Human decision-making algorithm: possible?

[Mellinia]

Humans typically brand themselves as illogical and so is their decisions, but I have a theory that could well explain our algorithms, and enable a breakthrough in the creation of human-like computer programmes. However, what I want to know is, can i post a theory paper without statistics, as my theory might prove near impossible to prove?

[ydoaPs]

Humans typically brand themselves as illogical and so is their decisions, but I have a theory that could well explain our algorithms, and enable a breakthrough in the creation of human-like computer programmes. However, what I want to know is, can i post a theory paper without statistics, as my theory might prove near impossible to prove?

Does your idea make predictions or not? Go ahead and post it.

 

I suggest reading On Intelligence by Jeff Hawkins if you're interested in truly intelligent machines.

Edited October 27, 2010 by ydoaPs

[Mellinia]

it can...provided that it has been provided with enough data..

[Mr Skeptic]

Just keep in mind we'll do our best to try to poke holes in it. But don't worry, if it is bulletproof we won't be able to.

[TonyMcC]

I sometimes wonder if "Swansont" is really a computer running an artificial intelligence program. I mean this as a compliment as he seems to be able to give precise answers to so many queries listing the relevant laws and theories. On the balance of probabilities he is probably a very intelligent human - but I do sometimes wonder! Please Swansont do not be offended - take this as a compliment.

[Mellinia]

Just keep in mind we'll do our best to try to poke holes in it. But don't worry, if it is bulletproof we won't be able to.

 

 

Ok, I'll try to skim it down to the basic rules of the algorithm:

 

 

1. The human decision-making algorithm is based on an input-process-output basis.

 

2. Input includes present-time data from the nevironment and stored memories. Output is the decision.

 

3. The algorithm is divided into four parts, the first is primary, following second and third is auxillary, fourth is special.

 

4. First, survival. This is divided into Need to Live (Maintaining continuity) and Pride (Respect for the ability to live). They are further divided into "Self", "Other", and "Human".

 

5. Next, material. This part controls the value of each category.

 

6. The value of self pride is always in a equilibrium. When there is a heightening in self pride, Material will adjust to lower the pride with the same degree. Think of it as a straight line; when there is heightening (someone praises you) a wave is formed. After the peak, the wave will proceed downwards to return to its original line, and vice versa for lowering of pride.

 

7. Emotional is next. It directs Material to the required categories by singling out the ones that are needed to process a decision.

 

8. Last is Emotions. Emotions heighten or lower pride based on their categories: Positive (Anger, Joy, Proud), and Negative (Frustation, Grief, Regret). Positive heightens; Negtive lowers.

 

9. All values are determined in numbers.

 

Loopholes...I'll try to cover them.

[Mr Skeptic]

Ah, I'm afraid your idea suffers from one of the most fatal of flaws -- you don't actually have anything. OK, so you do have a vague outline. But to actually make this work, you need to eventually put it into computer code. You'll have to define the terms you use since the computer knows only 1 and 0, and not just define them like a dictionary but in terms a computer can understand. It will have to end up as 1's, 0's, and the NAND operator (or equivalent), no more, no less. Same with the instructions.

 

Because of the vagueness it is impossible to tell whether any within the set of all possible interpretations of what you have said would actually work. I'd guess that several would work but that you don't know any of them.

[Mellinia]

Ah, I'm afraid your idea suffers from one of the most fatal of flaws -- you don't actually have anything. OK, so you do have a vague outline. But to actually make this work, you need to eventually put it into computer code. You'll have to define the terms you use since the computer knows only 1 and 0, and not just define them like a dictionary but in terms a computer can understand. It will have to end up as 1's, 0's, and the NAND operator (or equivalent), no more, no less. Same with the instructions.

 

Because of the vagueness it is impossible to tell whether any within the set of all possible interpretations of what you have said would actually work. I'd guess that several would work but that you don't know any of them.

 

Yeah...is there any way to correct this? Possibly transform it into a logical processor? I tried programming myself...but i lack much technical knowledge...Maybe the "if" program might work....

[Mr Skeptic]

Not really, that's why I said it is one of the most fatal flaws. Just to give a nicely discouraging example, people have been wanting to fly for thousands of years before someone up and invented an airplane. And then ~60 years later we landed on the moon. The first step is always the hardest, and I think that's roughly where you are.

 

There's various programming languages and they have a larger vocabulary than just NAND, they have addition, multiplication, and many other such things. They can then be turned into computer code. However, they don't have things like "emotion" in their vocabulary, and you'd have to define it with the vocabulary it has. If you could do that one bit you'd probably get a Nobel prize for it.

[Mellinia]

okay...that's difficult...hmm...what are Natural languages anyway?

 

okay...that's difficult...hmm...what are Natural computer languages anyway?

[imatfaal]

From a different point of view; the problems Mr S highlighted are in my opinion pretty insurmountable - and they are not alone. I am pretty sure that every decision I take has a multitude of factors - most of them unknown to my conscious mind. How do you plan to have your algorithm take account of the unconscious mind? The only way that you could do it was to have a decision process that had as its initial variable every sensation and every thought that the subject had ever had - and that is clearly not going to happen. I might posit that any machine capable of storing all the memories, sensations, thoughts of a human being and the myriad cross-relations and implications between them would be as sentient as the human being itself - and this machine might well tell you that it has more important things to do than predict another sapient machines decision.

[Mellinia]

Our brain seem fully capable to do that. However, I don't think every thought and sensation is needed to process a decision. The "rules of algorithm" is workings in the subconscious mind. You couldn't the workings and algorithms of a computer game (Say, warcraft), could you? Initially, I'd actually wanted to show that humans are (possibly) logical, as they followed rules when they think. However, I do agree with you on the part of "every sensation and every thought that the subject had ever had" is needed...possibly to do a reverse proof, and "this machine might well tell you that it has more important things to do than predict another sapient machines decision. " I really have too much time on my hands, don't I?

[TonyMcC]

I think your algorithm would need a measure of randomness about it. For instance I usually travel from A to B by a particular route. This route has much to recommend it such as shortest distance good roads etc.. However for reasons unclear to me, although I may have set out intending to follow my usual route, I suddenly decide to vary it and follow a diversion. If my passenger says "why are we going this way?" I have to reply "I don't really know - perhaps I just fancied a change".

[Mellinia]

I don't think it's simply random. The rules only apply subconsciously, and different data input means different decision output. Besides, if it's random, there's no rules to apply! Oh, right.

By the way, i wonder if i could predict what Osama Bin Laden will do next?

[TonyMcC]

It seems the logical conclusion of your aim would result in no free will. If every human outcome could be forecast from an examination of all data inputs what should we do with a murderer who followed your algorithm and killed someone? You might say that giving a numeric value based on emotion would cover that. However the murderer's emotional makeup would be beyond his control. If the murderer was sane, however angry or emotional he felt, he should be able to restrain himself. Perhaps a numeric value should be used for degree of insanity. After all IMO none of us could call ourselves completely sane all of the time! For some reason we all do daft things now and again - I do anyway!

Edited November 4, 2010 by TonyMcC

[Mellinia]

Well, naturally, that's why we have laws. The decision made by the murderer is based on previous data inputs, such as the victim was causing him immense lowering of pride or what so ever. In any case, my algorithm only apply to sane humans, psychos aside. Someone would kill because he thinks it's fun is categorised as "half sane and half insane" We must evaluate it from his viewpoint. Laws cause us to think twice before we do something bad because punishment will follow, along with lowering of pride, and the need to live will overcome this.

Free will? Unless you can control everything that happens around you. That's free will of controlling data input.

[TonyMcC]

I think the best outcome you could expect from your algorithm would be an answer to the question "what would be a sensible decision based on all available inputs?" This may be quite different from what decision would a human being make?". If you succeeded in only making a sensible decision your algorithm would have quite a lot to commend it. I remain convinced that two humans provided with identical inputs may well make different decisions. Anyway - best of luck. By the way programming languages mean you don't have to get right down to "1"s and "0"s to get a computer to do what you want.

Edited November 5, 2010 by TonyMcC

[Mellinia]

It's all relative. ah well, thanks for the luck.