Online research in philosophy

Research in many fields of biology has been extremely successful in decomposing biological mechanisms to discover their parts and operations. It often remains a significant challenge for scientists to recompose these mechanisms to understand how they function as wholes and interact with the environments around them. This is true of the eukaryotic cell. Although initially identified in nineteenth-century cell theory as the fundamental unit of organisms, researchers soon learned how to decompose it into its organelles and chemical constituents and have been highly successful in understanding how these carry out many operations important to life. The emphasis on decomposition is particularly evident in modern cell biology, which for the most part has viewed the cell as merely a locus of the mechanisms responsible for vital phenomena. The cell, however, is also an integrated system and for some explanatory purposes it is essential to recompose it and understand it as an organized whole. I illustrate both the virtues of decomposition (treating the cell as a locus) and recomposition (treating the cell as an object) with recent work on circadian rhythms. Circadian researchers have both identified critical intracellular operations that maintain endogenous oscillations and have also addressed the integration of cells into multicellular systems in which cells constitute units. Ó 2010 Elsevier Ltd. All rights reserved.

The mechanistic perspective has dominated biological disciplines such as biochemistry, physiology, cell and molecular biology, and neuroscience, especially during the 20th century. The primary strategy is reductionist: organisms are to be decomposed into component parts and operations at multiple levels. Researchers adopting this perspective have generated an enormous body of information about the mechanisms of life at scales ranging from the whole organism down to genetic and other molecular operations.

Two widely accepted assumptions within cognitive science are that (1) the goal is to understand the mechanisms responsible for cognitive performances and (2) computational modeling is a major tool for understanding these mechanisms. The particular approaches to computational modeling adopted in cognitive science, moreover, have significantly affected the way in which cognitive mechanisms are understood. Unable to employ some of the more common methods for conducting research on mechanisms, cognitive scientists’ guiding ideas about mechanism have developed in conjunction with their styles of modeling. In particular, mental operations often are conceptualized as comparable to the processes employed in classical symbolic AI or neural network models. These models, in turn, have been interpreted by some as themselves intelligent systems since they employ the same type of operations as does the mind. For this paper, what is significant about these approaches to modeling is that they are constructed specifically to account for behavior and are evaluated by how well they do so—not by independent evidence that they describe actual operations in mental mechanisms.

Neuroscience and cognitive science seek to explain behavioral regularities in terms of underlying mechanisms. An important element of a mechanistic explanation is a characterization of the operations of the parts of the mechanism. The challenge in characterizing such operations is illustrated by an example from the history of physiological chemistry in which some investigators tried to characterize the internal operations in the same terms as the overall physiological system while others appealed to elemental chemistry. In order for biochemistry to become successful, researchers had to identify a new level of operations involving operations over molecular groups. Existing attempts at mechanistic explanation of behavior are in a situation comparable to earlier approaches to physiological chemistry, drawing their inspiration either from overall psychology activities or from low-level neural processes. Successful mechanistic explanations of behavior require the discovery of the appropriate component operations. Such discovery is a daunting challenge but one on which success will be beneficial to both behavioral scientists and cognitive and neuroscientists.

Explanations in the life sciences frequently involve presenting a model of the mechanism taken to be responsible for a given phenomenon. Such explanations depart in numerous ways from nomological explanations commonly presented in philosophy of science. This paper focuses on three sorts of differences. First, scientists who develop mechanistic explanations are not limited to linguistic representations and logical inference; they frequently employ dia- grams to characterize mechanisms and simulations to reason about them. Thus, the epistemic resources for presenting mechanistic explanations are considerably richer than those suggested by a nomological framework. Second, the fact that mechanisms involve organized systems of component parts and operations provides direction to both the discovery and testing of mech- anistic explanations. Finally, models of mechanisms are developed for specific exemplars and are not represented in terms of universally quantified statements. Generalization involves investigating both the similarity of new exemplars to those already studied and the variations between them. Ó 2005 Elsevier Ltd. All rights reserved.

In the context of mechanistic explanation, reductionistic research pursues a decomposition of complex systems into their component parts and operations. Using research on the mechanisms responsible for circadian rhythms, I consider both the gains that have been made by discovering genes and proteins that figure in these intracellular oscillators and also highlight the increasingly recognized need to understand higher-level integration, both between cells in the central oscillator and between the central and peripheral oscillators. This history illustrates a common need to complement reductionistic inquiry with investigations at higher-levels. Unlike most other accounts of reduction, the mechanistic framework accommodates this complementary relationship between reductionistic and systems approaches.

Accounts of mechanistic explanation have emphasized the importance of looking down—decomposing a mechanism into its parts and operations. Using research on visual processing as an exemplar, I illustrate how productive such research has been. But once multiple components of a mechanism have been identified, researchers also need to figure out how it is organized—they must look around and determine how to recompose the mechanism. Although researchers often begin by trying to recompose the mechanism in terms of sequential operations, they frequently find that the components of a mechanism interact in complex ways involving positive and negative feedback and that the organization often exhibits highly interactive local networks linked by a few long-range connections (small-worlds organization) and power law distributions of connections. The mechanisms are themselves active systems that are perturbed by inputs but not set in motion by them. Researchers also need to look up —situate a mechanism in its context, which may be a larger mechanism that modulates its behavior. When looking down is combined with looking around and up, mechanistic research results in an integrated, multi-level perspective.

Arguments for the autonomy of psychology or other higher-level sciences have often taken the form of denying the possibility of reduction. The form of reduction most proponents and critics of the autonomy of psychology have in mind is theory reduction. Mechanistic explanations provide a different perspective. Mechanistic explanations are reductionist insofar as they appeal to lower-level entities—the component parts of a mechanism and their operations— to explain a phenomenon. However, unlike theory reductions, mechanistic explanations also recognize the fundamental role of organization in enabling mechanisms to engage their environments as units (as well as the role of yet higher-level structures in constraining such engagement). Especially when organization is non-linear, it can enable mechanisms to generate phenomena that are quite surprising given the operations of the components taken in isolation. Such organization must be discovered—it cannot simply be derived from knowledge of lower-level parts and their operations. Moreover, the organized environments in which mechanisms operate must also be discovered. It is typically the higher-level disciplines that have the tools for discovering the organization within and between mechanisms. Although these inquiries are constrained by the knowledge of the parts and operations constituting the mechanism, they make their own autonomous contribution to understanding how a mechanism actually behaves. Thus, mechanistic explanations provide a strong sense of autonomy for higher levels of organization and the inquiries addressing them even while recognizing the distinctive contributions of reductionistic research investigating the operations of the lower level components.

Explaining the complex dynamics exhibited in many biological mechanisms requires extending the recent philosophical treatment of mechanisms that emphasizes sequences of operations. To understand how nonsequentially organized mechanisms will behave, scientists often advance what we call dynamic mechanistic explanations. These begin with a decomposition of the mechanism into component parts and operations, using a variety of laboratory-based strategies. Crucially, the mechanism is then recomposed by means of computational models in which variables or terms in differential equations correspond to properties of its parts and operations. We provide two illustrations drawn from research on circadian rhythms. Once biologists identified some of the components of the molecular mechanism thought to be responsible for circadian rhythms, computational models were used to determine whether the proposed mechanisms could generate sustained oscillations. Modeling has become even more important as researchers have recognized that the oscillations generated in individual neurons are synchronized within networks; we describe models being employed to assess how different possible network architectures could produce the observed synchronized activity.

Abstract While agreeing that dynamical models play a major role in cognitive science, we reject Stepp, Chemero, and Turvey's contention that they constitute an alternative to mechanistic explanations. We review several problems dynamical models face as putative explanations when they are not grounded in mechanisms. Further, we argue that the opposition of dynamical models and mechanisms is a false one and that those dynamical models that characterize the operations of mechanisms overcome these problems. By briefly considering examples involving the generation of action potentials and circadian rhythms, we show how decomposing a mechanism and modeling its dynamics are complementary endeavors.