For more information contact Luis Rocha at rocha@lanl.gov
Last Modified: November 6, 2002
This is the abstract of a talk prepared for the International interdisciplinary seminar on new robotics, evolution and embodied cognition (IISREEC).12th to 15th November 2002, Lisbon, Portugal
Abstract: As we know, all forms of life on Earth have a single origin some 3.5 billion years ago. Life never happened de novo again, for every living being originates from another living being, embodying a common history through evolutionary time. The evolution of the extraordinary diversification that we can witness today owes, as we also know, to a few characteristics of living beings, which include heredity (of life itself to start with, with the historically derived “molecular organization” it presupposes), random genetic variation, molecular (and cellular) cooperativity in increasingly larger (self-organizing) networks on a background of marked degeneracy, followed by (epigenetic) selection for reproductive advantage. The process, as we understand it, has been entirely “purposeless”, entirely devoid of a “blue print”. Hence, biological evolution is the precise opposite of engineering in general, and “robotics” in particular. I submit, therefore that biology can only contribute to robotics with metaphors, the significance of which scientists systematically doubt.
The immune system is no exception, both in its evolutionary origins, as well as in its development inside every vertebrate. This is the first interesting characteristic of immune systems: they represent within each individual, some sort of fractal of biological evolution itself, following the same principles of random variation, cooperativity, degeneracy and selection, now applied to cells inside of a vertebrate organism. Because individual life spans are very short, however, evolution has endowed the immune system with means to accelerate genetic variation by many orders of magnitude, and with the continuous, life-long generation of large numbers of lymphocytes, such that the strategy may actually work.
The workings of immune systems have often been described as “cognitive”. Thus, the system “learns” (and here we go into metaphors) to “discriminate” between “self” and “non-self”, “decides” on the treatment to apply to molecules (eliminate them from the body, or maintain them and protect them from other attacks), keeps a life-long “memory” from past experiences. Some go as far as considering that immune systems play a general regulatory role for all other systems in the body, and speak of its operation (metaphorically, of course) as the “molecular consciousness of the body”. Moreover, as the great evolutionist Susumu Ohno put it, immune systems are Prometheic as their evolutionary value rests on the fact that they can “foresee” the molecular evolution of microbes.
In contrast, it is conventionally assumed that the evolution of life is Epimetheic, entirely based on “hind sight”. For the immunologist, however, this level of description has little interest, as it says nothing on the mechanisms that unable an individual to be “immune” to measles, for example, many years after having been infected with the virus. Interestingly, however, the apparently “intelligent” behaviors of the immune system, its very “consciousness”, were resolved into very simple rules of cellular dynamics, cooperation and selection for prolonged survival, recruitment and expansion, often resulting in very surprising “gains of function” and unpredictable behaviors. Again, on the solid basis of an extreme diversity of components and of a high degeneracy in their interactions. Again, on the same general mode of the “blind watchmaker”, starting from the genetic catastrophe that, by horizontal gene transmission, is today believed to have been at the very origin of the system. Again, as in evolution of life itself, the critical target for significant changes is the developmental time, such that the performance of the mature system owes entirely to its own development. As expected, the emergence of complex behaviors may be traced down to a few “developmental genetic programs”, if enriched by the randomness/degeneracy of a large diversity of molecular components. Interestingly, while degeneracy in molecular interactions is an a priori that has been evolutionary incorporated into the respective functional significance, its level itself is determined by developmental genetic programs, just like the randomness in the generation of diversity. This might be the useful metaphor for “robotics”: the immune systems’ structure, dynamics and organization, even some of the simple rules governing the behavior of its cellular components, are quite different in its “immature” stage (when it does not perform “intelligently”) and in its mature stage (when it does). These characteristics and their temporal unfolding, however, are under the control of evolutionarily selected developmental programs. Perhaps “robotics” would gain from getting closer to the developmental biology of complex systems.