How the Reductionist Should Respond to the Multiscale Argument, and What This Tells Us About Levels

Recent literature has raised what I'll call the 'multiscale argument' against reduction (see e.g. Batterman (2013), Wilson (2017), Bursten (2018)). These authors observe that numerous successful scientific models appeal to features and properties from a wide range of spatial/temporal scales. This is taken to undermine views that the world is sharply divided into distinct levels, roughly corresponding to different scales, and that each higher level is reducible to the next lowest level. While the multiscale argument does undermine a naive conception of levels and reduction, in this paper I argue that alternative views of reduction and levels can withstand this argument. After articulating the multiscale argument in more detail, I show that this does not undermine a version of reduction that accepts methodological pluralism in science, yet maintains that the adequacy of any model can be explained by appeal to details at smaller scales. I go on to discuss a case study – dislocations in steel –e used by Batterman and Wilson in defence of the multiscale argument. I argue that the version of reduction advocated above is available in this context. I conclude by arguing that, in the face of the multiscale argument, levels are either everywhere or nowhere.