[keghn]

How spatial navigation correlates with language: 

https://medicalxpress.com/news/2017-11-spatial-language.html#nRlv   

[ranch vermin]

My take on this-> you see letters on a page, just the same way as you see a path connecting together youd traverse,  you can see how two paths are travelled the same way if they are spacially congruent, just the same as "walk" and "move" are similar words to the symbolic meaning when reading the text!   or pancake and flapjack,  same dif.

Or...   the way a picture appears by text in a certain formation, as long as the text is a certain shape.   many things.

[keghn]

 I agree. A advanced brain need a out of body experience to view a the world, high above, as angel would do. Letters and world are
substituted for baths an roads. Out of body flying can happen in a 3d physics simulator in my AGI model.  https://phys.org/news/2018-03-dataset-human-strategies-foreign-networked.html: 

Open dataset of human navigation strategies in foreign networked systems:   
https://phys.org/news/2018-03-dataset-human-strategies-foreign-networked.html   

[LOCKSUIT]

nonono lololol I'm working on a NLP AI that does not need a body noooooooo hahaha lol nooooo

There's no way there is a connection between linguistics and navigation.

Similar brain activity in similar areas is bologna - not proof/evidence. This is not an explanation either. It is abductive (poor) reasoning.

I agree vision recognition is a NLP. But the motor thing um..........no. No body is needed. Bodyless AI for the win :)

[ranch vermin]

Well I guess you treat the objects around you as the literal thing,  and text as a symbolic thing,  but im pretty sure redundancies and synomic groups are detected in the exact same way in both.    dont believe me!!!

But err... maybe you were talking about that text link,  which I dont really agree with either, its just i have my own connection on this topic.

[LOCKSUIT]

That's because "hat" is sound and "hat" is a visual image plus it correlates with being able to pick up a hat.

Both mean hat.

[ranch vermin]

Basicly, but what you said was combining symbology with literality, and this actually requires distinction i think,  but i mean more closer to "hat, and sunglasses, both appear on peoples faces on sunny days" so the computer will generate both interchangably... but only given the context, and its quite explicit when it does it.  because in this example a hat is NOT a sunglass, but given a context they may be considered a synomic group.   but its a poor example because one goes on your nose and one goes on your head,  maybe that would only work in a text based system, not a graphical one...  but then no i doubt it,  a hat and sunglasses are completely different so it would need even more sorrounding context.

But the computer is supposed to generate them itself, and its fully unsupervised.  so its hard to say what groups would actually form as equatably equivicable. it helps the network see through useless difference when finding its solution.   u could also consider it getting large keys less active keys and breaking them up into smaller more active keys (that still dont error fire) as its cells count each others mutual appearances/fires in the system.

[Korrelan]

@ Lock

We exist in 4D space... an AGI also needs to understand and exist in 4D space.

Try describing 1mm with out referencing 2D/ 3D space/ distance (spatiotemporal).

Or… the hat goes on top of the head... describe on top of.

 

[LOCKSUIT]

Even I don't know what "on top" means when I hear it or see it. If I see, it's symbolic. Same for hearing it. Rewards mean more. I'm sure a good description would do.

Doctor who said who said there is no Time? Why there's too many dimensions already! Are you counting right? 3? Really?

That makes 7 million D + 1 for time brah.

[unreality]

AI will need some pattern recognition for it's navigation and tree search thinking functions. Space and time is just info for the visual algorithms. The tree search needs some unit of time, which can at times be non-linear. A detected object could be at a known location traveling at a known velocity. The tree search doesn't have to analyze the object being at a million locations from location A to B. It could have a function that estimates when's the next location it should analyze it, like chess. or Go. You can't analyze every position. Too many.