James Kennedy

Russell C.Eberhardt

SWARM INTELLIGENCE

Academic Press 2001
pg 3

Models and concepts of life and intelligence

As human cognition is really the gold standard for intelligence, we will, as artificial intelligence researchers have done before us, base our model on people's thinking. Unlike many previous AI researchers, we do not subscribe to the view of mind as equivalent to brain, as a private internal process, as some set of mechanistic dynamics, and we deemphasize the autonomy of the individual thinker.

The currently prevailing cognitivist view, while it is extreme in its assumptions, has taken on the mantle of orthodoxy in both popular and scientific thinking. Thus we expect that many readers will appreciate our setting a context for this new perspective.

This introductory discussion will emphasise the adaptive and dynamic nature of life in general, and of human intelligence in particular, and will introduce some computational approaches that support these views.

We consider thinking to be an aspect of our social nature, and we are in very good company in assuming this. Further, we tend to emphasise the similarities between human social behaviour and that of other species. The main difference to us is that people, that is, minds,"move" in a high dimensional abstract space. People navigate through a world of meaning, of many distinctions, gradations of differences, and degrees of similarity.

pg 12

THE TWO GREAT STOCHASTIC SYSTEMS

Stochastic adaption is seen in all known living systems. Probabilistic choices allow creative, innovative exploration of new possibilities, especially in a changing environment. There are other advantages as well - for instance, perfectly predictable prey would be a little too convenient brawl for predators.

In general, organisms can adapt by making adjustments within what Gregory Bateson called the "two great stochastic systems". These systems are evolution and mind.

Evolution operates through variation and selection. A variety of problem solutions (chromosomes or patterns of features) are proposed and tested; those that do well in the test tend to survive and reproduce, while those that perform poorly tend to be eliminated. This is what is meant by "survival of the fittest".

The second "great stochastic system" is the mind. Cognitive scientists see the mind as a stochastic system of neurons, adjusting their firings in the response to stimuli that include other neurons. The brain is an excellent examplar of the concept of a complex adaptive system - but this is not what Bateson meant when he described mind as a great stochastic system. The distinction between brain and mind is extremely important. We are talking about minds here, not brains.

Some theorists argue that mental processes are very similar to evolutionary ones, where hypotheses or ideas are proposed, tested, and either accepted or rejected by a population. The most prevalent opinion along these lines is the "memetic" view, proposed by Dawkins (1976) in the Selfish Gene, which suggests that ideas and other cultural symbols and patterns, called memes, act like genes; they evolve through selection, with mutation and recombination just like biological genes, increasing their frequency in the population if they are adaptive, dying out if they are not... The theory of memes itself has spread through the scientific community with surprising efficiency... The view that mind and evolution operates by similar or identical processes is widely accepted by those who should know.

The similarities between the two stochastic systems are significant, but the analogy becomes ambiguous when we try to get specific about the evolution of raw abstractions.

The ancient question is, what, really, is an "idea"? You could take the view that ideas exist somewhere in a Platonic World of Forms, independent of minds. Our insights and understanding, then, are glimpses of the ideal world, perceptions of shadows of pure ideas. New Ideas existed previously, they just had not been known yet to any human mind.

Next we would have to ask, how does evolution operate on such ideas? No one has proposed that ideas evolve in their disembodied state, that truth itself changes over time, that the World of Forms itself is in a dynamical process of evolving. It has only been proposed that ideas evolve in the minds of humans; in other words, our knowledge evolves. The distinction is crucial.

The other view is that ideas have no existence independent of minds. According to this view, ideas are only found in the states of individual minds. At its most extreme, this position holds that in ideas are nothing more than a pattern of neuronal connections and activations. The problem here is that the same ideas almost certainly embodied in different neuronal patterns in different people - individual's brains are simply wired up differently.

Luckily for us, we don't have to resolve millennia-old questions about the preexistence of ideas. For the present discussion it is only important to note that for memetic evolution to act upon ideas, they have to become manifest somehow in minds.

They have to take on a form that allows mental representation of some type, and if they are to propagate through the population, their form must permit communication. The requirement of communication probably means that ideas need to be encoded in a symbol system such as language.

Our argument is that cultural evolution should be defined, not as operations on ideas, but as operations on minds.

The evolution of ideas involves changes in the states of minds that hold ideas, not changes in the ideas themselves; it is a search - by minds - through the universe of ideas, to find a fitter ones. This will become more important when we discuss mental activity, intelligence, and culture. We will emphasise now, that cognition is a different process from genetic evolution.

The great stochastic systems of different from one another: one uses selection, removing less fit members from the population, and the another adapts by changing the states of individuals who persist over time. These are two different kinds of adaptation.

This is not to deny that ideas evolve in the minds of humans. The ideas expressed by people certainly change over time, in adaptive way. We are suggesting a change in emphasis, that a scientific view of the evolution of ideas should look at changes of states of individuals, rather than at the ideas themselves. There is something difficult about thinking of minds in populations. We are used to thinking of ourselves as autonomous thinkers, sometimes accepting beliefs and processes that others around us hold, but the most of the time figuring out things on our own. Our experience is that we are in control of our cognitive systems, perhaps with some exceptions, such as when we are surprised or overcome with emotions; we experience our thoughts as controlled and even logical.

If we are to consider the evolution of ideas through a population, though, we will need to transcend this illusion of autonomy and observe the individual in the context of a society, whether it is family, tribe, nation, culture, or pancontinental species. In order to do that, we need to remove self-interest and sentimental self-aggrandisement from the discussion.

In Mind and Nature, Bateson detailed what he considered to be criteria of mind, qualities that were necessary and sufficient for something to recall the mind:

1 A mind is an aggregate of interacting parts or components.

2 The interaction between parts of mind is triggered by difference. For instance, perception depends on changes in stimuli.

3 Mental process requires collateral energy.

4 Mental process requires circular (or more complex) chains of determination. The idea of reciprocal causation, or feedback, is very important one and is fundamental to mental processes.

5 In mental process, the effects of difference are to be regarded as transforms (i.e. coded versions) of the difference which preceded them. Effects are not the same as their causes; the map is not the same as the territory.

We will not attempt here to explain or elaborate Bateson's insightful and subtle analysis. We do want to note though that he hints - in fact it is a theme of his book - that biological evolution meets these criteria, that nature is a kind of mind.

This seems a fair and just turnaround of the currently prevalent opinion that mind as a kind of evolution. We will see that evolutionary processes are up can be encoded in computer programs used to solve seemingly intractable problems, where the problem is defined as the analogue of an ecological niche, and he combined and mutated variations are tested and competitively selected.

In other words, we can capitalise on the intelligence, the mental power, of evolution to solve many kinds of problems. Once we understand the social nature of mental processes, we can capitalise on those as well.

It appears to some thinkers that mind is a phenomenon on that occurs when human beings coexist in societies. The day-to-day rules of living together are not especially complicated - some relatively straightforward scraps of wisdom will get you through life well enough. But the accumulated effect of these rules is a cultural system of imponderable depth. If we could get a sufficiently rich system of interactions going in a computer, we just might be able to elicit something like human intelligence.



HOME      BOE     SAL     TEXTE