Can computers engage in concept-formation?

[amagi]

Hmmm, so I guess the conclusion is that consciousness can arise even from things that only do what they're programmed to do, like cells, because consciousness is an emergent property. Interesting.

AshRyan: argument refuted.

[Bowzer]

Cells only ever do what they are programmed to do.

Although I do not support this use of the word "programmed" in this context, Meriam-Webster seems to agree with you, amagi. So I will have to disagree with the both of you.

I believe that this use of the word is improper because it implies intentionality. A programmer intentionally constructs a program to do certain things. Since there is no god to program things in nature, there is no intention to be found there either. Even if I agreed that you could say that cells are "programmed" to do certain things, there would still be a fundamental difference between things that were programmed by evolution and things that were programmed by humans.

If you aren't familiar with Objectivism's distinction between the mand-made and the metaphysically-given, amagi, I can provide some references. It has everything to do with the claim that you are making and also explains the mistake behind your argument.

[Ed. Note: Added last paragraph]

Edited May 11, 2004 by Bowzer

[amagi]

Even if I agreed that you could say that cells are "programmed" to do certain things, there would still be a fundamental difference between things that were programmed by evolution and things that were programmed by humans

Why?

Obviously there's a difference in how the programs came to be there, but what's different about the programs themselves?

The genetic code of a cell is a program. There's no better word for it. (A [mostly] digital program no less!!) And cells, being machines, can only do what they are programmed to do.

The programs of cells make them goal-directed, but there's absolutely no reason why humans can't program a machine to be goal-directed as well.

Are you suggesting that a naturally-occurring program is inherently less deterministic than a man-made one? That cells can ignore their program and do something different every now and then, act a little non-deterministically, and that this is what allows them to give rise to consciousness?

The fact is, there's nothing inherent in the fact that a program is natural that makes it any less a program. (Moreover, whatever attributes of that natural program allow it to give rise to consciousness will ultimately be defined and understood by man. Why couldn't we then create a program with those same attributes?)

Bowzer, notice that AshRyan specifically removed the issue of human intent from the equation above. That's crucial.

It's not the source of the program that matters in this context. It's the nature of the program itself.

If anything, the fact that human intellect is being brought to bear in the programming means that, given time and enough scientific advancement, we will do better than anything in nature.

As for your references, feel free to give them.

[Bowzer]

After thinking about it some more I'm not sure that the man-made vs. metaphysically-given distinction is as much of an issue here as I previously stated. I was persuaded when I thought of synthetic drugs. This is an example of something man-made that is causally equivalent to something found in nature. You are making an argument from causailty, amagi, so I believe that I was mistaken to bring it up.

The distinction would be more critical in making an ethical judgment. Chemical A in nature is toxic to humans. Being a part of nature, it would be improper to evaluate Chemical A as evil. Chemical B on the other hand, is a synthetic reproduction of Chemical A and is just as potent. It was created in a lab and will be used to exterminate people. It is proper to evaluate Chemical B as evil. That is the distinction in action.

I still think that you are wrong on many levels, amagi, but I am not getting into all of the problems that I have with your position as I've done this elsewhere.

To learn about the distinction between the man-made and the metaphysically-given see The Metaphysical Versus the Man-Made The Ayn Rand Letter Vol. II, No. 12 March 12, 1973 and the section The Metaphysically Given as Absolute Chapter 1 of OPAR pp.23-9.

[Betsy]

The programs of cells make them goal-directed, but there's absolutely no reason why humans can't program a machine to be goal-directed as well.

Goal? Whose goal? The human programmer's goal or the machine's? It would have to be former because the latter isn't alive. Therefore it doesn't have a "What for?" of its own. See "The Objectivist Ethics."

Are you suggesting that a naturally-occurring program is inherently less deterministic than a man-made one?  That cells can ignore their program and do something different every now and then, act a little non-deterministically, and that this is what allows them to give rise to consciousness?

If you are discussing a conceptual creature like us, we are self-programmed. Our hardware gives us that capacity. As to how it does, that is yet to be discovered. As far as we know, lower animals, in addition to looking different and acting different than us, lack the capacity for volitional, conceptual thought.

The fact is, there's nothing inherent in the fact that a program is natural that makes it any less a program.  (Moreover, whatever attributes of that natural program allow it to give rise to consciousness will ultimately be defined and understood by man.  Why couldn't we then create a program with those same attributes?)

I suppose that could be done by copying natural systems just as we synthesize biological chemicals, but it wouldn't work unless it was imbedded in a living thing whether natural or man-made. The reason why is that conceptual thought is not limited by perceptually available data the way lower levels of consciousness are. At any given moment, there are millions of things you could choose to perceive, identify, abstract, integrate, evaluate.

What determines what you actually have in your crow-limited conscious awareness? What directs your thinking? YOU do. Why do you make the choices you do? Your purpose. Only living things have -- or need a purpose.

[stephen_speicher]

I believe that this use of the word is improper because it implies intentionality.

Yes, that is exactly right. Physical mechanisms simply act, producing a certain result. Period. It is a little like Ayn Rand's comment that clocks do not measure time, they simply produce a certain kind of motion. Biological mechanisms obey the laws of physics and chemistry, and any "programming" is a term used primarily by laymen and the novice. When you study the molecular biology of the cell you study the inner workings of the cell and its interactions, and any sense of "programming" would only arise as an extremely loose metaphor. DNA is the repository of genetics, not a program.

But, with that said, there is an interesting context in which programming and biology truly do unite. Several years ago I reported on the prospect of molecular-scale computers. Since that time there have been a number of interesting developments, but, by far, the most significant one will be reported on in an upcoming issue of Nature. I have read a preprint of the paper "An autonomous molecular computer for logical control of gene expression," Y. Benenson, et al., and this is really fantastic.

This group built a biological computer in vitro which has a computation module, an input module, and an output module, all three programmable. To prove the concept the biological system was programmed to identify and analyse mRNA of a diseased gene which is associated with both lung and prostate cancer, and to then produce a DNA molecule structured after an anticancer drug. The rules of operation which are programmed are all biological in nature.

The prospect of injecting a disease-seeking, identifying, and disease-destroying biological computer into a human body is a joy to antiicipate.

[isaac]

I believe that this use of the word [programmed] is improper because it implies intentionality. A programmer intentionally constructs a program to do certain things. Since there is no god to program things in nature, there is no intention to be found there either. Even if I agreed that you could say that cells are "programmed" to do certain things, there would still be a fundamental difference between things that were programmed by evolution and things that were programmed by humans.

Bowser, are you using the term "intentionality" in the philosophical sense? Or do you mean

adv - planned by a conscious agent in pursuit of a goal

Making a genetic fallacy about DNA.

That's classic. Is your wit intentional?

(Or is it just a series of states with no correllation to anything outside itself? Ok, I'll stop with the puns, I promise.)

The prospect of injecting a disease-seeking, identifying, and disease-destroying biological computer into a human body is a joy to antiicipate.

And how! Stephen, that's fascinating! I've read about projects that do things that could have conceivably yielded stuff like this, theoretically, blah blah, but someone's actually doing it? Wow! Do you know specifically when it will be published? Are there any web references you could provide us with?

Nature should thank you. You just sold me a copy of their magazine.

[stephen_speicher]

Do you know specifically when it will be published?  Are there any web references you could provide us with?

You can write to [email protected] and ask for the expected publication date. If you want a pdf of the paper I can email it to you, but the paper is a bit technical.

As to web references: Coincidentally, tonight's HBL had a post about this experiment. The poster provided this URL: http://tinyurl.com/2o928

If Reuters has the story, then I am sure it will be carried by all the usual online services.

[isaac]

Yep, the news has broken, alright.

If you could email me the pdf, or a link to it if it's really big, I'd absolutely love it. Technical is good ones. Thank you.

isaacschlueter is the beginning part, and then an @ of course, and then hotmail.com (Fighting spambots.)

[stephen_speicher]

If you could email me the pdf ...

Actually, I just realized that I do not have the author's permission to forward the paper, so I have to withdraw my offer. If you want you can email the lead scientist [email protected] and ask if he will forward you a copy.

[amagi]

I said:

The programs of cells make them goal-directed, but there's absolutely no reason why humans can't program a machine to be goal-directed as well.

And Betsy replied:

Goal? Whose goal? The human programmer's goal or the machine's? It would have to be former because the latter isn't alive. Therefore it doesn't have a "What for?" of its own.

First of all, I don't think this statement of mine was at all necessary for the point I was making in the post it was contained in. But I stand by it:

Why does Objectivism define life as the standard of goal-directed activity? Because, as Dr. Binswanger says:

The existence of inanimate objects is not conditional upon their actions: (1) inanimate objects are not capable of self-generated action, and (2) they will continue to exist as long as they are not acted upon by external forces

But it's definitely within man's capacity to design a machine that is both (1) capable of self-generated action and (2) dependent upon it's own action to continue existing. Moreover, the cessation of existence such a machine would act to avoid would not merely be destruction in the sense of a stone being "destroyed" when it is crushed. It would be the equivalent of death, as OPAR defines death (p. 191):

life is motion, a definite course of motion; if the motion is defaulted on or fails, what ensues is the antithesis of life: stillness, which is the essence of death.  Death is the irreversible cessation of vital processes.  Leaving aside the disintegration that follows it, death is a state that does not involve or require action.

A machine like the one I've described would act to avoid this "death," and so it's actions would be goal-directed. It need not be alive to have the same conditions imposed upon its actions by the risk of its own destruction. Additionally, if you accept, as I do, that a machine could be made to be conscious, then this issue becomes completely unambiguous--avoiding the cessation of that consciousness that would become the standard for its goal-directed actions.

[Edit: To make it even more clear, the only sort of consciousness required would be the consiousness of an insect. I'm not talking about AI here. I do not think it will be so very far in the future before we'll be able to build some sort of advanced robot equipped with sensory capabilities that could have the conscious awareness of a very simple animal.]

[amagi]

stephen speicher writes:

Biological mechanisms obey the laws of physics and chemistry, and any "programming" is a term used primarily by laymen and the novice. When you study the molecular biology of the cell you study the inner workings of the cell and its interactions, and any sense of "programming" would only arise as an extremely loose metaphor. DNA is the repository of genetics, not a program.

You are quite wrong. Arguing that DNA does not program the actions of the cell is just plain silly.

Everything a cell does is ultimately programmed by its genetic makeup.

Your argument amounts to this: you assert that you've redifined the word programming so that it can only refer to programs made with conscious intent, and then conclude that the genetic material of an organism doesn't constitute a program because it was not designed by intent.

The dictionary says the word "program" means

"operating instructions for a machine: a set of coded operating instructions that is used to run a machine automatically"

The genetic material represents a set of coded operating instructions that ultimately directs all of a cell's functions automatically.

DNA is the repository of genetics, not a program.

That's a meaningless statement as long as you neglect to define what genetics does for a cell. Genetics determines the form and function of the cell. In other words, it determines how the cellular machine runs, automatically. In other words, it's a program.

[Bowzer]

I know that this won't convince amagi but I'll put this out there for the rest of you.

"Life" comes down to this: lock a cat and a robot in separate rooms. Come back one year later and what do you find? A decaying pile of yuck that was once the cat. It has died since it was unable to obtain the materials needed to sustain its life. The robot, on the other hand, looks just as good as ever. Conclusion: a cat and a robot are different in a fundamental way--one is alive the other is not.

You won't find that scenario convincing unless you accept the Objectivist theory of concepts. A concept's meaning is in the units that it subsumes and amagi's robot is not a unit of the concept "life."

[stephen_speicher]

A concept's meaning is in the units that it subsumes and amagi's robot is not a unit of the concept "life."

Yes. Some would do well to understand the full meaning of Ayn Rand's words in Atlas Shrugged (p.31):

"There is only one fundamental alternative in the universe: existence or non-existence—and it pertains to a single class of entities: to living organisms."

And, in this regard, Ayn Rand's discussion in The Objectivist Ethics (p. 16) of an "indestructible robot" further underscores the point.

[amagi]

The point is it wouldn't be indestructible. It's possible to build a robot that must act continuously to avoid permanent cessation of it's functioning. Such a robot would cease to be a robot if it didn't act--just as a living creature would cease to be alive if it didn't act.

The body of the robot might be unharmed after it's "death"--it might not decay like the body of an organism (although it could)--but it would be an inert piece of metal or plastic or whatever. It's capacity to perform all of its functions would have been destroyed. Any data stored within it would also be destroyed. This destruction of function would impose the same alternatives on its actions--act or cease to exist--and it could therefore be called goal-directed.

Remember also that a cell is a machine--it's illogical to assume that it's the only kind of machine that could possibly be goal-directed.

Look, I realize this is tantamount to blasphemy to some here, so I'll make no further effort to convince anyone. Suffice it to say that I don't think this conflicts with the fundamental idea of life as the standard of ethics.

The very first primitive cell happened by accident, essentially, and it was goal-directed. The idea that it's now impossible for man to build something that is also goal-directed is an insult to human ingenuity, and it springs from a narrow, rationalistic conception of Rand's insights.

[Bowzer]

The idea that it's now impossible for man to build something that is also goal-directed is an insult to human ingenuity, and it springs from a narrow, rationalistic conception of Rand's insights.

Very interesting that in all of your posts here (in this thread at least), this is the first time that you have attempted to express knowledge of Objectivism and at that only in negative form, i.e., asserting our lack of knowledge.

I also am getting tired of fighting over an issue that is made abundantly clear in so many places in the Objectivist corpus. Seeing that Miss Rand used the example of a robot to disprove the position that you insist upon, amagi, maybe you can show us what we are misunderstanding here. I, for one, would see this as a breath of fresh air as it would turn the discussion to the philosophy of Objectivism which is, after all, the purpose of this BBS.

[amagi]

Bowzer, if you look again you'll see that I did not question your knowledge. I said that the idea that it is impossible for man to create anything capable of goal-directed action is blatant rationalism. In other words, if you hold that idea, I question, not your knowledge, but your understanding of the reasons why AR identified living things as the only entities capable of goal-directed action.

There is no reason that humans cannot build a machine (if not now, then at some time in the future) possessed of the same specific attributes of living things that indeed make them presently the only goal-possessing entities. Those attributes are-- (1) the capacity for self-generated action, and (2) the need to continually act to avoid destruction.

I have already explained why the destruction involved would be more than the meaningless "destruction" of an inert collection of matter; it would be the equivalent of death--the "cessation of vital processes"--in terms of the demands imposed on the actions of the machine.

Seeing that Miss Rand used the example of a robot to disprove the position that you insist upon, amagi, maybe you can show us what we are misunderstanding here.

Alright. Perhaps you and Mr. Speicher have misunderstood the fact that the robot in her example was "immortal, indestructible," and incabable of being "damaged, injured, or destroyed," or "affected by anything," or "changed in any respect."

[sv watson]

consciousness is an attribute or characteristic (of some) life, but it is not life.

a is a ......

sv watson

[Bowzer]

Amagi, I have never argued that man could not synthesize life in some form. I have been arguing that only living organisms are and can be conscious.

If this thread is now narrowing down to a linguistic quibble then I think that's a turn for the better. For I am saying that it is improper in this context to group living things--whether synthesized or naturally generated--with machines whereas you now seem to say that there is nothing wrong with this as long as they share certain essential characteristics (i.e., the essential characteristics of life). I'm not sure that you have been holding this context in mind this whole time but it's much better the way you are stating it now.

I'm sure that you are aware of academics' useage of the word "machine" and that they use it to argue that computers and thermostats are conscious. This is typically what people mean when they talk about "machines" and the nature of consciousness. That may be more my fault since I am so accustomed to this usage.

If we someday create something that has the essential characteristics of life, then I agree that consciousness is possible to it. But it would be alive and this is what I have been arguing all along.

I would like to explore the "robot" example but it will have to wait until later.

[stephen_speicher]

Amagi, I have never argued that man could not synthesize life in some form. I have been arguing that only living organisms are and can be conscious.

If this thread is now narrowing down to a linguistic quibble then I think that's a turn for the better. For I am saying that it is improper in this context to group living things--whether synthesized or naturally generated--with machines ...

Yes, exactly. Life is life, and that fact is fundamental.

And, I might add, contrary to the words that amagi continues to put in others mouths -- which is the reason I will no longer discuss with him directly -- my argument has not been proving the impossibility of this thesis, but rather putting the onus of proof on those who advocate it. All the scientific evidence supports consciousness as being distinctive to life, and there is no scientific evidence to the contrary. Anyone who asserts differently must provide evidence, not wishful thinking or fantasies.

[amagi]

Mr. Speicher, you continue to claim that I was misrepresenting your statements in the Artificial Intelligence thread, but not once have you bothered to explain why you think my understanding of your view is incorrect. It would have been wonderful, for the sake of clarity and to avoid a good deal of frustration, if you had addressed any criticisms of your position, but you did not.

Briefly, to review, you said that mental acts can occur without any physical expression, and that such acts are then followed by a delayed response in the brain. Then, when I said that you believe that mental acts can occur without physical expression, you first simply declared that I was putting words in your mouth, and then went silent.

Now, I also said that mysticism is the only explanation I can think of to explain how a process of thought could occur in the mind without any physical action whatsoever in the brain to account for it. If this is what you're characterizing as the view I'm falsely attributing to you, then you're greatly mistaken, for this reason:

"Putting words in your mouth" is fundamentally different than identifying the premises I believe must support that which you've explicitly stated. I never accused you of being deliberately a mystic--I assume you would recoil in horror if you ever suspected an idea of yours to be mystical. I stated only that it seems you've implicitly accepted a false premise, which is not the same as attributing arguments to you that you do not hold, and which is not inherently an act which you can fault anyone for engaging in.

The average environmentalist would no doubt protest that you were "putting words in his mouth" if you said his premises were fundamentally anti-man, because he does not think far enough to identify them as such. But would you actually be doing so?

I do not wish to start any discussion of this in this thread, but I cannot leave the accusation that I was falsely attributing arguments to you unchallenged. If you should ever decide to explain your mysterious reasons for why I'm wrong I'll be happy to consider them in the AI thread.

[amagi]

Mr. Speicher writes:

my argument has not been proving the impossibility of this thesis, but rather putting the onus of proof on those who advocate it. All the scientific evidence supports consciousness as being distinctive to life, and there is no scientific evidence to the contrary.

The issue is whether a man-made machine can be goal-directed, not whether it can be conscious. I've only mentioned consciousness in this thread as a side issue. Therefore, I did not attribute to you the argument you say I did.

The thesis I did refer to was this: "it is impossible for man to create anything capable of goal-directed action."

I suppose you could say that I attributed this to you, but only peripherally, in an indirect way. Refer to my previous post. But I'll grant that I attributed it to you by association.

As to whether this is a case of putting words in your mouth, Bowser has made me realize that neither he nor Mrs. Speicher (so, by extension, Mr. Speicher ) have said, "man cannot create anything goal-directed"--because they both allow that man could synthesize a new kind of lifeform.

In other words, this is to some extent merely a "linguistic quibble," because:

In my definition of "a machine created by man," I was including all machines that humans could make, even if they would be "living machines." Note that I have continually referred to the cell as a machine. What I have been saying all along is that man could build a machine that shared the essential characteristics of life that make life capable of goal-directed action. I did not specify whether this meant the machine would therefore be alive.

However, this is not entirely a linguistic or semantic issue, because sharing some essential characteristics of life is not necessarily the same as being alive. More on this later.

The point is that because of this misunderstanding, I erred in characterizing anyone's argument as denying man's ability to create anything goal-directed. But substitute the argument: "man cannot create a non-living machine capable of goal-directed action," and I still disagree.

[stephen_speicher]

Briefly, to review, you said that mental acts can occur without any physical expression ...

For the record, it should be known that "amagi" does not speak for me. Whatever views I have, stated or otherwise, should never be confused with whatever "amagi" chooses to attribute to me. Furthermore, my lack of response to any particular post or issue which references me, should never be taken as agreement on my part.

[amagi]

Are you serious?

Here are some quotes from you:

volitional conscious action precedes brain activity.

results showed the mental action to precede brain activity

That means the mental act occured before any physical expression to account for it existed--and only afterwards did the brain act at all.

This is not a "conclusion" based on your above quotes. It's what your quotes say. It contains all of the same information. And it's all I've ever attributed to you. Even if you thought that it was not what you said, the proper way to handle the situation would have been to explain my supposed misunderstanding of your position. Instead, you proceeded to arbitrarily assert, in several different threads, on several different days, that I was engaged in a deliberately dishonest campaign of "putting words in other people's mouths." Such actions are insulting, baseless, and undignified.

For the love of god Mr. Speicher, why do you persist in denying that you said this?

Until you explain how I'm in any way attributing views to you which you do not hold, even though you've explicitly stated those views, STOP accusing me of doing so. Please do not refer to me at all if you cannot refrain from engaging in these attacks on me.

This is completely ridiculous, and I'm through with you. Frankly, it's profoundly disappointing to find an Objectivist behaving in this manner.

[JPowersDC]

Emergent properties are very interesting.

Computers can model things in the real world, even very complex things, and properties can emerge from those models. I saw a documentary the other day where they modeled a tornado in a computer. The computer was not at all _programmed_ to simulate a tornado; it was programmed only to model a huge number of complex, interacting atmospheric phenomena, but a "tornado" emerged from it.

If/when we do discover the intricate inner workings of the brain that give rise to conciousness, I'm sure they could be modeled to some degree. But would such a thing be concious? I tend to think not, for the same reason that the above-mentioned tornado, no matter how close the model gets to approximating the real thing, will never destroy a real trailer.