excreationist

I think I'll change my mind on my ideas about information...
Just some questions - say you learnt someone's name and this became encoded in some neurons, would you say that those neurons contain that information? And is the "meaning" of those neurons the name of that particular person?

[ March 12, 2002: Message edited by: excreationist ]</p>

excreationist

I think this is the lowest form of intelligence - normally they don't learn new behaviours/problem-solving strategies by themselves - they just follow precise instructions. I think the machinery in a cell is like that as far as it processing the DNA goes.

It takes many years for humans to learn to understand things. And I believe that one of the fundamental human emotions is to seek "newness" and "connectedness" and so seek to explore the world and develop a coherent understanding of it.

BTW, you might be interested in this thing I'm working on:
I think that we *learn* to understand all the implications of things. And as that hierarchy showed, there are different levels of understanding.

Exactly... but I'd still call it intelligence...

Automaton

Since he defines "intelligence" in a way that included mindless molecules, the conclusion that it must have been "intelligently" created is merely word trickery. This proves nothing (as usual) about the actual ID hypothesis.

Vorkosigan

mturner's argument is that DNA is a symbol. Nonsense. DNA is a set of chemicals. Which we can represent by symbols. mturner has committed the fallacy of mistaking the symbol for the thing itself.

DNA is a thing-in-itself. A symbol stands for something. AGCT, that's a symbol.

Michael

Malaclypse the Younger

What is necessary in this discussion is to rigorously define "information", "meaning" and "intelligence" in an objectively determinable manner. Once these definitions are in place, it becomes possible to speak precisely about the possibility of intelligent design. Obviously, many definitions of these words are possible. But to be useful, a definition must be precise.

One good definition of "information" is determinable multi-valued ordered physical states of some system.

The states must be determinably multi-valued. It must be possible for parts of the system to be in one of at least two alternative states. For instance, information on your phone wires is expressed as different voltage states, in your computer by different physical states of the various transistors.

The information must be determinably ordered, either in space or in time. Again, in your phone line, the information is ordered by time, in your computer by space.

That's it for information. By applying some determinable state and ordering rules, any physical system can be turned into "information", from the electromagnetic background noise, to the arrangement of stars in the universe.

It is obviously clear that DNA contains information. The "state" is the presence of a particular amino acid which can have one of 4 possible states. The ordering is obviously spacial and linear.

Meaning can be defined as correspondence between some collection of information and some physical or abstract process outside that collection. For instance, the information on your phone line means (physically) various ascii characters--the characters together mean (abstractly) various words, etc. Clearly electromagnetic background noise or the arrangement of stars doesn't have meaning--there is no extrinsic physical or abstract system to which it corresponds. Of course, one could invent an abstract system to which they correspond (e.g. the constellations), but it is clear that the assignment of meaning is arbitrary and subjective.

Now it is clear that DNA does have physical meaning: The arrangement of nucleotides corresponds to specific arrangements of proteins.

The definition of "intelligence" is rather more difficult. One might simply declare that any system that manipulates information to create meaning is "intelligent". This definition leaves us with the interesting conclusion that the naturalistic process of evolution is itself intelligent, because, through RM&NS, the information in DNA is manipulated to be meaningful in terms of proteins (and subsequently organisms).

So, in a sense, "Intelligent Design" is obviously true--naturalistic evolution is itself an intelligent system that tends to "design" DNA to produce proteins which produce viable organisms.

[ March 12, 2002: Message edited by: Malaclypse the Younger ]</p>

owleye

Scientiae....

I think it's time to bring in Donald Davidson's paper "On What Thought Requires." It is illuminating because it tackles the difficult task that many folks (apparently including mturner) have either not confronted or played on the ambiguities inherent in 'intelligence'. Issues regarding information and meaning in this context are merely side-shows, that pretend to be relevent on the basis that it looks like there is some connection.

I have the article and I'm only too willing to draw from it, but let me try to encapsulate the outline of it that my instructor developed, adding my own interpretation as necessary.

1. What the problem is.

The problem for Davidson is to "determine what would turn calculation in the sense in which a fly or a computer can calculate into conscious thought."

My interpretation of Davidson would be to replace "calculation in the sense in which a fly or a computer can calculate" to "discrimination in the sense in which a fly can discriminate or calculation in the sense in which a computer can calculate." This interpretation is required, I think, because in some sense Davidson believes that human thought can be considered a combination of having the ability to discriminate sensory objects and perform complex calculations based on them.

2. First Thesis

"The thought of the thinking being must possess a definite structure, the structure of a language containing familiar features."

Animal behavior, despite its signficant capacity to discriminate, does not demonstrate possession of this structure.

To be able to discriminate requires, to be sure, a kind of concept that is able to carry out this function, but this kind of concept is quite inadequate to characterize how thinking beings use concepts. Concepts, in order to involve thought, require:

a. knowing something about their object -- what makes it what it is.

b. knowing how to fit a concept "into a complex conceptual scheme in which our concept of an x has logical and other relations to other concepts"

c. and speech is required to have such a conceptual framework

d. Thoughts and thoughts alone have logical relations (such as truth functional connectives) with each other.

e. An interpreter must notice the logical patterns among utterances which a speaker accepts.

f. Thoughts and thoughts alone also have evidential relations with each other.

g. An interpreter must also notice the evidential patterns among utterances which a speak accepts.

h. Speech behavior can be creative such that (1) predicates and demonstratives can be applied to an unlimited number of new objects; (2) truth functional connectives also have unlimited iterative capacity.

i. Speech must reveal the speaker has the concept of an object, and not just the ability to discriminate one object from another by: (1) the use of pronouns and cross references, quantifiers and variables; (2) must refer to different sorts of object, objects, to the obvious properties of objects, and to the various changes and activities of objects; (3) must refer to events, causes and effects.

Four other theses are given in the article but I will leave off for now, to see what reaction I get from it.

owleye

John Page

Scientiae:

Excellent definitions from Malaclypse's post and, given those definitions, I agree with his analysis as it relates to an intelligent system.

However, maybe we should differentiate between an "intelligent system" and an "intelligent being". Perhaps we're back to the Turing Machine experiment - arguably any system that produces a consistent result for a given input could qualify as an "intelligent system".

Attributes of an "intelligent being" could include consciousness, awareness, purpose, feelings etc. which differentiate it from a "mechanical intelligence". Unfortunately we don't really understand how our minds work, making definition of human intelligence an art rather than science. With this in mind I can only suggest that until mturner can prove that DNA can sense its environment (as opposed to just react), process that sensory information, posess and recall knowledge, make decisions and act on them in a conscious manner etc., his labeling of DNA as "intelligent" is unjustified and misleading. Anthropomorphism is misleading, it confers upon non-human entities qualities they may not possess.

Cheers.

excreationist

I did post my fairly comprehensive framework about the hierarchy of intelligence... oh well...

Well it is the <a href="http://www.microbiology.adelaide.edu.au/learn/tables.htm" target="_blank">triplets/codons</a> of basepairs that are symbols. It is also true that the letters "A, G, C, T" are symbols. They refer to the bases in DNA... but triplets of bases in the DNA are decoded as amino acids...
So they can be seen as symbols.
In a similar way, computers are said to manipulate symbols. But in silicon computers there are just electrons and transistors. We can say that 1's and 0's are involved - that is another symbol - but apparently computers still manipulate symbols even if we aren't there describing them.

It is the bases that can have one of 4 possible states... and triplets of basepairs (codons) are translated into 20 amino acids (and maybe more).

Malaclypse the Younger

John Page

Thank you.

Well, this is not quite my point. According to my definition, an intelligent system would have to create physically meaninful information. The canonical Turing Machine lacks any way to establish physical meaning, because it lacks inputs from the physical world.

Well, a gene pool can "sense" the environment through differential reproduction. It "processes" that knowledge by "remembering" what sequences produced reproductive advantage (by surviving) and "forgetting" which sequences produce reproductive disadvantage. It makes "decisions" by producing (or not producing) particular proteins. By all the objectively determinable criteria, a gene pool acts in an "intelligent" manner.

Whether a gene pool is "conscious" would depend on the definition of consciousness--a definition I would not care to attempt.

This is true, which highlights the need for objective definitions.

Malaclypse the Younger

turtonm

Well, mturner is still an idiot, but there's a reasonable definition of "symbol" which does make DNA obviously symbolic. A symbol is just meaningful information--information which corresponds to a physical system.

mturner's fallacy is to assume that symbolic information is indicative of an extrinsic intelligence. However this is obviously false under this definition. Our specific neural structures are symbolic, but they are not evidence of an intelligence extrinsic to our brains.


excreationist

Details details. I'm a computer scientist, not a biochemist.