No Moon, no life on Earth, suggests theory

NewScientist.com ^ | 18 March, 2004

[RightWhale]

our memories of our past self would get grainier and grainier

The computer model of memory would lead to that. However, the radio model of memory would allow continuation. It doesn't work for modern electronics where there are no user maintainable components inside, but the old tube radios could be resuscitated forever by replacing worn out tubes.

[general_re]

Hey, look at it this way - the Borg would be, as far as I know, the only ones who could potentially make socialism work ;)

[Modernman]

As my fiance says: Psstt.... Your geek is showing.....

[js1138]

Your scenario is perfect by definition, and if I found a flaw in it you would define around it, regardless of any engineering problems that might be posed.

It reminds me of some bosses I've had who defined applications and declared them ready for marketing, except for a small matter of programming.

What exactly does a neuron do? Aside from doing a bit of electrochemical switching, we know it alters its behavior in subtle ways in response to the chemistry of the brain. It is certain that chemical changes are an integral part of computation in the brain. They determine what gets remembered and what gets forgotten. They steer solutions to complex problems like what to do about the tiger that is circling you.

Mimicking all this in silicon will be a formidable problem. You would have to copy all the turtles, all the way down.

[general_re]

Can't run away from what you are, right? ;)

[js1138]

However, the radio model of memory would allow continuation.

I'm a bit curious about the radio model. Where is the seat of consciousness?

If I hit my thumb with a hammer, I am aware of the injury, and the pain has a profoung effect on my behavior. But I can put aside pain if my survival depends on it. I can "objectify" my thumb and see it as something that is part of me but not mu central self.

I'm curious how the radio brain accounts for mind altering drugs. Why can't our real self see our addled brain as a disfunctional component that needs to be put aside for a while until it recovers?

[general_re]

Your scenario is perfect by definition, and if I found a flaw in it you would define around it, regardless of any engineering problems that might be posed.

Obviously, this is purely hypothetical, but I don't see how imagining hypothetical solutions to hypothetical problems is beyond the pale. Every path begins not with the first step, but with someone imagining the journey ;)

Mimicking all this in silicon will be a formidable problem. You would have to copy all the turtles, all the way down.

Well, not exactly. I would suggest that you might be able to obviate all that by treating each neuron as a black box. You don't know what's in there, so there's no point in trying to perfectly replicate each and every aspect of a neuron. Rather, simply emulating how the neuron interacts with the outside world might be enough - I suspect that it's the way the aggregate collection of neurons interacts together that makes you "you", not something about the way each individual piece is made.

By way of an analogy, my quartz watch and my grandmother's hundred-year-old cuckoo clock are radically different from each other on the inside, and yet the core function of both - telling time - is essentially identical as far as everyone else is concerned. If I took out the gears and weights and springs from the inside of the cuckoo clock, and replaced it with silicon and a battery and an electric motor designed to act like a cuckoo clock, would anyone in your house even notice?

[RightWhale]

Where is the seat of consciousness?

It might be a 'whole is greater than the sum of its parts' kind of thing. It could be an aura around the entirety with no particular organ of its own. Replacng some parts wouldn't necessarily change the totality any more than changing violin strings or rehairing the bow would cause notes to be dropped from the music.

[RightWhale]

emulating how the neuron interacts with the outside world

That could be done within limited parameters. But your emulation program might not respond to unforeseen circumstances the way neurons do, and probably neurons are as different from each other as snowflakes so you would need several zillion emulation routines to simulate a single CNS under controlled conditions. Possible.

[js1138]

If I took out the gears and weights and springs from the inside of the cuckoo clock, and replaced it with silicon and a battery and an electric motor designed to act like a cuckoo clock, would anyone in your house even notice?

They would notice unless you emulated the need to wind it, and emulated all the tempermental aspects of a weight driven clock.

More importantly, they would notice if those eccentric aspects of the original clock were somehow important.

Neurons are not like transistors that merely switch. They are electrochemical, and their switching behavior is profoundly dependent on a zillion chemical receptors that simultaneously modify their trigger level. This is all part of the "computation". You can assert that this can be emulated, but I remain unconvinced.

In addition to switching, neurons grow new connections with their neighbors and retreat from others. Connections are strengthened or weakened by whatever happens in learning. In short, we don't know much about what they do, much less how they do it.

[PatrickHenry]

This "gradual-replacement-of-brain-cells-with-chips" scenario is a restated version of an ancient philosophical problem: Theseus' Ship.

[js1138]

It might be a 'whole is greater than the sum of its parts' kind of thing.

I agree in principal with that. One neuron in isolation is like one hand attempting to clap.

But I don't think you have to get all Buddhist about this. Our knowledge of neurons is conceptually at the same point as our knowledge of DNA was in 1950.

[RightWhale]

Sounds like my '86 Caprice. Of course, gasoline is part of the car that is replaced at [very] frequent intervals. But if we run out of gasoline altogether, would we be a nation of 500 million cars any longer? Would these piles of plastic and thin steel panels be cars if there were no gasoline?

[js1138]

Everything about our personalities, including our consciousness and our ability to experience pleasure and pain, functions to preserve and replicate a package of DNA.

Prosthetic brain cells will need to emulate all aspects of that functionality.

[Physicist]

Theseus's Ship is the problem exactly, except that we are facing it now. There is a constant turnover of cells in every body, and a constant turnover of atoms in every cell. None of us is made of the same stuff we were made of a year ago.

[Junior]

I remember visiting HMS Victory back in '90. The docent (a Royal Navy sailor) explained that only about 25 percent of the ship dated from the battle of Trafalgar as the various timbers had been replaced over the years.

[Moonman62]

It's part of a greater question that I have an interest in: what will human beings be like 100, 1000 or a million years from now?

Me too. What will we eventually know about the universe, existence, conciousness, and similar good things?

[Junior]

I wonder if one could create a "virtual neuron" independent of the platform it operated upon. It would take some sophisticated programming to replicate every aspect of the electro-chemical relations between neurons, but it's not inconceivable that this could be done.

[PatrickHenry]

Robert Nozick (who died recently) also dealt extensively with this issue:
Robert Nozick. (Click on "personal identity")

[js1138]

I wonder if one could create a "virtual neuron" independent of the platform it operated upon. It would take some sophisticated programming to replicate every aspect of the electro-chemical relations between neurons, but it's not inconceivable that this could be done.

I'm sure this is being attempted as we speak, in hardware, software, and combinations. Tortoise usually comes is about now to save the day. Someone could ping him. (I'm not sure I want his baleful gaze on my feeble posts.)

But there's a huge hardware problem. Neurons have thousands, maybe tens of thousands, of connections with other neurons. These connections grow and recede. Emulating this is not at all trivial.