Tools or Toys?
On Specific Challenges for Modeling and the Epistemology of Models in the Social Sciences

Eckhart Arnold

1 Introduction
2 The role of models in science
3 Why computer simulations are merely models and not experiments
4 The epistemology of simulations at work: How simulations are used to study chemical reactions in the ribosome
5 How do models explain in the social sciences?
6 Common obstacles for modeling in the social sciences
7 Conclusions

1 Introduction

Models play a central role in many sciences. One can safely say that the construction, analysis, discussion and validation of models is the daily bread of most researchers in the natural sciences and engineering. With the development of simulation technology the scope of modeling has vastly increased and it has become even more obvious that models have a “life of their own” independent from theories.

However, there is quite a difference between the extent to which models are used in the natural sciences and in the social social sciences. While in the natural sciences and engineering models are a standard tool and their use is undisputed, it is in the social sciences only economics, where models are the standard form of articulating hypotheses or causal assumptions. In some branches of the social sciences, like history, models are usually not used at all. And in other branches the use of mathematical models depends on the particular school that one adheres to. In sociology and political sciences, for example, it is the rational choice school that makes the strongest use of models.

But even in those parts of the social sciences that use models there exists a considerable unclairity about their epistemic role. It often remains unclear what conclusions can be drawn from models and if and how they can tell us something about the world. The kind of discussion about the role of models in economics that was triggered by Robert Sugden's credible world account (Sugden 2000) would appear surprising to a natural scientist. And when Robert Sugden wonders that in economics “authors typically say very little about how their models relate to the real world” (Sugden 2009, p.\ 25) then he expresses an embarassement that is quite uncommon in the the natural sciences. With the possible exception of economics, it is not only the epistemic status of models that is under discussion in the social sciences, but often it is disputed whether models are of any use at all. The respective discussion blends into the controversies about the usefulness or uselessness of mathematical methods in the social sciences in general (Green/Shapiro 1994, Shapiro 2005).

Leaving the ideological question whether everything that happens in this world can meaningfully be rendered in mathematical terms aside, these facts about the social sciences raise the question why there is such an unclarity about the possible role and function of models. In its broadest outline the lesson that I believe can be learned from examining this question, can be summarized in four theses:

  1. The possible role and function of models in the social sciences is often unclear, because models in the social sciences face specific challenges which indeed limit the scope of their useful employment.
  2. Because of this, the epistemic role of models in the social sciences cannot properly be understood in analogy to the role of models in the natural sciences.
  3. An epistemology of models that takes into account these challenges can help us to better understand the possible role and function of models in the social sciences. It might also help us to figure out when and how models and computer simulations can be applied usefully (where “usefully” means: “in such a way that we can learn something about the world from them”).
  4. As a side effect this may also help us to understand difficulties that models face in the natural sciences in those situations where similar conditions hold as appear to be the standard case in the social sciences (e.g. insufficient background knowledge about the processes involved, strong measurement inaccuracies).

In this article I shall mainly be concerned with the epistemology of models and simulations as traditionally understood and how it relates to the specific challenges that models face in the social sciences.

I first state what I consider to be the standard concept of models in the philosophy of science, namely, that models are links between theory and empirical reality (Morgan/Morrison 1999, Winsberg 2003). In this context I also discuss the more controversial question what status is to be ascribed to computer simulations. My position is that 1) computer simulations are just models that run on the computer 2) they, therefore, raise more or less the same epistemological questions as models and 3) computer simulations are not experiments, although they may under certain conditions replace experiments. I illustrate this with an example of a computer simulation from chemistry. The example shows that the conception of models as links between theory and reality is by and large appropriate.

After that I go on to review some of the viewpoints on models in the social sciences. These are highly diverse, which emphasizes that there is a considerable unclarity about what role models play in the social sciences. More importantly, the conception of models as a link between theory and empirical phenomena captures at best a rare exceptional case in the social sciences. This naturally raises the question of what distinguishes the social sciences and why models, therefore, play a different and generally much less prominent role in these sciences. I run through a number of distinguishing features which are apt to explain this difference and point out their epistemological consequences. I conclude with some suggestions for adjusting the epistemology of models so as to better capture the role of models in the social sciences.

t g+ f @