Well, obviously there is the cultural aspect like lots of emphasis on academics and stuff since very early ages. But I have more to say.

I noticed Chinese is the most "dense" language that there is. By dense, I mean the amount of information expressed per syllable in the Chinese language is pretty high. Also, I find Japanese to be similar too. Well, technically Japanese has the Hiragana/Katakana as well as the Kanji parts (adopted Chinese characters). Kanji is pretty frequently used and is also pretty vast, which again makes me expect that there is a relatively high density of information per syllable. (I'm not sure however, googling suggests that Japanese is supposedly a lower density language, don't know why. Native Japanese speakers please correct me. However the speed at which Japanese can be enunciated is quite fast and that does help in higher speed rehearsal which is somewhat relevant to the discussion below.)

The next piece of the puzzle I want to consider is the Phonological Loop component of working memory. Working memory is the component of our cognitive function, that is responsible for holding information and manipulating it in the short term. Phonological loop specifically, is that component of working memory that is responsible for "Storing Speech/Articulation".

Solving the puzzles in competitive programming (or math olympiads) involves heavy (subjective but whatever) amounts of reasoning, planning, and navigating a complex "search tree" so to speak. For me at least, everything I reason/plan about goes through the "articulation channel". Every tiny observation/claim that I have, is an "articulated sentence" that captures the complete claim. I don't know if its just me, but I am assuming everybody constantly "talks to themselves" while solving problems.

Now, the phonological loop is known to have a very small store. Well, in fact the literature claims working memory store as a whole can only store around $$$[4,7]$$$ chunks of information at any time, even though its possible to have a better "encoding" to maximally utilise these limited chunks in terms of the semantic value held per chunk. The phonological loop also rapidly decays information unless you "rehearse"(repeat to yourself) the information that you want to keep fresh in the phonological loop. (Well, its not literally that tiny there is lots of interaction with long term memory and complimentary retrieval structures are involved. But at the bleeding edge of processing incoming information it is pretty tiny.)

All of this background makes me believe that since all the raw processing of information is subvocalised, people that think in Chinese/Japanese/Russian(also pretty complex but not sure) have some non-zero advantage/assistance in being able to deal with longer chains of thoughts. Longer chains of thoughts should imply the ability to simply have a larger "context window" should help to reason more fluidly as well.

Another thing that I want to highlight about Chinese language is the kind of symbols used. They are basically pictographs. I'm not sure but I think that gives Chinese the option to "see a word" with the inner eye along with the usual "hear the word" with their inner voice. That can also be an opportunity to sort of distribute cognitive load between the phonological loop and the visuospatial sketchpad (another part of the working memory model). For example consider the word "calculate" vs "算".

Of course there might be many-many more factors involved for why the Asians are so good at STEM in general. And of course expertise coming from practice is much more nuanced and does circumvent lots of these limitations. I just intended to put this out and get others' opinions. Also, as you may have noticed, I've used the phrase "not sure" at many places throughout the blog, because I am no expert. The intention of writing this blog, again, is primarily to hopefully to hear from some other people in the community that might have some knowledge to share or other opinions around this. I'm sure other people might have thought about this stuff, its not really some novel/original thing.

Some questions for anybody that actually reads this,
1. Is it true that your style of problem solving is indeed mostly "talking to yourself" ?
2. Do you think the ability to handle longer chains of thoughts does indeed matter ?
Of course it doesn't matter as much when you could just jot down stuff. But do you think it does matter when thinking at the bleeding edge, when you take the first look at some information/structures ?

Just comment whatever comes to your mind, I am really interested in hearing other perspectives.