We investigate the context of discovery of two significant achievements of twentieth century biochemistry: the chemiosmotic mechanism of oxidative phosphorylation and the dark reaction of photosynthesis. The pursuit of these problems involved discovery strategies such as the transfer, recombination and reversal of previous causal and mechanistic knowledge in biochemistry. We study the operation and scope of these strategies by careful historical analysis, reaching a number of systematic conclusions: even basic strategies can illuminate “hard cases” of scientific discovery that go far (...) beyond simple extrapolation or analogy; the causal–mechanistic approach to discovery permits a middle course between the extremes of a completely substrate-neutral and a completely domain-specific view of scientific discovery; the existing literature on mechanism discovery underemphasizes the role of combinatorial approaches in defining and exploring search spaces of possible problem solutions; there is a subtle interplay between a fine-grained mechanistic and a more coarse-grained causal level of analysis, and both are needed to make discovery processes intelligible. (shrink)

Attempts to solve the puzzling problem of oxidative phosphorylation led to four very different hypotheses each of which suggested a different view of the ATP synthase, the phosphorylating enzyme. During the 1960s and 1970s evidence began to accumulate which rendered Peter Mitchell’s chemiosmotic hypothesis, the novel part of which was the proton translocating ATP synthase (ATPase), a plausible explanation. The conformational hypothesis of Paul Boyer implied an enzyme where ATP synthesis was driven by the energy of conformational changes in the (...) respiratory proteins. This was finally abandoned as an explanation of the overall process. Nevertheless the conformational understanding of the enzyme became an acceptable proposal during the early 1970s and eventually led Boyer to a view of the enzyme that incorporated both hypotheses. The correspondence between Mitchell and Boyer, both Nobel laureates, exposes their different approaches to both this enzyme and to the hypotheses of oxidative phosphorylation and illuminates a key step in the development of bioenergetics. In particular Boyer was suspicious of proton gradients, because he could not envisage a chemical mechanism for the synthesis of ATP, while Mitchell distrusted conformational arguments because he believed the proton must act vectorially at the active site of the enzyme. This resulted in two different views of the mechanisms operating in this enzyme. Ultimately while Boyer was able to marry the two approaches, Mitchell retained his insistence on the role of the proton at the active site and was thus unable to give significance to Boyer’s conformational ideas. The underlying issues in this debate are discussed particularly with reference to the differing styles of Boyer and Mitchell and the influence of molecular biology, especially the development of protein technology. (shrink)

Cell biologists, including those seeking molecular mechanistic explanations of cellular phenomena, frequently rely on experimental strategies focused on identifying the cellular context relevant to their investigations. We suggest that such practices can be understood as a guided decomposition strategy, where molecular explanations of phenomena are defined in relation to natural contextual (cell) boundaries. This “top-down” strategy contrasts with “bottom-up” reductionist approaches where well-defined molecular structures and activities are orphaned by their displacement from actual biological functions. We focus on the central (...) role of microscopic imaging in cell biology to uncover possible constraints on the system. We show how identified constraints are used heuristically to limit possible mechanistic explanations to those that are biologically meaningful. Historical examples of this process described here include discovery of the mechanism of oxidative phosphorylation in mitochondria, molecular explanation of the first steps in protein secretion, and identification of molecular motors. We suggest that these instances are examples of a form of downward causation or, more specifically, constraining relations, where higher-level structures and variables delimit and enable lower-level system states. The guided decomposition strategy in our historical cases illustrates the irreducibility of experimentally identified constraints in explaining biological activities of cells. Rather than viewing decomposition and recomposition as separate epistemic activities, we contend that they need to be iteratively integrated to account for the ontological complexity of multi-level systems. (shrink)