On Consciousness

[Piggy]

You can't tell the difference between yourself and a tree?

And neither can you, because "your experience of a tree" is simply part of what you are at that moment, if by "you" we mean the person I'm corresponding with at this moment.

When I'm looking at a tree, the green is me, the sound is me, the brightness is me....

When those things are, I am. When they stop, I stop.

 

[Piggy]

Ok, then. I'll chalk that up as "consciousness is mysterious". I hope Pixy answers my challenge with more substance, but if he does I won't have any opposing evidence to judge both your claims by.

Please don't misunderstand me.

You asked a question to which the only rational answer was to cite neurobiologists saying that the mechanism of consciousness is not known.

I can't help that, this is what you asked.

If you want to discuss the science, I'd love to.

What would you like to talk about?

 

[rocketdodger]

If you don't mind my saying, this is where your blind spot shows up again.

If you don't mind my saying, this is where you start off your response like every other post you make -- explicitly declaring everyone else to be wrong.

Our brains imagine unconsciously all the time. That's how we navigate our daily lives, and that's why we're surprised by a looming shape in the hall or a missing elevator.

So you are saying imagination plays no part in consciousness. Kind of a non-starter, since it is such an absurd proposition...

We can see this now. Brains use the same circuits for imagination as experience, and researchers have been able to observe some activity there as we (nonconsciously) imagine what's going to happen next at every waking moment of the day. (When we're dreaming, on the other hand, we don't seem to care what's going to happen next.)

Again -- you are claiming that this implies those circuits have nothing to do with consciousness ...

But we're not in any way consciously aware of this. It's not a brain process which our brain bothers to hallucinate about. And given the inevitable results if it did, it's no wonder that evolution didn't select for such an arrangement.

You know what is funny, in my last post I had actually made a sarcastic comment about how you will run back to some vague "evolution selecting for consciousness" argument the first chance you get, but I edited it out to be polite.

So that's really cool work, it's just off-topic here.

You didn't even read the paper, and you wouldn't understand it if you did, so don't patronize me.

It's amazing to me that you don't seem to understand that this research, while important for understanding the brain, no argument there, is nonetheless not about consciousness, because consciousness is not necessary here.

I think you misunderstand my intentions. I didn't expect you to change your position, or even think of changing your position, because I know you don't do that. In fact I expected you to just write off this real research as somehow irrelevant in your quest for the magic bean. Thanks for being so predictable !

The reason I posted that was so any observers would see that you are consistently dishonest when it comes to your representation of the opposition, and furthermore to show that there is real research that illustrates that the majority of your arguments don't hold much water.

I'm not really interested in a long discussion because I would rather read *actual research* than get bogged down in a typical piggy-esque merry-go-round of walls-of-text that, once fully parsed, contain zero new information for the reader. I also don't really care to be told, after I spend the time understanding some research that some very smart person has done, that "I don't understand why it is irrelevant." Hey, piggy, I'm the AI programmer, I'm the one that actually read the paper, I get to decide whether it is relevant or not.

 

[PixyMisa]

Great ! Help me, Pixy. Give me some links and references.

Sure.

Wikipedia gives a decent overview of computational neuroscience^WP.

The idea is that - since (a) mind is what brain does, and (b) the brain is a computer - if you model the activity of the brain at different levels of abstraction, you will get similar results.

So the Blue Brain Project^WP, for example, which started out with modelling a rat neocortical column at the molecular level and intends to continue scaling up until it models an entire brain (still in molecular detail). To quote project lead Henry Markram, "If we build it correctly it should speak and have an intelligence and behave very much as a human does." Which quote does nothing to establish my position, but does point out that researchers in the field believe that this sort of simulation will produce a conscious mind. (That's not Markram's primary reason for studying the brain, but it's his position.)

Here he is talking at a supercomputing conference if you're interested: .

And here's an article in Seed magazine about the project.

I also just discovered this: http://www.scholarpedia.org/article/Encyclopedia_of_computational_neuroscience

I haven't had a chance to do more than skim it yet, but it looks reasonable so far.

 

[PixyMisa]

About 12 minutes into that talk, Markram covers the point I was stressing earlier: That computational neuroscience is about building computational models at varying levels of granularity and testing them against each other, against real behaviour and against real biology.

Simulating a human brain at the molecular level will be enormously expensive, but it will allow us to derive more abstract models that accurately represent particular functions - to test and confirm what behaviours can be abstracted out without affecting the resulting behaviour.

 

[Mr. Scott]

If you want to discuss the science, you'll have to ask a different question, because there is no science dealing with Pixy's central claims, because they're false.

I get a whiff of potential circular logic there, but I think it's just wrong. Maybe computer science is not really science.

Is pixy's central claim that the brain is a computer?

I love it when tasks that are notoriously difficult for computers are cracked by computers simulating neural networks.

Please don't suggest that not knowing how to do something means it can't be done.

I just read How Many Computers to Identify a Cat? 16,000. Thanks for the link, guys!

The Google brain assembled a dreamlike digital image of a cat by employing a hierarchy of memory locations to successively cull out general features after being exposed to millions of images. The scientists said, however, that it appeared they had developed a cybernetic cousin to what takes place in the brain’s visual cortex.

One cool thing about computers is that any type of universal computer can simulate/emulate any other type. For example, an Apple can emulate a PC and vice versa, a PC can emulate a Nintendo, etc. Indeed, general purpose computers can emulate special purpose computers without difficulty.

Would someone explain again why a computer couldn't emulate a brain? Is it just because, if one could, we'd feel less special?

 

[punshhh]

This statement is meaningless, because there isn't a real distinction between "computational literalism" and the "neurobiological approach."

In fact I don't even know what "computational literalism" means. I just googled it and I don't see anything that looks relevant. Google doesn't lie, at least not these days.

Furthermore I don't even know what you mean to mean, because even after years of reading your rants I still honestly don't understand what your overall point is. People doing research in neurobiology and people doing research in computing deal with the same fundamental concepts of causation as every other scientist.

EDIT -- just so you can't say otherwise, here is a link: http://www.aist-pain.it/en/files/CONSCIOUSNESS/Consciousness,_Emotion_and_Imagination.pdf

That is about a robot some researchers got to move itself based on "imagining" the state that the resulting movement would lead to, using nothing but neural networks wired up somewhat similar to a known pathway in the mammalian brain. If you want me to explain in more detail I can, but suffice to say the research is 7 years old and yet it pretty much invalidates every single point you have ever tried to make in these discussions. In particular, it is 1) neurons that exist only in code controlling a real robot and 2) doing it in a manner that is similar to how our neurons might control us, at least on a very limited and simple scale.

I agree in principle with your position, accepting my lack of knowledge of the detail.

However I return to my principle point, that you are proposing an intelligent entity, which behaves the same as a conscious entity, is essentially identical in terms of performance.
Why would it ever need to be conscious?
Are you assuming or hoping that it would somehow become conscious as a sum of its parts?

In the examples we have of conscious entities, consciousness precedes intelligence. The evidence suggests that the principle requirement for consciousness is life and that consciousness is a consequence of life. An evolutionarily advantageous accident. Occurring in a chemical system or soup in which there is little intelligence and computation only in the broadest sense of physical causality in action.

I see no reason why AI would ever require consciousness (not the next few hundreds or thousands of years anyway).

 

[PixyMisa]

However I return to my principle point, that you are proposing an intelligent entity, which behaves the same as a conscious entity, is essentially identical in terms of performance.
Why would it ever need to be conscious?

Because otherwise you can't joots. You'd be perpetually sphexicised.

 

[PixyMisa]

I get a whiff of potential circular logic there, but I think it's just wrong. Maybe computer science is not really science.

Is pixy's central claim that the brain is a computer?

Which would be a very odd thing for anyone to dispute.

A computer is any system that performs computation.
The brain is a system that performs computation.
Therefore the brain is a computer.

 

[dlorde]

Name a process which is not an event.

It seems to me that a process is generally taken to be a causally connected sequence of events. But ISTR this has already been discussed, and it's Pixy you should address this to.

But yeah, consciousness does not have to be introspective at all.

Just out of curiosity, why would you think that it would have to be?

I don't think introspection is the right word - in my experience introspection refers to redirecting the focus of attention from the sensorium to the internal model of self, i.e. it's a higher level of abstraction, consciousness reflecting on itself.

The self-referencing that I see as the basis of consciousness involves feedback - feeding back the evaluated results of behavioural responses to environmental stimuli into the process that selects or generates behavioural responses. This permits adjustment of behavioural response based on the results of previous behaviour, leading to behavioural flexibility. At a minimally conscious level, this might just be a positive or negative association between environmental stimulus and behavioural response that biases subsequent behaviour selection. At a higher level of complexity, an accumulated history of such information can be used as an internal model of the organism's interaction with the environment, allowing the development of an ongoing sense of self, predictive 'what if' modeling, and at the highest levels, theory of mind and introspective self-awareness, self-referencing at a higher level of abstraction.

It's the amount and detail of the information accumulated and the sophistication of the evaluations involved that determine the level of consciousness, but it's all based on self-referential feedback.

I suppose the fly in all this ointment is the definition of consciousness itself. You may agree or disagree with some or all of the model I outlined, but if we each mean something different by consciousness, we're going to talk past each other anyway.

 

[dlorde]

Computational literalism (or opportunistic minimalism, to use Dennett's term) is the notion that your brain literally is a computer, and that consciousness may be reproduced in machines by reproducing only the syntax but not the semantics of the process... in other words, timing doesn't matter, the shape of the brain doesn't matter, you can make a conscious brain out of rope, etc.

OK. Well, for the record, I think timing is likely to be essential, the architecture of the system is crucial, and I can't conceive of a conscious brain made out of rope.

Having said that, timing and architecture are crucial to the effective function of many computer systems, and syntax or semantics, it seems to me that computational equivalence is the essential issue.

 

[dlorde]

When one of the aspects of waking existence that helps to define everything else is attempted to be defined, you will often encounter similar types of statements. Sensation, for whatever reason, is fundamental to how we go about figuring things out. Sensation as a part of our models in terms of it's ontological status, need not be as fundamental (although there is nothing precluding it either).

I understand, sometimes it's hard not to end up explaining or defining things in terms of themselves; but then it's not really a useful explanation or definition. But I got the gist of it.

Btw I meant 'circular' rather that 'tautologous' - it was late, I'd finished the wine

 

[dlorde]

Of course, by such a simple definition, any computer is conscious, but the "self-aware" part of the definition pretty much screams "introspective", no ?

I dunno. Does being aware of 'something within oneself' equate to what we generally mean by 'self-awareness'? I think of self-awareness as being aware of oneself [as an entity], i.e. more than being aware of 'something within'.

Maybe it is just ambiguous phrasing.

 

[Belz...]

That's right, "the brain is not a computer" is not a claim, because they are obviously different things (which is why we have different words for them) and if someone wants to say one's the other, then we need an explanation.

Rocketdodger has already mentioned the square vs rectangle thing. When someone says the brain is a computer, he or she is not saying that it's a pentium, and you know it.

Brains are bits of organic goo that are great at Frisbre and bad at math, computers (real ones we have now) are quite the opposite.

And yet they both do the same kind of work: computing.

If you want to say that two objects are the same thing, then you've got to show either structural of functional equivalence.

No. See, if I want to say that helicopters are aircraft, I don't have to show you that they are identical to planes.

Now if we start by comparing real brains to real computers, as they exist today -- and that's the safest place to begin -- we find that they're not similar in structure or in function. (See Frisbee/math above.)

Again, how ? They do the same kind of work.

It means we have to say "What is a computer" and explain why the brain and my PC are both computers, but my desk and the ocean and the stars are not.

According to wiki, computation is a form of calculation, which is "a deliberate process for transforming one or more inputs into one or more results, with variable change." Personally I'd add that it has to include some form of abstraction.

You raise the point of computation. If brains perform computations, doesn't that mean they are computers?

Actually, no.

Well, then, I see where our disagreement lies.

So you can talk about the computational capacity of hurricanes, for instance. And Wolfram does.

We've been there before. It's like saying running is a process, and then some philosopher jerk comes in and point out that everything, down to elementary particles, is a process. We know that, but we can use the narrower definition for the purposes of talking about running, can't we ? I don't know why we can't do it with computing.

Similarly, why not say that computers are brains, or can be brains? Hard to argue against that.

I wouldn't have a problem against that, but that would be using a definition of computation that specifically excludes brains, which would be weird.

 

[PixyMisa]

OK. Well, for the record, I think timing is likely to be essential, the architecture of the system is crucial, and I can't conceive of a conscious brain made out of rope.

You can't build a computer out of nothing but rope, so Piggy's objection there is nonsensical.

But timing is syntax. Or if that's too hard to swallow all at once, timing can be expressed as syntax. So that objection is also null.

 

[Belz...]

Computationalists seem to want to then jump to "self-awareness" as the next possible requisite for consciousness, perhaps because they can fit "data about the self" into their model.

Makes sense to me, since it's what we usually mean with the word.

Core consciousness is handled in the brain stem, and is evolutionarily early. Self-awareness, which we seem to share with dolphins and chimpanzees and gorillas at the moment but who knows what we'll find, is higher-order in the brain and later in evolutionary terms, so it's not a requirement for consciousness.

Sorry, it doesn't seem to follow, for me. Who cares if it's late development ?

Consciousness is what happens when you wake up or dream, and what isn't happening when you're deeply asleep. It's color and sound and odor and texture and hot and cold and all that.

Is it ?

 

[Belz...]

Please don't misunderstand me.

You asked a question to which the only rational answer was to cite neurobiologists saying that the mechanism of consciousness is not known.

I can't help that, this is what you asked.

If you want to discuss the science, I'd love to.

What would you like to talk about?

I'd like to do the bolded part. Saying that the mechanism is not known is fine, but not quite true because we do know quite a bit about it, even if we don't understand it completely.

 

[dlorde]

So what we're really saying in common usage is that a hallucination is a non-normative experience (at the least).

My point here is that our everyday experience is essentially a normative hallucination...

1. An hallucination is a non-normative experience.
2. Everyday experience is a normative hallucination.

Therefore, everyday experience is a normative non-normative experience...

This really doesn't work for me.

Nor does this:

The brain produces colors and brightness...

 

[Jeff Corey]

... Sensation as a part of our models in terms of it's ontological status, need not be as fundamental (although there is nothing precluding it either).

The sensation of reading that led to a complete lack of perception on my part.
I'm not kidding.
I thought I knew what all the words meant, apart from each other, but the gestalt is somehow less than the sum of itz parts.

 

[dlorde]

The sensation of reading that led to a complete lack of perception on my part.
I'm not kidding.
I thought I knew what all the words meant, apart from each other, but the gestalt is somehow less than the sum of itz parts.

My feelings exactly.