This is how I always thought about it until I actually thought about it. :)
You seem to be using "structurally different" in an odd way. If I'm 6'1" male and my wife is a 5'7" female you are telling me our bodies are structurally the same because we are both humans? We of course share many structural similarities, but in not structurally the same unless all the structures are the same.
Not only is the structure from the outside very different, but our brains are structurally very different as well - which is why she doesn't act like me, and I don't act like her - we have different programing which is 100% represented in the structurally differences of our bodies. That's how I use the term "structurally different".
Meaning a structure can exist which has no function? Or just that a structure can exist which doesn't have a specific function (a car dose not function as a hammer).
Not at all. It's _VERY_ different when it's powered. It has piles of extra electrons that aren't there for long after the power is removed. It's the physical structure that determines the memories function.
If you load Microsoft Word into the memory of a computer, this is not just a "logical" process (how we tend to think of that). The machine has physical been reconfigured into a word processor by the physical actions of it's parts.
A computer is just as physical as a clock with gears. It works for the same reason a clock with gears works. The atoms and electrons push each other around in physical ways constrained by the physical structure of the machines.
Computers are mechanical devices just like a mechanical clock is mechanical.
The brain is likewise a mechanical device just like a clock. It's just got much smaller parts that operate in more interesting ways that clock gears. Humans are robots.
Yes, process is nothing more than the documentation of physical motion.
Yeah, but we are built to respond to a lack of change as well. (hit the button if the light hasn't changed in 10 seconds). So I think a "change" centric view might be a bit limiting. In order for a machine to perform a task like that, it must have a physical clock which is constantly changing, so internally, it's all about time based change, but externally, we are not limited to responding to only sensory changes.
I think you are getting yourself in trouble by not sticking to one abstraction level when you talk about this. You say "at a conscious level we know it wasn't real". But what does it mean at the hardware level for the machine to consciously know something? What physically happens in the hardware to indicate we know something?
You talk about the brain injecting a hypothetical event into the sensory circuits. But what hardware in the brain is generating the event signal to be injected? Are you proposing special brain modules which generate the hypothetical event signals? How does that module know what type of signal to inject and when to inject it?
And if we understand the world in terms of our senses, how is it we are not fooled by the injection of a replacement signal? Why doesn't the brain react to the signal exactly as it would if it had come from the eyes?
What general structure are you suggesting, at the hardware level, to explain these things?
I think the only difference in our views is that I have a pure mechanical answer to what's going on, and you have a half mechanical and half subjective answer to how it works (you keep saying we "know" it's not real but don't explain in mechanical terms how the brain is structured to allow that to happen).
Hey, that's interesting. It's the first time I've seen an idea that actually comes up with what I consider a valid justification for dreaming while we sleep. Many people have suggested that sleep is needed to make the brain work correctly - to reset it self or something like that - and that in our AI we might have to duplicate the function - I've never believed that). I believe you are correct and that the dreams we have when we sleep actually do condition the brain to avoid predicted dangers - or seek out predicted rewards (we had a dream that "told us" to look under the rock for the gold).
I think however the main advantage to all this, as you have said, is that we can do this "dreaming" while we are awake (we can run scenarios and test options). The fact that they happen, and can be of use as we are falling asleep, is just a side effect of the same useful brain feature. (we sleep simply because we live on a planet that's dark half the time and our body is optimized for collecting food in daylight so we need to conserve our energy at night so we have it available in the day when it will do us the most good for food collecting).
But how is the hardware structured to separate the "base state" from the injected state?
The "human" is the current structure of our body.
The same problem happens with many effects in nature like waves, or clouds.
The physical material that makes up a wave changes just like the material that makes a human changes.
Well, the problem, is that the human you are talking about, is only in your mind. The human I talk about is the physical human. I don't see any advantage to talking about my body existing in your mind. I talk about my body as my body. Do you see the difference?
I can make a pile of 7 rocks. Every day, I can replace one of the rocks with a new rock. Every week, the pile is totally different from the previous week - 7 new rocks every week. But, when I talk to other humans, I can talk about "my pile of rocks", and we can all know exactly what we are talking about - it's the stupid pile of rocks that Curt has been keeping over in the corner for the past 10 years.
I use the same words to describe the rock pile today, that I did 10 years ago. It's still, "my rock pile". But yet, the rock pile itself is not the same over that entire period. It's physically very different. It's made of very different rocks, in a very different configuration.
So what is hard to understand about this rock pile? There is the physical rock pile, and there are the words that we use to describe it, and there are physical structures in my brain that gives me the ability to talk about my rock pile, and there's the physical structures in my wife's brain that lets her talk about my stupid rock pile.
Atoms are not stable either. They constantly reconfigure themselves from nanosecond to nanosecond second. But they maintain enough consistence to allow us talk about them as being a "thing".
The point I'm making here is that the best way to understand it all, is to use the physical layer as the lowest level abstraction of what "exists". The concept of "process" is a creation of our brain, which we use to help understand the physical world - just like we use language in general as a tool to understand the physical world. But the physical world exists on it's own, independent of the fact that we are trying to talk about, and define it as a process.
I think so, but you might be using the word story in a different way than I do. It can be very hard to find a common ground to talk about concepts like "process" and "function" and "information", and "knowledge" when we try to talk about these ideas independent of the observer. All these concepts are easy to understand when there's a human in the picture acting as the observer. But to talk about brain function, we are forced to remove the human observer from the picture and suddenly, the meaning of all these words become problematical. There were never intended to be used that way.
How can something exist in this universe and not be made of atoms or other sub atomic particles?
The atoms are all that exists. If there is something else besides the atoms, tell me what it is?
This talk you are using about something being the same, but not the same, at the same time, is just confusion in our language left over from the mind body confusion - the body soul confusion. This is where all these ideas of process came from and where they got messed up.
If you are going to reject the idea of a soul being separate from the body, you also need to fix all the other words in our language that are based on the assumption of the human soul being separate from the body.
The words software and hardware for example are just an extension of the soul body belief. Software is never "soft". It's always hardware. But since our entire language is based on the idea that it's the soul that is "seeing" and "understanding" the world instead of the body doing the "seeing", we developed this large and complex language based on the same split.
I believe that what you are trying to argue here (without even realizing why you are doing it), is that the split is real and valid. And that the mistake I'm making, is to deny that there is a split. I say there is only structure, but you say there is both structure, and process. And if I deny process as anything other than the motion of physical material, I'm leaving "something" important out. Yes, I'm leaving out the concept of a soul.
It's the same confusion of mind body. We don't have a mind and a body. We just have a body. The reason we were taught to use the word mind separate from the word body is because the language was designed by people that believed (knew for a fact) that the soul was separate from the body.
But if you believe they were wrong, and that body is just a body which moves according to the laws of physics, you need to find a new understanding of what all these other words really mean.
The body is just the body. It's not a body in a reality. If we choose to talk about a small part of the universe called one human body, that's something that is happening in the brain of the person doing the talking. Concepts are physical. They exist as physical brain structures in the person that understands, and uses those concepts. We have this abstract concept of what a human body is that we make use of when we talk about human bodies. But this abstract concept of "human body" is something that exists in the person doing the talking, not in the human body. Of course this is hard to separate when it's a human body talking about a human body. It would be easier to understand if we had intelligent robots with the same mental powers so then we could talk about the concept of a human body existing in the brain of the robot when it was talking, and thinking about, human bodies.
I thought it was the laws of physics that held the body together? What is this magic "glue" called process that holds the body together?
So, when I kill someone, it's God we hold responsible and not me? Or maybe Darwin?
You seem to have a very human centric view of reality - that only "humans" can be held responsible. (this is the typical view most people have and the one I constantly have to try to beat out of people for them to correctly understand a true materialistic view of reality).
It again, goes back, to the belief that we are "special" because we have a soul. That only a soul can be held responsible for an action, because it's the soul that is the cause of the action. Rocks and machines can't be held responsible because they don't have a soul. We must always track the cause back to a soul in order to find out who is "responsible" for the action. (you can't be a "who" unless you are a soul).
This is the mind-first view of reality that existed in our culture for thousands of years. But it's the view that's being demolished, slowly, piece by piece, by Darwin's Dangerous Idea (I put it in caps because it's the title of Dennett's book on this topic). Dennett talks about Darwin's ideas of evolution being a universal acid which is slowly eating through all our established beliefs and views - that it has extended far beyond the reach of biological evolution. And he's quite right.
Man's entire view of the world was based on an error made long ago - that the soul is separate from the body, and that only a soul has the power to act with purpose. Therefor, only a soul can be held responsible. And above the soul of man, was the super mind of God, the soul and the mind that is the cause of everything. This is a nice idea, everything seemed to fit in a nice order, but one which we now understand to be nothing but hogwash a few no more useful than the flat earth perspective.
This nice little mind first view of reality all started to fall apart when Darwin figured out that you don't need a soul, to explain the creation, and the design, of life on earth. You need nothing more than than simple physics and the law of motion. Of course, more than 50% of the people here in the US still don't understand this, and are still clinging to their mind-first view of reality, but that's another problem.
Designers are already creating intelligent robots. :)
Right, of course. The truth is that there is a causality chain that creates all physical actions in the universe. It's a causality chain that acts according to the laws of physics. In order for us to control the environment we live in, we learn what the causality chain is, and we manipulate parts of the chain to get the desired effects we want. We move the light switch in order to turn off the light because we understand the switch is part of the causality chain which is creating the light.
If someone is trying to kill us with a gun, we manipulate different parts of the physical causality chain to try and prevent this action - we might take the gun away, or we might try to block the path of the bullet, or we might try to manipulate the brain of the person to prevent it from causing the finger to pull the trigger by talking to the person. In all causes, we are attempting to use our knowledge about the causality chain to make a better future for ourselves.
The concept of "holding responsible" is just one way we talk about how we manipulate a causality chain that has a human brain as a major component in the chain.
Ah, not really. We don't blame the weather on the water that falls out of the sky because we could not predict the behavior of the water molecules do we?
We assign "blame" to souls. That's how the word is used in English. It comes from a day when people believed that it took a soul to animate matter. The idea that matter could animate itself, was mostly unthinkable in those days - just as the idea that matter could self organize to form life was unthinkable.
We are slowly extending the use of the word "blame" beyond where it started, but the progress is slow. Since souls are nothing but a special type of machine, where should we draw the line on the correct use of the word "blame"? What type of robot would it be valid to "blame" for it's actions, instead of blaming the designer? :)
Yes, there's no problem in being descriptive as long as you always understand where the line is. When we use these common English words to talk about AI, the line gets very confusing simply because most common English words were defined back in day when everyone believed in the soul. Even though some of us have totally rejected that old belief, we still use the old language with all it's old issues simply because our brain as been conditioned to think in those terms.
Yeah, that's the problem that happens. The language we use was built to talk about something where there was a clean split - the body and soul - the observer and the observed. Once you realize the two are one and the same, things turn back on themselves and vanish up their own tailpipe as you say. It gets confusing.
Right, my pulse sorting nodes perform all their actions as a function of the time space between pulses. Behavior of these nodes are not a function of a static datum, but on a measure of time. They act only when they receive a pulse, but how they react is not a function of the value of the pulse (a pulse has no value), the action is based on how much time has passed since the last pulse event happened. Everything in that type of network is time based. It's a temporal network.
Yes, that power you speak of is important in humans, I agree with that. But the question we must find an answer to is what do we need to build in order to duplicate that power? How do we build a robot that acts like a human?
The lowest level hardware is what we have to build, and understand first. What does that hardware need to be? We need to reduce all these problems of making a robot act like humans to the simplest answer we can find. The answer I've found, that is the only one I've every found that all other problems can be reduced to, is a reinforcement learning machine.
Other people working on AI for example have tried to reduce the problem of AI to a knowledge storage problem. Others have tried to reduce it to a logic problem. What are you suggesting we reduce it to? A machine which makes hypotheses? How does that work exactly?
Some common measure of reward defined in internally in the machine.
That is just the nature of reinforcement learning. In the end, the machine must pick one, and only one, behavior at each moment of it's existence. It can't move the arm up, move it down, and hold it sill, all at the same time. The laws of physics dictates that the arm must be in one place at a time.
When there are conflicting values, the machine must convert those values to some common currency and compare them. It must produce an answer as to which value the machine will select as the "better" value. This is the nature of reinforcement learning. The conversion to a constant currency can't be avoided. It happens by default.
In the brain, the neurons end up being rewired as a result of rewards and punishments. The conflicting rewards and predictions of rewards acting on the brain still ends up finding a single answer as to how to rewire itself. The probability of some past behavior being repeated in the future has to either go up, go down, or stay the same. However that probability changes is the decision the system has made when it weighed the conflicting values against each other.
Yes, I tend to do that. But I think it's the correct usage in this context.
And how do you know that don't have emotions or that they are reinforcement learning machines? Is that based on simple empirical evidence or on social convention of the use of the word "emotions"?
Well, I get into these long debates in comp.ai.philosophy all the time. It's very useful to get a strong grasp on all these concepts if we are going to try and build a machine that can act just like a human. I learn a little something new every time I argue these points.
If you want to continue to hash out these ideas we could move the thread there since most of this is not very connected with practical information on how to program robots even though the entire reason I work on passing out these ideas is so I can one day build a very interesting robot.
Or, we can drift more towards talking about how one might actually program any of this in a robot in which case it might not be too off topic for this group. I'm game to talk about it at any level.