Hawkins doesn’t know because even the scientists that study this stuff don’t know.
It seems to me that his approach (and he has studied and experimented this from the biological side too) is like a typical scientist that is doing a hands on approach, he has no need for a magic hypothesis. In fact it is precisely the approach needed going forward.
“Throughout history, every mystery ever solved turned out to be Not Magic” ~Tim Minchin
Nobody is claiming “magic”.
What people are correctly stating is that we simply do not know what consciousness is or how it arises.
What Hawkins is doing is creating a method for spotting sequential patterns (nothing more, nothing related to consciousness as he himself admits), and even there the scientists that are using deep recurrent networks are outperforming all others (including HTM, which is a form of a deep network also).
Whether it’s HTM or DBN’s or simple ANN’s, one thing we know for sure is that the secret sauce (i.e. the learning algorithm) is far from resolved. And even when the scientists truly understand the details of how best to train these networks, it’s all still just pattern recognition and nowhere near consciousness.
One final very interesting thing I just read recently:
A shallow feed forward neural network can be trained to mimic a deep belief network accurately (which has many layers, many more parameters/weights) using much fewer parameters/weights.
But, using current training algorithms, the shallow feed forward network can not be trained to mimic the dbn using just the same training material.
Conclusion:
1 - DBN is not optimal as it can be mimiced by much smaller/simpler network
2 - There exists a training algorithm for the simple network that can use just the training material (instead of the result of the dbn), which means we do not understand how to get to that training algorithm today
This is missing the point, my impression was that you are making the idea of the sauce the zombies are missing and the sauce not existing as being in equal terms, in reality when one approaches this scientifically one has to dismiss the magic side.
So far there is no need indeed to look for the “special sauce” and Hawkins also realized that neural networks are useful but not really the proper way to approach this.
Of course, you are correct on the item that we still do not know all, but my impression is that now that we have theories of the brain that are more grounded in reality that we are getting on the right track.
Correcting this and adding a bit:
Of course, you are correct on the item that we still do not know all, but my impression is that now that we have theories of the brain that are more grounded in reality, I think that we are getting on the right track.
What you showed is indeed a snapshot of the current situation, there is still a lot of work to do with neural networks, but the approach of Hawkins is adding another element: Hierarchical temporal memory (HTM)
I do think however that it is good to see an old fashion race to see if Hawkins or the just neural network approach will succeed. It is like the old fashion rivalry of the DC vs AC currents. IMHO like with the currents, both and other elements will turn out to be part of the big picture.
Well, but it’s really just a way of formulating how we usually arrive at judgements of possibility—when somebody asks you, ‘is x possible?’, then you will try to imagine x occurring, and if you succeed, conclude that it is indeed possible. Of course, we may falsely imagine something to be possible: but to show that one has made such an error, it is necessary to point out where the error lies, i.e. what prevents x from being possible. Somebody pointing exactly this out is what I hope will ultimately be the outcome of discussing the zombie argument (and related ones).
What more information do we need? I can certainly imagine a series of causal interactions that produce a reaction from a stimulus, without there being any conscious experience, in the same sense that I can imagine how a pixel on a screen lights up as part of a picture being displayed, etc. That’s all that’s needed. Of course, I may be wrong in my imaginings, and thus, falsely believe that I can ‘coherently imagine’ this happening; but this needs to be shown by pointing out exactly where my imagining goes wrong.
I’ve already discussed this: such an exchange is only possible if philosopher #2 does either not know what water is, or what H[sub]2[/sub]O is; that’s where his error in imagining lies. But in the case of consciousness, we know both what the subjective experience is—in fact, it is exactly what we know of it—, and what the physical processes are, at least broadly. And there seems no common ground at all between them.
Regardless, I do believe that (5) is certainly the right spot to attack the zombie argument, but it needs to be done by pointing out why I, in fact, fail to coherently imagine the physical processes involved without conscious experience, while it seems to me that I’m wholly succeeding at that task. But I think this will need some new insight, which will likely tell us a great deal about the nature of conscious experience; that’s why I’m so opposed to hand-wavy ‘solutions’ along the line that something will happen that somehow gives rise to consciousness.
There’s a difference between being right and being convincing, or being explanatory. If I produce a proof that 1 + 1 = 3, then you could say, well, that’s wrong, it’s actually 2. And of course, you’d be right. But without pointing out what’s wrong with the proof, that will neither convince me, nor teach me anything new.
It’s the same situation with the zombie argument. Merely claiming that it’s wrong, that something goes amiss somewhere, that I’m in fact erroneously believing myself to be able to imagine the physical processes, might be completely right (I certainly believe so)—but it’s also wholly uninformative. Point out what I miss in my imagining, for instance, and you’ll have provided me with some new insight. Unfortunately, I think that’s quite a hard problem.
Well, to put it in Leibniz’ terms, who sort of proposed the grandfather version of the zombie argument, if you have some machine (e.g. a brain) that you believe produces consciousness, imagine it enlarged to the size of a mill, such that you can walk in and take a look. And of course, there’ll be nothing relating to consciousness in there; there only will be gears grinding on other gears, pulleys pulling, levers levering, perhaps relays being engaged, or maybe atoms or electrons careening about—everything readily imaginable without the slightest hint of conscious experience being present. In fact, it’s always in imagining how conscious experience arises from this that problems abound. I just believe we’re doing ourselves a disservice by shallowly consoling ourselves that it’ll certainly occur somehow.
Not necessarily. There’s an interesting proposal by Achim Kempf that spacetime may be described as both continuous and discrete. And really, my own hunch is that the question mis-frames the problem: in all likelihood, like everything else, spacetime is neither continuous nor discrete, but rather, quantum, which incorporates both continuous variation (of states/probability amplitudes) and discreta—say, the quantized states of electrons in an atom.
What we do know, however—or at any rate, very very strongly suspect—is that you can only put a finite amount of information into a finite volume of spacetime—this is due to the Bekenstein bound (or its relativistic generalization due to Raphael Busso). Thus, in principle, any finite region of spacetime can be described using discrete mathematics. And since in a universe with a finite age (even if that universe is infinite in space), there can only be finite causally connected regions (umm, I think? I just realized I don’t know that for sure, there might be runaway inflation phenomena that blow up to infinity in a finite time—but anyway, such things seem not to happen in our universe), we never need anything more.
Well, that’s quite an empty statement to make, since the problem is precisely how anything can ‘feel like’ something in the first place.
I’ve never run across it being used for such a purpose. Do you have an example?
And I already did explain more. It is not empty, he is just falling for a default position, since consciousness is bound to not be mysterious the magical sauce the zombie is missing can be dismissed, in reality that sauce is the extraordinary item that needs extraordinary evidence.
Thing is that I found before that one big reason the zombies are proposed is that it is an effort to counter a materialistic view that would include consciousness, IMHO religion and faith is still an important component to this proposition.
I agree there is no magic sauce (and you’re really overusing that word), but just saying that ‘there’s no magic sauce so zombies can’t exist’ is at best dogmatism, and tells us precisely nothing about the nature of consciousness—it’s just a handwave without any actual content.
That article really isn’t very well written. And not believing in materialism or reductionism does not entail believing in magic, or in souls, or anything like that. There are many more positions between hardcore physicalism and ‘a wizard did it’, many of which are quite fascinating.
Besides, zombie arguments are quite often used against the idea of the existence of an immaterial soul—google ‘soulless zombie’, for instance. Interestingly, those claiming that the argument is obviously flawed in the p-zombie case (without, of course, being able to actually point to the flaw) are, in my experience, quite happy to accept it in the case of a soulless zombie. It’s all in which preconceptions you want to have satisfied.
But with also very little evidence to go by.
I’m just saying that eventually one needs to ask “I eso con que se come?” The point here to me is that besides the need for the proponents to show evidence for those fascinating items there is also the bit that eventually a use for those positions should be found or they will be ignored. So far the progress seen is not depending on those positions.
I changed my post because I didn’t read your post very thoroughly
Yes I agree that I’m not providing any new information to support my position (I’ve admitted that), and yes if I could that would be valuable.
P-Zombie certainly does force one to try to counter/prove it with specifics and yes that is very difficult.
But this is not simply a situation of one group positing the existence of some extra thing (“magic sauce” or whatever you want to call it), and having the burden of proof rest entirely on them.
Reductionism has been extraordinarily successful in helping us to understand the universe. But, the phenomenon of subjective experience, certainly at first glance, doesn’t appear to fit well with this kind of description. Indeed, it’s difficult to see how to even ask the questions about consciousness from such a bottom-up approach.
So I think it’s entirely valid to try to form alternative descriptions. And there is no burden of proof until someone claims that a particular description is more useful than reductionism in this case.
Then, in terms of Materialism / Physicalism, I would consider these philosophical questions, and not something science needs to get involved in. For example with materialism, science doesn’t need to claim that all that exists is matter, it’s enough to say there’s no reason to suppose anything exists beyond what we know exists.
Trying to tie Materialism to science has led to the situation where already the definition of “matter” has needed to be expanded several times, and the current definition is deliberately broad / vague.
Well, IMHO if this was equal we should had seen a lot of research that assumes that extra thing and showing progress. That has not been the case.
Well, regarding consciousness Hawkins has been successful at least in showing that he is on the right track.
Thing is that there is no good reason to leave the fruit on the bowl.
Eventually someone like Galileo will wonder if Aristotle was right and see if the grape and the orange do fall at the same time. The point here is that it may be a nice philosophical idea to have, and that is nice, but it seems to me that the progress on this issue (as in practical applications) is telling me that we are close to continue to deal with matter and energy that is not mysterious at all.
I asked a university computer science professor: “If you could attach a radioactive-sampling random number peripheral to a computer, is the result still ‘computation?’”
He said yes…but was only willing to commit to this on a linguistic and semantic basis, not on a formal computer-science basis. He said that he’d seen many papers with titles that included the phrase “Non-Deterministic Computation.” He said that if this is not an oxymoron, then a radioactive-sampling random number input would not prohibit the process from being computation, as he understands it.
He also noted that there are still people who believe in the “hidden variables” model of radioactive decay, so that, if they’re right, it becomes deterministic again.
He went on to warn against the very high sensitivity such a device would have to environmental variables, and that any “home” version that you could just plug in to your PC’s USB port would be terribly susceptible to biases in the random numbers it generates.
Anyway, I thought it was all amusing enough to share with y’all.
On the right track to consciousness?
Hawkins has been successful at that?
I challenge you to show any evidence whatsoever that Hawkins has done that.
Note: I seriously doubt Hawkins would make the claim posted above, he’s smarter than that (I think).
Sounds like the type of thing a p-zombie would say.
I think you are getting it wrong, no he is not claiming that we have consciousness yet, only that there is no reason to assume that the artificial consciousness he can get will be a magical or mysterious thing.
Well, the notion of evidence itself becomes quite difficult in this case. The scientific method, in the way it is currently understood, is the only reliable epistemological method to discover objective facts with ever-increasing precision. But the problem is that we’re not dealing with objective facts, that indeed the very problem is the existence of subjective facts, which are not even intersubjectively accessible.
On the other hand, evidence itself presupposes the notion of the experiencing self: what data ultimately really is, is subject-object correlation—I see, for instance, a red light flashing on an experimental apparatus, and from there, I conclude that, say, the electron’s spin was ‘up’. This implication only follows if I take the subjective side of this, i.e. my experience of the flashing light, as an unanalyzed primitive; only then am I licensed to conclude something about the objective world. In this sense, the subjective self is a pre-condition of the applicability of the scientific method.
But it’s not at all clear how to turn this on itself—i.e. study the self from within the self. Indeed, you often hear claims that such ‘first person science’ is impossible in principle, being unable to fulfill criteria such as objective repeatability, etc. I’m less pessimistic: people making this claim forget that the scientific method itself is not a god-given immutable framework, but has changed time and again to be able to better grasp phenomena—today’s framework (some mongrel of Popperian falsificationism and bits of Kuhn, maybe even some flirtations with Feyerabend’s anarchism, with the odd positivistic holdout here and there) is very different from that of Newton and Galileo’s book of nature that is written in the language of mathematics. But I think your reference to Galileo is apt: we will indeed need a new great reformer in order to bring the phenomena of subjective experience under the purview of the scientific method, and this may be the greatest change it has undergone so far.
The often proposed strategy of just accepting as conscious that which behaves consciously—perhaps related to some behaviouristic/positivistic hangover—, that has also been advocated in this thread, I find worrying, however. The reason is that we don’t know how an entity behaves consciously—we only know what human-like behaviour is, and since we know humans are conscious, then we can detect consciousness in this sense, or so the argument presumably goes. But this is unacceptably anthropocentric.
Consider what happens should we ever encounter an alien species—and here, I mean an alien alien species, not just one of the Star Trek bumpy-forehead person kind. There is no need for it to behave human-like, or even human-intelligible, at all—nevertheless, it may of course be perfectly conscious. The same thing goes for machine intelligences—without humanity’s background as an evolved species in a natural environment, the possibility for behavioural variation may be even greater. Yet, whether we judge something conscious or not has great impact on our moral behaviour—thus, we can’t afford having our judgments clouded by anthopocentrism. So it seems to me that ‘treat as conscious what behaves consciously’ is actually a pretty terrible approach (even without adding the difficulty that it seems very possible that human-like behaviour exists without attendant conscious experience).
That’s circular. You’re judging what research should be done by what research is being done.
I’d like to echo RaftPeople here in wanting to see a bit of substantiation to such a huge claim.
Nope, it is just the old fashion “extraordinary claims need extraordinary evidence” the ordinary thing here is the plain vanilla realization that the other side is not really showing any results.
Again, you are getting it wrong, he is not claiming that we have consciousness yet, only that there is no reason to assume that the artificial consciousness he can get will be a magical or mysterious thing.
Every claim here is extraordinary, because nobody has any sort of theory of how consciousness is produced; and there are just as many results from traditional approaches—none. Nobody knows how to create a conscious computer.
You’re just following the ‘something will happen somehow’-strategy. But the question is, of course—what happens, and how?
You made a definite claim—that Hawkins has been successful at showing himself to be on the right track towards consciousness. I’m just asking, how did he do that?
The part about being on the right track? I get it from the fact that there is experimental and independent papers that showed that his ideas have merit when applied (so far) to visual pattern recognition. And that is one reason why IBM and DARPA are working with his products and ideas.
Now can you show me where is something similar from the ones proposing that there is something mysterious operating there? Not demanding a complete thing, only any progress to support the alternative with supporting evidence.