Thanks for interesting post (sequence of posts).
Pattern recognition is highly subjective. This is why we need rules in order to calculate to some result only using those rules. There is not one correct answer for any of those example puzzles just as there is not one correct melody to be sung in this world.
You get rewarded for what the rule makers want, and that is what sticks as the answer.
youtu.be/aMEbboWmVCo?si=_FYkhWqqw1VAgCQA
You get rewarded for what the rule makers want, and that is what sticks as the answer.
youtu.be/aMEbboWmVCo?si=_FYkhWqqw1VAgCQA
OP said: "I've found quite a lot of chess research that proceeds from the assumption that pattern recognition in various guises is specifically important to playing chess well. My guess is that a lot of my readers have heard advice about building up pattern recognition via focused training on tactics (drill those puzzles!) but is that really the cognitive heart of the game?"
I would be cautious about associating tests like SPM with "pattern recognition" in general. Pattern recognition in one domain may not, (probably doesn't), translate to pattern recognition in another domain. If we're talking about dance choreography, Jennifer Lopez is going to score very highly. If we're talking about reading a defense in American Football, quarterback Tom Brady is going to score very highly but will Jennifer or Tom be able to transfer their pattern recognition expertise to chess? I'm guessing no.
While the ability to recognize patterns may be innate in humans, the patterns they eventually are able to recognize are not innate and must be learned. Or to put it in more concrete terms, Leela chess has learned to recognize chess patterns but probably sucks at dancing.
I would be cautious about associating tests like SPM with "pattern recognition" in general. Pattern recognition in one domain may not, (probably doesn't), translate to pattern recognition in another domain. If we're talking about dance choreography, Jennifer Lopez is going to score very highly. If we're talking about reading a defense in American Football, quarterback Tom Brady is going to score very highly but will Jennifer or Tom be able to transfer their pattern recognition expertise to chess? I'm guessing no.
While the ability to recognize patterns may be innate in humans, the patterns they eventually are able to recognize are not innate and must be learned. Or to put it in more concrete terms, Leela chess has learned to recognize chess patterns but probably sucks at dancing.
@SummerThereof said in #23:
> OP said: "I've found quite a lot of chess research that proceeds from the assumption that pattern recognition in various guises is specifically important to playing chess well. My guess is that a lot of my readers have heard advice about building up pattern recognition via focused training on tactics (drill those puzzles!) but is that really the cognitive heart of the game?"
>
> I would be cautious about associating tests like SPM with "pattern recognition" in general. Pattern recognition in one domain may not, (probably doesn't), translate to pattern recognition in another domain. If we're talking about dance choreography, Jennifer Lopez is going to score very highly. If we're talking about reading a defense in American Football, quarterback Tom Brady is going to score very highly but will Jennifer or Tom be able to transfer their pattern recognition expertise to chess? I'm guessing no.
>
> While the ability to recognize patterns may be innate in humans, the patterns they eventually are able to recognize are not innate and must be learned. Or to put it in more concrete terms, Leela chess has learned to recognize chess patterns but probably sucks at dancing.
Don't worry - cognitive scientists tend to be a cautious lot, too. :) I don't think there are many folks who study this stuff who would predict that domain expertise in football, dance, or any other domain would immediately transfer to comparable expertise in chess or another complex domain. Skill learning requires experience with a specific task. That doesn't exclude the possibility that there could be a broader mechanism for recognizing patterns that contributes to your performance across domains, though. It's no problem for a more general mechanism and specific experience to play a role.
To borrow your analogy for a second, JLo and Tom Brady might not be great at chess on Day 1, but maybe some general facility at recognizing patterns contributes to both of them becoming proficient very rapidly - unless this is what you're saying you're guessing no about? Such an outcome could mean that they had to learn specific things about the game (which isn't surprising), but a broader mechanism could also have supported high performance in multiple tasks (which would be interesting). It wouldn't have to turn out that way, but it could. Likewise, I agree that Leela is indeed not going to learn to dance anytime soon, but that also doesn't rule out humans having cognitive mechanisms that support pattern recognition across domains.
If you're interested in this sort of thing, you can check out the psychometric properties of the SPM to see what the validity looks like. No one would tell you that a single scale like this is a perfect proxy for a cognitive process, but you can see what kinds of factor structure we end up with after asking people to complete lots of different tasks and draw your own conclusions.
> OP said: "I've found quite a lot of chess research that proceeds from the assumption that pattern recognition in various guises is specifically important to playing chess well. My guess is that a lot of my readers have heard advice about building up pattern recognition via focused training on tactics (drill those puzzles!) but is that really the cognitive heart of the game?"
>
> I would be cautious about associating tests like SPM with "pattern recognition" in general. Pattern recognition in one domain may not, (probably doesn't), translate to pattern recognition in another domain. If we're talking about dance choreography, Jennifer Lopez is going to score very highly. If we're talking about reading a defense in American Football, quarterback Tom Brady is going to score very highly but will Jennifer or Tom be able to transfer their pattern recognition expertise to chess? I'm guessing no.
>
> While the ability to recognize patterns may be innate in humans, the patterns they eventually are able to recognize are not innate and must be learned. Or to put it in more concrete terms, Leela chess has learned to recognize chess patterns but probably sucks at dancing.
Don't worry - cognitive scientists tend to be a cautious lot, too. :) I don't think there are many folks who study this stuff who would predict that domain expertise in football, dance, or any other domain would immediately transfer to comparable expertise in chess or another complex domain. Skill learning requires experience with a specific task. That doesn't exclude the possibility that there could be a broader mechanism for recognizing patterns that contributes to your performance across domains, though. It's no problem for a more general mechanism and specific experience to play a role.
To borrow your analogy for a second, JLo and Tom Brady might not be great at chess on Day 1, but maybe some general facility at recognizing patterns contributes to both of them becoming proficient very rapidly - unless this is what you're saying you're guessing no about? Such an outcome could mean that they had to learn specific things about the game (which isn't surprising), but a broader mechanism could also have supported high performance in multiple tasks (which would be interesting). It wouldn't have to turn out that way, but it could. Likewise, I agree that Leela is indeed not going to learn to dance anytime soon, but that also doesn't rule out humans having cognitive mechanisms that support pattern recognition across domains.
If you're interested in this sort of thing, you can check out the psychometric properties of the SPM to see what the validity looks like. No one would tell you that a single scale like this is a perfect proxy for a cognitive process, but you can see what kinds of factor structure we end up with after asking people to complete lots of different tasks and draw your own conclusions.
I was wondering if you could update us on the experiments where people would wear inverting glasses for all their awake time, I have reminiscence of before internet documentaries reporting on this. It was something like month time scale before the person would have their sense of upside down adapted.
Do I remember well, and has this been found reproducible? If so what does that tell us about how adult synaptic or "neural" plasticity (bigger concept?) ability to adapt to such whole visual field of vision operations.
I would not think either that pattern having been internalized experientially in one field would automatically find the approriate remapping into another... Although, it has been my experience in mathematical models of certain topics, that if one is using a certain abstraction level that is adequate to have predictive value in context, one might figure out an angle of approach that would allow some patterns to be reused. So, not all experiences are completely distinct from each other, but thinking they would instantly "transpose" is something else. There is also notions of distances between filed of expertise.
Consider intrinsic conservation of grammatical linguistic patterns, while divergence betwee Spanish and Portuguese (and other latin based grammars maybe for conjugation of verbs). The amount of teaching rules need to teach portuguese as second language for native spanish compared to say english native. It becomes a smaller task to instead of teaching the whole thing from scratch, to rather tap on existing commonalities, and consacrate the learning energies to transpostioin rules.
This is second hand knowledge from a linguist friend long ago, who had actually practiced that, in applied linguistics.
I also have used my own french native knowledge to infer a bunch of things while in a semester course of Spanish as second (for me third or fourht, never quite made it though, verbally speaking or listening. Once some phonetic declinations were spotted, the way they would be applied would be tapping on the similarity of structure for the equivalent french rules across tenses and verb groups.
Or in biology of populatoin dynamics, from molecular populations (concentrations) in well stirred tanks (or not) to intro orgasnims tissue population (fixed or diffused in blood or other fluids or other tissues even), to ecological scale interacting populations, it might be that the dynamical questions would have some patterns in common, while being expert in the mathematical model forumlation would not make one an experiet in the cladistic knowledge in zoology, I think that one should allow noticing such things across apparently differnt objects of expertise..
The time to transpositon, might not be instantanesous. I am not talking about chess transpositions (although maybe one could make some arguments that way, but not me right now).
Do I remember well, and has this been found reproducible? If so what does that tell us about how adult synaptic or "neural" plasticity (bigger concept?) ability to adapt to such whole visual field of vision operations.
I would not think either that pattern having been internalized experientially in one field would automatically find the approriate remapping into another... Although, it has been my experience in mathematical models of certain topics, that if one is using a certain abstraction level that is adequate to have predictive value in context, one might figure out an angle of approach that would allow some patterns to be reused. So, not all experiences are completely distinct from each other, but thinking they would instantly "transpose" is something else. There is also notions of distances between filed of expertise.
Consider intrinsic conservation of grammatical linguistic patterns, while divergence betwee Spanish and Portuguese (and other latin based grammars maybe for conjugation of verbs). The amount of teaching rules need to teach portuguese as second language for native spanish compared to say english native. It becomes a smaller task to instead of teaching the whole thing from scratch, to rather tap on existing commonalities, and consacrate the learning energies to transpostioin rules.
This is second hand knowledge from a linguist friend long ago, who had actually practiced that, in applied linguistics.
I also have used my own french native knowledge to infer a bunch of things while in a semester course of Spanish as second (for me third or fourht, never quite made it though, verbally speaking or listening. Once some phonetic declinations were spotted, the way they would be applied would be tapping on the similarity of structure for the equivalent french rules across tenses and verb groups.
Or in biology of populatoin dynamics, from molecular populations (concentrations) in well stirred tanks (or not) to intro orgasnims tissue population (fixed or diffused in blood or other fluids or other tissues even), to ecological scale interacting populations, it might be that the dynamical questions would have some patterns in common, while being expert in the mathematical model forumlation would not make one an experiet in the cladistic knowledge in zoology, I think that one should allow noticing such things across apparently differnt objects of expertise..
The time to transpositon, might not be instantanesous. I am not talking about chess transpositions (although maybe one could make some arguments that way, but not me right now).
Could the previous to last paragraph about the the error detector delay specific to the chess crowd, indicator of more self awareness of the solution plan versus the board forecasting not well. That having done pre-planning implies having set some intermediate things along the way, and that a certain element of uncertainty in not having tested the plan, comes with executing the plan, hmm... having taken the time of preplanning might be a first testing, that the strategy to running first, might mean a more on the spot greedy plan never looking back (or present)...
babbling.. but I think planning in advance has some corroloaries or correlates or both about chekcing how it is going. so looking both at board and own "theory" of solution (=plan). or I throw that as question.
babbling.. but I think planning in advance has some corroloaries or correlates or both about chekcing how it is going. so looking both at board and own "theory" of solution (=plan). or I throw that as question.
This research still lacks the critical and most important component related to playing chess which is conceptual knowledge. When slow conceptual processes are added to the mix it shows that skilled chess players know when to look for a good and will use slow processes to find it. www.researchgate.net/publication/291363666_There_Is_Time_for_Calculation_in_Speed_Chess_and_Calculation_Accuracy_Increases_With_Expertise
Actual chess knowledge as opposed to experience also accounts for a significant part of chess memory.
general non-visual fluid intelligence plays a role as well. www.academia.edu/55428019/Chess_knowledge_predicts_chess_memory_even_after_controlling_for_chess_experience_Evidence_for_the_role_of_high_level_processes
The bottom line is if you want to get better at chess you need to develop a conceptual knowledge about the game and develop/use a chess domain specific fluid knowledge that is arrived at through internal conversation and calculation. Good chess players use chess visualization to look ahead and not make guesses based on visual pattern recognition. That is, pattern recognition is the beginning of a problem-solving process. Study of child prodigies show strong fluid memory as a more critical factor than time spent studying or playing chess.
Actual chess knowledge as opposed to experience also accounts for a significant part of chess memory.
general non-visual fluid intelligence plays a role as well. www.academia.edu/55428019/Chess_knowledge_predicts_chess_memory_even_after_controlling_for_chess_experience_Evidence_for_the_role_of_high_level_processes
The bottom line is if you want to get better at chess you need to develop a conceptual knowledge about the game and develop/use a chess domain specific fluid knowledge that is arrived at through internal conversation and calculation. Good chess players use chess visualization to look ahead and not make guesses based on visual pattern recognition. That is, pattern recognition is the beginning of a problem-solving process. Study of child prodigies show strong fluid memory as a more critical factor than time spent studying or playing chess.
I don't know man. If you are so knowledgeable, why are you bad at chess?