Short-term visual memory training

Please notice the page is built for the initial experience and not built for daily use. The page may get very slow. We are building a mobile app. For beta testing please apply to

Saccade width pixels.
Number of symbols symbols.

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

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.

Share and Enjoy: These icons link to social bookmarking sites where readers can share and discover new web pages.
  • MisterWong
  • Y!GG
  • Webnews
  • Digg
  • StumbleUpon
  • Reddit

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!

198 thoughts on “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. 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 :)

    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 :)

    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? :D 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 :) 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.

    1. Hi,
      There is no single solution.
      Typically 500 msec is OK.
      Play with it to see what works for you.

  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 :)
    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 :)

    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 :)

  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.

Leave a Reply