|
|
|
|
A shorter version of this article was published in the Guardian, October 1st 1998 Where Newton went wrongSteve Grand The answers to all the scientific questions we could possibly ask are sitting right in front of our noses, yet we don't see them. Sometimes it's because they involve things that are too small, too distant or otherwise lie outside the range of our senses. However, more often it’s simply that we fail to notice them. Darwin and Wallace went through years of painstaking observation before the theory of evolution by natural selection began to crystallise in their minds, but now that we know how evolution works, the evidence for it confronts us everywhere we look. In fact Philosophy can be facetiously but accurately described as "the art of stating the bleeding obvious". We just have to remember that what’s obvious after it has been stated might have been totally overlooked up until that moment. Take Newton’s famous laws of motion, for example. These are no more than "if you push something it’ll keep moving until you stop it", "the harder you push it the faster it accelerates" and "if you shove it, it will shove you back". All of which are so "bleeding obvious" that it took centuries of intellectual endeavour before the ideas actually occurred to anyone. Most, if not all progress occurs because of such a change of viewpoint. The snag is that these new paradigms can themselves quickly become the blinkers that prevent us from further advances. I work in the field of Artificial Life, trying to construct synthetic intelligent life forms for use in many kinds of technology application. A-life is a relatively new discipline, founded a decade ago on some quite revolutionary principles that were set to break out of the mental straightjacket that has surrounded Science for the past three hundred years. However, even A-life has lately begun to suffer from Physics Envy. Somehow the initial vitality and novelty of the approach is being replaced by Received Wisdom and hopelessly outdated thinking. I’m not trying to knock the classical scientific worldview, which has undoubtedly been stupendously successful. The problem is that we so easily fall back into modes of thought that, although very good for understanding the motion of planets, are wholly inadequate for things that are more than the sum of their parts, like intelligent living beings. Intelligent systems (whether natural or artificial) are messy, complex things, which simply aren’t amenable to the reductionist, materialist and mechanist approaches of classical Physics, from which most of our present-day intellectual tools derive. Physics itself is based on our intuitive understanding of the world, which in turn is a result of the way our senses and brains function. Ironically, then, the biggest hurdle to understanding how the brain works (and hence to making artificial brains possible) is the way the brain works. Take the way Physics is, almost by definition, concerned with matter. This obsession with stuff is a perfectly natural result of the way our senses respond to their environment. Our visual system converts a mass of coloured dots into discrete physical objects and classifies them, long before we become consciously aware of the scene and so, naturally, we assume the world really is divided up like that. Similarly our hearing and touch are responsive to material things but unable directly to discern any non-material phenomena, such as poverty, anger or, notably, minds. Consequently we tend to think of tangible things as real, while the intangible contents of the Universe are somehow disregarded as inferior or unreal. This is reflected (or perhaps compounded) by our strangely pejorative use of language. For some reason "material facts" are good, while if a thing is "immaterial" it is irrelevant. Likewise, "tangible assets" are better than "intangible ones" and "substantial" means something positive, while "insubstantial" is derogatory. Even the word "matter" carries emotive baggage when we discriminate between things that matter and things that don’t! But suppose we’ve got it all wrong. Suppose the distinction between matter and form is false and misleading. Suppose tables and chairs are made of the same "stuff" as minds, rather than the other way around. Suppose the world isn’t really divided up into discrete objects at all, but is one continuous, inseparable mish-mash. Suppose that it is not the things themselves that are "real" and significant, but the relationships between them. Suppose hardware is simply a subset of software. I submit that these facts are actually so "bleeding obvious" that our consequent total disregard of them is seriously hampering our ability to make sense of the world in general and intelligent systems in particular. I don’t have space here for all the illustrations that would help pave the way, so let’s jump straight in by bluntly denying that matter exists. It’s not my objective to upset physicists or put them out of work, but simply to outline a view of the world in which mind and intelligence can take their rightful place in the scheme of things. From such a standpoint we can perhaps begin to view all phenomena, including intelligent ones, in a coherent and manageable fashion. Let’s begin with a thought experiment. Imagine yourself by a swimming pool and mentally switch off the gravity. Now sculpt the surface of the water into a variety of shapes. Pick any shapes you like—in my mind I’ve created an authentic scale model of San Francisco, complete with the Golden Gate Bridge, but then I have a disturbed imagination. OK, now switch the gravity back on and watch what happens. It doesn’t matter what shapes you started out with—within a few seconds only two will remain: a mass of sinusoidal ripples moving at a uniform speed across the surface, plus maybe a few small whirlpools. The reason for this is that ripples and whirlpools are persistent phenomena on water, while the Golden Gate Bridge is not. This is a demonstration of the most fundamental, if tautological, principle of nature: Things that persist, persist; things that don’t, don’t. As always this is bleeding obvious, but nonetheless hugely important. This one law explains everything we see around us. We already recognise it as the central rule behind evolution, but it actually applies right across the board: on water, in space-time, in societies or in completely abstract spaces like the famous mathematical Game of Life. The reason that ripples are able to persist on water is that they are feed-forward mechanisms; they copy their shape forward in space in a kind of domino effect, otherwise known as propagation. Whirlpools are similar, but wrapped around in a loop. Even though photons of light and subatomic particles are rather different from ripples and whirlpools, in an important way they are the same: each is a disturbance in a field or fields (magnetic, electrical or Olympic-sized), which persists through propagation. It is not unreasonable to view photons as being like ripples, and particles and atoms as like whirlpools. Instead of the intuitive and classical perception of a universe like a canvas, onto which matter is daubed like paint, this gives us an impression of matter which is, well, an impression—a form embossed into space rather than stuff superimposed on it. Matter is therefore software, not hardware. One can then imagine a hierarchy of ever more sophisticated persistent phenomena superimposed on these simpler ones (like molecules superimposed on atoms). Science is essentially the study of how, what, why, when and where phenomena persist. Persistence through simple feed-forward mechanisms might interest physicists, but biologists get more excited when phenomena persist through feedback. Storm clouds are an example of simple phenomena that persist through positive feedback; in other words they are self-reinforcing systems. More complex phenomena also use negative feedback, in which the system responds in a way that tends to negate any external disturbance, just as a thermostat keeps the room temperature constant despite changes in the weather. Many phenomena even master the impressive trick of persisting by duplicating themselves, blueprints and all. This is called replication or reproduction, and we describe such phenomena as "alive". So we have a ladder of forms of persistence, each rung consisting of more sophisticated arrangements of "causal flow" (feed-forward and feedback) that ensures its ability to survive. Whenever such phenomena come into being, they tend to stay around, and sooner or later a "scientist phenomenon" will pop along and try to classify them. Scientists themselves (perhaps contrary to popular belief) are examples of some of the most sophisticated kinds of phenomena, which persist by means of intelligence. Intelligence equates with predictive power. The simplest living systems are those which persist by resisting change, using feedback from the environment to adjust their behaviour. Such phenomena are what we might class as the lowest, "adaptive" level of intelligence. Above that comes a much more powerful mechanism: the ability to adapt before the danger even arises. Here, instead of the environment feeding back on behaviour, it feeds back onto a memory of behaviour, allowing the response to differ next time around (learning). Such systems come into a second, broad, "predictive" category of intelligence. If we characterise the first level by the phrase "it’s raining so I’ll shelter", and the second by "it’s getting cloudy, I’ll go home before it rains", then the third level of intelligence is perhaps "it’s always bloody raining; I’ll invent the umbrella". This is "creative" intelligence; a form of prediction so powerful that systems which exhibit it can predict the behaviour of configurations never before witnessed. The ability to predict the outcome of strapping a circle of cloth to a stick and holding it over one’s head is essentially what appears to separate the human species from the rest. Our rough hierarchy of persistent structures starts with so-called matter and passes up through various kinds of adaptability and intelligence, to include higher-level entities such as "society". Everything is software; everything is made from the same (non-) stuff. Such a viewpoint at the very least weakens some of the more naïve dualistic notions of mind that hamper our understanding. What’s more notable, however, and the reason workers in Artificial Intelligence should take heed of this cock-eyed viewpoint, is that throughout this whole ensemble there are only a small number of basic mechanisms at work. Whether we’re talking about clouds, organisms, minds or societies, the same building blocks occur, albeit under different names. For example, one crucial element is the "modulator"—a mechanism that allows one flow of cause and effect to modify another. In electronics we call modulators "transistors", while in biological systems we call them "synapses" or "catalytic reactions". All of these are examples of the same fundamental device. It seems to me that by identifying these few cybernetic elements from which all software (matter, life, mind, society) is constructed, we find ourselves with a universal LEGO set. From these blocks we can create all kinds of organisations, including intelligent living ones. Biological systems in fact give us all the clues we need as to their nature, and after several years of practical exploration I now think I know what these LEGO blocks look like. The only remaining problem is to figure out better ways to put them all together to create intelligent life. Sadly there doesn’t yet seem to be a book called "Sentience for Dummies", so it looks like I’ll just have to make it up as I go along… |
Copyright © 2004 Cyberlife Research Ltd. |
