No, that is not what the p-zombie thought experiment is about, as envisaged by Chambers and Kirk. A true p-zombie would be physically identical to a conscious person, and have no physical differences at all from that person. That means all the difference would exist on a non-accessible, and non-analysable level of existence.
I’m quite hopeful about the possibility of non-conscious but competent entities, and it would be a great achievement if we managed to create AGI without consciousness and without the capacity for suffering. We could use non-conscious AGI to stand in for humans in performing dangerous and demeaning tasks that humans are not suitable for. Slavery of humans is not an acceptable practice; using a non-conscious AGI as a tool for any task would have no ethical problems.
However the internal design of a non-conscious AGI would be entirely different from that of a human (even if it looked like one externally), so it would not be a p-zombie.
As far as the whole billiard ball robot scenario, I question if it fully captures the complexity of living behavior. I’m not saying that billiard balls have consciousness or need consciousness. And I’m not discounting the hypothetical billiard ball machine.
Rather, look at the process of how cells interact to form a single cohesive being. Can we really explain how that works? How muliticellular clumps became mobile, reactive, autonomous beings? Hell, how do single cells operate, with organelles and complex structures that we can at least make sense of the chemical reactions at work?
Life is made of biochemical reactions, which should be just like those billiard balls. But how does the complexity of animals that move around, chase down prey or scamper for safety, and all the other things, arise out of chemistry?
We will never be able to examine this gap with billiard balls, but we might be able to with silicon chips.
I’m of the opinion that the nature of consciousness is a necessary outcome of the complex interactions that create an autonomous being, i.e. an animal.
Do insects have an internal experience? There is some rudimentary processing to avoid dangers and search out food or shelter. I doubt there is the full on experiential being that humans are, just because the seat of human experience seems to be our brains, and insects don’t have that.
But complex mammals? Birds? Even fish, or sea creatures like octopuses or squid?
It’s all well to project creating and programming a simulation, but making a fully autonomous robot is different in scale than, say, making a Mars rover that operates from a time delay from Earth that real time control is not possible.
Think about Pathfinder crawling across the face of Mars. It was an impressive run. Except it had daily commands sent from headquarters, and even that didn’t always work right. Imagine a Pathfinder that did everything the same, except it started out with every bit of knowledge that could be programmed, that had to solve the challenges it faced without any further progamming or input from Earth. You’re not going to get there coding every possible problem and solution directly.
The thing that makes animals possible is the place where awareness, consciousness, and sense experience and reacting takes place.
That doesn’t contradict what I said.
There are two aspects here; whether it is possible to make a p-zombie, and whether we can distinguish between a p-zombie and a conscious entity.
I said that we don’t know whether we can make a p-zombie (or whether p-zombies can exist; the two statements are equivalent). It may be that it is impossible, as an inner experience is intrinsically tied to higher cognition in ways that we are currently ignorant.
Whether we can distinguish between a conscious agent and a p-zombie is a different question. As you say, p-zombies were originally conceived as being physically identical, which on its face would seem to make distinguishing them an impossibility. But I’d have two objections:
What do we mean by physically identical? Does it include an identical past? A hypothetical model of consciousness might require a flow of states, right from the womb to arrive at the consciousness that we later enjoy. And if we find that consciousness is tied to things like entanglement states then statements like “physically identical” start to be quite fuzzy in meaning.
I don’t think these are likely to be a concern, but while we’re speaking abstractly we shouldn’t implicitly take things off the table.
The concept of p-zombie has been broadened in many discussions and I would be on board with that. Because, even if it is the case that the physical structure of a given human’s brain necessarily entails consciousness (thus eliminating the original concept of a p-zombie), it may still be possible for an entity that behaves entirely like a human to be a p-zombie.
Indeed, hypothetically, we could find at some future time that there’s a particular structure in the human brain responsible for subjective awarenes,s but there are neuropathies that either diminish subjective experience, or it may be absent entirely in some individuals. Discussing p-zombies helps us consider some of the implications of this.
Until recently, I didn’t believe that was possible (as I’ve said upthread). I thought that ‘if it walks like a duck and quacks like a duck, it is a duck’. But now I’ve seen the way large language models can mimic human speech responses, I think that entities which behave entirely like a human but which are not conscious may be both possible and a desirable goal for AI research. A ‘large behaviour model’ could replicate human behaviour so faithfully that it would be impossible for a true human to distinguish the difference, even over a lifetime of exposure. But it wouldn’t be conscious.
I am somewhat concerned, however, that by creating such incredibly competent and convincing replications, that we might not accidentally create a different kind of consciousness, possibly one that has non-human qualities that are equal but different to our own. Philip K. Dick, in his paranoid state of endless creativity, realised that this was a possibility long ago.
It’s massively more complicated than billiard balls banging into each other, no matter how many there are; there are not only collisions but also connections of multiple types and multiple periods of time from momentary to lifelong, electrical and chemical and spatial. And maybe somewhere in that wetware mess there is some combination of those things that, by necessity, produces the (by us poorly-defined) thing we call consciousness.
I can imagine it at least as well as I can imagine your p-zombie. Probably better.
I doubt that anything is going to behave “exactly like a human” unless the entire glandular system, at a minimum, is also replicated; along with all of its connections to all portions of the brain.
“Producing speech that comes across to humans as being human speech” isn’t remotely the same thing.
Maybe somebody’s working on that and I just don’t know about it.
And yet, for all of that, it’s true that the physical parameters fully determine the outcome of any interaction—or else, you’re already saying that there must be something beside the physical, thus accepting what the zombie argument sets out to establish.
Yes, maybe. But if the thesis of the causal closure of the physical holds, then—and this is a strict logical implication—that consciousness being produced does not influence what happens physically, hence, isn’t necessary to reproducing the same exact behaviors: we could imagine doing without.
I thought that, until then Chat GPT material came online. I thought we’d have to reproduce the way that adrenaline, serotonin, oestrogen, testosterone and oxytocin affect mood, as well as blood sugar levels and various recreational drugs including alcohol.
But now I think that these behaviours could all be reproduced by a large and comprehensive database of observed behaviours, with non-conscious systems learning how to mimic human behaviour over time. We’d be able to detect and probably dismiss such mimicry at first, but it would improve over time, until most observers would fail to detect the difference. Human behaviour is sufficiently varied and unpredictable that there would be plenty of false positives and false negatives over time, until there would be almost no difference at all.
Well, I think it’s important to separate opinion from fact on these things. When I have been saying “we don’t know” it doesn’t mean I don’t have an opinion about the phenomena. But it’s just an objective fact that there is no consensus, either among neuroscientists nor philosophers, on the mechanism of consciousness. And, again, no-one has a model that could be used to answer some of the practical questions I’ve listed upthread, on subjective experience, sleep, p-zombies, identity etc.
No-one should have any degree of conviction that p-zombies are impossible at this time, LLMs or not.
Agreed. I don’t think @Half_Man_Half_Wit is wrong to use this kind of description, but I think we should disfavor it, as it is extremely misleading.
This kind of description leads people to imagine behaviour as some singular, sequential process in response to some external stimuli. The reality is of course that it is massively parallel, and the vast majority of the processing is based on internal data. It’s a bustling city where data from the outside world is just a little pigeon landing on the ground.
I also think (and probably HMHW would agree) that there are likely many layers of abstraction employed by the brain and trying to understand consciousness is almost certainly going to involve figuring out those layers of abstraction, rather than a reductive account.
Anyway, I might be going off on a slight tangent. I’ve just been in these kinds of debate before, and once we start talking about the brain at the atom level the discussion goes down a very unproductive and unrealistic direction.
My contention is that p-zombies are irrelevant, even if they could exist. They have no physical difference from ordinary persons, and they behave exactly as if they were ordinary persons, so what possible way of testing would prove that they exist? The difference between a conscious entity and a p-zombie is a metaphysical difference, and cannot in principle be tested in this universe.
Even my ‘high-bandwidth linkage’ thought experiment would not prove anything much, since the two linked persons would create a new entity. If you already believed you experienced consciousness before linkage, you would presumably believe that the new entity was conscious as well; this, however, does not imply that the entity you were linked with was conscious before, or even if it was conscious after the link is broken.
I’m not discounting billiard balls can replicate any computer. Rather, I’m saying the complexity and scale for replicating a human would be so large and so complex as the energy to make the system work would be nontrivial. At the size of even marble-sized billiard balls, the number of them could approach Avogadro’s number and the size could dwarf the solar system.
In other words, while logically possible, practically it is unachievable.
In theory, yes. Building the thing would be a nightmare.
I don’t think anyone is proposing only using philosophy. The technical challenges you mention are there, and the technology to explore the topic doesn’t exist yet. Philosophy is there to point is in the direction of study and frame the kind of experiments we would need to devise.
While we are a long way from physical experiments, we are working on a version of simulating intelligence. We might just accidentally stumble onto a new consciousness through neutral networks and machine learning, and we might not even know it because we don’t know how to identify it except in the framework of assuming other humans have the same experiences we do.
To be clear, I doubt that is likely any time soon, either, but maybe it’s a reverse race to see which path takes the longest to achieve results.
Yes, but I’m also saying that consciousness may be a part of those physical processes. Without knowing how it manifests and how it’s even possible, I don’t think we can blindly dismiss it as unnecessary.
You talk about billiard balls bouncing around as a parallel or replacement of the physics and chemistry of the biological processes. I think it might just be a failure to adequately assess the complexity of the chemistry and physics.
It’s true that nothing we have discovered so far about physics and chemistry involve or seem to create a subjective experience. My proposed concept is that the level of our understanding of those topics is baby level compared to the activity occurring in biological processes.
It is my surmise that consciousness is a fundamental result of the interactions themselves, but that they are so far above where we are we can’t even see they are there.
I know, that’s conjecture, not data. I think it is at least as viable an option as the notion of p-zombies.
In English? I can’t imagine that a stone would be using any human language; and I can’t describe my imagination of a thinking stone in English, either. It’s not coming from the portion of my brain that uses words; it’s coming from other parts of me. The part that uses words mostly uses them to say that it has to get out of the way of that imagination.
I don’t imagine it would even be using the same timespans. A stone would be thinking in geologic time, most likely. Humans can think about geologic time, but we can’t think in it.
Well, the reasoning behind that is that people tend to get lost in complexity. For some reason, the argument for two billiard balls is easily followed, but then once one gets to complex assemblages, like cells, or perhaps electromagnetic interactions (fields tend to be especially mystifying, I find), people just transition into a state of ‘who knows what might happen, might as well be consciousness’, but the argument is of course exactly the same. The physical properties going in fully determine what’s going to happen, so we can always do without consciousness. Ultimately, nothing but billiard balls colliding—or elementary particles interacting or quantum fields oscillating or strings wobbling—is happening, and it doesn’t make a difference which of those it is: nothing in there necessitates consciousness.
If they could exist, then it would mean that the world wasn’t fully physical, provided conscious experience also exists. Sure, one could pragmatically say that it wasn’t relevant for day-to-day interactions and whatnot, but if you’re interested in how the world works, what sort of creatures we are, and so on, then it matters quite a great deal.
Sure, but logical possibility doesn’t require material realizability. It’s perfectly possible to say, ‘if I hadn’t had so much pizza yesterday, I wouldn’t feel so queasy today’, even without being able to turn back the clock and realize that scenario. So it’s just as valid to say, ‘if I build a brain out of billiard balls, nothing would necessitate the existence of any subjective experience despite its behavioral equivalence to a human being’.
You say yes, as if to agree with my comment, but the following bit contradicts it—if the physical parameters fully determine the outcome of any interaction, then consciousness can play no part in doing so, so yes, from that point of view, we can dismiss it as unnecessary.
It might be that you mean that out of certain physical processes, consciousness necessarily arises—but that’s where the billiard balls come in: a simple collision of billiard balls certainly necessitates no conscious experience. But any complex arrangement of billiard balls is just a collection of such simple collisions: whenever two billiard balls connect, only their respective energies and momenta matter for the outcome of the collision; there’s nothing that keeps track of whether they are colliding in the vacuum of space, or as part of a billiard brain. So if there’s no reason to see consciousness emerging from a single collision, there’s no reason to see it emerging for a collection of such collisions, since they remain single collisions all the way. Nothing’s changed about them for being part of a larger pattern.
And the same remains true if we change the physical substrate. Two single electrons interacting is mathematically marginally more complex than two billiard balls doing so. Even the most highly complicated interactions can be analyzed in a series of Feynman diagrams, and for quantum electrodynamics, which suffices to encompass all of chemistry, we essentially just have four different basic processes that occur over and over—an electron and a positron meet, and annihilate into a photon; pair production, a photon produces an electron and a positron; an electron (or positron) emits a photon; and an electron (or positron) absorbs a photon (and even these are just the same process oriented differently in spacetime, effectively). Using these, it’s possible to calculate any given interaction to an arbitrary degree of accuracy, but there is absolutely no hint or need for consciousness to occur at any given step. They would occur just the same way if there was no consciousness present, which is all the argument needs.
This would violate a core tenet of physics, namely, universality: even in full ignorance of the underlying microdynamics, we can formulate a complete description (upon fixing a few empirical parameters), a so-called effective field theory. The physics of the everyday world up to energy levels far beyond those relevant to living creatures is completely known. Effectively, the physics of lower levels of energy decouples from that at higher levels, such that the description is insensitive to what the actual microscopic levels of detail may be. (Here’s Sean Carroll laying out the argument in detail.)
So, you can’t tell me, can’t describe it, consider that stone’s thoughts would be fundamentally different from human thoughts, but still expect me to believe you can accomplish what amounts to a conceptual impossibility?
That’s the point: there’s nothing it is like to be a p-zombie.
And we have no idea whether there’s something in there that necessitates consciousness. But the existence of consciousness rather argues that there is.
Yup.
I can imagine that a stone is conscious, in a fashion that I can’t understand in a form that I can put into language. I can imagine all sorts of things that I can’t consciously understand. Hell, I can’t consciously understand or put into words what the scent of a cat or the tactile sense of putting my palm to a tree does in my own mind.
There is a whole lot of me that doesn’t use words. There’s a whole lot of you that doesn’t use words, also; though the part of you that does use words may not be recognizing it as part of your “I”.
Right. As a thought experiment, I follow the logic. My comment was not that the logic fails. My comment was that they’re does appear to be consciousness. I am a physicalist/materialist. I don’t believe in “souls” or other mystical entities. So consciousness has to be a manifestation of the physical interactions.
So finding that is what we’re looking for. The implication of the billiard ball thought experiment is that consciousness can’t exist, or else there is something besides the physical. I’m saying the third option is that consciousness is in the parts moving around.
The point of my comment about practical implementation was to suggest exploring the means consciousness arises from the physics. We will never be able to do that with billiard balls. We might with silicon chips.
And yet I presume you believe you are conscious. Are you proposing there is something besides the physics? Or are you illuminating the gap in understanding? Because I accept there is a gap in understanding.
And yet we cannot accurately predict weather beyond a few days. If weather, a fully physical process that we have a decent level of underlying physics, eludes prediction where eclipses can be predicted to eternity1, then how can we say we understand everything about the physics and chemistry of how the brain works when we can’t even describe what experiences are?
Exactly.
I was being facetious with my reply. The “conscious stone” line was a mix of comments from different people.
The intent of my facetious remark, however, is that projecting consciousness is what a humans do.
We experience the world from the position of intentional causality. We project intention onto the world. We create gods that are the intentional cause of acts of nature. We conceptualize the world as intention. It even infects our philosophical musings. We ask, “What is the meaning of life?”, i.e. why are we here? Why do our lives exist? Meaning is a word of intention. It necessitates an agent of intent who caused life.
So proposing a human replicant that is fully autonomous and self- sufficient, mobile, responsive to the environment and adaptive and reactive to the changing circumstances as they occur, except does not have the conscious experience that we have, is conceptually a challenge, arguably a greater challenge than conceiving a rock could have consciousness.
We have no reason to believe a rock has experiences. But we can imagine it. Hell, the movie “Everything, Everywhere, All At Once” does just that. The protagonists bounce between possible realities, and in one they are rocks.
It’s easier to imagine rocks have identity but no way to demonstrate it than to imagine a fully identical human exists that does not have identity.
1 Or at least until the Sun changes internal states and expands to a red giant.
Sure, there may be any number of inaccessible and undetectable metaphysical phenomena that are associated with our consciousness; there are plenty of real religious and philosophical possibilities available to choose from, and I’ve dreamt up a few myself, often inspired by discussions on this board.
But since they are inaccessible to scientific proof (and to falsification) they are indistinguishable, and therefore irrelevant to our investigations into the physical cause of consciousness. We need to explore the physical causes of consciousness in as much depth as possible, but the metaphysical aspect will be forever beyond our grasp (in this plane of existence) and belongs in a different magisterium.
Sure; I’ve imagined conscious rocks on many occasions. But the problem is that a conscious rock has no power to act, no effective volition, so it can’t really make choices and actually do anything. Maybe rocks have a bit more freedom to act in high-energy environments, where they are semi-fluid and capable of crystallising into multiple forms; but once they cool down and become strata, they are a prisoner of their own solidity.
Again, that’s what the argument is for: all of those physical processes are perfectly consistent with no conscious experience existing (read: those processes occurring without any conscious experience, just as all other physical processes do, entails no logical contradiction), hence, nothing in there necessitates consciousness. “Something unknown might be doing we don’t know what” isn’t really an argument.
These are inconsistent statements. If you think the logic of the thought experiment is sound, then consciousness is not a manifestation of the physical interactions, because those fail to determine whether there is any conscious experience present, much less what form it ought to take.
But there is no difference to an electron absorbing a photon whether it does so in the vacuum of space, or within all the complicated machinery of the human body. If it can occur without conscious experience in the former case, then so too in the latter. It doesn’t ‘know’ that it is part of some ‘parts moving around’; the physics of quantum field theory is dynamically local, i.e. every interaction takes part at some infinitesimal volume and nothing outside that volume determines its outcome (this isn’t touched by either the Bell or the Reeh-Schlieder theorem).
If you’re then adamant that there is still some difference in whether those interactions take place in vacuum or within a body, then you’re saying that there is a qualitative difference irreducible to the properties of the component parts that somehow appears when they are put into the right configuration. That’s a notion known as strong emergence, and I’ve just been discussing it here, so I won’t repeat myself. Suffice it to say that to me, it’s basically magic, and will take you out of a physicalist position anyhow: there need to be ‘further facts’ irreducible to the physical facts that fix the emergence of facts relating to conscious experience.
I’ve given my response to the argument several times in this thread, and posted a link to the paper where I go into detail a couple of times now. Since posting the same link too often makes Discourse cry, here’s a link to a popular-level discussion of my basic model:
In brief, I attack the conceivability → possibility link of the argument: zombies are conceivable, but not metaphysically possible. This is a fine (and to many, costly) line to tread: essentially, I argue that all third-personal knowledge is embodied in models of one kind or another, with only structural relations of the object modeled present in the model. Thus, intrinsic properties are necessarily absent, but it is those that are ultimately responsible for conscious experience. Hence, conscious experience and its absence seem consistent with our knowledge, but only its presence is also possible.
This entails that the hard problem is conceptually unsolvable, because it simply isn’t possible to find a model that encompasses it. It also means that I’m a different sort of physicalist than most others: while I think that the world is solely exhausted by physical properties, I don’t believe it can be completely subject to physical science. Whether this seems like an attractive position is up to anyone, but to me, it seemed the only way to rescue a physicalist commitment.
But again, that’s a practical, not a fundamental problem. The problem with conscious experience is a fundamental one.
Again, I think it’s in fact conceptually impossible: your experience is locked in your being that entity that has that experience; to imagine another’s experience (rock or otherwise) is just to fool yourself into believing that you could imagine what it’s like to be that other, but in fact, you’re just experiencing what it’s like to be you having a certain set of thoughts—your thoughts, not the rock’s (or whom/whatever’s).
If they exist, then there is no complete physical explanation of consciousness, much less a physical cause; so denying their relevance would be to embark on a fool’s errant in trying to find such things.
