Modeling without models

Modeling is an important scientific practice, yet it raises significant philosophical puzzles. Models are typically idealized, and they are often explored via imaginative engagement and at a certain “distance” from empirical reality. These features raise questions such as what models are and how they relate to the world. A number of recent accounts answer these questions in terms of indirect representation and analysis. Such views treat the model as a bona fide object (“the model system”), specified by the modeler and used to represent and reason about some portion of the concrete empirical world (“the target system”). On some indirect views, model systems are abstract entities, such as mathematical structures, while on other views they are concrete hypothetical things, akin to fictional characters. Here I assess these views and offer a novel account of models. I argue that regarding models as abstracta results in some significant tensions with the practice of modeling, especially in areas where non-mathematical models are common. On the other hand, viewing models as concrete hypotheticals raises difficult questions about model-world relations. The view I argue for treats models as direct, albeit simplified, representations of targets in the world. I close by suggesting a treatment of model-world relations that draws on recent work by Stephen Yablo concerning the notion of partial truth.