Intelligent Systems And Their Societies Walter Fritz



Elementary Concepts
A concept is the basic element of thought. It is a physical, material storage of information (in neurons or electronics).

Socrates, a great philosopher in ancient Greek, stated that ideas (concepts) are immaterial and eternal. But how does that explain concepts created in the last 100 years, such as "TV", "NATO" and "computer"? How does it explain all future concepts? They could not all have existed for all "eternity".
An IS creates concepts by the processing of what its senses tell it about its environment. Study has shown that all ISs, including humans, perceive their environment as: In another form, we can say they perceive (spatial and temporal) structures and transformations. The structures include objects and their relationships.

The sense organ is the first step in the process of obtaining this information. Later, when the brain receives this sense information (as nerve impulses) it processes it in such a way that it notes spatial and temporal relationships between some of the impulses. In the case that this relationship is similar to previously received information, it assigns the previous concept to the data. In the other case, it creates and assigns a new concept. These are the most elementary concepts.


Built-Up Concepts
Later, the brain of an IS uses these elementary concepts to build up higher level concepts, also called composite concepts. The mental processes then make use of these elementary and composite concepts.
All the concepts in memory are interrelated, they form a web, a net.

Total and parts
Concepts have links to those higher level concepts of which they themselves are parts. A wheel is part of a car; the concept "wheel" has a link to the concept "car". The concept "car" in turn is part of the concept "vehicle", and has the corresponding link. Concepts also have links in the other direction, namely to their parts. Naturally parts can themselves have parts. (The concept "tree" has links to its parts, to the concepts "roots", "stem", "branches" and "leaves").

Abstracts and concretes
Another set of links is to what we call more "abstract" concepts and more "concrete" concepts. The concept "tree" has a link to the more abstract concept "plant". We can also say a tree is an example of a plant. The other way around concepts also have links to their more concrete concepts, to their examples. The concept "animal" has a link to the concept "mammal" and the concept mammal has a link to the concept "mouse".

When the brain has to decompose a concept it can decompose it by using the links to parts. Here it would use all the parts in place of the concept, for instance in place of "fruit" it would use: kernels and flesh and skin and ...
or it could decompose by using any of the concepts in its link to more concrete concepts: For instance, the abstract concept "fruit" would be replaced by: apple or banana or orange or ...

Note that elementary concepts have no links to concretes. They are already the most concrete information that the brain possesses. In place of links to parts, they have detailed information of the sensation or elementary action.


The Concept - "Thing" Relationship
The relationship between the objects of an environment and concepts the IS uses to represent them is not easily understandable. Everyday experience does not show us the processes we use to think and speak. Let us, therefore, explore this in some detail.

To clarify this process, let's observe the human mind. We commonly say, for instance, that we see an apple there on the table. We say the apple is there and we can point at it with our finger. But is it really? Or are our eyes so perfect that they lead us to confuse the thing itself (the object on the table) with the concept (the representation) of the object?

This is a question that has been addressed many times in the history of IS on Earth. So let's sit back and try to think carefully about this for a moment. We will start with the "thing itself" out there in our environment and follow what happens in the course of our receiving information about it and converting this information into a concept. (The "thing itself" is a concept coined by the German philosopher Immanuel Kant; 1724-1804)


The Process
From this "thing itself" that we can "see" somewhere out there, electromagnetic radiation moves out in all directions. Some of this radiation reaches our eyes. The first layer of the retina of our eyes converts part of this radiation (the "visible light") into nerve impulses. These nerve impulses are then collected by other cells that form a second layer right behind the first layer of cells of the retina. These second layer cells and further layers process the nerve impulses, the communications, received by the first layer. (For instance, they detect and isolate edges.) This we already call "information". Now the eye sends all the nerve impulses to a certain location in the brain, which is dedicated to sensory data. The brain combines the different nerve impulses that communicate information about color, shape, and other aspects to form a coherent mental image. It is here, for the first time, that the brain tries to gives this lump of information a designation, a label. This designation is what we call a concept and consists of a pattern of excited and quiet neurons in a biological brain and a number in an electronic one. The content of the concept is the lump of information it refers to.

When an electric impulse is given to a point of the surface of the brain (visual cortex), we see a remembered object or a complete scene. The corresponding concept is activated. When a part of the brain is missing (by an accident), we cannot remember some things or some actions. The neurons that contained the information (the concept) does not exist anymore.

In linking and compacting the information that it stores, the brain creates (or recognizes, if known beforehand) that part of the concept for "apple" that indicates, for example, how an apple "looks". Later, with further experiences about apples, it adds other information parts to the concept for "apple". For example, it may add "skin information", and "meat", and "kernels" even though they are not always visible when an "apple" is identified. Likewise, it eventually also adds that an apple is edible, what tastes it has when ripe and when not ripe, how big it normally is, and so on. To all of this it also adds that it can communicate (in English) about the apple with some other IS by using symbols such as the letters a-p-p-l-e.

The concept, the label to this cumulative information construct, is what we use during the process of thinking about an "apple". This concept is actually a certain distribution of nerve impulses in the biologic brain and an (essentially binary) number in an electronic artificial IS.


The Concept is NOT the "Thing"!
This understanding of the relationship between "thing" and its concept is embodied in the following statement:

"the concept for an apple is not what is there on the table!"
Think about it. The concept is actually a material structure, a grouping of information in our mind. The information that an apple is edible and has kernels is not there on the table; it exists only in our mind. Still, there is something on the table.

This some "thing" is the "thing itself". It is the "thing itself" that has produced what we call "electromagnetic waves" and which resulted in part of a concept. What we "see" is this part of a concept, the result of the visible part of these electromagnetic waves. We do NOT see an "apple". An "apple" is much more than only the effects of visible electromagnetic waves of a certain type hitting our retina. When we "think about an apple" we are really thinking with (using) the concept "apple" (a data structure), which exists ONLY in the brain. We are not "thinking" with the "thing itself" (physical object) that is in our environment.


Let's experiment with "Reality"
Look at any object with one eye and press lightly with your finger on the side of that eye lid. You will see that the object jumps wildly. Obviously, the "thing itself" does not really jump around in front of us; it only appears to do so. We can explain this phenomenon by stating that during this experiment, light fell on different parts of the retina due to the deformation we made with our finger in the eyeball. Because of this deformation, the nerve impulses that we interpreted as the image of the object came from different parts of the retina. Thus, we see that the object moves, even though we know that this cannot be so.

This experiment shows something about the character of our sense and thinking organs. Sense organs receive communications (sometimes in the form of electromagnetic waves) and they transmit information (nerve impulses). The brain receives only these nerve impulses, and not the objects themselves nor the communications that they have given off. In the brain, these nerve impulses are summarized into concepts. Some of these concepts are related to shape, some to color, and others are related to other aspects. By comparing these concepts with previously stored information, the brain (usually) finds a match and concludes that the information came from some "thing" that it had previously labeled with a particular concept (or if no previous concept existed, it created a new one). With these concepts it builds up the present situation, and it is this situation that we "see", not the thing itself.


Communication of Concepts
For the purpose of simplifying communication, the brain also attaches a unique sequence of letters (a, p, p, l, e) to each new concept. This sequence of letters designates how we communicate this concept when using written communication media. We also store a sound sequence that we use when speaking about the concept. We must remember, however, that these two communication concepts are only part of our composite concept. That is, the letters and the sound are not the concept itself.

We can expand on this by showing that we use "concepts" and not words when thinking. For this, let's work with an example situation: Occasionally, when we are attempting to communicate something, we experience a difficulty that involves a situation where a concept is clear in our mind, but we are momentarily lacking the corresponding word that we need to communicate this concept. A phrase we English speakers often use for this particular difficulty is "we have it on the tip of our tongue". It can be reasoned that if we were thinking with words, the words we need for communication would be present: We would not have to look them up.

When you are thinking, you often express conclusions with silent words. You remember this conclusion and use it as a starting point for further thinking and a new conclusions.
You remember the conclusion, because sensations and actions are remembered, and speaking, even if only silently, is an action. You cannot remember the thinking which is done with concepts and is quite unconscious.

We have to distinguish between:

  1. the partially knowable "thing itself" on the table,
  2. the concept which we use in our brain while thinking,
  3. the word we write or speak, to communicate about the concept with others.

Thus, we have seen that concepts are not something that exist in our environment, its structures, or its transformations. Instead, concepts are something that exists only in the brain of an IS.


Using Concepts To Understand Environment
Above we saw that the connection between the brain's concepts and their source in the real environment is quite tenuous, quite loose -- in fact, much weaker than we would like it to be. This means, for example, that an IS cannot say that its environment is such and such. It can only say that it can view signals from its environment and attempt to link, store, and express them as a series of concepts. Amazingly, this is true, both for artificial and natural ISs.

This leaves us with the question of whether or not these "concepts" are sufficient to enable an IS to really understand its environment. The importance of this question is linked to the discovery that it is this understanding of the environment that, to a great degree, limits the amount of intelligence that a system can achieve.

It turns out that we can set up a way for an IS to test itself on its understanding of the environment. For instance, let's start with an IS that can manipulate concepts in its "imagination" and observe the results. If these results seem desirable, the IS may then put its limbs or other actuators in motion to repeat this imaginary concept manipulation in the real world. Then, if what it views of the result of these environmental manipulations is similar to what it found to result in its imagination, we can say that the IS understood its environment. On the other hand, if the viewed result is quite different, the IS certainly did not understand its environment, or at least did not understand enough of its environment.

For continuous reading, like a book - continue here.
Jump to the top of this document / Details of IS / Intelligent Systems /
A Scientific Philosophy / E-book Contents.

Last Edited 12 Nov. 2014 / Walter Fritz
Copyright © New Horizons Press