wp.complexity.co.krwp.complexity.co.kr/wp-content/uploads/2014/04/chapte…  · web view286chapter...

61
16 A Short Version of the Whole 16 전전전 전전 전전 This book has explored the origin and elaboration of interacting signals and boundaries in various complex adaptive systems, ranging from biological cells (membrane hierarchies) and eco- systems (niches) to markets (equity and derivative exchanges) and governments (departmental hierarchies). 이 이이 이이 이이(이 이이 이이)이 이이이(이이이)이이 이이(이이이 이이 이이 이이) 이이이 이이 (이이 이이 이이)이 이이이 이이이 이이 이이이 이이이이이이 이이 이이이이이 이이이이이이 이이이이이 이이이 이이이이 이이 이이. The fact that signal/boundary hierarchies are a common feature of all these systems suggests the possibility of a general, overarching explanation of many cas properties. 이이이/이이이이 이이 이이이 이이 이이이 이이이이이 이이이이 이이이이이 이이이 이이 이이이이이 이이이이 이이이이이 이이이 이이이 이이이이 이이 이이. The framework proposed here— dynamic (finitely) generated systems—centers on mechanisms that generate signal/boundary hierarchies. 이이이 이이이 이이이이이 -- 이이 (이이이이) 이이 이이이 -- 이 이이이/이이이이 이이 이이이 이이이이 이이이이이 이이이 이이.

Upload: lamkien

Post on 05-Mar-2018

216 views

Category:

Documents


4 download

TRANSCRIPT

16 A Short Version of the Whole16 전체의 짧은 버전

This book has explored the origin and elaboration of interacting

signals and boundaries in various complex adaptive systems,

ranging from biological cells (membrane hierarchies) and eco-

systems (niches) to markets (equity and derivative exchanges)

and governments (departmental hierarchies).

이 책은 생물 세포(막 계층 구조)와 생태계(니치들)에서 마켓(증권과 파생 상품 교환) 그리고 정부(부서 계층 구조)에 이르는 범위에 걸쳐 다양한 복잡적응계에 있어 시그널들과 바운더리들의 상호작용의 기원과 복잡함을 탐구 했다.

The fact that signal/boundary hierarchies are a common feature

of all these systems suggests the possibility of a general, overarching

explanation of many cas properties.

시그널/바운더리 계층 구조가 모든 이러한 시스템들의 일반적인 특징이라는 사실은 많은 복잡적응계 특성들의 일반적이고 중요한 설명의 가능성을 제시 한다.

The framework proposed here— dynamic (finitely) generated

systems—centers on mechanisms that generate signal/boundary

hierarchies.

여기서 제시된 프레임워크 -- 동적 (유한하게) 생성 시스템 -- 는 시그널/바운더리 계층 구조를 생성하는 메커니즘의 중심에 있다.

16.1 Agents16.1 행위자들

All complex adaptive systems exhibit obvious internal boundar-

ies that divide the cas into a diverse array of semi-autonomous

subsystems called agents.

모든 복잡적응계는 복잡적응계(cas)를 행위자들이라 불리는 준 자율적인 하부 시스템의 다양한 집합체로 분할하는 명확한 내부 바운더리들을 보인다. Each agent has a “program” that guides its interactions with other

agents and other parts of its environ- ment.

각 행위자는 다른 행위자들과 환경의 다른 부분과 상호작용을 유도하는 "프로그램"을 가진다. To provide a common framework for the programs used by

different kinds of cas agents, the book concentrates on classifier

systems for the following reasons:

상이한 종류의 복잡적응계(cas) 행위자들에 의해 사용된 프로그램들을 위한 공통 프레임워크를 제공하기 위해 책은 다음 사유로 분류자 시스템에 집중 한다.

A Short Version of the Whole 283

• Classifier systems use conditional (If/Then) signal-processing

rules, thus capturing the conditional interactions characteristic

of signal/boundary systems.

• 분류자 시스템들은 시그널/바운더리 시스템들의 조건부 상호작용 특성을 포착하는 조건부 (IF/Then) 시그널 처리 규칙들을 사용한다. • A classifier system executes many signal-processing rules

simultaneously without incurring problems of consistency, thus

capturing the simultaneous interactions typical of complex

adaptive systems.

• 분류자 시스템은 복잡적응계의 전형적인 동시 상호작용들을 포착하도록 일관성의 문제를 일으키지 않고 동시에 많은 시그널 처리 규칙을 실행 한다 .

Classifier systems are computationally complete, so any com-

putational procedure (e.g., any agent strategy) can be presented

by a set of classifier-system signal-processing rules.

• 분류자 시스템들은 임의의 계산 절차(예를 들어 임의의 행위자 전략)를 분류자 시스템 시그널 처리 규칙들의 집합으로 나타 낼 수 있을 정도로 계산적으로 완벽하다.• Tags in a classifier system play the role of active sites, motifs,

and the like in different complex adaptive systems, allowing

direct comparisons of tag-mediated signal/boundary networks

in different complex adaptive systems.

• 분류자 시스템에서 태그들은 상이한 복잡적응계에서 태그-중계 시그널/바운더리 네트워크의 직접 비교를 허용 하면서 복잡적응계에서 활성 사이트(부위), 모티프, 그리고 그와 같은 유사한 것 의 역할을 한다. • Classifier systems are specifically designed for use with genetic

algorithms, so that the coevolution of signals and boundaries

can be generated and studied.

• 분류자 시스템들은 시그널들과 바운더리들의 공진화가 생성되고 연구될 수 있도록 유전자 알고리즘을 사용하도록 특별하게 설계 되었다 .

Whereas classifier systems handle the signal-processing aspect of signal/boundary systems, boundaries are specified indirectly, via tags. 분류자 시스템들이 시그널/바운더리 시스템의 시그널 처리 부분을 다루는 반면에 바운더리들은 태그들을 통해 간접적으로 지정된다.Tagged urns (chapter 7), a generalized version of the urn models

A Short Version of the Whole 283used in probability theory, offer a direct approach to boundaries.태그가 붙은 항아리들(제 7 장), 확률 이론에 사용된 항아리 모델들의 일반화된 버전은 바운더리들에 직접 접근을 제공한다.Different colors of balls in the urns represent different tagged signals (resources), and the balls can be moved randomly from urn to urn in diffusive fashion, allowing signals to be “broadcast.” 항아리에 있는 공들의 상이한 색상은 상이한 태그가 붙은 시그널들(자원들)을 나타내고 시그널들이 "브로드캐스트" 되도록 허용하는 확산 형태로 공들은 무작위로 항아리에서 항아리로 이동한다. Boundaries with differing permeabilities to balls with different tags can be directly represented by assigning different entry and exit conditions to different urns. 상이한 태그들을 가진 공들에 대해 상이한 투과성을 가진 바운더리들은 상이한 항아리들에 상이한 진입출 조건들을 할당함으로써 직접 표현될 수 있다. These entry and exit conditions are defined using the same notation used to define conditions for classifier rules. Reactions between balls in the same urn are provided using a “billiard-ball chemistry” (chapter 3).이러한 진입출 조건들은 분류자 규칙들에 대한 조건들을 정의하는 데 사용된 동일한 표기법을 사용하여 정의된다. 동일한 항아리에서 공들간의 반응들은 "당구공 화학"(제 3 장) 을 사용하여 제공 된다.In more detail, tagged urns bring to the fore several characteristic features of signal/boundary systems:[bring something to the fore : ~을 표면화하다[전면에 내세우다]

좀 더 세부적으로 태그가 붙은 항아리들은 시그널/바운더리 시스템들의 몇 가지 특징들을 전면에 내세운다.

• Compartmented populations Using probabilities to

designate the concentrations of different colors of balls in an urn

allows calculations relevant to the multiplicity of signals in a

typical cas compartment; urns arrayed hierarchically mimic a

hierarchi- cal arrangement of cas compartments.

• 분리된 개체군들 : 항아리에 있는 공들의 상이한 색상들의 농도를 지정하는데 사용하는 확률들은 전형적인 복잡적응계 구획에서 시그널들의 다양성과 관련된 계산을 허용한다.

A Short Version of the Whole 283• Diffusion through semi-permeable membranes Moving balls from urn to urn, subject to entry and exit

conditions, models the effect of diffusion through semi-

permeable membranes and the counterparts of constrained

diffusion in other signal/bound- ary systems.

• 반투막을 통한 확산 : 진입출 조건들에 따르는 공들을 항아리에서 항아리로 이동하는 것은 반투막을 통한 확산의 효과와 다른 시그널/바운더리 시스템들에 있어 제한된 확산의 대응물을 모델링 한다.• Conditional action Using conditional probabilities to

repre- sent reaction rules paves the road for using Markov processes

to study signal processing in hierarchical arrays of semi-permeable

membranes.

• 조건부 행동 : 반응 규칙들을 나타내기 위해 조건부 확률들을 사용하는 것은 반투막들의 계층 구조 집합체들에 있어 시그널 처리를 연구하는데 마르코프 프로세스를 사용토록 길을 낸다.• Simultaneous processing Repeated random draws of pairs

of colored balls model the effects of simultaneous pairwise reac-

tions occurring in production lines, recycling, niches, and the

like.

• 동시 처리 : 색상을 가진 공들의 쌍의 반복된 무작위 추첨은 생산 라인, 재 활용, 니치들 그리고 그와 유사한 것에서 발생하는 동시 짝 반응(pairwise reaction)의 효과들을 모델링 한다.

• Coevolution Using a genetic algorithm to modify signal tags

and entry and exit conditions offers a direct approach to

adaptive, coevolutionary changes that modify boundary

permeability.

• 공진화 : 시그널 태그들과 진입출 조건들을 수정하는데 사용하는 유전자 알고리즘은 바운더리의 투과성을 수정하는 적응적, 공진화적 변화들에 대한 직접적인 접근법을 제공한다.

The foregoing makes it clear that conditional rules, reacting to signals on the basis of tags, play critical roles both in clas- sifier systems and in systems of tagged urns. 앞서 말한 것은 태그들에 기초한 시그널들에 반응하는 조건부 규칙이

A Short Version of the Whole 283분류자 시스템에서, 태그가 붙은 항아리들의 시스템에서 중요한 역할을 한다는 것을 명확하게 한다.The two systems can be linked by coding signals and reactants as binary strings, so in both cases conditions can be defined using #’s (“don’t cares”).두 개의 시스템은 바이너리 스트링으로 시그널들과 반응물들을 부호화 함으로써 연결될 수 있다. 그래서 양쪽 경우에 시그널들은 #("신경 쓰지마")을 사용하여 정의될 수 있다.

284 Chapter 16

Cross-breeding of tags provides grist for the evolutionary mill,

with emphasis on combinations of building blocks for tags that

have previously been exploited by the signal/bound- ary system.

태그들의 교차-교배는 시그널/바운더리 시스템에 의해 이전에 활용된 태그들에 대한 초석의 조합에 역점을 두고 진화에 도움을 준다.

For example, a single-point crossover between two conditions

typically yields one offspring condition that is more general (has

more #’s) and another that is less general.

예를 들어, 두 조건 간에 1 점 교차는 통상적으로 더 일반적인(더 많은 #을 가진) 하나의 자손 조건과 덜 일반적인 다른 자손을 낳는다 .

Because complex adaptive systems are organized around

populations, both offspring options can be explored without

abandoning advantages already attained by the parents.

복잡적응계가 개체군들 주변에 구성되기 때문에 양쪽 자손 옵션들은 부모에 의해 이미 획득된 장점들을 포기하지 않고 탐색될 수 있다.

As a result, succeeding generations make greater use of tags that

offer additional advantages.

결과로서 연속하는 세대들은 부가적인 장점들을 제공하는 태그들을 더 많이 사용한다.

In other words, progressive cross-breeding explores the levels

between “craftsman” generalists and “production-line” specialists

in the milieu of signals and resources available to the agents.

다른 말로, 점진적인 교차-교배는 행위자들에게 가용한 시그널들과 자원들의 환경에서 "장인" 제너럴리스트와 "생산라인" 스페셜리스트 간의 수준들을 탐색한다.

16.2 Niches16.2 니치들

The concept of niche has been used to discuss quite different

signal/boundary systems, including ecological niches, market

niches, personal niches, skills niches, and technological niches.

284 Chapter 16니치의 개념은 생태계 니치들, 마켓 니치들, 개인적인 니치들, 실력 니치들 그리고 기술적 니치들을 포함해서 상당히 상이한 시그널/바운더리 시스템들을 논의하는데 사용 되었다.When suitably generalized to multi-agent interactions, it offers

a way of comparing these different signal/boundary systems.

다중 행위자 상호작용에 대해 적절하게 일반화될 때 그것(니치)은 이러한 상이한 시그널/바운더리 시스템들을 비교하는 방법을 제공한다.

The mammalian immune system (chapter 9) provides a well-

studied example of the role of signals, tags, and boundaries in

delimiting a niche concerned with controlling invading patho-

gens.

포유류 면역 시스템(제 9 장)은 침입하는 병원체 통제와 연관된 니치 범위를 정하는데 있어 시그널들, 태그들과 바운더리들의 역할에 대해 잘 연구된 예를 제공한다.The invaders are identified by characteristics of their surface

molecules (tags) that distinguish them from body cells and

normal resident bacteria.

침입자들은 신체 세포와 일반 거주 박테리아로 부터 그들을 구분하는 그들의 표면 분자들(태그들)의 특징으로 구분된다. (The problem is complicated because a single mammal is in itself a

complex ecosystem, with hundreds of species of resident bacteria co-

existing with a large array of body cells. Typically the total

number of beneficial, or at least harmless, resident bacteria

outnumbers the number of body cells by a factor of 10 or more.)

(문제는 하나의 포유류는 신체 세포들의 많은 집단과 수백 종의 거주 박테리아가 공존하는 자신의 복잡한 생태계에 있기 때문에 복잡 해진다. 통상적으로 유익하거나 적어도 해가 없는 거주 박테리아의 총 수는 신체 세포 수의 10 배 이상 수가 더 많다.)

A Short Version of the Whole 285

Once identified, the invaders are sequestered in special compartments (vacuoles), where they are dismantled by enzymes keyed to their tags.

[keyed to : ~에 적합한]

한번 구분되면 침입자들은 그들의 태그들에 적합한 효소에 의해 분해되는 특별한 구획(액포-세포질의 액포)들 안으로 고립된다 .

However, that is only one stage of the evolutionary story.하지만 그것은 진화 스토리의 한 스테이지에 불과 하다.

Pathogenic bacteria have evolved signals to redirect the immune system’s activities, even converting a cell’s interior compartments into safe havens for their reproduction. So the “arms race” continues.병원성 박테리아(세균)는 그들의 재 생산을 위해 심지어 세포의 내부 구획들을 안전한 피난처로 바꾸면서 면역시스템의 작용을 우회 하도록 시그널들을 진화 시켰다. 그래서 "군비 경쟁"이 지속된다.

Body cells that manage to neutralize the invaders actually flour- ish in the niche provided by the invaders.침입자들을 중성화 하도록 관리하는 신체 세포들은 실제로 침입자들에 의해 제공된 니치에서 번성한다.

A rather different example of niche formation is provided by

the development of a eukaryote from a single fertilized egg.

니치 형식의 오히려 다른 예는 단일 수정란에서 진핵 생물의 발달에 의해 제공된다.Starting from the fertilized egg, cell division yields an ever larger

colony of cells, the cells in the interior experiencing higher

concentrations of some proteins than those on the exterior.

수정란에서 출발해서, 세포 분할은 끊임없이 더 커지는 세포 군집, 외부에 비해 일부 단백질들의 더 높은 농도를 체험하는 내부에 있는 세포들을 낳는다. Differential concentrations, along with induced signals, cause

different sets of genes to be repressed or de-repressed in cells at

different locations in the growing colony.

유도된 시그널들을 따라, 상이한 농도들은 자라는 군집에 상이한 위치에 있는 세포들에서 억제 되거나 활성 되는 상이한 유전자들의 집합을 야기 한다.

As a result of this differentiation, cells at different locations in

the colony con- struct different adhesion molecules for their

A Short Version of the Whole 285bounding mem- branes;

이런 분화의 결과로서 군집에 상이한 위치에 있는 세포들은 그들을 둘러싸고 있는 막들에 대한 상이한 부착 분자들을 구성한다.they also produce new signaling molecules, causing still further

differentiation. (See chapter 13.)

그들은 또한 여전히 더 분화를 일으키면서 새로운 시그널 전송 분자들을 생성한다. (제 13 장 참조)The development of the mammalian eye is a case in point. [a case in point : (논의 중인 문제·상황 등에) 딱 들어맞는 사례]

포유류의 눈의 발달은 딱 들어맞는 사례 이다.Cells in the portion of the ectoderm (exterior) that will become the

mammal’s eye undergo differential adhesion, with the result that

they form an interior colony and an exterior colony. [ectoderm (외배엽) : 정자와 난자가 수정하여 형성되는 초기 배아는 세 개의 세포층으로 나눌 수 있는데 , 이 중 초기 배아의 바깥 층을 뜻함]

포유류의 눈이 될 외배엽(외부) 부분에 있는 세포들은 내부 군집과 외부 군집을 형성하는 결과로 차별 접착(부착)을 겪는다.The interior colony bends to form a hollow niche, which is then

“invaded” by the exterior colony. [bend to : ~에 따르게 하다. ~에게 ~ 하도록 강요하다]

내부 군집은 외부 군집에 의해 "침입되는" 속인 빈 니치를 형성하도록 한다. The invading colony, after several further foldings, forms the

lens and the cornea;

몇 번의 접합을 더 진행 후에 침입하는 군집은 수정체와 각막을 형성한다. the interior colony forms the retina. (See figure 16.1.)

내부 군집은 망막을 형성한다. (그림 6.1. 참조)At each stage in this complicated folding process, the niches formed

by the dividing cells exchange signals that turn genes on and off,

thereby modifying adhesion and other parts of the development

program.

복잡한 접힘 과정에 있는 각 스테이지에서 세포들을 분할 함으로써 형성된 니치들은 유전자를 켜고 끄는 시그널들을 교환한다 그래서 부착과 발달 프로그램의 다른 부분들을 수정한다.

286 Chapter 16

Neural ectoderm Lens ectoderm

Prospective retina

Lens capsule

Opticvesicle

Lens placode

Cornea

Neural retina

Figure 16.1A schematic diagram of the morphogenesis of an eye.

눈의 형태 발생의 계통도[ectoderm : 외배엽 - 외배엽은 배아가 분열하면서 신경계로의 분화가 일어나는 중요한 세포층이다. 이는 주로 척추, 말초 신경, 뇌, 눈, 귀를 포함하는 신경계를 형성한다.Neural ectoderm : 신경 외배엽 - 표피 외배엽과 구분되어 신경계의 원기를 이루는 외배엽 세포Lens ectoderm : 수정체 외배엽, Optic vesicle : 안포 - 척추 동물의 배에서 전뇌 중 간뇌가 되는 부역의 좌우에서 측방을 향해 부풀어 나오면서 생기는 1 쌍의 포낭체. 장래 눈의 주요부(망막 및 망막색소세포층)를 형성하는 것으로, 그 벽은 신경아세포를 주체로 한 높은 원주상피로 구성되어 있다. 전뇌와 연속하는 안포의 기부는 가늘게생겨 눈자루라고 한다. 안포의 내강은 안병 속에 뻗어 있는 가는 구멍을 거쳐서 신경관 내강의 뇌실원기와 연락된다. 안포는점차로 측방으로 돌출하여 표피층에 가까이 가지만, 그 사이에 표피에 면한 쪽의 중심부역은 렌즈를 유도하면서 함몰되어 전체가 잔모양의 안배가 된다.Placode : 기원판(紀元板) : 플라코드 - 주로 척삭동물의 발생에서 여러 감각기관, 어떤 종의 신경절의 기원이 되는 외배엽의 비후. 기원판은 더 증식하며 대부분 모층에서 분리되어 특정 조형과정을 거쳐 각각 특수한 분화를 한다. 전형적인 기원판은 렌즈, 내이, 측선 등의 원기에서 볼 수 있다.Lens placode : 수정체 기원판Retina : 망막, Perspective retina : 시각 망막, Lens capsule : 수정체피막, Cornea : 각막, Neural retina : 신경 망막]

Unfortunately, the niche concept, though widely used, is

typically defined informally or in a narrow, context-dependent

way. A well-defined, widely applicable concept of niche should

encompass common features of the different versions:

불행히도 광범위하게 사용된 니치 개념은 통상 형식에 구애되지 않거나 협소한 맥락 의존적 방법으로 정의된다. 잘 정의된 광범위 하게 적용 가능한 니치는 상이한 버전들의 일반적인 특징들을 포함한다.

Multiplier effects Resources don’t disappear on consumption;

they are transformed in passing from agent to agent.

286 Chapter 16승수효과 : 자원들은 소비로 사라지지 않는다. 그들은 행위자에서 행위자로 전달 중에 변환된다.

As a result, an injection of an additional resource (say, from

outside the niche) propagates from agent to agent, multiplying

the effects of the injection. (The economic version of this effect

was proved by Samuelson.)

결과적으로 부가적인 자원의 투입(말하자면 니치 밖에서)은 행위자에서 행위자로 투입의 효과를 배가하면서 퍼져 나간다. (이 효과의 경제 버전은 사뮤엘슨에 의해 증명 되었다.)

Recycling Agents are distributed spatially, and they interact in

tangled loops of conditional responses and resource-passing

(e.g., phenotypic plasticity in ecosystems (Levin 1999)).[phenotypic plasticity (표현형적응성) : 표현형 적응성은 하나의 유전형이 환경의 변화에 따라 다양한 표현형을 만들어 내는 현상으로 이러한 현상들 중 몇몇 예들은 기존의 생물체라고 보기 힘들 정도로 너무나도 다른 형태의 표현형을 만들어 낸다. 과거에는 표현형적응성이라는 용어가 발생 과정에서 변화만을 말하는 것이지만, 지금은 더욱 넓어져 성체 상태에서의 표현형 변화도 표현형적응성이라 말한다.]

재순환 : 행위자들은 공간적으로 분산되어 있고 그들은 조건부 반응과 자원 전달의 꼬인 루프에서 상호작용 한다. (예를 들어 생태계에서 표현형적응성(Levin 1999)).Persistence Often patterns of interaction, imposed on a

“flow” of ever-changing agents, persist well beyond the lifetimes

of individual agents.

지속성 : 종종 항상-변하는 행위자들의 "흐름"에 부여된 상호작용의 패턴들은 개체 행위자들의 생명주기를 넘어 상당히 지속된다.These persistent patterns act as niches within which diverse agents can co-exist.이러한 지속적인 패턴은 다양한 행위자들이 공존하는 니치의 역할을 한다.

288 Chapter 16

The introduction of a new kind of agent to a niche—say, from some

other niche (e.g., an invasive species) or by recombination of tags

(e.g., a new flu virus)—can cause a torrent of further innovations

exploiting the new possibilities.

니치에 새로운 종류의 행위자의 도입은 -- 말하자면 어떤 다른 니치(예를 들어, 외래 유입종) 에서 오는 또는 태그들의 조합의 의해(예를 들어, 신종 독감 바이러스) -- 새로운 가능성들을 개발하는 더 혁신의 홍수를 불러 일으킨다.

The concept of a community within a network (Newman,

Barabasi, and Watts 2006) provides a starting point that leads

naturally to an overarching definition of niche.

네트워크 내에서 커뮤니티의 개념(Newman,Barabasi and Watts 2006)은 자연스럽게 니치의 핵심적인 정의에 이르는 출발점을 제공한다.

A niche is a diverse array of agents that regularly exchange resources and depend on that exchange for continued existence.

니치는 정규적으로 자원들을 교환하고 지속된 존재를 위해 교환하는 데 의존하는 행위자들의 다양한 집합체 이다.

Most signal/boundary systems exhibit counterparts of ecological

niche interactions— symbiosis, mutualism, predation, and the like.

대부분의 시그널/바운더리 시스템은 생태계의 니치 상호작용들의 대응물들을 드러낸다. - 공생, 상리공생, 포식과 그와 같은 것들

From this definition of niche, it is relatively easy to move to an

evolutionary dynamic for niches, because the conditional actions that

underpin the interlocking activities can be defined and compared

in a uniform way by means of tag-sensitive rules. (See chapter 3.)

니치의 이 정의로 부터 연동작용들을 떠받치고 있는 조건부 작용이 태그에 민감한 규칙들의 도움으로 동일한 방법에서 정의되고 비교될 수 있기 때문에 상대적으로 니치에 대한 진화 동역학으로 옮기는 것은 쉽다. Defining niches in terms of tag-based rules leads to an important mechanism-oriented question: 태그 기반 규칙의 측면에서 니치들을 정의하는 것은 중요한 메커니즘-지향적인 질문에 이른다.

Can mechanisms for manipulating tags (such as recombination), in combination with selection for the ability to collect resources, lead from simple niches to complex niches?

288 Chapter 16(재조합과 같은)태그들을 다루는 메커니즘들이 자원들을 수집하는 능력에 대해 선택을 가진 조합에서 단순한 니치들에서 복잡한 니치들에 도달 할 수 있는가 ?The conditional actions of the agents lead to non-additive effects that cannot be usefully averaged, so a purely statistical approach isn’t likely to provide answers to this question. 행위자들의 조건부 행동들은 유용하게 평균될 수 없는 비 가산 효과들을 유도 한다. 그래서 순수하게 통계적 접근법이 이 질문에 대한 답을 제공할 가능성이 없다.Statistical approaches wind up in the same cul- de-sac as statistical approaches to understanding computer pro- gramming.[cul- de-sac : 막다른 골목, 난감한 상황, wind up : (어떤 장소, 상황에) 처하게 되다]

통계적 접근법들은 컴퓨터 프로그램밍을 이해하는데 통계적 방법들 처럼 똑같이 난감한 상황에 처하게 된다.The “trends” suggested by a series of “snapshots” based on an average over agents’ activities rarely give reliable predictions or opportunities for control—instead of “clearing” of a market, we get “bubbles” and “crashes.”[clearing of market / market clearing : 시장청산 - 시장에서 수요와 공금이 일치된 상태, 주어진 시장 가격에서 생산자와 소비자가 원하는 물건을 모두 팔거나 살 수 있는 상태를 뜻한다. 그러나 시장청산은 몇 가지 가정을 충족해야 일어날 수 있다. 시장청산에 필요한 가정은, 모든 시장은 완전경쟁적이며, 시장 참여자들은 완벽한 정보를 공유하며, 시장에서 충격이 발생하면 그에 따른 조정이 신속히 발생하고, 시장에 불완전성이나 마찰 요인이 존재하지 않는다는 것이다.)]

행위자들의 행동들에 대한 평균에 기반을 둔 "스냅샷들"의 연속으로 제안된 "트렌드들"은 거의 믿을 만한 예측들 이나 통제에 대한 기회들을 주지 않는다 -- 시장 청산 대신 우리는 "거품"과 "공황"을 얻는다.

A more fruitful approach to questions of this kind requires an

examination of the coevolution of situated agents—agents that

have locations in both a geographic sense and a network sense.

이런 종류의 질문에 대한 더 유익한 접근법은 상황에 처한 행위자들의 공진화에 대한 조사를 요구한다. -- 지리적 센스와 네트워크 센스 모두에서 위치가 있는 행위자.

The “bandits” discussed in chapter 9 offer a boundary-oriented

setting for examining situated agents.

제 9 장에서 논의된 "슬롯머신"은 상황에 놓인 행위자들을 조사하기 위한 바운더리 지향된 설정을 제공한다.

Each bandit (slot machine) has a queue of players that share any

payoffs that occur.

각 슬롯머신은 발생하는 임의의 보상을 공유하는 플레이어들의 큐를 가진다.

288 Chapter 16The longer the queue, the less each agent in the queue receives.

큐가 길 수록 큐에 있는 각 행위자는 더 적게 받는다.

Each bandit can be looked upon as a tagged urn that dispenses a

mixture of colored balls (resources), with agents forming a queue

to share those resources.

각 슬롯 머신은 혼합 색상을 가진 공들(자원들)을 공유하는 큐를 형성하는 행위자들과 더불어 그러한 자원을 제공하는 태그를 가진 항아리로서 보일 수 있다.

From a niche point of view, an agent’s location (situation) is

determined by the queue it joins.

니치의 관점에서 행위자의 위치(상황)은 참여하고 있는 큐에 의해 결정 된다.

In the tagged-urn version, an agent can migrate to a queue only

if it has a tag that satisfies an entry condition assigned to the queue.

태그가 있는 항아리 버전에서 행위자는 큐에 배정된 진입 조건을 충족하는 태그를 가지고 있는 경우에만 큐로 이주 할 수 있다.

With this provision, a “generalist” agent accesses a broad array of

urns but often winds up in queues with low payoff per agent

because of crowding, whereas a “specialist” agent accesses a queue

that is less crowded but more difficult to locate.

이러한 조항으로 "제너럴리스트" 행위자는 항아리의 광범위한 집합체에 접근 하지만 종종 과밀로 인해 행위자당 낮은 보상을 가진 큐에 처하기도 한다. 반면에 "스페셜리스트" 행위자는 덜 밀집 하지만 위치하기에 더 어려운 큐에 접근한다.

As was shown in chapter 9, typical niche effects, such as crowding,

niche invasion, and specialization, can be examined via the

queued bandit approach.

제 9 장에서 보인것 처럼 밀집, 니치 침입과 분화와 같은 전형적인 니치 효과들은 큐를 가진 슬롯머신 접근법을 통해 조사 될 수 있다.

In a more complex, still more niche-like version of the queued bandits, agents within a queue exchange resources with other agents in the queue via the usual tag-based interactions.

여전히 큐를 가진 슬롯머신의 더 니치 같은 버전인 더 복잡한 상황에서 큐 내에 있는 행위자들은 통상적인 태그 기반 상호작용을 통해 큐에 있는 다른 행위자들과 자원들을 교환한다.

288 Chapter 16Displacement of one kind of agent by another can then arise when a specialist reproduces in a given queue at a faster pace than a generalist.

다른 종류의 행위자에 의한 한 종류의 행위자의 변위(이동)는 스페셜리스트가 제너럴리스트 보다 빠른 페이스로 주어진 큐에서 재생산 할 때 발생할 수 있다 .The generalist survives only in queues where it isn’t out-competed by specialists. [outcompete : 경쟁에서 이기다.]

제너럴리스트는 오직 스페셜리스트가 경쟁에서 이기지 못한 경우에 큐에서 생존한다.Symbiosis arises when two agents make different requirements on the queue’s resources and, via tag-based interactions, exchange their surplus shares of particular resources. 공생은 두 행위자들이 큐의 자원에 대한 상이한 요건을 만들 때, 태그 기반 상호작용을 통해, 특정한 자원들의 잉여 공유 교환 할 때 발생한다.Groups of agents can also organize into default hierarchies, which leads to more rapid accumulation of adaptations. (See chapter 6.)행위자들의 그룹들은 또한 더 빠른 적응의 축적에 이르는 상이한 계층구조를 조직할 수 있다.

A Short Version of the Whole 289

16.3 Theory16.3 이론

Progressive co-adaptation of tags causes continual change in tag-

based interactions in signal/boundary systems.

태그들의 점진적 상호적응은 시그널/바운더리 시스템들에서 태그 기반 상호작용에 있어 지속적인 변화를 야기 시킨다.That is a major complication, but it is also a source of insights.

그것이 주요 문제 이지만 또한 통찰력의 근원 이다. As we have seen, widely different agent-based signal/boundary

systems, ranging from biological cells to governments, exhibit

the same general features, notably the following:

우리가 봤던 것 처럼 생체 세포에서 정부에 이르는 광범위하게 상이한 행위자 기반 시그널/바운더리 시스템들은 현저하게 다음과 같은 동일한 일반 특징들을 드러낸다 .

semi-autonomous subsystems (agents)

반 자율적 하부 시스템들 (행위자들)

hierarchical organization (agents composed of agents)

계층적 조직 (행위자들로 구성된 행위자들)

sustained diversity (agents exploiting different strategies)

지속된 다양성 (다른 전략들을 활용하는 행위자들)

extensive recycling of resources (agents converting resources and passing them on).

광범위한 자원들의 재순환 (자원을 변환하고 전달하는 행위자들)As was emphasized in the preceding section, it is much easier to compare different signal/boundary systems when we can find a mechanism-oriented framework that encompasses all signal/ boundary systems. 이전 섹션에서 강조 했던 것 처럼 우리가 모든 시그널/바운더리 시스템들을 아우러는 메커니즘 지향적인 프레임워크를 발견 할 수 있을 때 상이한 시그널/바운더리 시스템들을 비교하는 것은 훨씬 더 쉽다.

With such a framework, we can organize the diverse array of

observations of signal/boundary systems in a way comparable to

the way in which the valence mechanisms suggested by the

A Short Version of the Whole 289periodic table organized the endlessly varied observations of the

alchemists.

그와 같은 프레임워크로 우리는 연금술사들의 끊없는 다양한 관찰로 구성된 주기율표에 의해 제안된 원자가 메커니즘 방법에 맞먹는 방법으로 시그널/바운더리 시스템들의 관찰의 다양한 배열을 구성할 수 있다.

Mechanisms that generate and modify tags via recombina-

tion and other evolutionary mechanisms play a pivotal role in

the framework proposed here.

재조합과 다른 진화 메커니즘들을 통해 태그들을 생성하고 수정하는 메커니즘들은 여기서 제안되는 프레임워크에서 중추적인 역할을 한다.The definition of fitness is critical when evolutionary mechanisms

come into play.

적합도의 정의는 진화 메커니즘들이 동작 할 때 핵심적 이다.The approach taken here emphasizes implicit fitness, in which

the fitness of an agent depends on its ability to collect the

resources necessary to duplicate its structure, as in the Echo

models (Holland 1995).

여기서 주어진 접근법은 행위자의 적합도가 에코 모델에서 처럼 자신의 구조를 복제하는데 필요한 지원들을 수집하는 능력에 따라 다른 암묵적인 적합도를 강조한다.For an agent defined via a hierarchy of tagged urns, this require-

ment entails collecting the “letters” necessary to produce copies

of the urns’ entry and exit conditions, as well as copies of the

“catalyst rules” that determine signal processing within the

agent.

태그를 가진 항아리들의 계층구조를 통해 정의된 행위자에 대해 이 요건은 행위자 내에 시그널 처리를 결정하는 "촉매 규칙들"의 사본들 처럼 항아리의 진입출 조건들의 사본들을 생산 하는데 필요한 "문자들"을 수집하는 것을 수반한다.

A Short Version of the Whole 291

Both conditions and rules are defined over a three-letter alphabet

{1,0,#), so the agent must collect sufficient numbers of these letters

to replicate its structures.

조건들과 규칙들 모두 세 개의 문자 알파벳 {1,0,#}을 통해 정의되고 행위자는 자신의 구조들을 복제 하기 위해 충분한 수의 이러한 문자들을 수집해야 한다.

It does this by means of its programmed interactions with other

agents, and by collecting sources of letters (if any) in its

environment.

다른 행위자와 자신의 프로그램된 상호작용과 자신의 환경에 있는 (만약 있다면) 문자들의 소스들을 수집 함으로써 이를 수행한다.

Implicit fitness, so defined, opens the way for a great diversity of

agents and niches, with recycling playing a critical role.

그렇게 정의된 암묵적 적합도는 핵심적인 역할을 하는 재순환으로 상당히

다양한 행위자들와 니치들에 대한 방법을 연다.Dynamic (finitely) generated systems, discussed in chapter

12 and 14, offer a framework that meets the foregoing desid- erata. (I have suppressed ‘finitely’ in the abbreviation dgs to keep it short.)

제 12 와 14 장에서 논의된 동적 (유한하게) 생성 시스템들은 앞서 말한 꼭 필요한것을 충족하는 프레임워크를 제공한다. (나는 짧게 유지하려고 약어 dgs 에서 "유한하게"를 숨겼다.)The dgs framework is a generalization of the constrained generating procedures developed in Holland 1998.

dgs 프레임워크는 1998년 홀랜더가 개발한 제한 생성 프로시저의 일반화 이다.Within the dgs framework, different signal/boundary systems

are described with the same alphabet and the same fixed set of

dgs operators (and meta-operators).

dgs 프레임워크에서 상이한 시그널/바운더리 시스템들은 동일한 알파벳과 동일한 고정된 dgs 오퍼레이터 집합( 그리고 메타-오퍼레이터)로 기술된다.

Each agent is represented by a set of tagged strings defining both its internal hierarchy of boundaries (an urn hierarchy) and the content of the compartments in that hierarchy.

A Short Version of the Whole 291각 행위자는 자신의 바운더리의 내부 계층구조(항아리 계층구조)와 그 계층구조에 있는 구획들의 내용 모두를 정의하는 태그가 있는 스트링들의 집합으로 표현된다.Tags, besides identifying parts of the agent, make possible the

sequencing of rules, thereby making the agent programmable

within the dgs.

게다가 행위자의 부문들을 구별하는 태그들은 규칙들의 시퀀싱을 가능하게 하고 그래서 dgs 내에서 프로그램 가능한 행위자를 만들 수 있게 한다.Overall, tags have a role similar to the role of motifs in molecular

genetics, determining the specificity of counterparts of enhancers,

promoters, silencers, ligands, catalysts, and the like (Alberts

2007).

전반적으로 태그들은 분자 유전학에서 증강인자(enhancers), 프로모터(promoters), 침묵인자(silencers), 리간드, 촉매 그리고 그와 같은 것들(Alberts 2007)의 대응물을 지정을 결정하는 모티프의 역할과 유사한 역할을 가진다.

In a dgs, the generated succession of sets (populations) of

tagged strings represents successive generations of agents.

dgs 에서 태그가 붙은 스트링들의 집합들(개체군들)의 생성된 연속은 행위자들의 지속적인 생성을 나타낸다..Operators and meta-operators—particularly meta-operators

acting as a genetic algorithm—modify the tags, thereby

modifying the urns and their content.

오퍼레이터들과 메터-오퍼레이터들 - 특히 유전자 알고리즘으로써

동작하는 메타-오퍼레이터들 - 은 항아리들과 그들의 내용을 수정 하도록

태그들을 수정한다. The result is a trajectory of progressive adaptations and

coevolution in successive generations.

결과는 연속적인 세대에서 점진적 적응과 공진화의 자취 이다.

Everything within the dgs, whatever the signal/boundary

system being modeled, uses the same operators and meta-operators

to manipulate the strings.

시그널/바운더리 시스템이 모델링 되는 것이 무엇이든 dgs 내의 모든 것은 스트링들을 조작하기 위해 동일한 오퍼레이터들과 메타-오퍼레이터들을 사용한다.As a consequence, the coevolution of signals and boundaries in

quite different signal/boundary systems can be compared within

A Short Version of the Whole 291the dgs.

결과적으로 꽤 상이한 시그널/바운더리 시스템에서 시그널들과 바운더리들의 공진화는 dgs 내에서 비교될 수 있다.The only differences between dgs models of different

signal/boundary systems arise from differences in the starting set

of strings (the X(0) of chapter 14).

상이한 시그널/바운더리 시스템들에 대한 dgs 모델들간의 유일한 차이점들은 스트링들의 시작 집합의 차이점에서 발생 한다 (제 14 장의 X(0) ).The corpus generated from X(0)—the sequence of generations—

follows entirely from the actions of the fixed set of operators and

meta-operators, much as theorems are generated from different

axiom sets use the same fixed rules of deduction.

X(0)로 부터 생성된 코퍼스 --생성들의 시퀀스--는 마치 상이한 공리 집합으로 부터 생성된 정리들이 추론에 대해 동일한 고정된 규칙들을 사용하는 것과 마찬가지로 완전히 오퍼레이터들과 메타-오퍼레이터들의 고정된 집합의 작용을 따른다.

The procedures for developing a dgs model relevant to a

particular set of signal/boundary observations are similar to the

procedures for writing a computer simulation.

시그널/바운더리 관찰들의 특정 집합에 관련된 dgs 모델을 개발하기 위한 절차들은 컴퓨터 시뮬레이션을 쓰는 것에 대한 절차와 유사하다. That is, the dgs generators and operators serve much like the

instruction set for a computer.

즉 dgs 생성자들과 오퍼레이터들은 마치 컴퓨터에 대한 명령어 집합과 같은 역할을 한다.

However, the dgs generators and operators are selected to

provide direct descriptions of the rules and conditions employed

by tagged urns.

하지만 dgs 생성자들과 오퍼레이터들은 태그가 있는 항아리들에 의해 채택된 규칙들과 조건들의 직접적인 설명을 제공하기 위해 선택된다. As a result, networks of urn and agent interactions become

important emergent phenomena in a dgs.

결과적으로 항아리와 행위자 상호작용 네트워크들은 dgs 에서 중요한 창발 현상이 된다.Moreover, network modifications are generated by dgs operators that produce agent adaptations.

A Short Version of the Whole 291게다가 네트워크 수정은 행위자 적응을 생산하는 dgs 오퍼레이터들에 의해 생성된다.Coevolutionary changes in tags can generate a cascade of subtle

network changes that are difficult to characterize in terms of

direct manipulation of nodes and edges.

태그들에 있어 공진화적 변화는 노드들과 엣지들에 대한 직접 조작의 측면에서 특징 짓기 어려운 미묘한 네트워크 변화들의 캐스케이드를 생성 할 수 있다.Roughly, the difference between emergent modification and direct

modification is the difference between modifying the genes that

specify a metabolic network and directly modifying the

reactants in that network.

대략 창발 수정과 직접 수정간의 차이는 대사 네트워크를 지정하는 유전자들을 수정하는 것과 그 네트워크에서 반응물들을 직접적으로 수정하는 것간의 차이 이다 .

Tags—in the guise of motifs, active sites, conditions, and the like

—have a large role in determining signal/boundary dynam- ics.

For that reason, a relevant dgs theory should provide ways of

generating new tags from extant tags.

태그들 - 모티프들, 활성 사이트들, 조건들 그리고 그와 같은 것들의 모습(외관)에서 - 은 시그널/바운더리 동역학을 결정함에 있어 큰 역할을 한다.

292 Chapter 16

In a dgs, a building block for a tag is a short string that appears

at the same loci in several different tags (akin to a motif in

genomics).

dgs 에서 태그를 위한 초석은 (유전학에서 모티와 유사한) 여러 다른 태그들에서 동일한 위치(유전자좌)에 나타나는 짧은 스트링 이다.

In the terminology of genetic algorithms, such building blocks are

called schemata.

유전자 알고리즘의 용어에서 그와 같은 초석들은 스키마타라고 불린다..

Combining a small number of such building blocks produces a

wide variety of “plausible” new tags for defining conditions and

signals.

그와 같은 작은 수의 초석들을 조합하는 것은 조건들과 시그널들을 정의하기 위해 대단히 다양한 "그럴듯한" 새로운 태그들을 생성한다.

Some of these “plausible combinations” facilitate the formation and

testing of new default hierarchies, “production lines,” and the like.

이러한 "그럴듯한 조합들"의 일부는 새로운 디폴트 계층구조들,"생산 라인들", 그리고 그와 같은 것의 형성과 테스팅을 가능하게 한다.

Because populations are involved, many variants can be tried

simultaneously without significant loss of capability.

개체군들이 참여 하기 때문에 많은 변형들이 기능의 큰 손실 없이 동시에 시도 될 수 있다.

16.4 Mathematical models16.4 수학적 모델들

Theory is used to suggest experiments and “laws” that explain

features appearing repeatedly in some wide range of observa-

tions.

이론은 어떤 광범위한 관찰들에서 반복적으로 나타나는 특징들을 설명하는 실험들과 "법칙들"을 제안 하는데 사용된다.

The dgs framework proposed in this book aims at the pervasive

occurrence of hierarchical signal/boundary interactions in

complex adaptive systems.

292 Chapter 16이 책에서 제안된 dgs 프레임워크는 복잡적응계에서 도처에 발생하는 계층적 시그널/바운더리 상호작용에 목표를 두고 있다.To provide the abstractions necessary for theory, the framework

centers on tagged urns that serve as semi-permeable boundaries

for agent compartments.

이론을 위해 필요한 추상화를 제공하기 위해 프레임워크는 행위자 구획을 위한 반투막 바운더리 역할을 하는 태그가 있는 항아리들에 중점을 둔다.The compartments, in turn, contain reactions (and interactions)

defined by a “billiard-ball chemistry.”

다음에 구획들은 "당구공 화학"에 의해 정의된 반응들(과 상호작용들)을 포함한다.Under this regime, the state of this dgs model at any time is a

probability distribution over the possible concentrations of signals

and resources within the tagged urns.

이런 체제 하에서 임의의 시간에 이 dgs 모델의 상태는 태그가 있는 항아리 내에 있는 시그널들과 자원들의 가능한 농도에 대한 확률 분포 이다.The probabilities defined by this framework open the way for

using the mathematics of Markov processes. (See chapter 15.)

이 프레임워크에 의해 정의된 확률들은 마르코프 프로세스 수학을 사용하는 길을 연다.

Markov matrices nicely capture the conditional probabilities

that define the changes imposed by diffusion/reaction systems,

as represented by diffusion and catalyzed reactions in the

tagged-urn models.

마르코프 행렬은 항아리에서 확산과 촉매 반응들로 표현된 것 처럼 확산/반응 시스템들에 의해 내포된 변화를 정의하는 조건부 확률들을 잘 포착 한다.

A Short Version of the Whole 293

Each entry in the matrix gives the condi- tional probability P(c’|c)

of going from one arrangement of balls in the urns (c, the row

label) to another arrangement (c’, the column label).

행렬에서 각 항목은 항아리들에 있는 공들의 한 배치(c, 행 라벨)에서 다른 배치(c', 열 라벨)로 이동하는 조건부 확률 P(c'|c).를 준다.

Because the system has a finite number of urns, the number of

possible arrangements (v) is finite, so there are v rows and

columns in the matrix, and there v components in the state

vector that gives the probability of any given arrangement.

행렬에서 각 항목은 항아리들에 있는 공들의 한 배치(c, 행 라벨)에서 다른 배치(c', 열 라벨)로 이동하는 조건부 확률 P(c'|c).를 준다.

Networks offer another theoretical approach, one that is

more “pictorial.”

네트워크들은 더 그림이 많은 다른 이론적 접근법을 제공한다.Each node c in a network represents a particular arrangement of

balls among urns, and two nodes c and c' are connected by a

directed edge just in case moving a single ball can take the

system from node (arrangement) c to node c'.

네트워크에서 각 노드 c 는 항아리들 사이에 있는 공들의 특정한 배치를 나타내고 두 노드 c 와 c'은 하나의 공을 움직이는 것이 노드(배치) c 에서 노드 c'으로 가는 시스템을 취할 수 있는 경우에만 방향이 있는 엣지로 연결된다. When the edges are weighted with the corresponding probabili-

ties, the network provides the same information as the Markov

matrix.

엣지들이 해당하는 확률들로 가중치가 부과될 때 네트워크는 마르코프 행렬 처럼 동일한 정보를 제공한다.

With the help of these theoretical devices, we can approach

a question of some importance in evolution of signal/boundary

systems.

이러한 이론적 장치들의 도움으로 우리는 시그널/바운더리 시스템의 진화에 있어 어떤 중요한 질문에 접근 할 수 있다.

Under what conditions does the system “settle down” (that is,

reach an equilibrium)?

A Short Version of the Whole 293무슨 존건하에서 시스템이 "정착" 하게 되나 (즉 평형에 도달)?When balls move randomly (diffuse) among a fixed set of urns

with no entry and exit conditions, balls of any given color are

equally distributed among the urns.

공들이 진입출 조건들이 없는 항아리들의 고정된 집합 사이에서 무작위로(확산된) 움직일 때 임의로 주어진 색상의 공들은 항아리들 간에 동일하게 분포되어 있다.How do entry and exit conditions on the urns modify this result?

어떻게 항아리들에 있는 진입출 조건들이 이 결과를 수정하는가?Standard methods for Markov theory can be used to show that,

under conditions usually satisfied by balls diffusing among the

tagged urns, the probability of each arrangement becomes fixed

after sufficient time elapses.

마르코프 이론에 대한 표준 방법들은 항아리들 간에 확산하는 공들에 의해 통상적으로 충족되는 조건 하에서 충분한 시간이 흐른 뒤에 각 배치 확률이 고정되는것을 보여 주는데 사용될 수 있다.More technically, the probability vector X(t) describing the state

of the signal/boundary system settles down to a fixed point.

That is, X(t + 1) = X(t)M = X(t), where M is the matrix

describing the diffusion process.

좀더 기술적으로 시그널/바운더리 시스템의 상태를 기술하는 확률 벡터 X(t)는 고정 점으로 정착 한다. 즉 X(t + 1) = X(t)M = X(t), 여기서 M 은 확산 프로세스를 기술하는 행렬 이다.

294 Chapter 16

However, most complex adaptive systems don’t settle down.

하지만 대부분의 복잡적응계는 정착 하지 않는다.

The adaptations and coevolution produced by mechanisms

such as genetic operators (recombination, mutation, etc.) change

tags and tagged urns more or less rapidly, thereby modifying the

Markov matrix M.

유전자 오퍼레이터들(재조합, 돌연변이, 등등)과 같은 메커니즘에 의해 생성된 적응과 공진화는 태그들과 태그가 있는 항아리들을 다소 빠르게 변화 시키면서 마르코프 행렬 M 을 수정 한다.

Nevertheless, a look at the steady states before and after one or

two applications of genetic operators can give some idea of the

resulting changes in fitness.

그럼에도 불구하고 전과 후 안정 상태들을 살펴 보면 하나 또는 두개의 유전자 오퍼레이터의 적용은 적합도에서 결과적으로 초래된 변경에 대한 어떤 아이디어를 줄 수 있다.

And, as was pointed out earlier, the changes that produce higher

fitness are the ones that persist.

그리고 이전에 지적했던 것 처럼 더 높은 적합도를 생성하는 변화들은 지속되는 것이다.

In considering mathematical approaches to signal/boundary

systems, it is important to recognize that agents situated in a

physical environment do encounter regularities.

시그널/바운더리 시스템에 수학적 접근법들을 고려하는데 있어 물리적 환경에 놓인 행위자들이 규칙적인 패턴을 접한다는 것을 인식하는 것은 중요하다.That is, the environments encountered are a highly constrained

subset of the set of all conceivable environments.

즉 마주한 환경은 모든 가능한 환경들의 집합의 아주 제한된 부분 집합 이다.It is a bad mistake to assume that nearly all conceivable

environments occur with non-zero probability.

거의 모든 가능한 환경들이 0 이 아닌 확률로 발생한다고 가정하는 것은 아주 좋지 않은 실수를 범하는 것이다.

That mistake leads to misconceptions such as Laplace’s

calculation that the sun will rise tomorrow or the “no free

lunch” theorems that assert that adaptive algorithms offer no

294 Chapter 16overall advantage (Wolpert 1992).

그 실수는 태양은 내일 떠 오를 것이라는 라플라스 계산이나 그 적응 알고리즘들은 전반적인 장점을 제공하지 않는다고 주장하는 [NLF(No Free Lunch ) 정리] "공짜는 없다" 정리 처럼 오해에 이르게 한다.Indeed, it is these observed regularities that make the tiered laws

of the physical sciences possible.

실제로 가능한 물리 과학의 단층로된 법칙들을 만드는 이러한 관찰된 규칙성들이 있다.

In realistic environments, regularities offer exploitable oppor-

tunities for the formation of niches involving persistent, inter-

dependent interactions of diverse agents.

실제적인 환경에서 규칙성들은 지속성, 다양한 행위자들의 서로 의존하는 상호작용을 포함하고 있는 니치들의 형성을 위해 활용 가능한 기회들을 제공한다. Once a niche is formed through adaptation and coevolution, its

persistence depends on using arrangements of signals and

boundaries to exploit environmental regularities.

적응과 공진화를 통해 니치가 형성되면 그것의 지속성은 환경의 규칙성들을 활용하기 위해 시그널들과 바운더리들의 배치를 사용하는 것에 달려 있다.In the signal/boundary context, recom- bination of tags offers

substantial advantages over uniform random search or searches

that depend on the law of large numbers.

시그널/바운더리 맥락에서 태그들의 재조합은 균일한 무작위 검색 또는 큰 수의 법칙에 의존하는 검색들에 대해 상당한 이점들을 제공한다.Agents that exploit the tags with conditional rules play a

pivotal role in sustaining the niche.

조건부 규칙들로 태그들을 활용하는 행위자들은 니치를 유지하는데 있어 중심이 되는 역할을 한다.

296 Chapter 16

And while the niche persists, new opportunities for interaction

proliferate, offering roles for still newer agents. (See section

16.1.)

그리고 니치가 지속되는 동안 계속 더 새로운 행위자들을 위한 역할들을 제공하는 상호작용에 대한 새로운 기회들이 급증한다. (섹션 16.1 참조)

16.5 Some questions16.5 몇 가지 질문들

Because bounded subsystems (especially adaptive agents) are

ubiquitous features of signal/boundary systems, priority goes to

questions concerning their origin, their interaction, and their

coevolution.

경계가 있는 하부 시스템들(특히 적응적 행위자들)은 시그널/바운더리 시스템들의 아주 흔한 특징들이 있기 때문에 우선 순위는 그들의 기원, 상호작용 그리고 공진화에 관한 질문으로 이어 진다.As was emphasized above, such questions center on the

coevolution of tags that mediate the routing and collecting of

signals within and between agents.

위에서 강조한 것 처럼 그와 같은 질문들은 행위자들 내부와 행위자들 사이에서 시그널들의 라우팅과 수집을 중계하는 태그들의 공진화에 중심에 있다.Coevolution, then, centers on various kinds of recombination,

ranging from somatic recombination in the immune system, through

various kinds of horizontal transfer in viruses and prokaryotes, to

crossover in eukaryotes, and technological innovation via new

combina- tions of tested components.

위에서 강조한 것 처럼 그와 같은 질문들은 행위자들 내부와 행위자들 간에 시그널들의 라우팅과 수집을 중계하는 태그들의 공진화에 중심에 있다.Answers to two broad questions would greatly improve our

understanding of this coevolutionary process.

두 가지 광범위한 질문들에 답하는 것은 이런 공진화 과정에 대한 우리의 이해를 상당히 증진 시킬 것이다.

The first of these questions is “What are typical steps in the

formation and evolution of niches?”

296 Chapter 16이러한 질문들의 첫 번째는 "니치들의 형성과 진화에 있어 전형적인 단계들은 무엇인가?"There are several closely related questions:

이러한 질문들의 첫 번째는 "니치들의 형성과 진화에 있어 전형적인 단계들은 무엇인가?"How is the development of niches affected by agents’ capabilities,

such as conditional response (e.g., phenotypic plasticity),

anticipation, and “lookahead” based on internal models?

조건부 반응(예를 들어 표현형 적응성), 예측과 내부 모델에 기반한 "통찰력" 과 같은 행위자들의 영량에 영향을 받는 니치들의 발달은 어떻게 되는가?What are the mechanisms that implement the transition from

individual “generalists” to networks of “special- ists”?

개별 "제너럴리스트"에서 스페셜리스트의 네트워크로 전이를 구현하는 메커니즘들은 무엇인가?What kinds of niches promote robustness in signal/ boundary

networks?

어떤 종류의 니치들이 시그널/바운더리 네트워크에서 강건함을 촉진 하는가?What kinds of niches encourage increasing diversity?

어떤 종류의 니치들이 증가하는 다양성을 촉진하는가?The second question is “What are the mechanisms that favor

the boundary hierarchies?”

두 번째 질문은 "바운더리 계층구조들을 선호하는 메커니즘은 무엇 인가?Again there is a closely related question:

다시 밀접하게 연관된 질문이 있다.

Can changes in the networks produced by coevolutionary processes

be described in terms of general operations on nodes and edges?

공진화 프로세스에 의해 생성된 네트워크에 있어서 변화들이 노드와 엣지에 대한 일반 오퍼레이션의 관점에서 기술될 수 있는가?

It is a thesis of this book that general answers to these questions

hinge on constructing a dgs based on operators and generators

extracted from the commonalities of signal/boundary system

296 Chapter 16hierarchies.

공진화 프로세스에 의해 생성된 네트워크에 있어서 변화들이 노드와 엣지에 대한 일반 오퍼레이션의 관점에서 기술될 수 있는가?

16.6 A short version of the short version16.6 간략 버전의 간략 버전

Complex adaptive systems—cells, rainforests, markets, lan-

guage, and the Internet, to name a few—are characterized by

complex, ever-changing interactions of signals and boundaries.[ to name a few : 두서너 가지 예만 들면]

복잡적응계는 - 세포들, 우림, 시장, 언어 그리고 인터넷등 두서너 가지 예를 들어 - 시그널들과 바운더리들의 복잡하고 지속적으로 변하는 상호작용들에 의해 특징 지어 진다.This book provides exploratory models for examining the adap-

tations generated by signal/boundary coevolution.

이 책은 시그널/바운더리 공진화에 의해 생성된 적응을 조사하기 위한 탐색 모델들을 제공 한다.The exploration centers on two kinds of models: adaptive agents and

tagged urns.

탐색은 두 종류의 모델의 중심에 있다. 적응적 행위자들과 태그가 있는 항아리들.Adaptive agents (defined by signal-processing rules that provide

for parallel processing and adaptation under a genetic algorithm)

model the evolution of hierarchical systems that employ many

signals (resources) interacting simultaneously.

(유전자 알고리즘 하에서 병렬 처리와 적응 시그널 처리 규칙들에 의해 정의되는) 적응적 행위자들은 동시에 상호작용하는 많은 시그널들(자원들)을 도입하는 계층적 시스템들의 진화를 모델링 한다.

Tagged urns (modifications of the urns used in probability

theory) use entry and exit conditions to control the flow of balls

(signals) between urns, thus providing explicit formal models of

semi-permeable boundaries.

태그가 있는 항아리들(확률 이론에서 사용된 항아리들의 수정)은 항아리들 간의 공들(시그널들)의 흐름을 조절하기 위해 진입출 조건들을 사용한다. 이는 반투막 바운더리들의 명확한 형식적 모델들을 제공한다.In these models, both signals and boundaries are constructed

296 Chapter 16from building blocks provided by tags and parts of tags—the

counterparts of active sites, motifs, and message headers.

이러한 모델들에서 시그널들과 바운더리들 모두 태그들과 태그들의

파트들에 의해 제공되는 초석으로 부터 구성된다. - 활성 사이트, 모티프

그리고 메시지 헤더의 대응물

The framework of dynamic generated systems combines adaptive-

agent models and tagged-urn models to provide a precise

formalism for exploring the evolution of signal/boundary systems.

동적 생성 시스템의 프레임워크는 적응 행위자 모델들과 태그가 있는

항아리 모델들을 시그널/바운더리 시스템들의 공진화를 탐색하기 위한

정확한 형식주의를 제공하기 위해 결합 한다 .

References

Alberts, B., et al. 2007. Molecular Biology of the Cell. Garland.

Allen, M. 1977. Darwin and His Flowers: The Key to Natural Selection. Taplinger.

Arthur, W. B. 2009. The Nature of Technology. Free Press.

Arthur, W. B., et al. 1997. Asset pricing under endogenous expectations in an artificial stock market. Economic Notes 26 (2): 297–330.

Ashby, W. R. 1952. Design for a Brain. Wiley. Asimov, I. 1988. Understanding Physics. Buccaneer.Axelrod, R., and Cohen, M. D. 1999. Harnessing Complexity. Free Press.

Babbage, C., and Babbage, H. P. 2010. Babbage’s Calculating Engines: Being a Collection of Papers Relating to Them. Cambridge University Press.

Birkoff, G., and MacLane, S. 2008. A Survey of Modern Algebra. A K Peters.

Braitenberg, V. 1984. Vehicles: Experiments in Synthetic Psychology. MIT Press.

Buss, L. W. 1987. The Evolution of Individuality. Princeton University Press.

Bybee, J. 2006. Frequency of Use and the Organization of Language. Oxford University Press.

298

References

Chomsky, N. 1965. Aspects of a Theory of Syntax. MIT Press.

Christiansen, F. B., and Feldman, M. W. 1986. Population Genetics. Blackwell.

Cole, L. C. 1954. Some features of random population cycles. Journal of Wildlife Management 18 (1): 2–24.

Davidson, E. H. 2006. The Regulatory Genome. Academic Press. Dennett, D. C. 1992. Consciousness Explained. Little, Brown. Dobzhansky, T., et al. 1977. Evolution. Freeman.Epstein, J. M., and Axtell, R. L. 1996. Growing Artificial Societies. MIT Press.

Felicity, A., Clements, A., Webb, C., and Lithgow, T. 2010. Tinkering inside the organelle. Science 327: 649–650.

Feller, W. 1968. An Introduction to Probability Theory and Its Applications, volume 1. Wiley.

Feynman, R. P., Leighton, R. B., and Sands, M. 2005. The Feynman Lec- tures on Physics Including Feynman’s Tips on Physics. Addison-Wesley.

Fisher, R. A. 1930. The Genetical Theory of Natural Selection. Clarendon.

Five Graces Group (C. Beckner et al.). 2009. Language is a complex adaptive system. Language Learning 59: 1–26.

Fontana, W. 2006. Pulling strings. Science 314: 1552–1553.

Forgacs, G., and Newman, S. A. 2005. Biological Physics of the Developing Embryo. Cambridge University Press.

Forsyth, A., and Miyata, K. 1984. Tropical Nature. Scribner.

Gao, H. 2001. The Physical Foundation of the Patterning of Physical Action Verbs. Lund University Press.

Gao, H., and Holland, J. H. 2008. Agent-Based Models of Levels of Con- scious. SFI Working Papers.

References

299

Gilbert, S. F., and Epel, D. 2009. Ecological Developmental

Biology. Sinauer. Goldberg, D. E. 2002. The Design of Innovation. Kluwer.Good, M. C., Zalatan, J. G., and Lim, W. A. 2011. Scaffold proteins: Hubs for controlling the flow of cellular information. Science 232: 680–686.

Han, J., Li, M., and Guo, L. 2006. Soft control on collective behavior of a group of autonomous agents by a shill agent. Journal of Systems Science and Complexity 19 (1): 54–62.

Hebb, D. O. 1949. The Organization of Behavior. Wiley.

Hofstadter, D. R. 1999. Gödel, Escher, Bach: An Eternal Golden Braid. Basic Books.

Holland, J. H., Holyoak, K. J., Nisbett, R. E., and Thagard, P. R. 1986.Induction: Processes of Inference, Learning, and Discovery. MIT Press.

Holland, J. H. 1992. Adaptation in Natural and Artificial Systems. MIT Press.

Holland, J. H. 1995. Hidden Order: How Adaptation Builds Complexity. Addison-Wesley.

Holland, J. H. 1998. Emergence: From Chaos to Order. Addison-Wesley.

Holland, J. H. 2002. Exploring the evolution of complexity in signaling networks. Complexity 7 (2): 34–45.

Holland, J. H., Tao, G., and Wang, W. S.-Y. 2005. Co-evolution of lexicon and syntax from a simulation perspective. Complexity 10: 50–62.

Holldobler, B., and Wilson, E. O. 1990. The Ants. Harvard University Press.

Ke, J., and Holland, J. H. 2006. Language origin from an emergentist perspective. Applied Linguistics 27: 691–716.

Kemeny, J. G., and Snell, J. L. 1976. Finite Markov Chains. Springer.

Kleene, S. C. 1956. Representation of events in nerve nets and finite automata. In Automata Studies, ed. C. Shannon and J. McCarthy. Princeton.

300

References

Konig, D. 1936. Theorie der Graphen. Akademische Verlag.

Krugman, P., Wells, R. and Graddy K. 2010. Essentials of Economics.Worth.

Lanzi, P. R., et al. 2000. Learning Classifier Systems. Springer. Levin, S. A. 1999. Fragile Dominion. Perseus.Lindsley, D. L., and Grell, E. H. 1967. Genetic Variations of Drosophila Melanogaster. Carnegie Institution of Washington.

Maxwell, J. C. (ed. W. Davidson). 1890. The Scientific Papers of James Clerk Maxwell. Cambridge University Press (BiblioLife).

McClelland, J. L., and Rumelhart, D. E. 1986. Parallel Distributed Process- ing: Explorations in the Microstructure of Cognition. MIT Press.

Mitchell, M. 1996. An Introduction to Genetic Algorithms. MIT Press. Mitchell, M. 2009. Complexity A Guided Tour. Oxford University Press. Newman, J. R. 2003. The World of

Mathematics. Simon & Schuster.Newman, M., Barabasi, A.-L., and Watts, D. J. 2006. The Structure and Dynamics of Networks. Princeton University Press.

Nowak, M. A. 2006. Evolutionary Dynamics. Harvard University Press.

Rapoport, A. 1960. Fights, Games, and Debates. University of Michigan Press.

Rochester, N., et al. 1988. 1955 Tests on a cell assembly theory of the action of the brain. In Neurocomputing, ed. J. Anderson and E. Rosenfeld. MIT Press.

Samuelson, P., and Nordhaus, W. 2009. Economics. McGraw-Hill. Simon, H. A. 1996. The Sciences of the

Artificial. MIT Press.Slack, E., et al. 2009. Innate and adaptive immunity cooperate flexibly to maintain host-microbiota mutualism. Science 325: 617–620.

Smith, A. 1776. The Wealth of Nations. Penguin.

References

301

Smith, E., and Morowitz, H. J. 2004. Universality in intermediary metab- olism. Proceedings of the National Academy of Sciences of the United States of America 101: 13168–13173.

Steels, L., and Kaplan, F. 2002. 2002 Bootstrapping grounded word semantics. In Linguistic Evolution through Language Acquisition, ed. E. Briscoe. Cambridge University Press.

Tarski, A., Mostowski, A., and Robinson, R. M. 2010 Undecidable Theories: Studies in Logic and the Foundation of Mathematics. Dover.

Tolstoy, I. 1981. James Clerk Maxwell. University of Chicago Press.

Turing, A. M. 1936. On computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society 42: 230–265.

von Neumann, J. 1966. In Theory of Self-Reproducing Automata, ed. A. Burks. University of Illinois Press.

Walker, P. M. B., ed. 1990. Cambridge Dictionary of Biology. Cambridge University Press.

Whitehead, A. N., and Russell, B. 1910. Principia Mathematica. Cam- bridge University Press.

Wolpert, D. H. 1992. On the connection between in-sample testing and generalization error. Complex Systems 6: 47–94.

Zelazo, P. D., Gao, H. H., and Todd, R. 2007. Development of conscious- ness. In The Cambridge Handbook of Consciousness, ed. P. Zelazo et al. Cambridge University Press.

Index

# (‘don’t care’), 164# (‘pass through’), 76

Activation, 64, 227Active site, 33, 34Adaptation, 30–32, 53, 99–115,

214Adhesion, 232, 234, 238, 243,

250, 259Agent, 2, 8, 24, 27, 28, 37,

56–63, 95, 114, 115, 139–141,281–284

conglomerate, 226, 227in dgs, 223–226language-using, 193–197, 205,258

meta-operator, 251, 252movement, 255ontogeny, 229–243programmable, 53, 54in queue, 162–165replication, 127, 128situated, 71, 72,

Agent-based model, 48

Algorithm, 108, 213, 217, 218bucket-brigade, 128–130genetic, 57, 126, 127

Allele, 67, 69, 70, 124, 126Alphabet, 28, 63, 64, 222,

290Amino acid, 43, 74Anticipation, 60, 192Asimov, 80Autocatalysis, 105Autonomy, 9, 202Axiom system, 51, 52, 245, 255,

256, 262

Babbage, 181, 217Ball stack, 251, 252Bandit (slot machine), 161–

166, 175–180, 288Billiard-ball mechanics, 29,

79–81, 131, 132, 151, 152, 223,224

Biological cell, 25, 83, 84, 96, 97,

100–108Bjerknes, 1

304

Index

Boundary, 93, 94, 139, 219–222,

260, 263Braitenberg, 186Building block, 30, 38, 41, 45,

46, 51–56, 92, 93, 102,108–113, 118, 119, 127

Bybee, 92

Canonical form, 209, 210Carrying capacity, 168cas (complex adaptive system),

2, 21, 281Cascade, 102Catalyst, 76, 80, 91, 103, 138,

152, 153cfs (classifier system), 50, 63–67,91, 92, 125, 126, 281, 282

Chomsky, 182Chromosome, 67–69, 101–103,

118, 119, 126Coevolution, 18–20, 219, 220,

283, 295Collision, 223, 224Colony, 113, 120Community, 159, 160, 197, 198,

287Compartment, 25, 83, 135, 136,

230, 233, 238, 242, 283Competence biological, 7, 8linguistic, 191

Competitive exclusion, 168–174Condition/action rule, 59Conditional probability, 132,

133, 264–277, 293Configuration, 82, 266, 267

Conglomerate, 53–55, 111–113,

226–228Consciousness, 186–189, 201,

202Conservation, 29, 74, 79, 127,

144Corpus, 206, 213, 255–258Coupled interactions, 148–158,

259Credit assignment, 125, 126, 129.

See also Bucket brigadeCross-breeding, 31, 32, 89, 147.

See also CrossoverCrossover, 31, 68–70, 118–123,

284. See also Recombination Crowding, 164, 167, 168, 288

Darwin, 7, 18, 96Default hierarchy, 122, 123, 260Detail, 22, 23, 44–46, 85, 218,

219, 251Detector, 71, 120, 121, 191–194dgs (dynamic generated

system), 218, 219, 222–228, 239,243–264, 291

Diffusionconstrained, 136–139, 271, 272,283

unconstrained, 269, 270Directed connection, 29, 30, 85,

86, 89, 91, 277. See also Graph

Distributed control, 8, 165Diversity, 8, 11, 12, 109, 198,

199, 218, 220, 289, 295Dobzhansky, 175, 178Dynamics, 22, 23, 59, 60, 82,

304

Index172

Index

305

Echo model, 139–141, 177, 179,

221, 222Effector, 66, 71, 72, 193, 195Emergence, 108, 113, 114Entry/exit conditions, 135–138,

141–147, 221, 224, 225, 264,271, 277, 282, 283, 296

Enzyme, 83, 84, 93, 103, 112Equilibrium, 167, 168, 173, 276,

278Estimation problem, 161Euclidean geometry,36, 52, 262Exchange reaction, 77, 78

Feedback, 15, 49, 103Feedforward, 49Feller, 131Fine tuning, 260, 261 Finitely generated group,

207–211Finitely generated system,

205–207Fisher, 67, 70, 124Fitness, 55, 56, 67–70, 124–130,

221, 225, 289, 290Five Graces Group, 190 Fixed point, 278Foodweb, 85, 86

Generalists/specialists, 19, 32, 58,

177, 284Generation, 67–70, 120, 126,

127, 166, 213, 245Generator, 52, 54, 111, 112,

139, 205, 206, 209, 246–251,291

Gödel, 262Grammar, 51–55, 142, 143,

181–185, 197. See also Finitely generated system

Graph, 29, 30, 91, 92Group, mathematical. See

Finitely generated group

Hierarchy, 15–17, 94, 145–148,

202, 206, 225, 226, 251, 252,290. See also Default

hierarchy

Idiolect, 190, 198IF/THEN rule, 41, 42, 50, 59, 73,

187–189, 194. See also cfs; Classifier system; Condition/ action rule

Immune system, 8, 27, 57, 284,

285Induction, 7, 8Innovation, 58, 61, 107, 109,

178, 287, 295Internal model, 60, 202, 215,

216, 221Interpretation, 36, 37, 46, 47,

257, 258Inversion, 124Invisible hand, 8

Jumping bean, 80. See alsoBilliard-ball mechanics

Kleene, 181Krebs cycle, 30, 31, 102, 111,

260, 261K-selection, 177

306

Index

Language, 9, 12, 20, 36, 60, 61,

91,181–203, 206, 211, 214,258, 263

Language acquisition, 9, 182,

185–190, 192, 199, 200Learning, 9, 49, 183, 187 Levels of Consciousness (LoC),

186–193, 200, 201Life cycle, 259Ligand, 175Linkage, 69, 119, 123, 126Lookahead, 202, 295Loop, 7, 15, 28, 30, 49, 50, 54,

90, 257, 286. See also Feedback; Recirculation

Lotka-Volterra, 47, 172Lyell, 100, 109

Market, 1, 8, 21, 27, 48, 59, 60,

129Markov matrix, 266, 267–277,

292–294Markov process, 269–277Mating, 68, 117, 118, 227, 228.

See also Crossbreeding; Recombination

Maxwell, 10, 47Maxwell-Boltzmann

distribution, 81McClelland and Rumelhart, 49 McCulloch and Pitts, 182 Membrane, 27, 28, 93, 104, 132,

135–138, 143, 145, 146, 283

Mendeleev, 52, 108Metabolic network, 15, 25, 26,

30, 31, 96, 97, 111, 112, 291

Meta-operator, 142, 186, 187,226, 230, 231, 245, 246,249–255, 290, 291

Migration, 174, 175, 178, 224,225, 255

Mimicry, 18, 19Model, 40–44, 47–51agent-based, 27, 28billiard-ball, 28, 29, 49classifier system, 28, 50computer-based, 40–42cellular automata, 48data-driven, 42equation-based, 38–42existence-proof, 42exploratory, 43Lotka-Volterra, 47, 48network, 29, 30, 50, 51neural network, 49, 50urn, 32, 33

Motif, 89, 102, 263, 264, 282Multiplier effect, 14, 19, 160,

169–172

Natural selection, 18, 68, 104,

105Network, 7, 17, 18, 29, 30,

49–51, 85–97, 100–105,131–136, 142, 143, 197, 198,201, 263, 277, 278, 291, 293

Newton, 10, 23, 35, 38, 39, 44,

45Next-level stack, 251–254Niche, 3, 11, 15–19, 87, 88,

159–168, 173–178, 278, 279,284–288, 294, 295

Index

307

Offense/defense, 140, 141Ontogeny, 226–243Operator, 201, 205–209, 212,

214, 217, 218, 250, 251, 290,291

conditional, 212genetic, 68, 119, 124

Operon, 227, 230–243, 249, 250,

256, 257

Particle mechanics, 29, 48,79–81, 223. See also Billiard-ball mechanics

Patternas building block, 109, 114, 199persistent, 37, 38

Payoff, 161–166Periodic table, 108, 110Persistence, 105, 112, 113, 286,

287Phenotypic difference, 203Phenotypic plasticity, 87, 88,

286Pin production, 19, 31, 99,

148–151Population, 47, 67–70, 120,

125–127, 163, 172, 177, 213,214, 224, 283, 284, 292

Population genetics, 67, 103Predator/prey, 12, 47, 87, 88,

169–172Prisoner’s dilemma, 4Probability vector, 270, 278,

293Production line, 19, 31, 32, 138,

149, 150, 157

Production, linguistic, 191 Program, for computation, 24,

28, 46, 53, 54, 72, 102–105,108, 111, 211–213

Protein, 7, 8, 25, 26, 30–32, 64,

65, 74, 75, 83, 84, 96, 97

Queue, 161–189, 288

Rainforest, 8, 11–20, 58, 257Rapoport, 38Reaction, 27–29, 72–79, 93,

103–108, 150–153, 232, 256,259, 272–275, 278

binary, 72, 73,coupled, 153–158rate, 81, 132, 133, 143, 150,151

Reaction-diffusion system, 275–277, 292, 293

Recirculation, 13–15, 159, 160,

278, 279Recombination, 58, 69, 107,

117–120, 123, 175–178, 260,294, 295

Regularity, 127, 177, 294Repression, 227, 231, 241Reproduction, 5, 43, 55, 56,

124–128, 139–143, 216Reservoir, 128–130, 140, 141,

231, 233, 235, 237, 243, 254,259

Resource, 12, 13, 16, 55, 159,160, 164–179, 225, 249, 257

Rigor, 40, 46, 47, 205–208r-selection, 177

308

Index

Samuelson, 286Search space, 68, 127Seed, 260,Semi-permeable membrane, 16,

93, 104, 135–138, 143–146,283

Signal processing, 27–30, 57, 62,

63, 83, 84, 142, 143, 223,224. See also cfs; Classifier systems

Signal routing, 97, 242, 243, 295.

See also Tag Simon, 206Simultaneous processing, 191,

282. See also cfs; Classifier system

Slime mold, 13, 14, 60Smith, 8, 19, 31, 99, 148–151Sphinx moth, 18, 19Spore, 13, 14, 229–239Stage setting, 128–130. See also

Credit assignmentState, 23, 67, 82, 83, 154, 267,

273–276, 292–294State transition, 266–268Strength, 72, 125, 128–130,String, 28, 63–67, 74, 75, 149,

176, 177, 182, 183, 205,209–214, 222, 245, 256, 257.See also Linkage; Tag

Subroutine, 28, 54, 72, 102, 103,

212, 223Symbiosis. See Multiplier effect s/b, 6, 36. See also Boundary;

Signal processing

Tag, 32, 75–79, 88–96, 131–133,

136–147, 164, 175–178,195–197

Theory, formal, 10, 18, 27,35–37, 40, 45, 46, 292

Throughput, 153–155, 158Transition, 198, 203, 218, 219Turing, 108, 218

Uniformitarian hypothesis, 100,

109Universal competitor, 58, 59, 61Universal grammar, 182,Universality, computational, 13,

212, 213, 217, 218, 223Urn model, 131–138, 141, 142,

145–147, 223–226, 242, 243,

265–268, 276, 277, 282, 283

Valence, 52, 108, 289

Whitehead, 46Wright, 22, 23

Zelazo, 186