Perceptual learning

daal · #1

Here is another interesting article from the New York Times that might have some relevance to improving one's go skills. It is about perceptual learning, which involves learning to quickly recognize what sort of problem one is facing. A group of cognitive scientists postulate that by focusing the brain's pattern recognition ability, it is possible to gain a deeper grasp on the underlying principles. Basically one practices quickly matching one of several questions to an answer or visa versa, and apparently with eyebrow raising results.

Numsgil · #2

I often study go problems (more or less) like this. I have SmartGo on the iPhone and it isn't too hard to just throw stones around until you get the right answer and then move on to the next question.

It doesn't train reading at all; it's more about identifying shape points and tesujis and the like and just internalizing it all into your gut. I have heard that this is how young inseis study.

EdLee · #3

Thanks, daal.

ethanb · #4

So in other words maybe an interesting thing to do would be to look at unlabeled life and death problems and -- without solving them first -- quickly jot down things like "under the stones" or "nakade" or "seki" or "ko" or "death in the hane." Then go back, solve them, and see if the first instinct is right?

I'm not sure if my process is good - it seems more likely to be a way to evaluate your current experience with those shapes than it is to train yourself to evaluate what to read at a glance. Can anybody find a better?

SoDesuNe · #5

Hm, what about determine the first move (the vital point in the specific situation) at glance? Hane, Pivot point, cut etc.?
If you get really good at this, I assume you can considerably shorten the reading time in a real game.

Bill Spight · #6

Verrry interesting! Thanks.

Differences are significant, so I thought that comparison problems might be good for perceptual learning. For instance:

Which response?

Black approached the top left corner with :b1:

. White replied with one of these moves. Quickly guess which one it was.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . O . . | $$ | . . . O . X . . X O O . . . O . O X . | $$ | . . . , . 2 . . X X . . . . . O X . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . O . . . . . . . . . . . . O X . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . 1 . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . X . . . . . . . . . . . . . . . | $$ | . . . X O . . . . . . . . . . . . . . | $$ | . . . . . . . . O . . . . . . O . . . | $$ | . . X , . O . . . , . . . . . , . . . | $$ | . . . X O . . . X . . X . . . . . . . | $$ | . . X O . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . O . . | $$ | . . . O . X . . X O O . . . O . O X . | $$ | . . . , . . . . X X . . . . . O X . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . O . 2 . . . . . . . . . . O X . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . 1 . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . X . . . . . . . . . . . . . . . | $$ | . . . X O . . . . . . . . . . . . . . | $$ | . . . . . . . . O . . . . . . O . . . | $$ | . . X , . O . . . , . . . . . , . . . | $$ | . . . X O . . . X . . X . . . . . . . | $$ | . . X O . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Often you see next move problems with 5 choices, which are graded from 2 to 10. Perhaps giving only two choices would work better for perceptual learning.

Bill Spight · #7

Let's try a few more.

Black has just invaded. White made one of these replies. Quickly guess which one it was.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . X . . . . . . | $$ | . X O , X . . . . , . . . . . X . . . | $$ | . O X X . . . . . . . . . . . . . . . | $$ | O . O . . . . . . . . . . . . . O . . | $$ | . O . . O . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . 1 . 2 . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O . O . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X X . . . . . , . . . . . , . . . | $$ | . . . . . . X . . . . . . . . O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . X . . . . . . | $$ | . X O , X . . . . , . . . . . X . . . | $$ | . O X X . . . . . . . . . . . . . . . | $$ | O . O . . . . . . . . . . . . . O . . | $$ | . O . . O . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . 1 . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O . O . . . . . . . . . . . . . . | $$ | . . . . . . 2 . . . . . . . . . . . . | $$ | . . X X . . . . . , . . . . . , . . . | $$ | . . . . . . X . . . . . . . . O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Bill Spight · #8

Black has just made a peep. Quickly guess which was White's reply.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . X . . . . . . . . . . . . | $$ | . . . X O X . X . . . . . . . . . . . | $$ | . . O X X O X X . . . . X . . . X . . | $$ | . . . X O O O . . , . . . . . X . . . | $$ | . . O O 2 . . . . . . . . . . . . . . | $$ | . . . . 1 . . . . . . . . . . . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . . . . . . . O . . . | $$ | . . . O X . . . . , . . . . X O O O . | $$ | . . O X . . . . . . . . . . . X O X . | $$ | . . O X . . . . O . . . . . . X X . . | $$ | . O X X . . . . . . X . . . . . X . . | $$ | . . O X . . . . O . . . . O . . . . . | $$ | . O . . . . . . O X . . X O . X . . . | $$ | . . X X . . O . . X . . . O . , X . . | $$ | . . . . . X X O O . . X . . . O X . . | $$ | . . . . . . . . . . . . O . . . O O . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . X . . . . . . . . . . . . | $$ | . . . X O X . X . . . . . . . . . . . | $$ | . . O X X O X X . . . . X . . . X . . | $$ | . . . X O O O . . , . . . . . X . . . | $$ | . . O O . . . . . . . . . . . . . . . | $$ | . . . . 1 . . . . . . . . . . . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O 2 . . . . . . . . . . O . . . | $$ | . . . O X . . . . , . . . . X O O O . | $$ | . . O X . . . . . . . . . . . X O X . | $$ | . . O X . . . . O . . . . . . X X . . | $$ | . O X X . . . . . . X . . . . . X . . | $$ | . . O X . . . . O . . . . O . . . . . | $$ | . O . . . . . . O X . . X O . X . . . | $$ | . . X X . . O . . X . . . O . , X . . | $$ | . . . . . X X O O . . X . . . O X . . | $$ | . . . . . . . . . . . . O . . . O O . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Bill Spight · #9

One more.

Black has just played an attachment at :b1:

. Quickly guess which reply White made.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . X . X . . X . O . . . . . . | $$ | . . . , . . . . . , . . . . X , X . . | $$ | . . . O 2 . . . . . O . O . . . . . . | $$ | . . . 1 . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . O . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O . . . . . . . . . . . X . X . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . O . . . O . . | $$ | . . X . . . . . . . O O . . . O . . . | $$ | . . . . . . . . . . O X . . X . . . . | $$ | . . . . . . . . . O X . X . . O . . . | $$ | . . . . . . . . O . X . . . X O . . . | $$ | . . X , . . X . . , . . X . X , . . . | $$ | . . . . X . X O O . X X O O X O . . . | $$ | . . . . . . . . . . . . . . O . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . X . X . . X . O . . . . . . | $$ | . . . , . . . . . , . . . . X , X . . | $$ | . . . O . . . . . . O . O . . . . . . | $$ | . . 2 1 . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . O . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O . . . . . . . . . . . X . X . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . O . . . O . . | $$ | . . X . . . . . . . O O . . . O . . . | $$ | . . . . . . . . . . O X . . X . . . . | $$ | . . . . . . . . . O X . X . . O . . . | $$ | . . . . . . . . O . X . . . X O . . . | $$ | . . X , . . X . . , . . X . X , . . . | $$ | . . . . X . X O O . X X O O X O . . . | $$ | . . . . . . . . . . . . . . O . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Oroth · **#10**

I found these problems interesting, not least because my first instinct is often wrong. What I often wonder about these choose-the-next-move multiple choice questions is the difference in playability between the answers. Is it like night and day, so the wrong answer is incontrovertably wrong, or is the 'wrong' answer within the realms of playable, so that a strong player might prefer it as a stylistic choice. For things like opening problem books are the answers so clear cut that all pros agree, without fail, on the correct answer?

daniel_the_smith · **#11**

Interesting problems.

Extra interesting to me because I could see automatically generating similar ones for my website. As a source of wrong moves I could either find a move a pro played in a different game, find a move an amateur played in an amateur game, or I could start (anonymously, of course) recording which incorrect moves my members are making in joseki positions... Hm...

Bill Spight · **#12**

Perhaps these are too easy, in the sense that you might do well by asking which one would I play and then picking the other one.

Alternatively, you can pick the more aggressive reply.

Bill Spight · **#13**

Oroth wrote:

I found these problems interesting, not least because my first instinct is often wrong. What I often wonder about these choose-the-next-move multiple choice questions is the difference in playability between the answers. Is it like night and day, so the wrong answer is incontrovertably wrong, or is the 'wrong' answer within the realms of playable, so that a strong player might prefer it as a stylistic choice.

The problems that give 5 choices typically have one incorrect answer that is barely playable, that gets you 2 points, and one that is playable, but (maybe) not as good as the "correct" answer, which gets you 8 points. The authors do not believe that it is merely a question of style.

Quote:

For things like opening problem books are the answers so clear cut that all pros agree, without fail, on the correct answer?

Pretty much. That does not mean that pros are right. For instance, the Kobayashi Fuseki would have been considered incorrect according to the opening books of the mid-20th century. As would the Chinese Fuseki. OTOH, we have made progress in the opening over time. There are fads, but usually opening ideas are rejected because they prove to be inferior.

BTW, Honinbo Dosaku, the hero of these problems, often played the Mini-Chinese as White in handicap games.

Back in the 17th century.

amnal · **#14**

Bill Spight wrote:

Perhaps these are too easy, in the sense that you might do well by asking which one would I play and then picking the other one.

Alternatively, you can pick the more aggressive reply.

Well...I got them all right by thinking 'if I were looking at a pro game, where would I expect the pro to play'. Perhaps this tells me something about how to improve my own games

Bill Spight · **#15**

daniel_the_smith wrote:

Interesting problems.

Extra interesting to me because I could see automatically generating similar ones for my website. As a source of wrong moves I could either find a move a pro played in a different game, find a move an amateur played in an amateur game, or I could start (anonymously, of course) recording which incorrect moves my members are making in joseki positions... Hm...

Maeda, God of Tsumego, used to have a column in one of the go magazines aimed at kyu players ("Igo Club", I think). In one he criticized the multiple choice problems (5 choices) that the magazine had every month for readers to send in answers and get ranked. (I sent in my answers one month and got rated as a 2 kyu, which is how I was playing, anyway.) He thought that, unless the answers were demonstrably correct, as with the tsumego problems, they were open to question. He mentioned one joseki that had been popular a couple of decades before, that he had always thought was inferior. He had seen pro fashion change so that his choice of plays had become the popular one. He thought that if one of the "incorrect" plays was chosen by at least 25% of the amateur (!) responders, you couldn't really say that it was wrong.

For these problems I looked at Dosaku's games, because I figured that the moves would be close to perfect. I also chose plays where I could see a plausible choice in the vicinity of the pro play that I thought was inferior. Not that a distant play would not fill the bill, but it would raise more questions. Also, I had in mind the perceptual notion of just noticeable differences. Local plays seem to meet that criterion better than distant plays. (Not that there is not a strong perceptual element in tenuki problems, though.)

Bill Spight · **#16**

amnal wrote:

Bill Spight wrote:

Perhaps these are too easy, in the sense that you might do well by asking which one would I play and then picking the other one.

Alternatively, you can pick the more aggressive reply.

Well...I got them all right by thinking 'if I were looking at a pro game, where would I expect the pro to play'. Perhaps this tells me something about how to improve my own games

I used to play regularly with someone (Mr. X) who was a much better tactician than I was. While considering my plays I would sometimes tease him by saying, "Where would X play?" Besides, I really did ask myself that. It helped.

I really do think that it helps to ask where a pro would play, even a specific pro. Once in a post-mortem somebody asked me why I made a certain play. I truthfully replied, "That's where I thought that Dosaku would play."

Now if I could only channel Cho Chikun.

Kirby · **#17**

Bill Spight wrote:

Perhaps these are too easy, in the sense that you might do well by asking which one would I play and then picking the other one.

Alternatively, you can pick the more aggressive reply.

This is exactly what happened with me. The answers you gave are all the opposite of the one I would have chosen.

entropi · **#18**

Numsgil wrote:

I often study go problems (more or less) like this. I have SmartGo on the iPhone and it isn't too hard to just throw stones around until you get the right answer and then move on to the next question.

It doesn't train reading at all; it's more about identifying shape points and tesujis and the like and just internalizing it all into your gut. I have heard that this is how young inseis study.

Very nice and interesting method of learning. Not only go but for anything.

But when you use it for go, my impression is that it can be more useful if you apply it to "normal" positions. What I mean by "normal" is positions that are likely to occur in real games, i.e. positions or sequences that you wouldn't say "wow" if you see it in a game.

But problems tend to have more "unusual" answers in order to force you to read accurately. My gut says that such a perceptual learning will be less useful for learning unusual tricky shapes from problems.

Nevertheless this is just my first impression of the idea. Maybe it's useful in any case. Would be nice hear results from the ones who experiment with it.

daal · **#19**

I'm having trouble constructing a worthwhile exercise because my own judgement is too inaccurate, but one idea would be to quickly recognize a weak group (i.e., when shown a few groups, correctly choose which one needs help first), and another would be to help teach positional judgement by showing several results of a clash and asking which one favors one player. Any other ideas?

entropi · **#20**

daal wrote:

I'm having trouble constructing a worthwhile exercise because my own judgement is too inaccurate, but one idea would be to quickly recognize a weak group (i.e., when shown a few groups, correctly choose which one needs help first), and another would be to help teach positional judgement by showing several results of a clash and asking which one favors one player. Any other ideas?

- Joseki selection
- Should you cut the opponets extension in a given position?
- Should you try to save the group or sacrifice it?

Perceptual learning

Who is online