Seeing red

[bennyboy]

I'd probably me more likely to skip the ANN theory and look at evolutionary theory. I like the idea of ideas evolving as they "bubble up," rather than adjusting weights to virtual neural connections at the end of complete trials (if that makes any sense).

[The Grand Nudger]

I don't think that the universe -does- know when information is happening.  I'm not really sure what that has to do with a mind occurring.  In any case, in a mechanistic explanation, the operations performed by the various portions of the brain -are- the things we are experiencing.  You are experiencing, for example, the effect of alcohol on your brain...specifically the effect of ethanol molecules on receptors, when you are drunk.  

I am a singular agent.  Neurons and brani parts are plural.  Unless there is some part of the brain which has the capacity to accept as input different kinds of processed information, then how do you believe this information gets coordinated?  Do you deny that a singular agent is experiencing sight and sound at the same time?  If there isn't one, then who's experiencing the illusion that there is one?

Of course, and that singular agent is me, my ears, my eyes, the whole me.  It makes no sense to say "I am a singular agent" as though all of these moving parts are not involved.  

-All- of the different parts of the brain appear to have the capacity to accept as input different kinds of processed information.  It's almost definitional as to what a component is as a piece of our CNS in the first place.........?   The cerebellum, for example, appears to be coordinating eye-hand movement (and perhaps a host of other sensory inputs with behavioral outputs).  Is that what you're looking for?   I suspect that it's coordinated computationally.  

What illusion?  The illusion that your ears hear, your eyes see, and your brain fills in?  I don't think that's an illusion.  I think you've taken it to mean something that it doesn't.  

Well, I have changing perceptions, but despite those changes, I still have what seems like a consistent, and persistent, sense of self and consciousness.  There's something at the core, which is why metaphors like a movie screen or a theater stage get mentioned.

People like to think that, that there's some immutable core to their being, despite all evidence to the contrary.  If there is a core, then it's you, the whole you, and not in any way immutable.  The many, many cases of traumatic brain injury and psychological disorder,as well as simple responses to open ended questions asked over large spans of time put the lie to the notion of consistent, persistent senses of self consciousness.  You may have that perception at any given time, and within that context it's almost a non-question (who else would you be, or think you were?), and while that perception doesn't seem to change..... the "self" certainly does, for a variety of reasons, regardless of what it's made out of or how it works.

I think that alot of confusion arises between our positions surrounding the idea of an illusion.  I'm asked "how do you explain the illusion of x".  There is none.  That "illusion" seems to occur at the level of rationalization, not perception.  This  perception does not have to speak to an objective truth of the universe and nothing in our experience would lead or suggest that it does, it need only refer to the current status of the system - which is always itself, by definition, no matter who it may be.

I like the idea of ideas evolving as they "bubble up," rather than adjusting weights to virtual neural connections at the end of complete trials (if that makes any sense).

Life, reproduction, and death -do- "adjust weights" to nueral connections at the end of complete trials. The ability of those connections to do work appears to be a strong selective pressure. If they under-perform in any given rep we'd expect that rep to have fewer offspring relative to an average or above-par performer. The squid that bolts first and fastest survives.

This is how a heuristic architecture is arrived upon, in terms of evolution, and that's important to comp theory, for example, because there was no top down design of circuits and bussing involved, so far as we can tell.

[emjay]

I'd probably me more likely to skip the ANN theory and look at evolutionary theory.  I like the idea of ideas evolving as they "bubble up," rather than adjusting weights to virtual neural connections at the end of complete trials (if that makes any sense).

Not really   But I take it that means you've changed your mind about the book? You could learn about competition in neural networks... so that's almost like an evolutionary struggle... and since I believe that neurons represent things, including ideas, it could well be not that far from what you've just said (in a way ) When a NN is in the process of settling it can be thought of as like a raging battle, with inhibitory neurons forcing one winner out of many.

[bennyboy]

Life, reproduction, and death -do- "adjust weights" to nueral connections at the end of complete trials.  The ability of those connections to do work appears to be a strong selective pressure.  If they under-perform in any given rep we'd expect that rep to have fewer offspring relative to an average or above-par performer.   The squid that bolts first and fastest survives.  

This is how a heuristic architecture is arrived upon, in terms of evolution, and that's important to comp theory, for example, because there was no top down design of circuits and bussing involved, so far as we can tell.

Yeah, both you and Emjay caught me out on that, and I was thinking about it while I was at work today. If I made a thread on how to build a game involve evolved traits in a simple space shooter or something, would you guys help out? My idea is to have the player make a planned ship, with bought upgrades or whatever, and fight wave after wave of minions that evolve traits in response to how I kill them. . . and show the player that against evolution, failure is inevitable no matter how well you try to plan.

[emjay]

Like the Borg in Star Trek? A few of them go down after getting shot until they evolve and suddenly phasers are useless? Sure, go for it, but I'd think it would get boring pretty quick   The game that is, not the thread 

[bennyboy]

Like the Borg in Star Trek? A few of them go down after getting shot until they evolve and suddenly phasers are useless? Sure, go for it, but I'd think it would get boring pretty quick   The game that is, not the thread 

The trick would be how to evolve new shapes and patterns with unique effects not predicted by the game maker.  You need the rules of the game to determine the outcome, but not through design.  So you couldn't just have a "whoever is stronger survives, so now we have stronger aliens" because that would suck.

Maybe a complex rock-paper-scissors scenario would do it.  Have like 50 variables which interact with each other in complex ways: shield makes better defense but heavier.  Heavier has higher momentum, but worse ability to turn.  Good ability to turn adds mobility, but increases stress on alien's innards.  Different resources (randomly placed in passing space debris) will allow monsters to maximize different traits, but limited by genetic capacity.  This way, the progammer knows the rock-paper-scissors rules, but cannot predict the result.

[emjay]

Like the Borg in Star Trek? A few of them go down after getting shot until they evolve and suddenly phasers are useless? Sure, go for it, but I'd think it would get boring pretty quick   The game that is, not the thread 

The trick would be how to evolve new shapes and patterns with unique effects not predicted by the game maker.  You need the rules of the game to determine the outcome, but not through design.  So you couldn't just have a "whoever is stronger survives, so now we have stronger aliens" because that would suck.

Maybe a complex rock-paper-scissors scenario would do it.  Have like 50 variables which interact with each other in complex ways: shield makes better defense but heavier.  Heavier has higher momentum, but worse ability to turn.  Good ability to turn adds mobility, but increases stress on alien's innards.  Different resources (randomly placed in passing space debris) will allow monsters to maximize different traits, but limited by genetic capacity.  This way, the progammer knows the rock-paper-scissors rules, but cannot predict the result.

Okay, I'm game to talk about it if you want to start your thread but not right now cos I'm just about to go to bed. What triggered this line of thinking?... did you want to use a neural network for the game mechanics?

[bennyboy]

Okay, I'm game to talk about it if you want to start your thread but not right now cos I'm just about to go to bed. What triggered this line of thinking?... did you want to use a neural network for the game mechanics?

I want to see how many variables need to interact to make the results completely unpredictable by the player or programmer.

[emjay]

Okay, I'm game to talk about it if you want to start your thread but not right now cos I'm just about to go to bed. What triggered this line of thinking?... did you want to use a neural network for the game mechanics?

I want to see how many variables need to interact to make the results completely unpredictable by the player or programmer.

Sorry about the slow reply I'm a bit distracted now because all our talk about the book has got me excited about it all over again and I want to go through it again, this time using Emergent. So I think I'm now going to bow out of this thread, but you guys are welcome to keep it going and I'll still read it from time to time. It's given me a lot of insight and I feel like I'm in a new phase of understanding with plenty enough to keep me going so it's served it's purpose for me

I think it would be pretty cool if you could do that cos it looks to me like it could have uses as a random number generator I'm afraid I don't know what to say except good luck with your project... I'm sure it makes sense to you You did get me thinking though about trying to model an 'evolving' shape in a neural network. Say you have a cube with nine 'panels' on each face, like a Rubik's cube, and each of those panels represents a pressure sensor and has a corresponding neuron in the input layer, then it would be interesting to see what representations and expectations form in the network given simulations of what could happen to it... such as just laying on a table (ie nine panels active)... being pushed off a table (nine panels + one on an adjoining face... representing pushing it along with a finger... then some of the nine opposite from where the pressure is turning off as the cube hangs over the edge before falling... then the random way it bounces, rolls, and comes to rest at the bottom, meaning 3 panels each from two adjoining faces if it bounces on an edge, and 1 each from three faces if it bounces on a corner etc... no pattern in the which faces/edges/corners hit first after the fall but you would still expect a general pattern of it's movement to emerge... the rolling motion. Then to 'evolve' this thing you would just add panels/mini-cubes to the surface and add corresponding neurons to the input layer. I'm just curious to see what sort of representations would form and if it could come to represent its own shape. I'm just thinking that about the question of what could be the simplest hypothetical organism to model where you could see, scaled down massively, the principles at work in the human brain.

[bennyboy]

Say you have a cube with nine 'panels' on each face, like a Rubik's cube, and each of those panels represents a pressure sensor and has a corresponding neuron in the input layer, then it would be interesting to see what representations and expectations form in the network given simulations of what could happen to it... such as just laying on a table (ie nine panels active)... being pushed off a table (nine panels + one on an adjoining face... representing pushing it along with a finger... then some of the nine opposite from where the pressure is turning off as the cube hangs over the edge before falling... then the random way it bounces, rolls, and comes to rest at the bottom, meaning 3 panels each from two adjoining faces if it bounces on an edge, and 1 each from three faces if it bounces on a corner etc... no pattern in the which faces/edges/corners hit first after the fall but you would still expect a general pattern of it's movement to emerge... the rolling motion. Then to 'evolve' this thing you would just add panels/mini-cubes to the surface and add corresponding neurons to the input layer. I'm just curious to see what sort of representations would form and if it could come to represent its own shape. I'm just thinking that about the question of what could be the simplest hypothetical organism to model where you could see, scaled down massively, the principles at work in the human brain.

Sounds a lot like this to me: