Redick et al 2012, "No evidence of intelligence improvement after working memory training: A randomized, placebo-controlled study"

[Gwern Branwen]

> Numerous recent studies seem to provide evidence for the general intellectual benefits of working memory training. In reviews of the training literature, Shipstead, Redick, and Engle (2010, in press) argued that the field should treat recent results with a critical eye. Many published working memory training studies suffer from design limitations (no-contact control groups, single measures of cognitive constructs), mixed results (transfer of training gains to some tasks but not others, inconsistent transfer to the same tasks across studies), and lack of theoretical grounding (identifying the mechanisms responsible for observed transfer).
>
> The current study compared young adults who received 20 sessions of practice on an adaptive dual n-back program (working memory training group) or an adaptive visual search program (active placebo-control group) with a no-contact control group that received no practice. In addition, all subjects completed pre-test, mid-test, and post-test sessions, comprising multiple measures of fluid intelligence, multitasking, working memory capacity, crystallized intelligence, and perceptual speed. Despite improvements on both the dual n-back and visual search tasks with practice, and despite a high level of statistical power, there was no positive transfer to any of the cognitive ability tests. We discuss these results in the context of previous working memory training research, and address issues for future working memory training studies.

Total n=75; RAPM was speeded. The multiple measures is something like
17 tests, which took the subjects 2 and a half hours to do just the
pre-test - I don't envy their subjects. My copy is very long and I had
to relatively skim it, but I have no initial objections (even the
speeding of the RAPM is a good thing inasmuch as it turned in a null
anyway). I already mentioned the very interesting survey results
suggesting placebo effects.

In press, estimated to be available online in 5-6 weeks. Don't bother
contacting me for a copy, I promised Redick I would not share it.

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

[☉]

Looks like things aren't coming to a close any time soon (not that I'm
surprised). Were it not for the publish or perish criterion that
shapes much of current scientific output, I think we would be saved
from the numerous false positives that unceasingly find their way
across the choppy waters of benighted uncertainty.

argumzio

[jttoto2]

I know you can't give a copy of the study but could you perhaps answer
a question? Were there no differences between the experimental group
and the active control group, or no differences between all three
groups?

[Gwern Branwen]

On Tue, May 22, 2012 at 8:09 PM, jttoto2 <john....@gmail.com> wrote:
>  Were there no differences between the experimental group
> and the active control group, or no differences between all three
> groups?

post - control: n=20 raw score=6 (deviation 3); control search: 29
6.24 (3.34); n-back: 24 6.25 (3.08)

I learned how to pool variances today, and when you do that to
collapse the two control groups, you get

n=49 6.14 (3.202)

As you can see, the differences are basically non-existent. The n-back
group has 0.11 more questions right, but the additional deviation in
the overall control group is 0.12...

--
gwern
http://www.gwern.net

[jttoto2]

Interesting. This supports previous criticism: studies with a
comprehensive battery of tests have less impressive results, and in
this case no results. Others have pointed out in the past that the
training methods were too similar to the measures of far-transfer.
(Chein has transfer to reading but not RAPM; Dnb looks like a RAPM
grid, etc.)

A caveat: is that data pulled from a composite score of all the tests
used, or the RAPM only? If it is the former, I'd say this solidifies
previous criticisms. If it is the latter, then it just raises more
questions, such as why do some studies using n-back show improvements
in fluid reasoning, but this one didn't? (as the researchers pointed
out, no improvements on any test? )

On May 22, 9:39 pm, Gwern Branwen <gwe...@gmail.com> wrote:
> On Tue, May 22, 2012 at 8:09 PM, jttoto2 <john.dem...@gmail.com> wrote:
> >  Were there no differences between the experimental groupr

[Gwern Branwen]

On Tue, May 22, 2012 at 10:19 PM, jttoto2 <john....@gmail.com> wrote:
> A caveat:  is that data pulled from a composite score of all the tests
> used, or the RAPM only?  If it is the former, I'd say this solidifies
> previous criticisms.  If it is the latter, then it just raises more
> questions, such as why do some studies using n-back show improvements
> in fluid reasoning, but this one didn't? (as the researchers pointed
> out, no improvements on any test? )

RAPM, since it was for my meta-analysis.

--
gwern
http://www.gwern.net

[Rotem Segev]

As time goes by, we see more and more evidence that we are wasting our
time with DNB?

bummer...

On 23 מאי, 06:39, Gwern Branwen <gwe...@gmail.com> wrote:

[Zaraki]

It says 20 sessions, but can you say how long these sessions were, and
how long a period of time these sessions were spread over?

[Pontus Granström]

Exactly, it almost becomes ridiculous that they neglect the most important and significant factor.

--
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.

[Michael]

Going to let my opinion fly amongst the clouds until I actually get to read the thing. Then & only then will I call for the rain to thunder down its message from above!

On Wednesday, May 23, 2012 8:40:52 PM UTC+10, King Of The Stars wrote:

Exactly, it almost becomes ridiculous that they neglect the most important and significant factor.

On Wed, May 23, 2012 at 11:05 AM, Zaraki <zara...@gmail.com> wrote:

It says 20 sessions, but can you say how long these sessions were, and
how long a period of time these sessions were spread over?

On 23 Maj, 09:30, Rotem Segev <hypnotiz...@gmail.com> wrote:
> As time goes by, we see more and more evidence that we are wasting our
> time with DNB?
>
> bummer...
>
> On 23 מאי, 06:39, Gwern Branwen <gwe...@gmail.com> wrote:
>
>
>
>
>
>
>
> > On Tue, May 22, 2012 at 10:19 PM, jttoto2 <john.dem...@gmail.com> wrote:
> > > A caveat:  is that data pulled from a composite score of all the tests
> > > used, or the RAPM only?  If it is the former, I'd say this solidifies
> > > previous criticisms.  If it is the latter, then it just raises more
> > > questions, such as why do some studies using n-back show improvements
> > > in fluid reasoning, but this one didn't? (as the researchers pointed
> > > out, no improvements on any test? )
>
> > RAPM, since it was for my meta-analysis.
>
> > --
> > gwernhttp://www.gwern.net

--
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-training@googlegroups.com.
To unsubscribe from this group, send email to brain-training+unsubscribe@googlegroups.com.

[Gwern Branwen]

On Wed, May 23, 2012 at 5:05 AM, Zaraki <zara...@gmail.com> wrote:
> It says 20 sessions, but can you say how long these sessions were, and
> how long a period of time these sessions were spread over?

The idea was to replicate Jaeggi 2008, so it will come as no surprise
that there were 20 sessions, and in each session, the subject played
20 rounds, each round being the usual 20 stimuli; the 20 sessions were
over something like 5 weeks:

> Subjects in the two training groups completed an additional 20 practice sessions, each of which took between 30-40 min. Subjects could not complete more than one experimental session per day, with a maximum of seven sessions per week. Dual n-back and visual search subjects completed all 20 practice sessions (and the mid-test session) in an average of 46 (SD = 13.7) and 41 (SD = 10.7) days, respectively; the time to complete the training did not differ for the two groups, t(51) = 1.59, p = .12. Dual n-back and visual search group subjects performed mid-test sessions after 10 practice sessions.4 Control subjects performed mid- and post-test sessions at approximately the same interval of days as the training groups, and the intervals did not differ for the three groups: Pre- to Mid-test, F(2, 70) = 1.23, p = .30; Pre- to Post-test, F(2, 70) = 1.58, p = .21. Compensation for each practice session was $10, with a 10% bonus at the end of the study for completing all practice sessions ($20).

41 / 7 = 5.85

On Wed, May 23, 2012 at 6:40 AM, Pontus Granström <lepo...@gmail.com> wrote:
> Exactly, it almost becomes ridiculous that they neglect the most important
> and significant factor.

If Jaeggi 2008 could see improvements, why can't Redick et al?

--
gwern
http://www.gwern.net

[jttoto2]

>If Jaeggi 2008 could see improvements, why can't Redick et al?

Yes, I find that extremely bizarre. Did the difference between the
experimental group and the no-contact (disregarding the active
control)group reach significance?


On May 23, 10:52 am, Gwern Branwen <gwe...@gmail.com> wrote:

> On Wed, May 23, 2012 at 5:05 AM, Zaraki <zaraki...@gmail.com> wrote:
> > It says 20 sessions, but can you say how long these sessions were, and
> > how long a period of time these sessions were spread over?
>
> The idea was to replicate Jaeggi 2008, so it will come as no surprise
> that there were 20 sessions, and in each session, the subject played
> 20 rounds, each round being the usual 20 stimuli; the 20 sessions were
> over something like 5 weeks:
>
> > Subjects in the two training groups completed an additional 20 practice sessions, each of which took between 30-40 min. Subjects could not complete more than one experimental session per day, with a maximum of seven sessions per week. Dual n-back and visual search subjects completed all 20 practice sessions (and the mid-test session) in an average of 46 (SD = 13.7) and 41 (SD = 10.7) days, respectively; the time to complete the training did not differ for the two groups, t(51) = 1.59, p = .12. Dual n-back and visual search group subjects performed mid-test sessions after 10 practice sessions.4 Control subjects performed mid- and post-test sessions at approximately the same interval of days as the training groups, and the intervals did not differ for the three groups: Pre- to Mid-test, F(2, 70) = 1.23, p = .30; Pre- to Post-test, F(2, 70) = 1.58, p = .21. Compensation for each practice session was $10, with a 10% bonus at the end of the study for completing all practice sessions ($20).
>
> 41 / 7 = 5.85
>

[Gwern Branwen]

On Wed, May 23, 2012 at 11:06 AM, jttoto2 <john....@gmail.com> wrote:
> Yes, I find that extremely bizarre.  Did the difference between the
> experimental group and the no-contact (disregarding the active
> control)group reach significance?

Eyeballing the scores I gave before, 6 (3) and 6.25 (3.08), I would be
surprised. Some more relevant material

> To facilitate comparisons with previous research, we re-analyzed all transfer data excluding the visual search group. The results matched the full analyses – no transfer for the dual n-back group relative to the no-contact control group. In addition, we re-analyzed all transfer data excluding the mid-test session, comparing only the pre- and post-test sessions, and, again, we found no transfer for the dual n-back group relative to the no-contact or active control groups. Finally, we conducted ANCOVAs for each transfer task, using the pre-test score as the covariate. The results were qualitatively the same as the ANOVA results presented in Table 3.6

And for Jonathan, the power analysis (which I appreciate more after
messing around with some power calculations for my own Zeo sleep
experiments):

> A post-hoc power analysis (G*Power; Faul, Erdfelder, Lang, & Buchner, 2007) indicated that with our sample size, we had sufficient power to detect a significant Group (between-subjects) x Session (within-subjects) interaction, if it was present in our transfer data. Our power to detect a large (f = .40) or medium (f = .25) effect was > .99, based on our sample size and the use of the within-subjects correlation of r = .53, which was the average correlation among the repeated measures across all 17 transfer tasks (the default value in G*Power is r = .50). We also re-ran the power analyses using a correlation among repeated measures of r = .30, which was the lowest observed correlation among pre-, mid-, and post-test performance, ignoring test version (Paper Folding). Using r = .30, the power to detect a large or medium effect was > .95. Note that we based our power analysis on a medium or large effect of dual n-back training, given the previous literature (JBJP, 2008: d = .68; JSBSJP, 2010: d = .98 for RAPM, d = .49 for BOMAT).

My current meta-analysis (including Redick) under the random effects
model says that the d = 0.47, BTW.

(In part because of the Stephenson 2010 outlier you can see clearly in
both the funnel and forest plots I posted earlier; I corrected my
pooling of groups, but that didn't make the outlier go away, so
yesterday I emailed Stephenson again to ask about it. Still no reply.)

--
gwern
http://www.gwern.net

[Aaron Hamlin]

>If Jaeggi 2008 could see improvements, why can't Redick et al? 
Yes, I find that extremely bizarre.  Did the difference between the 
experimental group and the no-contact (disregarding the active 
control)group reach significance?  

I think the possibility for this kind of outcome (no significant difference versus significant difference with Jaeggi) is due to randomness. If this is the true state of affairs, then Jaeggi made a Type I error (it happens). This is when you think there's an effect, but there's not. Run an experiment enough times, and some of the time just based on chance it will say there's an effect even when there isn't.

To get around this, you really need to do a bunch of well-designed experiments and run a meta analysis, which is what gwern seems to be driving at.

Also, significance is nothing special. If you have lots of statistical power (ex// from large sample size), you can get a significant outcome even though the actual effect going on is tiny. I think a more apt way to look at significance (p-value, assumedly after type-1 error correction like Bonferoni or something) is to see it as a gateway for whether you should check out the effect size measure. If it's significant, then you look at the effect size. And the effect size is really where the meaningful value is.

There are a bunch of effect size measures. I think the easiest (and therefore preferable) measure is r^2. This is the amount of the outcome measure (here IQ or whatever the dependent measure is) that is explained by the independent variable (whether you did n-back or did some control task).

So I'd go with being less concerned over significance alone. At this stage, I think you just want a bunch of well-designed studies with a meta-analysis reviewing them. Preferably, that meta analysis uses r^2 (or r, obviously) as its effect size so that it's easier to understand.

To make sense of gwern's meta analysis effect size (d= .47), perhaps this helps makes more sense of it: http://www.umdnj.edu/idsweb/shared/effect_size.htm

Using this as a reference you can imagine normal score distributions (from IQ difference or whatever was used) of the control and n-back groups. The n-back group would have about a third of its distribution not overlapping the control's (and for the better).So the n-back distribution would be shifted moderately more to the right than the control.

And, quite honestly, I think kind this of effect is a big deal.

[Gwern Branwen]

A fascinating item which I only just noticed while working on adding
IQ test type & test time to the meta-analysis:

> Data were collapsed across different transfer tests administered under different time limits. To assess transfer to intelligence, the 8-session groups completed the Raven Advanced Progressive Matrices (RAPM), and the 12-, 17- and 19-session groups performed the Bochumer Matrizen-Test (BOMAT). Although both tests assess matrix reasoning (presenting 3x3 vs. 3x5 matrices, respectively), they are not comparable in length (18 vs. 29 items, respectively). In addition, the 19-session groups were given 20 min to complete BOMAT, whereas the 12- and 17-session groups received only 10 min (S. M. Jaeggi, personal communication, May 25, 2011). As shown in Figure 2, the use of the short time limit in the 12- and 17-session studies produced substantially lower scores than the 19-session study. We argue that it is inappropriate to simply average across the number of problems solved from the four tests to create Figure 1a.

(Figure 2 is http://i.imgur.com/3g1KZ.png )

Can't believe I missed that.

--
gwern
http://www.gwern.net

[The.Fourth.Deviation.]

tried to reply but accidentally sent it to gwern only.

I was saying that this study seems to be missing the point, it's not necessarily Gf that is improving. In the originally Jaeggi videos, the trainees mentioned increased memory for things (i.e. not losing keys). This is exactly what I notice. Improved short term memory is key for many things. It does not mean you will score higher on matrix tests, but it does mean that you can learn faster, as you don't need to review as much etc.

I also notice increased focus after NBack, better problem solving ability due to better clarity of thought, improved verbal fluency and incidentally improved ability to create humor (as I don't lose my train of thought as easily after training Nback. (btw, i use modes much more advanced than plain nback, such as triple stim where I am following 3 stimuli at once, so my training is also much more advanced than these studies test).

next, everyone here seems to have forgotten about the tons of MRI/CT scan studies which have shown physical changes in the efficiency of key brain regions after WM training. It's very hard to claim that plasticity is a false positive. It is much more likely that this Redick study is coming up with a false negative, or more importantly, that it is testing the wrong things. For example, I don't expect my improved short term memory from Nbacking to make me better at finding patterns in matrices. However, it does help me at work by making me more organized and less forgetful, and therefore much more effective. This, along with the other benefits, really leads to improved quality of life. I should also point out that I have ADHD so my WM is a bit worse than the average, and the small boosts I get from Nbacking are marginally more impactful than they might be for someone who already has a strong WM. Then again, I am using much more advanced variants of DnB than most here, so it's difficult to say why I seem to benefit a lot more.

[Brain Train]

Adhd- self diagnosed or medically diagnosed?

> --
> 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.
> Visit this group at http://groups.google.com/group/brain-training?hl=en.
>  
>  
>

[The.Fourth.Deviation.]

Medically.

Aspergers:rather extensive self diagnosis

[hippocampus]

@fourth.deviation: nobody is saying that WMT is ineffective. It just probably doesn't raise intelligence - although it may improve other things. And as you have noticed, we can't conclude anything about different version of WMT - maybe longer sessions or something else is needed for far transfer to anything.

[The.Fourth.Deviation.]

It would appear then that the definition of intelligence being used is too rigid and therefore causing some researchers to commit a pretty bad type 2 error.

[whoisbambam]

agreed.

it may improve other things.

[Gwern Branwen]

On Wed, Dec 12, 2012 at 9:21 PM, The.Fourth.Deviation.
<david...@gmail.com> wrote:
> The other major point that no one here seems to remember is the massive
> array of MRI/CT scan studies that showed increased efficiency in key brain
> regions as a result of WM training. It's difficult to claim that physical
> adaptations in the brain are statistically insignificant.

You know that MRI and CT scans are pretty much the epitome of bad
statistics with regard to multiple correction, data mining, and
ambiguous interpretation, right? Look up the dead salmon study
sometime for a laugh.

--
gwern
http://www.gwern.net

[The.Fourth.Deviation.]

I'm not sure that argument is a logical justification to throw out the results of each and every MRI/CT study done on the subject. At the end of the day, based on my own observations, readings, and my own results from WM training, I think it's much more likely these guys committed a type II error..And if we are open to the fact that other studies may have committed errors, theres no reason to suggest that this one is any less likely to be flawed in some important way. At any rate, opinions from either side are obviously important, but I think this "debate" (if it can be called that, since I think, but am not sure, that the pro WM training camp has a preponderance of support) is starting to sputter in favor of pro WM training.

A collection of _all_ studies done on this, and their results, would help us determine what the overall literature seems to be saying, rather than illogically coalescing around the findings of single studies.

[Gwern Branwen]

On Thu, Dec 13, 2012 at 11:22 PM, The.Fourth.Deviation.
<david...@gmail.com> wrote:
> I'm not sure that argument is a logical justification to throw out the
> results of each and every MRI/CT study done on the subject. At the end of
> the day, based on my own observations, readings, and my own results from WM
> training, I think it's much more likely these guys committed a type II
> error..And if we are open to the fact that other studies may have committed
> errors, theres no reason to suggest that this one is any less likely to be
> flawed in some important way. At any rate, opinions from either side are
> obviously important, but I think this "debate" (if it can be called that,
> since I think, but am not sure, that the pro WM training camp has a
> preponderance of support) is starting to sputter in favor of pro WM
> training.

Alright, since you are persevering in this... let me point out, that
besides being statistically questionable in a way far worse than any
of the regular studies we usually discuss, the interpretation of those
results showing 'physical adaptation' is even more doubtful: *of
course* there are physical changes after learning new material and
training to some task. Given that the mind is what the brain does, how
could it be otherwise?! The existence of changes no more demonstrates
transfer to IQ or to meaningful endpoints or dubious anecdotes about
personal improvement, than the existence of words on a piece of paper
demonstrate that the author is a great novelist. Perhaps the words are
part of the Great American Novel, but one would have to test them
somehow... tests which you are rejecting because they just aren't
giving the answers you want.

--------------------------------------------

This is completely besides the point of my recent email, though: a
shocking detail about Jaeggi 2008 which has gone completely unnoticed.
They *increased the testing time for the final 19-day group*! What the
fuck? The group that they were counting on to show a dose-response
effect, remember! I have read Jaeggi 2008 carefully many times, and I
never noticed this because they don't mention it.

How many criticisms, how many dubious citations, how many embarrassing
details leaking out half a decade after the fact, how many failures
to replicate, do there have to be before we start to take seriously
the hypothesis that Jaeggi et al 2008 may be a pile of tenured grade-A
peer-reviewed highly-cited steaming-bullshit?

--
gwern
http://www.gwern.net

[☉]

To be quite honest, I thought that the Jaeggi study had long since
been thoroughly and adequately characterized as such. Any more nails
in the coffin just comes off as a bit in bad taste.

I think there's room for other kinds of research on this front, that
is, in terms of WM, but I seriously doubt any form of practice at
training will involve specific improvements in underlying cognition.
Heck, we've known this for a long time already, so it shouldn't prove
surprising that the study proved... less than adequate under the
appropriate amount of scrutiny.

argumzio

[The.Fourth.Deviation.]

Well in those MRI/CT studies it's shown that there is improved efficiency in the very exact regions of the brain that are more efficient in people with more WM capacity.  Efficiency of brain regions is often linked to performance. So i dont think the interpretation of the results is as dubious as you suggest...in fact theyre rather straightforward. You do bicep curls, your biceps get bigger. 

It's a very simple concept. If you concentrate on remembering random stimuli for 30 min, you will start to randomly remember more things after you finish for some particular interval of time, as your brain naturally adapts to demands placed on it. Is this tantamount to an increase in IQ? not necessarily. Will it improve RAPM? not necessarily. Will it improve quality of life? in my case it has, since I am less prone to making errors at work, as I can better remember what has been done and what hasn't. In addition, when I train DNB, my verbal fluency improves markedly (i.e. I am better able to express thoughts, which may have to do with improved WM but I am not really sure). 

At this point, I don't care about my RAPM score, as long as other aspects of my life have improved. Which brings us to the point of whether these researchers are even testing the right domains for improvement. Last I checked, none of these studies checked for improvements in verbal fluency, for example, but verbal fluency (not written, but spoken) and short term memory are the two things I always notice improving when I start DNB. And I also notice that they diminish when I cease practicing. The effect is great enough for it to be as if I am a different person, or at least a smarter version of myself.

Anyways, to conclude, logically, if there is a very large majority claiming benefit, (toss our Jaeggi, keep the others) and a small group claiming no benefit, one would be inclined to steer towards the majority opinion absent any other information. Studies claiming cigarettes don't cause lung cancer come to mind at this point.

[☉]

Unfortunately, there is no logical conclusion to be reached based on
your argument, which is merely a hasty generalization from your own
experience, and even more amusingly a case of ignoratio elenchi. The
whole point of such discussions as found in this list concerns the
claim made by Jaeggi that there is a discernible, positive effect on
fluid intelligence (as operationally determined by a few proxies
thereof). If so much has come to bear invalidating such claims as
counterfeit, what more hope do you have to suppose your grounds for
argument serving any sort of defense thereto? It is even more bizarre
that anyone would think that they could prove an empirical question
through such thoroughly misapplied "logical" argumentation, anyway.

And have you not in the very least considered that whatever effects
you perceive to occur after training might not at all be related to
the training itself? If you so much as remotely admit to yourself that
training your focus is sufficient for improving your focus, then
perhaps you should consider that improving your focus on the actual
tasks which you find so important in your life to be the primary aim
towards accomplishing such, rather than supposing the razzle dazzle
buzz of the task having any effect unlike a placebo. (However, it is
an open ethical question whether we should go about denying the
usefulness of placebos when they actually have an effect for those who
would otherwise not be responsive to other kinds of treatment. Okay, I
mention this only semi-seriously.)

In any case, perhaps you should reflect on the importance of reading
comprehension when addressing others' comments here, for no one denies
the possibility that training might have effects in other respects
besides those claimed by Jaeggi et al. which continually show their
faults and lack of support under repeated and prolonged criticism.
Hitherto, I think it is safe enough to say that the Jaeggi and related
studies, insofar as they claim any positive effect on non-verbal
reasoning tests, constitute an error of the first kind, i.e., an
erroneous rejection of the null hypothesis.

By the way, I've addressed such comments concerning fMRI and the like
before. It is bleeding obvious to anyone even somewhat familiar with
the question of neuroplasticity that the brain will adapt in the short
term to meet the demands of any given task; however, this is so far
from broaching the issue of such adaptation's effect on underlying
cognitive processes, the neurocorrelates of the claimed factors
involved in non-verbal reasoning tasks as seen in RPM or BOMAT (that
is, those factors besides practice, maturation, history, chance,
motivation, measurement error, etc.), that one wonders why such is
even being considered as relevant to the issue at hand at all. More
importantly, I have even questioned the supposed infallibility of the
claim that the original study and its forebears represents anything
tantamount to a demonstration of "far-transfer", even supposing such
an effect weren't illusory. There are gaping holes just about
everywhere. There are and remain unanswered questions, pretty much at
every infinitesimal point concerning the matter one could imagine.

If someone is satisfied with mere supposition, then by all means, let
such a person be satisfied. However, for those who are not, for those
who do not wish to waste their time with idle nonsense, perhaps a
moment of shared solicitude would be helpful so as not to undermine
continued discourse.

And to anyone with a modicum of logical fortitude, I need not remind
that appeals to (false) majorities are beneath consideration w.r.t.
matters concerning truth and logic. I think it is clear enough why any
one who "reasons" in such a fashion merely betrays the validity of
those measures designed to disclose such an individual's own reasoning
ability as broadly conceived on a generalized scale.

For my part, I think I will be better off responding only to those
comments worthy of real consideration. Of course, there are those
fanciful, pointless moments in one's life as well, where a little ball
of yarn makes the hours pass. If only it were a golden thread to tug
at the gods, or a golden thread out of the labyrinth. Ah, but the
Fates spun the thread such that even the gods couldn't overcome it.

argumzio, possibly related to the Mindotaur


On Dec 14, 12:12 am, "The.Fourth.Deviation." <davidsky...@gmail.com>
wrote:

[whoisbambam]

i suppose it may not improve other things.

i did 'want' to believe in 'something'

in any case, i did not perceive an increase in iq or intelligence...i
have already admitted to this.

[polar]

That's one possibility, and still another is, that things which do improve after some kinds of cognitive training, are more useful/important in real life success then intelligence is.

Dne čtvrtek, 13. prosince 2012 20:04:46 UTC+1 The.Fourth.Deviation. napsal(a):

[The.Fourth.Deviation.]

Argumzio thanks for your well written response - i shall ignore your little digs and insults and respond to points that were conceived of a higher intellectual plane.

If no one here denies that DnB may improve other domains besides those tested in Jaeggi, then it seems premature to suggest that DnB is debunked.

If this is the case, we have to consider the possibility that g itself may therefore be improving even as Gf, or isolated expressions of it, seems to be static.

Two, I think DnB is an outdated modality. I have not used DnB proper in over a year. More advanced modes with more complex stimuli and patterns should receive increased attention, especially given the uncertainty surrounding plain vanilla DnB.

[Green]

 I would like to reiterate a point I made in another thread, that there are prima facie reasons for thinking that cognitive training can facilitate far-transfer.

 

There's a whole body of literature on meditation, and a smaller (and sort of mysterious, IMHO) body of research on Space Fortress. It's reasonable to believe that there's a lot of untapped potential for properly designed cognitive training programs to improve work and social skills. 

 

Also, I agree with Gwern that Fourth.Deviation's ancedotal experiences are not good reason to think DnB facilitates far transfer, but I also think Fourth. Deviation has good reason to believe DnB has helped him. Ancedotes are good evidence, if they're first person. You are in a better position to know what works best for you than anyone else. But epistemic transfer requires different methods.

[☉]

On Saturday, December 15, 2012 9:35:53 AM UTC-6, The.Fourth.Deviation. wrote:

Argumzio thanks for your well written response - i shall ignore your little digs and insults and respond to points that were conceived of a higher intellectual plane.

Actually, considering I wrote nothing that was patently untrue, I'm not sure in what respect it is insulting. Seeing that I'm not a fighter pilot (a shocking admission, I know), which is a highly skilled vocation, I would not take it as an insult for someone to point out my incompetence in that respect. Of course, if you can point to a "higher intellectual plane" than that which is not untrue, then by all means, I would be interested in learning of such a plane, that is, assuming you meant to imply there is one.

 

If no one here denies that DnB may improve other domains besides those tested in Jaeggi, then it seems premature to suggest that DnB is debunked.

Just because it is possible that no one denies that n-back training of one form or another may have other effects beyond fluid intelligence, it does not follow that the original importance granted to such training is to be regarded as though it remains unaffected.  Your claim is a red herring here, for the debunking as such is clearly and unambiguously directed towards those spurious claims w.r.t. DNB training and its alleged influence on fluid intelligence. By contrast, consider the example of the effect of creatine monohydrate supplementation has been shown to have in precisely those respects DNB has otherwise been claimed to effect; there simply is little comparison. In other words, one would have a better chance by supplementing with a commonly found nutrient than wasting hours of tedious effort to get the result one had originally been told would come to fruition.

 

If this is the case, we have to consider the possibility that g itself may therefore be improving even as Gf, or isolated expressions of it, seems to be static.

Even granting that DNB isn't "debunked", your further suggestions are so vague and unclear that one has difficulty in determining what exactly you mean by "g itself" "may be improving" when it is clear to anyone with a well-grounded understanding in psychometrics that psychometric g is not an individual property, but a group property. It is clear, at any rate, that if there were a determinable effect on I.Q., then it is in principle spurious to suppose that one would see any transfer to psychometric g (e.g., as a Jensen effect) from this alone, because it is transparent to anyone with knowledge on the subject that training is a form of practice, and that any and all practice effects are reducible to Flynn effects, which are non-g-related increases in raw scores, with no direct impact on I.Q. proper and so to say on the g factor itself. (It is interesting to recall that if g were so adaptable in accordance with training, then it likely wouldn't be observed as the total intercorrelations among tests in a large sample of subjects, as an expression of the evolutionary result of those factors which contributed most strongly to survival, and hence being largely untrainable.) This is another conceptual issue regarding why it is so misleading as Jaeggi et al. originally (and perhaps continue to) claim that gF has actually been shown to improve, when it is so obviously the case that that wasn't demonstrated in the least by their original studies. That said, there are still open questions regarding the interpretation of the data being exclusively conceived as a g factor (a summary measure, or single quantitative latent factor), where the positive manifold of intelligence in accord with a mutualist model could serve in its stead, but this is still relatively new territory, not directly relevant to the issues at hand.

To be clear, though, I'm not sure what all of this fuss is about. It shouldn't be a matter of much contention that it is clear there has been no real effect demonstrated by n-back, and there are likely better things already out there which do have real demonstrated effects.

argumzio

[whoisbambam]

re:

and there are likely better things
already out there which do have real demonstrated effects.

elaborate, plz.

too vague.

[☉]

I already gave an example: creatine. One example is sufficient for the truth of the claim. And that there are other interventions out there for improving various skills needed for various tasks, well, that is hardly worth mentioning.

argumzio

[☉]

Oh, and I might as well mention another which continues to garner support: tDCS. But you already know about that, right?

In fact, we all already know about these things.

argumzio

[whoisbambam]

one may be enough for the claim to be true.

cognitively, creatine doesnt seem to do anything in my case--i do
continue to take it primarily because of my age of 44yrs--it may
assist in maintaining my muscle mass and also indirectly thereby
maintain testosterone levels, to some degree--something I may need to
increase/maintain now (i once read that between the ages of 30-60 a
man may lose 30lbs of muscle and gain 15lbs of fat (or vice versa,
cant quite remember)). Of course, I also take it with my daily casein
protein.

tDCS makes me 'feel better' mentally, esp. in these months.

i am still trying to learn about how to use it to possibly affect
cognitive ability, esp. beyond the 5-10minutes after the session
begins and the 2hrs post treatment period, ntm the potential different
lead placements (both anode and cathode placments) and the influence
of reference lead size, etc.

presently i have only been experimenting with 1ma for 20minutes, using
anode with about a 1.5inch by 1.75in lead and the cathode with 3inch
round reference lead on contralateral side.

[rgpddt]

Interesting you should mention creatine. I'm currently on it again (3rd time) and I am absolutely convinced that it increases my mathematical reasoning ability. To what extent, I can not be sure. It allows me to think in a more relaxed manner and without mental fatigue.

It is inexpensive and safe (at least short term), so there's is really no reason to not give it a chance and see for yourself.

[hippocampus]

offtopic: where can I try space fortress game?

[Gwern Branwen]

On Sun, Dec 16, 2012 at 2:59 PM, hippocampus <andra...@gmail.com> wrote:
> offtopic: where can I try space fortress game?

It's difficult. Search the Group archives for various attempts and guides.

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

[Michael]

I have a feeling it reduces my musical improvisational ability in spite of its breadth of other benefits.

 

I know there is a correlation between certain levels of testosterone and musical expression and I've read that creatine may increase 'T' levels. I'll think about getting tested soon enough.

[The.Fourth.Deviation.]

Considering you claim to have intellectual ability in the 5th sd, which I dont doubt, it's disenchanting that such a supposedly exceptional mind so frequently lunges at opportunities to weave intellectually primitive ad hominem attacks. On the other hand, I am more interested in exploring latent truths about N-Backing, while you bounce back and forth between attacking my ideas and discussing [attacking] my intellect. Such behavior is really unbecoming of someone theoretically so exceptional.

 

With that said, your method of responding to my points is verbose which clouds the meaning of your communique. For the sake of conversation, more succinct responses would be more productive.

 

You said my comment that Nback may improve G is vague. What I meant is that some domains may be impacted by Nback that enhance overall intelligence, although this may not necessarily be picked up by RAPM. Gf may be improved as well, but again, in such a way that it is not picked up by RAPM. For example, as you mentioned, if DNB improves focus, and focus is an important part of Gf, then it does improve Gf, although RAPM may not necessarily show this. If focus is just a part of G but not Gf, then we would say that DNB improves G , but not Gf. In such a case, I'm sure many would be glad to accept improvements in G, without any improvement in Gf. Therefore while it may be the case that DnB does not impact Gf at all, it may impact "G" or general intelligence in some way which hasn't been picked up. Again, the fact that DnB causes some changes in multiple regions of the brain, there exists a chance that some cognitive domain has been improved.

 

And secondly, modes of Nback which implement things such as task switching, for example, may be more effective than pure task-switching exercises alone. These advanced modalities of Nback are largely unstudied, and since they are prima facie more intellectually challenging than the original mode, and poly-modal, I hypothesize that they will show a more concrete impact on cognition. With that said, these modes are already available to us in Brain workshop...

[The.Fourth.Deviation.]

regarding focus: I remember you mentioned a point about training focus  on something static without DnB. I think it remains to be seen whether such a task would actually have the same outcome as focusing on DnB (meditation may be a proxy for such a static focus task), but my initial guess is that they would not be equivalent, probably as demonstrable by a fMRI test.

 

Regarding your critique of fMRI, i understand your point that we don't know what the brain changes after DnB mean or how they may impact cognition. However, that doesn't mean that the changes haven't had any impact on cognition, so again, I find it premature to rule those out, and they also give us a glimpse into what mechanisms are being targeted by DnB task training. The main point is that changes have occurred. The question is what impact those changes have had, if any, and also whether or not we are conducting the proper tests to measure such changes.

 

Therefore, my ultimate conclusion is that it remains premature to suggest with certainty that DnB has zero impact on general intelligence.

[☉]

On Dec 16, 6:54 pm, "The.Fourth.Deviation." <davidsky...@gmail.com>
wrote:

> Considering you claim to have intellectual ability in the 5th sd, which I
> dont doubt, it's disenchanting that such a supposedly exceptional mind so
> frequently lunges at opportunities to weave intellectually primitive ad
> hominem attacks.

Even some ad hominem criticisms can be non-fallacious, that is, when
such criticisms are non-tangentially related to the main issue under
discussion; and the last time I checked, that someone commits errors
in reasoning left and right should have this pointed out to them, and
that they should be more cautious when making claims which have an
elevated probability in being entirely wrong. The fact that you don't
acknowledge this and continue in a feeble attempt to one-up my valid
points only indemnifies my criticisms further. However, this is
besides any claims you seem to think I've made regarding my own
intellect. More misdirection on your part.

> On the other hand, I am more interested in exploring
> latent truths about N-Backing, while you bounce back and forth between
> attacking my ideas and discussing [attacking] my intellect.

Latent truths or not, I don't think they're your "ideas", whatever
they may be, that we've been discussing.

> Such behavior
> is really unbecoming of someone theoretically so exceptional.
>

Which further implies, considering I've committed no unbecoming
behavior (unless your simply cannot man up to being pointed out your
errors, as obvious as they are), you cannot recognize exceptional
behavior when it slaps you in the face, no?

>
> With that said, your method of responding to my points is verbose which
> clouds the meaning of your communique. For the sake of conversation, more
> succinct responses would be more productive.
>

I know, it must be a challenge for those with poor reading
comprehension. Unfortunately, I don't care to dumb down what needn't
be dumbed down.

> You said my comment that Nback may improve G is vague.

I did not merely say it is vague. I further scribed it is absolutely
wrong, for it is nonsensical to suppose that anything along the lines
of "g improving" is fundamentally incoherent. Sadly, I'm not here to
school those in psychometrics who don't care to learn about the
subject beyond the knowledge of their own psychological profile.

> What I meant is that
> some domains may be impacted by Nback that enhance overall intelligence,
> although this may not necessarily be picked up by RAPM. Gf may be improved
> as well, but again, in such a way that it is not picked up by RAPM.

It may do this, it may do that. Who cares what it *may* do? I care
about what it actually does. Once we know what that is, then there
might be something to talk about.

> If focus is just a part of G but not Gf, then we
> would say that DNB improves G , but not Gf.

No, we wouldn't, and no we don't. It is even more hilarious when one
realizes that first order factors, like gF, have been considered
identical to g itself. On such a view, which isn't spun from idle
verbal chatter, this is just nonsense in the highest degree. By the
way, focus is not exactly related to g, which we can see due to the
fact that focus can be trained. You might be interested in the
following: http://pss.sagepub.com/content/17/2/172.short . Admittedly,
we would commonly consider executive functions, such as "focus", to be
essential for intelligent behaviors, but that does not of itself imply
that EF and g are closely related; it seems, however, that there is a
sense in which EF can mediate the expression of g, but since
investment (where high learning ability gives way to the acquisition
of accumulated skills) has more or less be corroborated, it would
further appear that gF is essential in the investment process, not EF
as such. In other words, there are ways to improve non-g-related
factors and still give rise to seemingly more "intelligent" self-
management behaviors. In such a scenario, which is entirely feasible,
one does not arrive at the conclusion that the intelligence of those
involved in such remediation has "increased", only that it has become
better managed and expressed.

> In such a case, I'm sure
> many would be glad to accept improvements in G, without any improvement in
> Gf.

Uh huh.

> Therefore while it may be the case that DnB does not impact Gf at all,
> it may impact "G" or general intelligence in some way which hasn't been
> picked up.

Sure, whatever you say...

> Again, the fact that DnB causes some changes in multiple regions
> of the brain, there exists a chance that some cognitive domain has been
> improved.
>

Not necessarily. Golly, isn't there just as much a chance that some
domain of skills has been unimproved, if we want to play that line?

> And secondly, modes of Nback which implement things such as task switching,
> for example, may be more effective than pure task-switching exercises
> alone.

Yes, they certainly would be more effective, if they're effective,
when compared to something that isn't effective at all. I.e., it is
still an unsettled question if such is effective on any level
whatever. You mention in the next post "that it remains premature to

suggest with certainty that DnB has zero impact on general

intelligence". However, this is a case of reversing the burden of
proof. We are not trying to demonstrate there is no effect. We are
saying there is no evidence that there is an effect, that the claims
by others that there is such an effect are spurious and without merit,
and that we need not be swayed to suppose that there might be an
effect until such a time as some evidence has come forward to make us
reconsider our position on the matter. It is tiresome to point these
things out, but so they must – more for some than others, I guess.

> These advanced modalities of Nback are largely unstudied, and since
> they are prima facie more intellectually challenging than the original
> mode, and poly-modal, I hypothesize that they will show a more concrete
> impact on cognition.

I've said this before, but it's not as though I actually believe this
is true.

argumzio

[☉]

Your feeling seems to fall in line with what I would intuitively
expect with creatine supplementation. Namely, since it improves those
capacities involved in focus and stamina, whereas improvisation in
music necessitates the turning off of one's DLPFC, it is clear that
supplementation would not be helpful towards improvised improvement.
However, the data may seem mixed, but it is most likely the case that
DLPFC deactivation relates strongly to improvisation ability:
http://www.wpi.edu/Images/CMS/HUA-CIMA/Brain_Research_V6.pdf . I think
this makes a great deal of sense, for improvisation is supposed to be
utterly spontaneous, unthought creation in the act itself.

argumzio

[Pontus Granström]

Crap study btw, since they spread out the training over 46 days or
something. It's not even close to the Jaeggi studies.

> --
> 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.
> Visit this group at http://groups.google.com/group/brain-training?hl=en.
>
>

[☉]

On Dec 17, 7:01 am, Pontus Granström <lepon...@gmail.com> wrote:
> Crap study btw, since they spread out the training over 46 days or
> something. It's not even close to the Jaeggi studies.
>

Hah. Yeah, I knew you would have had to chime in eventually.

argumzio

[Pontus Granström]

Well there are many positive and a few negative studies, what makes
them different? There's your answer to what makes brain training work
or not.

[☉]

I explain your response as an ad hoc attempt at rescuing the
hypothesis. Let's go ahead and wait for the next study to show that
this is another pellet of drivel. Not to mention the fact that your
figure of 46 days is only an average. Plenty of subjects took more or
less time than that. If the difference in time to complete all
sessions is so stark, don't you think Redick et al. would have noticed
the discrepancy, if it is as real as you seem to think?

Besides, why should training duration matter, if the only thing that
seemed to matter was training dosage? There isn't even a theoretical
basis for why we should be given to think duration has much of an
effect.

argumzio

[Pontus Granström]

It's still very interesting that so many studies gives positive
results. This study might show that during some conditions like high
ability with non-standardized training gives null results.

[Zaraki]

Actually, it's pretty obvious that many of the studies with positive results get them because of an absence of active control groups.

[jttoto2]

Agreed.  Taken as a whole, active control groups have wimpy results.

[The.Fourth.Deviation.]

Your fixation on ad hominem attacks is laughable;  attempts to bait others, by insult, to engage in such jousting is intellectually inferior and useless, yet you've an insatiable desire to partake in it.

To your points of importance: 

You express doubt about the efficacy of N-Back, however your doubt clearly stems from uncertainty rather than certitude of inefficacy. You admit that we don't clearly understand why the brain changes the way it does after nback training, and also admit that you don't know what effects those changes have. Your lengthy analysis of the topic converges to this expression of uncertainty. If you disagree with my characterization and you are certain about the probability of Nback being effective, why not put forth a 99% confidence interval of the probability that it is completely ineffective [at improving Gf, or any area of intelligence], as such a quantified degree of certainty is magnitudes more meaningful and useful than dozen-sentence paragraphs which ultimately convey uncertainty with great certitude.

In one sentence you to strongly imply and suggest that people who train Nback are wasting their time, and in your next admit that such training may potentially have effects that are not known to us nor to you. This confounds logic; which one is it? Again, how about some confidence intervals so we can quantitatively see how certain you are???

[☉]

On Dec 17, 9:15 pm, "The.Fourth.Deviation." <davidsky...@gmail.com>
wrote:

> Your fixation on ad hominem attacks is laughable; attempts to bait others,
> by insult, to engage in such jousting is intellectually inferior and
> useless, yet you've an insatiable desire to partake in it.

Speak for yourself, David. Really, this crap is tiresome and
childish.
I know it must be hard for you, "The Fourth Deviation", to realize
that other people in this world see the obvious errors in reasoning
and logic you continue to commit, but that's just the way it is.

> To your points of importance:
> You express doubt about the efficacy of N-Back, however your doubt clearly
> stems from uncertainty rather than certitude of inefficacy.

Oh, please. This is a cute straw man. The first requisite of proper
communication is to be able to restate the other person's position,
and that's only the very basic level which you failed to meet here.

> You admit that
> we don't clearly understand why the brain changes the way it does after
> nback training,

No, I never "admitted" that. In fact, I've frequently mentioned that
it is obvious, on account of neuroplasticity, that the brain will
adapt to meet the demands of any given task. Any more straw men? Let
us see.

> and also admit that you don't know what effects those
> changes have.

Never said that either. I can tell you that the immediate effect such
changes would have (whatever such changes must involve) pertain to
improvement on a given task, obviously. It's almost a no-brainer!
Hyuk
hyuk!

> Your lengthy analysis of the topic converges to this
> expression of uncertainty.

No, it doesn't. However, the more you reply in this vein, the more
certain I become that your responses are ridiculous, incompetent, and
not worth the time to continually reply to by correcting,
recorrecting, and so on. Apologies if this sounds too ad hominem, but
you haven't demonstrated why I should really be concerned with
anything you might have to say....

> If you disagree with my characterization and you
> are certain about the probability of Nback being effective,

Oh, there is no doubt I disagree with your characterization. It is
utterly false. The fact that you continue to sidestep my valid points
– and I'm generously assuming such, for it would be thoroughly
surprising if they haven't even been recognized as such – shows that
you're only here to waste my time. As for n-back being "effective" it
is clear to date that it has not been demonstrated to have any real
effect.

> why not put
> forth a 99% confidence interval of the probability that it is completely
> ineffective [at improving Gf, or any area of intelligence],

Why not look at the current data and do that yourself? What the hell
does this have to do with what has been said hitherto, anyway? Oh,
that's right, nothing at all. More misdirection.

> as such a
> quantified degree of certainty is magnitudes more meaningful and useful
> than dozen-sentence paragraphs which ultimately convey uncertainty with
> great certitude.

As I've argued, in very clear terms, that the data is what is
important, I am willing to venture you have nothing more to
contribute
than "dozen-sentence paragraphs which ultimately convey [certainty]
with great [incertitude]".

> In one sentence you to strongly imply and suggest that people who train
> Nback are wasting their time,

Yes, I would not only imply but outright express this, particularly
given the fact that it hasn't been demonstrated to lead to the
benefits claimed.

> and in your next admit that such training may
> potentially have effects that are not known to us nor to you.

Well, I do not know what effects the rings of Saturn and the farts of
a fruit fly might have on the quantum particle interactions in your
brain, but I'm willing to venture you don't either. Who the hell
cares? Oh wait, you seem to be putting forward an argument from
ignorance. How quaint.

> This
> confounds logic; which one is it?

Pffft. Are you for real, dude?

> Again, how about some confidence
> intervals so we can quantitatively see how certain you are???

Actually, I've argued against the use of subjective confidence
intervals (as in the Bayesian mode of subjective probabilities),
because they are largely arbitrary and reflect subjective biases.
Hence, my sentences are more meaningful than an artificially
quantified confidence interval would be. I much prefer that we look
at
the hard data, and they do not speak much to n-back as far as
improving WM or measures of gF to any degree that would justify the
amount of time people think is required of them to see such
improvements. Those prone to the snake oil of placeboas may want to
inquire into finding some kind of anti-venom, for example, by
exercising their feeble reasoning powers they so desperately need to
improve and read the data that is out there on the subject; however,
such is a matter of taste, so I am quite willing to let people
believe
whatever they want to believe. Even so, this doesn't mean I actually
think a Jamesian "will to believe" line of thought is at all valid or
sensible either in the short term or long term.
Then again, with any genetic luck, someone could pop a few pills and
see the gains in WM they'd like.

argumzio

[☉]

Too, too obvious...

argumzio

[Pontus Granström]

EVERYBODY who has engaged with a real effort in n-backing knows the
benefits of greater clarity, stamina/focus, updating of working memory
and quicker thinking, all which plays a MAJOR role in what we lump
together as matrix reasoning, which we call intelligence. But then
again REAL STRENUOUS EFFORT it what makes it work. When I do not put a
real effort into it, I feel nothing from it. So given high ability
groups with an already high WM that spread out their training over 46
days it's not so surprising that they don't improve. Again the
principles of training. If I take people who run 3k in say 11:30
minutes and put them through vo2max training load way below their
maximum and more than double the time really needed it might not work.

[whoisbambam]

argumzio n the.fourth.deviation,

enlighten me: why do you two 'converse' when neither of you are able
to change/influence/improve the other (in any 'meaningful' way, that
is)?

it seems illogical n wasteful/futile (this hasn't been the first time)

there is another purpose?

some 'emotional'/amygdala/reptilian origin?

because it doesn't seem to be within the domain of the neocortex

i could be mistaken, and thus my inquiry.

[polar]

David we're here for years, and I would like to remind you, that argumzio doesnt mean to be like that (I mean opposing, verbally exhibiting, self-important, patronizing, never giving credit or acknowledging "you're right"). You know he's like that, not only to you, but to virtually anybody who openly disagrees with him (usually me, pontus). I bet several people here are at his or higher level of intelligence, but he simply has the time and need to defend his perspective to the absolutely meaningless point. Lets not waste our time anytime somebody in this forum gets into similar modus (it happens to me and you too of course). For it doesnt say anything about the your intelligence or the topic itself at this point, but it says something about your decency. peace

Dne pondělí, 17. prosince 2012 2:05:28 UTC+1 The.Fourth.Deviation. napsal(a):

[☉]

Of course polar would take this opportunity, as misplaced as it is, to side with another individual incapable of discussing these matters at all seriously or intelligently, for he himself has on many occasions stooped to low-level insult all the while playing the victim. Sadly for him, I do not take kindly to such games.

In response to whoisbambam, the matter never centers around convincing the other of anything. To my mind, this is a mere exercise, a suitable playground meant only for those willing to respond in equanimity to the points raised. I have clearly fulfilled my duty in this, whereas all I typically receive is nonsense from who cannot bear the thought that they are beholden to such ideas with no meaningful justification. Besides, David does not deserve civility as his past behavior towards me shows, so he shouldn't be surprised at any outcome when I must interact with his character.

Any further post irrelevant to this thread will not receive a response from me. Fair notice.

argumzio

[whoisbambam]

very well.

thanks for your explanation.

[The.Fourth.Deviation.]

"Never said that either. I can tell you that the immediate effect such 
changes would have (whatever such changes must involve) pertain to 
improvement on a given task, obviously. It's almost a no-brainer! 
Hyuk 
hyuk! "

Improving the underlying "task" is equivalent to improving the ability to recall and organize increasingly lengthy parallel streams of arbitrary stimuli. At which point does changing said arbitrary stimuli to objects encountered in real life cause said improvements to suddenly become null? Certainly at least some things encountered in real life can be considered "arbitrary stimuli" and if we have been training to recall more and more such arbitrary stimuli, its not certain how their source (non-computer based or computer based) is of massive material importance when it comes to processing them in memory.

The closer any "real-life" stimuli is to resembling something 2d as seen in Nback, the more transfer might be seen. In addition, such problems of comparison only stem to the visual portion of nbacking, as the auditory component (which can be changed to music notes, beeps, etc in Brain Workshop) is a rather decent proxy for sound in general.

On Tuesday, May 22, 2012 1:00:29 PM UTC-5, gwern wrote:

> Numerous recent studies seem to provide evidence for the general intellectual benefits of working memory training. In reviews of the training literature, Shipstead, Redick, and Engle (2010, in press) argued that the field should treat recent results with a critical eye. Many published working memory training studies suffer from design limitations (no-contact control groups, single measures of cognitive constructs), mixed results (transfer of training gains to some tasks but not others, inconsistent transfer to the same tasks across studies), and lack of theoretical grounding (identifying the mechanisms responsible for observed transfer).
>
> The current study compared young adults who received 20 sessions of practice on an adaptive dual n-back program (working memory training group) or an adaptive visual search program (active placebo-control group) with a no-contact control group that received no practice. In addition, all subjects completed pre-test, mid-test, and post-test sessions, comprising multiple measures of fluid intelligence, multitasking, working memory capacity, crystallized intelligence, and perceptual speed. Despite improvements on both the dual n-back and visual search tasks with practice, and despite a high level of statistical power, there was no positive transfer to any of the cognitive ability tests. We discuss these results in the context of previous working memory training research, and address issues for future working memory training studies.

Total n=75; RAPM was speeded. The multiple measures is something like
17 tests, which took the subjects 2 and a half hours to do just the
pre-test - I don't envy their subjects. My copy is very long and I had
to relatively skim it, but I have no initial objections (even the
speeding of the RAPM is a good thing inasmuch as it turned in a null
anyway). I already mentioned the very interesting survey results
suggesting placebo effects.

In press, estimated to be available online in 5-6 weeks. Don't bother
contacting me for a copy, I promised Redick I would not share it.

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

[☉]

On Dec 19, 8:30 pm, "The.Fourth.Deviation." <davidsky...@gmail.com>
wrote:

> Improving the underlying "task" is equivalent to improving the ability to
> recall and organize increasingly lengthy parallel streams of arbitrary
> stimuli.

Yes, but that does not therewith imply that one will see the same
result in other tasks which also involve "the ability to recall and

organize increasingly lengthy parallel streams of arbitrary stimuli".

In other words, the stimuli may have an effect such that one's
improvement on the task is purely task-specific, with little to no
transfer to other tasks. One can further observe such examples in
memory, where someone can train themselves in the fashion of multiple
mentalist Harry Kahne to do many things in a rapid task-switching
manner, but whose overall ability clearly isn't all that improved over
normal subjects. Better yet, one can consider examples of persons who
have trained many years on certain tasks, while such training doesn't
lead to seemingly obvious improvements in related domains. This is the
central issue regarding the claim of far-transfer which is
increasingly called in to question w.r.t. DNB training: seeing that
there isn't much improvement on seemingly similar tasks, it is
unreasonable to suppose that task improvement will correlate or be
representative of improvement on other tasks of one's arbitrary
choice.

> At which point does changing said arbitrary stimuli to objects
> encountered in real life cause said improvements to suddenly become null?

I will assume that this isn't a rhetorical question. For purposes of
discussion, I suppose a suitable answer would be to consider that
there is little ecological validity in the tasks themselves. In other
words, no matter the demands of the task itself, one will not readily
observe similar levels of improvement to non-related tasks in one's
workaday life, unlike how such could be determined with a pre-test,
post-test design with some intelligence tasks. Indeed, one is better
off practicing the specific tasks in which one desires to see
improvement, than practicing a task only supposedly superficial to
those tasks. In such a scenario, It could be pointed out that anyone
who practices on DNB would probably hope to see improvement in other
tasks, and in being so motivated they would try harder at them by
being primed to perform at their optimal level of responsiveness. In
this way, the improvement to those other tasks wouldn't actually
reflect transfer on account of training and task-specific improvement,
where the training task has nothing to do with those skills one uses.
Only insofar as the training task relates to employed skills would one
have hope of observing improvement on other tasks which make use of
those skills in a similar fashion. And where two tasks employ similar
skills, one cannot further argue that there is evidence of far-
transfer per se. For instance, speaking hypothetically, were I to use
visualization in my n-back training, where I otherwise hadn't used it
nearly as much in my daily life, I am sure that I would see a general
improvement in visualizing things in my daily life - but that is only
on account of my making use of such a skill (visualizing), not
necessarily having much to do with the training itself. Such secondary
effects could lead to dubious results in the literature, which is
probably why the usefulness of active control groups has become all
the more pressing.

> Certainly at least some things encountered in real life can be considered
> "arbitrary stimuli" and if we have been training to recall more and more
> such arbitrary stimuli, its not certain how their source (non-computer
> based or computer based) is of massive material importance when it comes to
> processing them in memory.

I understand what point you're making here, David. However, it is not
clear to what extent we can argue that the general improvements people
claim to perceive are real or imaginary. I do not wish to imply that
people are always fooling themselves, but it is more common than one
might suppose, which means we must be cautious in our observations.
Introspection is remarkably unreliable, especially given the spate of
cognitive biases we've come to know. That said, I have personally
liked the idea of making life more and more like an n-back task; this
feeling was very strong while I was involved in assiduous and
prolonged training. Throughout the days, my tendency to remember idly
all previous instances up to that time would be engaged considerably.
However, the fact that my daily environment isn't even 1/20th as
demanding as QNB is, this tendency naturally waned due to its
fruitlessness. I would by no means deny that it is possible for n-back
tasks to be highly relevant in dynamic, information-sensitive
environments, but there are only so many such environments which one
would not already be able to meet their demands through training and
familiarity therewith. This I leave as an open question, for it is
generally the case that one cannot bring a lab just anywhere and query
people every ten minutes throughout the day, granting us the kind of
resolution we would likely need to be able to determine the daily
impact such training might afford. Hence, the inferential methods we
see in studies published to date. Do not think I am satisfied with
this situation, as I would be very interested in learning what is
possible. But the case needs to be made, not assumed.

> The closer any "real-life" stimuli is to resembling something 2d as seen in
> Nback, the more transfer might be seen.

As I said, yes, I have thought this as well. Apologies if I'm
repeating myself here, as I'm only responding to your post point by
point as I read it. The issue there is by no means unclear.
Nonetheless, it is difficult to ascertain to what extent very much in
real-life is all that analogous to an implementation of n-back like
the one we know. For one, there is the artificial layering of
abstraction which we would use to treat such real-life data as a task.
Surely this requirement isn't forced upon us while we are training on
the task. I, therefore, submit that training at n-back is like
learning a new skill entirely, which will only have relevance insofar
as one uses that skill, like mental arithmetic, in one's daily life;
and hence, the skill only becomes relevant when a problem is amenable
to such treatment. Not all problems in the world require much in the
manner of remembering n-length sequences and updating them at each
discrete data point; were such the case, we would already be able to
see people starting at D6B from the start without any sort of problem:
we all had to start at n=1 or n=2 and develop proficiency from there.
Besides this, it is interesting to ask why some cannot break the n=5
plateau while others can go as far as n=14 (or greater) at DNB; is
anyone else curious about that?)

> In addition, such problems of
> comparison only stem to the visual portion of nbacking, as the auditory
> component (which can be changed to music notes, beeps, etc in Brain
> Workshop) is a rather decent proxy for sound in general.

You by no means make a case I haven't previously considered. However,
this line of thought is quite different from what is being pursued in
the research. I by no means possess the resources to settle the matter
to any degree of reliability I would like, as indeed very few of us
here do. The issue with sounds, however, is quite different, for even
as a proxy, it doesn't exactly represent the typical tasks we are
forced to manage in our daily lives. For instance, listening to boring
lectures of indeterminate length is quite different from the process
of listening for repeated letters. Music notes are perhaps the most
appropriate representation of the task which people have to manage,
namely, listening to and reproducing music; however, there is such
evidence of task-specific learning in that domain beyond n-back that
one wonders to what extent it is even seriously relevant at all.
(Think of examples of persons who can listen to a piece of music once
and play it back exactly. In no way would the auditory portion of an n-
back task represent any kind of challenge to them; however, the
reverse isn't true for who desires to do this by learning musical
notation through n-back to do the same later on. The only way to get
that good at musical performance is to practice performing and develop
the desiderata of long-term memory, dexterity, etc.) As such, I don't
think this bears on much that has been discussed so far.

argumzio

[jotaro]

most of the times argumzio is pretty amusing, though i find it odd that he devotes so much time into ti.

 

[Michael]

Ahh, cheers for your help! Need to do more research here.

Mike