Evolutionary biology has been enriched by theory, from simple models that test the soundness of verbal logic to fully-parameterized models that make quantitative predictions directly applicable to empirical systems. There are great misconceptions, however, about the goals of simple, “proof-of-concept” evolutionary models and the relationship that they have (or should have) to empirical systems.
In this article, Yaniv Brandvain, Sumit Dhole, Courtney Fitzpatrick, Emma Goldberg, Caitlin Stern, Jeremy Van Cleve, Justin Yeh and I explain the purpose of proof-of-concept models in evolutionary biology, and their relationship to empirical work.
Many important ideas in evolutionary biology were initially developed using verbal logic. Simple evolutionary models are then used to test this logic: converting verbal models to mathematics requires making all assumptions explicit, while analyzing mathematical models tests the underlying logic, analogously to the way in which experiments are used to falsify a hypothesis.
Biologists generally think of hypotheses being tested with data, but with proof-of-concept modeling, mathematical models themselves are tests of whether verbal hypotheses are sound. These models thus serve a very different function than do descriptive or statistical models, which indeed must be tested or parameterized with data. Proof-of-concept models are valuable even if they do not directly produce empirically-testable predictions or enable data analysis.
In the article linked above, we explain the function of proof-of-concept modeling in evolutionary biology, in the context of the broader functions of modeling in the biological sciences. We describe the relationship between this type of modeling and empirical data, which, although not used to "test" the models in a specific system, 1) provide the basis for realistic assumptions for models, and 2) in some cases, show whether broad patterns predicted by these models are observed.