the prescribed rules of procedure are to begin with data, and by induction or abduction, to generate an explanatory hypothesis. (In general, it is a good idea to consider several distinct hypotheses, and choose that which seems most likely for further testing.) Then the reasoner assumes the truth of the hypothesis, and deduces further consequences that can be tested observationally or experimentally. That is, we begin with a bottom-up approach, studying the basic components of a system and attempting to generate a synthesis that describes and explains the system's structure and behavior. Once this synthesis is formalized as a hypothesis, we are in a position to take a top-down approach in which we can deduce further system properties, and make predictions about components and their interactions. These, in turn, are compared to empirical data. If there is a fit, we feel more confident in our hypothesis. If there is a misfit (contradiction), we need to change or adjust the hypothesis until we are able to find a fit.