Finally, let us consider the implications of ‘exploration’ for our view of the rela- tion between models and theories. The semantic view helps us see one important fact about models: a collection of models of the same phenomenon (e.g. emer- gence of money) can be considered as supplying the resources for explaining sin- gular cases if they form a more or less consistent body, that is, if the models that emphasise different explanatory factors are more or less complementary. But the semantic view holds that models and theories are related in a certain way: models describe possible abstract systems and theory is a collection of these models.
Our examination of the models of unintended social phenomena and their ex- ploration suggests that the relation between models and theories may be more complex. Models may derive from existing theories, from observation, as well as from other models. For this reason, they may be characterised as somewhat independent objects, or tools that mediate between theory and the real world, and between models and other models.
A recent defence of the view that models are mediators is presented in Models as Mediators (Morgan and Morrison 1999). It argues that models may be seen as ‘autonomous agents’ that serve as ‘instruments of investigation’ (Morgan and Morrison 1999: 10). The main idea is that the construction of models is a complex matter and that models do not have a defined relationship with theory or data.46 For example, by studying economic models, Boumans (1999) argues that these models use a rich pool of conceptual resources.
On one hand, model construction is a complex process and scientists use varie- ties of resources to be able to reason about the phenomenon under question. On the other hand, models are usually partial and cannot alone account for particular cases. Explaining particular exemplifications of the explanandum phenomenon requires the use of other resources, such as other models of the same phenom- enon or of similar phenomena. For this reason, abstract models should not be evaluated in isolation from the rest of the body of thought developed on a certain topic. Models serve different purposes. Some of them provide partial potential theoretical explanations, some others test other models, some help us conduct experiments, some help us measure and some help us explore the implications of our thoughts. But if considered all together, distinct but related models of a certain phenomenon help us explain its particular exemplifications. As a collection, they form something like an incomplete theory, serving the conceptual resources to provide singular explanations. For example, Schelling’s model of residential seg- regation, its reconsiderations, and the other models and explanations of segrega- tion that emphasise organised action and economic factors may be considered as forming an incomplete, partial theory or a meta-model of segregation. This ‘theory’ supplies the conceptual resources of a social scientist when confronted with particular cases. Similarly, Menger’s verbal model, its formal reconsidera- tions and historical / anthropological accounts of money may be considered as forming an ‘incomplete theory’ of the emergence of money.
Considering models as mediators also helps us to understand the nature of models of unintended social phenomena and of many models in economics. Models constitute the toolbox of many economists and, hence, models that do not explain any particular case are useful: they prepare the base for explaining particular cases and alert us to certain types of mechanisms. Any economist will be familiar with the fact that most of the theoretical journals contain what seem to be extremely abstract models with little or no explanation of particular real-life situations. An economist will be able to publish a model which is a modification of another (known) model, if he or she can show that under such and such (differ- ent) conditions such and such (different) results are obtained, or if he or she could prove that the same results hold under a wide variety of conditions. Many people, including philosophers, have criticised this practice. However, the fact that no particular singular explanation is provided cannot immediately be used against these models, because these models are usually parts of the preparation process prior to a singular explanation. Many of these models, as exemplified by models of unintended social phenomena, provide partial potential theoretical explana- tions that alert us to certain ways in which the real world may be working. Such models are mediators between our theories, hypotheses and explanations. They are further explored to have a better understanding of the real world.
Source: Aydinonat N. Emrah (2008), The Invisible Hand in Economics: How Economists Explain Unintended Social Consequences, Routledge; 1st edition.