Well, yes, all true. But look at the things we can do that nature never provided for. Moon rockets, supersonic transports, nuclear reactors, superconducting magnets, calculating the billionth digit of pi, etc.
We can do, in decades, what nature took billions of years to do – and some things, nature never would have gotten around to, ever, no way, ever. Nature was not going to put mammals on the moon…but we did it, and only some 70 years after the first powered heavier-than-air flight.
We have the advantage of intentional design, something nature lacks.
Well, fair enough. I am definitely a strong AI proponent, and think it is possible. I’m also pessimistic (or just realistic) enough to know it won’t happen any time real soon.
Still, I’m old enough to remember skeptics declaring, most positively, that a computer could never play a game of chess. Not merely that a computer would never be a world champion (but that’s happened too) but that a computer couldn’t possibly even play the game at all, not even at a beginner’s level.
Total agreement.
Total disagreement… But, of course, you phrased it as a subjective belief, and those are always to be respected. I am not in any kind of position to say, “You’re wrong.” The best I have is, “I think something different from what you think.”
I’d even be happy if that view could be demonstrated conclusively. Knowing that something can’t be done (squaring the circle, trisecting the angle, solving an arbitrary fifth-degree polynomial, etc.) is, in itself, hugely educational.
(The mathematical proof regarding fifth degree polynomials is intensely fascinating! It’s one of the true triumphs of human knowledge. It’s negative, but it’s still an amazing insight into the nature of mathematics.)
Such as human bones made of the mineral calcium phosphate?
Grin! No worries, my last sentence was also a “zinger” and not any kind of formal point of debate. This is the SDMB, not the Harvard Debating Society. We get to chew the scenery a little!
Grin! I’ve got relatives like that!
I’m a big believer in the “exponential curve” model of technological advance. For instance, I think we’ll soon see quantum computing come in to its own, and that might lead to new advances in algorithmic approaches to problem solving that will stagger the imagination. It won’t lead to AI directly, but will lead to the solving of so many other problems that AI (and other incredible things) will fall out as side-stream benefits.
(The way the space program led to so many other cool technologies.)
Anyway, that’s my optimistic view.
(My pessimistic view comprises runaway nanotech, ecological collapse, new continental-scale wars, cyberterrorism, theocratic regimes at home and abroad, and my lumbago getting worse.)
Sort of like what the judge said about pornography: we know it when we see it.
I confess, I’m partial to the “behavioral” model of communication. I can’t know what you are thinking, only what I see you do. Your words are actually very clear – you write well – and your posts are convincing. But I can’t ever really know what you mean by what you say. I can only take the words as external objects and address myself to them.
I put a lot of reliance on the positive behavioral traits of social animals. A good part of our concept of “good” comes, I think, from social traits.
Cats, as lone predators, have fewer of these social traits – but they make up for this by being cute and friendly. But I also have to confess, I’m more of a dog person than a cat person…
No…but I am a “furry” and have been since C’Mell and Mam’selle Hepzibah!
Oh, here, I’m quite with you. This is why a self-contained AI would have to have a simulated environment for it to work with. A naked brain in a bottle isn’t enough; it has to have wires going down the neck feeding it sensory data for it to process. Some sort of external world is absolutely vital to consciousness.
You might want to reconsider, given that the only time I’ve ever been out sailing, I ended up overboard and having to push the boat out of the mud flats. The cut of my jib was about fifteen seconds too late!
Seriously, I think we may agree on more than we disagree on…and much of that is opinion, belief, and maybe even faith. I’ve got nothing against faith, in discussions where the science isn’t available. What will alien life-forms look like? Totally alien…or vaguely along “animal” lines, with limbs, sensory clusters, some kind of endoskeleton, etc.? Until we actually meet them, any degree of speculation is completely valid!
Formally, it’s exactly the same sense: to perform computation on a computer, we set it up in some initial state, and from there, it simply follows its usual evolution according to the laws of physics, a sequence of states, every one following causally from the previous one. The initial state will typically encode, in some sense, the program and input—i.e. you can view it as just the bits set in the memory. We have some mapping that we apply in order to translate the evolution into something human intelligible.
Now, with the rock, we don’t set up the initial state, that’s true; but we define the initial state of the rock’s evolution to code for the initial state of the computation. But that’s not an essential difference to the case where we use a special purpose computer—there, we simply set up a state that we have defined to code for the initial state of the computation, but since the coding is arbitrary, this is no salient difference (alternatively, we could also define some special rock state to be the initial, and simply wait until it comes about naturally). From there on, the rock follows its natural physical evolution, and we have a mapping translating that into something human-intelligible. (Consider that if we didn’t set up the computer in a special initial state, we could simply have it cycle through states, and if it ever hits on the ‘initial’ state of some computation we wish to perform—which it will, since it’s a finite system—then it will just as well perform that computation as if we had set it up in that state.)
In both cases, we have physical systems evolving according to the phyisical laws that apply to them, and a mapping taking their states to computational ones.
I wasn’t aware that metaphysical questions could be decided based on ethical principles (and besides, if you oppose my noncomputationalism based on justice, it would be unjust if you didn’t oppose Tibby or Not Tibby’s as well :p).
The problem is that you’re setting up a false dichotomy: basically, you seem to hold that either consciousness is computational, or there is no naturalistic explanation. I think that’s wrong: there are plenty of things that are not computational, but which nevertheless can be perfectly well explained naturalistically. I’ve given several examples, such as the mass of a black hole, the heat of a volcano, and so on. So this shows that there’s a third logical possibility, which is ‘having it both ways’ as you called it: there are perfectly natural things that are not captured by any computer simulation. Thus, even if consciousness is noncomputational, doesn’t rule out the possibility that it is explicable without reference to souls and other woo.
Because there are certain properties that are in principle not simulatable, i.e. properties such that if X has that property, the simulation of X doesn’t—because X and its simulation (emulation, whatever) are not the same thing.
I’m not saying that computers can’t attain consciousness; I’m saying that consciousness can’t be attained solely by performing a certain computation. As Tibby or Not Tibby proposes, one might need certain physico-chemical processes/properties in order to generate consciousness, just as you need certain physical properties (mass, energy) in order to generate a gravitational field or heat.
No computer will ever generate the gravitational field of a black hole solely through computation, but obviously, atoms can do so—so there’s nothing that forces the implication ‘if atoms can do it, it can be done by computation’, and in fact, that’s typically a wrong statement.
Actually, you need not appeal to any abstract properties to see this: consider simple motion. Certainly, atoms can move, and can come together to produce moving complexes, such as animals. But movement is not attainable through a computer executing some program: it can simulate every kind of moving system imaginable, and will nevertheless always stay put. But we can build moving computers: just give them a robot body. In the same sense, even though we can’t get a computer to produce consciousness solely by having him compute the right program, it’s nevertheless in principle possible to build a conscious computer, or robot, perhaps. We know that to be the case because we are exactly such conscious robots. But saying that consciousness follows from computation, because evidently, we perform computation, is too strong.
1 - We set the initial state where in the rock we don’t (you agreed to that)
2 - The evolution of the rock’s physical states is not impacted by us in any way whereas the computer’s physical states are directly impacted by us
3 - In the case of the rock, we actually PRE-COMPUTED the entire sequence of states that we were interested in and then created a mapping after the entire computation had been performed, prior to any involvement by the rock - this is a key point
With a computer, we do not pre-compute the entire sequence of states, rather we setup of the rules of state evolution and it follows them.
The rules we created in our computer for state transition are critical and they are the secret sauce.
The rules we created in the case of the rock when we pre-computed the sequence of states (prior to any involvement by the rock) is this exact same secret sauce.
I fully understand that the state inside a computer has an arbitrary interpretation, I just disagree with the statement that the computation is the “same” because there are clear physical and logical differences.
Yes. but that’s not relevant to performing some computation: no matter how the system enters that state, if it does, then it performs the associated computation.
That’s also not necessarily the case for ordinary computation; indeed, any computation can be performed inputlessly.
If we were to implement the computation in this way in practice, then perhaps; but all the argument needs is for the mapping to exist.
Perhaps as an analogy, consider an orchestra versus an organ-grinder: the former has many specially engineered parts that come together in a complex, well-designed way, while the latter just needs some dude turning a crank. But if all we care about is the melody—the computation—they’re the same in every relevant way.
I can only attest to what I think consciousness is. It is generally accepted that there are 3 states of consciousness: deep sleep, dreaming, and waking. These states are considered relative consciousness. Deep sleep is being asleep, then the brain goes into a dream state where the dream is, at the time, perceived as real. After that, the brain awakes and realizes that it was a dream, becomes awake and enters what we perceive as the world and then we can think and do things.
Going further, we can meditate, and achieve what some call transcendental consciousness, which is basically just glimpsing at the source of being. Some people would call this the spirit, god, being, etc. In this state we “wake up” from the dream called life and all the stress that goes with it and experience the stillness of being (no thought). Achieving this creates a feeling of bliss, because the mind is going to the natural state of consciousness. Once we leave this state, we go back to the wakeful state and after time the bliss fades away. It’s an in and out state.
Eventually, after entering the transcendental state multiple times, our nervous system becomes accustomed to the feeling of being in this state. Then, unconsciously, we are in a transcendental state and we connect with the feeling of bliss at all times even during relative consciousness. This is cosmic consciousness. Cosmic just meaning all encompassing.
When we can finally pull back the layers of perception and see things for what they really are, then we are achieving ever higher states of consciousness. For example, if we look at a flower and we see a flower, that is the mind seeing the perception of that flower not the actual flower. Now, if we look at a flower and we see everything else, mountains, trees, rain, etc. Then we’ve achieved god consciousness. We can see that everything is connected within the universe. Ultimately, when we can look at that same flower and see ourselves or our spark, then we’ve made a complete circle and have become fully self-aware. This is unity consciousness. Some call this enlightenment.
Consciousness is just a state of being with different perceptions.
It’s a meta-level rule, the idea that debates have rules we need to follow. “Having it both ways” violates a procedural rule of debate. It also is tricky, logically, because it could lead to one endorsing a contradiction, whereas, classically, once one is involved in a contradiction, he’s considered to have lost.
I don’t believe there is any evidence of a third option.
Computational systems are perfectly happy interacting with the physical world. I have a USB turntable that accepts sample input from old vinyl records. Every computer has a BIOS that interacts with the physical world: keyboards and monitors.
I believe that the information content of consciousness is consciousness. I can’t prove this, but I do think that any attempt to disprove it leads to violations of naturalism, and to an inevitable appeal to some variety of vitalism.
Yes, a computer simulation of heat transfer doesn’t produce actual heat (well, you know what I mean; CPUs generate a blortload of actual heat, but that isn’t what either of us means.) But a computer simulation of a block of text…is a block of text. A computer simulation of a grammar or a syntax…is a grammar and a syntax.
To claim that a computer can never really comprehend the meaning of the data it processes does not seem illogical. How could a chess playing machine ever really understand the notion of a game, two players, the representation of war, the abstraction of peering into another’s mind, etc.?
But I always come right back to the immediate problem: how do the atoms of our brains do those things? How is it a few pounds of stiff bloody lard can comprehend anything?
I can’t buy it as a meaningful possibility. I see it as intrinsically woo-ish. It is, to me, exactly the same as those who claimed a computer can’t play chess at all. Why not?
If it’s possible, then I’m right and you’re wrong.
If it isn’t possible, then whatever is missing, whatever “spark of life” or intrinsic property of chemical matter that is absent from electronic matter cannot be anything other than a “spiritual” – or at best, quintessent – element. It would evoke a “Platonic” quality of chemistry that electronics lacks.
To convince me, you would have to demonstrate a perfectly ordinary task – factoring large numbers – that chemical matter can accomplish but which electronic matter cannot.
And the killing issue is that I don’t accept “feel emotions” as such a task, because I am perfectly convinced that an electronic system would, indeed, feel emotions if it were built (or evolved) with that as the design goal.
Is all we have here is a difference of faith? I think that “understanding” is an informational quality, which can be meaningfully emulated just as grammar can. You think it is (or might be) a physical quality, and requires a specific physical gizmo to be made real.
Seems to me we’re done here. Let’s slap a Good Housekeeping Seal on it and put it in the basement for storage. We’ll pay up our bets at a later date, when such a system is designed. (If it permits us, being our new computer overlord and everything…)
Minor nitpick: the dream is not actually perceived as real. There is always a distance, a reservation, a degree of doubt regarding dream adventures. It’s a little like reading a book or watching a movie: you can become very absorbed in it, but you never quite “perceive it as real.”
(This was always my experience, anyway, even before I discovered the joys of semi-lucid dreaming.)
When moving between these states, that is, from lower to higher consciousness, we seem to be blissfully unaware at the lower state that a higher state even exists. We know what we know, but not even that sometimes, because in dreamland we forget all of those rules of physics and what not. It’s more like being in a hypnotic state as compared to what we call reality. The hypnotized person doesn’t know they are hypnotized, and even if they do, they can’t do anything about it. I wonder how many higher states there actually are beyond the one I’m in now, beyond even cosmic consciousness, that I am blissfully unaware of right now.
I understand the point about a computer and a rock are just 2 sequences of states and any mapping can be applied to either one.
In that sense they are the same, but that is a very very abstract view.
Calling them the same in ordinary conversation ignores the setting of initial state and it ignores the way in which we create the program (due to the difference in state evolution between the two).
In the case of the computer, we created the state evolution rules to more easily allow us to create instances of state evolution that are interesting to us.
To achieve the same thing with the rock requires a different type of effort. We need to pre-compute things in a different way and in more detail than we do with the computer to end up with a valid sequence of states.
Yes, I believe I understand and agree about that abstract argument, or at least don’t have a counter. Essentially the states and rules and transitions would appear to be too limited of an explanation for consciousness due to the fact that the interpretation of state is arbitrary and sequences of states in any changing entity can be mapped to an enormous (maybe infinite) set of other states and interpretations.
As stated multiple times, I don’t disagree with the argument about mappings.
What I disagree with is just your statement that the rock and a computer perform the computation in the “same” manner. There are differences:
1 - We set the initial state (this can’t be discarded, it makes our life easier and/or saves time, it is a very real physical action)
2 - The way the computer transitions from state to state follows a set of rules we created for the express purpose of saving time and energy when trying to determine what the initial state should be. The computation follows these rules of state transition.
It’s like saying that a person running on a treadmill and a person running around a track are doing the same thing because they are both running, but the difference between performing that action on a treadmill vs around a track makes a significant physical difference to various measures of state within the universe.
There’s only a logical problem if there’s a genuine dichotomy, which there isn’t—I have shown explicit examples that neither fall in the category of being computable, nor are woo.
Well, but I showed examples of that third option! Your stance is, either something is computational, or it has no naturalistic explanation. But gravitational fields, magnetism, heat, and many, many more properties are not brought about by computation, and nevertheless, have a perfectly naturalistic explanation. So there are at least some aspects of the world that don’t fall into your dichotomy, which therefore isn’t universal. I’m just pointing out that consciousness may be one such aspect.
Yes, and where they do, they don’t do so solely through computation, but by accessing a physical interface. A computer can produce a magnetic field by switching on the current running through a solenoid, but not by simply carrying out some computation. Likewise, a computer may produce consciousness by, say, triggering the release of certain chemicals interacting a particular way, but not simply by carrying out computation.
What’s your reason for believing this? Why don’t you believe it, for instance, in the case of mass, or heat?
Some things are identical to their simulation; others aren’t. You seem to believe that consciousness is of the former kind, but so far, I can’t see any reason for accepting this.
You could make the exact same pronouncements with regard to, say, mass: claiming that you can’t see it as a meaningful possibility that mass isn’t generated by computation, that there would be some ‘spark of gravity’ beyond physics necessary if it can’t be generated by computation, etc. But that’s not the case: mass is indeed not produced by a simulation of something massive, and there is nothing needed but physics to account for it. Hence, the logic you’re using is faulty.
I’ve already provided you with several—producing gravitational fields, producing heat, producing magnetic fields, and so on. Factoring large numbers, of course, is perfectly possible for a computer: it’s a computational task. But that doesn’t exhaust the possibilities, and the fact that some systems or properties do carry over to a simulation does not mean—and indeed, does not even argue—that all do. There are examples to the contrary, and it’s completely conceivable, logically speaking, that consciousness is one, without any attendant wooishness.
But what are your reasons to be so perfectly convinced?
I think there’s good reason to doubt statements such as the above, and have argued at length accordingly. Basically, computers manipulate information solely syntactically. Consider these two pseudo-programs:
IF water_level < low
THEN valve = open
IF water_level > high
THEN valve = close
IF temp < cold
THEN heater = on
IF temp > hot
THEN heater = off
They control two different systems: one keeps the water level in a tank in between some predefined values, and the other controls room temperature. Nevertheless, of course, to any computer, they’re the same program: their structure is identical. The kinds of facts a computer is sensitive to are structural, syntactic, relational—ultimately, they can always be expressed as a series of differences, which is why you can code them as patterns of 1 and 0, high and low voltage, on and off. But it isn’t sensitive to what those differences are between: it could be voltage levels, could be water levels, could be temperature levels—the structural, algorithmic level underdetermines the physical level. The intrinsic facts, the underlying physical reality, is completely inaccessible to the computer—it deals only in relations, only in structure.
This is very different to human minds. There, we have first and foremost the intrinsic level, which is how the world is given to us—not in terms of relations, but as concrete things embodying those relations. We do not experience the world in terms of the abstract structure of the two programs above, it is given to us in terms of water- or temperature-levels, which nobody would—indeed could—confuse. In fact, we have to work hard to bring out the abstract structure of a situation in order to, say, write a program that models it. The way we experience the world is in terms of intrinsic facts—things just of the sort that are not captured by simulations: heat, gravity, concrete things that have certain properties; the way computers model the world is in terms of extrinsic facts, relations between properties, abstract structure.
We see, in the first example, the temperature level, and if it gets too hot, or too cold, we adjust the thermostat; we see, in the second example, the water level, and turn off the valve if it gets too full, and turn it back on if it gets too empty. From there, we can abstract the description that yields the two programs above. But the computer has nothing but that abstract description: it only sees ‘IF {value stored at this particular address} < {value stored at that address} THEN {store particular value at third address}’. The program is neither about water nor about temperature—no program is ever really ‘about’ anything. But our perceptions always are intrinsically about something.
Take a computer, and remove all external interfaces. Run some program on it. Then give it to a third party, and ask them to determine what the program does. In general, this question has no answer: there is nothing that connects the sequence of states that the computer cycles through to any particular computation (although if the computer follows common design principles, you can probably get a reasonable idea of what it computes—but we may suppose, for the moment, that this is an entirely alien system, perhaps running some cellular automaton or a ballistic computing device, take your pick). Nothing makes the sequence of states the machine traverses be in any way ‘about’ any particular set of entities. It could be a computation of the digits of pi, a simulation of a black hole, or indeed a perfect simulation of the Earth with all its inhabitants right now. But it is not either of those to the exclusion of the others: you can find a mapping such that each of those computations is brought out, but it is neither one more than the others.
The world we experience, however, is one of concrete things, it’s viewed from a given perspective, it has particular properties, and most of all, is only exactly this world to the exclusion of all others. Saying of the black box computer that it contains a simulation of me having some particular experience is not any more justified than saying that it does not contain such a simulation, or that it only computes pi to the googolth digit (this, I think, is the correct response to Bostrom’s Simulation Argument, by the way). Reality, however, manifestly does contain me having a certain experience. A computation is not intrinsically anything, whereas all the world really is, and all our experience really is, is its intrinsic properties (from which structure, which is what computations are all about, is only abstracted).
Ah, I see, I misunderstand you as saying that those differences made a difference regarding whether some computation is accompanied by conscious experience. OK, so in place of saying that the rock and the computer perform computation in the same way, I’ll just say that they perform computations in a way that’s equivalent for the purpose of the argument (although in fairness, I did qualify my saying they’re the same by the word ‘formally’ and pointed out the relevance to performing the computation).
Personally, I don’t care much for the discussion about artificial intelligence/conciousness.
As per this thread we aren’t even sure what the mechanics are in living creatures, such as ourselves.
If we can’t first agree on what it actually is, how then can we say we can build it artificially?
No. My stance is that consciousness is computational, or specifically, that it is informational.
It has become very obvious that we have a disagreement based on opinion. I believe that consciousness is an informational process, and thus amenable to informational simulation. You (seem to) believe it is a physical process, and that it can only be performed by certain chemicals, yet not by others.
And I hold that consciousness is a computational, or informational, task. No evidence has been presented that this is necessarily the case…or that it is not necessarily the case.
(Which is to say, lest you once more accuse me of rejecting the possibility that I am wrong, I am willing to acknowledge, yes, I might very well be wrong.)
As a (possibly?) interesting side-issue: are prime and composite numbers a “physical” property, or an informational one? I hold that prime numbers exist, even if numbers of “things” – peaches, pears, horse-apples, etc. – aren’t around to be counted. I say that primeness is an informational property, and thus, that a computer simulation of a prime number actually is a prime number, even if a computer tabulation of the number of apples in the warehouse is not actually “apples,” but just labeled “apples.”
My opinion is that consciousness is more like prime numbers in this than like apples.
This is why I want us to try to build it artificially…so then we’ll know what it is. Sort of the way the Wright Brothers settled the question of whether powered flight is possible…by bloody well doing it!
As I see it, we’re in the position of people in 1901 arguing, “Powered flight? Nah, can’t be done!” “Oh, yeah? Why not? Put a powerful enough motor in a kite, and why shouldn’t it fly?”
Certainly we’re arguing in advance of the full facts, but that’s part of the fun.
In the meantime, I’m also in favor of other means of learning what consciousness is, and I believe that current brain physiology observations are teaching us a lot about the subject.
But the rock argument is interesting in this regard.
I had always assumed computation was enough, and the Chinese room never bothered me because it was designed as an extreme that avoided the types of calculations we perform (compression, pattern matching, modeling, prediction, etc.) that I always thought resulted in “understanding”.
But pondering the rock thing definitely has me scratching my head. I can see that the computer’s transitions between states really has multiple (maybe infinite) interpretations and there doesn’t seem to be any link between those states and what we experience.
The rock thing strikes me as contradictory. “It’s an exact duplicate…oh, but, wait, it leaves something out.” It only works by assuming that consciousness must be “material” – like heat or gravity – but since that’s what’s being debated, there is an element of circularity.
It also runs up against practical limits of computation.
Here is a reduction ad absurdum.
There is a rock with one set of states. Side A is up, or side B is up. You can count to 1 in binary. 0, 1 corresponds to A-up, B-up.
There is a rock with three sets of states: Side A/B up, end X/Y points north; rock is wet/dry. You can count from 0 to seven in binary.
There is a rock with so many states that it can count to 127 in binary. I leave the states as an exercise for the student.
There is a rock with so many states it can emulate an abacus, summing a short column of three-digit numbers.
There is a rock with so many states, it can emulate an old 1980’s Atari play station, running Frogger.
There is a rock with so many states, it can emulate a Cray supercomputer, doing hurricane simulations. And, say, this is interesting: the Coriolis force wasn’t explicitly coded into the sim, but fell out as an emergent property.
There is a rock with so many states, it can emulate a futuristic supercomputer that doesn’t just enjoy terabytes, but godzillabytes (10^712) and cthulhubytes (10^4089) of working data. It can model the entire solar system at the level of individual atoms…including all of the above rocks. (Oh, yeah, and including human brains and human consciousness, too.)
And…so on. There is a rock that can emulate the previous rock. Invoke recursion infinitely, or at least indefinitely.
At what point in this progression does an actual physical rock cease to exist that can fulfill the requirements of the hypothesis?
You know something? I think it’s between the third and fourth somewhere. Beyond that, real rocks don’t have enough physical states.
But then, on what grounds do you justify your claim that if consciousness isn’t computational, it’s woo?
Well, that’s in the end what most debates boil down to, no? If any of us had access to all the facts, there would be no debate. But the idea—a meta-level rule, if you will—of a debate is to present arguments in support of one’s opinion, to try and see which opinion is more justified, or where logical problems lurk. You seem to be somewhat reluctant to do so—you’re quick to brush off my arguments, or not engage them at all, but I don’t think I’ve seen you offer any justification of the idea that consciousness is informational, which to be honest is a little frustrating.
That last part is not true—the very existence of other logical possibilities shows that it is not necessarily the case that consciousness is computational. Consciousness would only be necessarily computational if that were the only logically consistent alternative, but it’s not, hence, consciousness is not necessarily computational.
An informational one: it’s a property of the structure of sets of certain sizes. You can easily exhibit an algorithm testing for primality.
But what’s the reasoning behind that opinion?
You’ve got it the wrong way round: the rock argument assumes that consciousness is computational, and from there, establishes that every rock, every physical system, must be conscious in every way possible. But since this is an absurd conclusion, it throws doubt on the assumption of computationalism.
Every macroscopic rock will have enough states in order to simulate a human brain. A typical rock will contain on the order of 10[sup]23[/sup] atoms, while a human brain has around 10[sup]11[/sup] neurons, and 10[sup]14[/sup] synapses, so if a state of the rock is an arrangement of those atoms (even if it’s just a fraction of those atoms), and a state of the brain is a firing pattern of neurons, it’s clear that you have enough to work with.
Even if you think things are more complicated (they may well be), at the utmost, you’d have to invoke Putnam’s open systems proviso: if you introduce arbitrary outside influences, then the rock will effectively have infinitely many states, as long as you’re prepares to wait long enough.
)