Show HN: Memory Chimp – How good is your working memory? | Hacker News Reader

Show HN: Memory Chimp – How good is your working memory?

awendt.github.io

53 pointsawendt53 comments

  Uncaught SecurityError: Failed to read the 'localStorage' property from 'Window': Access is denied for this document.

I feel like this happens on the majority of the demos on the HN frontpage too. I just tried searching for what percentage of users disable cookies, and it seems like ~10% is a common figure. (Though I got numbers as low a ~1%, and a lot of confusion about third-party cookies.) I know this isn't technically a cookie, but I think most of the major browser vendors lump it under the same preference.

Anyways, hackers and demo writers: degrade gracefully. On first brush with your random website, I might not want to let you track me. (I don't know what you're using this for.) Show me value. Or at least tell me why you need this.

Thank you for reporting this, I've create an issue: https://github.com/awendt/memory-chimp/issues/1. To explain why I do this: There's one write to localStorage, to persist your personal best across sessions.

In my defense, I did read the guidelines before posting this and it explicitly said:

> HN users are comfortable with work that's at an early stage.

This is my first "Show HN." I'll put "graceful degradation" on my list for next time :)

This is now fixed.

Remembering a pattern seems easier than remembering the actual order. Maybe it's because of years of using pattern unlock on my phone.

Very cool. Is there any theoretical basis to it? Some research? Does the ability to solve this test correlate with other qualities?

See http://www.youtube.com/watch?v=zsXP8qeFF6A for a video that made me build this.

Google for "ayumu chimp" for a number of articles on the subject.

"it's really very very difficult"

Um... no. Cool concept, but I find it hard to see how someone could find this hard though. I also don't see how this is a sign of "intelligence"

Just imagine the line the points draw, and when you have to split it into one directional patterns you can easily remember..

Ps. Can we get a version where it's just 9 circles?

I can't see any mention of this in the post or on the page where it claims this is an indicator for intelligence.

now, since we're on the topic, we probably need to discuss whether or not this could be an indicator for intelligence.

I don't know how much research has gone into seeing whether or not there's a link between intelligence and working memory, but according to wikipedia [1], memory is one of the things that encompasses intelligence, so there might be a link.

1 http://en.wikipedia.org/wiki/Intelligence

The very first lines in the video refers to intelligence:

> The scientists […] have devised intelligence tests for chimps. One test is for what's called working memory.

http://www.gwern.net/DNB%20FAQ

That's incredible how quickly and consistently the chimp could do it.

I didn't find it difficult, but I still had to take a bit of time to look for a pattern and review the pattern a couple times in my mind before I started. Yet, it's clear for the chimp there is no 'thinking' about it going on. Trying to imagine what it would be like to experience it as a chimp.

I could use some help over at https://github.com/awendt/memory-chimp

I'd like to use a fixed-width layout like 2048 has but cannot get it to work. The code is in a feature branch, https://github.com/awendt/memory-chimp/tree/fixed-width. I'm having trouble that the viewport is scrolling horizontally and I don't know why.

I was hoping to get rid of the zooming bug using a fixed-width layout together with the meta tag "viewport". That new layout could also be a chance to suggest users to try it on touch devices.

Just forked, I have some other ideas also.

Are there any proven methods for improving working memory?

I forget so much of what I hear day to day, and think I could be far more effective in most regards if I could remember (something from the news, what a certain client said about X, or name of the cool company I read about etc.).

I've read several books about memory techniques, and can use the techniques actively to remember a number or a list. That is cool, however I really want to improve my working memory without needing to actively try to remember everything.

From what I've roughly read, all behavioral interventions (such as n-back training) on working memory and its correlates or antecedents are basically disappointments, whether the effect size is too boring or the effects are not generalizable.

I did this the first time no problem. I didn't memorize the order itself, but instead memorized sequential pairs (1-2, 3-4, etc...). I think memorizing each one individually would have been more difficult.

Working memory is a fascinating concept. I've played with techniques like the Memory palace and found them to be wildly effective. Attaching a meaning to any piece of information seems to make it much easier to remember.

Y'all need to help me with the next levels I have planned: The numbers will only appear for a limited time...

From http://www.livescience.com/7444-chimps-numbers-humans.html:

> However, they discovered the three young chimps could remember many numerals with a glance, with virtually no change in performance even when the numbers were flashed for just 210 milliseconds

I think it would be a lot more difficult if the numbers actually moved. Our brains are very spatially oriented, making remembering positions or lines (as sillysaurus3 pointed out) much easier.

Would help out myself, but I'm on vacation and it's late here.

You're right, memorizing pairs makes it easier. There's a trick that seemed to work even better: Trace lines with your eyes and then memorize that. I found myself memorizing "up down up down up down left right right" by the end, like some old Nintendo game, which was easier than memorizing pairs.

I just remembered "swirls" of numbers (2s and 3s) that made a sweeping pattern (alternating or straight.) It worked so well I passed first time too!

That sounds like an interesting method! It seems that attaching some kind of spatial meaning to the sequence makes it a lot easier.

I think everyone got them all, I thought maybe the point was how quickly you could do it?

Did the same thing but with pairs of 3 (1-2-3, 4-5-6..)

1910 on the first try. 2614 on the second. Can't get past 2741.

Here is a trick. It is very easy to rememeber up to 6 spots, but over that becomes hard. Split the numbers into groups and remember the patterns for each one and where the transition to the next group happens.

OP and author here. Over time, you'll get better. A lot.

8 spots is okay for me now, used to be next to impossible at first.

Have you found this to increase your working memory in other contexts as well?

Unfortunately, no :-/

dammit human brain too flexible

Got some high scores initially from drawing the pattern with my mouse and letting the pattern "burn in" to my brain.

BTW, points are awarded for the time you take from the first to the last spot.

Thus, using a mouse costs you points. Better play it on a touch device.

So memorization time isn't taken into account?

No, it's not. Should I start the clock right when the new level is displayed instead?

Hangs on iPad+Chrome after a few levels (the background/banana shows up but no numbers), but otherwise a fun game.

Yes, I've seen this sporadically. Problem is: The algorithm keeps a list of empty rectangles to draw the next circle in and sometimes the list is empty before all circles are drawn.

I'll try to catch that and probably just attempt a redraw...

Issue: https://github.com/awendt/memory-chimp/issues/3

http://cl.ly/Xa4f/Image%202014-09-17%20at%203.39.34%20pm.png

I used my phone camera to help on last step. :)

Very cool. But one thing: please prevent zooming. Sometimes after tapping a circle I accidently tap it twice, the site zooms in and I instantly miss the next one :)

Darn, I've heard that before. But it never happened to me so I forgot all about it.

Can you send a pull request or give me any pointers how to prevent it?

Just add <meta name="viewport" content="initial-scale=1.0, user-scalable=no" /> to the <head>. Cool game!

Thanks. However, I'm having layout troubles with that. See https://news.ycombinator.com/item?id=8329031 for details.

It just says "Congratulations, you won!" after 9 numbers; why not more levels?

Also, is the time needed to remember the numbers measured or not?

I saw this game in an exhibition about monkeys, and it stopped at 9 numbers. The limit is arbitrary and could be increased. But at some point, the algorithm drawing the numbers may run out of screen estate.

You can take any time you need to remember the numbers, it does not matter. However, I'm planning to add more levels that only show the numbers briefly.

> I'm planning to add more levels that only show the numbers briefly

I think it would be better (and simpler) to just decrease the potential score according to the time the player takes to look at the numbers (time of first click minus time the level was first displayed).

Also, I played several games and ended with a different score each time; if time doesn't matter, what does the score come from?

The score depends on how fast you can get from 1 to x. A stopwatch is started when you tap 1 and stopped when you tap the last number.

If you improve in each level and remain a constant rate, you get more points in higher levels. Suppose you take 1 second to finish each level, your points in all levels will be: 320, 500, 720, 980, 1280, 1620.

See https://github.com/awendt/memory-chimp/blob/master/game.js#L...

Ok; I wanted to try again and click super-fast, but the game stopped between levels 8 and 9 with the error "Uncaught TypeError: Cannot read property 'width' of undefined" in board.js, line 48

I'm on Chrome 39.0.2150.5 dev-m on Windows Vista on a 1280x800 laptop.

Can't seem to do better than 1770; but isn't what you're measuring mouse ability rather than memory... ;-)

Nice game. I finished the game. Had some trouble with the last level. What I memorized was what my hand movements (mouse) needed to do.

Thats exactly how I did it except I visualised a line joining them together in order and then "ran" the line.

I memorised by memorising the movement my hands need to do. Nice game!

Nice game, André! Would you mind if I to port this to Android?

I would also like that. I would pay.

i'd like to regularly & longitudinally evaluate my own cognitive performance.

can anyone recommend an app which does this?

http://www.cambridgebrainsciences.com/

seems like a clone of the lumosity game, which features a chip in space playing for poker chips using memory recall

Good Game.!

Waiting for more levels..

Reply on news.ycombinator.com