(May 13, 2015 at 3:01 pm)Rhythm Wrote: My own models, lol..nothing at all approaching mind in it's totality or effectiveness.  Don't want you to get the wrong idea.  I model (and sometimes build) circuits largely with an aim to design better sensory for commercial ag - and no one's buying my little toys...they're all private use..lol.
(I'm butchering one of those parrot drones right now to haul an ir camera and plan paths that follow ir trace for watermanagement...and trying to sucker a buddy into programming a mobile app to handle the data, so as to replace the functionality lost when I destroy the native systems in order to jerry rig it.)

-IR can provide a means for detecting a whole range of problems in crops, from water to pests to disease.  Imagine having a bunch of cheap, tiny, resource efficient drones that did all of our detection work for us real-time, and could feed that data to our smartphone?  Trick is getting it to fly itself, eh?

Wow, that's impressive   I won't pretend to know all you're talking about basically you're an electrical engineer, into robotics? I find electronics really hard to grasp unfortunately. I'm just a programmer, or I was until I got RSI; still do it for fun sometimes but never again as a job   I have done a bit of Android programming in Java and have a small app on my phone but I really hate it - Android programming is the most laborious, boring, unintuitive programming I've ever come across. But it's nice to know that if I could come up with an x-factor idea, I would stand a chance of being able to make it into a reality.

Anyway I've looked at your link and watched the video. I wish they'd taken that approach to computing when I was at college: building a system from scratch and letting you play Linus Torvalds or that other guy, was it Richard Matthew ?Stallman? I think I have a better understanding of where you are coming from now; the guy in the video talks about increasing levels of abstraction and I always talk about that as well in reference to neural networks, as does my book, so I think we're on the same page. But the question of whether a neural network can be conceptualised in terms of logic gates like NAND is going to be a lot harder to answer. I do remember reading at some point that some people thought some neural circuits could have that sort of effect but I didn't really look into it much at the time. It might be fun to look and see how you could in theory create each of the logic gates with a neural network but whether they're possible or not doesn't make any difference if they're not configured that way in the brain; that will be the big test - finding them in the brain if they do turn out to be feasible. Anyway today I ordered a book that may be helpful called "Computing the Mind: How the Mind Really Works". I don't know if it's about ctm specifically but it is certainly about identifying all the features of mind, so could possibly be. That should give me a better idea of the distinct features to be looking at.

LOL, no, electrical engineers are accredited, and have certifications......I'm just a guy with an HDL that knows how to print boards and solder comps.
(got into it when I was a kid, worked under the table at a rapid prototyping house - then I went back into it professionally, as management after the service and some school (mech engi, not electrical). It's actually really simple. You, as a programmer, could increase your value greatly by learning hardware architecture.

On topic, though - yeah, NAND is brilliant. The answer that ctm might offer, regarding brain structure..is that our nuerons are most definitely -not- nand gates..but nand simply describes an implementation of an implicational relationship within the boundaries of whatever material it was formed from. So, say, a string NAND is not capable of everything that a digital NAND is capable of (that's why string comps like our targeting systems from ww2 went out as fast as they came in). Similarly, a nueral net may be performing NAND, it may be arranged so that a NAND can be applied to data, and even though it's not, strictly speaking, built -as- a NAND...it is capable of performing the NAND function. The NN has the better explanation for implementation, imo...while CTM has a better explanation of the underlying principles. That's all just my opinion. In short, it's much more likely, between the two, that our brains are structured as NN, however, the manner in which the NN works leans on classical computing, CTM.

(May 13, 2015 at 4:07 pm)Rhythm Wrote: LOL, no, electrical engineers are accredited, and have certifications......I'm just a guy with an HDL that knows how to print boards and solder comps.
(got into it when I was a kid, worked under the table at a rapid prototyping house - then I went back into it professionally, as management after the service and some school (mech engi, not electrical).  It's actually really simple.  You, as a programmer, could increase your value greatly by learning hardware architecture.

Well, I still think it's really impressive, even if it's just a hobby   Thanks for the tip. Would one of these, what is it, Raspberry Pi or something, things be helpful, or is that a different subject altogether?

Those things are awesome!  I was looking at those but they're too heavy.   Mine, after 200grams or so...wobbles...and I get no flight time.  Pretty much full throttle all the time.  I believe mine works slightly under specs (I looked them up but I can't remember how much it can lift, only that it didn't live up to the specs).  It's not the board itself, but the associated peripherals..like the IR camera - but even more than that the battery.  Also, alot of the power of that thing would be wasted for the task I have in mind, at least in the manner I want to try and make it do the work.

I just need a board that can detect IR variance, and then focus it's efforts around that area - all while flying a parrot within defined boundaries (property lines associated /w an onboard nav, a radio fence..who knows, haven't got that far yet). Ideally it would be simple, cheap, and built entirely to-purpose...like a hammer. It's farm equipment, after all. I'm sure that an actual engineer could whip something up in no time, professionally....but that doesn't sound like a hell of alot of fun.

(I want to get my daughter one, though, just for fun)

-or did you mean helpful in learning architecture? The answer to that is no, absolutely not - though I've been told they're great for flaunting what you -already- know in attaching peripherals..lol.....

(May 13, 2015 at 4:07 pm)Rhythm Wrote: On topic, though - yeah, NAND is brilliant.  The answer that ctm might offer, regarding brain structure..is that our nuerons are most definitely -not- nand gates..but nand simply describes an implementation of an implicational relationship within the boundaries of whatever material it was formed from.  So, say, a string NAND is not capable of everything that a digital NAND is capable of (that's why string comps like our targeting systems from ww2 went out as fast as they came in).  Similarly, a nueral net may be performing NAND, it may be arranged so that a NAND can be applied to data, and even though it's not, strictly speaking, built -as- a NAND...it is capable of performing the NAND function.  The NN has the better explanation for implementation, imo...while CTM has a better explanation of the underlying principles.  That's all just my opinion.  In short, it's much more likely, between the two, that our brains are structured as NN, however, the manner in which the NN works leans on classical computing, CTM.

Sorry you kinda lost me again. Are we looking for a NAND gate or not? If we are, then at the level of individual neurons I'd say not a chance but in terms of groups of neurons it's a definite possibility. Take a simple NOT for instance. That could be achieved, possibly, with inhibition which is a prevalent feature in the brain's neural networks. Inhibition serves to stop neurons getting too excited and to turn them off. In fact I think the knee-jerk reflex uses an approach very much like that and with very few neurons so it should be very illustrative. It uses inhibitory interneurons to switch between the various muscles that it needs to trigger in opposition to each other. You'll have to bear with me while I look into the details of that.

---

(May 13, 2015 at 4:37 pm)Rhythm Wrote: Those things are awesome!  I was looking at those but they're too heavy.   Mine, after 200grams or so...wobbles...and I get no flight time.  Pretty much full throttle all the time.  I believe mine works slightly under specs (I looked them up but I can't remember how much it can lift, only that it didn't live up to the specs).  It's not the board itself, but the associated peripherals..like the IR camera - but even more than that the battery.  Also, alot of the power of that thing would be wasted for the task I have in mind, at least in the manner I want to try and make it do the work.

I just need a board that can detect IR variance, and then focus it's efforts around that area - all while flying a parrot within defined boundaries (property lines associated /w an onboard nav, a radio fence..who knows, haven't got that far yet).  Ideally it would be simple, cheap, and built entirely to-purpose...like a hammer.  It's farm equipment, after all.  I'm sure that an actual engineer could whip something up in no time, professionally....but that doesn't sound like a hell of alot of fun.

(I want to get my daughter one, though, just for fun)

-or did you mean helpful in learning architecture?  The answer to that is no, absolutely not - though I've been told they're great for flaunting what you -already- know in attaching peripherals..lol.....

I meant the latter but your first answer is cool too 

We're looking for something that performs the gate function - which, obviously, anyone doing comp architecture would just call a gate.  It may not be structured like a digital gate, but a string gate isn't structured like a digital gate either, and so it shouldn't surprise us to see that a biological gate is different from a digital or string gate.  Yes, we're looking for NAND, but we're not looking for any specific -type- of NAND.  No one expects to crack someones skull open and see a bunch of analog tubes rendered in biological material fall out, amiright?

(May 13, 2015 at 5:01 pm)Rhythm Wrote: We're looking for something that performs the gate function - which, obviously, anyone doing comp architecture would just call a gate.  It may not be structured like a digital gate, but a string gate isn't structured like a digital gate either, and so it shouldn't surprise us to see that a biological gate is different from a digital or string gate.  Yes, we're looking for NAND, but we're not looking for any specific -type- of NAND.  No one expects to crack someones skull open and see a bunch of analog tubes rendered in biological material fall out, amiright?

OK got yer again. I think it's very well possible that any gate you care to mention could be achieved in the neural networks of the brain because it's actually a lot more complex that most people, including me, are capable of imagining. In the cerebral cortex for instance, I believe, the axons of neurons can have about 1000 axon terminals and there are exponentially more post-synaptic sites on the dendrites and cell bodies of the receiving neurons. So though the popular conception of neurons is of a cell with one process coming out the front (the axon) and a dendritic tree of branches coming out the back onto which axons synapse, in fact the single axon has several 'terminals' where it can synapse with another cell and it can even synapse directly with the cell body of a cell. Anyway add to this inhibitory neurons, along with feedforward and feedback connections between six or seven layers in the cerebral cortex and I wouldn't be surprised at all that there could be many logic operations in action there.

Anyway give me a few days and I'll make it my project to figure out how it could work and do some research into possible work already done on it.

Exactly, they are more robust than any digital gate, and -could- function as universal gates.  Charles Peirce first showed that NOR could execute the function of any other gate in series or parallel (or some combination).  Henry Sheffer proved (and importantly, relative to Peirce, published...lol), using NAND, that there was such a thing as a universal gate, upon which any logical function could be built.  

see Peirces Arrow, Peirce generally being the more well known not only because he was the first to publish his work, but also because he was a pioneer when it came to demonstrating that electronic gates could be built. Obviously useful in hindsight.....lol

It is important, for any theory of mind invoking computation or logical operations, that this was found to be true.  Because whatever we may choose to conceptualize natures role in all of this one thing can be said with confidence.  Nature is not a designer, not a programmer.  It is a much simpler construction paradigm that nature toes the line on.  If nature were capable of "blindly producing" a thinking machine, repetition of components would, by the nature of it's constructive means, feature heavily.  We would not expect to find task specific gates in greater abundance than universal gates, if this were an "accidental computer".  We would expect, instead, to find something very much like a neural net comprised of universals, in structure - which is what we -do- find....yet another reason I feel that computation is a powerful explanation for at least -some- of the effects we attribute to mind, even if I could not, with certainty, say that computation explains it all, to everyones satisfaction (or even my own).

Computers, huh? I guess that works too. I've always thought we were more like elevators though. The circumstances of our upbringing and genetic inheritance send us to the appropriate floor and presto. Here we are.

Next week how we are all equal to a toaster.

LOL, not so much equal to, or equivalent to - as operating upon the same principles.  It's just a possible explanation of a specific function, not a description of the human/toaster/computer clade...hehehe. OPs language was overreaching, but the comparison may have a great deal of merit if we take care to keep it appropriately limited.