Many philosophers have explored the extensive use of non-universal generalizations in different sciences for inductive and explanatory purposes, analyzing properties such as how widely a generalization holds in space and time. We concentrate on developmental biology to distinguish and characterize two kinds of scientific generalizations—mechanisms and principles—that correspond to different explanatory aims. Our analysis shows why each kind of generalization is sought in a research context, thereby accounting for how the practices of inquiry are structured. It also diagnoses problematic assumptions in prior discussions, such as abstraction always being correlated positively with generalizations of wide scope.

It is common to distinguish classificatory, physical, and formal unification. Of these, only physical unification seems to have anything to do with explanation and hence understanding. In this paper, I argue that that view is incorrect. Classificatory and formal unification facilitate understanding. Moreover, I argue that theories that physically unify do not necessarily facilitate understanding better than non-unified theories.

New mechanists forward an influential account of mechanisms in which entities (or parts) and their activities are organized so as to produce the phenomenon that calls out for explanation; and to explain is to describe that mechanism. However, critics charge that new mechanists have not provided a standard for identifying and individuating parts that blocks gerrymandered parts. To remedy this, I defend a carving principle that justifies the standard parts that are included as components of mechanistic explanations. My account grounds good parthood in robust explanatory relations I call the explanatory mosaic of science.

Philosophical accounts of biological mechanisms have only recently attended to the crucial role free energy plays in enabling the operation of mechanisms and have not addressed how scientists discover the role of free energy in the operation of biological mechanisms. To do so, I examine research on two mechanisms—the myosin motor in muscle contraction and the cyanobacterial circadian clock. I describe the discovery process in which researchers compare static images to determine the conformation changes in proteins produced by ATP hydrolysis and infer how these conformation changes generate forces that result in the phenomenon produced by the mechanism.