Ideas and Theory - Animat - Behaviour, Needs, Innate Reactions, Innate Reaction Adaptation, and Novel Reactions

When one considers my definition of intelligence, behaviour is not implied.  A medium does not have to be behavioural to be intelligent.  Instead, it is the other way round - a medium must be intelligent before it can be successfully behavioural.  Why?  Consider the equation of processing o = f(i).  In section 1.1.1 I said that for a given i, we should be able to determine the performance of function f() by considering o.  Well, behaviour and animat / environment interactions directly demonstrate that performance, by translating i into stimuli, and o into reaction - it is the nature of these reactions that will directly represent the efficience of the function f().

This means that, if I could endow the Farve iModule with behaviour, the nature of that behaviour would be a perfect representation of the degree of intelligence of the iModule.  It would be a convenient (and biologically well-known) means by which the performance of the iModule can be judged.

So, how would one go about making an intelligent medium behavioural?  This section deals with that.

What is Behaviour?

Behaviour is defined as all of the actions that an intelligent medium performs.  These include everything from yawning to eating to looking to hunting to sleeping.

The above definition is true but rather simplistic, though.  I define 3 kinds, or levels, of behaviour - innate behaviour, adaptable innate behaviour, and learned behaviour.  These 3 kinds of behaviour builds one on the other.  It can be seen with lifeforms of differing evolutionary status: very simple creatures, like Hydra, are totally reflex-based.  They can not learn, but they have evolved in such a way that their innate reactions to stimuli, their innate behaviour, complements their environment.  More complex creatures, like flies or birds, (interresting fact - earthworms can learn), have an additional ability - they can learn.  Their innate behaviour is complemented by higher neural centers that can influence their innate behaviour, according to past experience.  And finally, even more complex creatures, like monkeys and humans (and I suspect many more, like dogs, cats..) can come up with totally new behaviours, that is not innate, but that they realize can help them in some way.  At this extreme, the brain can totally override many innate behaviour - some trained people can, for example, press sharp metal bars through their bodies, or make their metabolisms slow down considerably.  Someone who gives his/her life for another is also an example.  All of these things in fact goes against the action of innate behaviour, which tries to protect the individual from harm.

Needs Drive Behaviour

To get tips for how to make the Farve iModule behavioural, I will start by looking at the first level of behaviour, as found in all behavioural biological lifeforms.  This is innate behaviour.  What causes them in these biological lifeforms?  If I can find this out, I will be closer to being able to endow the same to Farve.

Let me first discuss the functions of these innate behaviours.  Innate behaviour is nothing but predefined (by nature) reactions.  All of us know that our ‘base reactions’, like closing our eyes when a bright light is shone into them, or our ‘escape responce’ from a very warm object, for example, are designed with one goal - to protect us.  Actually, preserve would be a better word that protect, because drinking and eating behaviour also falls under innate behaviour.  It is more or less under our control, but as hunger and thurst increases, these actions become more and more stereotyped, more and more innate.  Have you ever seen a normally-hungry high-class man eat, with neat utensils and postures and actions?  And have you ever seen a very hungry man get a plate of food, and seen him eat, with a very simple range of actions - take food - move to mouth - chew briefly - swallow.

We have come to the crux of behaviour - behaviour is actions designed to fulfill needs.  This definition holds up on all three previously mentioned levels of behaviour.  We do things, because we need to.  ‘Want’ is also a need.  Level 1 behaviour (innate behaviour) exists to try to satisfy needs.  Level 2 behaviour is level 1 behaviour with added flexibility (because the environment is not static, but always changing - unpredictable).  Finally, level 3 behaviour is more ‘personal’ - the entity may want to learn chess, or want to watch babylon 5, because he/she finds pleasure in those actions.  These behaviours are not nescessarily linked to biological needs, but they are needs nonetheless.

It is now clear that, to program an intelligent software entity, that entity must consist of not only a nervous system, but also some form of a metabolism to drive needs, to in turn drive level 1 behaviour.