Progrès et « urgence » de la modélisation en sociologie. du concept de « modèle générateur » et de sa mise en Œuvre

Abstract

L'article propose une analyse critique du concept de « modèle » et de la place que celui-ci devrait occuper en sociologie. La première partie discute la possibilité et l'intérêt de concevoir l'activité de modélisation théorique comme construction systématique de « modèles générateurs ». La deuxième partie discute trois différents types de « modèles » (les « modèles statistiques », les « modèles informatiques », et les « modèles mathématiques »), et elle s'interroge sur leurs capacités respectives de « mettre en oeuvre » un « modèle générateur ». Dans cette partie, l'article consacre une attention particulière à une forme récente de « modélisation informatique » (les systèmes multi-agents) : l'idée mise en avant est que ces « modèles » constituent l'outil le plus approprié pour formaliser et pour étudier des « modèles générateurs ». En guise de conclusion, l'article esquisse une manière possible d'intégrer les quatre acceptions mobilisées du concept de « modèle » afin de proposer un cadre de recherche cohérent apte à supporter un type de sociologie fondée sur une activité constante et systématique de « modélisation ». ABSTRACT.-The article proposes a critical analysis of the concept of « model » and of the place that it should hold in sociology. The first part discusses the possibility of, and the interest in conceiving theoretical modelling as a systematic construction of « generating models ». The second part deals with three various types of « models » (« statistical models », « computational models », and « mathematical models »), and assesses their respective capacities to implement a « generating model ». In this section, the article devotes a detailed attention to a recent form of « computational modelling » (the multi-agents systems) : the idea put forward is that these « models » constitute the most suitable tool for formalizing and studying « generating models ». By way of conclusion, the article outlines a possible way of integrating the four meanings of the concept of « model » mobilized in order to propose a coherent research framework able to support a « modelling-based sociology ».

Emergence in Agent-Based Computational Social Science

“…This diachronic dimension of social emergent has been underlined by Archer (1995) for whom social structures emerged in the past from actions of agents; continue to exert effects in the present. As a consequence the only pertinent concept of emergence is diachronic (see also Manzo 2007). As underlined previously, if the non-reducibility argument is problematic in a synchronic context (and by extension the related downward causation) this is not the case in a diachronic context.…”

Analytical Sociology and Agent-Based Modeling: Is Generative Sufficiency Sufficient?

“…In 2005, Hedström stated that ABM “is destined to become the dominant type of theoretical formalism in sociology” (p. 87), and in 2010 he presented it as an “essential tool for analytical sociology” (Hedström and Ylikoski 2010:62). Manzo (2007b, 2014b) also clearly defended the affinity between the theoretical agenda of AS and agent-based model techniques. This affinity seems indisputable and has been responsible for remarkable developments in the social sciences (some chapters of Manzo 2014a are proof of that).…”

“…A second problem refers to the theoretical convenience of this definition. Hedström himself proposed the category of “transformational mechanisms” (Hedström and Swedberg 1996, 1998), and Manzo (2007b:21) stated that the importance of those mechanisms in sociology is paramount, “owing to the fact that any outcome as far as society is concerned is reached through them.” It thus follows that any definition of mechanisms must include the process triggered by the set of cogs and wheels (i.e., the transformational mechanisms).…”

“…Because we do not have a simple aggregation phenomenon but an emerging effect, the generative power of a set of micro specifications must be proved. Many scholars consider ABM the most suitable tool for this purpose (Axtell 2000; Cederman 2005; Epstein 2006a, 2006b; Manzo 2007b). In ABM, agents are (or can be) heterogeneous, autonomous, physically conditioned by space, socially embedded in a network, and guided by realistic rules of behavior and decision processes.…”

Understanding the Coleman boat

“…This is an empirical question; the purpose of the arrow is to remind the researcher that it is important to look for these The idea of a loop is not entirely new. Some authors (Abell 2000;Manzo 2007;Abell, Felin and 6 Foss 2008) have attempted to represent the loop by a chain of boats (where each D is followed by A of the next boat). However, this does not fully solve the problem because the loops do not necessarily involve A and D facts.…”