Malfunctions

A persistent boast of the historical approach to functions is that function al properties are normative. The claim is that a token trait retains its fu nctional status even when it is defective, diseased, or damaged and consequ ently unable to perform the relevant task. This is because historical funct ional categories are defined in terms of some sort of historical success - success in natural selection, typically - which imposes a norm upon the per formance of descendent tokens. Descendents thus are 'supposed to' perform t he associated task even when they cannot. The conceit, then, is that malfun ctions are explicable in terms of historical success. The aim of this paper is to challenge this conceit. My thesis is that the historical approach to functions lacks the resources with which to account for the possibility of malfunctions. If functional types are defined in terms of historical succe ss, then tokens that lack the defining property due to defect, and tokens t hat have lost the defining property due to disease or damage, are excluded from the functional category. Historically based malfunctions, in consequen ce, are impossible. The historical approach is no better than its non-histo rical competitors in accounting for the presumed normativity of functional properties.