Wednesday, April 22, 2009

autism/ genius

two more thoughts related to the autism/ Rowson/ gender/ chess/ capacity for hard work/obsession conversation:

1. Gary Kasparov's speech at SuperNationals: "Hard work is a talent"
2. An Economist article on autism and genius:

"A standard diagnosis of autism requires three things to be present in an individual. Two of these three, impairments in social interaction and in communication with other people, are the results of autists lacking empathy or, in technical jargon, a “theory of mind”. In other words they cannot, as even fairly young neurotypicals can, put themselves in the position of another being and ask themselves what that other is thinking. The third criterion, however, is that a person has what are known as restrictive and repetitive behaviours and interests, or RRBI, in the jargon.

Until recently, the feeling among many researchers was that the first two features were crucial to someone becoming a savant. The idea was that mental resources which would have been used for interaction and communication could be redeployed to develop expertise in some arbitrary task. Now, though, that consensus is shifting. Several of the volume’s authors argue that it is the third feature, RRBI, that permits people to become savants."

Coming from a chess background, that seems intuitively true, right? Think about people who play on ICC 10 hours a day. The article also contains hope for the "neurotypicals" among us:

"Given such anatomical differences, then, what hope is there for the neurotypical who would like to be a savant? Some, possibly. There are examples of people suddenly developing extraordinary skills in painting and music in adult life as a result of brain damage caused by accidents or strokes. That, perhaps, is too high a price to pay. But Allan Snyder of the University of Sydney has been able to induce what looks like a temporary version of this phenomenon using magnetism.

Dr Snyder argues that savant skills are latent in everyone, but that access to them is inhibited in non-savants by other neurological processes. He is able to remove this inhibition using a technique called repetitive transcranial magnetic stimulation.

Applying a magnetic field to part of the brain disrupts the electrical activity of the nerve cells for a few seconds. Applying such a field repeatedly can have effects that last for an hour or so. The technique has been approved for the treatment of depression, and is being tested against several other conditions, including Parkinson’s disease and migraines. Dr Snyder, however, has found that stimulating an area called the left anterior temporal lobe improves people’s ability to draw things like animals and faces from memory. It helps them, too, with other tasks savants do famously well—proofreading, for example, and estimating the number of objects in a large group, such as a pile of match sticks. It also reduces “false” memories (savants tend to remember things literally, rather than constructing a mnemonic narrative and remembering that).

There are, however, examples of people who seem very neurotypical indeed achieving savant-like skills through sheer diligence. Probably the most famous is that of London taxi drivers, who must master the Knowledge—ie, the location of 25,000 streets, and the quickest ways between them—to qualify for a licence.

The expert here is Eleanor Maguire of University College, London, who famously showed a few years ago that the shape of the hippocampus, a part of the brain involved in long-term learning, changes in London cabbies."
update: maybe only tangentially relevant, but today's NY Times has an article on the importance of professional coaching in sports. It makes an interesting counterpoint to the idea in chess that adults almost never improve, even with coaching.


ATH2044 said...

I read the Economist article on autism and genius today & a survey popped up, so I just couldn't resist (due to my subclinical ADD) filling it out. The last question was "What is your primary source of information about news & current affairs?" so of course I typed in "".

You are here!Since "impairments in social interaction and in communication with other people" seem to be requirements for this type of savantism, one might conclude that conversely, excessive facility in social interaction and in communication with other people leads to idiocy. A quick look at today's video game & social networking culture would readily confirm this.
RRBI is jargon for "If you keep doing something a whole lot, you're bound to get better at it (& worse at everything else)", right? Is "repetitive transcranial magnetic stimulation" jargon for hitting yourself in the head with a magnetized hammer (I guess that would be "auto repetitive transcranial magnetic stimulation")?
I think it was Will Rogers who said, "We're all ignorant, just on different subjects."
I obviously spent more time filling out the survey than I did reading the Economist article on autism and genius.
There's also some interesting stuff about talent & hard work on the Kasparov link that I'll have to get into later.

Elizabeth Vicary said...

how do you make the hyperlinks in comments?

ATH2044 said...

Hey Liz,
I thought you'd never ask!
All you do is type [less than symbol < ], then [a href=""] [greater than symbol > ]then [Clickable text][less than][/a][greater than] & put the hyperlink (including the whole between the double quotes.
In case this thing tries to interpret my HTML code, it's also explained over here. I think the "a" stands for "anchor", so you can also find it by searching for HTML anchored text. If you're really brave, you can go up to your browser menu & click on view then page source to see how it was actually done on any web page.
There's only a few HTML tags that are allowed in postings, but the "a href=..." one is definitely useful. It's also good to spell out the links sometimes in case someone can't get to it on his current browser, but the vast majority of people will be fine with just clicking hot links.

Anonymous said...

How can someone who has trouble guessing what other people are thinking play a decent game of chess? Yet we know that some autistic people are good at chess, so something is wrong with that criterion.

Elizabeth Vicary said...

I don't ever try to guess what my opponent is thinking when I'm playing chess.

Howard Goldowsky said...

There is a common misconception that adult chess players never improve. This is unquantifiable, subjective rhetoric. Nobody has ever done a study or produced statistics proving this claim. Even Jonathan Rowson, in his 2007 lecture (it was at a conference in Sweden, and the link to the lecture can be found at Dennis M's "Chess Mind" blog), stated that adults can improve, but it's just not as common. Saying that adults "never" improve is just a vague and false statement. What age defines an adult? What rating gain defines improvement? A rating gain from 500 to 1500 is not the same as a rating gain from 1500 to 2500. Chess skill is non-linear. It takes much less training and ability to go the 1,000 points from beginner (500) to intermediate player (1500) as it does to go the 1,000 points from intermediate player to grandmaster (2500). It is rare to see "adults" (say, over 30) go from class-A (1800) to master (2200), but I have seen it done. But think about this: a master is in the top 1% of ability of all competitive chess players (higher than the top 1% if you include all competitive players who have dropped out of OTB chess and out of the rating pool). If, say, approximately 1% of adult class-A players make it to master, how is this rare? This is just the norm, because it is difficult to make master, in general. Perhaps my analysis is not perfect; all I'm saying is that a rigorous, objective analysis of the "no adult ever improves" phenomenon has never been undertaken.

Howard Goldowsky

Howard Goldowsky said...

One more note: I noticed that you qualified your last sentence with "almost never improve." OK. Believe me, as a 37 y/o (currently) class-A player striving to improve, I'm very interested in this topic. One possible reason Rowson gives as a show-stopper when adults try to improve is that they must unlearn bad habits. These bad habits could have been formed anywhere, but most likely they formed from beating poor players. If poor moves win, then the brain may be tricked into thinking that the moves are actually good.


es_trick said...

This may be only tangentially relevant, but makes for an interesting comparison.

I taught English overseas as a foreign language for fifteen years, mostly to adults age 18 – 70.
It was apparent to everyone in every class I ever taught that the college age students could catch on and assimilate new material much more rapidly than the older students. Those in their 30s could achieve a reasonable rate of progress if they worked hard enough. People in their 40s could make some progress. Those over 50 made very limited progress, if any.

For the past nine years I have been working as a performance consultant, monitoring the English language training of a large multinational corporation that has training sites in many countries around the world. The standardized test score data on nearly a thousand trainees that I have collected and analyzed as they go through their training goes far beyond the level of anecdotal evidence and impressions one may get from classroom observation.

As a group, young hires in their 20s make much more rapid progress than those in their 40s. Trainees in their 30s fall in between. Of course, there are exceptions. Some of the younger fellows may slack off or otherwise not take the training seriously. It’s also possible that the younger trainees with low rates of progress are simply lacking in aptitude for learning a foreign language. Most of the older trainees are able to improve, but at a significantly lower rate. The result is that the younger trainees typically ‘graduate’ from the first stage of training in a fraction of the time it takes the older ones to do so.

In some cases, I’ve had the opportunity to track these trainees’ progress over a three year period. Using a 0 – 10 language proficiency scale, trainees in their twenties are usually able to reach Level 5 (the company’s minimum for being allowed to take on work assignments in the field) within a reasonable amount of time. Some of the more talented ones get to Level 6, and a very small number may eventually reach Level 7.

Note that this 0 – 10 scale is non-linear in the same way as the ELO rating scale. That is, it takes much less learning to advance from Level 0 – 1 and 1 – 2 than it does to go from Level 5 – 6, or 6 – 7. A two level gap probably represents an exponentially greater amount of language skill acquired by the individual with the higher level. Also note that children learning a second language have the potential to reach 10 which is native speaker proficiency, if given the right conditions. Adult learners have a theoretical limit of Level 8 or 9.

Trainees in their 40s typically reach their “ceiling” after a gain of one or two levels. So, if they enter the training with Level 3 or 4 to begin with, they’ll probably make it to Level 5 and secure permanent employment with the company. But if they start at Level 2 or below, it is unlikely that they will get to Level 5, with some not even making it to Level 4. It’s almost certain that none of them will ever get to Level 6. The few trainees in their 50s who have been brought in rarely gain more than one level, no matter how much language training they get.

Recently there has been a fair amount of research documenting some of the physical changes that occur in the brain throughout childhood and adolescence. I don’t think there are many people who doubt that the adult brain gradually loses plasticity and that people no longer have the ability to become truly proficient in totally new fields of endeavor taken up in the later stages of life. This doesn’t mean that mature people can’t take up new hobbies and attain a level of skill where they can derive great enjoyment from it. But it’s extremely unlikely that they’ll ever reach an expert or master level at it (if it were possible to measure expertise in those fields.)

This, btw, comes to you from someone who is 50, who in the past couple of years was actually able to advance from a Class C rating to Class B, after not playing in any tournaments for almost three decades.

Another btw, I believe that USCF has done some fairly in depth research on the relationship between ratings and age. They have found that players between the age of 35 – 45 are the most stable group, in terms of rating fluctuation. In general, but not always, players over the age of 45 tend to experience decreases in their rating over time. This applies especially to players who have accumulated more than six years of tournament experience. This means that if you’re over 45, but still relatively fresh to the game, without too many hardened habits in your style of play or thought processes that cause people to become set in their ways, then there is still a learning curve that can be ascended!

ATH2044 said...

Geez Howard, you hit so many nails on the head that I don't know where to start. First the "dirty little secret" of chess ratings that "Chess skill is non-linear." is pretty obvious to anyone who's ever played, but the rating formula (which is anything but linear) supports this as well. Part of the problem that people encounter when they try to improve is that they run into a whole lot of other people who are trying to do the same thing. Rolf Wetzell addresses this in his (very useful by the way) book on the topic: "Chess Any Age". I'd long supposed that each 200 rating points represents approximately a doubling of the amount of knowledge involved. Rolf suggests it's more like every 100 points! That means that even the 100 points between 1800 & 1900 will require about twice the amount of work (however you define it) as the 100 points between 1700 & 1800. Getting to 2200 takes a whole lot of something that most people don't have much of.
Michael De La Maza makes a lot of noise about how intense training in tactics will add gobs (400+) points to your rating, but he also readily admits that this works best for people (like himself) starting out in the 1100-1500 range. (Read this review before reading his book.) No offense, but if there are any "easy" rating points to be had, you're a hell of a lot more likely to find them below 1800 than above. It's also readily apparent that the stuff to be learned by a 1200 player is much more likely to be simple tactics & easy to explain training methods than what a 2000 player would need.
I definitely agree with the idea that you can pick up "bad habits" by "training" with weaker players, which is one reason I dread entering high priced events occasionally after months of club play even when I seem to be doing well.
One glimmer of hope for the "more settled folk" that I've observed is that the "quality" of your games may be a more reliable indicator of your true strength than your results alone. At tournaments, I often watch other players' games that I consider to be particularly interesting & a while back I found myself repeatedly impressed by (surprise) an older player who typically finished near the top of his section but rarely first place yet almost always played great chess. He seemed to handle both tactics & positional play very well, could squeeze a lot out of a "lost" or "drawn" ending & (are you ready for this) didn't seem to be much of an opening theory monster from what I could tell. If you look at his rating history, you'll see what looks almost exactly like a typical graph of the performance of a young player through his/her high school & college years into their 20's. The only difference is he was in his 60's (born 1930) when he did this!

ATH2044 said...

Yikes, the es_trick post went up before I finished typing! It looks like John Maynard Keynes was right.
"In the long run we are all dead."
No doubt external extraneous factors may partly explain why older people don't improve as consistently, but I'd like to think that some of the degenerative physiological processes involved can at least be slowed. Psychological factors also weigh against plasticity to a great extent. I think people develop habits because they gravitate towards what works best for them (but I'm open to diverging opinions).

Howard Goldowsky said...


Actually, I played David Tylevich (the owner of the rating chart you present) once (I lost), and I believe that he is a Soviet immigrant; his first non-provisional rating was actually 2070, and he was probably close to USCF master strength before he came to the U.S. This is just a guess, based on my memory of him, but you present a good example nevertheless.

Good point about Wetzell and his book. The guy actually plays at my club (Metrowest CC, Natick, MA). Unfortunately, he's back to the 2000 to 2050 range, but he's still a tough player at around 70 y/o.

es_trick presents some interesting data: excellent stuff.

Yes, the USCF uses the 35-45 y/o crowd to "stabalize" the rating system, based on the observation that this age group shows the greatest rating stability compared to other 10-year-range age groups. So there certainly is this data/observation that this age range is the most stable (or least likely to improve). But that doesn't mean there are no exceptions! :)

I also agree that the quality of your games need to somehow become a performance metric, because wins and losses are determined by an amalgam of factors. By isolating different performance metrics (openings, tactics, positional play, time management, even specific types of tactical motives), one can isolate their strengths and weaknesses from results. This is where a coach comes in. And now we have come full circle back to the original topic!


Elizabeth Vicary said...

"First the "dirty little secret" of chess ratings that "Chess skill is non-linear." is pretty obvious to anyone who's ever played, but the rating formula (which is anything but linear) supports this as well."

The rating formula supports the argument that chess mastery is nonlinear? I don't think that makes any sense.

"One glimmer of hope for the "more settled folk" that I've observed is that the "quality" of your games may be a more reliable indicator of your true strength than your results alone."

That also only makes sense if you change the meaning of strength from "able to win games" to "able to think of yourself as a strong player."

es-trick-- thanks! nice to see some empirically based thinking!

Elizabeth Vicary said...
This comment has been removed by the author.
Howard Goldowsky said...

I shouldn't say that the 35-45 y/o age range is the least likely to improve. They are merely the most stable. The least likely to improve would probably be the oldest members, because, as experience shows, they are the most likely to decrease in skill. Thus, improvement for them is toughest. Arpad Elo, in his book The Rating of Chessplayers Past and Present, did a simple study of rating performance of masters as they age. Jeff Sonas' also has some graphs for this. Most players start to lose skill around 60+.

So I guess there is definitely a data trend that suggests the older you get the harder it is to improve, but there still has not been a study done with USCF data. Perhaps it is easy to retroactively track each 25-45 y/o USCF player, and see how much they improved over their USCF career, as a function of age. The results would be the following: on the x-axis we would have age; on the y-axis we would have average rating gain/loss from this age to, say, age 50. We could also track how many players made master, starting at various ages and ratings.


Howard Goldowsky said...

The ability to win games is the ultimate performance metric, and this, of course, is mirrored in ones rating; however, a player may be a 1600-level player in five out of six aspects of his game, and a 1000-level player in one aspect of his game (say, tactics), which brings his rating to, say, 1100. All ATH is saying is that if one can isolate the weaknesses in ones game, the player may make a performance leap. Thus, assessing the quality of ones games is another way of saying "finding ones weaknesses." This is a good reason to hire a coach, and why coaches may help us older folk.

The rating formula is nonlinear. A 200 point rating difference means that the higher rated player will win 75% of the time (1 game out of 4); a 400 point difference means that the higher rated player will win about 95% of the time (1 game out of 20); a 600 point difference means like 99.9% (1 game out of 100+). This is nonlinear. Win expectancies are relative, based on rating differences, not absolute values. This is a nonlinear process.


es_trick said...

". . .each 200 rating points represents approximately a doubling of the amount of knowledge involved. Rolf suggests it's more like every 100 points!"

I recently made the argument (elsewhere) that a player rated 400points higher than another probably has an exponentially greater amount of chess skill and knowledge. That fits right between your supposition and Wetzell's.

Using Wetzell's assertion, a 300point increase could be represented as 2 to the 3rd power (a factor of 8) while a 400 point rating increase would be 2 to the 4th power, or 16 times as much skill and knowledge on the part of the higher rated player.

My idea of an exponent in base 10 would be higher than your formula of 200 points = double the knowledge, but less than Wetzell's formula. (Of course, that's all guess work, on everyone's part.)

But what to make of the child prodigies? They can't possibly have acquired the same amount of "knowledge" that an equally high rated player has spent many years acquiring. Their skill seems to come from the ability to perceive and apprehend a tremendous amount of things that we mortals must have input as "knowledge" into our brains byte by byte.

ATH2044 said...

Nice try Liz,
"The rating formula supports the argument that chess mastery is nonlinear? I don't think that makes any sense."
The reason that doesn't make any sense is because it's not what I said.
A.) I never used the word "mastery". &
B.) The astute reader (i.e. someone other than Angie Tempura) might observe where I placed the quotes within my paragraph.
First the "dirty little secret" of chess ratings that "Chess skill is non-linear."
The quotes around "Chess skill is non-linear." were placed there because I believe the OP intended to say something like "Chess ratings as measured by the current rating formula have a non-linear relationship to chess skill."
Yes, this is an interpretation, but I then mention the actual formula, which doesn't conveniently fit into the familiar y = mx + b model, is certainly a non-linear relationship.
This makes perfect sense once we recognize that a typical 2500 rated player isn't simply "about 2.5 times stronger than a typical 1000 rated player". I suspect if each win resulted in an equal point gain, then GM's would have much higher ratings.

On that other thing about the definition of "strength":
Your unnecessarily narrow statement:
"That also only makes sense if you change the meaning of strength from 'able to win games' to 'able to think of yourself as a strong player.' " also rules out any possibility for inconsistencies, or mismatches of specific skill sets. The rating formula mostly works most of the time, but is particularly anemic in the area of tracking instantaneous playing strength/weakness as well as specific skill mismatches & intermittent over or under performance which are of course a few of the main reasons why occasionally a 2200+ player actually loses a game to a 1600 player. (Yes, I'm suggesting that it happens more than the formula predicts & yes that's at the extremes of the formula.) The rating formula also doesn't correctly reward/punish draws, but that's another topic.
If the rating formula were an accurate representation of your only allowed criterion, 'able to win games', then one might suggest saving (the aggravation of) an entire weekend (of all that chess stuff) by simply sorting all the tournament entrants in decreasing order of rating & handing out prizes accordingly.

ATH2044 said...

Angie Tempura has moved to here, sorry.

Anonymous said...

Fascinating stuff. One quibble:

"If, say, approximately 1% of adult class-A players make it to master, how is this rare? This is just the norm, because it is difficult to make master, in general."

I hate to add to the off-the-cuff generalizations, but that 1% figure seems really high to me. I've been playing on and off for 30 years, and I can think of one player who fits this description.

Believe me, I wish I could think of more. It is finally dawning on me that, while I picked myself up off my 1600 floor pretty quickly when I started playing seriously again almost three years ago, I've been in the mid-1700s for nearly two years and I'm 44, so 2000 is probably not going to happen. Certainly losing to two 1400 players in a row is not something someone destined for 2000 would do.

Rick Massimo

Anonymous said...

I just saw a really good documentary, Billy the Kid, directed by Jennifer Venditti in 2007. It's about a teen with special needs. He might have Asperger's, I'm not sure. It's such a well-made, thoughtful film.