What Chess Tells Us About the Value of Perception

As a physics student, I found that I could solve most of the problems simply by looking at derivations and listening carefully to my reactions to the equations. A soft voice inside me would say, “No, that term just doesn’t seem right. Go and find out what went wrong there.” Or, “Ah, these terms hang together and the result feels right. It must be okay.” And it almost always worked out. My piano teacher would do the same when playing an unfamiliar piece of music. She could play it just by making sure it sounded right.

Were these just party tricks? Or was a more fundamental process going on?

The answer is in my favorite study from the field of expertise and expert performance; like the best research in this field, it teaches us general lessons about how one reaches high performance in almost any domain, whether music, chess, poker, mathematics, or teaching.

The study is by Fernand Gobet and the late Herbert Simon, winner of the 1978 Nobel Prize in Economics. They wanted to measure the relative importance of analysis (calculation) and perception (insight) in expert performance. And they thought of a beautiful measurement based on objective data. [“The roles of recognition processes and look-ahead search in time-constrained expert problem solving: Evidence from grandmaster level chess”, Psychological Science 7:52-55 (1996)]

Comstock

Their subject was Garry Kasparov, chess champion of the world for 15 years (1985-2000). As world champion, he often demonstrated his skill by playing a “simul”: games of chess against several masters and grandmasters simultaneously. Kasparov would have to rotate between games. As soon as Kasparov reached a board, his opponent on that board had to make his or her move. Kasparov would then think for roughly 20 seconds before moving on to the next opponent and game. In contrast, his opponent would think for 3 minutes, until Kasparov’s return from cycling through the other games (typically six others).

At 20 seconds per move, Kasparov mostly used his perception and judgment of chess positions rather than his ability to calculate chess variations (the “I take, he takes, I take, etc.” kind of thinking). Thus, simultaneous chess is a real-life laboratory for measuring the value of perception. How well did Kasparov play, in comparison to his normal strength when playing at the usual tournament rate of 3 minutes per move? His normal strength at the time was 2750 on the Elo scale of chess skill. (To give a feel for the Elo scale, a beginner would be rated about 1000, an average tournament player is rated about 1600, a master is rated at 2200 or above, and a grandmaster is usually above 2400.)

The amazing result: At the rapid “simul” pace, Kasparov performed at a rating of 2650: higher than all but half a dozen players in the world! In other words, most of his world-class expertise comes from how he sees and looks at the chess board, not from his calculation ability. The traditional picture of the chess master as a calculating prodigy is bogus.

This unexpected result explains my (lack of) success playing chess against a friend in college, Adam Lief, who at the time was a very strong master. To make the games even remotely interesting, Adam started with less and less time, eventually reaching 30 seconds to make all of his moves while I had my 5 minutes for all my moves. In my best effort at those odds, I managed, in one game, to get Adam on the run. As I was gloating to myself, he paused for a longer think. As his time was running out, I sensed victory, even if it would come from Adam’s running out of time. After an eternity (7 seconds), he sacrificed his queen and announced “mate in 6”, i.e. that I would be checkmated in 6 moves even if I played perfectly.

Adam had only 5 seconds total for all his moves, so I thought, “I’ll just move fast, so that he doesn’t have time to think while I think, and might then not manage the checkmate.” Bad choice. With my first move, I made a mistake. Adam instantly announced “mate in 2.” And so vanished my victory, the only time I managed to even come close—even though Adam barely even had time to move his piece and hit his chess clock, let alone study the board. For the master, perception and insight are king!

This idea, as I will discuss in a subsequent post, explains the insanity of most mathematics teaching.


Krish

Is there any research on the transition from "calculative thinking" when someone is a beginner to "intuitive/perceptive thinking" when that person becomes an expert? Does engaging in intensive "calculative thinking" sufficient to build the intuitive muscles? Or is it more complex than that?
One thing I would be very interested in knowing is understanding the decline in the efficiency of this transition with age? That is can 'older' people still master arts like mathematics, music with sufficient practice?

Jay L

It's a combination of practicing calculation and learning the theory. Here's one analogy - it's possible that someone could advance from beginner to concert pianist based solely on practicing pieces and learning more songs - start with Twinkle Twinkle Little Star, work your way through the Mozart minuets, and eventually you're ready for the Rachmaninoff concertos. That said, I doubt many pianists have actually been able to do that. Musical patterns on the keyboard tend to make more sense when the student has training in scales and arpeggios. It accelerates the learning when you have that theoretical side to back up the rote calculation, and this is the same for chess. Students learn strategic fundamentals like opening principles (take the center, develop knights/bishops before rooks/queen) and endgame technique (outside passed pawns, bad vs. good bishops). These ideas are like scales, and they complement and reinforce the tactical side of chess. I suppose some gifted chess players can figure these ideas out, but they're more commonly learned through study and instruction.

Read more...

Lancaster

speed golf suggests that it's not only chess where performance doesn't drop as much as one might expect when put under time pressure.

tobias

This doesn't in any way dispel the notion that chess masters are calculating prodigies. It tends to confirm that calculation often occurs without conscious computation.

Sarah

It seems to me a lot of calculating (of the she takes, I take variety) must have taken place behind the scenes during individual practice. The insight can only come after hours and hours of analysis. When you watch Kasparov play "simul", you are seeing the end result of a lifetime of study. I don't think Kasparov started out learning to play chess using just perception. But you make an interesting point, experts are so good that what they do becomes intuition or second nature.

After 30 years of playing the cello off and on, I no longer make the mistake of leaving out sharps or flats. I used to have to check the key signature, but now I infrequently play wrong accidentals because I hear (intuit) the notes a split second before I put my fingers down. Still have many years to go before I become an expert though!

Eric M Jones

Right on. Here's my math lesson on fractions: "There aren't any fractions.... It's just made up."

Intuitive understanding is not the way our school system is structured, but it should be. Paul Hewitt writes physics books designed to bring out intuitive understanding in students. (Does a big rock fall faster than a small rock? Imagine that the big rock is just a bunch of small rocks flying in formation.)

I claim vast but entirely intuitive knowledge of etymological matters and probably annoy Fred Shapiro's postings followers, but some thing just ring right, and some just don't. I'll go to the mat on many matters for which I have absolutely no objective evidence. In these matters "experts" have no sway with me.

But I also claim vast but entirely intuitive ignorance, or that is, I don't feel anything I have heard rings right...thus a smooth and unrippled fabric of ignorance stretch before me, and I am a pain in the butt in the presence of people who claim to know. Example: Global Warming.

Read more...

Miley Cyrax

I suspect a similar phenomenon in language acquisition as well. Fluent individuals understand the language "automatically," whereas non-fluent individuals first convert it to their primary language.

Kelsey H

As inferred to in this article, a key skill to succeeding in chess is pattern recognition--developing as effective as possible in as few moves as possible. A good chess player has developed complex patterns of recognition on the board where they can instantly perceive how different pieces on the board can support or mount attacks with other.

Most people struggle so much with chess because they think the whole game is about calculating a perfect execution of a checkmate rather than playing for positional and developmental strength.

Examples of basic patterns chess players come to recognize: effective use of pawn structures to break gridlock, how to effectively develop in as few moves as possible, the importance of knights in closed games and bishops in open games, how controlling the centre can open attacking opportunities, effective development of rooks, effective use of the king was a defending piece during the end game, etc........

The list is endless. A good player is able to identify these patterns, but most critically, evaluate in any given situation which set of strategy from these patterns is the most effective to achieving victory. That's the really hard part.

Read more...

njc

An alternative possible explanation as to how Kasparov won is shown in this video of Derren Brown, playing 9 simultaneous games including grandmasters:

http://www.dailymotion.com/video/x8a6st_derren-brown-beats-9-chess-players_videogames

He basically played them off against each other.

Neil (SM)

Certainly he would not have been able to play at 2650 using that trick. He's known for getting 100% scores in simuls.

Antonio Capillo

What is the role which emotivity to high psychological pressure plays in those situations. It looks like to me it is not an aspect that can be ignored. That is, under stress, our capability to calculate decreases significantly. Then, our subconscious could fill up this gap with perception. It could be, eventually, that controlled for stress, the one with higher calculation abilities is the winning one. Or, that perception is a form of unconscious analysis. Brain is so powerful.

Tom H.

This may be an apocryphal story, but it was told by a visiting lecturer in an artificial intelligence course at Stanford in 1974. He said in the early days of trying to create chess programs some computer scientists were studying how humans played and doing experiments. In one experiment a group of players including Mikhail Tal (a world champion) were shown board positions flashed briefly on a screen and were asked to recall them. As the positions were flashed for shorter and shorter intervals, weaker players dropped out, and finally only Tal could correctly itemize the position. Then the interval was further shortened and even Tal could not correctly itemize all the pieces, but.... he could specify the best move from the position! Supposedly at this point the AI researchers realized analyzing human experts might not be a viable approach.

MW

From the description, Kasparov's opponents were also at a disadvantage compared to tournament chess (although a lesser one.) They had to move once every 3 minutes, rather than *on average* once every 3 minutes (i.e. they can think longer in tricky situations, if they make up for it by making other moves faster). This means that Kasparov's simul rating of 2650 is something of an overestimate, as he was playing against handicapped opponents.

Marc W.

"At 20 seconds per move, Kasparov mostly used his perception and judgment of chess positions rather than his ability to calculate chess variations (the “I take, he takes, I take, etc.” kind of thinking). "

I'm not sure this is necessarily true. It's likely that Kasparov was still calculating just more efficiently. I would imagine he becomes more aggressive with his elimination of strange variations. Also Kasparov likely wasn't thinking "I move here, he moves there," instead he would be thinking "After this is acceptable, after this is unacceptable" etc. If Kasparov is able to block moves together more efficiently than his opponent, and wastes less time considering unlikely variations, he can still be effective under the time constraints.

Likewise, I fear your example with your friend may confuse the calculation and perception. Perception would be your friend announcing "white is in a losing position" however "mate in 6" shows that clearly he has calculated out the exact variations (your playing fast can't change what he has already calculated). The fact that "mate in 6" became "mate in 2" after your move shows that he was able to prune the variation at 2 moves (a big time saver for him) while you presumably missed that shortcut. In a time-handicapped game, the stronger player does most of their calculating on the opponent's time, and only needs their own time if a novel move is played.

Read more...

Jeff Yablon

Interestingly, Kasparov himself said something just like this about fifteen months ago.

Like me (sorry), he basically stated "Trust Yourself": http://answerguy.com/2010/04/02/new-business-process-kasparov-stop-thinking-play-chess/

Joshua Northey

I used to do something similar when I was my state's "Knowledge Bowl" star. When you buzzed in they stopped reading the question, and in addition to simply having a large knowledge base, I was an absolute beast at ringing in just a few words in an inferring what the answer would be based on the types of subjects they were likely to ask, typical phrasings, knowledge that would be uncommon but not impossible for a top-shelf high schooler to have. Yet unlike most of the other participants I didn't prepare by memorizing reams and reams of past questions, it was just all on the spot intuition and drawing on the thousands of books I had read.

So when they would ask "What liquid m-" I would ring in and answer "quicksilver". A question about which metal is liquid at room temperature is too easy, so they would be looking for something a little more esoteric, like an alternate name. I was accused of cheating on more than one occasion. It was great times.

This is also how I could stagger into standardized tests with a hangover and no prep/sleep and get great scores, its always too clear what answers they are looking for. Too many patterns. It is really too bad I came from such a screwed up background and never developed a work ethic or appreciation of jumping through life's ridiculous hoops until my late 20s, I could have made something of my life...

Read more...

Bald Idiot

The author's description of clock simuls is inaccurate (and when Kasparov played simuls against groups of strong players, they were clock simuls). Kasparov's opponent in a clock simul was free to move whenever he chose to move. (In a normal, n0-clock simul, the opponent must move when the master reaches his board; Kasparov did not play no-clock simuls against strong opposition.)

More importantly, the evidence the author cites is flawed in a manner that tends to overstate the role of Kasparov's chess understanding. When Kasparov initially played a clock simul, he played "cold" and his result was unimpressive. (He scored 3.5 out of 8 against a team of only moderate strength.) For his subsequent clock simuls, Kasparov would prepare beforehand by examining his upcoming opponents' games, identifying weak areas, and devising strategies to target those weak areas. The evidence clearly demonstrates that the pre-simul preparation was crucial to Kasparov's success.

Here is a link to a ChessBase article on the subject: http://www.chessbase.com/newsdetail.asp?newsid=6251

Read more...

ezhaya@gmail.com

This makes a lot of sense. When I played football. I was a defensive lineman. For many years I was taught muscle memory for which way the running back would be going. This not from watching the backfield because it is very deceptive. It is from which way are the blockers trying to block you. You cannot think about it because it would be to late. Learned from taking one step into the initial blocker and instinctively realizing which way he was blocking you and going in the opposite direction or pushing the blocker that way to plug a hole so the running back could not get through. We did drills for many many hours.

Also if the blocker talk a step back it meant a pass. If you think about it, you only have less than a second to make a decision on which way to go to get the running back or to get around the blocker. If you wait to see where the running back or quarter back is going. Or if you try to figure out what the blocker is doing with out knowing instinctively it will be to late. People who do not do this instinctively is typically why deception in the backfield works so well. Counter plays always get people running in the wrong direction.

I am sure the same could be in baseball with line drives, hard ground balls, and deep fly balls. If you start running in the wrong direction your in trouble or hesitate.

Read more...

Tim

FYI - Kasparov's first name is normally spelled "Garry" not "Gary" http://en.wikipedia.org/wiki/Kasparov