Indistinct conceptions of ‘hypothesis' account for much confusion about the epistemology of contemporary biological research. Systems biology—particularly the omics disciplines (genomics, proteomics, metabolomics and the like)—challenges the received notion that all science is hypothesis-driven. Closer examination reveals multiple levels at which hypotheses or alternate-drivers can operate. Elsewhere I have suggested that contemporary biological research can be broadly categorized as either hypothesis-driven or non-hypothesis-driven. The latter is comprised of at least two subcategories, system-driven and data-driven. In system-driven research the complexity of a biological system is addressed directly; in data-driven research a collection of data already assembled is interrogated to find new information ‘hidden' there. Omics research is system-driven research, ostensibly without a hypothesis.

Yet, when asked, the omics researcher will contend that s/he has a hypothesis. Typically it is something like “there are numerous unidentified proteins in this system” or “multiple genes contribute to this cellular process”. I propose that biological research involves a hierarchy of ‘hypotheses'. First, as Kuhnian normal research, the research enterprise has a hypothesis locating it within its research paradigm. Secondly, it features what I call a hunch hypothesis: a broad motivation of the research project. Finally, there is an operational level immediately proximate to experimental design. For hypothesis-driven research, a detailed hypothesis governs experimental design. For system-driven research, the system being investigated directly informs experimental design. The character of ‘hypothesis' in biology has been elusive because multiple layers of hypothesis are at work; however, what really counts is what governs experimental design.