Holistic Science | April 1994

FMBR Editorial: April, 1994

Holistic Science

Marshall Pease

Recent research on systems of cellular automata is yielding some fascinating insights into the possibility of holistic science. "Cellular automata" are autonomous units that still must interact with their environment and each other. In the real world we can regard individuals, cultures, species, economic systems and many other entities as cellular automata. Research generally models such systems on computers using autonomous program modules that interact with the computer's environment and with each other according to rules contained in each module. In addition, the automata in a computer system may mutate randomly on occasion. The environment commonly includes a limited resource of some kind for which the automata must compete, leading to a kind of Darwinian selection that eliminates unsuccessful mutations. The system is allowed to run to see what emerges. Some of these systems mimic life so successfully that they become examples of what is called "Artificial Life." Whether the automata in these systems can or should be considered as actually alive, or as merely modeling how life behaves, is a subject of epistemological and philosophical debate. In any case, they provide an extremely interesting new realm of scientific research.

A characteristic of such systems is that they seek to discover holistic phenomena. How the system changes as the cellular automata evolve is determined by the entire system, not the details of any particular type of automaton or any special instance of interaction. In a very real sense, researchers in this field have had to transcend the limitations imposed by the reductionist approach so basic to much of science.

To illustrate, the system may begin with automata competing for some limited resource that can be regarded as "food." As successful mutations occur and propagate through the system, the automata learn to improve how they seek and acquire the "food" they need. All of this is quite reasonable. The system becomes more interesting when a few predators are introduced into this comfortable and gentle world. This sets up a typical predator-prey system.

The fascinating thing is that the prey species evolves much more rapidly than before and often reaches a substantially higher level of competence. Indeed in some cases the prey species becomes so competent that the predator species actually dies out. Furthermore, in some of those systems, the prey species then proceeds to die out also! The prey species has become so skilled at surviving the predators that they can no longer maintain themselves without the threat! This phenomenon is one of co-evolution as the system responds to dynamical processes that reflect the entire system as a single whole.

I am not going to argue what this says about the human condition. What does interest me here is what it says about the researchers doing the work. These are hard experiments that have been deliberately designed and executed to exhibit phenomena that are properties of an entire system -- holistic effects. They yield hard data (mountains of it, sometimes!) in a way that violates the reductionist principle that has long been the scientific ideal. Perhaps there is hope for science yet!

Marshall Pease


This editorial is based in large part on the book Artificial Life by Steven Levy, Vintage Books, Random House Inc., New York, 1992. The book is described on its cover as "a report from the frontier where computers meet biology." It is quite readable and I recommend it for those who find the topic as interesting as I do.