Has consciousness been fully explained?

[Beth]

I don't think anyone here is sure about understanding consciousness (but I can only speak for myself really). My understanding of what they have been saying is that it depends first on how you define it -- that's why they bring up SRIP, but no one pretends that is anything like human-type consciousness -- and second that it should be computable, but that's about it. That really doesn't tell us anything about how it works or how the brain does it. Everyone is pretty much in the dark about how it happens, because saying that it should be computable doesn't really tell us much.

I think there is a great deal of contention over whether or not it is computable. Certainly some people are claiming that it not only should be, but it is. Others seem quite certain it is not. A few of us are not certain about it. It is interesting that most debaters here seem so certain that they are correct.

 

[Ichneumonwasp]

You'll have to keep implementing more and more of these "why couldn't you?"s until you have an actual model of a human, not a simulation, if you want to keep with the claim that we can interact with it in any way that we can interact with a real human.

All I am saying is that since we have the Church-Turing hypothesis, which seems to be correct, unless there is some knock out argument that shows that we can't deal with all of this computationally, there should be a theoretical way of doing it. Whether we actually could or not is a whole other issue.

I mean, if there is only one substance (and I take that as my starting point because there are horrible philosophical problems with substance dualism), then on what grounds is computation of everything not possible?

Actions, in their implementation, are actions however we go about bringing them about -- at least as far as I can see.

 

[Ichneumonwasp]

I think there is a great deal of contention over whether or not it is computable. Certainly some people are claiming that it not only should be, but it is. Others seem quite certain it is not. A few of us are not certain about it. It is interesting that most debaters here seem so certain that they are correct.

Basically what I am asking, is "on what grounds would it not be computable?"

I assume as my starting point that there is only one substance. If we simulate that substance, how could we not be able -- in theory -- to simulate everything?

If you believe there is more than one substance, then we may not be able to do so.

 

[rocketdodger]

It's not an adequate part. And this can be shown very quickly in the description of a simple real time program, which will not conform to the Turing model.

There's no need for abstruse speculation about relativity. This is something known to people who actually program control systems. Programs have to be able to handle asynchronous events - that is, events where the sequence is not known in advance.

Suppose we have events A and B in real time.

Suppose we enumerate every causal interaction between A, B, and anything in between, at plank time granularity. Including no-ops. It would look like this:

event A -- planck time 0
event A + 1 -- planck time 1
event A + 2 -- planck time 2
.
.
.
event X -- planck time X

.
.
.
event B - 1 -- planck time <some huge number - 1 >
event B -- planck time <some huge number>


Can this not be done in a machine limited to the "Turing model?"

How is a linear sequence of discrete operations, including no-ops, not well within the capability of a Turing machine?

Here is another linear sequence for you: ? ? ? ? ? ? ? ? ? ? ? ? ?

 

[Ichneumonwasp]

This is the crucial question now isn't it. It is here that the issue of duality comes into focus. If you think that the simulated 'you' would be conscious this implies to me that you are arguing for the existance of minds that have no physical body. Which is the definition of duality as I understand it.

That depends a bit on what you mean by mind and physical body. There is a physical body -- the computer. That the mind in question seems to be associated in the simulation with a body that is also merely an action is just a minor issue. Ultimately the body would be the computer passing electrons around because that is where the action in the real world is happening. The code that creates the simulation is just the instructions that start telling the the electrons to go, which gates to open.

 

[rocketdodger]

Do you feel persecuted by the anti-computationalists?

I am not sure how one could feel persecuted by an entity they felt vastly superior to.

 

[rocketdodger]

The simulation argument depends on too many hypotheticals to be entirely convincing. If we could play poker with Commander Data, would we accept that he was as conscious as we were? Well, we'd have to think about that one - but for the time being, we can't play poker with Commander Data, and there's no particular reason to assume we will ever be able to do so.

This just betrays your predjudice.

Why would you even need to "think" about it?

The dude would be sitting there playing poker with you, able to talk with you about anything, in a way more convincing than many humans to boot.

The point of questions like that -- why I brought it up, in other words -- is to show that you don't really care about any of the objectively valid arguments brought forth in the entire history of this discussion. You are just here to reject any notion that anything besides a normal human might actually be capable of consciousness.

 

[Meed]

All I am saying is that since we have the Church-Turing hypothesis, which seems to be correct, unless there is some knock out argument that shows that we can't deal with all of this computationally, there should be a theoretical way of doing it. Whether we actually could or not is a whole other issue.

Huh?

I mean, if there is only one substance (and I take that as my starting point because there are horrible philosophical problems with substance dualism), then on what grounds is computation of everything not possible?

I didn't say it wasn't. I thought we were talking about whether we could interact with a simulated mind in every way that we could interact with a real mind.

Actions, in their implementation, are actions however we go about bringing them about -- at least as far as I can see.

Yes, but is a full scale robotic replica of a human the same thing as a simulation? That is the path you're headed for when you start with the simple question "Why couldn't it be linked to an audio system so that you could hear a voice?"

 

[Ichneumonwasp]

Huh?

Church-Turing says all this physical stuff should be computable. Consciousness is based in physical stuff -- it is the action of physical stuff -- so it should theoretically be calculable.

I didn't say it wasn't. I thought we were talking about whether we could interact with a simulated mind in every way that we could interact with a real mind.

We have been discussing many different things. I don't think anyone wants to claim that we could interact with a simulated mind in every possible way that we can interact with a real one, just that we should theoretically be able to interact with one.

Yes, but is a full scale robotic replica of a human the same thing as a simulation? That is the path you're headed for when you start with the simple question "Why couldn't it be linked to an audio system so that you could hear a voice?"

That all depends on how one does the computations. The simulations that we have been discussing would not be part of a full scale robot -- that would require a different type of architecture. There are two ways of doing it that I can see with a robot -- one would be a simulation of the neural networks of a brain. The other would be the physicalist solution -- linking up physical 'chips' to do the same things that neurons do physically and passing information along pathways the same way that information is passed in brains. The second option is easier to see as probably conscious, though there are folks who I am sure would argue with this scenario as well. It is the first one that seems to create the biggest controversy. (as far as actually carrying this out in a realistic way I'm all for the second way of doing it)

But what I was referring to was in a big simulation of the 'universe' there are still electrons passing through gates and some of those actions correspond to a particular individual, so why couldn't we just translate that electrical activity in a one-to-one way so that we could hear what was said in the simulated world, and speak to the simulated world?

 

[Beth]

Basically what I am asking, is "on what grounds would it not be computable?"

I assume as my starting point that there is only one substance. If we simulate that substance, how could we not be able -- in theory -- to simulate everything?

If you believe there is more than one substance, then we may not be able to do so.

To my knowledge, there is only one basic substance that everything is composed of. I'm not certain that it is possible to simulate that substance in every respect. We certainly don't know enough at this point to be able to do so.

As far as on what grounds would it not be computable, Roger Penrose has speculated that the mind may not be computable due to making use of QM - i.e. true randomness. Whether or not he is correct in that speculation, I don't know. I am only arguing that we cannot be certain that minds are computable, not that they are or are not.

That depends a bit on what you mean by mind and physical body. There is a physical body -- the computer.

The computer holds the entire simulation - it would be analogous to our entire universe - making this argument analogous to the claim that disembodied consciousness can exist. Which is typically termed dualism in this forum.

That the mind in question seems to be associated in the simulation with a body that is also merely an action is just a minor issue.

Is the fact that our minds are associated with our bodies just a minor issue? I don't see these issues are separate. After all, our bodies (and all material things) can be thought of as merely the actions of subatomic particles. It seems to me that either minds are associated with physical bodies or they are not. It is inconsistent to claim that our consciousness cannot be separated from our physical being but that consciousness in a simulation is 'real' and yet does not similarly require a physical substrate with which to interact with other conscious beings, be they simulated or human.

 

[Ichneumonwasp]

To my knowledge, there is only one basic substance that everything is composed of. I'm not certain that it is possible to simulate that substance in every respect. We certainly don't know enough at this point to be able to do so.

As far as on what grounds would it not be computable, Roger Penrose has speculated that the mind may not be computable due to making use of QM - i.e. true randomness. Whether or not he is correct in that speculation, I don't know. I am only arguing that we cannot be certain that minds are computable, not that they are or are not.

I'm not sure of any of this either, just making the argument that I haven't heard a knock down counter. Penrose's speculations are, from what I can tell, pure bunk because the level of neuron function is many orders of magnitude above QM effects. Until someone shows reliably good evidence that this occurs at a QM level, and I don't think any of Stuart Hameroff's speculations about how anesthetics work count as that, I see no reason to pursue that line of enquiry.

The computer holds the entire simulation - it would be analogous to our entire universe - making this argument analogous to the claim that disembodied consciousness can exist. Which is typically termed dualism in this forum.

It wouldn't be a disembodied consciousness within the simulation because it only exists in the function of the body one would see. You could only speculate about it being disembodied from an entirely different framework -- namely ours looking at the simulation from the outside. From anyone within the simulations point of view, they would see exactly what we see in our world -- minds are always and only functions of bodies.

From the other framework it would be obvious that it is electrons passing through gates that are responsible for the whole thing.

Is the fact that our minds are associated with our bodies just a minor issue? I don't see these issues are separate. After all, our bodies (and all material things) can be thought of as merely the actions of subatomic particles. It seems to me that either minds are associated with physical bodies or they are not. It is inconsistent to claim that our consciousness cannot be separated from our physical being but that consciousness in a simulation is 'real' and yet does not similarly require a physical substrate with which to interact with other conscious beings, be they simulated or human.

Yes, yes to the issue of our bodies being merely the actions of subatomic particles because that shows this in starker detail. Everything ultimately is just interactions of some stuff we can't define. The simulation consists of interactions alone.

Consciousness in a simulation is 'real' in the sense that if you carry out in a simulation exactly the same actions that occur in the real world in precisely the same way then you should see the same result -- in the simulation. The body in simulation is an action, which is why you cannot touch it. But the actions in the simulation are still actions and actions are not bodies.

If you have what we could call isomorphic actions, for the sake of argument -- they are identical in form in every way -- with actions defined as changing relations of parts, then in what possible way are the changing relations of parts in the simulation different from the changing relations of parts in the real world? I don't see any way to separate them -- they are just changing relations of parts. Sure in one situation the parts are simulated, but if the relations of the parts are identical, how are they different? I don't see a way to consider them different. This is true of all actions, not just consciousness. There is nothing special about consciousness in this regard.

 

[rocketdodger]

The computer holds the entire simulation - it would be analogous to our entire universe - making this argument analogous to the claim that disembodied consciousness can exist. Which is typically termed dualism in this forum.

No.

Within the simulation, the consciousnesses still need simulated bodies.

Think about it -- how ingrained in our consciousness is our body map, our past experiences, our sensory perception, etc?

To say that there could be a human-like consciousness without all the stuff we get from our bodies is just incorrect. Even with sensory deprivation, our entire memory base is built upon how our bodies interacted with the world in our past.

So a simulated consciousness, in a full universe simulation, still needs something like that.

 

[Meed]

Church-Turing says all this physical stuff should be computable. Consciousness is based in physical stuff -- it is the action of physical stuff -- so it should theoretically be calculable.

Yes. In other words, simulable. I haven't been disputing that.

We have been discussing many different things. I don't think anyone wants to claim that we could interact with a simulated mind in every possible way that we can interact with a real one, just that we should theoretically be able to interact with one.

This is what I was responding to when you jumped in:

You can't interact with simulated water in all the ways you can interact with water in your world.

You can interact with a simulated mind in all the ways you can interact with a mind in your world.

 

[Ichneumonwasp]

This is what I was responding to when you jumped in:

Ah, sorry, I would assume he meant that if we take away all the physical aspects of interaction -- like body language, etc. There are clear constraints on how such interactions would be possible. Though I guess if we are talking about linking the output to a robot, then we should be able to interact in all ways that didn't involve it being carbon based.

 

[Ichneumonwasp]

Cornsail,

Let me try this one out on you; I don't know if it will help, but it might explain more how I envision this.

Again, I don't know how possible such a scenario could be, but we are really just using it as a thought experiment.

So, we try to create a simulated world. Let's say we start with a very simple set of commands that simulates a particle. That tells the comuter which gates to open and close. We repeat this simple instruction many times and we have another set of instructions that tell this simulated world what the 'laws of physics' are. So, there is very little coding going on here.

Those simple sets of instructions tell the electrons (in the actual computer) where to go and in what sequence. If motion is introduced into the simulated world, then the simulated particles should interact according to the 'laws of physics', again just a few lines of code. But shouldn't those simple few lines of code be able to produce all sorts of different changing ways that electrons (in the actual computer) move through gates as the 'particles' interact? The code won't be growing, but the way the electrons move through the gates will change over time according to the rules of the game. Now assuming that we were God and could recreate our world precisely, knowing all the contingencies, we should need just a few bits of code to ensure that things happen in this universe just as it did in 'the real world'.

What I imagine is a similar pattern of those electrons passing through gates that match what occurs in someones brain with ions passing through channels, etc.

If a brain can produce consciousness through physical means, why could not a computer? The difference I see is that the "rules of the game" simply come from different places -- in 'the real world' we speak of them being inherent to the universe, and in the computer simulation they are imposed from the top-down.

 

[blobru]

... But what I was referring to was in a big simulation of the 'universe' there are still electrons passing through gates and some of those actions correspond to a particular individual, so why couldn't we just translate that electrical activity in a one-to-one way so that we could hear what was said in the simulated world, and speak to the simulated world?

Sorry to jump in, Ich_wasp -- Beth and others are posing some excellent questions and I'm having fun just following along as someone who's not convinced either way by the simulation arguments -- but I'm curious on a couple of points that you keep coming back to [hilited above] in your replies; how much does your conviction in the 'reality' of the simulated consciousness depend on:

-- electrons passing through gates? Would it matter if we replaced them with some other mechanism, like the push-pegs or gear-teeth in Babbage's jacquard looms and analytical engines?

-- one-to-one correspondence? Would it matter if the computer program used compression algorithms to store information (so that an array of 1000 neurons say, all in the same 'zero' state, was stored as "1000" (9 bits, 9 switches), rather than "000... rpt 1000 times ...000")?

-- ignoring the computer overhead that a simulation requires (as in "The Matrix", or "The Wizard of Oz" for that matter, when you pull back the curtain of the simulation, there's a massive machinery that generates it; so within the computer generating the simulation, some instructions will be 'assembly instructions' to tell the computer machinery what to do to generate the switch sequences which will be interpreted as the data for the simulation, thus simple state changes will require more operations: in physical reality, changing '1' to '0' seems a single operation; for virtual reality, the simulating program may have to translate a higher-level instruction like "change_state(quantum_string[1234567890])" into machine code to eventually locate the address of the bit tracking that state and direct the computer to mechanically change it from '1' to '0'; iow, there's a lot going on "behind the scenes" with virtual reality we needn't assume with physical reality)...

...to encapsulate the above point better (or worse): is there an inevitable ontological difference between virtual reality, which is prescribed by the computer program; and physical reality, which isn't -- necessarily at least, as far as we know (the universe itself may be a quantum computer, who knows? but not everyone is ready to abandon basic substance for "turtles all the way down" just yet) -- where the physical laws we observe are situated in a pre-existing spacetime vacuum that doesn't have to be described and to which things seem to conform, which seems to form things -- that's just the way the cookie crumbles / quanta fluctuate, so to speak -- rather than rule-following -- where the "physical laws" are specified by a computer program; that is relevant to consciousness? Even briefer: are relations really sufficient for consciousness, or must there be something there to relate? (this last two-part long-winded question elaborates a point others have been raising)

Note: I don't have a clue, as usual; just asking...

 

[Beth]

It wouldn't be a disembodied consciousness within the simulation because it only exists in the function of the body one would see. You could only speculate about it being disembodied from an entirely different framework -- namely ours looking at the simulation from the outside. From anyone within the simulations point of view, they would see exactly what we see in our world -- minds are always and only functions of bodies.

Again, this doesn’t eliminate the problem of consciousness existing without a body. If their consciousness can be disembodied from our point of view which supersedes theirs, why cannot our consciousness be disembodied from a point of view that supersedes ours? I do not think that you can be consistent in your view of what consciousness is and claim that simulated consciousness is ‘real’ but not dualism while also maintaining that the idea that consciousness could exist separate from human bodies is dualism and therefore not 'real'.

From the other framework it would be obvious that it is electrons passing through gates that are responsible for the whole thing. Yes, yes to the issue of our bodies being merely the actions of subatomic particles because that shows this in starker detail. Everything ultimately is just interactions of some stuff we can't define. The simulation consists of interactions alone.

If the simulation consists of interactions alone, then you are talking about disembodied consciousness. If you also maintain that a simulated consciousness is ‘real’ consciousness, then you are concluding that consciousness need not be tethered to a particular physical structure, such as a body belonging to the mind. If you believe that consciousness is ‘just’ interactions independent of the substrate it is enacted on, then how can you avoid concluding that our own consciousness could be disembodied from our physical being – i.e. that dualistic souls embodying the essence of our consciousness are possible?

 

[Ichneumonwasp]

Sorry to jump in, Ich_wasp -- Beth and others are posing some excellent questions and I'm having fun just following along as someone who's not convinced either way by the simulation arguments -- but I'm curious on a couple of points that you keep coming back to [hilited above] in your replies; how much does your conviction in the 'reality' of the simulated consciousness depend on:

-- electrons passing through gates? Would it matter if we replaced them with some other mechanism, like the push-pegs or gear-teeth in Babbage's jacquard looms and analytical engines?

-- one-to-one correspondence? Would it matter if the computer program used compression algorithms to store information (so that an array of 1000 neurons say, all in the same 'zero' state, was stored as "1000" (9 bits, 9 switches), rather than "000... rpt 1000 times ...000")?

-- ignoring the computer overhead that a simulation requires (as in "The Matrix", or "The Wizard of Oz" for that matter, when you pull back the curtain of the simulation, there's a massive machinery that generates it; so within the computer generating the simulation, some instructions will be 'assembly instructions' to tell the computer machinery what to do to generate the switch sequences which will be interpreted as the data for the simulation, thus simple state changes will require more operations: in physical reality, changing '1' to '0' seems a single operation; for virtual reality, the simulating program may have to translate a higher-level instruction like "change_state(quantum_string[1234567890])" into machine code to eventually locate the address of the bit tracking that state and direct the computer to mechanically change it from '1' to '0'; iow, there's a lot going on "behind the scenes" with virtual reality we needn't assume with physical reality)...

...to encapsulate the above point better (or worse): is there an inevitable ontological difference between virtual reality, which is prescribed by the computer program; and physical reality, which isn't -- necessarily at least, as far as we know (the universe itself may be a quantum computer, who knows? but not everyone is ready to abandon basic substance for "turtles all the way down" just yet) -- where the physical laws we observe are situated in a pre-existing spacetime vacuum that doesn't have to be described and to which things seem to conform, which seems to form things -- that's just the way the cookie crumbles / quanta fluctuate, so to speak -- rather than rule-following -- where the "physical laws" are specified by a computer program; that is relevant to consciousness? Even briefer: are relations really sufficient for consciousness, or must there be something there to relate?

Note: I don't have a clue, as usual; just asking...

1. I'm not sure why it would matter what the medium was, computer, banging cans, etc. But I don't know. I think it's easier to see this with ions flowing through channels and electrons through gates because we are used to thinking about the constraints on the systems and can visualize the movements to some extent.

2. I don't know enough about it to answer your second point, but I would guess that there are many ways to skin a cat. For the purposes of the discussion I think it is easier to see consciousness in a one-to-one correspondence.

3. Yes. And I don't know much about all that stuff.

4. I would tend to say that ultimately there has to be something there to relate or relation is not possible. But the weird thing thing that others have brought up in the past with this discussion is that maybe it really is turtles all the way down and relation is all there is? I mean, what is energy after all? Something that can relate to itself? Something that can vibrate at different frequencies in different dimensions to produce different effects?


Here's one of the problems I have with it all -- whatever instructions we give the hardware will have to be carried out by electrons through gates, but the actions that anything in a simulation will carry out will also be electrons moving through gates; so both the rules and what happens will be the same type of process in the computer. That seems kind of strange to me and might be a very big problem. It is certainly nothing like what we see in the 'real world', except that everything seems to be either energy or space-time and somehow they are tied together?

 

[Ichneumonwasp]

Again, this doesn’t eliminate the problem of consciousness existing without a body. If their consciousness can be disembodied from our point of view which supersedes theirs, why cannot our consciousness be disembodied from a point of view that supersedes ours? I do not think that you can be consistent in your view of what consciousness is and claim that simulated consciousness is ‘real’ but not dualism while also maintaining that the idea that consciousness could exist separate from human bodies is dualism and therefore not 'real'.

Because it is not actually disembodied. It's a framing problem. They would see their 'neurons' causing thought to occur, but it would really be electrons passing through gates; same issue with their bodies. We could be in a simulation right now and it would mean not that we have disembodied minds but that we are wrong about the nature of our reality.

If the simulation consists of interactions alone, then you are talking about disembodied consciousness. If you also maintain that a simulated consciousness is ‘real’ consciousness, then you are concluding that consciousness need not be tethered to a particular physical structure, such as a body belonging to the mind. If you believe that consciousness is ‘just’ interactions independent of the substrate it is enacted on, then how can you avoid concluding that our own consciousness could be disembodied from our physical being – i.e. that dualistic souls embodying the essence of our consciousness are possible?

It's only reality is that it performs the same functions -- same interactions -- as in the real world. It is still only occurs because of electrons passing through gates.

Let me think about that. This actually might mean that all other actions in a simulation that are not mental really are just digital representations of an action, which would be Cornsail's point as I take it.

So, to answer one of your points from previous posts, as I process through all of this, there may well be something different about mental actions as opposed to other types of actions. Or it may just be the way I look at them?

We could still speak of a simulated ball rolling as opposed to a simulated ball being simulated to roll -- which would have to do with the way the simulation was carried out -- but it may be that mental action is in its own category. Or that we could only get close to the way that it occurs in a computer simulation but never actually reproduce it in full, which probably wouldn't matter all that much.


ETA:
No, I think I'm falling into the same framing error.

I think the analgous situation, assuming monism is correct, to what you are proposing would be idealism. When we, in this world, think that the nature of reality is that there are physical things and consciousness arises from it we think we have a grasp on the nature of reality. But it might be the case that what really exists is the mind of God, which acts to produce the stuff that we see as matter; and the actions of that matter produces what we see as human consciousness. We could simply be wrong about the nature of reality, but assuming monism is correct this does not produce disembodied minds. The folks in the simulation would simply be wrong in their conclusion about the nature of reality. We might be too.

 

[rocketdodger]

To be technically correct, computationalists assert that consciousness is relations between relations between things, not just relations between things.

That is, there is (at least one ) extra level of relation on top of the base level of relation.

So really the question is whether or not a relation between relations between real things is ontologically different from a relation between relations between simulated things.