Tadeusz Szuba

A formal definition of the phenomenon of collective intelligence and its IQ measure

Abstract This paper presents a formalization of collective intelligence (CI). A molecular, quasi-chaotic model of computations allows us to model CI in social structures, and to define its measure (IQS). This methodology works for bacterial colonies and social insects as well as for human social structures. With the CI theory some patterns of human behavior receive formal justification, others can be explained as IQS optimization. The CI formalization assumes that it is a property of a social structure, initializing when individuals interact, and as a result, acquiring the ability to solve new or more complex problems. CI amplifies if the structure improves synergy, which further increases the spectrum and complexity of the problems, which can be solved together. The formalization covers areas where CI results in physical synergy and mental/logical cooperation.

A molecular quasi-random model of computations applied to evaluate collective intelligence

The paper presents how the Random PROLOG Processor (RPP), a bio-inspired model of computations, can be used for formalization and analysis of a phenomenon — the Collective Intelligence (CI) of social structures. The RPP originates from the question of why inference processes are quasi-chaotic in real life. In the RPP, clause-molecules (CMs) move quasi-randomly around in abstract Computational_PROLOG_Space(CS). CMs can carry clauses of facts, rules, and goals, or CMs can even be moving sets of facts, rules, and goals enclosed by membranes. When CMs rendezvous, an inference process can occur iff the prerequisite logical conditions are fulfilled. The RPP can be considered an implementation proposal of the NonDetenninistic Turing Machine. With the RPP, Cl can be evaluated as follows: l) the mapping is done of a given social structure into the structured computational space of the RPP; 2) beings and their behavior are translated into PROLOG expressions, carried by CMs; 3) the global or temporary goal(s) of the social structure (of ants, humans, etc.) are translated into an N-step inference (NSI); 4) on this basis, the efficiency of the NSI will be evaluated and given as the Intelligence Quotient of a Social Structure (IQS) projected onto NSI. The concept of IQS can be mathematically developed or used for practical evaluation of a given social structure.

Parallel Evolutionary Computing with the Random PROLOG Processor

The Random PROLOG Processor (RPP) is an abstract model for logical computations based on the concept ofinformation_moleculesquasi-randomly traveling and inferring in abstractcomputational_space. The simulated process or problem is defined in terms of PROLOG clauses (or sets of clauses encapsulated by membranes) seeded into the RPP by the programmer as a program. During the computational process,information_moleculesmove about, rendezvous, exchange internal information, modify themselves, assert otherinformation_molecules(or are retracted), disintegrate, or expire. Computations are unification-based logical inferences. In the RPP, the computations are parallel, randomly evolving processes. The RPP exhibits some properties of the DNA-computer, but can be considered for any computational environment. The RPP is proposed as a promising tool for parallel evolutionary computing, because PROLOG is a widely known, powerful language; because prices of multiprocessor architectures are falling sharply; and because parallel RPP implementation is easy. Since they are declarative and nondeterministic, RPP programs are extremely short. This paper presents the structure of the RPP application for parallel Evolutionary Computations with different types of problems.

A Formal Definition of the Phenomenon of Collective Intelligence and its IQ Measure

This paper formalizes the concept of Collective Intelligence (C-I). Application of the Random PROLOG Processor (RPP) has allowed us to model the phenomenon of C-I in social structures, and to define a C-I measure (IQS). This approach works for various beings: bacterial÷insect colonies to human social structures. It gives formal justification to well-known patterns of behavior in social structures. Some other social phenomenon can be explained as optimization toward higher IQS. The definition of C-I is based on the assumption that it is a specific property of a social structure, initialized when individuals organize, acquiring the ability to solve more complex problems than individuals can. This property amplifies if the social structure improves its synergy. The definition covers both cases when C-I results in physical synergy or in logical cooperative problem-solving.

Universal Formal Model of Collective Intelligence and Its IQ Measure

This paper presents proposal of a universal computational theory of Collective Intelligence (CI),. The toll for formalization, analysis, and modeling is a quasi-chaotic model of computations RPP. In the RPP, molecules (CMs) of facts, rules, goals, or higher-level logical structures enclosed by membranes, move quasi-randomly in structured Computational _Space (CS). When CMs rendezvous, an inference process can occur if and only if the logical conditions are fulfilled. It is proposed that Collective Intelligence can be measured as follows: 1) the mapping is done of a given structure of beings into the RPP; 2) the beings and their behavior are translated into expressions of mathematical logic, carried by CMs; 3) the goal(s) of the social structure is(are) translated into N-Element Inferences (NEI); 4) the efficiency of the NEI is evaluated and given as the Intelligence Quotient of a Social Structure (IQS) projected onto NEI. IQS is computed as a probability function over time, what implies various possibilities, e.g.: to order social structures according to their IQS, to optimize social structures with IQS as a quality measure, or even to compare single beings with social structures. The use of probability allows estimation of IQS either by simulation, or on the basis of analytical calculations.

Toward a computational model of collective intelligence and its IQ measure

This paper presents an attempt to build a universal computational theory of Collective Intelligence (Cl). which will serve social structures of beings including humans, ants. bacteria and other species that collectively solve problems king their families. social structures, or biological kind. The basic toll for formalization. analysis, and modeling is a quasi-chaotic model of computations, the Random PROLOG Processor (RPP). In the RPP. Clause-Molecules (CM) of facts, ales, goals, or higher-level logical structures enclosed by membranes move quasi-randomly in s@uctu& Computational-PROLOG-Space (C’S). When CM rendezvous in CS, an inference process can occur if and only if the logical conditions are h~lfXed. It is proposed in this theory that Collective Intelligence can be measured as follows: I) the mapping is done of a given social structure of beings into the RPP; 2) the beings and their behavior are tram&ed intO PROLOG expressions, carried by CMs; 3) the goal(s) of the social structure is(are) lranslatal into N-Element infere~es (.VEI); 4) the eficiency of the NEI is evaluated and given as the !ntelligence Quotient of a &ciai Structure (I@) projad onto NEI. Since fQS is computed as a probability function over time. them are various possibilities, eg.: to order social structures according to their IQS. to optimize social structures with IQS as a quality measure, or even to compare single beings with social structures. The use of probability allows estimation of IQS either by simulation, or on the basis of analytical calculations. I.BASICCONCEPTSOFMODELING COLLEC~IVE~NTELLIGENCE It is a paradox that the evaluation of the Collective Intelligence of social structures can be easier than the et aluation of the IQ of a single being, Individual intelligence has only been evaluated on the basis of external results of behavior during a problemsolving process in real life or during IQ tests. Neuropsychological processes accompanying problemsolving in the brain are still very far fkom being observable [2]. As a result, it is necessary to create abstract models of brain activity based on neuropsychological hypotheses (e.g. Luria [2j). or to use computer+riented models like Artificial Intelligence. Permission to make digital or hard copies of all or part of this wuk for pexsonal oc classroom use is gnnted without fee provided that copies UC not made or distributed for poffi or commercirl advantage and that copies bear thir notice and the tkll citation on the lint page. To copy otheawise. to republish, to post on servera or to redistribute to lists. requires prior specific permission and/or a fee. SAC 99. San Antonio, Texas 01998 ACM l-58113486-4/99/0001

Was Collective Intelligence before Life on Earth

5.00 In contrast. many more elements of Coiledively Intelligent activity can be observ& tneasw& and evaluated in a social structure. We can observe displacements and resultant actiats of being, u we-0 as exchange of infonnatcm between beiryg (cg human language. the ant’s pheromone communication system. the dance of ho to direct toward a source of honey, the crossover of genes between bacteria resulting in spreading specific resistance to ar&oti~ CtC. Individual intelligence and behavior is scaled &MI as a factor to accidental, local, and pmMiiiic pwresses. Our fimdamentai assunt~ion is that

Universal formal model of Collective Intelligence and its IQ measure

Zoktivq * lntelli ence can . with the helo of absbact chaoac models of ann~UM& . and statistical evaluation of the dobal beham of bcn ‘na in structured environments (111, [IO], [13). Underlying the design of the Random PROLOG Processor model of computations and justifjing its use ior Collective Intelligence Rxmaiiiat and modeling are these basic Mars: