Short-term visual memory training

Saccade width pixels.
Number of symbols symbols.

S (novice)
S/T (novice)
S/T/U (intermediate)
STUX (expert)
STUXL (expert)

We are building a mobile app and want your help! To join our beta testing please email Please notice the web page is built for the initial experience and is not appropriate for daily use.

In short term visual memory training you see several letters in front of you for a second, and then you need to count the number of occurrences of each letter. The more letter of more types you need to count over a larger area, the harder the exercise. You can choose which part you want to work on (saccade size, capacity of memory, accuracy of memory). A normal superlearner should be able to read a flash of 9 letters of 3 types over saccade of 400.
Do not count the letters from the screen: form a mental image of the screen and count the letters from the mental image. Practice for 10 min per day and after several months, you will have better photographic memory.

Get 4 Free Sample Chapters of the Key To Study Book

Get access to advanced training, and a selection of free apps to train your reading speed and visual memory

You have Successfully Subscribed!

219 Replies to “Short-term visual memory training”

  1. Would it be possible to leave the answer fields blank (empty string) ? I have a lot of trouble using it on my other devices because of the zero always being there… Thx for reply

      1. I have the same issue.
        Just try the exercise on an iphone for example.
        It is very hard to fill in the numbers, because of the magnifying glass and pointer movement.
        If you click the box the pointer “|” would be set in front of the 0
        So it would look like this |0.
        Then you will have to move the pointer to the right side of the 0 to delete it. Since there is no “DEL” button on the Keyboard.
        You would have to do this for every Checkbox!

        Leaving the fields blank would make it way easier at least for an iphone user. I can not speak for android though.

        Thank you once again.

        1. Sorry, these exercises are not intended to be used with mobile devices.I plan to build eventually mobile apps just for that purpose, but do not have resources right now.

          1. The same exact issue is relevant to any device one could use, however. It is quite tedious to have to re-type certain fields because the caret was placed before the 0 that is pre-entered into the field.

            Leaving the field blank would be a huge improvement for any device, not just mobile devices.

    1. When I try to use this tool the bottom part of the box disappears
      under the bottom of the screen. Is there an answer to this?

    1. I am trying to write exercises that address different skills first. The way Anna teaches is trying little bits of all exercises and not focusing on one exercise till perfection. Definitely there is also an incentive to address or A/B test various aspects of the same skill, e.g. using colors and shapes instead of letters, various forms of animation etc. I will be releasing more exercises very soon.

  2. Would be really great (I could squeeze more into my 15 minutes of training) if once you press FLASH, see the characters, and then input your answers, the previous parameter settings are maintained (currently it falls back to the standard 4 symbols, intermediate). Would be much more productive to me!

    Nice work, thanks

  3. Hi,
    Is there anyway to isolate the flash exercise from the rest of this page? I am unable to carry out the exercise properly on my 13-inch laptop simply because the screen length is too small and covers the bottom half of the box.

    Thank you

    1. The exercise is PHP embedded in wordpress plugin and not flash, so there is no “full screen” support. I will think if I can preserve the screen scroll position instead. Please provide me the exact dimensions (resolution) of your screen and I will try to simulate the situation.

    2. i had the same problem but you can fix it by changing to full screen on your browser or in the function keys. on my Dell it is F11.

  4. Hey, I notice sometimes the letters overlap each other, making it extra difficult to figure out what letter is on top of the other. When the program picks a location, could you set it that it cannot appear i within another letter’s area.

    Also, I’ve noticed that sometimes the flash is a split second (or 2) longer than other times. Was this intentional? I just find the longer visibility easier, and doable.

    Thanks πŸ™‚

    1. Dear Steve,
      These are interesting observations showing attention to detail. I will address them in lower priority right now. Please send me your mail, I might want to ask you to beta-test some other software.
      Thank you,

  5. I put the code before the text to adapt for small laptops, and the code preserves the settings. Please see if it works for you πŸ™‚

    1. Hi Lev,

      First of all, thanks a lot for working on fixing all the issues that are poping up so fast.

      After your last modifications of the code, the problem is still there. The issue is that when you press the Flash button, the whole page refreshes sending you to the top of the page again. Now you have to scroll down to check the exercise making it impossible to use.

      I think the problem is that when you press in one of the elements of the “Subjects to read about” post, you always get that post first, then the “Proposed schedule training” and finally the post you were looking for. Probably if you don’t include the first two posts and only show the one you wanted to see, the problem will be solved.

      If you want to have always visible the Subjects to read about and the Proposed schedule, maybe you cound add those items in the left column or create another column in the right including them.



      1. I noticed the issue. I am using “sticky category posts for wordpress”, unfortunately the plugin is buggy and I am getting posts sticking to the wrong places. I fix one bug and get another… I think I fixed most of it now…

  6. Regarding Steve’s comment about sometimes a few letters overlapping, it does make it more difficult, but I don’t find it harmful, since the idea is also to practice focus.

    One thing, however, Dr. Lev: when practicing with two letters S/T, at random 8 symbols, on the 400×400 pixels sq, sometimes I can, for example, count all 5 S’s real quick, and deduce the 3 T’s, or vice-versa. I’m not sure that if that counts as a right answer, since I did not actually count all the letters (read all the material). Would I be somehow cheating by working the answers like that?


      1. Hello, same thing with me. I’m doing 3 letters, 400 x 6 and I’m mostly getting 100%. I’m counting the most letters I see, and quickly look on the other 2 letters, but I’m feeling I’m doing more math. I mean, the math is helping me remember the letters better than if I wouldn’t know how many total letters there are in the game. For example, if I see 3 Us and remember seeing about 2 Ts, I can deduct there was an S also (especially if I think I saw it). Is this okay?

  7. I love the exercise. It works great on my computer.

    Question: Am I suppose to keep my eyes fixed in the middle of the square and count the letters without moving me eyes?
    I find my eyes racing across the box to where the letters are to read them but I don’t know if that defeats the purpose of the exercise.

    Thank you in advance!

  8. so, to be clear, counting the letters up and deducing is cheating or not? πŸ˜€ If yes, my life is a lie. Cause I can get the 9 letters of 3 different types, but not without deducing…

  9. Sorry… Sometimes I just do not know the right answer. Probably you just need a harder exercise. I am trying to devise such exercises using simple words and images instead of single letters. Please be patient….

    1. Perhaps if it is possible…to have an option besides choosing the number of symbols ourselves, exercise can generate random number of symbols. If I don’t know how many are there (if I haven’t chosen that number) I will have to count them all.

      1. Your recommendation is hard to implement with current code. I will generate other exercises based on this one, this is a simple and proper way to continue. I just need a quiet weekend, which I hope to get mid-October.

  10. There has been some piracy attack on my content recently. I am considering to activate legal persecution.
    Please do not duplicate my content or develop stuff based on my code without consulting me first.
    Please remove the content you already copied from public domain and contact me for further discussion.

  11. Is it possible to turn off autocomplete for the input boxes where you type in the counted values? Its very distracting.
    And I think that random generator mentioned above would be a nice upgrade.

    1. Not sure. I can add ” autocomplete=”off”” the next time I touch the code. Hope it helps…
      Thank you for motivating me. I do need to create several versions of this exercise, but the current version is SO effective…

  12. Hi, I love this exercise and I can tell within a week the renewed firing of brain engines πŸ™‚ One thing that I find could be better is the navigation. Right now I need to use the mouse and click on different places to navigate. This is inefficient and takes the focus away from task at hand. It would be great if the exercise navigation were redesigned to be completely mouse free – just by using Tab and Enter keys one could very quickly run through many iterations. Thank you!

  13. Not aligned well for the screen. It pops up and down so when I push the start button the window pops upward and part of it is hidden. So I can’t even see all the letters that are there.

    1. The way I built the exercise I do not preserve the page scroll. I wrote down your comment, but it is a low priority. Please do not scroll when playing. Sorry for inconvenience.

  14. Felt like sharing how I did this exercise.
    9 letters with S/T/U flashes too damn fast for me to count them all. To get around this, I instead I memorized the pattern of the letter types and numbers during the flash. Then afterwards start counting from the image in my mind. Then deduce 1 or 2 letters if I missed them.

    1. When you speedread you add details to markers during the saccade masking, e.g. in the millisconds when your eyes see nothing. This exercise trains the required skill.

  15. Hi Lev, thanks a lot for developping this great exercice. I have been doing this exercice but I seem to hit a wall with a flash of 5 letters of 3 types over saccade of 400.Any advice on how to get passed that?

  16. Dr. Lev,

    This exercise is awesome. The first day, I did it with the required 15 minutes only on 2 letters and built up from wherever it starts up to like 9-10-11 letters at once. I did very well. But today, which is only my second day practicing with this exercise.. I did the 3 letters, 9 points, as it is standardized and what is the average for a speed reader, and didn’t start getting 100s until after my 4th / 5th time through.

    What would be the higher level of speed readers skill set with saccade masking, in regards to how many letters and points they can saccade in the average time (2 seconds?) that the letters stay on the screen?

    Do you think it’s a good idea to go first from starting out to reaching the average, and then pushing higher in such a short period of time?

    1. That’s an interesting question.
      1. A high level speedreader can handle ~10 words per saccade, that is ~60 letters – but we do not read the letters, we recognize whole words. At 3000 wpm this means 50 words per second, and .2 sec per saccade. This is pretty consistent with .1 sec reading and .1 sec masking per saccade. At 2 sec we can get ~10 saccades.
      2. I usually put the letters on the screen for ~500 msec to handle them in 2-3 saccades (e.g. top, middle and bottom). So this exercise also includes visual angle training.
      3. In Anna’s course we run this training 2 times a week for 3 weeks. This is much less than the ~20 times each of the readers refreshes this page in a single day. I do not have enough statistics to recommend you on a strategy.

  17. I find it is better to try to visualize the text image, instead of trying to count each character. Visualize the image and count it after the image disappear. This really helps me to better intuitively visualize image, not to focus on text.

  18. Hi:D
    Thank you for making such a nice visual training game but i have a question about it.
    is it ok to deduct the rest of the letters?(for example,play with S/T and 9 symbols and if i just see 2 S, i can deduct there are 7 T)

    1. I have the same question!

      But let me put my thoughts out there.

      The answer is probably NO… AND YES.

      It’s good central executive/working memory training to practice deductions quickly, but obviously you are not training your visual memory capacity if you’re not working your ability to SEE and VISUALLY REMEMBER those characters.

      So, since this is about the latter, you’d be better off working yourself up to actually being capable of VISUALLY remembering 9 characters rather than deducing some of them.

      But perhaps Dr Gold wasn’t strict about that distinction when he set that goal, for that is a real challenge to get to. (I believe the average is just 4-5 items, so we’re asked to double it!)

  19. “A normal superlearner should be able to read a flash of 9 letters of 3 types over saccade of 400.” Is this the goal we should be shooting for?

    1. More of a milestone…
      You need to succeed in multiple skills, there is little point in pushing a strong skill when you are limited by other issues.

  20. Dr Gold, what a simple yet fun game! Just played it for 45 mins surprisingly.

    I’m curious about some points though. I can tell that I’m relying on audial memory to remember the number of characters I saw. Does this defeat the purpose of the exercise? I mean, should I be referring to images in my head rather than sounds (ie subvocals)?

    Also, would there be a benefit, in your opinion, in being able to adjust the amount of time the characters are presented for? It seems like it’s currently set at under 1000ms, which tells me we’re using sensory memory. (Can sensory memory even be improved?)


  21. Suggestion: I think it would help improve the game if you could set a range for how many symbols show up in the box. When ever I set the amount of letters, most of the time I find myself trying to know how many of one symbol, and then add up the others to make the total number of symbols that I put in the box. If you put a range instead of the set amount it creates a greater need to focus to get the %100.

    1. I disagree with you here. When you read a text there is a prereading response that functions similar to 100% focus. Prereading is a very strong tool and I want to encourage using it. If you train yourself to have some prerequisite before you start reading, this will force you to preread the texts later on.

  22. A little bug in the latest addition: In one of the Expert levels: STUX, the letter that appears is the L instead of the X while the result asks to fill the number of X.

  23. I’m not sure if I am doing it right, I’m on day 2 of my training schedule, up to 7 symbols, with 3 types and 400px saccade. 80 – 100% accuracy

    At the moment I try and count the symbols in the second they flash, should I be trying to scan the area and count in the second, or should I try and recall the image after the flash and count the symbols on the recalled flash?

  24. Hi Lev,

    Just wanted if there are any tips on increasing your visual short term memory.

    I am having trouble progressing past 5 characters and 9 symbols.

    I’ve done characters 2 – 4 with 8 symbols, and all the character options for 1 – 7 symbols, only progressing once I’m able to comfortably get 3-4 100% scores in a row without guessing of summing up the characters to reach the right number of characters.

    It seems I have become a little stuck with this and do not want to go on without mastering this element.

    My eyes seem to have trouble taking in all the characters in a glance.


    1. 1. I suggest each time you increase number of symbols decrease the number of characters to 2, and slowly increase it back to 4
      2. If you get 9 symbols, this means you are maximizing your “working table” memory. The key above 9 is to take a “snapshot” of the screen and read characters from your visual memory. This is similar to processing we do between saccades.
      3. If you cannot snapshot the whole screen in one glace you could reduce the working area from 400(?)px to smaller number and
      4. There is no reason to obsess over one exercise, even a very good one. Continue with other exercises and come back after a week. You will get “unstuck”

  25. Is it wrong to see these words and assemble them into a word type structure, even though these aren’t real words of course. Example: It flashes S U T AND X randomly. If i were to put those together as SUTX, then input it as 1 S, 1 U, 1 T, and 1 X, is that the wrong mentality to have when doing this exercise? Same with S U T T if i make a word like Sutt?

    1. This is an interesting question. Some people [especially people with a talent for language studies] find story method or “sounds like” method very useful for markers. It is OK even to vocalize 2-3 words out of each paragraph as markers. In speed vs retention tradeoff this trick moves you closer to retention, which is perfectly OK as far as I am concerned.

  26. Hello,

    I feel like I am doing this exercise incorrectly. I tend to quickly scan all the letters on the the screen and quickly determine which letters have a smaller number and count those only and inference the amount for the rest of the letter. I have been trying to force myself to count them but it is very hard for me to not apply this pattern. It is almost like a habit.

    I have a suggestion which is that the user specify the range (max & min) for the number of symbols. This way the user is more likely forced to count the number of letters and can’t rely on inferencing.

    Also, thanks for creating this course. Just started it and am looking forward to see how I progress in the coming weeks

    1. Basically you should remember the pattern you see and count the letters from memory. It is OK to use inference between letters, this is similar to focusing on links between markers and their structure. So do not try to change your routine, but increase amount of characters you see.

  27. This works well on my P C, however, on the ipad, a 2 second screen picture is like a blink, a nano second, sometimes going directly to the answer screen. I am assuming this is only designed for P C’s?

    1. It should be responsive, e.g. work everywhere. Works on my devices…. I test it on Chrome, so there may be some browser compatibility issues.

  28. When i do the exercise say 400 pixel, 10 symbol s/t i see say 7 s , but don’t see or catch the 3 ts but know there are 3 more so it should be 3 ts. Is that cheating or is the main point to see all 10 symbols? If so is there a trick to it or just need to work on it daily so see all 10?

    1. This exercise has several complexity levels. If 2 symbols are too easy for you, increase the number of symbols STUXL.

  29. Hi,

    i have a big problem to make a use of this exercise on my 15` laptop 1600:900.
    Every time i click “flash” button i need to use scroll to catch the widget …
    I can ofc reduce the size of everything by ctr+scroll but than i cant see letters

    on my tablet where i can use screen upright its much better ofc, but …
    maybe it is posible to export this game flash on a new landing page without half of screen taken by headlines and textures?

    i think it could help a lot

      1. @ Konsnos
        What you mean by “fullscreen” mode?

        @ Lev
        Thank you.
        Anyway, i have one thought that keep me awake.

        As i am trying to train my spatial visual memory. I noticed that its easier for me not to focus on letters “one by one” but to “catch a big picture” – with all letters at once (without moving an eyeball).
        But than i lose a lot of quality and i cant sometimes recognize symbols.

        Maybe i am doing something wrong?

        1. In Chrome by pressing F11 the browser goes to fullscreen, giving you more space for the page. Other browsers must have a shortcut for fullscreen mode too.

        2. Saccade training here.
          Need to capture ~4 pictures, not one big one.
          Try to divide the image into squares.
          I am trying to write strategies into workbook. The work goes significantly slower than anticipated…

  30. This is a really great and addictive exercise πŸ™‚
    One little suggestion though:
    When warming up I usually start with 7-9 S/T letters.
    The problem there is once you have all the S letters for example together, you can just calculate the rest without really thinking what you saw. That happens to me a lot.
    Maybe you can put in an option so that the number for two letters varies by -1-2. So that one does not know how many will be displayed, and so is forced to always think visual not mathematical πŸ™‚

    Just a suggestion.
    Great work though!
    Really admire your effort!

    1. Thank you. This is an interesting concept, but most people do not like uncertainty. You should warm up with STU instead πŸ™‚

  31. Pressing Flash button reloads the whole page every time! This is at least 3 seconds of wasted time! Please write it in java script, remember the options and avoid reloading the whole page!

    1. I think this is not a problem for PCs due to fast connection and hardware [please correct me if I am wrong]
      For mobile, I prefer to develop a small app… This is something we definitely consider doing.

  32. Any chance that you could make a memory box that keeps an average of the percentage when we play, that also has a button that can be cleared and reset? This will help to keep track of our average result when we play. Thanks.

  33. When should we consider this skill at a reasonable level? From the instructions, “a normal superlearner should be able to read a flash of 9 letters of 3 types over saccade of 400”. I am able to do this, so I am uncertain if it is best for me to keep practicing with 10+ letters and/or 4+types, or to continue with other exercises and occasionally assess my skill level with this SVTM exercise. Thoughts?

  34. When i do this with 10 leters or more, normaly i can’t get it. But when i close my eyes imidietly after seeing the picture i se picture in my head and i can coun’t it like the picture is in front of me. Why is that? And is it a good thing or should i be able to quickly see and remember?

  35. When watching the SuperLearner lecture, the instructor states that he sees different groups of 4 to 5 letters…like “s,t,t,s,t”. My eyes actually trace a line of the same character, breaks it into reasonable portions like say all the Ts on a screen form sort of a V and there are 5 of them, I see a v if Ts across the screen broken up into groups of 2 and 3. This happens naturally and is easy for me. Do I need to retrain myself to see different groups of letters bunched together in different areas of the screen and try to remember what I saw like the instructor does?

    1. Interesting. I think your technique is good. Not sure how it will work with more letters – you may need to use some hybrid method.

  36. I have started doing this exercise. I am able to do 3 characters, 400 saccade, 9 no. Of symbols. But from last 2 days I am struck on that. I think my fault is that I try counting out in the second of visualisation. Should I count or should I just visualise. Kindly answer coz my problem regarding reading is also subvocalisation. That’s bothering me here also.

  37. Hey team,

    Sorry if this has already been covered. Is there an upper limit we should be aiming for before moving on from this exercise? I’m fairly comfortable on 500-600 pixel, 10-12 characters of 5 different kinds. I’ve set my personal goal at 20 characters but wonder whether my time might be better invested in other areas?

    1. This has never been asked. When you look at text, can you remember 3 lines of it photographically for a couple of seconds?
      Eidetic memory allows fewer saccades and faster reading [if you want to go above 1000wpm]
      This exercise will help.

  38. I imagine that this has already been asked, But I want to check the accuracy of the mental image I am making, I have High accuracy on remembering the letters and their quantity by using the techniques taught, I would however like to also compare my image with what I actually saw, could you please post the image that we saw next to the accuracy box after you input the quantities?

  39. Judging by other posts, it seems that if I’m “counting” one of the letters, and subtracting that from the total to get the number of the other letter, then we should step up to intermediate (or expert depending on where you are).

    Something that I’m a little unsure of, is that I’m hitting 100% a lot of the time on intermediate with 7 symbols, but sometimes hitting below that. Does that mean I’m at the correct settings? Or should I figure out what is causing me to drop below 100%, and drop that setting down slightly?
    I’m not sure if I’m actually working just within my comfort zone, or just outside, so not quite sure when to bump up each setting. My assumption is we should be just outside of the comfort zone…

    I think it putting in a little guide at the beginning just to explain when you are know you are ready to move onto the next level, or bump up the number of symbols, would be great.

  40. Hello Mr. Goldentouch, thank you for offering this short-term visual memory training for free. I started recently to learn with it. Unfortunately the trainer changed in the last 24 hours, so the field is bigger than my screen.

    Did you make any changes to the script?

    Best regards
    Luis Arteaga

  41. Hi, I have tried to remember the visual image, but it did not work. In order to remember the image, I tend to subvocalize the group of letters such as SSU or TSS. I feel that since it is not an vivid thing like a flower or a flag, I have to subvocalize the letters in order to remember them. Can you suggest me a solution for that please?
    Thank you.

  42. Hi Lev,

    Thank you for creating this activity. One question I had (and perhaps I am missing the point), I find myself knowing how many of one letter I saw, and then subtracting that from the total to know how many of the other letter(s) there are. I feel like this is problematic because I’m not really developing my ability to remember the picture, but rather am focusing on one single aspect.

    Would it be possible to put in a randomize feature, either for the number of characters, or for the characters themselves. For example, you would tell the system you want a range (4-7 characters) and you don’t know how many total there are. Alternatively maybe randomizing the characters, so that you have to say both how many characters you saw and also what they were (two entry fields required instead of 1).

    In either case, thank you for all of the work that you, Anna, and Jonathan have put into this course!

    1. Hi Ryan.. U relieved me by asking this. I am also fighting with the same que. The mind calculates by substraction. What to do ?? Is it right ??

      1. As my experience, you should increase the number of letter type up to 4. That way you can’t do the subtraction unless you know the quantity of other letters which is good enough to make you remember the picture.

  43. do you suggest listening to any music while performing this exercise or would music distract from what we are trying to do? Also we need to be chunking them together, where should i be looking?

  44. Am I allowed to move my eyes from place to place, or am I supposed to look only at the centre of the box?

    1. In this particular exercise you can do whatever you want, but if you minimize eye movement you will improve performance.

  45. Is it possible to randomise the number of letters that appear? e.g. instead of saying “7” … and then i know i am seeing 7 letters …. could we say between “6 and 9”, this way we don’t know the number of letters before seeing the flash. This would definitely add a level of complexity to the exercise

  46. I’m forming my mental image but not exactly counting all of the occurrence of each letter. There are frequent times when I can only remember and count 2 letters from my mental. I can get the count of the 3rd letter by subtracting the total of those 2 letters from my total characters so I can still get perfect score.

    Is this an acceptable strategy? Should I aim to remember all the characters (in a clustered mode) in my mental image?

    1. It is OK for beginner level. When you choose more symbols you will be forced to remember everything.

  47. I’m getting longer and shorter flashes, sometimes no flash at all! Are the flashtimes supposed to be equally long and if so, is there a solution like downloadable script/software. Let me guess, the app? πŸ˜‰

    1. Totally. I cannot solve your connection stability. Basically, if you use PC with reasonable connection speed and memory, there should be no issues… Yes, we do want to offer the app….

  48. How can you chunk these letters? There is no time to think about how to chuck them. There is only time to remember en recall.

    1. Start with few letters. As you increase the number of letters and train, you will see the letters in geometric patterns. These patterns are your basic chunking mechanism.

    1. I am not an athlete, I do not need to train. When I do my research, I am pretty fast. I think I read 3000wpm last time I measured, and I am not trying to go beyond this speed.

  49. the app(on novice mode) always shows the max qty.
    this app to be useful should run through a series of screen displays.
    say 2(novice) or 3 or 5 or 10(hardest).
    also speed setting as well.

  50. I average nearly 80 – 100 percent on a 9 letters S/T/U level should I move on to S/T/U/X and still keep this exercise active on a daily basis or should I stick with S/T/U till I get 100 percent accuracy on it. It is hard to determine if you have acquired mental improvements since there is no way to monitor this factor unless you write or remember each round percentage it may be useful to add a login that allows you to monitor your daily progression. It may also be helpful to also place a timer on this because 10 min is not long but it is quite easy to get lost in trying to master it.

  51. Great exercise! Thanks.
    So 9 letters of 3 types at what wpm skill correspond, more or less (leaving aside comprehension and recall for a moment)?

  52. Hello, i just came across this exercise game and just wondered the total accurracy value doesnt make any sense sometimes. How is it calculated in case of false values?

    1. Number of symbols: 4
      Level: S/T novice
      Accuracy in S: (1/3),
      Accuracy in T: (3/1),
      Total accuracy: 33.333333333333%
      Accuracy in S: (2/1),
      Accuracy in T: (4/3),
      Total accuracy: 66.666666666667%
      Accuracy in S: (7/3),
      Accuracy in T: (1/1),
      Total accuracy: 50%

      1. Try to get the whole image in one glance, and not follow the letters. 90% @ 7 letters is realistic for one week of training.

    2. There may be bugs in computation. We are in the process of moving to mobile version with a different code. Focus on memory not the score.

  53. When I’m on 8 symbols (3 letters), most of the time Im not been able to grasp the whole picture before its gone, its too fast for me.. usually when I get 100% its because I get right 2 letters and the rest is what left. How can I improve that?
    I’ve also noticed that couple of minutes into the drill Im getting weaker

    1. Try generating/recognizing patterns. If the symbols appear independent there is only so much you can grasp.
      Also, the visual angle may be not optimal for you. You need to get the whole picture with peripheral vision without scanning with your eyes.
      There are specifical exercises for visual angle….

  54. Hi Dr Lev, I do face the same problem of gravitating towards S and T, getting those two right and subtracting to get U. Despite focusing in the middle of the screen I still naturally notice S and T first. Is there any way to remedy this? Thanks.

    1. No need. This is a good strategy. When we memorize text we do not need to remember EVERYTHING, but just enough to reconstruct the whole text.

  55. My accuracy is 80%-100% on 400/6/TS but after 4-5 minutes of practice on same level it become much harder and I my accuracy reduce to 30%-50%, any idea why? what should I do about it?

    1. Probably you are thinking about every step. Once you get used to the process it should be automatic (almost effortless). Try to “go with the flow” and see what happens.

  56. The way I tend to play the game is a bit of and i’m not entirely sure if its the correct way. I imagine weird shapes around the same letters.
    For instance if there are 4 letters (T and S) one S is in between the two T’s and the other is in the far right corner of the square i’ll imagine a weird shape connecting the 2 T’s by overlapping the S, and another shape connecting the S in between the T’s and the S in the far right corner on the screen.
    (They are not necessarily shapes but doodles weaving around the connecting the same letters).
    Should I stop doing performing the exercise this way or continue ?

    1. I’m curious about this too. I actually do the same thing, and I don’t think it’s the same as the strategy found in the KeyToStudy schools. Is this really a good strategy that gets at the heart of what this exercise is trying to train?

      1. Not sure what you mean. Please notice there are no KeyToStudy schools currently available. Anna teaches 1:1 via Skype. Jonathan uses/teaches his own classes based on a small subset of the keytostudy methodologies and some memory techniques he collected along the way.

  57. Hi, great exercise, Dr. Levi!

    I am doing it for a couple of weeks now.

    For those who want to make it faster, just press F5 + enter when the letters dissappear. As there is no progress tracking, it wonΒ΄t matter. It is helping me a lot! πŸ˜€

    Cheers, superlearners!

  58. Hi Dr. Lev,

    how long am I supposed to keep the mental picture in memory? I’m able to form a mental picture and recall the letters, but very soon (usually when I’m counting the letters, or right after I’m done) the mental image is gone, so I guess the mental image only stays in my memory for 2-3 seconds. Is it normal or I need more exercise to better my short term memory?


    1. In this particular exercise, the mental picture should be stable for several seconds, enough to recreate what you read from the mental picture.
      When we consider speedreading, the speed is the essence. for memorization, we aim for stable recall. So we pause at the key positions (usually after each section and longer after each chapter) to strengthen the visualization, link it to other stuff we know, and even take notes on paper/computer.

  59. In the source of the answer form, please change:
    <input type="text" value="0"
    <input type="text" onfocus="if(this.value == '0') { this.value = ''; }" value="0"

    this will remove the 0 if you put in a different value instead of creating 02 for instance or 20.

    For the rest, tough but nice excercises.

  60. Hi there, I am not encouraged by my progress after one day of practice. I am usually unable to find groups of letters fast enough (level 400/6) and if I do, I often forget what I saw before I can mentally count what I saw. I realize day 1 is just starting out, but from what I read/heard from Levi’s class I thought I would be at least able to put letters into groups. Is this normal at all? Thanks!

  61. Hi, in two days, I was able to reach the intermediate level with ten word but I am not really sure if I am doing it right. To perform the task, I link all the similar letters to form a figure and then try to keep those figures in my mind. But it seems like it’s easier for me to create few figures with five or seven letters than to create several figures with two to three letters. And also, it is easier tp create figures with similar letters than figures with different letters. I am not sure if what I am doing is right or wrong. Thanks!

  62. Cannot play properly. After you click on the Flash button, the next page shows up. It is below the fold. You have to scroll down. Once you scroll down, the letters are all gone.

  63. The page is laggy…after clicking on “flash” it takes sometimes too long time to refresh and you can barely see the letters, sometimes it does it automatically. Is something wrong with the website? Perhaps if you hide the comments it would help.

    1. This blog is not optimized for gaming. I wanted to test some concepts. If you want to do some real training, please apply for mobile beta.

  64. I’ve just used this a few times. It seems like knowing the total number of letters I’m looking for makes my mind behave very differently than if I don’t know the total number of letters that I seek. I suggest (and perhaps the app has it already) randomly adding or subtracting a letter from the target number would make my mind less likely to try and game the system and hence be more realistic.

    1. If you have 5 types of letters, knowing the exact number does not really help. If you feel you are cheating, use more letters.

  65. Hi, I am having issues getting the image in single window. the image kind of shows only one half of the letters and to see other part , I have to scroll but the screen is only there for a second. What do I do to fix this ?

  66. The layout of the text fields are messed up for higher levels, because of the number of letters. Please fix.
    I have tried this one multiple OSs and browsers. Used to work just fine.

  67. I’m barely getting a flash of the image. At first there was no flash. Now there is a flash, but the flash is way too short (like a quarter of a second) and barely visible.

    1. It should be about .5 sec by design. This is supposed to be more than enough time to get a glimpse of a screen and then count from the memory. There could be some additional browser-specific issues, so you are welcome to apply for the mobile app beta testing.

  68. Dr. Lev, Call me crazy. But how about trying to recall 2 different flashes after we have seen both? Flash 1 of one type, Flash 2 of one type. Then you question the number of letters in the 2 different flashes. Thank You.

  69. Its been a week with this exercise and I am able to recognize 7 symbols (S/T/U ) most of the time and 9 symbols only about 1 in 10 tries even then I most remember atmost 2 of 3 symbols and simple calculate the third. To remember I try to draw curved lines made of symbols in my mind . Am I on the right track?

  70. Hello, may I suggest something?

    When you set novice difficulty, the number of symbols “S” should be random and not set by the player. When you put 20, 30 symbols even in novice itΒ΄s really difficult to follow up

    1. We are not maintaining the web version of exercises.
      Most of the ideas people come up with have been implemented in our mobile app version.
      Please apply for the mobile app beta testing to

Leave a Reply

Your email address will not be published. Required fields are marked *