functional shifts to sofic shifts. We prove
that a class of functional shifts of finite type, having bi-infinite
sequences of function s given by a shift of finite type, is equivalent
to that of sofic shifts. From this theorem, we also prove that the
topological entropy of general functional shifts of fini te type is
computable, and we provide an algorithm to compute it. Next, we show
some properties of the topological entropy of rules and meta-rules. A
meta-rule is a rule that governs the change of rules. Since the
relation between meta-rules and rules corresponds to that of rules and
states (namely, a relation of operators and operands), it is clear
that states, rules, and meta-rules form a hierarch ical relationship.
Furthermore, since we can construct rules that govern change of
meta-rules, i.e., meta-meta-rules, the hierarchical relationship
continues recursively. The significant result is that the topological
entropy of meta-rules provides the upper limit for that of rules.
This means that the higher the level of rules in the hierarchy is, the
stronger the non-linearity o f these rules. The hierarchical
relationship is discussed in terms of properties of rules and
meta-rules. The presented framework can be used to study
`self-modifying' systems in which rules are used to ch ange the rules
themselves. We investigate self-modifying systems using a network of
operator-operand relations. We argue that the network is invariant
under shift maps. Finally we discuss the significance of this result
to cognitive development in which the developmen t of rules of
self-modifying type can be observed.
Keywords
Rule Dynamics, Functional Shifts, Topological Entropy,
Self-referential Shifts
Abstract
This paper studies the relationship between social institutions and
individual behaviour through the development of the cognitive framework
of individuals. Social sciences have had interests in social
institutions or norms. Much of the studies treat the problem from the
viewpoint of social costs and lack focusing on individual action and
cognition. To consider the problem as a whole naturally means that the
discussion have to be made according to subjectivism. Setting
subjectivism forth as a premise, the problem of relativity of the
cognition is studied through reconsideration of the conception of
information and its transfer. By adopting a multi-agent model having
the cognitive framework and the interaction among agents, the formation
process and the feature of institutions are computationally
investigated. From results of simulation, it is confirmed that the
cognitive frameworks of agents are affected action mimicing others'
superficial behaviour. Moreover, it is eventually shown that the
assumption of isolated individuals is unsuited even in studying in less
communicative communities.
Keywords
Formation of institutions, Institutional action, Cognitive framework,
Multi-agent model
Abstract
This article deals with the process of formation of norms in
societies in which individuals act according to their own
cognitive framework under communication. The individuals acquire
information from others and interpret it. The way of individual's
action is revised through changing the source of information and
reforming the method of interpretation. Through the revising
mechanisms, assemblages sharing cognitive frameworks establish.
At first individuals adopt the same source of information and then
arrange the shared cognitive framework. The assemblage are
considered as groups with common norms, since the same cognitive
framework gives actions coherency. In the process of formation
the two revising mechanisms function in turn. The intensity of
interaction among individuals affects the period to build norms
and the size of groups sharing norms. The size develops under
strong interaction but the period comes to long. The dependency
of average size of norms on the strength is a power.
Keywords
Formation of Norms; Cognitive Framework; Multi Agent Model
Abstract
The role of social norms is to process `fundamentally' imperfect
information. Information about future events, often needed in economic
activity, is inevitably imperfect, since there's no way to check its
correctness from our subjective viewpoint. The reason to be able to act
well in spite of `fundamentally' imperfect information is that we have
social systems available to process and complement imperfect
information. In this article, the problem of transferring information
between agents is investigated with a multi-agent model. The model
represents dual subjective interpretations of information by agents,
information senders and receivers. Two types of norms emerge in the
agents' system. One is the norm between senders and receivers, and the
other among receivers. The former facilitates clear communication
between senders and receivers. On the other hand, the latter intervenes
to form the former norms. The receivers decide their action by
referring to others' behaviour, so some clusters are formed. It is
notable that the relationship between intervals of referring to others'
behaviour by receivers and the average size of clusters is a power.
This means that even if the agents seldom refer to others, there is a
possibility of clusters emerging. Economic implications of our model
and results are discussed.
Keywords
Multi-Agent Model, `Fundamentally' imperfect
Information, Norm Formation
Abstract
Significance and implication in the constructive study of
dynamics of language are discussed. The constructive studies
admit three important viewpoints in study of language. These are
dynamics, emergence and subjectivity. However, showing global
dynamics in language structure based on internal dynamics of
subjects has not been succeeded. An attempt to synthesize
language dynamics with the developing word-web model is
introduced.
Kerword
Dynamics of Language, Developing Word-web Model
Abstract
Categorization dynamics as the clustering of words in word
relation is studied by a constructive approach which is suited
to inquire evolutionary linguistics with dynamical view on
language. Word meaning is represented by relationship among
words. Tthe relationship should be derived from usage of
language. Being founded on this usage-based view, we define an
algorithm to evaluate word relationship. Using the algorithm,
cluster structure and its dynamics of words are shown in a model
with communicating artificial agents. The relevance of
clustering with linguistic categorization is discussed.
Keywords
Dynamic View of Language; Categorization;
Usage-based viewpoint; Constructive approach;
Word similarity; Word clusters
Abstract
Dynamics of structure in relationships among words via
communication is studied by a constructive approach. Our study is
based on the dynamical view of language, which is that meanings of
words are dynamically created through activities of sense-making
by individual language users. We simulate ``conversation'' in
which agents having word relation matrices as their internal
structure speak and listen to sentences. The sense-making
activity is modeled by situating words in a web of relationships
among words according to the usage of the words. We observe that
agents develop cluster structure in word relationships, which is
endowed with both stability and adaptability. It is considered
that the cluster structure shares some features with prototype
category. The structure changes dynamically by appearing new
usages of language. The coexisting commonality and individuality
of cluster structure is observed in an ensemble of individuals.
Kerword
Dynamical View of Language (動的言語観),
Relationship among Words (単語間関係),
Constructive Approach (構成論的アプローチ),
Dynamics of Cluster Structure (クラスタ構造のダイナミクス),
Individualty and Commonality (個別性と共通性)
Abstract
Development of category structure in communication is studied by
a constructive approach. Individuals having a word relation
matrix as their internal structure communicate by uttering and
accepting sentences. Words in sentences uttered are situated in
relation with other words by each individual. Words make
clusters according to their interrelationships. The structures
and dynamics of clusters are studied. Coexisting commonality and
individuality of clustering structure is observed in an ensemble
of individuals.
Keywords
Categorization; Dynamical View of Language; Complex Systems;
Constructive approach; Evolutionary linguistics; Commonality and Individuality
Abstract
Development of meaning structure is studied from a usage-based
viewpoint by a constructive approach. The meaning structure is
represented by relationships between words. A word's relationship to
other words, which represents meanings of the word, is derived by
analyzing similarity of the word's usage in sentences. Words make
clusters according to their similarity. The word clusters are
classified into several types and show some remarkable dynamics. The
clusters and their dynamics are studied by means of hierarchical
cluster analysis and principal coordinate analysis.
Keywords
Usage-based viewpoint; Constructive approach;
Evolutionary linguistics; Word similarity; Word clusters
Abstract
The development of structure of relationships between words is
studied with a constructive approach by means of artificial agents
with grammar systems. The agents try to recognize given sentences in
terms of their own grammar. A word's relationship to other words,
which represents meanings of the word, is derived by analyzing the
word's usage in sentences, and which is calculated via the mutual
dependency between words and sentences. The agents differentiate
recognized words into clusters in a space of relationships among
words. The structures of clusters can be classified into several
types. The dynamics of clusters such as merging, boundary
expansions, structural changes are observed. These clusters and
their dynamics have some relevance with linguistic categorization.
Keywords
Evolutionary linguistics; Usage based meaning; Meaning as word
relationship; Word similarity; Categolization; Prototype
Abstract
Development of syntax and semantics of language is studied with a constructive approach in a network model of agents. Agents with generative grammar systems evolve through communication in terms of their own grammar in a language game. They form a community sharing usages of symbols and develop to equip recursive structures. After development, agents become to have double articulation mechanism in their grammar. Semantics is expressed by word-space which is a space to represent interrelationship among words. The relationship is formed according to ways of usage of language and calculated by means of mutual dependency between words and sentences. An agent can develop categorization in its word-space from sentences.
Abstract
Evolution of symbolic language and grammar is studied in a network
model. Language is expressed by words, i.e. strings of symbols, which
are generated by agents with their own symbolic grammar system. Agents
communicate with each other by deriving and accepting words in terms of
their own grammar. They are ranked according to their communicative
effectiveness: an agent which can derive less frequent and less
acceptable words and accept words in less computational time will have
higher scores. They can evolve by mutational processes, which change
rewriting rules in their symbolic grammars. Complexity and diversity of
words increase in the course of time. The emergence of modules and loop
structure enhances the evolution. On the other hand, ensemble structure
lead to a net-grammar, restricting individual grammars and their
evolution.
Keywords
Net-grammar; Algorithmic evolution; Module-type evolution Evolution of
language; Symbolic grammar systems
Abstract
Evolution of symbolic language and grammar is studied in a network
model. Language is expressed by words, i.e. strings of symbols,
which are generated by agents with their own symbolic grammar
system. By deriving and accepting words, the agents communicate
with each other. An agent which can derive less frequent and less
acceptable words and accept words in less computational time will
have higher scores. Grammars of agents can evolve by mutationally
processes, where higher scored agents have more chances to breed
their offsprings with improved grammar system. Complexity and
diversity of words increase in time. It is found that the module
type evolution and the emergence of loop structure enhance the
evolution. Furthermore, ensemble structure (net-grammar) emerges
from interaction among individual grammar systems. A net-grammar
restricts structures of individual grammar and determines their
evolutionary pathway.
Keywords
Net-grammar; Algorithmic evolution;
Module-type evolution Evolution of language; Symbolic grammar
systems
Abstract
Interacting agents with symbolic grammar are proposed in order to study
the evolution of computational ability of agents. The algorithmic
evolution of the formal grammar system is characterized by Chomsky's
hierarchy$^1$. Agents with a higher grammar can speak/recognize many
more words than those with a lower one. However, when agents form a
network, the higher Chomsky grammar is not always advantageous. It is
shown that to speak/recognize commonly used words is more favorable in a
network.
Abstract
A problem of self-referential paradox and self-reproduction is discussed
in a network model of machines and tapes. A tape consists of a bit
string, encoding function of a machine. Tapes are replicated when it is
attached by an adequate machine. Generally, a tape is replicated but it
may be different from the original one. In this paper, external noise
evolves diversity in a system. New reaction pathway induced by external
noise will be reproduced deterministically by an emerging autocatalytic
network. Hence it will remain stable after external noise is turned
off. Low external noise develops a minimal self-replicative loop. When
external noise is elevated, a more complex network evolves, where a core
structure emerges. Tapes in a core network can be bifurcated into
either a RNA-like or a DNA-like tape with respect to its usage in an
autocatalytic loop.
Abstract
The origin and evolution of genetic system is studied by the metaphor
of machine-tape coevolutionary system. Mutation is taken as a
rewriting process which machines act on tapes. A tape consists of a
bit string, encoding a machine function. Tapes can be replicated only
being read by adequate machines. It is reported in our previous
studies, complex but stable autocatalytic structures evolve under the
effects of external noise.
An ensemble of such autocatalytic networks are the present study. Each
network, embedded in a cell structure, can exchange machines with
other networks. Each cell is assumed to duplicate when defined
conditions are met. A duplicate cell has the same kinds and
populations of machines and tapes as its parental cell. It is
reported that developed cells are classified into two different types.
One cell type has unstable dynamics which generates cell
differentiation. The other cell type has stable dynamics, those cell
types replicate unlimitedly with leaving identical cell types. The
former differentiating cells form an ecology with essentially large
degrees of freedom. On the other hand, the latter cell types are
cancer-like cells, whose dynamics cannot be effected by the other
existing cells. Hence those cancer cells are characterized as an
independent small degrees of freedom.
Artificial Life, 2, (1995), 305--318
Abstract
A self-reproduction via description is discussed in a network model of
machines and description tapes. Tapes consist of bit strings, encoding
function of machines. A tape is replicated when it is read by an
adequate machine. Generally, a machine rewrites a tape without doing
correct replication. The variation in a reproduced tape is taken as
mutation. Since this mutation is caused by a machine's program, we call
it active mutation. Which machine is translated from a given tape is
dependent on what kind of a machine reads the tape. External noise is
introduced in machine's reading process to make errors. A new reaction
pathway is induced by external noise via machine's error action. We
find that the induced pathways will be mimicked deterministically in an
emerging core structure. This core structure will be remained stable
after turning off external noise. Low external noise develops a core
structure of minimal self-replicative loop. When external noise is
elevated, a more complex network evolves. Machines composing a complex
core network, which has been bred in high external noise, will actively
rewrite tapes rather just replicate them. Self-replication not as an
individual but as a network now becomes important.
Keywords Active mutation; Self-reproducing machines; Tapes; Core network; Double loop replication
Back to TH & Colleague' Works
Hashimoto's lab
Complex systems analysis
School of Knowledge Science
JAIST