Customer Reviews

Self-Modifying Systems in Biology and Cognitive Science, Volume 6: A New Framework for Dynamics, Information and Complexity (IFSR International Series on Systems Science and Engineering)

5 star:		(3)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

As another reviewer notes, this book will be very helpful to those interested in the complexity research of theoretical biologist Robert Rosen. Whereas Rosen has a tight, highly rigorous focus on his goal in "Life Itself", Kampis paints on a somewhat broader canvas, referencing the work of many other researchers (including Rosen). However, Kampis is similarly detailed and methodical.

Kampis first describes the limits of dynamical models, and state-based approaches, including the limitations inherent in the 'canonical formalism' of mechanics.

He then goes on to introduce 'component-systems'. This is a general formal representation of a system as being composed of some number of components out of an essentially unlimited number of possible components. In component systems, the "rules" for the dynamics of the system are not independent of the components themselves. Self-modifying component systems generate new components and delete others, thereby changing the identity of the system itself. In mathematical terms, a self-modifying system is like a function f that belongs to its own domain and range ("f:f-->f"). The result is that such systems are non-algorithmic, nor are their dynamics describable in a state-based formalism (e.g., Newtonian, Hamiltonian, etc.). This has notable consequences for approaches that attempt to treat such systems as algorithmic, or via modelling their state-based dynamics. By comparison to component systems, cellular automata and similar algorithmic formal systems are entirely trivial.

Kampis devotes many chapters to what I have cursorily mentioned, and there is much, much more in this book that is worth reading. Although there is not alot of math, what is there is important to understand. It would be helpful for the interested reader to generally understand the basic notation of mechanics, first-order differential equations, basic logic, Godel's Incompleteness Theorems, Turing machines, basic set theory, system theory, a modicum of philosophy, and linguistics. Most of these aspects are fairly well-explained, so a diligent reader can pick them up as he goes along.

This is not a book of vague handwaving arguments. It will take some studious effort to read and grasp the concepts and profundity of what he presents. However, it will be well worth the effort, and afterward you will never be able to look at dynamical systems and models, complexity, and self-modifying systems, in the same way.

Although there are alot of similarities between Kampis' and Rosen's works, they are sufficiently distinct in approaches and conclusions that both are well worth reading.

One final note: the "typewriter" font used throughout may be a bit surprising to see in the 21st century, but I found it entirely legible and comfortable once I got used to it.

The theme that Kampis examines in Self-Modifying Systems is the self-generation of information by the nontrivial change (self-modification) of systems. Such a system is a network of many components, which have the property of being able to transform each other and organising themselves into larger components. It is this feature that makes such component systems closed to efficient cause. Component-systems, then, are not algorithmic, but this is not a reversible equation in that component-systems can, Kampis argues, give rise, in fact, to any particular algorithm. Kampis describes the difference as that between known complexity, that is to say complexity-to-be-realised, and unknown complexity, or complexity-to-be-explained. The first of these is relatively easy to realise, the second being impossible in that "a complex operation operating on components and bringing forth yet unknown and unidentified components cannot be described as an algorithm" (Kampis 1991:239).
Component-systems, therefore, have a high degree of creativity, but they also have characteristics that avoid many of the problems that other forms of nonlinear models.Kampis argues that nothing that such a process gives rise to can be predicted before hand, and no identity can be traced back to an origin. From this, Kampis states that the creation thesis emerges. This thesis can be stated in the following way:
The organisation of the world is continually self-creating; this process is at any given stage incomplete. Information about the future is not only inaccessible but does not exist in any form. Creation is a basic and general phenomenon that cannot be explained logically. (Kampis 1991: 258).
Self-creation occurs in the form of self-modification. A system that exhibits creativity, then, has to be continually redefined because, in the course of time, all variables and their interrelations will change in so far as each component is replaced by another. It is a system which will be defined (and constructed) by the very processes it undergoes. (Kampis 1991: 490).
The book unfolds, then, as a wonderfully sophisticated model to account for the very process of change and the important limitations of prediction the process of change implies. This book deserves to be one of the key texts of autopoiesis.

George Kampis follows in the ground breaking tradition of Robert Rosen. Examines the notions of reproduction and construction. His scope is wide and through. A must read for anybody interested in Rosen complexity.