Modularity in Complex Natural Systems (Remembering Herb Simon) |
Herbert A. Simon's landmark work on nearly-decomposable systems is the premier source for ideas about thinking in terms of modules. This session will honor Simon, who died recently, by addressing the modularity of biological systems (in development, evolution, and mind).
Organized by: Werner Callebaut, Diego Rasskin-Gutman & Wiliiam Wimsatt
- David Walton, University of Notre Dame
"The Intellectual Roots of Simon's Views on Complexity and Computation"
- Although Herbert Simon's ideas on near-decomposable systems resonate with concerns about evolution and biological complexity, his approach to the subject actually developed in the social sciences, not the biological. From the earliest years of his career, Simon was interested in applying computational ideas to the study of rational behavior, and his eventual concern with complexity was largely a response to developments in organization and administrative theory, political science, economics, operations research, as well as formal attempts to describe computation, such as von Neumann's program in automata theory. It is my argument that Simon's program of simulation in artificial intelligence emerged as an approach to modeling rational behavior which sought to introduce computational concerns while simultaneously seeking to minimize any methodological commitments to the nature of the underlying computational hardware.
- Werner Callebaut, Konrad Lorenz Institute for Evolution and Cognition Research
"Bounded Rationality and Modularity"
- Although Simon received the Nobel Prize for Economics in 1978, most practitioners of the dismal science have been reluctant to adopt his satisficing approach, which has revolutionary implications for the foundations of economic theory. Their typical response to the challenge of bounded rationality has been to reinterpret satisficing behavior as optimizing under constraints, although arguments to this effect have been shown to be flawed from the 1960s on. In this talk, I address the reception of Simon's computational view of rationality in biology (ecology, evolutionary game theory, etc.), where it meets with equal resistance from pan-adaptationists and others. Focussing on the promise of Simon's "ontology" of nearly-decomposable systems rather than on his "epistemology" of satisficing (which are intimately related!), I try to assess recent work on the modularity of living systems and mind from a consistently Simonian perspective.
- W. C. Wimsatt, University of Chicago, and J. C. Schank, University of California, Davis
"An Apparent Evolutionary Paradox"
- Modular adaptive systems would seem to have great advantages over comparable systems with less modularity. Herbert Simon pioneered arguments that "nearly-decomposable" systems modularly organized in "stable subassemblies" should evolve more rapidly than things that are not. Lewontin urges related principles of quasi-independence and continuity: with any variance for modularity, there should be selection for more of it. But these intuitions may be misleading, as our simulations show. With selection in complex multi-locus systems, even apparently unrelated elements become tied in a kind of unanticipated interaction spanning the whole organism. The resulting system illustrates interesting characteristics of near-decomposability with unanticipated results.
- Carl Bumba, University of Vienna
"Intrinsic and Extrinsic Regulation of Developmental Gene Modules"
- For this talk, I present a standard model for gene regulatory networks in development based on likely, functional relationships among developmental gene families and splice variants and the agonistic and antagonistic interactions known to occur among members of these groups. This description invites a community ecology perspective that should include, I suggest, the concept of keystone species. While changes in "keystone genes" may result in small, gradual changes to the phenotype, macro-evolutionary changes likely result from stabile changes outside the network, such as environmental, nucleotypic, chaperone system changes, and even, perhaps, clustered Hox gene changes. I use my studies in morphological evolution involving genome size and cell cycle parameters and, more recently, heat shock proteins to demonstrate that large, balanced changes in phenotype may occur during evolution from changes outside of developmental gene "modules". In addition, this distinction between the micro- and macro-evolutionary aspects of gene regulatory networks helps reconcile a contradiction, I propose, between two common concepts to explain major, integrated changes in evolution - genetic assimilation and the 'decoupling of developmental constraints'.
Conference Main Page