- Blind mode tutorial
lichess.org
Donate

Why Are People Good At Chess?

Great article, but think the conclusion misses the point of 'Chunking Theory', which till today is one of the best theories of memory, maybe even consciousness (predictive processing / Active Inference).

Almost all introductory psychology books include Chunking Theory when describing short-term, and long-term memory (Miller's experiment). Miller's experiment had nothing to do with chess, and till today for psychology, Miller remains much more important than deGroot, Chase, Simon,..., who are actually more useful for computer science than psychology. (Note Simon won the Nobel Prize for economics, and Chess Chunking and Chase Simon 1973 might not even be in his top 10 accomplishments.

But within Psychology, Miller stands above any of the chess researchers, and Chunking stands alone just on the reputation of Miller, even had it never been applied to chess studies.

From that Chunking Theory is useful for a 'general theory of expertise', where there is a universal path in all fields from Novice to Master, that can be explained by Chunking Theory.

Chess is only useful as a field of experiment, as other simple games like Tower of Honoi, chess just being the most complicated (requiring most amount of memory), and the rating system to differentiate skill levels with clear multiple skill levels from chess novice to master.

But other than being a subject of study for expertise and memory, Chunking has nothing to do with chess, and is a general theory that applies to all mental activity.

Also note, after the mid 1990s, chess is hardly an interest of pyshological study at all, and now Chunking Theory is mostly used by trainers for Sports and Music, and sometimes Math. And for competence training for professionals like pilots and medical doctors.

So an accurate history of chunking and chess, must include why chess was the main area of study (drosophila of psychology and computer science research from mid 1940s till 1980s), then once chunking theory was developed to be useful for sports training and music, chess has been diminished to the annuls of history, as now chunking demonstates the limited faculty of chess skill.

Great article, but think the conclusion misses the point of 'Chunking Theory', which till today is one of the best theories of memory, maybe even consciousness (predictive processing / Active Inference). Almost all introductory psychology books include Chunking Theory when describing short-term, and long-term memory (Miller's experiment). Miller's experiment had nothing to do with chess, and till today for psychology, Miller remains much more important than deGroot, Chase, Simon,..., who are actually more useful for computer science than psychology. (Note Simon won the Nobel Prize for economics, and Chess Chunking and Chase Simon 1973 might not even be in his top 10 accomplishments. But within Psychology, Miller stands above any of the chess researchers, and Chunking stands alone just on the reputation of Miller, even had it never been applied to chess studies. From that Chunking Theory is useful for a 'general theory of expertise', where there is a universal path in all fields from Novice to Master, that can be explained by Chunking Theory. Chess is only useful as a field of experiment, as other simple games like Tower of Honoi, chess just being the most complicated (requiring most amount of memory), and the rating system to differentiate skill levels with clear multiple skill levels from chess novice to master. But other than being a subject of study for expertise and memory, Chunking has nothing to do with chess, and is a general theory that applies to all mental activity. Also note, after the mid 1990s, chess is hardly an interest of pyshological study at all, and now Chunking Theory is mostly used by trainers for Sports and Music, and sometimes Math. And for competence training for professionals like pilots and medical doctors. So an accurate history of chunking and chess, must include why chess was the main area of study (drosophila of psychology and computer science research from mid 1940s till 1980s), then once chunking theory was developed to be useful for sports training and music, chess has been diminished to the annuls of history, as now chunking demonstates the limited faculty of chess skill.

If anyone is interested in Chunking Theory, I would actually look into the literature regarding 'speed reading', and practice speed reading, which is almost exclusively based on chunking theory, and unlike chess, speed reading is actually one of the most useful skills one can master.

In the last few decades, there are now complete linguistic theories of language acquisition based on Chunking Theory that is cutting edge, very little interesting research regarding chess and chunking in last 20 years.

Also, Chunking Theory has theoretical implications for the collapse of the wave function and theories of quantum consciousness, where forming a chunk requires a 'collapse of the wave function'.

If anyone is interested in Chunking Theory, I would actually look into the literature regarding 'speed reading', and practice speed reading, which is almost exclusively based on chunking theory, and unlike chess, speed reading is actually one of the most useful skills one can master. In the last few decades, there are now complete linguistic theories of language acquisition based on Chunking Theory that is cutting edge, very little interesting research regarding chess and chunking in last 20 years. Also, Chunking Theory has theoretical implications for the collapse of the wave function and theories of quantum consciousness, where forming a chunk requires a 'collapse of the wave function'.

Very nice review of the literature. This is how I look at it. Piano players who memorize lengthy compositions cannot think about every note or even phrasing. They must chunk things together which takes 10s of thousands of hours of practice to learn how to do that. A music composer is a different animal. They must be familiar with all of the phrasings and the notes and how they can be combined meaningfully to invoke emotion. Chess players can memorize long variations but ultimately they have to compose. That is, they have to chunk but also use formal operations thinking to derive the correct move. In particular, what can be chunked in chess is tactics and pattern recognition as it applies to positional chess. However, it requires formal operations to put all of this together. That is, recognition is all that is required of piano players but composers and chess players must learn how to put it all together. This is what is so disappointing about most of the research in this area. One should be designing experiments to test how much is chunking and how much is formal operations thinking in real life situations. Early experiments focused on looking at a position for a very short period of time as opposed to giving individuals time to show that they can improve their understanding of the position by having more time to think. Kotov's Think Like a Grandmaster is an excellent example of this. Kotov trained to think about chess positions in a certain way. Does training ppl in his method really work? Modern trainers focus purely on tactics but then you have players like Magnus who take advantage of that by putting them in strategic positions that they don't understand. I believe studies that induce blundering is a good way to understand thinking processes because it shows how we can trip ourselves up. Is it because we see a tactic and play a move before looking a little deeper and finding the best tactic? Or is it an overall understanding of the position that avoids blunders? I find all of this very interesting. Chess can teach us how we learn, understand and evaluate.

Very nice review of the literature. This is how I look at it. Piano players who memorize lengthy compositions cannot think about every note or even phrasing. They must chunk things together which takes 10s of thousands of hours of practice to learn how to do that. A music composer is a different animal. They must be familiar with all of the phrasings and the notes and how they can be combined meaningfully to invoke emotion. Chess players can memorize long variations but ultimately they have to compose. That is, they have to chunk but also use formal operations thinking to derive the correct move. In particular, what can be chunked in chess is tactics and pattern recognition as it applies to positional chess. However, it requires formal operations to put all of this together. That is, recognition is all that is required of piano players but composers and chess players must learn how to put it all together. This is what is so disappointing about most of the research in this area. One should be designing experiments to test how much is chunking and how much is formal operations thinking in real life situations. Early experiments focused on looking at a position for a very short period of time as opposed to giving individuals time to show that they can improve their understanding of the position by having more time to think. Kotov's Think Like a Grandmaster is an excellent example of this. Kotov trained to think about chess positions in a certain way. Does training ppl in his method really work? Modern trainers focus purely on tactics but then you have players like Magnus who take advantage of that by putting them in strategic positions that they don't understand. I believe studies that induce blundering is a good way to understand thinking processes because it shows how we can trip ourselves up. Is it because we see a tactic and play a move before looking a little deeper and finding the best tactic? Or is it an overall understanding of the position that avoids blunders? I find all of this very interesting. Chess can teach us how we learn, understand and evaluate.

@DIAChessClubStudies said in #3:

Great article, but think the conclusion misses the point of 'Chunking Theory', which till today is one of the best theories of memory, maybe even consciousness (predictive processing / Active Inference).

The article is about chunking theory in relation to explaining chess skill. I don't dispute that chunks are an important concept for other areas. But I don't agree that chunking theory is what explains chess skill predominantly.

So an accurate history of chunking and chess, must include why chess was the main area of study (drosophila of psychology and computer science research from mid 1940s till 1980s), then once chunking theory was developed to be useful for sports training and music, chess has been diminished to the annuls of history, as now chunking demonstates the limited faculty of chess skill.

But how does chunking demonstrate the 'limited faculty of chess skill'? The 'Criticism' and 'Summary' section of the blog shows plenty of observations that chunking theory doesn't explain in the context of chess.

@DIAChessClubStudies said in #3: > Great article, but think the conclusion misses the point of 'Chunking Theory', which till today is one of the best theories of memory, maybe even consciousness (predictive processing / Active Inference). The article is about chunking theory in relation to explaining chess skill. I don't dispute that chunks are an important concept for other areas. But I don't agree that chunking theory is what explains chess skill predominantly. > So an accurate history of chunking and chess, must include why chess was the main area of study (drosophila of psychology and computer science research from mid 1940s till 1980s), then once chunking theory was developed to be useful for sports training and music, chess has been diminished to the annuls of history, as now chunking demonstates the limited faculty of chess skill. But how does chunking demonstrate the 'limited faculty of chess skill'? The 'Criticism' and 'Summary' section of the blog shows plenty of observations that chunking theory doesn't explain in the context of chess.

@defense57 said in #5:

In particular, what can be chunked in chess is tactics and pattern recognition as it applies to positional chess. However, it requires formal operations to put all of this together. That is, recognition is all that is required of piano players but composers and chess players must learn how to put it all together.

Yep.

This is what is so disappointing about most of the research in this area. One should be designing experiments to test how much is chunking and how much is formal operations thinking in real life situations. Early experiments focused on looking at a position for a very short period of time as opposed to giving individuals time to show that they can improve their understanding of the position by having more time to think.

It seems to be difficult to separate chunking and formal operations. How could this be done? maybe it has to be done by analyzing thought processes in de Groot style.

Kotov's Think Like a Grandmaster is an excellent example of this. Kotov trained to think about chess positions in a certain way. Does training ppl in his method really work? Modern trainers focus purely on tactics but then you have players like Magnus who take advantage of that by putting them in strategic positions that they don't understand.

Would be interesting to see. Kotov's claim that you should stick with one line at a time sequentially is not supported. de Groot showed that the Masters would often go back to lines at the beginning, before narrowing it down two about two competing lines. Going back and forth between lines is beneficial as it adds more context.

I believe studies that induce blundering is a good way to understand thinking processes because it shows how we can trip ourselves up. Is it because we see a tactic and play a move before looking a little deeper and finding the best tactic? Or is it an overall understanding of the position that avoids blunders? I find all of this very interesting. Chess can teach us how we learn, understand and evaluate.

First blunder needs to be defined, to see if big positional mistakes count as blunders. If its both piece loss and strategic blunders than it means that it just depends of whether there was a calculation error or some sort of overall misunderstanding. But defining understanding is much harder than defining calculation error which leads to material loss.

@defense57 said in #5: >In particular, what can be chunked in chess is tactics and pattern recognition as it applies to positional chess. However, it requires formal operations to put all of this together. That is, recognition is all that is required of piano players but composers and chess players must learn how to put it all together. Yep. >This is what is so disappointing about most of the research in this area. One should be designing experiments to test how much is chunking and how much is formal operations thinking in real life situations. Early experiments focused on looking at a position for a very short period of time as opposed to giving individuals time to show that they can improve their understanding of the position by having more time to think. It seems to be difficult to separate chunking and formal operations. How could this be done? maybe it has to be done by analyzing thought processes in de Groot style. >Kotov's Think Like a Grandmaster is an excellent example of this. Kotov trained to think about chess positions in a certain way. Does training ppl in his method really work? Modern trainers focus purely on tactics but then you have players like Magnus who take advantage of that by putting them in strategic positions that they don't understand. Would be interesting to see. Kotov's claim that you should stick with one line at a time sequentially is not supported. de Groot showed that the Masters would often go back to lines at the beginning, before narrowing it down two about two competing lines. Going back and forth between lines is beneficial as it adds more context. >I believe studies that induce blundering is a good way to understand thinking processes because it shows how we can trip ourselves up. Is it because we see a tactic and play a move before looking a little deeper and finding the best tactic? Or is it an overall understanding of the position that avoids blunders? I find all of this very interesting. Chess can teach us how we learn, understand and evaluate. First blunder needs to be defined, to see if big positional mistakes count as blunders. If its both piece loss and strategic blunders than it means that it just depends of whether there was a calculation error or some sort of overall misunderstanding. But defining understanding is much harder than defining calculation error which leads to material loss.

@RuyLopez1000 said in #6:

But how does chunking demonstrate the 'limited faculty of chess skill'? The 'Criticism' and 'Summary' section of the blog shows plenty of observations that chunking theory doesn't explain in the context of chess.

Sorry for my poor wording, but chess skill unlikely transfers to other areas, As building huge chunks for chess is only useful for chess, as say mathematics is useful for countless fields not just math.

@RuyLopez1000 said in #6: > But how does chunking demonstrate the 'limited faculty of chess skill'? The 'Criticism' and 'Summary' section of the blog shows plenty of observations that chunking theory doesn't explain in the context of chess. Sorry for my poor wording, but chess skill unlikely transfers to other areas, As building huge chunks for chess is only useful for chess, as say mathematics is useful for countless fields not just math.

I would define a blunder as an incorrect solution to a chess problem on a chess tactics server such as Lichess puzzles.

Kotov's argument is that it confuses the mind to go back and forth between variations where you end up blundering because you mix up the variation in your mind. This is why many GMs don't look at the board when they're playing. Seeing the pieces stuck there is distracting. Would training someone in Kotov's manner lead to greater chess improvement over just practicing tactics (improving chunking) ? My guess is it depends on how much you know about chess to begin with. Perhaps beginners would improve better by just learning simple tactics but stronger players would benefit from taking a more systematic approach. My guess is Kotov's approach is superior or it wouldn't have been used by the Russian chess school for so long a period of time. It's true many master's don't use Kotov's approach but I'm not talking about what ppl do but how they can improve which I also believe is how one could begin to do some more real world experiments by testing how various training approaches influence ability to solve puzzles. Lichess is nice in that it breaks down what sorts of positions you do well out and what you don't. For example, I'm rated at 2650 for quiet moves (intermezzos) but only 2000 for an exposed king. Where I should be training becomes obvious.

The Russian Chess school focuses on tactics, opening training (learning two openings for each), positional chess and endgame play throughout the week and then playing in all time controls on weekends. They claim an IM rating within 2 years should be the result with 4-6 hours of study per day. This gets back to the idea that most of what players study isn't chunking but understanding -- openings, endgames, pawn structures and piece placement.

Can one throw all of this out the window and just focus on tactics and be successful? We seem to have a world champion that has taken the calculation approach to the extreme.

The way most GMs describe top player chess nowadays is they prepare by memorizing some variations for opening play and then develop a deep understanding of how the resulting positions should be played. If the opponent knows how to play these positions then the result will be a draw but if they don't than one has a chance to win. It seems like this is more deeply understanding certain positions than chunking them. Magnus' strength is a deep understanding of almost all positions and his knowledge of what positions his opponents are not expert at.

I believe the operational definition of understanding can be studying positional chess.

Studying chunking could be using various teaching method that focus on various types of tactics.

Having a systematic "Thinking" approach to calculation might involve training in Kotov's method. The dependent variable would be consistency in solving problem (reducing blundering) . The research seems to show that extreme overlearning is the key (the chunking approach) but are there shortcuts?

One experiment might be to ask players to write down the moves they are considering and then guess the correct moves using Lichess puzzles. Then train them in Kotov's method or in a control group ask them to do more puzzles in anyway they prefer and determine if the more systemic approaches leads to a higher tactics rating. My personal experience is that Kotov method works better especially when I haven't played in a while. Of course, it's quite hopeless for someone as old as I am. However, I can tell you I was stuck at 1800 FIDE for more than a year. in 1975 I read and applied Kotov's method and I moved to 2290 FIDE in a year. I never had an expert rating, I skipped it. A more dramatic example would be Larry Christian who skipped the IM title and went straight to GM. He felt he didn't blunder because he was in very good physical shape (a national swimming champion) which is another important element to reducing blundering. Similarly with Sierawan who was a martial artist. I don't recall their approaches to studying tactics it would be interesting to take a survey of GMs. James Tarjan studied in Yugoslavia which gave him a very solid and blunder proof approach to chess. The Yugoslavian approach is similar to Kotov's.

There was a round table a few years ago where some of the top players began explaining their training methods and they all discovered that they study composed chess problems to induce greater creativity. That is, just studying different types of problems (e.g., as those categories described by Lichess or more simple categorizations) might be another approach to tactics training. It's a truly facisinating area of study. After so many years, I believe there is so much left to explore.

Of course David Lane's methodological approach to determining formal operational thinking is useful. Lane is a chess master but gave it up long ago. I suggested to him that chess must have an element of problem solving to it which may account for why he had a student to a dissertation on the topic many years after I suggested the idea to him. I still believe the idea has merit.

I would define a blunder as an incorrect solution to a chess problem on a chess tactics server such as Lichess puzzles. Kotov's argument is that it confuses the mind to go back and forth between variations where you end up blundering because you mix up the variation in your mind. This is why many GMs don't look at the board when they're playing. Seeing the pieces stuck there is distracting. Would training someone in Kotov's manner lead to greater chess improvement over just practicing tactics (improving chunking) ? My guess is it depends on how much you know about chess to begin with. Perhaps beginners would improve better by just learning simple tactics but stronger players would benefit from taking a more systematic approach. My guess is Kotov's approach is superior or it wouldn't have been used by the Russian chess school for so long a period of time. It's true many master's don't use Kotov's approach but I'm not talking about what ppl do but how they can improve which I also believe is how one could begin to do some more real world experiments by testing how various training approaches influence ability to solve puzzles. Lichess is nice in that it breaks down what sorts of positions you do well out and what you don't. For example, I'm rated at 2650 for quiet moves (intermezzos) but only 2000 for an exposed king. Where I should be training becomes obvious. The Russian Chess school focuses on tactics, opening training (learning two openings for each), positional chess and endgame play throughout the week and then playing in all time controls on weekends. They claim an IM rating within 2 years should be the result with 4-6 hours of study per day. This gets back to the idea that most of what players study isn't chunking but understanding -- openings, endgames, pawn structures and piece placement. Can one throw all of this out the window and just focus on tactics and be successful? We seem to have a world champion that has taken the calculation approach to the extreme. The way most GMs describe top player chess nowadays is they prepare by memorizing some variations for opening play and then develop a deep understanding of how the resulting positions should be played. If the opponent knows how to play these positions then the result will be a draw but if they don't than one has a chance to win. It seems like this is more deeply understanding certain positions than chunking them. Magnus' strength is a deep understanding of almost all positions and his knowledge of what positions his opponents are not expert at. I believe the operational definition of understanding can be studying positional chess. Studying chunking could be using various teaching method that focus on various types of tactics. Having a systematic "Thinking" approach to calculation might involve training in Kotov's method. The dependent variable would be consistency in solving problem (reducing blundering) . The research seems to show that extreme overlearning is the key (the chunking approach) but are there shortcuts? One experiment might be to ask players to write down the moves they are considering and then guess the correct moves using Lichess puzzles. Then train them in Kotov's method or in a control group ask them to do more puzzles in anyway they prefer and determine if the more systemic approaches leads to a higher tactics rating. My personal experience is that Kotov method works better especially when I haven't played in a while. Of course, it's quite hopeless for someone as old as I am. However, I can tell you I was stuck at 1800 FIDE for more than a year. in 1975 I read and applied Kotov's method and I moved to 2290 FIDE in a year. I never had an expert rating, I skipped it. A more dramatic example would be Larry Christian who skipped the IM title and went straight to GM. He felt he didn't blunder because he was in very good physical shape (a national swimming champion) which is another important element to reducing blundering. Similarly with Sierawan who was a martial artist. I don't recall their approaches to studying tactics it would be interesting to take a survey of GMs. James Tarjan studied in Yugoslavia which gave him a very solid and blunder proof approach to chess. The Yugoslavian approach is similar to Kotov's. There was a round table a few years ago where some of the top players began explaining their training methods and they all discovered that they study composed chess problems to induce greater creativity. That is, just studying different types of problems (e.g., as those categories described by Lichess or more simple categorizations) might be another approach to tactics training. It's a truly facisinating area of study. After so many years, I believe there is so much left to explore. Of course David Lane's methodological approach to determining formal operational thinking is useful. Lane is a chess master but gave it up long ago. I suggested to him that chess must have an element of problem solving to it which may account for why he had a student to a dissertation on the topic many years after I suggested the idea to him. I still believe the idea has merit.

Interesting, thanks for bringing to light. My perception is that there is no one simple model for describing chess thinking and that it is a complex combination of many factors of which Chunking could be an important one. In particular, I can think of good answers to the deficiencies proposed in your Conclusion that build on Chunking with other factors. For example, candidate moves could be of better quality with better and more chunks remembered by masters and a system of priorities to govern which are the suggested moves should be at the top of the list.

Interesting, thanks for bringing to light. My perception is that there is no one simple model for describing chess thinking and that it is a complex combination of many factors of which Chunking could be an important one. In particular, I can think of good answers to the deficiencies proposed in your Conclusion that build on Chunking with other factors. For example, candidate moves could be of better quality with better and more chunks remembered by masters and a system of priorities to govern which are the suggested moves should be at the top of the list.