Human intelligence and learning.

[Pharaoh Cat]

A half-baked notion that occurred to me after reading your post, Tank:

Maybe humans are best viewed as rule-makers.  Other creatures, including artificial ones, follow internal rules and discover external ones.  We, meanwhile, add to the world rules that never existed before.

Computer software, of course, is nothing other than rules made by humans to be followed by computer hardware.

The great question in my mind has always been, how did internal rules (what some would call "instincts") first evolve?  How is it that brains come with firmware?


EDIT: Split from another thread - Tank

[Tank]

A half-baked notion that occurred to me after reading your post, Tank:

Maybe humans are best viewed as rule-makers.  Other creatures, including artificial ones, follow internal rules and discover external ones.  We, meanwhile, add to the world rules that never existed before.

Computer software, of course, is nothing other than rules made by humans to be followed by computer hardware.

The great question in my mind has always been, how did internal rules (what some would call "instincts") first evolve?  How is it that brains come with firmware?

The bad firmware literally died out along with the first faulty slimeware. Natural selection is the ultimate test proceedure.

[Pharaoh Cat]

The bad firmware literally died out along with the first faulty slimeware. Natural selection is the ultimate test proceedure.

Yes, but why is there any firmware at all?  The existence of any at all is astonishing to me.  And where does it reside?  The cells of the brain are the hardware.  Where is the software?  Neurologists seem to me to be reluctant to consider that question.  They act as if the cells don't need software.  Yet computer hardware without software is inert.  Brain cells without software would simply absorb and expel chemicals like all other cells do.  I speak of "firmware" because it seems to me we animals are born with certain rules already in place.  Rules we're born with would be firmware.  Yet how is it possible that animals are born with rules in their brains?  Until we take this question seriously, we will never understand the brain.

[Davin]

Yet computer hardware without software is inert.

This is not quite true, lots of hardware does many things without software. All software does is emulate hardware. When you break the software down to the machine code, what you get are emulations of hardware gates: or, and, and/or, nor, nand... etc. are all the software does trillions of times a second (expecially while playing video games). So while this analogy may be useful in describing some processes, don't conflate analogies with reality, they only go so far.

[Tank]

The bad firmware literally died out along with the first faulty slimeware. Natural selection is the ultimate test proceedure.

Yes, but why is there any firmware at all?  The existence of any at all is astonishing to me.  And where does it reside?  The cells of the brain are the hardware.  Where is the software?  Neurologists seem to me to be reluctant to consider that question.  They act as if the cells don't need software.  Yet computer hardware without software is inert.  Brain cells without software would simply absorb and expel chemicals like all other cells do.  I speak of "firmware" because it seems to me we animals are born with certain rules already in place.  Rules we're born with would be firmware.  Yet how is it possible that animals are born with rules in their brains?  Until we take this question seriously, we will never understand the brain.

The 'brain' has been built over hundreds of millions and years and is never more complex than it needs to be to express the behaviours required for the organism to survive and reproduce. The basic 'brain' would initially have involved a few neurons, very much like some jelly fish currently have. These are simple stimulus:response machines. In humans this level of expression is controlled by the spinal cord in simple reflexes. Then one gets more complex activity in the cerebellum then overlaid by the cerebral cortex. Initially when a baby is born all of its motor control is built into the cerebellum. Primitive reflexes are the basic abilities a baby needs to survive for long enough for its cerebral cortex to develop and take over with learned behaviours.

The key point about the human brain that is different from run-of-the-mill computers is that the human brain learns, it isn't programmed. And it learns through a process of repetition and reinforcement. Its instinctual behaviour is to explore its environment, get its senses functioning, and mimic what it sees and hears. So there are only a few very simple behavioural rules at the base of the learning process, but by applying those rule repetitively and reinforcing the positive actions the required behaviours develop.

I wasn't particularly aware of this process with my own kids, I was too involved with the basics of keeping them clean, fed and out of danger. But quite unknowingly I was doing exactly what my children required. I now have a grandson and I can watch his development a little more objectively and one can see the development. When a child's brain first connects up is has way more neurons than it needs and they saturate the cortex but there are very few connections. As the brain developes some neurons make more connections than others and those that don't make enough die off. Repetitive stimulation causes neurons to make connections. As stimulation is normally associated with positive activities the brain becomes physically biased towards carrying out rewarded (good) behaviours.

An example of post natal cortex development can be demonstrated in mice. A new-born mouse has its whisker buds removed on one side of its nose so the whiskers never develop. When the mouse is full grow it is killed and dissected. One one side of its brain a pattern of cells can be found that mirror the whiskers on the other side of its head. The other side of its brain is smooth where the whisker mirror structure should be because the required stimulation was never provided to the cortex by the missing whiskers.

If you want to put in computer terminology (and it's not a good idea as a brain isn't like any human computer) one would say a child has a 'Built in Operating System' (BIOS). In the a computer this would be stored in ROM of some kind, in the baby's brain this is the cerebellum. Beyond this the computer analogy is not only bad, it's misleading as it implies the requirement of a program and therefore programer. The human brain programs itself through experience.

[Pharaoh Cat]

Yet computer hardware without software is inert.

This is not quite true, lots of hardware does many things without software.

You're talking about things like toasters and chainsaws.  The reason they don't have software is that they only do one thing so we didn't have to give them software.  Their behaviors are so simple they could be hard-wired.  Cars used to be the same way.

Robots, meanwhile, have software, because they do complicated things.  The closest mechanical analogy to organisms would be robots.

All software does is emulate hardware. When you break the software down to the machine code, what you get are emulations of hardware gates: or, and, and/or, nor, nand... etc. are all the software does trillions of times a second (expecially while playing video games).

I was a programmer in my younger days so I have a good sense of all that - and it's precisely my experience as a programmer that convinces me the brain couldn't possibly do what it does without software.  Now someone might tell me the DNA is the software but that would be false, since DNA controls cell growth only.  It doesn't control what the cells do.  It controls where and what they are.  What they do is built into their structure.  But brain cells are doing things that couldn't possibly be built into their structure.  There's a gap in our understanding and it won't be filled until we ask the right question.  Where's the code?  Where do the rules physically reside?  People like to say humans are hard-wired for the rules of language.  OK - show me.  Identify a rule sitting there in the brain.  This is the kind of question we need to ask. 

So while this analogy may be useful in describing some processes, don't conflate analogies with reality, they only go so far.

Hey, you said "conflate"!  I love that word. 

The brain is typically described as an organic super-computer and this description, when made, isn't intended as an analogy, but as concrete fact.  Where's the code?

[Pharaoh Cat]

If you want to put in computer terminology (and it's not a good idea as a brain isn't like any human computer) one would say a child has a 'Built in Operating System' (BIOS). In the a computer this would be stored in ROM of some kind

Yes - what I'm calling "firmware."

...in the baby's brain this is the cerebellum.

Not the cerebrum?  Interesting.  How does the cerebellum act as the firmware?  Are you saying the rules of language are built into the physical structure of the cerebellum?

Beyond this the computer analogy is not only bad, it's misleading as it implies the requirement of a program and therefore programer.

Yes - that's what scares people away from the question.  "Oh no, he mentioned software - that means a soul!  Yikes!"  But I'm not necessarily talking about anything we would recognize as a soul.  I'm certainly not talking about anything that could survive death.  Here, consider this: gestation is controlled by software.  The software is the DNA.  The firmware of cell growth is DNA.  That's the code.  That's where the rules physically reside.  A purely natural, physical, chemical, reality.  We need to find whatever it is that plays the role of DNA in the processes of perception, curiosity, judgment, imagination, desire, and decision-making.  I'm talking about something natural, physical, perhaps even chemical.  DNA didn't need a programmer (so far as we know) so the firmware of consciousness shouldn't need one either.

If we want knowledge we have to ask every question, even the ones that scare us because they might imply something metaphysically dubious to us.  If we can have DNA without God, then we can have brain firmware without God too.

The human brain programs itself through experience.

Based on rules.  Where do the rules physically reside?  You said the cerebellum and that intrigues me.  I look forward to learning what you mean by that.

[Tank]

DNA codes for a lot of things, most importantly in this case embryological development. A lot of the so-called 'junk' DNA does not code for proteins but cellular behaviour.

It's important to recognise human ego plays a huge roll in anthropomorphizing the way the brain works. Throughout history the brain has been likened to humanity's highest level of technology. In the Victorian era there were many mechanical analogies to brain operation. The fact is that we don't understand exactly how the brain works so any analogy of its operation is by definition incomplete and flawed.

Nural networks that can self program in a black box situation come out with identical responses but when opened up are wired differently. Simple rules, applied repetativly often create vastly different outcomes. As you know twenty binary choices can give a million possible combinational outcomes. The human brain has billions of nurons each with hundreds of connections, the possible wiring patterns are practically limitless, yet we all walk by putting one foot in front of the other and then into a controlled fall stopped by the other foot. And none of us can walk when we are born. But a baby Gnu can get up virtually instantly and run to keep up with it's mother. Its cerebelum is already pretty much wired up, in a black box sense, in the womb.

[Davin]

Yet computer hardware without software is inert.

This is not quite true, lots of hardware does many things without software.

You're talking about things like toasters and chainsaws.  The reason they don't have software is that they only do one thing so we didn't have to give them software.  Their behaviors are so simple they could be hard-wired.  Cars used to be the same way.

Don't make the mistake of trying to tell me what I'm talking about, because I'm very aware, I know full well what I'm talking about. I've done many complicated things without the use of software, like a mechanical arm that will perform complicated movements on it's own.

Robots, meanwhile, have software, because they do complicated things.  The closest mechanical analogy to organisms would be robots.

They do, do many complicated things, but it could still be done with just hardware (of course "could" would mean very large things because you can fit more emulated gates into a smaller space).

All software does is emulate hardware. When you break the software down to the machine code, what you get are emulations of hardware gates: or, and, and/or, nor, nand... etc. are all the software does trillions of times a second (expecially while playing video games).

I was a programmer in my younger days so I have a good sense of all that - and it's precisely my experience as a programmer that convinces me the brain couldn't possibly do what it does without software.  Now someone might tell me the DNA is the software but that would be false, since DNA controls cell growth only.  It doesn't control what the cells do.  It controls where and what they are.  What they do is built into their structure.  But brain cells are doing things that couldn't possibly be built into their structure.  There's a gap in our understanding and it won't be filled until we ask the right question.  Where's the code?  Where do the rules physically reside?  People like to say humans are hard-wired for the rules of language.  OK - show me.  Identify a rule sitting there in the brain.  This is the kind of question we need to ask.

I'm a programmer now and have been since I was eight (23 years total), now that we have the appeal to authority out of the way, we can get back into discussion? Just because you can't imagine a brain doing things without software, doesn't mean that it can't, this is the argument from ignorance fallacy.

Now the analogy to software is very useful, but it's still an analogy, it only goes so far (like that pesky DNA code analogy that trips up evolution deniers). The people that say things like, "the brain is hardwired for X" should have the show part, and I've seen many examples of this. They don't actually mean that there are wires in the brain that have been soldered into place so that a certain part can only do one thing and nothing else. In fact, the "hardwired" analogy breaks down with the adaptive nature of the brain. So while I agree with using the analogies for explanatory purposes, you seem to be taking the analogies for realities.

So while this analogy may be useful in describing some processes, don't conflate analogies with reality, they only go so far.

Hey, you said "conflate"!  I love that word. 

The brain is typically described as an organic super-computer and this description, when made, isn't intended as an analogy, but as concrete fact.  Where's the code?

But it isn't a concrete fact, it's an analogy. The brain being described as an organic super-computer is an analogy, because it's not a super-computer, it's a brain. There are many things one can do to a super-computer that they can't do to a brain, this is because, while they have many similarities (which is the purpose of the analogy), they are not the same all the way through. I can analogize eating and reading, however ones belly is not any more full after reading.

Edit: Sure Tank, just rip the conversation right out from under me. Move this too please.

[Tank]

The 'rules' of learning and motor control are built into the brain stem and cerebellum. All humans can create all the phonetics required in all languages. But through the learning process some phonyms are reinforced and some not. This is why Japanese and Chinese people can't pronounce 'R'. And what's worse they never can do it properly without immense practice as the original pathways were not reinforced while they grew up. They have to force their brain to build a control pathway from scratch. There is a phonym in Arabic that non-native Arabic speakers have terrible trouble getting right.

We don't have rules of our extended adult behaviour built-in, we have basic behaviours that allow us to learn and thus in a very crude sense we 'program' ourselves through trial and error.

[Stevil]

Humans seem to be much less hard wired than other animals, it generally takes us a year to learn to walk, a couple of years to learn to talk.
Many animals can walk within minutes of being born. Imagine a fish that couldn't swim.

But I guess that our lack of hard wiring gives us a much bigger potential to learn. We are only able to do this because our parents are able to cater for our needs. Until animals are able to look after every need of their young, the young must retain hard wiring.

[Pharaoh Cat]

A lot of the so-called 'junk' DNA does not code for proteins but cellular behaviour.

Interesting!  I didn't know that!  Time to do some googling. 

It's important to recognise human ego plays a huge roll in anthropomorphizing the way the brain works. Throughout history the brain has been likened to humanity's highest level of technology. In the Victorian era there were many mechanical analogies to brain operation. The fact is that we don't understand exactly how the brain works so any analogy of its operation is by definition incomplete and flawed.

Oh, you and I are in violent agreement there. 

Nural networks that can self program in a black box situation come out with identical responses but when opened up are wired differently. Simple rules, applied repetativly often create vastly different outcomes. As you know twenty binary choices can give a million possible combinational outcomes. The human brain has billions of nurons each with hundreds of connections, the possible wiring patterns are practically limitless, yet we all walk by putting one foot in front of the other and then into a controlled fall stopped by the other foot. And none of us can walk when we are born. But a baby Gnu can get up virtually instantly and run to keep up with it's mother. Its cerebelum is already pretty much wired up, in a black box sense, in the womb.

So you're indeed saying it's the cerebellum (not the cerebrum) that's wired, say, for the rules of language.  OK.  If the wiring is somehow defined by the shape of the network, the pattern of connections, then the shape of the network, the pattern of connections, is the software, the encoded rules.  OK.  I'll buy that if and when someone seriously presents it as a testable hypothesis, tests it, and gets confirming results.  I want someone to say out loud, "See that shape?  See that pattern?  That's a language rule.  Let me show you the data that makes me say that."

We can't just hand-wave away the encoded rules.  The neurologist tells us, "the brain is hard-wired for X."  Fine, my dear neurologist - isolate X in the shape or pattern of connections.  Or at least ask the question and look for the answer.  Assume the encoded rules are in there somewhere (because they damn well have to be) and search for them.

[Tank]

Edit: Sure Tank, just rip the conversation right out from under me. Move this too please.

Soz, done 

[Pharaoh Cat]

I've done many complicated things without the use of software, like a mechanical arm that will perform complicated movements on it's own.

Cool!  Can the arm make decisions?

I think I see what you mean by emulating hardware.  My father used to program room-sized computers in the very early days, when processes we moderns would use software for, were hard-wired.  Still, in that case, we could point to the wires and say, "See these wires?  These wires make the computer do X.  These other wires make it do Y."

We have to be able to point to something and say, "See this?  This makes the brain do X."  Maybe we'll be pointing to a set of chemicals that get excreted by one cell and absorbed by another, similar to how ant colonies govern themselves.  Maybe we'll be pointing to "junk" DNA that isn't junk after all.  Maybe we'll be pointing to patterns of connection.

Just because you can't imagine a brain doing things without software, doesn't mean that it can't, this is the argument from ignorance fallacy.

I think my use of the word "software" has confused rather than clarified.  See above for the kinds of things I would consider software.  Basically anything at all that encodes or actualizes rules governing processes is software as far as I'm concerned.  I'll start talking about rules instead.  The meta-rules that govern rulemaking have to physically exist somewhere, either as entities, or as patterns, or as events, but we have to be able to point to them.

[Davin]

Edit: Sure Tank, just rip the conversation right out from under me. Move this too please.

Soz, done 

Whatever... //_¬