In lieu of an abstract, here is a brief excerpt of the content:

CHAPTER 9 "Emergent" Phenomena in Interconnected Networks 1. INTRODUCTION: DISPENSING WITH MODULES The more complex localizationist explanations we examined in Part III are still recognizably mechanistic. Tasks involved in performing a function are divided between components, and system behavior is explained by showing how it is accomplished through the combined performance ofthe component tasks. Although one might prefer explanations in which the component tasks can be thought of as following a linear, sequential order, so that the contributions of each component can be examined separately, natural systems are not always organized in such a manner. Component tasks are often dependent on one another, so we cannot understand the operation of the system by imposing a linear order on it. Cyclic rather than linear organization occurs when the activity of any given component is dependent on a variety ofother components that, in turn, depend on it. In integrated systems, the explanation ofthe behavior ofthe whole system depends in a nonlinear way on the activities ofthe components and on the modes of interaction found within the system. To the extent that organization is important in affecting system behavior , a system is nondecomposable. As we discussed in Chapter 2, there is a continuum of cases. At one end are simply decomposable systems for which the major explanatory task is to identify the components and understand their behavior. These include the kinds ofcases considered in Chapter 4, in which a single part is held to be responsible for the behavior ofthe whole system. Toward the other end of the continuum lie cases of integrated systems in which the behavior of the system is largely due to the interaction of the components. In all of these cases, the operations of the components can, nonetheless, be understood in terms of the operations performed by the whole. Conversely, the behavior of the whole is explained in terms of the behaviors of component parts. There are other systems, yet farther out on the continuum, in which localization and decomposition appear to be hopeless, or even misguided. The hallmark of these cases is that, given a principled structural analysis, the activities of the parts seem to be different in kind from, and so far simpler than, those performed by the whole. The parts can be so simple, in fact, that they do "Emergent" Phenomena . 203 not seem individually to contribute anything of interest to understanding the behavior of the whole; in some cases it is possible to destroy or disable much of the system without significantly affecting performance. 1 With systems in which the parts do not seem to be performing intelligible subtasks contributing to the overall task, classical mechanistic strategies -and, in particular, decomposition and localization-fall short. What alternatives are there to pursuing a program of mechanistic explanation? In Chapter 5 we considered one possibility: rejecting the mechanistic program and settling for descriptive accounts of behavior. But this is not a strategy for developing an explanation; it is a denial of any explanation. In this chapter we consider an explanatory strategy that abandons localization and decomposition. We leave open whether it constitutes a properly mechanistic approach. We examine three cases in the following sections. Hughlings Jackson provides a transitional case. Jackson rejects the localization of Broca and proposes a complex, hierarchical model of the nervous system in which control of tasks was distributed over different neural structures at different levels, with higher-level systems regulating and modulating performance oflower-level systems. Our other two cases come from contemporary research, which use newly developed formal mathematical tools for modeling the behavior of complex systems. These two cases are, respectively , models of cognitive performance and genetic regulation. In these cases performance depends primarily upon the interaction of the components in the system. The components do not perform tasks that would appear in a functional decomposition of the system. 2. HIERARCHICAL CONTROL: HUGHLINGS JACKSON'S ANALYSIS OF THE NERVOUS SYSTEM In the late nineteenth century, John Hughlings Jackson developed a hierarchical model ofthe nervous system that was intended as a repudiation of the kind oflocalizationist claims advanced by Bouillaud and Broca.2 These latter "neophrenologists," as we have seen, constructed localized models on the basis of correlations between neurological lesions and pathological symptoms such as the loss ofcoherent speech. Losses ofspecific capacities were traced to injury or destruction of specific regions. While such correlations certainly can be found, they are not as precise as would be required to justify the strong conclusions of these localizationists. The clinical syndromes are not...

Share