Yeah, and a mandolin is tuned just lime a violin. And a modern violin even has f holes like a violin.

Re: ice

Color is an interesting domain for studying language and thought.

My pondering were as follows:
From a young age, I have wondered whether what I call "blue" is the same experience as what someone else experiences when they see the same shade of "blue". Looking at the same blueberry, we both might call it "blue". But if I had a chance to experience consciousness from EdLee's perspective, would blueberries be experienced in the same way?

Since color is something we can describe in words, but kind of just experience, it's difficult for me to know what the experience is like for others - maybe it's just the same.

But then the link goes further and uses color as a way of measuring the impact of the language itself.

I don't know if I'm a universalist here or not. I do recall someone mentioning a study where cultures where people had more names for certain colors were better on tests distinguishing those colors.

Could this ability be trained? Or has the language itself impacted the thought?

Interesting stuff.

It'd be cool if I could experience life through another person's eyes for a bit to find out what it's like, but probably I'm stuck with being Kirby

_________________
be immersed

Then there's also the question.. let's say it were possible to experience life as EdLee, and blue looked different through his eyes than it did to me. Would I be able to describe it? At a glance, it seems like it'd be difficult to express.

_________________
be immersed

Was thinking about colors more, and on computers, you can express a wide variety of colors by RGB values, numerically. Mix 57 red + 32 green + 16 blue, and you get a unique color.

So we have a way to express colors uniquely to computers through language. Though, the way a computer may "experience" the color could be different. Well, computers don't have feelings, so it's hard to say.

_________________
be immersed

Hmm, multiple words... The sense of galloping in haengma makes me want to translate it as active shape— the term active ingredient should help English speakers have a feeling for it that is different to the original, but still useful, I think; it implies an active shape is often the key one in a position with passive shapes around it.

You learn concepts, and your brain rewires itself. Knowledge turns to perception as it gets finer. The word used to describe a concept in written language may be nuanced or simple. This could impact how quickly it takes to improve an understanding of a concept into its accurate perception.

I read the first paragraph of this and found it interesting. McCullough effect edit: this effect can last a long time. Rotate the image by 90 degrees and look at it for half the time you did originally to reverse the effect.

The brain seems designed to figure out how different a colour is relative to the surrounding ones.
Hermann grid illusion. Encore. Colourful relativity. Encore.

_________________
On Go proverbs:
"A fine Gotation is a diamond in the hand of a dan of wit and a pebble in the hand of a kyu" —Joseph Raux misquoted.

More on the possibility that the more limited go vocabulary in the west inhibits growth of perception.

Prof. Daniel Levitin, neuroscientist, in This is Your Brain on Music:

Musical training appears to have the effect of shifting some music processing from the right (imagistic) hemisphere to the left (logical) hemisphere, as musicians learn to talk about - and perhaps think about - music using linguistic terms.

Me (??):

Go training appears to have the effect of shifting some go processing from the right (pattern-matching) hemisphere to the left (logical) hemisphere, as players learn to talk about - and perhaps think about - go using technical terms.

However, it may be relevant to note that "Children show less lateralisation of musical operations than do adults, regardless of whether they are musicians or not."

Also possibly relevant: "The most important way that music differs from visual art is that it is manifested over time. As tones unfold sequentially, they lead us - our brains and our minds - to make predictions about what will come next."

Go seems to be an activity "manifested over time" and the mental activities are designed to predict the next move. So what neuroscientists can tell us about the brain and music may have lessons for go.

It's beginning to sound a lot like go is a language with stones as letters and shapes as words. Maybe strength comes by learning not only shapes, but also grammar with which to makes sense between shapes and their order of play.

Pros who learn at about the same time they speak are natives. The average amateur must be a naïve foreigner wandering the streets .

A word in English, Japanese or any other is simply a translation of many positions, but most of it is not accessible on sight, which exactly why talking about go seems beneficial— naming a concept can be just the jolt of realisation one needs. It probably helps to understand the basic concepts, which can be used to learn harder to define ones. How many concepts fall under thickness? Do the Japanese words for thickness's different types allude to them?

Another reason I'm more convinced of this is a tactic to mend abysmal drawing that I've recently applied to go. Sometimes when reading and my threshold runs out, trying to 'hear' the sequence improves depth of search, especially in positions like kos or those with many captures and replacements. It's like hearing a sight . Now which sign points to territory...

_________________
On Go proverbs:
"A fine Gotation is a diamond in the hand of a dan of wit and a pebble in the hand of a kyu" —Joseph Raux misquoted.

Bill Spight and I have been the main proponents of a go "grammar" on L19 (in my case because I'm a linguist - not sure where Bill's starting point was), but Levitin discusses this too in a much more advanced way vis-a-vis music, focusing on the concept of structure. It seems to be applicable to go.

One key aspect, still being hotly debated by neuroscientists, is whether the brain memorises (or sees) things in a relational or absolute way. There seems to be significant evidence for both camps, so presumably the brain can use both methods, but I get the impression that the relationists may have the upper hand. In this view, the brain stores/sees just a few key elements of an object and their relationships to each other, and when the time comes to use that object, it retrieves that information from memory and fills in the missing bits by inference (an interesting reverse use of perception!). It's easy to see how this can apply to reading in go. We learn some key shapes/sujis etc and check the result by some confirmatory reading (as Bill has described). This is tantamount to pruning the tree at a high level, of course, and so is highly efficient for humans, but it also runs the risk of leaving some blind spots - many for amateurs, a tiny handful but not zero for pros.

The absolutist model is where the brain learns every single detail. In go reading this can perhaps be likened to knowing the L shape is dead: you know the shape, you know the starting moves, you know the result, and if you're good you also know the equivalent details for L-shapes with legs and hanes. Perhaps our brains use absolute models like this for very high frequency situations.

I think it's easy to accept that a grammar of language can fit into this schema. We can speed read quite accurately (though never anything like as accurately as the adverts claim) by focusing on the shape of words and even whole phrases, but if a text is written in a dense, tortuous way with lots of rare words we may need to plod through very slowly, word by word, and with individual unusual words letter by letter, trying to work out the etymology.

We also have potential blind spots such as the famous:

PARIS
IN THE
THE SPRING.

On the other hand, we can also achieve some remarkable results. I remember standing next to Kato Masao as he was presented with a hard unknown tsumego problem. He solved it in a second. He was a pro. But I also saw Zoran Mutabzhija walk past a board where a player was pondering a life and death situation on the side and almost without breaking stride he picked up a stone and played the key move. I was obviously agog, but I later caught myself doing something similar. Colleagues in the newsroom would be typing (clickety clackety manual machines in those days) and I (and others) could likewise walk past without breaking stride and spot typos in a whole page of text (yet at other times we could make typos of our own!). I've seen this work in other fields, too, so I'm inclined to believe it's universal. Universal but not necessarily useful: it's like a bot spotting a bad move instantly but it can't tell us how it knows it's bad, no more than I can spot someone else's typo in passing but can't tell you how either.

This seems to support the absolutist view because we were seeing go positions/pages of text for the very first time, and instantly, so no detailed memorisation was possible, but at the same time we must have been relying on a memory bank of something built up by 10,000 hours of work. So at least two skills are at play together, but there has to be a higher one that combines them all in some still unknown way. Levitin refers to a section of the brain that appears to be concerned with "structure." Especially as he also refers to Chomsky, I take this to be similar to grammar or syntax.

Making big assumptions that I've vaguely understood all this stuff about the human brain, I have to say that I can't believe that current AI bots operate in anything like the same way, and the use of terms like "neural network" is actually fundamentally deceptive. At any rate, they don't seem to have the slightest on higher structure.

But even assuming the bots can handle grammar perfectly, if they producing real language and not go moves, I think they would be producing sentences like this:

Twas brillyg, and the slythy toves
Did gyre and gymble in the wabe:
All mimsy were the borogoves;
And the mome raths outgrabe.

In other words, there's a level even higher than grammar.

What is grammar for go terminologies? A metaphor for theory applying go terms well? A linguistic aspect of how to formulate phrases or sentences with go terms? A way of hiding a preference for subconscious thinking instead of explicit go theory? Something else?

Richness of vocabulary has been mentioned as an advantage but there is no linear relation. It can also be an advantage to unlearn or ignore go terms with no or little value and acquire a more relevant vocabulary. I have done this successfully. It is irrelevant whether something is correctly called a dog shape but relevant whether its connectivity is assessed is correctly.

I beg to differ. A beginner looking at a complex go position sees a writhing mass of stones. Without help he can't sort out what is going on. Picking out a common shape and giving it a name to help him identify it so that it can be talked about and explained is vital. The only part that's not relevant is whether you call it a dog shape, a sake bottle shape or a Gefurtel.

It's just like astronomy. The ancients saw the Little Dipper (but called it very different names in each culture) and were able to use its identifiable shape to navigate across the oceans. That's relevance for you. They didn't need to know the apparent magnitude or spectral type or number of each star, and even if they did know that, it wouldn't have helped them one little bit. In fact, with just that they would have got lost.

Dealing with humans as if they were machines is stupid. However, humans dealing with machines as if they are machines is sensible. Just don't mix the two up.

Had there been a linguistics major when I was an undergraduate I may well have majored in it. As it was, I managed to study stratificational linguistics under Lamb. Stratificational grammar is quite general, and can be used as a fairly low level computer language, but one that produces structured programs. When I learned to program I used it rather than flow charts. I have also used it to analyze music. It can also be used to analyze go.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Can you clarify what it is that the human brain is capable of, which modern neural networks are not? You referred to a particular structure. Is that what you mean? Why do you think today's neural networks incapable of achieving this structure?

_________________
be immersed

I think what you are getting at is that using a model to simplify reality can be useful, and language provides such a model.

I agree with this, though, I don't think that it's the only type of model that exists. My mind can chunk and recognize a shape for which I have a particular feeling, even if I don't have a name to express that chunk of information.

Having a word, however, can definitely help communicate that feeling.

_________________
be immersed

I think that John may be mistaken here about structure. Isn't the superior ability at go of modern neural nets based upon a deeper structure than those of 25 years ago? (OC, actual neurons in the brain are relatively sparsely connected, so the structures are different.)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I'm not an expert on neural nets, but "deep learning" does imply a deeper structure, I.e. more intermediate layers in the network.

There have probably been lots of improvements to the algorithms behind neural networks, but my opinion is that improved computer hardware has also made training neural nets more feasible.

_________________
be immersed

I have no doubt I'm talking through my backside - but that's form of reverse engineering and it helps me understand.

I talked about "higher structure" (certainly not deeper) but put that badly. The portion of Levitin's work that I leapt on was something higher that attended to structure.

On the principle that when you are in a hole stop digging, it will best if I just quote Levitin:



A little later he says:



It is this chronological aspect that makes me wonder whether music and go may significantly share some brain functions.

As far as I know, neural networks don't have the concept of left and right brain. I'm not sure what this implies about the learning process.

_________________
be immersed

Left and right brain are structure at a wildly different scale than our neural network. A brain has around 100 billions neurons and hundred trillions of synapses.

It's like comparing the urbanism and structure of the whole US east coast metropolis to the urbanism and structure of a small village.

There's different scale, but the left/right-ness is also a concept that doesn't exist in neural networks, right?

The human brain has a lot of different areas performing different functions, but a neural network is basically a graph of connected nodes. The intermediate layers are connected to different components, but no explicit functions are designed to match the human brain afaik.

For example, a neural network doesn't have a portion of the graph set aside to simulate the frontal lobe vs the temporal lobe.

Intermediate layers may have different weights after being trained, but this is implicitly learned as opposed to being designed to be similar to sections of the human brain.

_________________
be immersed