December 15, 2008

Review of "The Complementary Nature"

Review of: Kelso, J.A.S. and Engstrom, D.A. (2006). The Complementary Nature. MIT Press, Cambridge, MA.
By Bradly Alicea

This is a potentially momentous book. That being said, it is far from a synthesis. From a superficial perspective, it seems more like a long-winded manifesto with nice headshots of famous people. Nevertheless, the core idea is clear; namely that mentally-represented physical phenomena come in "complementary pairs", and that they form an interstitial and heterogeneous continuum between them. There is even a pairings glossary at the end of the book; each set of concepts is modified by a tilde (~) which denotes the link between two discrete states represented by linguistic titles.

The universal tilde designation is my major objection to this approach. The pairs actually seem to come in one of three varieties: binary oppositions, causal pairs, and hierarchical nestings. This enables higher-level mathematical operations, and scalable models to be constructed. Yet pairs of all categories are annotated in much the same way. Think of the tilde as a mathematical operator, which I'm sure was the authors’ ultimate intent given their tone. Following this logic, if pairs come in qualitatively different types, then the authors should have used different operators for each type. It would make the entire enterprise much more straightforward, especially when mapping pairs to a phase space as occurs later in the book.

Conceptual Taxonomy and Functional Information
Binary oppositions (linear~nonlinear) are by far the most straightforward. People tend to be most comfortable the outcome of such pairings, and can most intuitively analyze their outcomes. Consider the physical and mental aspects of hot~cold. Conditions in a physical
system range from hot to cold; indeed, not only is there a linguistic dichotomy, but a physical one as well. Because the mapping between the two is relatively seamless, we can easily quantify "hot" vs. "cold" using both a dichotomous representation coupled to quantitative instruments. It is the pairings that do not fall cleanly into this category that cause potential confusion. For example, causal pairs (reaction~anticipation) and hierarchical nestings (individual~society) might be considered differentiated states in a superficial sense, but treated as such may not map to a formalized phase space well.

One of the most intriguing ideas in the book is the way in which the authors conceptualize functional information. One example from brain science involves the specificity of COMT expression in prefrontal cortex. The initiation of gene expression at certain points in life history relies on the correct environmental conditions; interactions with surrounding proteins lead to specific types of emergent structures and specific phenotypes. No surprise there; classifying such processes as the flow of information is increasingly commonplace. The potential food for thought offered here is that this is part of an emergent process. In general, complex systems use functional information to build complexity. While information is a necessary prerequisite for emergent complexity, it is most powerful when coordinated by concurrent processes.

Treatment of Complexity
Their treatment of emergence (micro~macro) is one of the best I've seen, and is at once
mathematically rigorous and intuitive. They treat "individual" and "collective" as a metastable system (two local minima on opposite sides of a metastable "saddle point"). The system is driven by the values of a few key parameters; these parameters represent the reciprocal forces of downward and upward causation. Instability in these parameters drives the system towards a phase transition; more intuitively, the system climbs out of one stable state to a metastable plateau. At this point, it is free to return to its original state, remain unstable, or change to a new state. Dealing with the effects of causality on the initiation of phase transitions front and center makes for a much cleaner model than many of the other approaches out there.

More formal complex systems models becomes the thrust of this book's second part. Readers not familiar with Kelso's 1995 book "Dynamic Patterns" would do well to go there for a formal mathematical treatment. Once you understand the underlying concepts of coordination dynamics, go back and read "The Complementary Nature" again. Fresh eyes will provide you with a new perspective on the pairings. For example, pairings might be viewed as the boundary conditions of an n-dimensional phase space, or as discrete states in a multistable system. In any event, it is the space between the discrete states that are of interest to the authors. The take home message seems to be that this space is complex, unstable, and potentially fertile ground for the gray areas that humanities, brain science, and complexity scholars alike must understand.

No comments:

Post a Comment