Abstract

The claim that mechanisms are essential good science is widespread. I argue, however, that these claims are ambiguous in multiple ways. I sort out different version of the mechanism idea: (1) mechanisms that are horizontal —between cause and effect— and mechanisms that are vertical —they realize in lower-level terms causal properties—: and (2) different purposes or uses mechanisms may have. I then focus on the claim that various senses of mechanism are necessary for the confirmation of causal claims. The paper shows that mechanisms can be useful, essential, or harmful depending on context, using the now standard graphical causal structure framework. These conclusions also support the larger philosophy of science moral that methodological norms in science are often context specific and empirical, not a priori and universal.