spinback software

[polar]

Hello everyone,

I realized my contemplations from this post
http://groups.google.com/group/brain-training/msg/e933da6f994e3254
and created an n-back task with mental rotation and semantics.

This means classic (triple) n-back with rotating grid (mode 0), and
semantic n-back with rotating grid (mode 1). In semantic n-back, you
have to watch whether n-th-back audio word (e.g. the one presented 6
seconds ago), belongs into category being visually presented at the
moment.

It's made in processing (based on java), so I was able to export
versions for windows, linux and mac. I tested only windows - there is
one executable, one folder ("data") with sounds and 2 conf files (both
are .xls, sorry for that :). You can configure the most important
options (self explaining "conf.xls"), but so far there are provisional
result only, no fancy menus, and keys are hardcoded:

Mode 0: A-position, S-color, F-sound.
Mode 1: A-position, F-semantic hit.
In both modes there are P-pause, and +/- rotation speed.

Last thing, it would be quite useful to have more *objects*
(categories are fine with cca 10). We can cooperate on this, I
uploaded semantics.xls conf file on google docs where anybody who has
this link can edit it:
http://spreadsheets.google.com/ccc?key=0AoDHizcNs2tWdEZ3UWIxaGJHbnFZNVJmd3Y0RTdJRlE&hl=en&authkey=CIbdhNMI

What you can do, is add another *column* - I'll transform the word
into sound and implement it. "1" means that object definitely belongs
into that row's category, "0" means object definitely doesn't belong
there, and "9" means undecisive / not sure (software won't ask for
this relationship). You can even try to add new category, but it's
row's content has to be reasonably varied (not only 0 or 1 or 9).

Here are the links (no installation, just unrar, pass = sk8isgr8)

http://rapidshare.com/files/404470362/spinback.windows.rar.html
http://rapidshare.com/files/404470366/spinback.linux.rar.html
http://rapidshare.com/files/404470367/spinback.macosx.rar.html

I'm leaving today for few days, but I'll definitely get back and
appreciate all your comments. Enjoy :)

[αrgumziΩ]

This is looking really good, polar!

Unfortunately, none of the buttons work, and the missing overhead
(trials left, time in, item missed) will certainly leave many in a
daze. Furthermore, I noticed that there's very little variability in
the items (every time the program starts, it starts roughly the same
way with the same items).

That said, I have one suggestion as far as semantics go: it is
probably preferable to go with antonyms of the items. That way there
won't be as much trial and error involved in understanding the
semantic dimensions of the words. Another semantic possibility is word
matching (a kind of combined n-back setup), where the word displayed
has to match up with the word spoken. There would be three inputs for
the latter: audio match, visual match, and semantic match. These two
suggestions with their own merits merely serve to simplify the
program, because we should keep in mind that the semantic world is
monstrously complex. I can already see the list of categories becoming
unwieldy and probably a huge impediment to future development.

Anyway, this is definitely a great way forward! I was about to suggest
that BW should have a rotating grid like the one you've developed! I
look forward to seeing what you can do with this, polar.

Kind regards,
argumzio

On Jul 2, 6:56 am, polar <pol...@gmail.com> wrote:
> Hello everyone,
>

> I realized my contemplations from this posthttp://groups.google.com/group/brain-training/msg/e933da6f994e3254

> and created an n-back task with mental rotation and semantics.
>
> This means classic (triple) n-back with rotating grid (mode 0), and
> semantic n-back with rotating grid (mode 1). In semantic n-back, you
> have to watch whether n-th-back audio word (e.g. the one presented 6
> seconds ago), belongs into category being visually presented at the
> moment.
>
> It's made in processing (based on java), so I was able to export
> versions for windows, linux and mac. I tested only windows - there is
> one executable, one folder ("data") with sounds and 2 conf files (both
> are .xls, sorry for that :). You can configure the most important
> options (self explaining "conf.xls"), but so far there are provisional
> result only, no fancy menus, and keys are hardcoded:
>
> Mode 0: A-position, S-color, F-sound.
> Mode 1: A-position, F-semantic hit.
> In both modes there are P-pause, and +/- rotation speed.
>
> Last thing, it would be quite useful to have more *objects*
> (categories are fine with cca 10). We can cooperate on this, I
> uploaded semantics.xls conf file on google docs where anybody who has

> this link can edit it:http://spreadsheets.google.com/ccc?key=0AoDHizcNs2tWdEZ3UWIxaGJHbnFZN...

>
> What you can do, is add another *column* - I'll transform the word
> into sound and implement it. "1" means that object definitely belongs
> into that row's category, "0" means object definitely doesn't belong
> there, and "9" means undecisive / not sure (software won't ask for
> this relationship). You can even try to add new category, but it's
> row's content has to be reasonably varied (not only 0 or 1 or 9).
>
> Here are the links (no installation, just unrar, pass = sk8isgr8)
>

> http://rapidshare.com/files/404470362/spinback.windows.rar.htmlhttp://rapidshare.com/files/404470366/spinback.linux.rar.htmlhttp://rapidshare.com/files/404470367/spinback.macosx.rar.html

[polar]

Thanks argumzio. Btw google put the message three times to group
summary mail, sorry for that.

Now, what do you mean by buttons (there shouldn't be any buttons in
the game)? I dont use time or trials left indication, nor hit/miss,
but I'll think of that in next release. I acknowledge bad random seed
too (and result screen on 16:9 monitors is not properly readable).

The antonyms are very interesting idea, thanks for that. I see your
next point too, but if we keep low number of categories, it should be
ok. Till later

> >http://rapidshare.com/files/404470362/spinback.windows.rar.htmlhttp:/...

[Bewater]

What do you think of making some kind of pattern like in the one in
the odd one out task?
I am thinking of filling the whole dnb grid with figures. And the user
would have to memorize the position of the odd estimuli like in odd
one out.
That way you train pattern recognition at the same time you exercise
working memory. (so if it doesn't produce a new tipe of processing
workout, at least it produces more cognitive load)
It seems compatible, and i had that idea because ive been recently
noticing some positive effects due to the odd one out task practice.
While doing math in college, i feel that the way i imagine the terms
of the ecuations has some kind of similarity to the task.
It´s some kind of visual comparison between caracteristics of the
figures/terms.
What do you think?

[polar]

I see your point Bewater and agree, that OOO is a very good reasoning
task. But in my opinion, it is not really consistent with n-back - you
need different time duration for every OOO task (like 1 - 20 seconds),
while the shorter time is not reasoning-loaded enough, longer times
are quite off the WM time limit even by n-back level 3.

There are many ways how to increase cognitive load, we need to find
the most useful ones. This could be useful, but it exercices more of a
specific skill than WM. But it's interesting and I'll keep it in mind,
really.

What I thought about, was to present more visual stimuli in classic n-
back. Like 3 squares. And for example on n-back level 3, you would
always had to click on the "overlapping" square (the one which was
presented in *all* previous three rounds). This is not reasoning task,
but it would really tax your visual working memory. And you can do
this for all kinds of stimuli, but for higher N its too hard to play.

[polar]

Thanks for reminding likeprestige, but let me remind too, that it was
just (IT) conference proceedings, and they used only SPM (which is,
frankly speaking, very outdated and easy test measuring cognitive
speed at best). And I'm afraid I have to say explicitly too, that
scientific level of the paper is VERY low (and its not because of its
bad english, typos or serious lack of citations).

Lets just take one paragraph, with the subtitle "Dual N-back Working
Memory Theory" (nothing like this exists, nevermind). First sentence
is "The concept of working memory was put forward by Baddeley and
Hitch in 1974" - this is not true, they were already an alternative to
Atkinson and Shriffrin (1968), not mentioning Miller's short-term
memory in 1956. Next sentence: "N-back task is based on the Baddeley’s
experiment" - no it is not. Next sentence "The principle of N-back is
that two tasks competed to use the same limited resource." - that is
principle of mulitasking, not n-back. Next they say that n-back
includes reasoning task, which is not true.

So these guys can be real IT experts one day, but they seriously lack
knowledge of WM or intelligence today. And as david sky said, they
used dual-n-back in their study. Taken together, gabor stimuli
definitely works for people with amblyopia, and can be useful in dual-
n-back - but only marginally regarding WM or intelligence. Anyway,
thanks for the idea, and I'll appreciate any following.

On 6. Júl, 11:18 h., likeprestige <plastic...@live.com.au> wrote:
> Hey Benjamin,
>
> In respect to incorporating a reasoning task in Brain Workshop I'm
> thinking along the same lines...
>
> What happened to the idea of deciphering a way to include "Gabor
> Stimuli"?
>
> Let me take us 'back in time'... to this article...
>
> "Study on Improving Fluid Intelligence through Cognitive Training
> System Based on Gabor Stimulus"
>
> http://brain-training.googlegroups.com/web/Study+on+Improving+Fluid+I...
>
> "Analyses of the training data revealed that the participants of
> training group improved in their performance on the Dual Nback
> task in ten sessions. We can see in the Figure 6, all the
> participants in the control and trained group got more IQ
> scores than pre-test after ten days, because we adopted the
> same standardized fluid intelligence test. But participants in
> training group increase more IQ scores than control group.
> This data indicated that the effect on Gf scores went beyond
> an increase in working memory capacity. With the enhance
> performance of trainers, the reasoning ability has been
> strengthened. Moreover, the performance has strong
> relationship with the degree of concentration during the
> cognitive training."
>
> "During the cognitive training, Gabor visual target with two
> different kinds of spatial frequency at four different directions
> asked the user to take the Gabor indentify task, which can
> effect spatial frequency in visual system and improve vision.
> We can get a conclusion that the system not only help to
> improve fluid intelligence but also help to improve vision."
>
> There is information contained in the article that highlights the
> "how" and "what" in respect to the Gabor stimulus functions. Is it
> possible for us to replicate these functions in Brain Workshop or even
> enhance the difficulty level of these functions?
>
> I would like to donate money towards the effort of incorporating this
> function into Brain Workshop as well as the rotating grid for I
> recognize that these additions may take some time. I hope fellow
> members are also able to appreciate the difficulty level of this task
> by providing a degree of support. I have some other ideas floating
> around that I'm hoping can be done at the same time but for now I was
> just hoping we could estimate a projected cost.
>
> Help Jonathan Toomim???
>
> Kind regards,
>
> likeprestige
>
> On Jul 6, 10:29 am, Bewater <bewate...@gmail.com> wrote:> What do you think of making some kind of pattern like in the one in

[Gwern Branwen]

On Wed, Jul 7, 2010 at 4:35 AM, polar <pol...@gmail.com> wrote:
> Thanks for reminding likeprestige, but let me remind too, that it was
> just (IT) conference proceedings, and they used only SPM (which is,
> frankly speaking, very outdated and easy test measuring cognitive
> speed at best). And I'm afraid I have to say explicitly too, that
> scientific level of the paper is VERY low (and its not because of its
> bad english, typos or serious lack of citations).
>
> Lets just take one paragraph, with the subtitle "Dual N-back Working
> Memory Theory" (nothing like this exists, nevermind). First sentence
> is "The concept of working memory was put forward by Baddeley and
> Hitch in 1974" - this is not true, they were already an alternative to
> Atkinson and Shriffrin (1968), not mentioning Miller's short-term
> memory in 1956. Next sentence: "N-back task is based on the Baddeley’s
> experiment" - no it is not. Next sentence "The principle of N-back is
> that two tasks competed to use the same limited resource." - that is
> principle of mulitasking, not n-back. Next they say that n-back
> includes reasoning task, which is not true.
>
> So these guys can be real IT experts one day, but they seriously lack
> knowledge of WM or intelligence today. And as david sky said, they
> used dual-n-back in their study. Taken together, gabor stimuli
> definitely works for people with amblyopia, and can be useful in dual-
> n-back - but only marginally regarding WM or intelligence. Anyway,
> thanks for the idea, and I'll appreciate any following.

Well, I've already criticized this paper, but you seem to've
understood it better than I.

What *is* a Gabor stimulus and how was their DNB variant using it?

--
gwern

[polar]

As I got it, gabor is a visual effect, which can add special kind of
"blur" to any image. This "blurriness" makes your brain think it
should try harder to recognize the object. This way gabor increases
visual perception load, which is useful when you have problems with it
(amblyopia) and need to train. It can improve even healthy vision a
bit, but this improvements seems to be pretty much restricted to
occipital cortex and visual perception. These chinese researchers
simply used in dnb some "blurred geometrical shapes" instead of
classic squares/triangles etc.

[Pontus Granström]

Thanks for reminding likeprestige, but let me remind too, that it was
just (IT) conference proceedings, and they used only SPM (which is,
frankly speaking, very outdated and easy test measuring cognitive
speed at best). And I'm afraid I have to say explicitly too, that
scientific level of the paper is VERY low (and its not because of its
bad english, typos or serious lack of citations).

I've taken a look at SPM type of questions, and I would not say the questions are very much like WMC tasks, remember some lines, remove some others etc. Al though this is very common for all IQ-tests and thinking in general. As for the English, since they are Chinese they've got more difficulty with English than other mainly European countries, very much like you and I had to write a WM-training report in Arabic. I also wonder why that would affect the outcome of the training as well!

I do not know anything about the status of SPM, but from what I know it should be highly G-loaded and most definitely at least as good as the very praised OOO.

As for speed claims etc, in the 10-min RAPM case speed is not a significant factor in the performance, dual-n-back is not a speed training regime either for that matter.

Pontus

[Pontus Granström]

sorry remove not

I've taken a look at SPM type of questions, and I would  say the questions are very much like WMC tasks

[Pheonoxia]

What is SPM?

[Pontus Granström]

Standard progressive matrices an easier version of the RAPM or RAPM is rather a harder version of the SPM.

On Thu, Jul 8, 2010 at 6:02 PM, Pheonoxia <b...@brockman.info> wrote:

What is SPM?

--
You received this message because you are subscribed to the Google Groups "Dual N-Back, Brain Training & Intelligence" group.
To post to this group, send email to brain-t...@googlegroups.com.
To unsubscribe from this group, send email to brain-trainin...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/brain-training?hl=en.

[Shamanu999]

Very interesting approach polar.
Would you mind if I use the semantics list as a default in my own
implementation?

Greetings,
Shamanu

Am 02.07.2010 13:56, schrieb polar:
> Hello everyone,
>
> I realized my contemplations from this post
> http://groups.google.com/group/brain-training/msg/e933da6f994e3254
> and created an n-back task with mental rotation and semantics.
>
> This means classic (triple) n-back with rotating grid (mode 0), and
> semantic n-back with rotating grid (mode 1). In semantic n-back, you
> have to watch whether n-th-back audio word (e.g. the one presented 6
> seconds ago), belongs into category being visually presented at the
> moment.
>
> It's made in processing (based on java), so I was able to export
> versions for windows, linux and mac. I tested only windows - there is
> one executable, one folder ("data") with sounds and 2 conf files (both
> are .xls, sorry for that :). You can configure the most important
> options (self explaining "conf.xls"), but so far there are provisional
> result only, no fancy menus, and keys are hardcoded:
>
> Mode 0: A-position, S-color, F-sound.
> Mode 1: A-position, F-semantic hit.
> In both modes there are P-pause, and +/- rotation speed.
>
> Last thing, it would be quite useful to have more *objects*
> (categories are fine with cca 10). We can cooperate on this, I
> uploaded semantics.xls conf file on google docs where anybody who has
> this link can edit it:

> http://spreadsheets.google.com/ccc?key=0AoDHizcNs2tWdEZ3UWIxaGJHbnFZNVJmd3Y0RTdJRlE&hl=en&authkey=CIbdhNMI

>
> What you can do, is add another *column* - I'll transform the word
> into sound and implement it. "1" means that object definitely belongs
> into that row's category, "0" means object definitely doesn't belong
> there, and "9" means undecisive / not sure (software won't ask for
> this relationship). You can even try to add new category, but it's
> row's content has to be reasonably varied (not only 0 or 1 or 9).
>
> Here are the links (no installation, just unrar, pass = sk8isgr8)
>

> http://rapidshare.com/files/404470362/spinback.windows.rar.html
> http://rapidshare.com/files/404470366/spinback.linux.rar.html
> http://rapidshare.com/files/404470367/spinback.macosx.rar.html
>

[polar]

Shamanu if you want to use it, I surely agree (actually the semantics
mode is very unstable in this version). I really believe it is worth
to use semantics, but it's hard to implement it in a reasonable way
(I'm not really satisfied with the algorithm I used).

> >http://spreadsheets.google.com/ccc?key=0AoDHizcNs2tWdEZ3UWIxaGJHbnFZN...

[Shamanu999]

Ok, thanks.
Yes, i will also have to play/experiment with the principle a little bit
to see in which way
to implement it. The concept is a bit different to basic n-back as there
is a mix between
the visual and audio part to create a new one.

Greetings,
Shamanu

Am 13.07.2010 11:52, schrieb polar:
> Shamanu if you want to use it, I surely agree (actually the semantics
> mode is very unstable in this version). I really believe it is worth
> to use semantics, but it's hard to implement it in a reasonable way
> (I'm not really satisfied with the algorithm I used).
>