Reviewed by
Frederic Amblard
IRIT - Université de Toulouse 1 Capitole
The book New Frontiers in the Study of Social Phenomena: Cognition, Complexity, Adaptation, published by Springer in 2016, was edited by Federico Cecconi, researcher from the Laboratory of Agent-based Social Simulation (LABSS) at the Institute of Cognitive Sciences and Technologies, CNR, Rome, who works in particular on the agent-based modelling of economic phenomena.
The book is divided into eleven chapters mostly authored by researchers issued from the LABSS. At a first glance, one could wonder if the new frontiers in question are not those between the LABSS and the outside world. However, this is not the case. But just like most books built on chapters written by different authors, it suffers from a lack of global coherence and articulation between chapters; it gives a good overview of what could be presented by an Italian school in the field of agent-based social simulation.
The book starts with a very nice introduction written by Rosaria Conte, which presents the history of Computational Social Science (CSS) and its positioning towards different disciplines (mainly physics and computer science) all along its emergence. The second part of the chapter is an insightful description of today’s challenges related to Social Science (in particular Big Data) and the possible integration of CSS within the framework of existing tools and models to better understand social phenomena.
Then, the book is divided a bit artificially in two parts: new theories and new applications. The architecture of these parts is rather misleading. On the one hand, even if the chapters taken separately are overall quite interesting, grouping them under the label of “new theories” does not reflect what the reader may expect. On the other hand, either based on theories or applications, several chapters could just as well be integrated with the other part.
The first part (let’s call it “Part 1” rather than “New Theories”) consists of five chapters. The chapter 2, written by Marco Campennì, brilliantly presents the state of the art concerning cooperation within an animal society, strongly influenced by a game-theory perspective. The author generalizes such a perspective towards the study of social ecological systems and so the earth system dynamics. He points out the important role that agent-based modelling could play to investigate cooperation theories, especially regarding the controversy around reciprocal cooperation. Chapter 3 is quite redundant with the former one, although it focuses on human societies. This chapter is linked to the previous one but extends the perspective to examine institutions and norms. Chapter 4 refers to an application rather than a theory and is dedicated to extorsion racket systems with well-known work on the articulation between legal and social norms, the immergence and salience of norms and their application. Chapter 5 focuses on tax evasion and provides a good and updated overview of models and approaches in this field, ranging from classical approaches (following an utilitarian point of view), laboratory experiments and agent-based social simulation, seen as a third way to study tax evasion. The last chapter of this part, chapter 6, shows a model concerning the reputation-based selection of partners and cooperation in groups, and in particular the effect of network structure (small-world vs. bipartite graphs) over the emergence of cooperation.
The second part is divided in five chapters as well. Chapter 7 is a very promising interdisciplinary work on the modelling of sport crowd’s behavior, built from an articulation between sociology (concerning micro-interactions), ontology and computer vision (for the analysis of stadium videos). The methodology is pragmatic and straightforward and could be inspirational for similar studies referring to the data integration onto agent-based simulation approaches. Chapter 8 presents a short bibliographic study based on Google Scholar concerning the evolution of agent-based simulation compared to other methods (system dynamics, game theory, etc.) in different fields (physics, sociology, economics, etc.). Even though quite interesting, it seems other available studies on this subject are more comprehensive. Chapter 9 is rather theoretically oriented as it is focused on the quality evaluation of collective decision and on judgement aggregation, thereby more oriented towards decision theory. Chapter 10 is a good contribution on research on crime diffusion. It presents a comprehensive state of the art on the subject and focuses in particular on crime-learning through imitation, either rational or social. The last chapter, chapter 11, is not directly linked to agent-based modelling but concerns data analysis of news, using natural language processing approaches, for stock prices prediction.
To sum up, the book has certain weaknesses, i.e., the lack of a clear overall structure, which could be solved by at least an editorial introduction explaining the plan of the book. This said, if taken separately, many chapters are very well written and interesting. Moreover, all considered, the book provides a good overview of the Italian tradition of research on ABM especially focused on cognitive approaches, cooperation and normative behavior. These are topics that have been well covered by the LABSS research for quite a long time now.
Return to Contents
of this issue