Can we eliminate predation?

[RandFan]

Earthborn,

Learned behavior as I understand dictates some trial and error. According to ethologists an orb web spider can spin a perfect orb web first time every time and can do so without ever seeing another spider spin a web and inspite of its environment.

I note that orb web spiders have already been discussed. If you addressed this specifically already then just say so and I will try and find it.

I have to say that the lack of trial and error and the ability to build webs without example suggest no learning.

 

[uruk]

Then it appears to me that you are wrong. Even the part you highlighted says the exact opposite of what you are arguing.

I am sorry but it does not. The articles, even the ones you posted, clearly state that the source of the morphogenes is the DNA. The DNA is the template from which RNA is produced and from the nucleic acids in the RNA bond with peptides and produce protiens.

Please re-read the quote here:

"An important class of molecules involved in morphogenesis are transcription factor proteins that determine the fate of cells by interacting with DNA."

This quote from the article you linked to says this:

"Gene Transcription: DNA → RNA
DNA serves as the template for the synthesis of RNA much as it does for its own replication. "

"Summary
Gene expression occurs in two steps:
transcription of the information encoded in DNA into a molecule of RNA (described here) and
translation of the information encoded in the nucleotides of mRNA into a defined sequence of amino acids in a protein (discussed in Gene Translation: RNA → Protein)."

The site you posted CLEARLY states DNA to RNA to PROTIENS. And the morphogenes ARE protiens.

The neural pathways that make this possible grew during its time in the womb and may have had a different function inside the womb than outside.

Show me proof of this statement, give me an example from an outside source. Even you admit this is an assumption on your part.

Of course it does. Every part of the body has a different structure than every other part. But that does not answer the question: why assume that genes determines how every part of the brain is structured, except the neo-cortex? Saying that it has a different structure does not answer why its structure develops in an entirely different way.

The structure of the cortex is determined by our genes via morphogenesis. The neural connections in the neo-cortex which stimulated by our senses is laid down as we experiance i.e. memory, dexterity, etc.. Other connections which procces certain senses like sight, sound and touch are preconnected as the embryo is developing. That is why we all use the same part of our brains to processe those signals. The structures or circuits in the limbic system of the brain are also laid down during gestation i.e. physiological emotional response autonomic responses, etc..

Morphogens come from the cells that make them. Those cells use genes to determine their molecular shape and thus their function. It is not the genes that determine the shapes, but the cells using those genes.

Yes it is the genes. The cells are using the morphogenes to determine thier shape. And the morphogenes come from the DNA strand. Remember DNA to RNA to PROTIEN!
How else does your son or daughter inhereit your eye color or your hair color or your build or even any genetic defects you have. The genes determine eye color, hair color and other characteristics. It is encoded in the DNA.

You have shown enough proof of it yourself. Young spiders make smaller and simpler webs, and in the end they all make similar webs. That means they have opportunity to learn, and in the end all must have learned it in pretty much the same way. The only alternative would be to assume that the spider's nerve cells somehow grew the necessary circuits without learning, but that's an impossibility because nerve cells cannot not learn (they'd die if they tried) and forming (or altering) neural circuits equals learning.

A small orb web is still an orb web. The infant spiders do not stay together. They leave the egg sack immediately after hatching. The necessary circuits do grow and connect while in gestation. This does happen otherwise the infant heart will not beat in the proper sequence. The muscles in the diaphragm will not contract to breath, the infant will not suckle, the infant will not cry when spanked on the bottom. It is necessary otherwise the infant will die.
You are making quite a bit of statements and assumptions without backing them up. You assume them because it seems to make sense to you. Way back it seemed perfectly sensible to believe that an ether permeated the universe. Michelson/ Morley proved that "sensible" notion wrong while trying to detect this ether.

Here is proof of such your line of thinking:

Observation shows that those spiders all manage to make a web, and therefore they must have learned how to do it.

You make the jump from observation to assumption. The spider manages to make a web so they must have learned how to do it.
No doubt by comparison to humans. Spiders are much less complex then we are. Thier brains amount to not much more than ganglion and notocords. There is very little room for "learning" in the human sense. And if you do your researching you will find that an isolated infant orb web spider will start to build a small orb web. The small size being in proportion due to its diminutive size. Enlarge that web and the pattern will still be the same as an adult orb web, just smaller.


This is becoming a bit exhasperating. Clearly, we both see two different interpretations of the same data. And so I must agree to disagree. Be well.

 

[Earthborn]

Originally posted by RandFan
I have to say that the lack of trial and error and the ability to build webs without example suggest no learning.

I disagree. Often what seems to be efficient behaviour can be divided into a whole lot of small 'trials and errors' that are easy to miss because the animal quickly repairs its errors and tries something different. Just because errors seem absent because they are quickly spotted and restored does not mean they do not exist and that the behaviour must be automatic. It just means the animal is a quick learner.

Remember also that the spiders that never get the hang of web making have a much smaller chance of survival, and quickly disappear from the scene making the remaining spiders appear as if they made no mistakes. You simply never hear from the spiders that made the biggest mistakes.

Originally posted by uruk
The DNA is the template from which RNA is produced

Indeed it is. A template: yes. A program: no.
A template does not equal a program.

The site you posted CLEARLY states DNA to RNA to PROTIENS. And the morphogenes ARE protiens.

The words in all caps make me think what you are saying is a shocking revelation that somehow proves me wrong. The sentence itself doesn't, though.

Show me proof of this statement

Show you proof that the nervous system of a human being develops during its time in the womb? I think it is fairly obvious, actually.

give me an example from an outside source.

What do yu mean with 'outside source' ?

Even you admit this is an assumption on your part.

It is an assumption on my part that the neural pathways that seem only to have function outside the womb, must have had a function to the fetus inside the womb. I admit that. I do not think it is an unreasonable assumption: neural pathways grow with use, and since they grew when the fetus was inside the womb it is hard to imagine that wasn't used for anything and still managed to grow.

The neural connections in the neo-cortex which stimulated by our senses is laid down as we experiance i.e. memory, dexterity, etc.. Other connections which procces certain senses like sight, sound and touch are preconnected as the embryo is developing.

All true. But you seem to forget that an embryo is not neurologically dead. It is experiencing things on its own, and these experiences help form the nervous system.

The genes determine eye color, hair color and other characteristics. It is encoded in the DNA.

It is obvious that characteristics that are caused by a limited number of proteins can be found more directly in the genome. The genome can be seen as a sort of list of ingredients (also an analogy that should not be taken too seriously). In the ingredient list for soup, the characteristic of redness is directly encoded in that part of the list that calls for the use of tomatos. If you inherit an ingredient list from someone that misses this 'gene' you won't get red soup.

Way back it seemed perfectly sensible to believe that an ether permeated the universe. Michelson/ Morley proved that "sensible" notion wrong while trying to detect this ether.

So something that seems to make sense -- even something that seems to make sense to the entire scientific community -- must be wrong? Is that what you are trying to say?

There is very little room for "learning" in the human sense.

Correct. But there is a lot of room for learning in the neuronal sense. Spiders still have many thousands of neurons, much more than the most powerful learning machines, and each neuron is also much more complex than any artificial neuron in any artificial neural network. And those machines can learn to some degree, so why not spiders?

 

[uruk]

I know I said i was not going to respond anymore but...Here I am.

Indeed it is. A template: yes. A program: no.

A template is a pattern, a guide. A program is a pattern, a guide.

Show you proof that the nervous system of a human being develops during its time in the womb? I think it is fairly obvious, actually.

No, proof that the neural pathways that facilitate the crying behaviour in infants had a different purpose while in the womb.

What do yu mean with 'outside source' ?

A source other than what your opinion on what seems obvious to you.

I do not think it is an unreasonable assumption: neural pathways grow with use, and since they grew when the fetus was inside the womb it is hard to imagine that wasn't used for anything and still managed to grow.

And sometimes they just grow because the cells are told to do so by the DNA during embryonic development.

So something that seems to make sense -- even something that seems to make sense to the entire scientific community -- must be wrong? Is that what you are trying to say?

Here is anothe example of your assumptions. (maybe Im just not making myself clear)
Not "must" but the entire scientific community has been wrong in the past. Thats why we rely on data, experimentation and follow the evidence where it takes us reguardless of what seem obvious or reasonable to us.

 

[RandFan]

I disagree. Often what seems to be efficient behaviour can be divided into a whole lot of small 'trials and errors' that are easy to miss because the animal quickly repairs its errors and tries something different. Just because errors seem absent because they are quickly spotted and restored does not mean they do not exist and that the behaviour must be automatic. It just means the animal is a quick learner.

I'm confused. How would a spider spot an error? How does a spider know what an error is? What is the feedback?

 

[uruk]

It just occured to this slow old boy that we are debating from two sides of the same issue. It appears to me that you are stressing the mechinizm while I am stressing that which guides that mechinisim.

The reason for the analogy of DNA to a computer program is a valid one. Take for instance the process of putting an image on a computer screen. The mechinizims for displaying the image are the electronincs and parts that make a monitor. But without the program to guide the electronics and the information that contains the image, nothing is diplayed on the monitor. It is the patterns of ones and zeros of which the program is composed that activate certain circuits in the proper sequence and guide those circuits to display an image. Now do those one and zeros play a role in the shape or configuration of the components that make up the monitor? Yes they do, in a manner of speaking. The manufactures of the monitors know that the programs which control the process of displaying images is digital in nature. i.e. ones and zeros. so the circuits have to be designed and built with that in mind. ( I know in the electroninc industry electroninc components and devices evolved instep with the software. The design of one drove the design of the other which is more than likely the case with DNA and Physiology) In some cases the hardware has to be design to conform as to which operating system (program) is being used. i.e is the product "windows compatable" or "Mac compatable" (this was truer in the old days) In this manner the program defines configuration. And I, agree so does the environment. The DNA has to respond to the surrounding environment as to which configurations are more successful in that environment.

The same holds true (more or less) for DNA and Physiology. There are biological mechinism that shape the configuration of the organisim, ( and these do include neural connection configurations. Remember it is neural chemicals that cause neurons to connect and disconnect. This process had been observed in the labratory and there are many forces which can guide and instigate this process and DNA is but one of them) but those mechinisism are guided by the the information encoded in the DNA. It is your DNA that tells you that you are human or canine or ceatation or red headed or blue eyed. That configuration is encoded in the DNA and that code guides the mechinisims that creates the form. The environment of the cells within the body is also guided by the information encoded in the DNA.

I feel the program analogy IS a valid one. Otherwise the geneticists themsleves would not refer to it so often.

 

[uruk]

Correct. But there is a lot of room for learning in the neuronal sense. Spiders still have many thousands of neurons, much more than the most powerful learning machines, and each neuron is also much more complex than any artificial neuron in any artificial neural network. And those machines can learn to some degree, so why not spiders?

What the spider "learns" is how to place the web within the immediate surroundings. What it already "knows" is how to weave the particular pattern of web. (i.e. orb rather than funnel)

And before you say: "But there are little programs inside those microchips", here's Mark Tilden's website showing of some simple robots that can do similar things without programs. In fact, without any digital electronics at all!

Yes that's because the behavior is hardwired in the design of the electronics. I'm familiar with Tilden's work. The robot discovers the way to walk not by "learning" in the conventional human sense where we think about something before we do it) But by trial and error of built in behaviours. A photovore robot has a built in behaivior "move toward a light source" it has mechinizims that facilitate this behaivouir. Optical sensor, motors, wheels. You turn on the robot and the optical sensors control the motors. If one optical sensor senses more light than the other it cause the motor to spin a wheel in a particular direction that will increase the amount of light being recieved by the sensor. the motors keep moving until the maxium amout of light is recieved. No thinking oro "learning" in the conventional sense.
Something analogus exists for the orb spider. it has a built in behaviour to spin an orb web. It has the mechinizims for doing such. What it learns by trail and error is how to place that orb web into it surroundings. The orb spider will not try top place a funnel web into that environment because the funnel web design is not built into the orb spider. That is the difference between built in behaviour and "learned" behaviour.

 

[Hellbound]

Having followed the conversation on DNA and environemnt, I have a few things to say.

Earthborn:

You seem to disregard the fact that programs can have conditional statements, not just a flat, unchanging set of rules. If-then, While-Do, Case and Select, etc, etc, etc are all programs. Take Microsoft Word (or whatever word processing software you use). Do you get the exact same document everytime you run the program? No, you apply input to the program to guide the result. The program reacts to the input. That does not mean it isn't a program. It will always produce a word processing document (barring errors and such), but those documents can be varied.

I could easily write a computer program that reacts to the environemnt, and does almost everything as a reaction to the environment. That does not mean it isn't a program.

Likewise, DNA progvides the program, and the environment provides the inputs to that program. Dna sets the framework that the inputs (environment) works on. In this sense, the similarity of program to DNA is even more clear. Without the DNA, there would be nothing to react with the environment. Without the environment, the DNA would have nothing to spur it's growth of cells and tissues and protiens.

Finally, just as in some programs there are portions that are inaccessible to inputs and some portions that react to input, DNA acts in a similar manner. Your DNA may be hard coded to produce a certain protein in a certain way...no amount of learning can change that. However, the way that protien acts once created is affected by environment, both internal and external. Especially in lower-order animals and the basic instincts in higher-order animals, some behavior is hard-coded, in the sense that, barring extremes of environment (a lack of sufficent input, to carry on the program analogy, or a variable mis-match or something similar), the organism will react that way because the DNA codes for the behavior to be built-in as a reflex.

Regarding learning machines and neurons, here you're falling into the trap of equating neurons to programs, a point which you labelled as incorrect before. In any case, even the learning computer must be built with certain hard-coded functions already in place, otherwise the computer does not have the basic abilities required to learn. And many computers are designed for simple, hard-coded tasks...and likewise organisms can be constructed as such. Just because a computer has a fast and powerful processor, does not automatically mean it can learn...although if it can learn at a useful rate, then there is a certain minimum processor power than can be calculated. The number of neurons in a brain/nervous system does not mean an animal can learn, but animals that can learn will most likely have at least a certain amount.

It seems you are trying to paint this as an all-or-nothign proposition, and example of the Either-Or fallacy. Currently, evidence and scientific opinion lean toward the truth being somewhere in the middle. There are some hard-coded behaviors and others that are learned, and the lower on the scale an animal is the more it's behavior tends towards hard-coded and vice versa (virus to bacteria to fungus to spider to cat to monkey to human).

 

[Earthborn]

Originally posted by uruk
What it already "knows" is how to weave the particular pattern of web. (i.e. orb rather than funnel)

How do you know it knows this?

Yes that's because the behavior is hardwired in the design of the electronics.

I brought it up to show that consistent behaviour is possible without a program. The behaviour is only hardwired because it cannot grow. Animals can.

Originally posted by Huntsman
You seem to disregard the fact that programs can have conditional statements

No, I don't.

not just a flat, unchanging set of rules.

In a program, the program flow can change, but it can only change into the things that were programmed into it.

If-then, While-Do, Case and Select, etc, etc, etc are all programs.

Take an IF...THEN statement:

IF Condition = True THEN
DoOneThing
ELSE
DoOtherThing
END IF

Every IF THEN statement splits the program flow in two, and depending on the condition, the program can do one thing or another thing. But the things it can do are programmed in. The program above can do two things. If it has two IF THEN statements it can do four things, with three it can do eight things etc... It is theoretically possible to calculate every possible position a program consisting of IF THEN statements can be in. It maybe an incredibly large number, but it is finite.

To get complex behaviour, the program basically needs to have an answer in store for every possible occurance. You'll get a very complex program. Behaviour as complex as that of insects or spiders requires a program that will easily exceed the few tens of thousands of genetic 'instructions' that are available. So if it is a program, where is it? And where is the computer that runs it?

Take Microsoft Word (or whatever word processing software you use). Do you get the exact same document everytime you run the program? No, you apply input to the program to guide the result.

But it will never grow beyond its programming. It will not adapt to new surroundings, or make associations between new challenges or opportunities. And even though it is large and complex, it can only deal with a very low bandwidth of sensory input: maybe a hundred and ten keys from the keyboard, movements and clicks from the mouse and maybe one or two other devices.

Compared to that, living organisms have to deal with an incredible barage of sensory information. The only way they can deal with it is not running any algorithms at all, but employing parallel processing.

I could easily write a computer program that reacts to the environemnt, and does almost everything as a reaction to the environment. That does not mean it isn't a program.

Yes, you could. But to make it do exactly what a spider does would require a very large program, a very powerful computer and you'll find that it will run incredibly sluggish. And it also will not be as adaptive as a real spider. Place the real spider in a new environment where it needs to exhibit new behaviour to survive, and it quickly will. Your computer will find it very tough to invent new behavioural patterns from scratch. You'll quickly find that it is more efficient to model (preferably in hardware) how the spider does it than to try to do the same algorithmically.

Your DNA may be hard coded to produce a certain protein in a certain way...

DNA doesn't produce anything. The living cell does. It produces proteins using DNA.

Regarding learning machines and neurons, here you're falling into the trap of equating neurons to programs

No, I am not. Neural networks don't have to be programs, they can also be made with hardware. They cannot run any programs and because they use parallel processing they are a closer analogy (but still only an analogy) to nervous systems than algorithmic programs are. They can exhibit all kinds of complex organic behaviours and they don't need to be programmed as they are able to learn. So why assume animals need to be programmed?

In any case, even the learning computer must be built with certain hard-coded functions already in place, otherwise the computer does not have the basic abilities required to learn.

That's because it cannot grow and can only exist when it is made directly in its 'adult' form.

Just because a computer has a fast and powerful processor, does not automatically mean it can learn...although if it can learn at a useful rate, then there is a certain minimum processor power than can be calculated.

A computer is not a learning machine. It can be programmed to simulate one, but that would mean it must simulate parallel processes as an algorithmic, step-by-step process. It will therefore be slow. Much slower than a machine that has hardware capable of learning.

The number of neurons in a brain/nervous system does not mean an animal can learn, but animals that can learn will most likely have at least a certain amount.

If the animal has one neuron or more and a few simple senses, it can learn. Not much, but it can. Even the artificial neurons of a neural network (a very simplified analog of real neurons) can.

Currently, evidence and scientific opinion lean toward the truth being somewhere in the middle. There are some hard-coded behaviors and others that are learned

I think the evidence and scientific consensus lean towards the idea that animals grow and develop, and are not made in a factory where people build all their parts in completed form. Nothing is built in ready to go, and every part of the animal underwent a development instead.

But I could be wrong about what scientists believe these days...

 

[Hellbound]

Earthborn:

I don't know how to say this tactifully, but it is obvious you have a very basic grasp of programming, and what consititutes a program.

Not pnly can program flow change, the possible outcomes of the program can change and grow with time. This is done already. For a simple example, look at the various "Twenty Questions" type websites that exist. They grow and adapt by adding new information to their database whenever they are stumped. The program/blueprint (DNA) stays the same, but environmental factors allow them to alter their database of responses (much as memory and conditioning work in animals).

Also, I'd suggest a bit of study on emergent behavior. Complex behavior of insects does not "easily overflow the tens of thousands of genetic instructions". Simple instructions can give rise to very complex behavior. One of the first examples of this was the simple game "Life", that had very limited rules but produced behavior strikingly similar to that of real organisms. Additionally, insect behavior is precisely what has been modelled in many small, robotic systems...because they can use relatively few, simple instructions that, when multiple units are acting in concert, develop into complex emergent behaviors.

Also, you seem to use parallel processing as a synonym for nueral network.. Parallel processing is, quite simply, the ability to run multiple tasks at once. My home PC does this now. Parallel processing does not eliminate the need for programming, it simply allows multiple programs to be run simultaneously.

As to writing a program that does what a spider does, I doubt it would be that large. As I mentioned, studies are undergoing in these areas as we speak, but I'll touch on that more later. It is purely your assumption that this would require some huge, complex program and a huge, powerful computer.

Yes, I know DNA doesn't produce anything itself, I assumed you would understand what I meant with that statement.

Next, on neural networks:

No, I am not. Neural networks don't have to be programs, they can also be made with hardware.

Just because it's hardware, does not mean it isn't a program. It simply means it isn't a software program. ANY logic that can be done in hardware can be programmed in software. The first computers were all hardware based programs. Punch cards and paper tape were the next extension of this. Current software programs work by directing the CPU (a general purpose electronic IC) to emulate various combinations of hardware gates. Neural networks still require programming (done in the blueprint, at the least...all that I know of still contain software code, as well), and MUST be able to be reprogrammed on the fly in order to learn (by assigning differring weights to various linkages between elements of the network). The reason this is done in neural networks is not because it isn't possible in software, but because a dedicated hardware program is much more efficient at the task it is designed for (as opposed to a general purpose computer).

It seems you next contradict yourself by claiming that the learning computer must be programmed, when in the paragraph before you stated neural networks don't require programming. Neural networks are still computers. In any case, neural networks need some programming as well, both hardware and software, so I don't see how this supports your argument.

A computer can be a learning machine, neural networks being a case in point. And not all computers have to simulate parallel processing...my home computer is perfectly capable of parallel processing (dual CPUs), and massivley parallel processors are being used as supercomputers as we speak, capable of true parallel processing of hundreds of tasks simultaneously.

Finally, I have NEVER, in ANY of my posts stated that animals are "completely hardwired", nor have I ever denied that they grow or develop.
I am not arguing that all behavior is hard-wired. I am not arguing that organisms don't develop. I am not arguing that learning isn't possible for some portion of animals. What I am arguing is that some behaviors are built-in to the organism, specifically certain instinctual behaviors that are necesary for the animal to survive. Learning can add to this, and in extreme cases it can counteract some of it in individuals...if that individual has sufficient capacity to learn. To my knowledge, this has never been done successfully with spiders.

The consensus is that some instinctual behaviors are hardwired, while others are learned. The greater the capacity for learning in an animal, the fewer behaviors are hardwired. It's a scale, from virii and bacteria (which are completely hardwired systems incapable of learning), to humans (who have a few instinctual behaviors, many of which can be overwritten by later learning).

Edited to remove inappropriate remarks and clarify some formatting errors. It's been a long day, I apologize.

 

[Capsid]

There are some 100 000 million neurons in the brain, far more than those 3 000 million base pairs, most of them junk DNA. Far more than the 30 000 genes. And then there is the huge number of interconnections between all those neurons, which must be somewhere in the gazillions. There is simply no way to describe the complexity of the brain in something that is as small as DNA.

I would argue that viruses and bacteria are very complex and yet they have very few genes.

 

[Earthborn]

don't know how to say this tactifully, but it is obvious you have a very basic grasp of programming, and what consititutes a program.

Well, I certainly don't claim to be an expert.

much as memory and Compelx behavior of insects does not "easily overflow the tens of thousands of genetic instructions".

I think it would. You do realise that insects do have quite a lot of sensory data to process, don't you?

Additionally, insect behavior is precisely what has been modelled in amny small, robotic systems, because they can use relatively few, simple instructions that, when multiple units are acting in concert, develop into complex emergent behaviors.

True, but so far only a limited number of insect behaviours are simulated, and the robots involved have much fewer sensors to perceive the world around them. If they had many more, even the simple instructions that cause the reactions to all the stimuli will add up very quickly.

Paralell processing is, quite simply, the ability to run multiple tasks at once.

Not according to Wikipedia, which tells that we are both wrong:

Parallel computing is the simultaneous execution of the same task (split up and specially adapted) on multiple processors in order to obtain faster results.[/url]. I meant the ability to run multiple tasks at once, and I specifically mean the ability to run multiple contradictory tasks at once and having them compete against eachother.

Parallel processing does not, in any way, shape, or fashion, eliminate the need for programming.

Not necessarily, no. But a system that can have multiple tasks compete against eachother and vary their influence on overall behaviour, no longer needs programming after it has been set up. It can sort of 'program' itself.

As to writing a program that does what a spider does, I doubt it would be that large.

You think so? Reading out all its sensor cells, and controlling all of its behaviours? I think it would be rather big.

As I mentioned, studies are undergoing in these areas as we speak

True, but those usually limit themselves to specific behaviours (such as web weaving) and use a very simplified model of a spider.

Also, not all programming is purely algorithmic, either

What kind of programming is non-algorithmic?

Jusdt because it's hardware, it does not mean it isn't a program. It simply means it isn't a software program.

Then when is something a program according to you, and when is it not?

Of course, the next thing you do is contradict yourself by claiming that the learning computer must be programmed, when in the paragraph before you stated neural networks don't require programming.

I meant, after the network is built, it does not need to be programmed. If you consider the assigning of different weights to be programming, then you could say that each neuron needs to be able to 'reprogram' itself. But the network in its totality does not need a program.

NO ONE HERE IS ARGUING THAT ALL BEHAVIOR IS HARD-WIRED. NO ONE HERE IS ARGUING THAT ORGANISMS DO NOT DEVELOP.

If you think that the nervous system also develops and grows, you have to explain to me what you mean with 'hardwired' and 'built in' as it would mean that there is never a period where the neural pathways are 'built in' but not yet used.

 

[uruk]

How do you know it knows this?

All of entomology (the part concerning spiders) says so.

Now I am not arguing that web weaving is limited to only genetic influences. The world is not black and white.

Here is a quote from a Baylor university web site:
http://bioedonline.org/slides/slide01.cfm?q=species&pg=1

"Even though it is useful to think of behavior as being strictly genetic (inborn) or strictly learned, the line between these two types of behavior is blurred. Even a rigidly automatic behavior depends on the environmental factors for which it evolved, and most learning is guided by internal neural programs.

In general, instinct is considered to consist of heritable, genetically specified behavior organized and guided by circuits in the nervous system. In other words, innate or instinctive traits are those existing in, belonging to, or determined by factors present in an individual from birth. In the 1800s, C.O. Whitman coined the term “instinct” to describe behaviors he observed in pigeons. Innate behavior is that which appears to be developmentally fixed, meaning a particular behavior occurs in all members of a species, regardless of environmental influences. Innate behaviors are present at birth and do not require experience or practice, but do require a stimulus. Many of these behaviors improve with practice or physical maturation. Examples of innate behavior include hibernation in black and brown bears, nest building in birds, web weaving in spiders, and mating in all animals."


But I will add that entomologist at one time thought that the spider web pattern was limited to solely genetic influences, but that has changed. Web patterns will vary from individual spiders of the same species due to the local environment, molt cycle or age and feeding cycle and size of the individual spider. But the underlineing pattern or type of web is still the same. Not one entomologist will tell you that if you raise a orb weaving spider with a funnel weaving spider that the orb weaver will start weaving funnel webs. The spiders are incapable of that level of learning.

originally posted by Huntsman
The first computers were all hardware based programs. Punch cards and paper tape were an extension of this.

True. One of the first digital computers built, ENIAC, was programed by physicaly rewiring the vaccum tubes. To program the computer, you had to have a soldering iron and wire cutters. That's where they get the term "hard wiring". (consequently, also the term "bug in the system". A moth or roach would get into the computer and short out a connection.)

 

[Hellbound]

Parallel computing and parallel processing are not the same thing. Parallel Processing is exactly as I described it (see http://dictionary.reference.com/search?q=Parallel Processing). This can include multiple processors running the same task or multiple processors running one task. In either case, my home PC is capable of this (and is doing it now, as we speak). The Wikipedia definition, if that's the listing for Parallel processing, is incorrect (or, more accurately, incomplete). It is correct for parallel computing. In computing, minor changes in terminology can denote changes in meaning, as well

In any case, modern system can run multiple tasks at once and have them compete, it's not that hard to set up. It does require a bit more baseline programming, but can be done.

As to sensory data, much of the processing is done in the sensory organ itself, in many lower-order animals.

And also, just because they compete, does not mean it doesn't require additional programing, but I agree that you can create a system that programs itself. That was one of the points I was making earlier. However, you need some basic framework set up to guide this self-reprogramming.

As to what is a program, a program is simply a set of logical steps executed to perform a task or function. In hardware systems (dedicated electronics), these logical steps are in the form of electrical connections and IC chips, logic gates and transistors. Each of these componenets can be represented in formal logic (based on function) or in a software program. As I stated, software programs used today developed from the and are direct extensions of the hardware programs used before them. IN fact, even today dedicated systems are developed using hardware over software for boosts in efficiency and durability.

As to neural networks not needing programming after they are built, I would agree (to an extent, they can often corrupt themselves into pointless loops and dead ends). However, that's pretty much the point we've been trying to make. DNA is responsible for that initial programming, for guiding the physical construction of components that make up an organism.

As to your last question, I have no idea what you mean. The hardwired portions are grown with the organism as it develops. Once they are they, they are functional. whether they are used or not then depends on environmental factors. Your own pain reflex is an example. The human nervous system has a built-in program that says, basically, IF Ouch! THEN Move Away. I don't see where the dichotomy you are positing comes into play.

 

[uruk]

Not according to Wikipedia, which tells that we are both wrong:

I think the computer term he meant was multi-threading Which is the CPU's ability to run more than one program or function at the same time. By the way, parallel processing is extremely good at multi-threading

I think it would. You do realise that insects do have quite a lot of sensory data to process, don't you?

Actually, not really. They have rudimentary sight (they can see patterns rather than detail), touch hairs (which usually triggers a flee response), something that resembles taste in thier antennae, but no hearing in some cases. They do not process sensory input any where near as complex as we do. It usually amounts to "there's food" and "run away from this or strike that" and "seek mate".

You think so? Reading out all its sensor cells, and controlling all of its behaviours? I think it would be rather big.

Not as big as you'd think. Remember that PDF about the guys that programmed a computer to weave webs I think it only required a desktop computer.

What kind of programming is non-algorithmic?

The kind that is in the physical design of system or device.

Then when is something a program according to you, and when is it not?

When you design a system to perform a certain function that works in conjunction with other systems. What's not a program? Hmmm. A rock sitting in the middle of nowhere serving no particular function.

I meant, after the network is built, it does not need to be programmed. If you consider the assigning of different weights to be programming, then you could say that each neuron needs to be able to 'reprogram' itself. But the network in its totality does not need a program.

The network in its totality can be considered to be a program.

 

[luchog]

There are deviation in nature. Otherwise spiders wouldn't be able to weave their webs under different conditions.

They aren't able to weave their webs under different conditions. They weave exactly the same type of web, regardless of the conditions; and they stay within a range of micro-environments appropriate to that type of web. You will never seen an orb web spider weaving a cob or funnel web. You will also never see an orb web spider weaving a web on the ground, but only in a space where such a web would work.

To adapt their web, they will have to constantly deviate from what to us seems a common plan, or else the web could not fit between the branches it is made.

No, the web type does not deviate at all from the style specific to the species. The only deviation is size and location. There is no variation in the silk type or construction pattern within members of the same species. For example: the Subfamily Argiopinae spiders, such as the St. Andrew's Cross spider, always weaves a specific pattern into their webs, using a specific type of silk; a pattern which no other species of spider uses. The similar Garden Orb spider, Subfamily Araneinae, weaves a web that lacks the distinctive cross pattern. Yet both spiders are found in the exact same environments. One can even find them practiaclly side by side. If the web is the result of environment, why do they both have such dramatically different patterns? Then there is the Tent spider, Subfamily Cyrtophorinae, which is nearly identical to the St. Andrew's Cross spider, is found in identical habitat; but doesn't weave orb webs at all, but tent webs nearer the ground.

All three of these spiders can be found in the same environment, in the same backyard even, yet they are all very distinctively different.

If they learn from their environment, why is that?

While making a web, it constantly alters its own behaviour. See the computer simulation article.

This is meaningless. It doesn't alter it's own behaviour, it behaves in the same way. It makes a web, the same type of web every time, in the same way every time.

Its degree of reaction is limited by the ways it has previously developed/grown its body.The difference is between programmatic behaviour and adaptive behaviour. Programmatic behaviour is fixed, and program flow can only change it fixed IF statements: all possible reactions of the program are defined from the beginning. It is algorithmic, program flow always exists at only one point in the program, and all instructions are followed one by one, one after the next.

Adaptive behaviour is different and is caused by many simultaneous processes happening together and influencing eachother. There is not one single program flow, but instead there are many. Therefore the behaviour that results from it is not algorithmic, but chaotic and only predictable probabilistically. Such behaviour is much more efficient when dealing with the real world, because not every possible occurance in a chaotic and unpredictable world needs to be programmed in.

This is gibberish. I'm not sure you really understand what a program algorithm is.

One of the first robots where such behaviour was implemented was Genghis. It can walk very rugged terrain almost effortlessly. If you want to do this programmatically, you'll need a very powerful computer, an enormous array of sensors, very complex algorithms and even then you'll probably only get very slow movement. Genghis uses only a few microchips, has no 'walking program' and very limited sensors. Note also that it is moves very similarly to an anthropod. It is not programmed to walk, but it learns to do it very quickly.

And before you say: "But there are little programs inside those microchips", here's Mark Tilden's website showing of some simple robots that can do similar things without programs. In fact, without any digital electronics at all!

You are quite clearly not understanding what you're reading, and clearly do not understand what an algorithm is; since the information you've posted above here is mostly gibberish; and what little isnt, directly contradicts your comments. All of these robots do have very clear, hardwired programs. Even the small ones. Extremely basic ones, but they are still hardwired logic.

Hell, Tilden's own BEAM robotics system refers to his work with AI programming and genetic algorithms for building his robots; all of which so far have embedded controllers with specific logical algorithms for determining behaviour. Non-algorithmic logic is his stated goal, but he has clearly not achieved that yet; and some of his work shades into some pretty woo speculation. None of the robots can actually learn anything, they just operate within the limits of their onboard mechanical or electronic programming.

The rest of the post was just more reiteration of the same circular reasoning; and unparseable gibberish, so I've cut it; since it's been effectively refuted already.

 

[Earthborn]

Originally posted by Huntsman
As to what is a program, a program is simply a set of logical steps executed to perform a task or function.

That tells me very little. What is a 'logical step' ? Does there need to be an order specified in these steps for it to be logical? Are there also 'illogical steps' ?

DNA is responsible for that initial programming

Is DNA a set of logical steps? Can it not just be a set of templates for proteins?

The hardwired portions are grown with the organism as it develops. Once they are they, they are functional.

the problem is this: the neurons are already active and functional from the very start. They don't wait until the pathways they are part of are fully grown. They don't start to behave any differently when you think the circuits are 'hardwired'. In some cases, they can 're-hardwire' circuits which means the wiring was not so hard as it seemed. If hardwiring is something different than learning, what exactly is the difference?

Originally posted by uruk
They do not process sensory input any where near as complex as we do.

Don't compare them to humans. Compare them to the robots used to simulate emergent behaviour. We're probably decades away from robots that can process that much sensory information in real-time.

Remember that PDF about the guys that programmed a computer to weave webs

Which used a very simplified model of a spider and those spiders didn't even have to deal with the chaotic real world.

What's not a program? Hmmm. A rock sitting in the middle of nowhere serving no particular function.

I think a definition of 'program' that includes almost everything is fairly useless. The genome would be a program, and everything you said about it is irrelevant because it would be a 'program' regardless. It would still performs a certain function in conjunction with other systems. It would of course not solve any of the misconceptions people have about genes because of the use of this word.

Cupboards are programs, tapwater is a program, I am a program, even computer is itself a program. Whatever.

 

[uruk]

I think a definition of 'program' that includes almost everything is fairly useless. The genome would be a program, and everything you said about it is irrelevant because it would be a 'program' regardless. It would still performs a certain function in conjunction with other systems. It would of course not solve any of the misconceptions people have about genes because of the use of this word.

I think Huntsman is right about you. You are a literalist. The subltleties of analogies seem to be lost on you as does sarcasm.

Don't compare them to humans. Compare them to the robots used to simulate emergent behaviour. We're probably decades away from robots that can process that much sensory information in real-time.

Presently robots can process sensory information in real time sufficiently to drive across the nation without human input.

Not to mention drive across a debri strewn field several million miles away from earth.

Or handle the intricacies of bipedal locomotion. (Robosapien is an example of true bipedal movment with very little computational power.)

 

[RandFan]

I think a definition of 'program' that includes almost everything is fairly useless.

I would be interested in your definition of "program"?

pro·gram Audio pronunciation of "program" ( P ) Pronunciation Key (prgrm, -grm)
n.

6. A set of coded instructions that enables a machine, especially a computer, to perform a desired sequence of operations.

When I was born my DNA enabled my cells to perform a set sequence of operations. The code is so precise that I could theoretically reconstruct myself using the DNA.

I would also be interested in your definition of "template"? As a programmer I can tell you that often programs are referred to as templates. Template and program are not necessarily exclusive so I don't even know what you mean when you separate the two.

When water travels down hill it seeks the course of least resistance. This is logical. Is this a program? It could be called that but I might not. However one could use water, gravity and gates to construct a computer program.

So what is special about "program" and what does DNA do that disqualifies it to be a "program"? BTW, did you know that you can use DNA to solve mathematical problems?

How DNA Computers Will Work

You won't believe where scientists have found the new material they need to build the next generation of microprocessors. Millions of natural supercomputers exist inside living organisms, including your body. DNA (deoxyribonucleic acid) molecules, the material our genes are made of, have the potential to perform calculations many times faster than the world's most powerful human-built computers. DNA might one day be integrated into a computer chip to create a so-called biochip that will push computers even faster. DNA molecules have already been harnessed to perform complex mathematical problems.

Why can this be done?

A Fledgling Technology

In fact, DNA is very similar to a computer hard drive in how it stores permanent information about your genes.

Adleman is often called the inventor of DNA computers. His article in a 1994 issue of the journal Science outlined how to use DNA to solve a well-known mathematical problem, called the directed Hamilton Path problem, also known as the "traveling salesman" problem. The goal of the problem is to find the shortest route between a number of cities, going through each city only once. As you add more cities to the problem, the problem becomes more difficult. Adleman chose to find the shortest route between seven cities.

You could probably draw this problem out on paper and come to a solution faster than Adleman did using his DNA test-tube computer. Here are the steps taken in the Adleman DNA computer experiment:

1. Strands of DNA represent the seven cities. In genes, genetic coding is represented by the letters A, T, C and G. Some sequence of these four letters represented each city and possible flight path.
   
2. These molecules are then mixed in a test tube, with some of these DNA strands sticking together. A chain of these strands represents a possible answer.
   
3. Within a few seconds, all of the possible combinations of DNA strands, which represent answers, are created in the test tube.
   
4. Adleman eliminates the wrong molecules through chemical reactions, which leaves behind only the flight paths that connect all seven cities.
   
   The success of the Adleman DNA computer proves that DNA can be used to calculate complex mathematical problems.[/list=1]

1. What does Adleman's DNA computer do that our DNA doesn't? Doesn't encoding determine the outcome? You would agree that bits of protein strung together in a double helix constitute encoding, right?

 

[Hellbound]

Randfan gave a better, more precise definition of program than I, and I'll second that definition.

Yes, neurons are active and functional from the start, but the pathwy isn't. The pathway and how the cells within it interact are what make the hardwired behavior. Reflexes and instincual activities are emergent behavior, if you insist on looking at it from the level of single cells then of course it will seem absurb, much as any other function a multi-celled creature performs. Your protest in this part is like claiming that bridges can't be built, because each brick begins supporting weight as soon as it's lain, and you can't have the bridge buitl all in one swoop. Reflexes and instincual behaviors are controlled by more than single neurons. They are controlled by groups of neurons that connect in specific ways, and specialized neural areas in the CNS (for higher-order mammals) or other nervous tissue. The development of these areas is coded for in the DNA. The DNA also codes some areas (for those animals capable of learning) as more general purpose areas, used for learned behavior and social evolution.

As to "re-hardwiring" circuits, do you have an example of this? I do not understand what you mean. If you could show me a case where a persons reflex actions were turned into something else (baring something obvious such as paralysis or physical damage to the nervous tissue), I would be suprised. If you simply mean that nerves can regrow, then that does not defeat my point. They'll regrow in pathways that will redevlop those instintcual reactions (or attempt to, barring physical limitations).

Uruk:

No, multi-threading was not what I was referring to. Multi-threading is simply the capability to run multiple tasks, and can be done with a single processor via various strategies such as time slice or demand. Parallel Processing refers specifically to the capability to multi-thread multiple applications, or parts of a single application, across more than one processing unit, working on seperate parts of the problem in parallel and re-linking them at the end. It can be a single program running on multiple processors, or multiple programs spread across multiple processors. All depends on context. (I might refer to what my body does as a sinle unit, or might refer to the actions of my individual cells as seperate, likewise with parallel processing. Each processor works on a seperate part of the program, application, but at the lowest level each CPU has it's own "piece" or program to run).