Abstract
Similarity accounts of scientific representation have been a matter of controversy. After reviewing some of the objections, I introduce the notion of pragmatically and cognitively constrained stipulation which takes precedence over similarity judgments, and which (partially) characterizes the way in which the representational relationship is established.
While similarity does not ground representation, it nevertheless plays a major role in scientific reasoning, and in some cases, it also feeds into representational considerations. To illustrate, I will discuss number of examples, ranging from biometrics to model organisms.
In doing so I will make several distinctions regarding the multifarious roles of similarity judgments. By construing representation along the lines of pragmatically and cognitively constrained stipulation we dodge the usual objections against similarity accounts.
Similarity, then, is, for the most part, a success term having to do with representational accuracy rather than with representation as such. Resituating the notion of similarity in the suggested way allows us to maintain its importance and to remain faithful to the scientific practice.