Computational Simulations in Social Sciences (Les simulations computationnelles dans les sciences sociales) (p. 17-49)
Since the 1990’s, social sciences are living their computational turn. This paper aims to clarify the epistemological meaning of this turn. To do this, we have to discriminate between different epistemic functions of computation among the diverse uses of computers for modeling and simulating in the social sciences. Because of the introduction of a new – and often more user-friendly – way of formalizing and computing, the question of realism of formalisms and of proof value of computational treatments reemerges. Facing the spreading of computational simulations in all disciplines, some enthusiastic observers are claiming that we are entering a new era of unity for social sciences. Finally, the article shows that the conceptual and epistemological distinctions presented in the first sections lead to a more mitigated position: the transdisciplinary computational turn is a great one, but it is of a methodological nature.
Keywords: Social sciences, computational sciences, model, simulation, epistemology, realism, iconicity, epistemic function of models.
Simulation: Changes to Our Epistemologies (La simulation : des déplacements de notre épistémologie) (p. 51–57)
In social sciences, we need simulations in order to know how collective forms can result from complex interactions. The more complex they are – for example when integrating the effect of actors’ representations of collective forms on collective phenomena- the more divergent they can be, leading to indeterminacy. Moreover we are more sensitive to more stable, more differentiated forms, the recognition of which is easier. Combining these two trends leads us to substitute to the ideal of correspondence with social reality the comparability of simulations, in order to get means of criticizing and revising one simulation by the others. We keep in mind together the divergent scenarios that are the easier ones to be distinguished while paying attention to possible divergences that were neglected at first.
Keywords: Simulation, indeterminacy, form recognition, comparability, multiple scenarios.
Computerized Simulation in Humanities and Social Sciences (Simulation informatisée en humanités et sciences sociales) (p. 59–67)
This short text on computer simulations in humanities and the social sciences is devoted to agent-based simulations. The ingredients of an agent-based simulation are essentially two: agents and interactions. Agents with their own properties interact among them and interactions may be indirect or direct. As a consequence of an analogy with physical systems, the relevance of statistical equilibrium in economics is also discussed.
Key-words: Agent-based simulations, stochastic processes, Markov chains, Markovianism, statistical equilibrium.
Constructing Artificial Societies to Understand Real Social Phenomena (Construire des sociétés artificielles pour comprendre les phénomènes sociaux réels) (p. 69–77)
We briefly summarize the agent-based modeling and simulation approach, its main characteristics as well as the interest for the social sciences. In particular, we insist on the proximity between such a formalization and some classical frameworks in the social sciences like the methodological individualism. We then propose a use of the agent-based approach as a tool enabling to formalize and investigate the representations of the social systems by the social scientist.
Keywords: Modelling in the social sciences, simulation, agent-based modelling.
And What If Cooperation Were a Myth? A Pillar of Social Sciences Shaken by Simulation (Et si la coopération était un mythe? Un pilier des sciences sociales ébranlé par la simulation) (p. 79–89)
Many authors consider cooperation as a binding force holding human society together. More than other animals, human beings would be able to abandon immediate benefits to others, in the hope of future reciprocity. This hypothesis is, however, refuted by simulation: less cooperative agents always prevail eventually. I consider an alternative hypothesis, inspired by the study and modelling of natural language. Seemingly cooperative acts would rather be signals. Showing prosocial attitudes is a way to advertise qualities that are crucial, in our species, to attract friends and thus build up one’s social network.
Keywords: Cooperation, costly signal, evolution, language.
Agent-Based Simulation in Social Sciences: A “Crutch for the Human Mind”? (La simulation à base d’agents en sciences sociales : une « béquille pour l’esprit humain »?) (p. 91–100)
Computer-assisted simulation allows us to explore and better understand phenomena which may at first appear counterintuitive. It also, however, entails certain risks which the modelization/simulation specialist must compose with. The object of this paper is to show that when it comes to simulation, dangers and virtues cannot be dissociated. Thus the use of a formalized model is as likely to empoverish as it is to enrich our understanding of social phenomena; the formalized model therefore plays a very important heuristic role in extracting complexity (enriching our vision) from simplicity (poorness of the model). The flexibility of the model as well as that of its use allowing to repeat the experience whilst varying the angles of approach and adding or modifying certain parameters of the experience, the researcher is given great opportunity to master the complexity which she/he may then instill in her/his investigation.
Keywords: Complexity, formalization, heuristic, agent-based model, simplicity, simulation.
Why an Ontological Framework for Multi-Agent Modeling in Humanities and Social Sciences? (Pourquoi un cadre ontologique pour la modélisation multi-agents en sciences humaines et sociales?) (p. 101–133)
From Philosophy, ontology is “the science of what is, of the kinds and structures of objects, properties, events, processes and relations”. In computer sciences and knowledge management an “ontology” is a specification of a conceptualization of a given knowledge domain. For multi-agent simulation, the domain is models rather than data. To answer the question “Why an ontological framework for the multi-agent modelling in the Social Sciences?”, this paper deals first with three dimensions: (1) model engineering, (2) thematical and epistemological issues and (3) model assessment and comparisons (ontological test). Contrary to several ontologies, this paper does not propose a single representation of the knowledge domain, but a possible plurality, based on the concept of “knowledge framework” building to integrate the plurality of “point of view” co-existing in the Social Sciences within a general framework. Accordingly, the last part presents some examples of ontological points of view on a model of residential segregation derived from the Schelling’s one.
Keywords: Ontology, agent-based model of simulation, knowledge framework, model design, model building, multi-agent simulation, social simulation, ontological test.
Formal Models for the Reproduction of Agent-Based Simulations (Modèles formels pour la reproduction des simulations à base d’agents) (p. 135–149)
Since the results obtained by means of computer-based simulations have an experimental character, their validation requests their reproduction. These results are obtained by observing the outputs of the runs of a software entity, which is the concrete device of the experimentation; thus, it is this software simulation model that must support a new implementation. Although the replication of a model brings many advantages, in many cases it is problematic. This paper proposes to present simulation models as systems, to ease their replication.
Keywords: Social simulation, agents, experimentation, replication, system, formal model, theory.
Cities As Agents: Simulation of Possible Futures of the European Urban System (Les villes comme agents : simulation des futurs possibles du système urbain européen) (p. 153-180)
The systemic approach has long since been used for modelling the dynamics of systems of cities. Self-organization principles and differential equations have been broadly applied in the 1980’s. Agent-based models open new ways for simulating cities’ evolution. This paper discusses different registers and levels of explanation when it comes to cities’ growth rates differences. It also shows the interest of the agent approach for formalizing hypotheses at the meso-geographical level of cities. A short state of the art on the concept of system of cities and on the associated spatio-temporal models is given. The EuroSim model, developed with a multi-agent system, is presented. The evolution of European cities is simulated between 1950 and 2050 giving different scenarios on the opening of European borders to outside immigration and on the existence of internal economical barriers.
Keywords: Micro/macro, multi-scale, self-organization, explanation, system of cities, urban growth, agent-based simulation, multi-agent system, dynamic model, network of exchanges.
Mireille Ducassé and Sébastien Ferré
Help for the Multi-Criteria Decision: Coherence and Equity Thanks to Concept Analysis (Aide à la décision multicritère : cohérence et équité grâce à l’analyse de concepts) (p. 181–196)
Many decisions are taken by committees, for example in order to allocate resources. The decision criteria are difficult to express and the global situation is in general too complex for the participants to grasp it fully. In this article, we describe a decision process for the selection of job applicants where concept analysis is used to address these problems. Thanks to formal concept analysis and logical information systems, fair play people have the possibility to be fair to the applicants and to be consistent in their judgments across the whole decision process.
Keywords: Multi-criteria decision, decision support, social choice, formal concept analysis, logical information systems, case study.
Aurélien Décamps, Nathalie Gaussier, Philippe Laroque, and Philippe Gaussier
Segregation and Spatial Cognition (Ségrégation et cognition spatiale) (p. 197–226)
Considering traditional models of segregation and the analysis of the determinants of spatial segregation, the paper emphasizes and explains the role of spatial cognition in segregation emergence. We use a MAS based on cognitive agents who are able to build a cognitive map of their environment, depending on their range of vision, as they explore, discover and learn places of their environment. The agents develop spatial behaviours that create situations identified in the literature as “chosen segregation” or “involuntary segregation”. We discuss the specification of the model and the segregated configurations that emerge. These configurations underlie the importance of space on the analysis of individual and collective dynamics.
Keywords: Urban segregation, spatial cognition, individual signatures, collective dynamic.
The Historian and His Models (L’historien et ses modèles) (p. 227–279)
Speaking of models in history is a delicate matter. Since its emergence at the time of World War II in the area of our academic studies, the word covers different ways of thinking the historical method and practice. A minority of historians, mostly in the field of economic history, take models as new tools establishing a link with mathematics, and especially today, with game theory. On the other hand, all of this is mostly a question of semantic change for most of them, and does not mean anything deeper. Historians create and use models basically to explain their works and to establish comparisons. In some cases though they might be taken hostage by their own models, by misusing models designed for other purpose or by using them without knowledge.
Keywords: Model (definitions), mathematization, explanatory models, determinism, relationship in the reality, transposition (of models).