Rich as a class of models, and equally prolific as a modelling philosophy, the numerous approaches which make up the class of dynamic social impact models emphasize a deep paradigmatic shift toward modelling political attitude change as a dynamic, context‐sensitive, path‐dependent process. In conceptual terms, all this makes more and more obvious a shift of focus from empirical toward generative scenarios of attitude change able to provide for the typical experimental settings achieved by means of artificial life technologies. At the beginning of the 1990s, political attitude research ran around the clock to keep pace with the amazing advances in cellular automata, social networks and agent‐based systems modelling methodologies which were coming of age. Modelling research seemed fundamentally attracted by the idea of generating scenarios of change which were much richer than the empirical experimental settings and at the same time much more complex. ‘Would‐Be Worlds’, the concept defined by John Casti (1997) introduced a modelling paradigm able to provide for the evaluation of the potential developments of an artificial system (model) by generating and examining simulation scenarios able to reproduce in the model the dynamics of real‐world change phenomena. Casti’s work reveals a philosophy about simulation modelling which emphasizes the generative characteristic of a model as a means to achieve in virtual experimental media (e.g., simulation runs of a computational model) path‐dependent outcomes. Such modelling techniques were soon afterwards created with other artificial life technologies as well, like multi‐agent distributed systems (MAS) and agent‐based systems (ABS). This philosophy of simulation replaced the classic computer simulation idea of testing a model against (empirically provided) data by a constructivist idea of achieving model dynamics and complexity by employing a generative computational architecture denoted ever since as the ‘bottom‐up’ architecture. Bottom‐up artificial systems are designed to evolve as the multiple artificial agents inside them repeatedly interact with each other. Based on the methodological individualism idea that individual initiative is the true engine of social development, such systems achieve ‘upward causation’ (macro‐level emergence) by introducing an interaction rule at the micro level. The bottom‐up paradigm has been intensively employed in artificial society simulations (Epstein and Axtell, 1996), and in the study of the dynamics of culture dissemination and change (Axelrod, 1997). These two modelling approaches have fundamentally revolutionized modelling research as they have decisively influenced the subsequent orientation of political attitude modelling research in both North American and European schools of thought. The agent‐based systems provide the appropriate framework for revising the computational and simulation models developed earlier by McPhee, Latané, Abelson and their collaborators. It is this technology which makes possible the revision of the social persuasion scenarios of political attitude change in networks of interpersonal relationships and dyadic interactions. The new models elaborated by Robert Huckfeldt, Paul Johnson, John Sprague and their numerous young collaborators bring to the front old conceptual and technical problems: universal agreement and the survival of diversity in democratic societies, the dissemination of culture and the dynamics of culture change in homogeneous/heterogenous societies of artificial agents, and the collective action and the role of elites/activists in enhancing political protests.
Part IV introduces the main modelling approaches which dominated political attitude research during the 1990s. They focused on the revision of the classic theories about the social context, the role of interpersonal relationships in social and political persuasion and the complexity‐based modelling of self‐organizing social systems (artificial societies). These modelling approaches include: Chapter 7 briefly describes Axelrod’s Culture Dissemination Model (1997). The subsequent chapters present agent‐based models which either revise or extend Axelrod’s ‘thought exercise’ about culture change in self‐organizing social systems. Chapter 8 introduces the Diversity Survival Model (Huckfeldt et al., 2004) and Chapter 9 presents two collective action modelling approaches concerned with political attitude formation and change: the Political Contagion Model (Johnson, 1999) and the Cooperative Social Action Model (Johnson and Brichoux, 2002). The selected models bring forth in various ways the issue of social persuasion by emphasizing that agent‐based as well as social network technologies help revisit interpersonal networks and dyadic interactions as the basic context of individual interactions. The Diversity Survival Model (Huckfeldt et al., 2004) thus reminds the reader about the classic opposition between the typical Columbia modelling approach based on dyadic interactions as a means of preserving the contextual dynamics of social influence processes in the small worlds, on the one hand, and on the other hand, the type of modelling approach, mainly advocated by the Michigan group, which emphasizes the atomization of political opinion and attitude once the individual is considered an autonomous decision‐maker among many others at the national level (Huckfeldt, 2014).