Epistemic Opacity

This chapter shows that—and how—simulation models are epistemically opaque. Nevertheless, it is argued, simulation models can provide a means to control dynamics. Researchers can employ a series of iterated (experimental) runs of the model and can learn to orient themselves within the model—even if the dynamics of the simulation remain (at least partly) opaque. Admittedly, such an acquaintance with the model falls short of the high epistemic standards usually ascribed to mathematical models. This lower standard is still sufficient, however, when the aim is controlled intervention in technological contexts. On the other hand, opacity has to be accepted if the option for control is to remain in any way open. This chapter closes by discussing whether epistemic opacity restricts simulation-based science to a pragmatic—“weak”—version of scientific understanding.

Validation, Holism, and the Limits of Analysis

This chapter has two parts. The first part boils down to the thesis that simulations simply extend or amplify the validation problem, because they include steps that are not part of traditional mathematical modeling—but they arguably do not pose a conceptually new type of validation problem. The second part deals with the problem of holism that emerges when complex interactions together with a modular design govern how the model behaves. Simulations then gain a particular twist with dramatic consequences. Modularity is the very basis for handling complex systems, but it erodes for reasons inherent to simulation modeling. In a way, simulations undermine their own working basis, and as a consequence, the problem of holism emerges to reveal the limits of analysis.

Solution or Imitation?

This chapter distinguishes two fundamental but opposing conceptions of simulation. The first conception conceives simulations as numerical solutions of equations. The second approach does not involve the concept of solution, but takes simulation as the imitation of the behavior of a complex system by a computer model. This chapter claims that simulation modeling combines both conceptions. Large parts of the sciences involve a compromise (in one way or another) between two diverging forces. Theoretical understanding and epistemic quality stand on the one side; applicability and tractability on the other. What is interesting about simulation is the way in which a balance is achieved—that is, how the conflicting types are combined. The chapter analyzes the relationship between the simulation pioneers John von Neumann, who advocated the solution, and Norbert Wiener, who advocated the imitation concept.

Introduction

The chapter provides a brief overview of the history of simulation modeling and of philosophical accounts dealing with simulation. Computer and simulation modeling, it is stated, do form a new exploratory and iterative type of mathematical modeling. Four aspects are introduced: experiment and artificiality, visualization, plasticity, and epistemic opacity. The key thesis is that the novelty of simulation modeling rests on how these aspects are combined into a combinatorial style of reasoning. The computer as an instrument does not only speed up calculations but also channels mathematical modeling. This is exerting transformational power on central concepts like solution, validation, and the real—instrumental divide.

Visualization and Interaction

This chapter addresses the role of visualization in the process of simulation modeling. The claim is as follows: from a methodological perspective, visualization supports the exploratory and iterative mode of modeling; from an epistemic perspective, it assigns a special role to judgment. Of central importance for both perspectives is that visualizations offer opportunities to interact with models. Researchers may vary representations on screen or highlight particular aspects of processes. Such interactions combine the impressive human powers of comprehension in the visual dimension with the computational capacities of digital machines. A visual presentation fosters the interplay between experimental approach and instrumentalist assumptions discussed in chapter 1. Visualization can be decisive when adapting such assumptions in line with the performance of the simulated system. Studying galaxy formation and predicting hurricanes serve as case studies.

Novelty and Reality

This last chapter summarizes the findings presented in earlier chapters, and the major part of the chapter presents an outlook on critical challenges for a philosophy of simulation. One of these challenges is to take into account the science–technology nexus. Another is to account for the relationship between human activity and reality that results from this nexus. This makes it necessary to rethink the instrumentalism versus realism divide in the philosophy of science. It is argued that simulation evades the stalemate of this divide. Finally, this chapter provides an outlook on rationality. To the extent that mathematization forms a constitutive part of modern scientific rationality, the philosophy of simulation faces the question whether and how the new type of mathematical modeling affects and potentially changes the conception of rationality.

A New Type of Mathematical Modeling

This chapter synthesizes the findings of chapters 1 through 4. The phenomena investigated there correlate positively; that is, they interact in such a way that one strengthens the other. As a result, they converge to form a new type of mathematical modeling. Simulation modeling is distinguished by its exploratory and iterative mode that presents a multidimensional picture. The properties spanning the dimensions can occur to different degrees; that is, to what extent the characteristic properties are realized depends on the particular classes of simulation strategies and the concrete applications. From a systematic perspective, it is argued, simulation synthesizes theoretical and technological elements. Hence, simulation-based sciences are application oriented and show a greater proximity to engineering.

Plasticity

This chapter introduces the property of plasticity. A model has this property when its dynamic behavior can vary over a broad spectrum while its structure remains unchanged. Simulations present a special case, because plasticity does not appear as a shortcoming that needs to be compensated for but, rather, as a systematic pillar of modeling. The structure of the model is designed to include a measure of plasticity that creates room to maneuver. It is argued that the model dynamics can be called structurally underdetermined. Only the process of specification—which involves the activities of experimenting and visualizing—fully determines the model behavior. In short, plasticity and exploratory modeling complement one another. Various illustrations from neural networks, finite differences, and cellular automate are discussed.

Experiment and Artificiality

This chapter works out in what way or ways experimentation is fitted into the process of simulation modeling: how much do numerical experiments contribute to making simulation modeling a special type of mathematical modeling? The main point of the chapter is that the discreteness of the computer makes it necessary to perform repeated experimental adjustments throughout the modeling process. Experimentation and modeling, it is argued, build an explorative cooperation. Experimental practice (in the ordinary sense) is bound up with adjustments such as calibrating instruments. With simulation, they become essential to mathematical modeling, as well. Atmospheric circulation models are discussed as an illustrating case.