Abstract

Winsberg's "handshaking" account of inter-model relations is a well-known theory of multiscale modeling in physical systems. Winsberg argues that relations among the component models in a multiscale modeling system are not related mereologically, but rather by empirically determined algorithms. I argue that while the handshaking account does demonstrate the existence of non-mereological relationships among component models, Winsberg does not attend to the different ways in which handshaking algorithms are developed. By overlooking the distinct strategies employed in different handshake models, Winsberg's account fails to capture the central feature of effective multiscale modeling practices, namely, how the dominant behaviors of the modeled systems vary across the different scales, and how this variation constrains the ways modelers can combine component models. Using Winsberg's example of nanoscale crack propagation, I distinguish two modes of handshaking and show how the different modes arise from the scale-dependent physics involved in each component model.