I agree with Eschew. The processes matter; the physical substrate doesn't. This view does lead to a kind of dualism but at least it's a dualism that is scientifically compliant.

This all reminds me of Dr. McCoy's objections to the transporter on Star Trek. He knew the machine was a duplicator and a destroyer. Despite this, he used it again and again every week knowing it killed him and made a copy every time.

Death doesn't matter as much when you can be restored or copied from backups.

Kirk or Picard gets pulverized by a meteor. No problem. We've got a snapshot of his atomic profile stored in the ship's memory from a few hours ago. Just throw some sea water, several lumps of coal, some chalk, a few nails and some other trace elements on the transporter platform and impose the profile on it. Bang! Picard walks again minus a few memories.

Trek always seemed to avoid this implication. They never dealt adequately with it.

If I learned I was a copy, it would certainly intrigue me but, it wouldn't make me suicidal or feel any less human.

Abolitionist wrote, "It may be possible to replicate a neuron (seems likely that they are just complicated on/off switches) but for the copy to be identical it would have to have the same exact properties as the original in all aspects."

There are a few things wrong with the assumptions behind this sentence:

• The neuron is more than just a switch. Synapses don't work in a strictly digital fashion. It's a lot more complicated than that. A synthetic neuron would have to capture this analog processing and communication to be useful to us.
  
• A synthetic neuron doesn't have to be exactly the same if we are only interested in certain aspects of it--information processing and communication. We don't need to capture the cell's reproductive and defensive organelles for example because those aren't necessary for information processing or communication. This is why I constantly repeat in these discussions that synthetic neurons need only capture the essential details.
  

Having said that, I see no scientific reason why synthetic neurons shouldn't be possible. Once you have duplicated the behavior of units that compose the whole in sufficient detail, it then becomes a matter of arranging them properly so that higher level behaviors also emerge. That's one of the implications of emergent holism.

The Blue Brain Project is one of the humble steps along this path: a path that will eventually fullfill the strong formulation of AI research.

EschewObfuscation:

Abolitionist:

the copy believes it is me, but is programmed to live in the same Darwinian condition - though I think it might be difficult to keep the truth a secret from the copy, maybe not...

I don't understand.

If I learned that I was a copy, I'd probably become psychotic.

Why? As I explained, being a copy doesn't make you any less valuable...

It may be possible to replicate a neuron (seems likely that they are just complicated on/off switches) but for the copy to be identical it would have to have the same exact properties as the original in all aspects. Are there any theories about how the rest of the brain might be replicated?

Don't know. It's just physics; it's possible to emulate, however hard.

Google search turns up many differing perspectives on the dancing qualia thought experiment.

I understand that these are theories to be put to the test and anything is possible.

The way I see it, Occam's razor suggests that carbon-basedness probably isn't important to consciousness at all - only information processing matters.

How would the copy deal with learning that he is not in fact the original when his memories say he is,  and that he was created (using a flawed design) in order to make someone feel better about dying? To me it's a valid concern.

I agree that anything is conceivably possible. Yes, carbon-basedness may not be critical for consciousness, but anything not functionally identical would not be an identical copy. How can we know that the artificial substrate has the exact same functionality as the original substrate? As all aspects of the mind's functionality are related to our conscious experience - we'd have to choose what to preserve based on our current definition of the self.

Mr. Farlops:

Abolitionist wrote, "It may be possible to replicate a neuron (seems likely that they are just complicated on/off switches) but for the copy to be identical it would have to have the same exact properties as the original in all aspects."

There are a few things wrong with the assumptions behind this sentence:

• The neuron is more than just a switch. Synapses don't work in a strictly digital fashion. It's a lot more complicated than that. A synthetic neuron would have to capture this analog processing and communication to be useful to us.
  
• A synthetic neuron doesn't have to be exactly the same if we are only interested in certain aspects of it--information processing and communication. We don't need to capture the cell's reproductive and defensive organelles for example because those aren't necessary for information processing or communication. This is why I constantly repeat in these discussions that synthetic neurons need only capture the essential details.

"Seems likely that they are just complicated on/off switches" - this is distinct from saying that they are digital - in fact, I think we're on the same page here.

What are the essential details? (It will be an interesting debate)

 

Abolitionist:

"Seems likely that they are just complicated on/off switches" - this is distinct from saying that they are digital - in fact, I think we're on the same page here.

Saying something is an on/off switch is exactly the same thing as saying it's digital. It doesn't matter if it's a two state, four state, twenty state or million state switch. The mere fact it has discrete, quantizable states means it's digital but, let it stand. You get my point.

Please note that just because I'm saying a neuron isn't digital doesn't mean I'm saying a neuron isn't a formalizable machine. It is. It's a machine whose behavior we will one day duplicate.

What are the essential details?

Understanding and duplicating the information processing and communication rules that neurons follow. Those are the details that matter.

There are many different types of neurotransmitters sorted into three general categories. Neurotransmitters, as you probably know, are the chemicals that relay, modulate, diminish or amplify signals between neurons or neurons and other cells.

Already at this point we've departed from the switch concept because the signals can change in strength, speed and pattern continously depending on how much of a particular type of neurotransmitter is suffusing the synaptic gap between neurons. These changes are not discrete nor digital.

Neurons are also bathed in cerbrospinal fluid which soaks them in hormones and other chemicals that change their processing and communication behavior outside the vector of neurotransmitters.

We need to learn more about how neurons react to these two communications paths. How do they they process the signals they recieve? What are the protocols, the rules they follow? Can we model this in software and hardware?

Things that aren't essential:

• Internalized cell reproduction machinery (Since we are manufacturing these synthetic neurons externally they don't need their own reproductive apparatus.)
  
• Cell defense machinary. (Disease fighting may not be necessary.)
  
• Cell differentiation machinary. (We don't need synthetic neurons to have the potential to turn into bone, blood or fat cells.)
• Communication vectors that have nothing to do with brain activity
• Others I can't think of

Computational neurology gets better and better at modeling these behaviors. Currently it takes our most powerful supercomputers to model cortical columns (collections of 80 to 100 neurons) to a high degree of accuracy.

But that level of accuracy may not even be necessary if all we're concerned with is communication and processing. We may be able to optimize the simulations, while still capturing the facets we want, so they are much less computationally intensive. We may discover higher level behaviors that can be modeled in other ways rather than requiring the duplication of individual neurons at a lower level in the cortical column system. We may find shortcuts.

Also machines get steadily more powerful. Today Blue Brain requires IBM's big iron. Ten years from now, it may fit within your mobile phone.

One way or another we'll eventually figure this out. It took blind evolution billions of years to develop neurons in the phylum of Cnidaria (Anemones, sea jellies, etc.). We've been at this, what? 500 years? Seems to me we're getting pretty good at it.

Remember you can lose a lot of your brain and still remain pretty much the same person... too...

So the copy wouldn't need to be perfect to be fit to live... unless we insist on everything being 100% surely perfect... which to me, is, like, weird...

Mr. Farlops:

Abolitionist:

"Seems likely that they are just complicated on/off switches" - this is distinct from saying that they are digital - in fact, I think we're on the same page here.

Saying something is an on/off switch is exactly the same thing as saying it's digital. It doesn't matter if it's a two state, four state, twenty state or million state switch. The mere fact it has discrete, quantizable states means it's digital but, let it stand. You get my point.

Please note that just because I'm saying a neuron isn't digital doesn't mean I'm saying a neuron isn't a formalizable machine. It is. It's a machine whose behavior we will one day duplicate.

What are the essential details?

Understanding and duplicating the information processing and communication rules that neurons follow. Those are the details that matter.

There are many different types of neurotransmitters sorted into three general categories. Neurotransmitters, as you probably know, are the chemicals that relay, modulate, diminish or amplify signals between neurons or neurons and other cells.

Already at this point we've departed from the switch concept because the signals can change in strength, speed and pattern continously depending on how much of a particular type of neurotransmitter is suffusing the synaptic gap between neurons. These changes are not discrete nor digital.

Neurons are also bathed in cerbrospinal fluid which soaks them in hormones and other chemicals that change their processing and communication behavior outside the vector of neurotransmitters.

We need to learn more about how neurons react to these two communications paths. How do they they process the signals they recieve? What are the protocols, the rules they follow? Can we model this in software and hardware?

Things that aren't essential:

• Internalized cell reproduction machinery (Since we are manufacturing these synthetic neurons externally they don't need their own reproductive apparatus.)
  
• Cell defense machinary. (Disease fighting may not be necessary.)
  
• Cell differentiation machinary. (We don't need synthetic neurons to have the potential to turn into bone, blood or fat cells.)
• Communication vectors that have nothing to do with brain activity
• Others I can't think of

Computational neurology gets better and better at modeling these behaviors. Currently it takes our most powerful supercomputers to model cortical columns (collections of 80 to 100 neurons) to a high degree of accuracy.

But that level of accuracy may not even be necessary if all we're concerned with is communication and processing. We may be able to optimize the simulations, while still capturing the facets we want, so they are much less computationally intensive. We may discover higher level behaviors that can be modeled in other ways rather than requiring the duplication of individual neurons at a lower level in the cortical column system. We may find shortcuts.

Also machines get steadily more powerful. Today Blue Brain requires IBM's big iron. Ten years from now, it may fit within your mobile phone.

One way or another we'll eventually figure this out. It took blind evolution billions of years to develop neurons in the phylum of Cnidaria (Anemones, sea jellies, etc.). We've been at this, what? 500 years? Seems to me we're getting pretty good at it.

Complex on/off switches - you're right, it's important to spell these distinctions out clearly and we have made great progress at understanding the brain.

I'm thinking that the capacity to experience suffering may define the self - what do you think?

 

 

Abolitionist:

I'm thinking that the capacity to experience suffering may define the self - what do you think?

I think you and i are using different definitions of selfhood.

Let it stand though. Sure, one charactistic of the human self is the capacity to suffer.

Mr. Farlops:

Abolitionist:

I'm thinking that the capacity to experience suffering may define the self - what do you think?

I think you and i are using different definitions of selfhood.

Let it stand though. Sure, one charactistic of the human self is the capacity to suffer.

"let it stand"? - please explain. Do you think this discussion is not edifying? I'm hoping that my definition will stir debate. Incidently, my goal is not to be right but to promote discussion - that's why I'm throwing out a definition that I know will be dissected.

Come on, Mr. Farlops - you know you love this stuff ;-)

 

 

Frankly I don't have much to say about the subject of suffering. This why I didn't weigh in when you've brought it up in other threads. It's really hard for me to avoid being glib about it.

Someone else want to step in here?

Abolitionist:

How would the copy deal with learning that he is not in fact the original when his memories say he is,  and that he was created (using a flawed design) in order to make someone feel better about dying? To me it's a valid concern.

The copy... is... not... distinct from... the... original. Both are conscious. Both have the same memories. Both have the same body. Same name, same ability to interact with others, same rights (hopefully), etc. Making a distinction between "original" and "copy" is not helpful. The copy is not created to "make someone feel better about dying" - the copy is created so that that person does not die. The copy is for all intents and purposes the same person.

Or is your real objection to the "using a flawed design" part? I.e., do you think that making a copy that suffers is unethical? I agree that we should improve ourselves and our descendants/copies/what-have-you to reduce suffering, but personally I wouldn't mind being a copy with the current human design, however flawed, if the alternative were death. Or even if it weren't. I don't know about you, but I consider my life, however full of suffering, to be on the whole pleasant and very much worth living.

I agree that anything is conceivably possible. Yes, carbon-basedness may not be critical for consciousness, but anything not functionally identical would not be an identical copy. How can we know that the artificial substrate has the exact same functionality as the original substrate?

Experiment. Probably something along the lines of a Turing test.

As all aspects of the mind's functionality are related to our conscious experience - we'd have to choose what to preserve based on our current definition of the self.

I do not understand this sentence.

Abolitionist:

I'm thinking that the capacity to experience suffering may define the self - what do you think?

This sounds interesting, but if the self is defined by suffering, then wouldn't the self be lost if suffering were abolished? I doubt you actually want that, so I can't help but think I'm understanding this incorrectly.

Personally, I consider a "self" to be an entity with memory, consciousness, continuity, and concepts of time and other consciousnesses - well, at least that's what I can dash off in 30 seconds, there's probably actually more to it.

Eschew wrote, "This sounds interesting, but if the self is defined by suffering, then wouldn't the self be lost if suffering were abolished?"

There we go! That's an angle I can say something about. Thanks Eschew.

There are many different kinds of suffering. If we take physical pain to be a kind of suffering then Ab's proposal is flawed. My selfhood doesn't vanish when the dentist numbs my mouth before starting a drilling. My selfhood doesn't vanish if I take an aspirin for a headache.

In terms of emotional pain, my selfhood doesn't vanish if I take prozac.

My selfhood does vanish if the drug makes me pass out however. General anesthesia shuts off consciousness and, thus, selfhood.

EschewObfuscation:

Abolitionist:

How would the copy deal with learning that he is not in fact the original when his memories say he is,  and that he was created (using a flawed design) in order to make someone feel better about dying? To me it's a valid concern.

The copy... is... not... distinct from... the... original. Both are conscious. Both have the same memories. Both have the same body. Same name, same ability to interact with others, same rights (hopefully), etc. Making a distinction between "original" and "copy" is not helpful. The copy is not created to "make someone feel better about dying" - the copy is created so that that person does not die. The copy is for all intents and purposes the same person.

Or is your real objection to the "using a flawed design" part? I.e., do you think that making a copy that suffers is unethical? I agree that we should improve ourselves and our descendants/copies/what-have-you to reduce suffering, but personally I wouldn't mind being a copy with the current human design, however flawed, if the alternative were death. Or even if it weren't. I don't know about you, but I consider my life, however full of suffering, to be on the whole pleasant and very much worth living.

I agree that anything is conceivably possible. Yes, carbon-basedness may not be critical for consciousness, but anything not functionally identical would not be an identical copy. How can we know that the artificial substrate has the exact same functionality as the original substrate?

Experiment. Probably something along the lines of a Turing test.

As all aspects of the mind's functionality are related to our conscious experience - we'd have to choose what to preserve based on our current definition of the self.

I do not understand this sentence.

With the last sentence, I'm saying that if we are interested in preserving ourselves, we must first define what the self is - surely we don't want to preserve all the nastiness of our Darwinian design.

Practically speaking, a copy will always be distinct from the original.

I'm very glad you feel your life is worth living - hopefully this will always be the case for you and all humans.

Can you elaborate on how we should experiment to determine whether or not the copy would be similar to the original?

If you are creating a copy so that you do not die - then you will not be successful. You will still die and experience death even if a copy is made. So, what is it that you are really aspiring to do?

 

Simply: If you can reference yourself, you are conscious you exist. Once an AI can reference itself and make independent decisions (i.e. not programmed beyond its source code), then it will have self too.