This chapter addresses a number of subjects related to learning and thinking from a philosophical perspective. These discussions help to put the subjects of thinking and learning in context.
If any attempt is to be made to model intelligent behaviour then an understanding of what is 'intelligence' becomes critical. There are many books and articles that describe intelligence and all of its many facets. Intelligence is a general property or characteristic that we bestow on an animal when it shows certain behavioural traits. We may then use the word as an adjective e.g. an intelligent person.
Intelligence is not absolute. It is not black or white; one does or does not have intelligence. It exists at many levels. There are many degrees of intelligence. It is also specific to a subject area. I am relatively intelligent with respect to mathematics but in music my degree of intelligence is very poor. A person or animal can also be very knowledge intelligent i.e. knowledgeable, in a subject area while showing a low degree of skill in the same subject area. I may know copious amounts of facts about snakes but when it comes to handling them I may easily be bitten. Or visa-versa, I may be very skilled in handling snakes and have the knowledge to perform this task but know very little about snakes in general.
An animal that would normally be called intelligent exhibits behaviour that is based on one or more of the following capabilities:
- Recognition / pattern matching
- Learning / adapting / goal seeking
- Remembering / recalling
- Thinking / Reasoning / conceptualizing
- Communicating / speaking in some language / signing
- Responding / performing complex actions
But not all of these abilities are necessary to be intelligent. The difficulty is that the required subset is unknown. Communication is an example of an ability that may not be necessary. It is unfortunate that we have often erroneously assumed that any animal that can not communicate with us is not intelligent. This is most likely because we find it difficult or impossible to judge the intelligence of an animal without communicating with it. We are beginning to realize that other animals such as chimpanzees, dolphins, elephants and horses just to pick a few, all exhibit degrees of intelligent behaviour even though their ability to communicate with us is limited or non-existent.
The two abilities from the above list that I feel are most necessary for intelligence and lie at its core are the ability to learn and think.
Learning occurs when an intelligent animal is placed in a novel environment and it exhibits behaviour that can be interpreted as being in the pursuit of some set of goals.
Only recently have animals been studied from a thinking perspective. Ants and birds can learn and they communicate with each other effectively but we rate them low on the intelligence scale. We usually try to judge the intelligence of such creatures based on their behavioural characteristics. For example, do they have goals? Do they use tools? Can they use reasoning to solve a problem? Can they understand and remember concepts? Do they use these concepts to aid in learning?
The ability to produce models and use them is one of the most powerful human skills. A model is a representation of something real. A small sample of models that exist in nature or that we produce and often use without thinking are:
- A model aeroplane
- An architect’s cardboard model of a building
- An architects drawing of a building
- A blueprint for manufacturing an engine part
- A recipe for cheesecake
- An equation describing the laws of gravity
- A detailed written description of an apple
- The instructions for opening a combination lock
- A painting of an apple
- A road map for a city
- A photo of an apple
- A computer simulation of the traffic in a city
- A theory of how the sun works
- An organization chart for a company
- The chemical formula for citric acid
- The DNA in an apple seed
A model is captured either in the same medium as the original such as doll house furniture made from wood or it is captured in some other medium such as a model aeroplane made of plastic. If our medium is a solid substance such as wood or plastic then theoretically we can model the shape of any other solid object using this medium. The same applies for liquids and gases. But models don't have to be the same shape as the real thing. Their appearance and medium all depend on what aspects or features of the original are being modelled. Often models are purely information and we capture them on paper in some symbolic form, such as the map of a city. Using information as the medium for modelling is more flexible than using a solid, a liquid or a gas. More aspects of the original can be captured.
Most of the time it is not possible or desirable to capture all the aspects or features of the original thing. If one did capture all its aspects the result would be the real thing. A model is said to be an abstraction of a real thing because only those aspects that are of interest are captured. The reasons why models are so powerful include:
- a model costs less and takes less time to build than the real thing because only those aspects of interest are captured.
- a model is usually smaller than the real thing which aids in its use.
- we can experiment with a model and perform tests on it that might not be possible without damaging or destroying the real thing. This only works though if the model accurately represents the aspects of the original that are being tested.
- it represents our understanding of those aspects of the original that are of interest.
- we can prove facts about the original by using the model, provided it is an accurate representation of those aspects involved in the proof.
- we can take the model apart and show others its internal structure and how it works.
- If the right modelling medium is chosen the model can be transformed into another model that may be more useful, easier to test, prove facts about or demonstrate.
- A model can be copied inexpensively and given to others.
It is not only touchable, tangible, three-dimensional things that can be modelled. Things such as processes, events and relationships between things can also be modelled. An organization chart or a family tree is a model of relationships between people. The instructions for assembling a toy or a recipe for baking a cake are models of processes. A contract is a model of an agreement made between two parties and an itinerary is a model in time order of a number of events. Notice that all these things require an information model for their representation because they are not tangible, touchable objects.
There are both static and dynamic models. A static model does not change or operate in time. An architect’s model of a building and a plastic aeroplane model are quite static. They are meant to model only static aspects of the real things. However, a balsa wood model aeroplane that has a motor, remote control and can be flown is a dynamic model because it can simulate the behavioural as well as the structural aspects of the real thing. Certain information models such as a sequence of instructions to bake a cake can also become dynamic. However, to do so requires something or someone to interpret the steps. For example, a recipe becomes dynamic when you become the interpreter and follow it to bake the cake. The recipe does not model the cake; it describes the process required to create the cake. And when you interpret and perform the recipe you are creating an instance of the process. The process is the thing that the recipe describes. So this is a unique feature of information models of processes. This feature does not apply to information models of objects, events or relationships.
A computer program is also a model of a process because it captures in a sequence of instructions a description of some steps to follow to produce some kind of behaviour or result. The computer that interprets and carries out those instructions brings to life the process being modelled. This becomes more complicated when we realize that there is also information involved that the computer program manipulates. For example, a computer database that keeps track of employees in a company does not contain employees; it contains a static information model of the employees and their working relationships. It models those aspects of employees that are important from a company perspective. The computer software that is interpreted by the computer describes the operations to be performed on the employee database. And when the software is run (interpreted by the computer) the employee model is changed and its accuracy maintained. In other words, the software models a process to perform the update of the employee model exactly as the employee’s changes take place in reality.
A model also is a real thing. We can touch it, use it and examine it. This means that we could build a model of a model. This would mean modelling the types of components that comprise the model. But this does not seem to make much sense. Why would you want to model a model? For the narrow class of information models of processes it becomes very useful.
Self referential models - models of self - An example is a compiler written in its own language or a data model of data models. People watching a movie of people watching a movie of people watching a movie etc.
Consciousness and free will are the results of the power of self-referential systems to model themselves or to model some aspect of themselves. In the case of consciousness we have the ability to model ourselves and compare the observations of ourselves with our model of ourselves. We can examine the reasoning used for making a decision and decide that a better decision making process is possible. Consciousness is also self-awareness. You have a model of yourself as a being and also a model of your mental activity that you can use to predict what will happen when. A mental model is not a complete description of oneself. Our memory contains a model of our experiences in the real world plus a model of the mental activity we performed. It's the latter that allows us to perform self-reflection and have self-awareness.
Before the physics of electromagnetic radiation such as light was fully understood, there were many debates in the scientific community about the nature of luminescence, which is the property of some things to be bright and others to be dark. Similarly the true nature of life was not understood until the underlying chemistry and biology was discovered. Our current understanding of consciousness is rather like those early states of understanding of luminescence and life. It is a complex phenomenon that results from the nature of the underlying processes. And once those operations are better understood, we will spend less time dwelling upon the nature of consciousness.
Consciousness is a noun like whiteness. A noun made out of an adjective. Anything that can be described as white has whiteness and anything that can be described as conscious has consciousness. This is the same grammatical structure explained between Intelligence and intelligent. An object that is conscious has and exhibits the properties that are attributed to consciousness. These properties include:
Awareness vs self-awareness, sentience, self referential, will - ability to make decisions.
Survival instinct - an animal would only be thought of as being intelligent if it had some interest in continuing to exist. The fact that they continue to exits makes them interesting. But continuing to exist is not the only criteria. A rock exists for a long time but that’s not interesting. If it survived being melted and then hardened again, that would be interesting. Can an object be called intelligent if it has no purpose? Can it be conscious and have no purpose?