No transfer (WT Chooi 2011)
XFMQ902SF
able to improve fluid intelligence (gf) by training working memory.
Subjects who trained their working memory on a dual n-back task for a
period of time showed significant improvements in working memory span
tasks and fluid intelligence tests such as the Raven’s Progressive
Matrices and the Bochumer Matrices Test (BOMAT) after training
compared to those without training. The current study aimed to
replicate and extend the original study conducted by Jaeggi et al.
(2008) in a well-controlled experiment that could explain the cause or
causes of such transfer if indeed the case. There were a total of 93
participants who completed the study, and they were randomly assigned
to one of three groups – a passive control group, active control group
and experimental group. Half of the participants were randomly
assigned to the 8-day condition and the other half to the 20-day
condition. All participants completed a battery of tests at pre- and
post-tests that consisted of short timed tests, a complex working
memory span and a matrix reasoning task. Participants in the active
control group practiced for either 8 days or 20 days on the same task
as the one used in the experimental group, the dual n-back, but at the
easiest level to control for Hawthorne effect. Results from the
current study did not suggest any significant improvement in the
mental abilities tested, especially fluid intelligence and working
memory capacity, after training for 8 days or 20 days. This leads to
the conclusion that increasing one’s working memory capacity by
training and practice did not transfer to improvement on fluid
intelligence as asserted by Jaeggi and her colleagues (2008, 2010).
http://etd.ohiolink.edu/view.cgi?acc_num=case1301710207
Pheonoxia
et al.'s is because its post-training intelligence tests didn't use
time constraints. When participants have unlimited time, there's no
significant increase in performance. Not what anybody on this forum is
trying to hear.
But this study's authors do assert that working memory training may
have its uses,
"Although repeated practice on a challenging task such as the dual N-
back task did not show any significant transfer effects to g or
related mental abilities comprising g, it may not be wise to
completely reject the notion that there are no practical benefits to
training one’s working memory. There are many studies in the
literature that suggest positive outcome from working memory training,
such as reduced inattentive symptoms in ADHD children (Klingberg et
al., 2002; Klingberg et al., 2005), increased memory performance in
older adults (Buschkuehl et al., 2008), increased math performance in
children with working memory deficits (Holmes et al., 2009), improved
short term memory in adolescents with borderline intellectual
disability (Van der Molen et al., 2010), reduced cognitive deficits in
schizophrenic patients (Wykes et al., 1999), significant reduction in
symptoms of cognitive problems in patients with stroke (Westerberg et
al., 2007) and improved fatigue symptoms in adults with multiple
sclerosis (Vogt et al., 2009). Some of these studies reported no
improvement in fluid intelligence (Westerberg et al., 2007; Holmes et
al., 2009; Van der Molen et al., 2010) and some reported significant
improvement (Klingberg et al., 2002; Klingberg et al., 2005).
"Anecdotal examples in the current study have been encouraging where
some participants in the study claimed that they have improved their
focus in general. One participant commented on the post-test
questionnaire:
'I feel like sometimes I struggled to keep my attention after 12-15
minutes but have noticed during my voice lessons I can focus more on
multiple techniques at once, which takes a lot of focus!'
"This particular participant also mentioned to the experimenter that
she first noticed increase in focus when her voice lesson instructor
commented on her improved techniques."
Gwern Branwen
This is an excellent find.
Notes:
- more subjects than Jaeggi 2008
- trained some longer than Jaeggi 2008 (19 vs 20 days)
- Chooi approvingly discusses Moody, pg 16-19
- RAPM was administered to all participants *untimed*
- pretty decent variety in the test-battery:
> "The tests administered in the pre- and post-tests include the
Mill-Hill vocabulary test, vocabulary tests (part I and II) from the
Primary Mental Abilities test battery, Word Beginning and Ending test,
Colorado Perceptual Speed Test, Identical Pictures, Finding A’s, Card
Rotation and Paper Folding from the ETS test battery, Shepard-Metzler
Mental Rotation Test (1971) and Raven’s Advanced Progressive Matrices
(1990)."
- more similarities:
> "Participants in the 2-week (8 days) condition had a 34%
improvement and those in the 5-week (20 days) condition improved by
44%. Participants in the Jaeggi et al. (2008) study displayed similar
trends. From the data that they published, it can be estimated that
participants in the 8-day training condition improved by 34% and those
in the 19-day condition improved by 47% (Jaeggi et al., 2008). The
numbers suggested that participants from the current study and the
original Jaeggi et al. (2008) study showed very similar performance on
the training task."
- not sure how to interpret this:
> "A simple regression with improvement in N-back task as
predictor was conducted using data from participants in the 5-week
training group. This predictor did not significantly contribute to any
variance in gain scores for all the variables in pre- and post-tests."
- amusing:
> "Some gender differences were observed – male participants in
the study achieved better SAT Math scores (Cohen’s d = 0.76). They
also performed better on Mental Rotation (Cohen’s d = 0.60) and Card
Rotation (Cohen’s d = 0.52) tasks. Female participants, on the other
hand, scored significantly higher than their male counterparts on Word
Beginning and Ending tests (Cohen’s d = 0.35). They also did better on
the perceptual speed test Finding A’s at both pre- and post-tests
(Cohen’s d = 0.67 and 0.60 respectively). In this test, participants
went through long lists of words and crossed out words that have the
letter A."
- results; penalties to training?
> "Results from dependent sample t-tests suggested no significant
improvement overall after training. There were significant declines in
performance on some of the verbal fluency and all of the perceptual
speed tests, but this trend is consistent with a previous study
suggesting that the items used in the post-test are potentially more
difficult than the items used in the pre-test. Looking at the results
for participants who trained for 2 weeks, they did not show any
significant decline in performance, which may suggest that they
actually improved on speed after training. These participants also
improved significantly on the Card Rotation test, and they showed some
improvement on the Paper Folding test even though it was not
significant. The same cannot be said of those who trained for 5 weeks.
They showed significant declines in performance on the verbal fluency
and perceptual speed tests except for Finding As. They did show
improvement on Paper Folding and Mental Rotation albeit insignificant.
Interestingly, those who did not train took the post-test 5 weeks
after pre-test improved significantly on Card Rotation."
- ceilings:
> "Jaeggi and her colleagues (2008, 2010) argued that the speeded
administration of the transfer tasks was comparable to the non-speeded
administration, and they decided to administer the tests with time
limits to avoid ceiling effects. Results from the current study did
not suggest any ceiling effects of any for the tests administered
(RAPM pre-test mean = 12.5, SD = 2.65; RAPM post-test mean = 12.3, SD
= 2.55; there were 18 items on pre- and post-tests)."
And finally, something I've said many times - that it looks like the
benefits do not include IQ but other things:
> "There are many studies in the literature that suggest positive outcome from working memory training, such as reduced inattentive symptoms in ADHD children (Klingberg et al., 2002; Klingberg et al., 2005), increased memory performance in older adults (Buschkuehl et al., 2008), increased math performance in children with working memory deficits (Holmes et al., 2009), improved short term memory in adolescents with borderline intellectual disability (Van der Molen et al., 2010), reduced cognitive deficits in schizophrenic patients (Wykes et al., 1999), significant reduction in symptoms of cognitive problems in patients with stroke (Westerberg et al., 2007) and improved fatigue symptoms in adults with multiple sclerosis (Vogt et al., 2009). Some of these studies reported no improvement in fluid intelligence (Westerberg et al., 2007; Holmes et al., 2009; Van der Molen et al., 2010) and some reported significant improvement (Klingberg et al., 2002; Klingberg et al., 2005). Anecdotal examples in the current study have been encouraging where some participants in the study claimed that they have improved their focus in general. One participant commented on the post-test questionnaire:
>
>> 'I feel like sometimes I struggled to keep my attention after 12-15 minutes but have noticed during my voice lessons I can focus more on multiple techniques at once, which takes a lot of focus!'"
--
gwern
http://www.gwern.net
XFMQ902SF
Perhaps in 20 years or so there will be effective ways to become more
intelligent either through nanotechnology or whatever.
On Feb 20, 6:38 pm, Gwern Branwen <gwe...@gmail.com> wrote:
>
> - Show quoted text -
Michael
polar
skeptic and critical to this ONE, as you are to many others
(particularly to those confirming jaeggi). Of course Jaeggi research
had many flaws and NOBODY here takes for granted they "raised
intelligence ten points in 19 days." But I can tell you right now,
were these researches serious about measuring intelligence (or
replicating jaeggi), they would use bomat, or actuallly any iq test
that is not 20 years old... and they would use more of them (like we
are - and there is a clean difference between raven and bomat).
Nevertheless, this are very important findings and any research on n-
back is seriously welcomed, great find indeed. Looking forward to
study it in detail in one week.
On 21 ún, 00:38, Gwern Branwen <gwe...@gmail.com> wrote:
Gwern Branwen
> Gwern we all have our favourite researches :) - I hope you will be as
> skeptic and critical to this ONE, as you are to many others
> (particularly to those confirming jaeggi).
The only criticism I could think of for this one was the significant
attrition: ~120 to ~90 subjects. But she checked that the attrited
were not different on any of the metrics she had for them, so I didn't
bother to include it. The gender balance - 60 females (greater
Conscientiousness, eh?) - was also odd, but there's no way an
imbalance of 15 females could possibly lead to the observed null
result.
> Of course Jaeggi research
> had many flaws and NOBODY here takes for granted they "raised
> intelligence ten points in 19 days." But I can tell you right now,
> were these researches serious about measuring intelligence (or
> replicating jaeggi), they would use bomat, or actuallly any iq test
> that is not 20 years old... and they would use more of them (like we
> are - and there is a clean difference between raven and bomat).
Yes, so just like Jaeggi didn't use a flawed 20 year old IQ test...?
This study redid Jaeggi 2008, with the major flaw removed, with bigger
n, and more tests to boot, and without the questionable post hoc
analysis strategy of Jaeggi 2010. This is better than Jaeggi 2008, so
you can't debunk it without also _a fortiori_ debunking Jaeggi 2008,
and it supports what I've been saying all along, against the strenuous
opposition of posters like Pontus: Moody was right.
--
gwern
http://www.gwern.net
XFMQ902SF
Gwern Branwen
> so is n-back debunked officially?
Officially? No. Any one study is untrustworthy (as I emphasized a
while back). But there are, what, 2 or 3 studies due out this and next
year which we recently discussed; if they all come back null that it
is looking pretty bad for the strong IQ claims.
Which still leaves the non-IQ benefits, of course. I really need to
compile them all into a good section for the FAQ, but I keep getting
distracted! Like yesterday and today, by lithium -
http://www.gwern.net/Nootropics#lithium
--
gwern
jttoto2
transfer on some tasks.
I think we have to be open to the possibility that the ones that
benefit the most are the ones most cognitively afflicted.
Gwern Branwen
> I think we have to be open to the possibility that the ones that
> benefit the most are the ones most cognitively afflicted.
That's pretty boring though. The afflicted do not number most of us
here in their ranks, nor will they for decades to come, and we already
know of a great many interventions which help the afflicted. (We've
covered a few here already.) For some such groups, like the elderly,
it seems like any mental workout helps! This is not what made
n-backing famous.
--
gwern
http://www.gwern.net
jttoto2
cognitively enhance the average person (or above-average). I just
don't think we will find it in n-back.
On Feb 20, 10:06 pm, Gwern Branwen <gwe...@gmail.com> wrote:
whoisbambam
i do believe that for some ppl, myself included, that it helps with
focus/attention. i can do more study of material when i do dnb
regularly.
i have resumed practice, 3times a week--i can not STAND my distracted
mind.
it seems dnb gives me more 'willpower' to study and to focus, stay on
task.
also, my working memory was horrendous-- i probably had some other
issue at 42 that either dnb or some of the supplementation i took has
addressed.
Jonathan Toomim
He had 93 subjects, but only 13 of them were assigned to the group that
they expected gains from, and only 22 in either of the two training
groups. The other 80 subjects were assigned to one of the 5 control
groups he had.
On 2/20/2012 2:29 PM, XFMQ902SF wrote:
> A study by Jaeggi and her colleagues (2008) claimed that they were
> able to improve fluid intelligence (gf) by training working memory.
> Subjects who trained their working memory on a dual n-back task for a
> period of time showed significant improvements in working memory span
> tasks and fluid intelligence tests such as the Raven�s Progressive
> Matrices and the Bochumer Matrices Test (BOMAT) after training
> compared to those without training. The current study aimed to
> replicate and extend the original study conducted by Jaeggi et al.
> (2008) in a well-controlled experiment that could explain the cause or
> causes of such transfer if indeed the case. There were a total of 93
> participants who completed the study, and they were randomly assigned
> to one of three groups � a passive control group, active control group
> and experimental group. Half of the participants were randomly
> assigned to the 8-day condition and the other half to the 20-day
> condition. All participants completed a battery of tests at pre- and
> post-tests that consisted of short timed tests, a complex working
> memory span and a matrix reasoning task. Participants in the active
> control group practiced for either 8 days or 20 days on the same task
> as the one used in the experimental group, the dual n-back, but at the
> easiest level to control for Hawthorne effect. Results from the
> current study did not suggest any significant improvement in the
> mental abilities tested, especially fluid intelligence and working
> memory capacity, after training for 8 days or 20 days. This leads to
> the conclusion that increasing one�s working memory capacity by
Gwern Branwen
> He had 93 subjects, but only 13 of them were assigned to the group that they
> expected gains from, and only 22 in either of the two training groups.
And presented power calculations to justify the smallness.
--
gwern
http://www.gwern.net
jttoto2
backing. (no gains in fluid intelligence) I have the feeling that n-
back is enhancing 'something', but whether that 'something' is
detectible by current cognitive measures is another story, and what
that 'something' is just raises more questions. Perhaps it is simply
sustained attention.
And yikes, I just read this: "The same cannot be said of those who
They showed significant declines in performance on the verbal fluency
and perceptual speed tests except for Finding As."
into a single task can't be good for the brain. Familiarity kicks in
and the cognitive demand of that task could become less demanding than
everyday life. Whether these losses are caused directly by n-back or
simply people zoning out because the task has become too easy needs to
be addressed. What is disappointing is that it has been almost a
decade since Klingberg and no one has been able to filter out why some
studies show benefits while others don't. Further studies need to
identify mediators as to who benefits and why. The why needs to be
identified in particular. Why? If these positive studies show
transfer simply due to a novelty effect, trying anything novel and
demanding enhances cognition, then training wouldn't benefit in the
long-run (because it is no longer novel and demanding), and once
learning sets in, may actually do some harm (because people are going
through the motions as opposed to actually exercising their brain).
XFMQ902SF
be too rude or something.
A couple things before I throw in the towel-
1. Bickel et al and Houben et al were able to demonstrate an
improvement in delay discounting/ reduced impulsiveness after working
memory training tasks. N-back was not part of their training-some
other tasks like reverse digit span were used. Maybe working memory
training can help in other ways which might benefit healthy
adults(reduced impulsivity).
2. Certain cognitive training games can improve spatial ability which
led to better grades.
http://www.psychologicalscience.org/index.php/publications/observer/obsonline/longitudinal-impacts-of-3-d-spatial-training-among-gifted-stem-undergraduates.html
So it might be that the n-back task is not effective for some reason
in improving fluid intelligence but other cognitive training games
not yet developed can be helpful. Just because this n-back game does
not work, does mean other interventions won’t. How long we must wait
for this new game, or drug, or whatever to augment our intelligence is
the real debate.
Jonathan Toomim
and larger than has been found in almost all of the other DNB research.
If the true effect size were more typical of earlier DNB research,
Chooi's study would only have a 17 to 60% chance of detecting it.
On 2/20/2012 8:06 PM, Gwern Branwen wrote:
> And presented power calculations to justify the smallness.
In that power analysis, they used an effect size estimate of (Cohen's) d
= 0.98, and calculated a power for his study of 0.8 based on that effect
size. Yes, that's what was found for DNB on the Raven's in Jaeggi 2010,
but it's still really high. What if the actual effect size is lower than
that, like d=0.5? This page says their power would have been about 0.25
(http://www.stat.ubc.ca/~rollin/stats/ssize/n2.html
<http://www.stat.ubc.ca/%7Erollin/stats/ssize/n2.html>). For d=0.75,
about 0.48. For d=0.40, about 0.17. For a power of 0.8 with an effect
size of 0.5, they would have needed a sample size of 63 in each group.
With their 6-group design (and still using a paired t-test, which is not
optimal for this design), that would mean 378 subjects total to have an
80% chance of detecting an effect size of 0.5 standard deviations (or ~7
IQ points). And that's without correcting for multiple testing.
(Note: that page assumes that the two groups being compared have the
same sample size, whereas Chooi's study had heavily unbalanced groups.
Though in practice, the size of the smaller group matters more, this
still means that the power estimates spat out by that page (and which I
reported above) are underestimates for Chooi's study, so maybe their
actual power with d=0.75 is about 60%, not 48%.)
It's worth mentioning also that 0.98 was the largest effect size seen
for any of the treatments and transfer tasks in Jaeggi 2010 (DNB to
Raven's). Looking at DNB to BOMAT, the effect size was only 0.49; for
SNB to Raven's and BOMAT, it was 0.65 and 0.70, respectively. Also, the
no-contact group had an effect size of 0.09 (Raven's) or 0.26 (BOMAT),
which really should be subtracted from the treatment groups' effect
sizes, since we don't care if people get better, but only if they get
better because of the training. For comparison, Jaeggi 2008 found an
effect size of 0.65 for the treatment group vs. 0.25 for the no-contact
group. In any case, 0.98 is definitely *not* a conservative effect size
estimate, which is what you should be using when performing power
calculations.
Actually, Chooi's choice of effect sizes to use for his power analysis
kinda makes it seem like he's trying to gloss over the fact that his
sample size per group was so small. "Yeah, I didn't have many subjects,
but look! Power analysis! Can I have my PhD now?" (I'm not usually one
for ad hominem attacks, but this particular instance seems like the
equivalent of using 1.5 inch margins on a 3 page homework assignment.
Sure, there are instances in which 1.5 inch margins can be appropriate
-- such as when you expect your grader to be running out of space to
write comments -- so it could just be a case of bad judgment, but in
this case it feels pretty slimy. Especially since in most of the paper
he writes about Jaeggi 2008 (which has a smaller effect size), but when
it comes to the power analysis he chooses the biggest effect size from
Jaeggi 2010. Plus there's the fact that Chooi's dissertation is dated
March 2011, so if he did a power analysis using Jaeggi 2010's effect
size, it was clearly a post-hoc power analysis.)
What I get from this study: The effect size for DNB training is probably
less than 0.98. (Of course, that's what I believed anyway before I saw
this.) The effect size could quite reasonably still be as high as 0.75.
Even if the true effect size of DNB training were only 0.3 standard
deviations (~5 IQ points), I think that would still be a pretty big deal
and totally worth my time.
Still haven't read most of Chooi's paper, unfortunately. If only I had
minions to do my engineering work for me, then I could spend more time
on science... sigh.
Jonathan
Jonathan Toomim
http://psychology.case.edu/faculty/thompson_grads.html
On 2/20/2012 11:16 PM, Jonathan Toomim wrote:
>
> Actually, Chooi's choice of effect sizes to use for his power analysis
> kinda makes it seem like he's trying to gloss over the fact that his
> sample size per group was so small....
polar
Sorry, you're plain wrong. They used more tests, but less iq tests.
They used more participants, but less actual n-backers (to the point
where your standard error is maybe bigger than the possible gains).
Gwern do you really think that all the studies which regularly are
finding iq improvements after cognitive training are flawed or fraud?
Of course moody was partially right - but do you really think that
speed is NOT a part of intelligence? Yes, for some people dual n-back
doesnt work AT ALL. For others through, it definitely works. Maybe
there is some interesting trait or factor thats intervening, whaddya
say?
When there are two extremes (nback can make 15 points in one month VS
null hypothesis), there's not solution in the middle - there's a
problem. A fascinating one and of pretty high impoortance. So please
let us remind us, that the truth is what we seek here - if n-back (or
nootropics ;) proves really inefficient in improving iq, I'll be 100%
dedicated to it and spread the word. But in now way I can do that now
and the story continues, actually we are in still the most interesting
part of it. Greetz.
Arbo Arba
fact the WM varies over our lifetime and is even effected by how
stressed we are and how much sleep we get, but IQ stays steady -- So
people with IQ's of 130 will test at that steadily even as their WM
capacity changes with age, stimulant use, etc. From that it seems to
me that the assumption that WM contributes to IQ is flawed and I've
thought this way for the past two years
However, I'm more excited about WM improvement than I am IQ
improvement because most of the effects I want in my thinking come
from WM gains -- So for people who're trying to boost IQ, maybe they
won't use dual n-back anymore, but I think that WM and executive
function are extremely important for your overall intellectual
capacity, independent of IQ
So if WM training improves WM, then it's still a great advancement :)
Thanks for posting the study OP!
- Arbo
> --
> 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.
>
>
Seth
working memory capacity and focus. I mean, lets pretend you have no
working memory, do u think you'll do well on an IQ test? Therefore,
working memory capacity is important for intelligence. Now maybe if
you already have a good working memory, and you increase it, and you
don't see an IQ gain its because you don't know how to effectively
utilize your new capacity because you don't need it for everyday life?
While if you had a deficit, you would definitely use your new capacity
in everyday life, and thus demonstrate an IQ gain.
I am also confused why people throw in the towel after one negative
opinion. Its like they want to be confirmed wrong and as a result have
an excuse not to try.
Anyways, in the end, IQ is irrelevant. What is relevant is increased
performance. Brain n-back increase performance across a wide range of
activities. Thats enough for me ....
Gwern Branwen
> It's worth mentioning also that 0.98 was the largest effect size seen for
> any of the treatments and transfer tasks in Jaeggi 2010 (DNB to Raven's).
> Looking at DNB to BOMAT, the effect size was only 0.49; for SNB to Raven's
> and BOMAT, it was 0.65 and 0.70, respectively. Also, the no-contact group
> had an effect size of 0.09 (Raven's) or 0.26 (BOMAT), which really should be
> subtracted from the treatment groups' effect sizes, since we don't care if
> people get better, but only if they get better because of the training. For
> comparison, Jaeggi 2008 found an effect size of 0.65 for the treatment group
> vs. 0.25 for the no-contact group. In any case, 0.98 is definitely *not* a
> conservative effect size estimate, which is what you should be using when
> performing power calculations.
That's a good point. I didn't realize there was such an effect size
difference between Jaeggi 2010 and 2008 or that the choice made such a
difference. I'll add that to the section as it definitely limits what
Chooi 2011 shows. I'll remember to pay closer attention to effect size
& power in the upcoming studies.
On Tue, Feb 21, 2012 at 5:25 AM, polar <pol...@gmail.com> wrote:
> Gwern do you really think that all the studies which regularly are
> finding iq improvements after cognitive training are flawed or fraud?
The burden of proof is *heavily* on those claiming an IQ intervention
in healthy young adults, or children for that matter too. (Chooi
includes some of the standard examples in a small history of such
failures, like Headstart, but I guess you didn't read it...)
> Of course moody was partially right -
I guess I should be happy with what I can get.
> but do you really think that speed is NOT a part of intelligence?
It may be 'part of intelligence' - like vocab is. I believe I have
made this point a few dozen times now. Correlation, causation...
> Yes, for some people dual n-back doesnt work AT ALL. For others through, it definitely works.
Just like faith healing definitely cures cancer for some people amirite
> Maybe there is some interesting trait or factor thats intervening, whaddya say?
Epicycles, special pleading, etc.
--
gwern
http://www.gwern.net
Shell
as J's study? 20 sessions of 24 trials?
XFMQ902SF
jotaro
seth said enough, he basically said all what was needed.
James F
> I feel like emailing this study to Jaeggi et al but maybe that would
> be too rude or something.
Glenn Henshaw
Jonathan Toomim
On 2/21/2012 3:06 PM, Glenn Henshaw wrote:
> "I found this guys..."
Girl's.
Mr. Menesus
Would it have any effect on change in IQ?
--
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
> Has anyone considered that the effect of motivation on n-back training? What
> if people were not very motivated to work hard on the training and just went
> through the motions?
>
> Would it have any effect on change in IQ?
Then presumably they won't improve very much on n-back, and the
statistics won't be much affected by their simultaneous lack of
improvement on the IQ tests?
--
gwern
http://www.gwern.net
whoisbambam
ppl.
even if dnb does not improve iq, and i dont think it does, i am pretty
sure it has benefits.
it seems to have increased my willpower
it seems to have increased my attention deficit (as in i can get more
done--reduced the deficit, increased the focus factor)
it seems to have increased my ability to study longer, which is
probably related to the previous two
for me, that is enough.
i cant imagine life without having found dnb now.
the effect was that huge for me.
of course, i may be some sort of minority group as i believe i had a
'problem' as this didnt seem to be a problem for me in my 20s in
college. i am nearly restored after dnb and the many other supplements
i tried.
hard to say for sure what did it, but every time i restart dnb
training, my focus again increases.
Arbo Arba
performance but that WM doesn't improve how we score on an IQ test --
We know this because WM is variable over a lifetime and even
day-to-day but IQ remains steady
So I agree with you that there's more to our thinking skill than our
IQ, but I also think that WM can't be a large contributor to our IQ
scores
Do you believe that our WM is constantly changing?
Do you believe that our performance on IQ tests is constantly changing
to match the change in WM?
(these are questions for a lot of people who think that improving WM
will lead to IQ gains -- I'm a huge fan of WM improvement and have
argued so on this newsgroup, so don't think I'm raising critiques of
WM training)
jotaro
polar
cognitive journals are publishing studies based on placebo or faith
effect (polarity to your rigorousness perhaps? no offense, just
exploring). Approximately 50% of WM training studies (based on this
metaanalysis: http://www.springerlink.com/content/a29v30l61626184t/ ,
p. 51+52) found significant iq improvements. Thats what matters for me
- I cant be sure, I dont know, I'm exploring. I dont believe you
finished already.
On 21 ún, 19:34, Gwern Branwen <gwe...@gmail.com> wrote:
Gwern Branwen
> Gwern it also needs hell of a superstition to think that world class
> cognitive journals are publishing studies based on placebo or faith
> effect (polarity to your rigorousness perhaps? no offense, just
> exploring).
On the contrary, I do not believe that is superstition in the least
bit. I commend to your attention the numerous citations in
http://www.gwern.net/DNB%20FAQ#fn52 and point out that we discussed
previously the lessons of Bem's time-traveling pornography for the
methods & rigorousness of 'world class cognitive journals'.
--
gwern
http://www.gwern.net
XFMQ902SF
training could be driven by inadequate methodology, the way Jaeggi et
al now seems be. Only 50% of wm training studies showed transfer? Why
not 90-100%? The fact that not all studies could produce the same
results/replicate findings, and only half produced transfer makes me
more skeptical and unsure not less.
$.02
On Feb 22, 11:26 am, Gwern Branwen <gwe...@gmail.com> wrote:
> On Wed, Feb 22, 2012 at 9:59 AM, polar <pol...@gmail.com> wrote:
> > Gwern it also needs hell of a superstition to think that world class
> > cognitive journals are publishing studies based on placebo or faith
> > effect (polarity to your rigorousness perhaps? no offense, just
> > exploring).
>
> On the contrary, I do not believe that is superstition in the least
Jonathan Toomim
(http://en.wikipedia.org/wiki/Type_I_and_type_II_errors#Type_II_error)
and statistical power. The short version: if you performed Chooi's study
(say) 1000 times, and if DNB training actually improves IQ by 7.5
points, you would only get positive results in about 250 of those
studies. Larger studies can detect small effects, but small studies can
only detect large effects. If a small study fails to detect an effect,
that means almost nothing. If a large study fails to detect an effect,
that means something.
If there's no actual effect of DNB (and no publishing bias or systematic
error), one would expect for 5% of studies to find positive results. 50%
is a lot more than 5%. That means that there is either systematic bias
(the most common of which is publishing bias, where people who don't
find anything tend not to publish), there is a real effect, or something
relatively improbable happened.
(P.S.: I've tried a couple of different online power analysis
calculators, and I get different results from each one. So far they've
all given powers of less than 0.5 for an effect size of 0.5, but the
variance is still troubling. (I've gotten 0.25, 0.38, and 0.48 so far
for d=0.5, n=13.) So I'm uncertain about just how uncertain we should be
about Chooi's findings, but I'm pretty certain that we should be at
least pretty uncertain.)
Jonathan
>> methods& rigorousness of 'world class cognitive journals'.
>>
>> --
>> gwernhttp://www.gwern.net