That’s a nice diagram you did there. Although the diagram looks simple but it touches on large number of complicated sub components. For example, the diagram mentioned about somatic does that mean you will implement the motor side? And with the limbic system will you also implement emotion also? Not sure if I miss it in the long post, I don’t seem to find the area it talks about the testing. Would love to hear your thought on how to test. I will spend more time on re-reading the post and all the related posts.
I am not sure if I am one of those since I really haven’t heard of the term pseudo code since I was in high school. And not sure what “C language formatted language independent Pseudo code” really is, isn’t that just C?
I don’t like getting into the discussion on subjective things in coding. Too religious and everyone thinks their code is better. Any language is fine to me. The only reason why I mentioned about OO is because the design of OO was inspired by biology and so would it would seem to be a good fit to use it to develop a biological system and would be easy to read because each class can reflect the corresponding actual biological part of the system. The challenging part would be how to manage the massive parallelism efficiently, both energy consumption and processing latency. Our brain can potentially have chemical reaction simultaneously in trillions of synapses in close proximity.
But I am going to focus on the conceptual side first. Will worry about the implementation later.
I have no idea if you know C but if you did you would know that the natural way to organize and access complicated data structures in C is with “structures” and pointers to members of these structures, things that are not really all that compatible with Python programming.
Most modern languages use C formatting and this is a good way to to write things so most programmers can understand it.
If I use the C notation to access a table of pointers to pointers to members of an array of structures most Python programmers would be utterly lost. Pointers to functions? Yes, I plan to use that in dispatch tables. Again, not a thing that Python people have experience with. Example code exposition riddled with interactions between structures of axons and structures of dendrites based on these data access methods is not going to be very clear for most reader and I would be peppered with a never-ending stream of clueless newbie questions.
This is not a good way to communicate when I know that a significant fraction of my audience will never have seen anything but python OO code. So no - C code is not the best way to speak to a wider audience. Not optimal.
And the PERL code that I want to use for the master control panel? Even worse.
This means that I have to express algorithms and data structures using pictures and C like pseudo code code fragments to be clear to programmers no matter what language they are familiar with.
Unfortunately couldn’t get much out from this discussion on defining intelligence but the topic got into more about whether it is necessary to define what intelligence is. Personally, without defining what the fundamental elements of intelligence really are I am certain any brain building project will fail because massive amount of time will be spent on building something unnecessary. For example, I do agree with Jeff Hawkins at https://www.youtube.com/watch?v=fiZESDyPPag 43.28 that emotion is not a necessity of intelligence. If we spend time on building the parts related to emotion then the system will be so complicated and will never get done. And without the list, there is no way to test if the implementation achieves the objectives. I will continue to search for what I am looking for. Thanks everyone for the help in this!
2 Likes
Oh, they are definitions available. Plenty. What you won’t get nowadays is a consensus.
Consider this: at this point, any “brain project” is in fact a way to maybe reach the goal of finally defining what it is in a more concrete way.
What we’re quite sure about is that there exists biological implementations that work. And that Newton had neurons and synapses.
Failure of a brain project is not necessarily a stall for researching about it. Failure is informative.
2 Likes
Assuming you get your definitions correctly. Once people thought if they figure out how to perform at chess game, whatever plays good chess must be intelligent.
A definition can be used as a test, (which as “chess test” might later prove insufficient). We need more than assumptions about what a target should be, we need good intuition about what the right building blocks are.
And speaking of “building” stuff, I think intelligence in humans at least isn’t built. It grows within the brain in a process.
Once one have built the “brain”, it might be much more difficult to have it built “already intelligent” than to provide the right framework in which may grow by itself (well within a right environment, the point is in humans it doesn’t pops out of the brain instantly and it isn’t carved or sculpted in by parents or “education”.
It’s a developmental process.
For this process to work right emotions might be as vital as currency for an economy to work right. They shape directions in which to grow, otherwise we’ll end up as dysfunctional idiots with big brains.
Specially when the path is inspired by human brain, is very tempting to say oh, we don’t need to do what cerebellum (apparently) does, or what amygdala (apparently) does.
This clip is quite interesting: https://www.youtube.com/watch?v=zeCoO5U9h1Q
1 Like
I was trying to achieve that here. But unfortunately other than on Emotion, I couldn’t get much else.
This is an interesting statement, you mention “that work”, what works? If we don’t know what intelligence is, how do you know the biological implementations work?
That’s conventional thinking. But think this through, If Tesla was out to build Model S right away instead of electrify the MB Smart Car first, then I don’t believe Tesla will still exist today. If we are set out to build the full human brain with full emotion instead of building a brain that satisfies the absolute minimum of what intelligence is, I think we purposely try to reduce the odds to succeed.
But then the definition is not correct then. Isn’t it?
Test is important. Definition also defines the scope. For this complicated subject, you don’t want to go so big that is important to achieve.
Make sure we are on the same page, I NEVER say “human”. That’s way too big. I don’t believe our designer, if there is one, build human on day 1. And that is what I am trying to avoid.
Also, I also did not say building intelligence. My objective is to build a software version of the biological brain (not human) to be able to satisfy the minimum of what intelligence is.
If you still think in the human intelligence perspective, then yes. But I am not. Too big for me. I am trying to do a baby step.
Again, want to emphasize not “human” in my context. Hope that clarifies.
Thanks for the comments. Always makes me think more!
Now that you make your goal clear I can give you a better definition. But first - you will need a body to run because that is THE primary purpose of a brain. Your goal is to learn the behaviors your puppet body needs to survive in it’s environment and enough smarts to select the best behavior in enough of the situations it encounters to be considered a success.
In biology that definition of success is successful reproduction. You may provide a different measure of success.
That is the basic measure of intelligence that nature uses. Anything beyond that is a second order effect.
2 Likes
the fact that you wouldn’t get a consensus on a single definition does not preclude consensus upon the existence of the hard-to-define phenomenon.
in other words…
“I don’t know what intelligence is, but I bet @Bitking is intelligent”
maybe you’d find that to be an interesting statement too 
3 Likes
What would this ‘minimum’ look like in action? (and pardon if you already answered this somewhere).
I’ve asked myself this question too, and my current basic idea is an agent which can move intelligently in a novel environment toward some not-perfectly-specific goal(s).
When I say ‘move intelligently’ I imagine it takes efficient paths, doesn’t run into stuff and adapts immediately to other moving objects – which involves predicting their movements, even for objects its never seen.
When I say ‘not-perfectly-specific goal(s)’ I mean instructions like: ‘Put all fragile objects off to the side near all soft objects and all the hard or heavy objects to another side, and bring back any recent financial documents you find’.
The theme is things which you could reasonably ask any adult person to do, but would be very non-trivial for current A"I" systems.
The OP asked about “minimum intelligence” and building a small brain to support that. The proposals that pass by insects and worms are missing the bit about “minimum” as the referenced critters clearly have a brain and expressed behavior. Go back to Cisek’s phylogenetic chart in the earliest examples to see things on the minimal level, which is what was requested. Reiterating, acquiring (learning or genetic gifts) behavior and expressing it in an appropriate manner to survive to reproductive success is the minimal intelligence required by nature. It may not look like much to clever mammals but that is what brains started out doing.
At this level we avoid scary looking things, look for food and mates, seek water and shelter. Oh, and grooming. That about wraps it up for basic intelligence.
1 Like
Unfortunately I do not find that interesting.
It doesn’t look like a good starting point for discussion. All definitions exist and useful only in some contexts. People study intelligence for absolutely different purposes, so different definitions make seance in every specific case. What is your reason to study it?
Also, as pointed here Intelligence vs Cognition, the term intelligence is too broad and ambiguous, what can be another reason for such a long, but not productive discussion.
So I wonder if I can strip the body down to the minimum and see if we can see the bottom of this.
If there is a very unfortunate person born blind, one ear has no hearing capability, no limbs and cannot move his/her neck, I take the assumption everyone here can agree that person has intelligence still. If so, what criteria we use to determine that person still has intelligence?
Since that person cannot see, the brain does not have opportunity to build location model.
Since that person cannot move, the brain does not have opportunity to build motor model.
Again, I am not trying to build a human brain simulator to achieve a human intelligence. But I want to strip it as down as possible to see if we can find the minimum set of criteria to define what intelligence is.
But agree with you on:
and
Ability to “Learn” and Ability to “Decide” seem to be something I can agree on. If so, we can further break down what elements in intelligence are needed to “learn” and what elements in intelligence are needed to “decide”.
This is the reason:
And to fully clarify the objective, I am NOT referring to human intelligence and human brain. I want to define the most fundamental list of elements for intelligence (again not human intelligence) to avoid being too broad and ambiguous.
I am confident we can define the list if we are on the same page on the objective and we are open mind about it.
Like others in the thread, I also think that this discussion is not going to be productive because we still lack fundamental understanding of the mechanisms of intelligence to precise a broad and general definition of intelligence.
In the short term, it will be easier if you could pick a specific context and purpose to get the definition that fits your needs.
Not sure if it will help, but here is an extract from a presentation I am working on:
1 Like
With such a reason, you can take any definition, and you are good to go. It would be smart to take the simplest one, then you would need just several lines of code to build it 
However, perhaps you need your simulation for something practical - that would be a good reason, and it would let you define your objectives more accurately and get the result easier.
Intelligence isn’t a physical thing, it’s just a way to talk about some emergent phenomena. If you go to the simplest forms of intelligence you won’t get much interesting. If you go to something more complicated, you’ll get some types of intelligences, so you need to specify what kind of it you are interested in.
1 Like
You are speaking of a locked-in person. At this point the expression of intelligence is still an action. In this case, moving the eyes.
This is still a very high level intelligence.
You really should be looking at a worm or similar creature for the most basic intelligence.
When you simplify the body enough the intelligence expressed is not very interesting.
Forum member @Gary_Gaulin has posted a bug creature that has a very basic intelligence. Check out the video in the linked post. That “bug” is driven by a very small chunk of “cortex like” intelligence.
1 Like
And that’s exactly the purpose of my posting. Maybe I did not convey that clearly with my original verbiage:
And for your suggestion
It is exactly what I am trying to achieve in the short term to get everyone’s help to identify the most fundamental and basic elements to define the lowest form of intelligence (not necessarily human) such that it will allow me to start building the most basic simulation of the brain
And after seeing many replies I will add the sixth one in there which is “Decision”.
In terms of your quoted definition of Intelligence, not sure about the “measure” part but the ability to “achieve goals in a wide range of environments” it has some common theme as @Bitking “goal is to learn” and “select the best behaviour”. Does that mean we can at least agree that the ability to “learn” and the ability to “decide” are the elements (not fundamental yet) of intelligence?
Thanks for your reply on this. I can feel that we are getting somewhere.