My System, let's share strategies

[surcer]

I've been doing dual n back on and off since last summer, when the
paper was first published. I've been diagnosed ADHD, but am majoring
in math, so am always looking for an intellectual leg up, so I figured
why not? Since then, I've been unconsciously developing strategies. I
did not sit down and think these up rationally, they just came to me
as I was playing and seemed most natural. I wonder if anyone else had
the same experiences or new ones they use?

From DNB2-3 I use a simple, single minded strategy of counting or
replaying in my head the last letter spoken and position appeared.
When a new one comes, I discard the oldest one and replace it in my
head with that one. I call this the updating strategy, constantly
replacing the oldest with the newest in your mind and replaying them
over.

At DNB4 this gets tougher to do because as the DNB increases it's
harder to replace them in my mind while keeping the order of the ones
in between in memory. To keep the order of the ones in between for 2-3
I could quickly replay them back if needed, but here it begins to take
up too much time and I fall behind. Nevertheless, I can still usually
get through DNB4 in or two goes this way without bringing out the big
guns.

DNB5 requires a completely different strategy for me. I realized I was
making two "sets" in my brain when I started getting good at these.
The first set, or group, in my mind is taken up by the first 5. I keep
them in my mind and do not replay them after they have all appeared,
then I put them away somewhere, and use the 2nd group to remember the
next 5. During this time, I am not really paying attention to the
first group which needs to be keyed in then. Instead, I am focusing on
remembering the 2nd group. While doing that I'm able to output the
first group of 5 instinctively, intuitively, drawing on the distant
memory of the 2nd group in the back of my mind while using my
conscious focus to remember the new group. I still do not completely
understand that part.
After the second 5 have come and gone, I discard the first group from
my mind and fill it up with the new 5 I was focused on. I repeat the
process for each 5. I call this "set strategy".

DNB6 is exactly the same as DNB5, only I split the 6 into groupings of
3 for each set. I don't think this is because it's harder to draw them
out from the instinct, intuitive set, but because it's harder to focus
on and remember all 6 at the start for the first set.

This is also where I started subconsciously counting when I can stop
focusing on new n's. For example, the computer will say/show 26
letters and positions here, but one needs only to remember 20, not the
whole 26. This is true for all dual n back's. They thus can be broken
down all kinds of ways that do not utilize the remaining numbers. With
6 for instance: 4+4+4+4+4+2=26, but remember, only the first 20
matter, so we can break it into 4 sets of 5, 4+4+4+4+4=20, and ignore
the remaining 4+2, making 26, when the computer says them. I hope you
see the potential for this. Why stop at 4? I like to stick with the
sets I already have made up and not mix new ones, so for DNB6 I do:
6+6+6+2, ignoring the remaining 4 that would normally go into my first
mental set, AND ignoring the two additional 'n's the computer has
remaining completely.
This is immensely useful if used efficiently, once learned correctly,
as it frees up a ton of memory to use at the very end and allows me to
concentrate on the 2nd set.
Here's another picture explaining the process at the very of
completing DNB6:
-6 (first 1st set, first 2nd set, forgotten)
-6 (second 1st set, second 2nd set, forgotten)
-6 (third 1st set, third 2nd set, in process of outputting them)
-2 (last 1st set, can combine with third 2nd set since so few, or just
make a fourth 2nd set of 2.)
-Ignore remaining 6 since there won't be time to output them before
ending.

Summed up, there are two purposes to the ignore strategy for me:
1. Reduce the number of 'n's per set. (I will call ignore1)
2. Reduce number of overall sets since I can ignore remaining n's past
20. (I will call ignore2)
This can be very confusing in writing, so go try it for yourself in
Manual mode if you don't understand. I haven't tried it at levels
lower than 4, but I assume it could work for 3 or 4 as well.

DNB7 This always causes problems. Whereas previously if I am having a
good day, I can go thru 1-6 without problems, I can get hung up on
DNB7 for at least several rounds, maybe a lot more. I'm convinced this
is because seven is an odd number and so it's harder to divide this
one into two groups like for DNB6. I still do it anyway, and usually
find myself breaking them up by 4-3 for the first focused set. It's a
little harder to recall them intuitively, but I am unsure this is
because 7 is just larger than 6, or because in my mind the two groups
are not symmetric, and don't feel as natural. Occasionally I break
them up 3-4, but I try to avoid mixing 4-3 and 3-4 in the same round.

The ignore1 strategy is not very practical here, since the remainder
is a meager one 'n'. It would go 7+7+6, and ignore only the last one
normally put in the first set. But ignore2, like always, can be used
to ignore the last 7 n's.

DNB8 This actually is not much harder than 7. This is because 8 is
even, and this compensates moderately for the larger number. It's
exactly like DNB6, only I break them up 4-4 always. Ignore strategy is
more useful here, but not quite as useful as it was for 6 I find for
some reason. It goes 8+8+4, ignore 4 to reduce set load by 4
(ignore1), then ignore remaining 4 to reduce over all set number
(ignore2).

DNB9 This one is hell. It has the odd numbered problem of DNB7 plus
being two more than that one. This is the point where my set strategy
starts to break down because I am just unable to focus for 9 in a row
while remembering them. I break them into 5-4 or 4-5 about evenly,
even within rounds, depending on what letter 4 or 5 is and if it fits
nicer with one group or the other, it can make it worth doing.
I admit that this is a mnemonic device, and so if I catch myself, I
try not to use it, because if you just care about increasing your IQ,
I highly doubt that strategy is the way to go. The ignore strategy is
also suspect imho, but probably isn't as bad at the low levels since
it doesn't make a huge difference, and it seems kind of obvious and so
I would assume more people would have used it in the study anyway. I
don't see a problem at all with the main intuitive set strategy I use
at DNB5+, in fact it might be ideal because there is little repeating
and counting.
Anyway, I don't frequently make it to DNB9, and when I do, rarely
sustain it for more than 3-4, maybe 5 if I'm lucky, so the lack of
habitual practice is another problem.
I've made fantastic use of the ignore strategy here when I can
remember/have time to use it, since my mental resources are so
stretched, and 9 naturally lends itself to it. It goes 9+9+2, ignore1
for 7, then ignore2 for the remaining 2.

DNB10 My experience with this one is so shrift that I cannot draw
meaningful conclusions-- I've only made it here naturally twice. I've
tried it artificially a couple more times for the experience, and I
think my "set strategy" is just barely still workable, mainly because
10 is an even number, and so like 8 to 7, is not a huge jump over 9. I
divide it in 5-5 and go from there. Unfortunately, ignore1 bottoms out
here. 10+10=20, ignore 0. Ignore2 is still fabulous for another 10.

I can actually see how somehow could get to DNB11 like what the faq
says, just from being lucky on 10 while using a strategy like mine.
But surpassing 11 would be insane since it’s odd too. At that point
I'm very curious for different strategies. What do you think?

[Ashirgo]

Honestly, 10-back for me is much more difficult than 11-back, I divide
11-back on 5-back and 6-back and all goes well, but 10-back is for me
2x5-back, a true killer, I swear! :) And updating the data in my mind
is now an unconscious process.... Anyway, look for my "How to conquer
(...)" guide, I hope it can help a little.

And, most of all, have fun!

[Chris]

surcer wrote:
> This is also where I started subconsciously counting when I can stop
> focusing on new n's. For example, the computer will say/show 26
> letters and positions here, but one needs only to remember 20, not the
> whole 26.

I do think this is something of a weakness of dual n-back. I'm sure
what makes it difficult is that one has to take in new information
while simultaneously checking for matches with the old information.
But for the first n steps, one has only to take in new information,
and for the last n steps, one has only to check for matches. So for
n=7 and higher, one is having to do both simultaneously for only half
the time. In default mode, at n=10, correct memorisation of the first
n steps alone would give a 50% score, sufficient to remain at the same
level. And at n=16 and above, it would be sufficient to increase the
level.

[Pontus Granström]

I think there's research that shows that you are only able to hold 7 +/- 2 in your head. When reaching higher levels people use chunking etc. So in your case with 10-back it will become 5 back?

[Ashirgo]

Well, I can hold as much as 13 items in my head (alluding to my 13-
back experience), so we can assume that I am "different" :) , I do not
think that chunking reduces the working memory load, it is just a kind
of grouping, and in case of 10-back being divided on two 5-back parts,
these become logical parts of the whole 10-back. And well, I use
chunking even at 6 back, reducing it to two 3-backs.

And how to describe the whole idea.... You do not remember the exact
letters/positions, but the sequence, or maybe a cycle - in a form of
pattern (it is a good word), as it is mentioned in my previous guide
(picture, image). That is why it is good to envisage positions in
one's mind - not to repeat, to immediately learn them as a pattern, as
if you were to remember them for your entire life.

[Pontus Granström]

But chunking does reduce the memory load. I took a digitspan test when applying for the airforce and I scored 12, still I used chunking. There's no way of remebering 12 numbers without chunking!

[Pheonoxia]

Surcer, I'm in the process of mastering D6B and so far all you've
written is what I've already employed.

Being a math major, has DNB made math easier for you? I'd imagine it'd
affect mental math most precipitously, allowing you to solve problems
much faster. Is your GPA better this year than last year? Since you've
made some rather remarkable gains, seemingly at least doubling your
working memory, have you experienced greater fluid intelligence or any
corresponding symptoms of being smarter?

> But chunking does reduce the memory load. I took a digitspan test when
> applying for the airforce and I scored 12, still I used chunking. There's no
> way of remebering 12 numbers without chunking!

Why does the AF test digit span? What does it mean to them if you
score exceptionally high?

[Pontus Granström]

It's in the test battery you have to take if you want to apply as an airforce officer.

[sutur]

the question is if what we're doing here really is full blown
chunking. what chunking really means is that if we had the letter
sequence B-B-Q we remembered "barbecue" thus reducing three units of
information into one. but what people do while playing is more of a
grouping for convenience or to make a certain algorithm work.
algorithms greatly simplify the task at hand, but nevertheless:
mentally arranging trials into groups of four doesn't make them a
single unit of information. what i do think is that audio and visual
memory work semi-separately. that means that your dnb N, after some
practice, should only be slightly lower then the N you reach using
only one modality. that meant that according to the "magical number
seven" approach most people should be able to reach at least 7-back.
if they don't i think the problem is not the working memory per se but
some other aspect. maybe it's attention, maybe executive control,
maybe its a lack of the proper cognitive strategy, or whatever...

On May 6, 2:57 pm, Pontus Granström <lepon...@gmail.com> wrote:
> But chunking does reduce the memory load. I took a digitspan test when
> applying for the airforce and I scored 12, still I used chunking. There's no
> way of remebering 12 numbers without chunking!
>

[Pontus Granström]

To always update and work fast is a key ingredient to succeed since you always have the same amount of time to complete the task no matter your dnb-level.

[Ron Williams]

I may be wrong, but aren't all of these 'strategies' counter
productive? If chunking reduces the workload, then you're not really
gaining anything. We're not trying to 'beat' DNB - it's not really a
game, as such. Strategies just muddy the water w.r.t. whether there's
actual improvement or simply a more facile use of a strategy to avoid
doing the work!

IMHO, anyhow (and I really do mean H :) )

[Ashirgo]

Jaeggi states that the task is meant to reduce the workload, or maybe
optimize our ways of using so that we are able to improve our
performance. It is the optimisation, not the sheer power, that
matters. My two cents. :)

[Pontus Granström]

What's the reason behind the increase in Gf by the way?

[Ron Williams]

However the 'optimisation' shouldn't be by using a technology. It's
the difference between lifting 300 kg using developed muscles, and
lifting it with a forklift.

[Pheonoxia]

On May 8, 11:54 am, Ron Williams <rhwil...@gmail.com> wrote:
> However the 'optimisation' shouldn't be by using a technology. It's
> the difference between lifting 300 kg using developed muscles, and
> lifting it with a forklift.

Unless one is using some software to beat DNB, that's a terrible
analogy. You're still doing it on your own, just figuring out how to
do it better. More effective and efficient use of chunking can
transfer over to non-DNB related tasks.

[Ron Williams]

No, you're wrong - yes, you're using your own brain, however you're
not using a wider working memory - the chunking just lets you use *the
same* working memory to achieve more levels - ie. the analogy isn't
'terrible' at all. Chunking is a technology - technology doesn't have
to be a 'machine' made of metal. Do I really have to explain?....

[Pontus Granström]

 We could make the analogy of saying that everybody has the same amount of "RAM" in the brain, but some people use it more efficiently.

[surcer]

It's very hard to objectively know how much or even if one has
improved their intelligence or not without taking a test. I've done
this on and off for months, and I don't think it has impacted my math
classes very much, if at all. This is because in upper level courses
you're not doing addition or multiplication, easily done on the
calculator, which DNB might help, but other abstract things working
memory doesn't really deal with. I think I read an experiment that
showed working memory training aided in reading comprehension, can't
remember by whom though.
An interesting effect I notice is that whenever my mind is jittery,
and I can't seem to concentrate, 20 rounds of challenging DNB will
actually calm me down and allow me to focus afterward.

It's not surprising that other people are using the same strategies as
me. I think ignoring the last numbers is kind of obvious, and a DNB
weakness. I still read about people though here who after a month or
so still struggle with 3 or 4 DNB, so I wondered how similar our
strategies were. It'd be very interesting for the next DNB to
determine which strategies the participants were using, and which ones
were most correlated with DNB score and IQ gain. Until that happens,
it's hard to predict whether "chunking" is harmful or helpful to IQ
gains.

[Pontus Granström]

I reached level 6 for the first time and managed to stay there for one round, however I never use chunking until I am above 9. I think dnb has helped me, I feel "happier" and more "motivated" and I can easily handle "unease" (research shows that dopamine is affected by memory training). I also feel that I have better memory and focus. Which of course might have a positive impact on intelligence. It's important to understand that "IQ-test" do not measure one single factor, with something like "you think logically or not", it tries to quantify something they do not fully understand. They are only interested in the distrubtions of the test scores but never study the process that actually leads to it. I am fully convinced that this  kind of training will benefit many people in getting higher scores, because it mimics an IQ-test from a memory and executive process point of view. "Higher mathematics" for example often opens up to many "attack angeles" which it might be wrong to say that there is only way of thinking of you are intelligent.

[Anthony Reyes]

Before devising a strategy for anything in life, one must first analyze the purpose. Our purpose here is not necessarily to increase our  score but to increase our working memory. My thought process is: what strategy can I implement that would help me achieve this purpose?

Breaking down the input into size N chunks ensures the highest level of stress on your working memory. For example: in a 5-back trial, it would be best to memorize five units as they come to you. Then, in a different memory bucket, start memorizing the next 5. Every time a new unit arrives, compare the current unit in your new bucket to the unit in the old bucket. Once the new bucket is full, it becomes the old bucket and the new incoming units start filling up yet another new bucket.

This has worked well for me and it is certainly a good way to put stress on working memory. The traditional approach of keeping one bucket of size N and pushing old units out as new units come in may have the advantage of putting more stress on the manipulation of working memory. The double-bucket approach requires more memory but less processing, so it's hard to say which strategy will result in the best overall results. Less processing might also result in higher N-Back levels since a large part of not being able to advance with the single bucket method is that there's not enough time between incoming units to finish processing.

Thoughts?

 

[jotaro]

but wait man so in the end, you hold 5 n and 5 n both at the same time? before filling one again>?

if you can do that ,then this level 5 back is just too low for you in the first place.