An Expedient Mind: REPRESENTATION

AN EXPEDIENT MIND: ISSUES

Tartan Hen Publications : Home | more books | Contact : feedback@tartanhen.co.uk

A NOTE ON FORMS OF REPRESENTATION

Much of this text is concerned with the topic of representation. But before I describe how things migght be represented in my system, I want to make this point strongly. The actual format of a representation does not matter at all. It does not matter whether it is couched in the language of mathematics with funny Greek letters and squiggly symbols, whether it takes the form of verbose descriptions in linguistic form, takes “clausal form”, is constructed by drawing on sand or appears in the form of data structures (as I have used here). What matters, is how a representation is processed - how it can be manipulated - how it is used to trigger and predict further events.
That is What intelligence “is about”. It is about the prediction of future events, even if (or perhaps especially if) those events are only a few seconds in the future. I make no apology for frequent repetition of that assertion. It is fundamental to my thesis. What I mean, is that an intelligent system should be able to take a representation of the current state of affairs (plus representations of past events) and use that information to generate a representation of a predictable future state of affairs. Apply that idea to the signals received by the emergency control centre. The “meaning” of those signals, or the “interpretation” placed on them by the operators, can be understood in terms of what would be observed by someone who got into a fire engine or whatever, and drove to the scene of the emergency. Thus a fire alarm can be regarded as a prediction that if someone responded to the alarm by travelling to the scene, they would see flames. A poison gas alarm means that the person would be well advised to take a gas-mask.
One representation is used to generate another. That is the essential characteristic of an acceptable, or useful form of representation. It can be manipulated to create a prediction in the form of yet another representation. That is possible only if the different forms in which a representation can be presented, are linked by rules of inference. These rules will take the form ...

If
(this representation) is accurate
then
(that representation) is also accurate.

And that is exactly what the rules of inference in the indefinite calculus are for. We will also come across other rules of inference, which express the relationships between other structures.

Note: The philosophical position known as ‘representationalism’ is a current topic of animated debate (see for example Lycan 2000). The form of representationalism I am putting forward here takes, much of the force out of a traditional objection which has been raised against the idea. Briefly, that if representationalism is correct, the mind would never be able to escape from ‘the world of ideas’ (as it was called by many philosophers). My response is that indeed the mind cannot escape, and that it has no need to escape. It is not even clear that it has anywhere to escape to. It can, however, serve the purposes (in the evolutionary sense) of the mind’s owner perfectly well by operating entirely within the representational world. I shall return to this topic in the discussion.

Tartan Hen Publications : Home | more books | Contact : feedback@tartanhen.co.uk