Individual-Based Modelling and Ecology

Order this book

Reviewed by Marco Janssen
School of Human Evolution and Social Change & Department of Computer Science and Engineering
Arizona State University

Individual-based modeling (IBM) has been applied in ecology since the 1970s/1980s. Early advocates of IBM stated that the new methodology is revolutionary and would be the right method for ecology (e.g. Huston et al. 1988). The last 20 years a lot of experience is gained on the potential and limitations of IBM, and this book addresses the lessons learned and challenges left from these experiences.

This book is a very interesting volume for scholars in social simulation who started their journey somewhat later than the IBM scholars, but experience similar challenges. Grimm and Railsback provide a wonderful volume of examples of practices and approaches of doing rigorous research using IBM which might be helpful for the social simulation community in order to speed up the learning curve.

The book consists of four parts. The first part is on "Modeling" where the authors discuss a brief history of IBM, heuristics for modeling and the modeling cycle, and close with a discussion on pattern-oriented modeling. Their focus on pattern-oriented modeling is fueled by the strength of IBM, a tool to understand how patterns may form.

In part 2, Grimm and Railsback purposely formulate a theoretical framework Individual-Based Ecology (IBE). IBE "formulates theories of the adaptive behaviors of individuals and tests the theories by seeing how well they reproduce, in an IBM, patterns observed at the system level." (italics in the original) (p. 54). Although a complex adaptive system perspective is used as a starting point for this framework, they stress the importance to formulate alternative hypotheses in order to test alternative theories by model analysis and comparison with empirical data. They stress the importance to work together with scholars doing field observation and laboratory experiments.

Ten different concepts are high lighted in order to design IBMs: emergence, adaptive traits, fitness, prediction, interaction, sensing, stochasticity, collectives, scheduling, and observation. Then they discuss a large number of example models and highlight how these examples address the above ten concepts.

In part 3, the engine room, the authors discuss practical issues in formulating IBMs, such as including space, stochasticity and adaptation. They also discuss a number of lessons learned from implementing IBMs in software, and stress the importance of working with computer scientists and software engineers, the development of common programming frameworks and the verification process of the software. Rightly, they point to the most time-consuming part of the modeling process, the analysis of the models, and discuss a number of ways IBMs have been analyzed by using statistical techniques, performing sensitivity and robustness analysis of the simulated patterns, and ultimately predicting behavior of situations on which the model is not parameterized. Interestingly, the authors conclude part 3 of the book on communication of IBMs. Frequently, not enough space is available in journals to discuss the models in detail, and they suggest to have electronic appendices or reports with detailed model descriptions as well as documentation on model testing and verification. Project websites might be a useful way to have full documentation of the model and implementation. Such a practice is also something journal editors and reviewers need to enforce in order to derive a norm that scholars have a decent documentation of their work that can be replicated and used by others.

The book closes with a discussion on the relation of IBE with the use of analytical models. Rightly they stress that analytical models still have an important role to play, for example by formulating simplified analytical models of IBMs that can be analyzed more thoroughly and communicated more easily. The book ends with a vision how IBM and IBE might be used in practice in the longer term.

Although the book is primarily focused on ecology, it is a very valuable volume for the readers of JASSS since it discuss challenges we social simulators face too, and provide potential solutions. For example, social simulation scholars may develop a common theoretical framework in order to stimulate a research agenda where findings of simulation models contribute to a common body of knowledge, and translate this knowledge to theories in the social sciences. IBE is not aimed to cover all topics where scholars apply IBM, but is more a core theoretical framework to enable knowledge exchange. Given the larger variety of theories within the social sciences compared to ecology, this might be an ambitious challenge, but worthwhile to pursue.

Another example is to develop more rigorous practices on calibration, verification, and documentation, which need to be enforced by journals where social simulators regularly publish in. One of the related activities within the social simulation community is the model to model workshop series, which illustrate the importance to increase the standards of model implementation, testing and documentation (Hales et al. 2003). This might be derived by developing a model library of simulation models where scholars need to deposit their models and documentation when they like to publish a simulation model.

A healthy attitude of Grimm and Railsback is to see IBM as a complementary research approach to existing traditional approaches, like traditional population ecology in their world. IBM stimulates a more interdisciplinary approach of ecology, like social simulation might be in the social sciences. However, they do not claim superiority of their "hammer" like some of the early advocates of IBM, as some of the current advocates of social simulation. Social simulation is in its infancy and it is encouraging to see our older nephew IBM overcoming the challenges of its early years and enter a new phase where it may blossom. I recommend social simulation scholars to study this book carefully so that we can speed up our learning process.