INTELLIGENCE ARTIFICIAL OR OTHERWISE
The problem for many of us when considering what constitutes intelligence is that animals are, in terms of surviving in their environment, clearly just as intelligent as we, that is humans, are even to the point of manipulating it in personally positive ways e.g. digging burrows, and developing complex constructions such as nests or beaver dams which require a degree of engineering "nous" if they are to succeed.
Clearly brain size is not particularly related to species success, other than in that man, who has a large brain, might be said to be successful, although is often unsuccessful at controlling pest species, particularly of the insect kind.
In my youth (some 30 - 40 years ago) it was a generally held belief that mans success and superiority stemmed from his manipulation of the environment to meet his needs rather than just responding to whatever mother earth threw at him, indeed it was a generally held view that man would eventually be able to control every aspect, including the weather of his environment. A bit like GOD really. However I notice as we start the 21st century quite a few voices have been raised suggesting that we must stop attacking the environment head on but to go with it, let it do it's own thing, and make use of what we can in as non intrusive manner as is possible.
Where man clearly differs from others species may be as follows:-
The degree to which mankind is willing to accept long term goals to modify short term behaviour, and the degree to which intellectualized goals can be substituted for real ones e.g. the success of religions promising heaven after death verses a good shag now! (This is a deliberate attempt at not modifying current behaviour to enhance long term goals. The long term goal of acceptance of this paper by my peers verses the short term desire of a joke to shock etc. By adding this note I am attempting to modify you the readers, perception of what is going on. I am attempting to alter it back to your perception of a statement which compares religion to the sublimation of the satisfaction of basic instinctive motivations)
(You are probably aware by now that the original unacceptable form was in fact a much clearer communication than the the more complex acceptable form).
The degree to which man is motivated to manipulate his environment, or his refusal to accept the hand dealt and to set out to do something about it. Bruce's spider is relevant in this context.
The degree to which man can conceptualize, communicate and activate environmental manipulation
Towns are by and large the most amazing constructs some of which go way beyond any useful grouping of facilities.
Tokenised language development
Language is not the same as communication as most animals communicate and many use this as a basis for cooperation. The concept of communication is clearly understood by many animals, a short spell of dog training will soon convince you that the dog is well aware of your futile attempts to communicate, and finally cooperates by learning your communicants having clearly given up on you learning theirs. The skill of horse whisperers has recently been put down to the ability of some people to observe the horses communicants and to learn them and use them in a way horses can recognise).
Intelligence is therefore I suggest, nothing more than language consisting of a grammar (rules) and a lexicon of words, some of which represent real world facts, and others which represent groupings of other words. All school children have doubtless played the game of looking up the meaning of a word in a dictionary, then looked up the meaning of the keywords used to give the meaning, then repeated this until the find themselves back at the original word. The larger the dictionary the greater the number of words checked.
How does this relate to the real world. The fox gets into the chicken run bites all the chickens heads off (yes they really do this) gets labeled --instinctive behaviour, hardware intelligence. The farmer sends his chickens in a crate to the slaughterer where they are killed and their heads removed (in some parts of the world killed by having their heads removed for instance compare to France and the guillotine) --intelligent behaviour, software intelligence.
In tokenised forms there is little to choose between the two actions except that the farmer has language tokens to describe what he did the fox does not (at least as far as we know, and the fox certainly doesn't act as if it has which is all that really matters). This can be developed to show that many sophisticated forms of human behaviour only exist because they are tokenised and have grammatical type rules. The stock exchange has jargon words and rules which allow it to operate, Chess has jargon words and rules so does a compiler for a computer, but here the affinity to language has been recognised "Computer Language" but why not "Operating as a stockbroker language". there has long been a dichotomy between arts and science, but many jobs which look more akin to science have been filled by arts graduates especially those with a language "classics" background. What they are good at is language and to quote a gamekeeper I know "gob beats brains every time", and in most non physical non research based occupations (i.e. 90% of non manual jobs ) those good at language are usually the winners. If you can learn one language easily you can learn others just as easily, so a German language scholar should be able to learn mathematics***, or how to be a stockbroker.
Certainly this means the dyslexic person is more widely disabled than has originally been thought and why there is such a wide range of activities at which they under perform, even football has a language consisting of jargon words and rules and must be learnt. but learning language is really difficult for dyslexics. Oh, and yes I do recognise that mathematics*** may require other hardware brain firmware than the hardwired brain firmware which facilitates language skills. An interesting aside is that occupationally some languages might be harder than others, for instance the doctor language is probably more difficult to learn than the estate agent language, and clearly the soccer football game language is much simpler than the American rules football game language.
At the first level language difficulty might well be related to size, i.e. the number of jargon words to learn and the number of rules to memorise. It is interesting in this context that English used to be a difficult language to learn, with it's strong emphasis on correct grammar and large lexicon, however when psychologists studied the language they found you could successfully use it with very few words and rules, about 200 to 500 words without jargon (the nearest comparable language needed 2000 words) and since this time it has , based on this information been relatively easy to get going and now is, I understand, the largest second language in the world. There is "simplified" English, a small compact language easy to learn, and "correct" English which has the largest lexicon of any language and more rules than you could shake a stick at.
The more difficult a language is the less people who understand it, and the more ordinary people look for a summary language to handle it with, I suspect that the only time people speak the word 'faeces' is when they go to the Doctor, but in ordinary, as against occupational, life doctors say shit same as the rest of us. We actively look for a simplified way of dealing with an activities language, often by talking more loudly.
Other factors may well be how different is the lexicon to one we all ready know, it may well be easier to add a lexicon which is a sub set of one we know. Learning novel words is more difficult especially if we do it in quantity. It might be more efficient to get this part over with right at the beginning of learning a new language/occupation/game by learning the lexicon by rote. It is clearly more efficient to do mental arithmetic having learnt the multiplication tables by rote and this might be applied to other activities.
Next, simple rule sets are more easily learnt, both quantity and complexity play a part, and more complicated rules are more easily learnt if they can be broken down into simple rules or grammars. Moreover there would appear to be scope for the transfer of grammar from one situation to another and for the transcription of grammars from things difficult to learn to be more closely allied with the Grammars of things easy to learn. There has been considerable work done by computer scientists and by linguists to develop rules for good grammar, and clearly a "Good Grammar" language is beginning to evolve, hopefully as a simple language.
There appears to be no problem with a language operating at different levels of complexity but it generally only gets wide acceptance if their is a simple version. One of the real problems is that ,Human nature being what it is, when there is only a simple version satisfactory as that might be, clever buggers come along and develop a complex language in order to keep ordinary people out, and of course to make money from it. You can't therefore play our game unless you learn our rules and often achieve some initiation process set up by us, usually age or time related, and to the benefit of those already there.
It would seem possible that the human brain is hard wired to learn tokenised language, that is lists of words, the meanings of these words as symbolic objects, actions or constructs of other words and the grammatical rules to manipulate this knowledge base and derive further constructs. Such constructs may well be actions (move over there -walk) expressions (I do not want to move over there -speak) or ideas, a string of words held in memory which describe something and may well have a considerable number of non word tokens attached to it, such as "when needed fetch to THE BRAIN'S IMAGE TO WORD PROCESSOR that image of the Taj-Mahal so I can add the description of it's dome to my description of this new building 'MODIFIED' by the word string in memory location [**?**--**)(**...] ".
If we are dealing in artificial intelligence terms then the brain has an inference engine capable of handling complex languages, pre and post processing data to and from that language and causing the grammatical rules of that language to be constantly parsed. Triggered rules are then fired in accordance with some means of conflict resolution. The system has a means of running more than one of these languages at once, a primary language or operating system might be called a language of life, and runs more or less permanently, although it might have more than one language to deal with differences such as sleeping or waking. Additionally other minor languages are swapped in and out to deal with small situations. Consider going to your favourite restaurant, everything is familiar and you go easily to a table having recognised the words and gestures of the maitre D . compare this to going to an unfamiliar restaurant, there is no exact language available to you so you fall back on the nearest substitute and make minor faux-pas. Now consider the person who has never been to restaurant before, near panic nothing even remotely fits, and finally watch the Michelin Guide examiner go to a restaurant, he has a language for handling unfamiliar restaurants and so it is no big deal!
Over and above these small languages for specific situations there are major ones loaded up, and capable of being reloaded remarkably quickly, notice how quickly someone you know only socially changes when they enter their work environment in your company, and the changes are often of a gross (large) nature, amounting to almost a personality swap. The inference here is that there is only room for one of these major languages, possibly because if two were trying to run at once there would be huge conflicts. Similarly if a language got damaged then the results might be catastrophic, the evidence from stress related and mental illness suggests that languages even after extensive learning never become fixed but are constantly updated in the light of input data and may well go wrong if subjected to adverse inputs. It is also conceptually possible to think in terms of the affect of a damaged inference engine.
The purpose of this text is to outline a hypothesis which may have far reaching effects on all aspects of human and machine intelligence, and models of these. On the other hand it might not.
(c)C. J. Barnes 2000
~~---oOo---~~
