Robustness, solidity, and multiple determinations

Abstract

William Wimsatt’s 1981 essay ‘‘Robustness, Reliability, and Overdetermination’’ was a seminal contribution to philosophical discussions about robustness. In it, Wimsatt offered a synthesis of available concepts and ideas related to robustness. According to Wimsatt, robustness can be understood as invariance of a phenomenon, object, or result under multiple independent determinations. Since the early 1980s, philosophical discussions about robustness—often inspired by Wimsatt—have ranged over a number of issues, including the analysis of the structure of robust evidence and the question of whether robustness is a criterion for attribution of reality. Recent conceptual analyses have yielded distinctions among three concepts of robustness within the broad Wimsattian notion with regard to the nature of the phenomenon, object, or result that is considered robust (e.g., Woodward 2006; Calcott 2011). According to this threefold distinction, robustness concepts may pertain to models and model outcomes, experimental results (‘‘robust detection’’), or complex biological, technical, and epistemic systems and causal relations ‘‘in the world.’’ Most discussions in philosophy of science about robustness have been concerned with models and model outcomes. The epistemic force of arguments pertaining to the robustness of model outcomes has appeared particularly problematic, and doubts have been raised specifically about the confirmatory role of robustness (see, e.g., Orzack and Sober 1993, and more recently Odenbaugh and Alexandrova 2011). The volume Characterizing the Robustness of Science, a collection of essays stemming from a conference on robustness in Nancy in 2008, deals with one aspect of the discussion about robustness that has received comparatively little attention: the functions and uses of robustness in actual scientific practice. The collection opens