At Aeon, philosopher Philip Goff argues for panpsychism:
It’s a short essay that only takes a couple of minutes to read.
Goff’s argument is pretty weak, in my opinion, and it boils down to an appeal to Occam’s Razor:
I maintain that there is a powerful simplicity argument in favour of panpsychism…
In fact, the only thing we know about the intrinsic nature of matter is that some of it – the stuff in brains – involves experience… The theoretical imperative to form as simple and unified a view as is consistent with the data leads us quite straightforwardly in the direction of panpsychism.
…the brains of organisms are coloured in with experience. How to colour in the rest? The most elegant, simple, sensible option is to colour in the rest of the world with the same pen.
Panpsychism is crazy. But it is also highly likely to be true.
I think Goff is misapplying Occam’s Razor here, but I’ll save my detailed criticisms for the comment thread.
I am not seeing an important distinction there.
That seems right to me — that it isn’t processing information.
I take out a thermometer to measure the air temperature. I look at the mercury level in the tube. It’s half-way between 37 and 38. Should I call that 37 or should I call that 38?
Suppose that I decide it is 38. Then that “38” is information. But what I was doing was measuring. It wasn’t information processing. There wasn’t information until I had completed the measuring.
Information processing is a syntactic operation on symbols, and those symbols might just as well be meaningless. Measurement is a semantic operation, where meaning is very important.
What the neurons are doing is closer to measuring than to information processing.
The accountant is balancing the books. The accountant’s brain is not balancing books. Doubtless, the accountant couldn’t do it without using his brain. But it does not follow that the brain is balancing books.
That is why you are my favorite
Neil,
I keep asking you what the brain actually does, if it doesn’t process information. You keep avoiding my question, for an obvious reason: you don’t have a good answer.
Let’s take another look at the example I gave to petrushka, which disproves two of your claims:
First, note that there was information present in the list of numbers. That information was received by your visual system. Therefore your claim that the senses don’t receive information, but instead only “manufacture” it, is false.
Next, note that you add up the numbers in your head, without writing down any intermediate results. All you write down is the final result. The addition obviously took place in your brain, unless you want to venture that it was farmed out to the liver, or something like that. Addition is information processing. Obviously, then, your brain processes information. Therefore your second claim is false. Brains do process information.
Up to here, we agree.
I disagree with all of that. I did the adding. My brain did not do the adding, though I could not have done it without using my brain.
The difference between me doing something and my brain doing something borders on equivocation. It is invoking a petty dispute over terminology rather than digging for common understanding.
I use brain as a shortcut to saying person or animal having a brain.
Adding numbers is something a person can do. The first people hired to do this as a profession were called computers. Electronic computers are named after these professionals.
I would agree that people can process data. It is an activity or behavior.
But it has nothing to do with what neurons are doing when a person sees blue or feels pain.
This is pitiful, Neil.
I just demonstrated, clearly, that your two claims are wrong. You have no rebuttal. Your errors are obvious. Why deny them?
The information was already there in the list of numbers. Your visual system did not “manufacture” the information — it received it. This is obvious. If you had been blindfolded, or had kept your eyes tightly shut, you wouldn’t have seen the numbers. The information wouldn’t have gotten to you. Your first claim is therefore incorrect.
And you’re still avoiding this question, which I am now asking for at least the sixth time: If the brain doesn’t process information, then what does it do?
And in terms of my example, you say that the brain isn’t performing the addition, but that you are. What is your brain doing, then? And if your brain isn’t performing the addition, then how is it getting done? Are your two kidneys taking on the job, perhaps with help from your left femur?
Your position is beyond goofy.
petrushka,
But you’ve denied that brains process information. That makes no sense. It’s the same error that Neil is making.
Neither does keiths, apparently, so keep bickering you two.
But with regard to mind, consciousness, and awareness, it is a vital distinction. A computer or robot is decidedly unconscious because it only rearranges objects or bits of data (mechanically, as per a pre-fed program). Humans are conscious because they comprehend messages. Computers and robots do not see the information value in the data. They only act-react. People comprehend, and then they can decide whether to act-react or not. Or sometimes people are inattentive and fail to act-react where they should, whereas computers never fail, as long as they are plugged in and switched on.
The difference is as important as alphabet versus a written message. Of course you cannot write a message without an alphabet, but feeding an alphabet to the computer does not make it able to comprehend or produce messages. Anyway, I guess you are one of those dudes who thinks syntax equals semantics or, when syntax is perfect, semantics will take care of itself. It won’t.
yes an algorithm.
Therefore comprehension is non-algorithmic!
I think this is a very important insight. One that can perhaps be exploited to help with determining whether the source of a particular “artifact” is conscious or not.
peace
fifthmonarchyman,
Others are talking about conscious beings, but you are talking about “whether the source of a particular “artifact” is conscious or not”. Your confusion is so deep that nobody can help you. Ever. But here’s a little clue.
You think that when a sequence is non-algorithmic then the source is (could be) conscious. But what sequence are you looking at? You can extract all sorts of imaginable sequences out of all sorts of things.
There are computers that store and execute programs. Programs are algorithms. But who made those programs? Humans did, a conscious source. Therefore your logic was a dead end from the start and nobody wants to talk with you about it unless they are in for some laughs.
Erik,
It was just a throwaway comment.
Relax and have a nice day.
peace
A throwaway comment that you keep throwing around for years…
If you had said “Turing machine computations” instead of “information processing” I would agree with you completely. Turing machine computations are syntactic operations on meaningless symbols.
However, to say that all information processing must be understood in terms of Turing machines, and in no other way, would imply that the entire field of computational neuroscience rests on a fundamental error. It takes more chutzpah than I possess this morning to say that a whole field of science is completely mistaken.
I was talking about all this stuff with a friend of mine the other day (also a philosopher), and he raised a nice problem that I need to worry about a lot more than I do: suppose we have, on the one hand, a phenomenological description of lived experience at the person-level. And suppose, on the other hand, we have a nice neuroscientific theory or family of theories about neuronal composition, chemistry, structure, patterns of interaction, and so forth. What else do we need?
The founding idea of cognitive science was that there’s a real need for a distinct science, neither phenomenology nor physiology, but a tertia quid. Something that would be at the sub-personal level, genuinely explanatory and not merely descriptive, but conceptually distinct from neurophysiology and irreducible to it.
The question was, how to conceptualize this tertia quid? And the founding idea, developed by Ulric Neisser and Zeno Pylyshyn and others, is that we can conceptualize this tertia quid in terms of computation. The nice thing about computing machinery, as Turing stressed from the outset, was that syntactic operations over meaningless symbols can be understood without concern for the machine is actually built. As long as the machine can run a program, it doesn’t matter what the machine is made of.
The software/hardware distinction was appealing to cognitive scientists and philosophers of mind for two reasons, one empirical and one conceptual.
The empirical reason was that neuroimaging didn’t exist yet, and neuroscience was relatively crude. Only EEGs allowed for recording of what living brains were doing and they were very limited. A lot of neuroscience involved lesion studies, which is basically trying to figure out how a radio works by getting a lot of them, breaking different pieces in each one, and then turning it on and seeing what happens.
The conceptual reason was that mind-brain identity theory, as in the old “pain = firing of C-fibers” of mid 20th century philosophers like Place and Feigl, had very serious problems. These problems became much more serious when Kripke closely examined the concept of identity using possible world semantics.
But now the empirical problem is not as serious, because we have ways of examining brain activity much more precisely. The conceptual problem is still serious, but the upshot of Kripke and Chalmers for philosophy of cognitive science has been minimal. Their views about mind depend on their views about meaning, and how to understand how intensions are mapped onto possible worlds. Chalmer’s argument for the possibility of zombies depends on two-dimensional semantics, because there’s no other way of getting from conceivability to possibility. And even if one were to establish possibility, it wouldn’t matter to neuroscience, because neuroscience (like all sciences) is concerned with the actual world, not possible worlds.
In light of those considerations, I think it’s going to be a live concern going forward as to whether the need for a tertia quid between phenomenology and neurophysiology is as compelling now as it was fifty years ago.
I do not believe brains (neurons and synapses) perform logical operations on tokens.
Suppose I aim three Hollywood wind machines at a structure and turn them on one by one. After the third one reached speed, the structure collapses. Is this an addition operation?
One could build a computer out of wind machines blowing things down, and perhaps program it to do your taxes. That is a bit like brains (or humans) doing math and logic.
Math and logic are necessarily sequential, but brains are massively parallel. Humans have language, which is necessarily sequential, but it is a recent add-on. In CPU design terms, it is a bag. ( http://www.thefullwiki.org/The_Soul_of_a_New_Machine ) Emulating language behavior is not emulating the underlying architecture of brains.
But while we are doing thought experiments about zombies and such, let me propose a hypothetical.
Suppose it had turned out to be easy to build machines that behave like humans.
Suppose in 2001 we had HAL computers discussing their inner thoughts and feelings.
I guess something that doesn’t process information, the brain, can conceive of and produce machines that do process information, apparently without a model of anything that does.
How remarkable.
Glen Davidson
Incidentally, unlike both Comte and petrushka, I haven’t claimed (and don’t believe) that these questions can’t possibly be answered. (Just one more to add to keiths’ nearly endless list) What I’ve said (and believe) is that they are not and will never be empirical questions that can be answered by scientific investigation. They’re “heavyweight” philosophical questions. I don’t say they can’t be solved once and for all….but based on the history of philosophical disputation, I doubt it.
But I would not have said that.
I see the Turing machine as part of a formal model for computation. I don’t see it as involved with information processing. What the bank computer does with financial records is information processing, because the financial records are information. What a Turing machine does is entirely solipsistic — disconnected from reality.
When I disagree with something that keiths said, I am not asserting that what keiths said is false. I am merely expressing disagreement. My disagreement is with the way that he is using the word “information”. I know what keiths means, and my disagreement is not with what he means but with his way of expressing it.
Personally, I want to use “information” in a very constrained way. I’m well aware that the word is used more broadly. But I am also well aware that such broad use has resulted in a great deal of confusion. The “hard problem” is one example of that confusion.
Let’s go back 50 years. Maybe I was listening to the radio. The news program on the radio would have been considered to be information. The music program would not have been considered information. Perhaps the music announcer might have said something about the composer of that music. And what he said about that would have been information. The music itself would not have been considered to be information.
Back, at that time, “information” normally referred to speech or similar. It was always symbolic. Light reflected off a surface would not have been considered to be information. Thinking about the news would have been called “thinking about the news”. It would not have been called “processing information.”
What has changed since then — the changes mostly come from information processing with computers. We have Shannon’s theory of communication, which actually fits pretty well with traditional usage of “information” as symbolic expression. As far as I can tell, the idea that light reflected off a surface counts as information, arose from the physical symbol system hypothesis of AI.
The physical symbol system hypothesis was a terrible idea. One of the reason for my constrained use of “information” is to avoid that kind of mistake.
I just ate breakfast. Maybe that cereal counts as information, and maybe eating breakfast counts as processing information. Common language usage changes with fads. But when the context is discussion of human cognition, I think we should put those fads aside and be more strict about how we use words such as “information.”
In my preferred way of talking, brains do not conceive of and produce machines. People conceive of and produce machines. Yes, the people are using their brains while doing that. But the conceiving of and producing — that is properly attributed to people, not to brains.
Electronic computers are to human computers as power tools are to human hands. They are functional extensions.
The problem with the phrase “process information” is it reifies information.
Saying people can take and do logical operations and arithmetic is sufficient. There is no value added by saying they process information. People were computing before electronics and before Babbage. The information processing metaphor is derived from the world of electronic data processing, which derived from human computers, which are a rather recent thing in human history.
But electronic computers mimic just one thing that humans do. We built cameras in imitation of eyes, but the activity of seeing is a great deal more complex than forming images. Cameras can resolve better than eyes, but self driving cars have run over people in a situation where a competent human driver would not. the car computer was not asleep or drunk or inattentive. It merely did not actively see in the way that humans see.
I’m pretty sure that this defect will be overcome, and autonomous cars will soon be safer than human driven cars. They might achieve the brainpower of a housefly.
The “therefore” — what is it there for?
I can agree that comprehension is not algorithmic. But you cannot conclude that from your starting “premise”. Your “therefore” is just wrong.
The “it” being discussed is the distinction between “data” and “information”.
In all honesty, Shannon’s theory of communication greatly changed what we mean by “information”. And the distinction that you still see as vital, no longer matters.
FWIW, Samuel Butler said much the same thing a long time ago.
Let me try to get at the point in a different way.
Consider the bar code reader at your local store.
While reading a bar code, it is continually moving where it is pointing. That motion is behavior.
Eventually, it gets a number recorded in memory from reading that bar code. How does it get that number? It isn’t just from the light. Reading the bar code takes input from the light sensor, but it also takes input from the system clock. The precise timing of when the light signal is detected, is involved in reading that bar code. If you describe it as the light carrying information which is then processed, then you are missing almost everything that happens.
That’s why I am being fussy about “information”.
If we look at human visual perception, we also see behavior. And timing is probably important. There’s far more to it than just receiving information carried by light and processing that information. The meaning or interpretation of what is seen is going to depend on the behavior used to get that information. And that meaning is likely an important part of perceptual experience.
That passage suggests to me that you DO think the information is out there to be processed.
I would say the information metaphor is unhelpful rather than wrong.
Analogies and metaphors are not reality. If they are useful in discussion or useful in generating research ideas, then keep them.
But the brain as computer analogy has not really generated AI.
What has generated useful AI is attempting to do things like play games and drive cars. Until literacy became mandatory for living in civilization, there was really no great penalty for having an IQ in the mildly retarded range. People who lack aptitude for academics are really not distinguishable in a world where hunting and gathering are the primary means of making a living.
It says something about AI research that Deep Blue plays chess, but computers have trouble driving a car. I think this is relevant to discussing what it means to be aware or to have qualia experiences.
Sure, I happen to be interested in things like comprehension and consciousness. I think it’s probably one of the most important topics we can ponder
If you are going to point out that out one of the things that makes comprehension special is that it’s non-algorithmic. You can expect me to speculate that perhaps that quality is something we can recognize in other entities besides ourselves.
If on the other hand you think that there is no way to know if another entity is conscious (has comprehension) that is interesting to me as well.
Either way there is no reason to derail a thread you can take it all with a grain a salt if you like.
peace
Recall the starting premise was.
quote:
A computer or robot is decidedly unconscious because it only rearranges objects or bits of data (mechanically, as per a pre-fed program).
end quote:
Why exactly can we not conclude that consciousness is not about rearranging objects or bits of data (mechanically, as per a pre-fed program) from that premise.???
peace
Because the conclusion doesn’t follow from the premise.
It seems to me that the conclusion is simply a restating of the premise positively rather than negatively.
Premise) Computers are not conscious because they are only algorithmic.
is simply flipped to
Conclusion) conscious agents are more than just algorithms
What am I missing?
On a related note check this out. It suggests an interesting way to test if a computer is like a human.
http://theconversation.com/how-to-build-a-computer-with-free-will-92234
I think it’s a step forward but needs to take randomness into account to be viable.
peace
Shannon’s theory has greatly confused you. When you want to learn about comprehension of meaningful messages, you should reasonably turn to psychologists and linguists, not to mathematicians or computer scientists.
Define randomness.
quote:
Random : lacking a definite plan, purpose, or pattern.
end quote:
from here
https://www.merriam-webster.com/dictionary/random
As you know I don’t think that actual randomness exists. So when I speak of randomness most often I mean apparent randomness and by “apparent” I usually mean apparent from my particular perspective as the observer/evaluator.
peace
petrushka:
keiths:
petrushka:
So? The question is whether brains process information, not whether they “perform logical operations on tokens”. The latter is just one form of information processing.
And of course “performing logical operations on tokens” is something that brains can do, as when a philosophy student analyzes a logical argument.
Since you agree that people can process information, but deny that brains do so, perhaps you can answer the question that Neil is avoiding:
Neil:
walto:
He’s already admitted that in my addition scenario, the list of numbers contains information. Obviously, he takes that information in through his visual system before performing the addition.
He just doesn’t want to admit that. It blatantly contradicts his claim that perceptual systems don’t receive information from the outside world, they only “manufacture” it.
walto,
Which is exactly what Comte believed regarding the composition of stars. His verdict was premature, and so is yours.
Neil:
Good grief, Neil. A Turing machine can do everything the bank computer can do, including processing financial records. That’s precisely why it works as a model of computation.
You were a professor of computer science. Do you really not understand this?
Neil:
It doesn’t in any way depend on that hypothesis.
Consider my addition scenario. Before you look at the list, you don’t know what the numbers are. After you look at the list, you do know what the numbers are. Your uncertainty has been reduced. That’s precisely what Shannon information does, and the information was carried to you by the light reflecting off the list.
If you had been wearing a light-tight mask, you wouldn’t know the numbers and your uncertainty regarding them wouldn’t have been reduced. In other words, no Shannon information.
The light carries the information. Block the light and you block the information.
This is frikkin’ obvious, Neil.
Turing machines don’t actually exist, except as abstract models.
That is not at all obvious. The visual system is picking up ink marks on paper. That’s not the same as picking up information.
Neil:
Guess again. (That’s a very cool project.)
In any case, didn’t it occur to you how dumb your objection was, given that you had just told us that “what a Turing machine does is entirely solipsistic”?
You talked about “what a Turing machine does”, believing that no such machines actually existed, and then criticized me for talking about what Turing machines can do. Goose, gander.
keiths, to walto:
Neil:
Christ, Neil. Read this again:
Consider my addition scenario. Before you look at the list, you don’t know what the numbers are. After you look at the list, you do know what the numbers are. Your uncertainty has been reduced. That’s precisely what Shannon information does, and the information was carried to you by the light reflecting off the list.
If you had been wearing a light-tight mask, you wouldn’t know the numbers and your uncertainty regarding them wouldn’t have been reduced. In other words, no Shannon information.
The light carries the information. Block the light and you block the information.
This is frikkin’ obvious, Neil.
Also, regarding your odd “solipsism” objection: Why on earth wouldn’t the bank computer be just as “solipsistic” as the Turing machine, if both were carrying out the same computation?
The question is not whether people with brains can process information, but rather, what is the most useful model, going forward, for imitating brains.
Assuming that is a useful project.
If you can imitate brains, you are on the road to understanding how they work and eventually, “why” people experience qualia.
I see nothing in this discussion that contradicts my personal observation that the imitation game has focused on verbal behavior and on logical operations. That has been a very useful path, and John Henry the human computer has been defeated. Verbal and logical reasoning is how we got things like penicillin and Novocain and really cool digital watches.
But the underlying architecture of humans and animals is not anything like an electronic computer, even though some narrowly observed behavior looks like logic circuits.
The hard problem seems to have arisen from the failure to imitate brains. If we had HAL or Data, we would be asking different questions. Or the same questions differently.
petrushka:
Correct. We all agree on that.
Here’s the question:
Since you agree that people can process information, but deny that brains do so, perhaps you can answer the question that Neil is avoiding:
No, not a Turing machine.