Piero Scaruffi(Copyright © 2013 Piero Scaruffi | Legal restrictions )
These are excerpts and elaborations from my book "The Nature of Consciousness"
A Self-organizing Universe
The main property of neural networks is feedback: they learn by doing things. Memory and learning seem to go hand in hand. Neural networks are "self-organizing" objects: response to a stimulus affects, among other things, the internal state of the object. To understand the behavior of a neural network one does not need to analyze the constituents of a neural network; one only needs to analyze the "organization" of a neural network.
Physics assumes that matter has no memory and that the laws of Nature entail no feedback. Physics assumes that all objects in the universe are passive and response to a stimulus does not affect the internal state of the object: objects are non-organizing, the opposite of self-organizing objects. To understand the behavior of a physical object, one needs to analyze its constituents: the object is made of molecules, which are made of atoms, which are made of leptons and quarks, which are made of...
There is no end to this type of investigation, as history has proved. The behavior of matter still eludes physicists even if they have reached a level of detail that is millions of times finer-grained than the level at which we operate. There is no end to this type of investigation, because everything has constituents: there is no such thing as a fundamental constituent. Just like there is no such thing as a fundamental instant of time or point of space. We will always be able to split things apart with more powerful equipment. The equipment itself might be what creates constituents: atoms were "seen" with equipment that was not available before atoms were conceived.
In any case it is the essence itself of a "reductionist" (constituent-oriented) science that requires scientists to keep going down in levels of detail. No single particle, no matter how small, will ever explain its own behavior. One needs to look at its constituents to understand why it behaves the way it behaves. But then it will need to do the same thing for each new constituent. And so on forever. Over the last century, Physics has gotten trapped into this endless loop.
Could matter in general be analyzed in the same way that we analyze neural networks? Could matter be explained in terms of self-organizing systems? Neural networks remember and learn. There is evidence that other objects do so too: a piece of paper, if folded many times, will "remember" that it was folded and will learn to stay folded. Could we represent a piece of paper as a self-organizing system?
Nature exhibits a "hierarchy" of sorts of self-organizing systems, from the atomic level to the biological level, from the cognitive level to the astronomical level. The "output" of one self-organizing system (e.g. the genome) seems to be a new self-organizing system (e.g. the mind). Can all self-organizing systems be deduced from one such system, the "mother" of all self-organizing systems?
We are witnessing a shift in relative dominant roles between Physics and Biology. At first, ideas from physical sciences were applied to Biology, in order to make Biology more "scientific". This led to quantifying and formalizing biological phenomena by introducing discussions on energy, entropy and so forth. Slowly, the debate shifted towards unification of Physics and Biology, rather then unidirectional import of ideas from Physics. Biological phenomena just don't fit in the rigid deterministic model of Physics. Then it became progressively clear that biological phenomena cannot be reduced to Physics the way we know it. And now we are moving steadily towards the idea that Physics has to be changed to cope with biological phenomena, it has to absorb concepts that come from Biology.
In order to accommodate biological concepts, such as selection and feedback, in order to be able to encompass neural and living systems, which evolve in a Darwinian fashion and whose behavior is described by non-linear equations, Physics will need to adopt non-linear equations and possibly an algorithm-oriented (rather than equation-oriented) approach.
Almost all of Physics is built on the idea that the solution to a problem is the shortest proof from the known premises. The use and abuse of logic has determined a way of thinking about nature that tends to draw the simplest conclusions given what is known (and what is not known) about the situation. For example, it was "intuitive" for scientists to think that the immune system creates anti-bodies based on the attacking virus. This is the simplest explanation, and the one that stems from logical thinking: a virus attacks the body, a virus is killed by the body; therefore the body must be able to build a "killer" for that virus. The disciplines of life constantly remind us of a different approach to scientific explanation: instead of solving a mathematical theorem through logic, nature always chooses to let things solve themselves. In a sense, solutions are found by natural systems not via the shortest proof but thanks to redundancy. The immune systems creates all sorts of antibodies. An invading virus will be tricked into "selecting" the one that kills it. There is no processor in the immune system that can analyze the invading virus, determine its chemical structure and build a counter-virus, as a mathematician would "intuitively" guess. The immune system has no ability to "reason" about the attacking virus. It doesn't even know whether some virus is attacking or not. It simply keeps producing antibodies all the time. If a virus attacks the body, the redundancy of antibodies will take care of it.
This represents a fundamental shift of paradigm in thinking about Nature. For many centuries, humans have implicitly assumed that the universe must be behaving like a machine: actions follow logically from situations, the history of the universe is but one gigantic mathematical proof. It is possible that the larger-scale laws of nature resemble very little a mathematical proof. They might have more to do with randomness than with determinism.
The distinction between instruction and selection is fundamental. Physics has evolved around the concept of instruction: mathematical laws instruct matter how to behave. Selection entails a different set of mind: things happen, more or less by accident, and some are "selected" to survive. The universe as it is may be the product of such selection, not of a logical chain of instructions.
Physics is meandering after the unified theory that would explain all forces. What seems more interesting is a unification of physical and biological laws. We are now looking for the ultimate theory of nature from whose principles the behavior of all (animate and inanimate) systems can be explained. Particles, waves and forces seem less and less interesting objects to study. Physics has been built on recurring "themes": planets revolve around the sun, electrons revolve around the nucleus; masses attract each other, charged particles attract each other. Still, Physics has not explained these recurring patterns of Nature. Biology is explaining its recurring patterns of evolution.
A new scenario may be emerging, one in which the world is mostly non-linear. And somehow that implies that the world self-organizes. Self-organizing systems are ones in which very complex structures emerge from very simple rules. Self-organizing systems are about where regularity comes from. And self-organizing systems cannot be explained by simply analyzing their constituents, because the organization prevails: the whole is more than its parts.
Back to the beginning of the chapter "The New Physics" | Back to the index of all chapters