<div dir="ltr">Hi Tsvi,<div><br></div><div>What does it take for an algorithm to feel?</div><div><br></div><div>If you have the answer to that question, then perhaps you can create mean field, bayesian, and neural network algorithms with differing amounts of detail, all of which feel human.</div>
<div><br></div><div>Brian</div><div><br></div><div><a href="http://grey.colorado.edu/mingus">http://grey.colorado.edu/mingus</a></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Jan 29, 2014 at 5:41 PM, Tsvi Achler <span dir="ltr"><<a href="mailto:achler@gmail.com" target="_blank">achler@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I can't resist commenting anymore.<br>
<br>
It is interesting that the subject of quantum mechanics is being<br>
discussed here. From what I understand, Einstein never liked quantum<br>
mechanics because it replaces mechanics that are unknown with<br>
statistics. Subsequently electrons and atomic particle locations are<br>
represented by a probabilistic density functions. In this way, yet<br>
unknown mechanisms can be quantified and characterized statistically<br>
moving the physics forward to some degree.<br>
<br>
However neuroscience has an equivalent. It is Bayesian networks.<br>
Using a statistical framework, neural connections and activations are<br>
modeled with statistical distributions. Using the the conditional<br>
density function, priors, their distributions, and the Bayes equation<br>
the posterior probability (recognition) is calculated.<br>
<br>
Bayesian methods have been extremely successful in explaining<br>
cognitive and neural data. However they are not true neural networks<br>
because the connections are statistically defined. In some sense I<br>
think connectionists are being squeezed by Bayesian networks because<br>
they are now the go-to for cognitive modeling, not neural networks.<br>
<br>
I think the neural network community is in the unique position to<br>
truly explain neural function, and that's its biggest contribution.<br>
Here are my 2 cents: I think if we do not stay close to the biology we<br>
will be swallowed up by statistical methods.<br>
<br>
I would like to hear what others think.<br>
<br>
Sincerely,<br>
-Tsvi<br>
<div class="im HOEnZb"><br>
On Wed, Jan 29, 2014 at 2:26 AM, Włodzisław Duch <<a href="mailto:wduch@is.umk.pl">wduch@is.umk.pl</a>> wrote:<br>
</div><div class="HOEnZb"><div class="h5">> Dear all,<br>
><br>
><br>
><br>
> QM has yet to show some advantages over strong synchronization in classical<br>
> models that unifies the activity of the whole network. There is another<br>
> aspect to this discussion: we need to go beyond naïve interpretation of<br>
> assigning functions to activity of single structures. We have to use a<br>
> formalism similar to the quantum mechanical representation theory in Hilbert<br>
> space, decomposing brain activations into combinations of other activations.<br>
> In wrote a bit about it in sec. 2 of “Neurolinguistic Approach to Natural<br>
> Language Processing”, Neural Networks 21(10), 1500-1510, 2008.<br>
><br>
><br>
><br>
> QM seems to be attractive because we do not understand how to make a<br>
> transition between brain activations and subjective experience, described in<br>
> some psychological spaces, outside and inside (3rd and 1st person) points of<br>
> view. I have tried to explain it in a paper for APA, Mind-Brain Relations,<br>
> Geometric Perspective and Neurophenomenology, American Philosophical<br>
> Association Newsletter 12(1), 1-7, 2012<br>
><br>
> QM formalism of representation theory may be useful also for classical<br>
> distributed computing systems.<br>
><br>
><br>
><br>
> Best regards, Włodek Duch<br>
><br>
> ____________________<br>
><br>
> Google W. Duch<br>
><br>
><br>
><br>
><br>
><br>
><br>
><br>
> From: Connectionists [mailto:<a href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu">connectionists-bounces@mailman.srv.cs.cmu.edu</a>]<br>
> On Behalf Of Carson Chow<br>
> Sent: Tuesday, January 28, 2014 10:01 PM<br>
> To: <a href="mailto:connectionists@mailman.srv.cs.cmu.edu">connectionists@mailman.srv.cs.cmu.edu</a><br>
> Subject: Re: Connectionists: Physics and Psychology (and the C-word)<br>
><br>
><br>
><br>
> Brian,<br>
><br>
> Quantum mechanics can be completely simulated on a classical computer so if<br>
> quantum mechanics do matter for C then it must be a matter of computational<br>
> efficiency and nothing more. We also know that BQP (i.e. set of problems<br>
> solved efficiently on a quantum computer) is bigger than BPP (set of<br>
> problems solved effficiently on a classical computer) but not by much. I'm<br>
> not fully up to date on this but I think factoring and boson sampling or<br>
> about the only two examples that are in BQP and not in BPP. We also know<br>
> that BPP is much smaller than NP, so if C does require QM then for some<br>
> reason it sits in a small sliver of complexity space.<br>
><br>
> best,<br>
> Carson<br>
><br>
> PS I do like your self-consistent test for confirming consciousness. I once<br>
> proposed that we could just run Turing machines and see which ones asked why<br>
> they exist as a test of C. Kind of similar to your idea.<br>
><br>
> On 1/28/14 3:09 PM, Brian J Mingus wrote:<br>
><br>
> Hi Richard, thanks for the feedback.<br>
><br>
><br>
><br>
>> Yes, in general, having an outcome measure that correlates with C ... that<br>
>> is good, but only with a clear and unambigous meaning for C itself (which I<br>
>> don't think anyone has, so therefore it is, after all, of no value to look<br>
>> for outcome measures that correlate)<br>
><br>
><br>
><br>
> Actually, the outcome measure I described is independent of a clear and<br>
> unambiguous meaning for C itself, and in an interesting way: the models,<br>
> like us, essentially reinvent the entire literature, and have a conversation<br>
> as we do, inventing almost all the same positions that we've invented<br>
> (including the one in your paper).<br>
><br>
><br>
><br>
> I will read your paper and see if it changes my position. At the present<br>
> time, however, I can't imagine any information that would solve the<br>
> so-called zombie problem. I'm not a big fan of integrative information<br>
> theory - I don't think hydrogen atoms are conscious, and I don't think naive<br>
> bayes trained on a large corpus and run in generative mode is conscious.<br>
> Thus, if the model doesn't go through the same philosophical reasoning that<br>
> we've collectively gone through with regards to subjective experience, then<br>
> I'm going to wonder if its experience is anything like mine at all.<br>
><br>
><br>
><br>
> Touching back on QM, if we create a point neuron-based model that doesn't<br>
> wax philosophical on consciousness, I'm going to wonder if we should add<br>
> lower levels of analysis.<br>
><br>
><br>
><br>
> I will take a look at your paper, and see if it changes my view on this at<br>
> all.<br>
><br>
><br>
><br>
> Cheers,<br>
><br>
><br>
><br>
> Brian Mingus<br>
><br>
><br>
><br>
> <a href="http://grey.colorado.edu/mingus" target="_blank">http://grey.colorado.edu/mingus</a><br>
><br>
><br>
><br>
><br>
><br>
> On Tue, Jan 28, 2014 at 12:05 PM, Richard Loosemore <<a href="mailto:rloosemore@susaro.com">rloosemore@susaro.com</a>><br>
> wrote:<br>
><br>
><br>
><br>
> Brian,<br>
><br>
> Everything hinges on the definition of the concept ("consciousness") under<br>
> consideration.<br>
><br>
> In the chapter I wrote in Wang & Goertzel's "Theoretical Foundations of<br>
> Artificial General Intelligence" I pointed out (echoing Chalmers) that too<br>
> much is said about C without a clear enough understanding of what is meant<br>
> by it .... and then I went on to clarify what exactly could be meant by it,<br>
> and thereby came to a resolution of the problem (with testable predictions).<br>
> So I think the answer to the question you pose below is that:<br>
><br>
> (a) Yes, in general, having an outcome measure that correlates with C ...<br>
> that is good, but only with a clear and unambigous meaning for C itself<br>
> (which I don't think anyone has, so therefore it is, after all, of no value<br>
> to look for outcome measures that correlate), and<br>
><br>
> (b) All three of the approaches you mention are sidelined and finessed by<br>
> the approach I used in the abovementioned paper, where I clarify the<br>
> definition by clarifying first why we have so much difficulty defining it.<br>
> In other words, there is a fourth way, and that is to explain it as ...<br>
> well, I have to leave that dangling because there is too much subtlety to<br>
> pack into an elevator pitch. (The title is the best I can do: " Human and<br>
> Machine Consciousness as a Boundary Effect in the Concept Analysis Mechanism<br>
> ").<br>
><br>
> Certainly though, the weakness of all quantum mechanics 'answers' is that<br>
> they are stranded on the wrong side of the explanatory gap.<br>
><br>
><br>
> Richard Loosemore<br>
><br>
><br>
> Reference<br>
> Loosemore, R.P.W. (2012). Human and Machine Consciousness as a Boundary<br>
> Effect in the Concept Analysis Mechanism. In: P. Wang & B. Goertzel (Eds),<br>
> Theoretical Foundations of Artifical General Intelligence. Atlantis Press.<br>
> <a href="http://richardloosemore.com/docs/2012a_Consciousness_rpwl.pdf" target="_blank">http://richardloosemore.com/docs/2012a_Consciousness_rpwl.pdf</a><br>
><br>
><br>
><br>
><br>
> On 1/28/14, 10:34 AM, Brian J Mingus wrote:<br>
><br>
> Hi Richard,<br>
><br>
><br>
><br>
>> I can tell you that the quantum story isn't nearly enough clear in the<br>
>> minds of physicists, yet, so how it can be applied to the C question is<br>
>> beyond me. Frankly, it does NOT apply: saying anything about observers and<br>
>> entanglement does not at any point touch the kind of statements that involve<br>
>> talk about qualia etc.<br>
><br>
><br>
><br>
> I'm not sure I see the argument you're trying to make here. If you have an<br>
> outcome measure that you agree correlates with consciousness, then we have a<br>
> framework for scientifically studying it.<br>
><br>
><br>
><br>
> Here's my setup: If you create a society of models and do not expose them to<br>
> a corpus containing consciousness philosophy and they then, in a reasonably<br>
> short amount of time, independently rewrite it, they are almost certainly<br>
> conscious. This design explicitly rules out a generative model that<br>
> accidentally spits out consciousness philosophy.<br>
><br>
><br>
><br>
> Another approach is to accept that our brains are so similar that you and I<br>
> are almost certainly both conscious, and to then perform experiments on each<br>
> other and study our subjective reports.<br>
><br>
><br>
><br>
> Another approach is to perform experiments on your own brain and to write<br>
> first person reports about your experience.<br>
><br>
><br>
><br>
> These three approaches each have tradeoffs, and each provide unique<br>
> information. The first approach, in particular, might ultimately allow us to<br>
> draw some of the strongest possible conclusions. For example, it allows for<br>
> the scientific study of the extent to which quantum effects may or may not<br>
> be relevant.<br>
><br>
><br>
><br>
> I'm very interested in hearing any counterarguments as to why this general<br>
> approach won't work. If it can't work, then I would argue that perhaps we<br>
> should not create full models of ourselves, but should instead focus on<br>
> upgrading ourselves. From that perspective, getting this to work is<br>
> extremely important, despite however futuristic it may seem.<br>
><br>
><br>
><br>
>> So let's let that sleeping dog lie.... (?).<br>
><br>
><br>
><br>
> Not gonna' happen. :)<br>
><br>
><br>
><br>
> Brian Mingus<br>
><br>
> <a href="http://grey.colorado.edu" target="_blank">http://grey.colorado.edu</a><br>
><br>
><br>
><br>
> On Tue, Jan 28, 2014 at 7:32 AM, Richard Loosemore <<a href="mailto:rloosemore@susaro.com">rloosemore@susaro.com</a>><br>
> wrote:<br>
><br>
> On 1/27/14, 11:30 PM, Brian J Mingus wrote:<br>
><br>
> Consciousness is also such a bag of worms that we can't rule out that qualia<br>
> owes its totally non-obvious and a priori unpredicted existence to concepts<br>
> derived from quantum mechanics, such as nested observers, or entanglement.<br>
><br>
> As far as I know, my litmus test for a model is the only way to tell whether<br>
> low-level quantum effects are required: if the model, which has not been<br>
> exposed to a corpus containing consciousness philosophy, then goes on to<br>
> independently recreate consciousness philosophy, despite the fact that it is<br>
> composed of (for example) point neurons, then we can be sure that low-level<br>
> quantum mechanical details are not important.<br>
><br>
> Note, however, that such a model might still rely on nested observers or<br>
> entanglement. I'll let a quantum physicist chime in on that - although I<br>
> will note that according to news articles I've read that we keep managing to<br>
> entangle larger and larger objects - up to the size of molecules at this<br>
> time, IIRC.<br>
><br>
><br>
> Brian Mingus<br>
> <a href="http://grey.colorado.edu/mingus" target="_blank">http://grey.colorado.edu/mingus</a><br>
><br>
> Speaking as someone is both a physicist and a cognitive scientist, AND<br>
> someone who has written papers resolving that whole C-word issue, I can tell<br>
> you that the quantum story isn't nearly enough clear in the minds of<br>
> physicists, yet, so how it can be applied to the C question is beyond me.<br>
> Frankly, it does NOT apply: saying anything about observers and<br>
> entanglement does not at any point touch the kind of statements that involve<br>
> talk about qualia etc. So let's let that sleeping dog lie.... (?).<br>
><br>
> As for using the methods/standards of physics over here in cog sci ..... I<br>
> think it best to listen to George Bernard Shaw on this one: "Never do unto<br>
> others as you would they do unto you: their tastes may not be the same."<br>
><br>
> Our tastes (requirements/constraints/issues) are quite different, so what<br>
> happens elsewhere cannot be directly, slavishly imported.<br>
><br>
><br>
> Richard Loosemore<br>
><br>
> Wells College<br>
> Aurora NY<br>
> USA<br>
><br>
><br>
><br>
><br>
><br>
><br>
><br>
><br>
<br>
</div></div></blockquote></div><br></div>