(Untitled)

[shinmatsuz]

Has anybody ever tried to measure the percentage of the human brain devoted to the grounding problem? That might give us some idea of how hard it actually is. A quick Google search turns up nothing, but I have a whole bunch of cognitive science people on my flist.

Comments

[endril]

The grounding problem only exists in a purely virtual domain, I think? The brain is not its own universe, unlike computer programs. "Grounding information" is inherited from the way it is constructed, and from the senses.

[shinmatsuz]

Well the way I was thinking of it was the following: There are structures in the brain that deal with such abstract things as "cats". There are structures in the brain that deal with non-abstract things such as "the input from my eyeballs". In order to use the higher-level structures, the brain has to "ground" it somehow. We also know that certain areas of the brain are associated with various senses. So if we take the "sight" region of the brain and subtract the neurons associated with higher-level constructs and the neurons associated with direct input, we have the "grounding problem area".

One problem with this way of going about things is that we can develop new hierarchical ideas. So if we know cats and we know dogs and then we see a cat-dog-thing, we can still categorize it as being in a third, higher category of "similar to both cat and dog". The terrible secret of the grounding problem: There are far more concepts than we have words for, many of which are so basic that we have trouble thinking of them.

[endril]

The entire cortex uses the same pattern of organization, arranged so that the lowest levels gather signals from the senses, and the higher levels get signals from the lower. Essentially the same exact memory->prediction framework is at work at the "what shape am I looking at" and "the difference between cats and dogs" levels.
That said, I believe approximately 50% of the cortex does really non-mental-sounding stuff like guesstimating angles or telling noises apart.

And yes, you're right, there's a large space of precepts that our brains use for organization that we don't have much conscious awareness of, at least partly because it's not something we ever communicate.

[zosimos]

Not coming at this from a direct Cog Sci issue but in information science "cats" is what gets called a category. Categories are not exclusive though, so cat-dog is not necessarily in some larger category but can be on the edges of both the cat and dog category. Usually something is sort of the platonic idea of cat at the center, and things that are less like it are lacking. See tigers and lions being "big cats", they are cats but bigger than most modern western people are used to so they are apart from the basic form of cat somewhat.

There's also this little tidbit which seems potentially relevant to the issue. Sadly the link to the abstract is not working right now, so you get the less scientific link.