"Putting brain training to the test" - Nature paper

[Jonathan Toomim]

http://www.nature.com/nature/journal/vnfv/ncurrent/pdf/nature09042.pdf

Abstract:
> ‘Brain training’, or the goal of improved cognitive function through
> the regular use of computerized tests, is a multimillion-pound
> industry, yet in our view scientific evidence to support its
> efficacy is lacking. Modest effects have been reported in some
> studies of older individuals and preschool children, and video-game
> players outperform non-players on some tests of visual attention5.
> However, the widely held belief that commercially available
> computerized brain-training programs improve general cognitive
> function in the wider population in our opinion lacks empirical
> support. The central question is not whether performance on
> cognitive tests can be improved by training, but rather, whether
> those benefits transfer to other untrained tasks or lead to any
> general improvement in the level of cognitive functioning. Here we
> report the results of a six-week online study in which 11,430
> participants trained several times each week on cognitive tasks
> designed to improve reasoning, memory, planning, visuospatial skills
> and attention. Although improvements were observed in every one of
> the cognitive tasks that were trained, no evidence was found for
> transfer effects to untrained tasks, even when those tasks were
> cognitively closely related.

--
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.

[Pontus Granström]

This is nothing new, but does not state in any way that the brain is immune to improvement,nor does it discuss dual-n-back or other similar studies. It also fails to explain the flynn effect.

[jttoto]

Actually, this is more important then some will likely give credit
for. Since the newest evidence shows that single n-back improves
intelligence just as well as dual, then we can imagine that training
in any WM task will cause results in intelligence

. This study shows the opposite. The game used in the study likely
functions similar to n-back, since the WM games escalate in difficulty
as the user improves. Yet, no improvement.

> For more options, visit this group athttp://groups.google.com/group/brain-training?hl=en.- Hide quoted text -
>
> - Show quoted text -

[jttoto]

And in fact,this is more compelling evidence that, while one can
improve in a small number of cognitive domains given training, one
won't see an improvement everywhere. It is very likely that Jaeggi's
measure for "general fluid intelligence" was in fact very training-
specific.

> > For more options, visit this group athttp://groups.google.com/group/brain-training?hl=en.-Hide quoted text -

[Pontus Granström]

WM-tasks although labeled as "equal"  are in fact not from a neuroactivity point of view. N-back is something that stresses the central executive while just "remembering stuff" does not.

[Pontus Granström]

How can IQ-tests then measure something that should be common to all mental acitivty? Then IQ-tests by themselves are specific if there's not overlap.

[Pontus Granström]

I think you use the word WM-task a bit sloppy. All "WM-tasks" are not equal from a executive function point of view. N-back activates roughly 3 times more areas than just "maintaining information". The reason dual-n-back increases IQ is because Working memory and attentional control is a big part of IQ-tests, this is widely accepted and almost certainly the reason behind the flynn effect. In fact only 0.25 of the variation is explained by "procedual knowledge" or what some of the dnb critics want to call intelligence. This is true, but still IQ-tests demands more than this.

On Wed, Apr 21, 2010 at 1:54 PM, jttoto <jtde...@uncc.edu> wrote:

[jttoto]

Let's keep it in focus. In the end, Jaeggi used a very specific
measure of visuospatial reasoning. Yes, it has a strong correlation
with other functions, but we still are not 100% sure why these
correlations exist. Now lets look at this evidence this study has in
comparison to Jaeggi.

The sample size was large, much larger then Jaeggi's.

The study used games that train and improve WM.

The study measured a broad range of cognitive functions, like one
would measure in g. Therefore, it is more comprehensive then Jaeggi's
study.

No result.

> > brain-trainin...@googlegroups.com<brain-training%2Bunsubscribe@go­oglegroups.com>
> > .

> > > > For more options, visit this group athttp://
> > groups.google.com/group/brain-training?hl=en.-Hide quoted text -
>

> > > > - Show quoted text -
>
> > > --
> > > 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<brain-training%2Bunsubscribe@go­oglegroups.com>
> > .

> > > For more options, visit this group athttp://

> > groups.google.com/group/brain-training?hl=en.- Hide quoted text -
>
> > > - Show quoted text -
>
> > --
> > 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<brain-training%2Bunsubscribe@go­oglegroups.com>
> > .

[Pontus Granström]

You can not put all WM-tasks into one I am afraid.

1. Working memory was often measured using a small, unsystematic set of tasks. The
resulting representation of the construct is therefore contaminated by task-specific
variance and does not necessarily reflect all relevant aspects of working memory.
2. In most studies on individual differences, working-memory capacity was operationalized
as an undifferentiated construct. This contrasts sharply with numerous experimental findings
that suggest a multicomponential view of working memory (e.g., Baddeley, 1986).
Different components of working memory, i.e., distinguishable cognitive resources, could
be expected to contribute to different extents to different intellectual abilities.
3. Similarly, studies relating working memory to intelligence constructs usually focused on
a single mental ability, such as reasoning or reading ability, as the criterion. This does
not provide a clear picture of how working memory relates to the structure of
intelligence, in other words, which abilities depend to what degree on working memory.
4. The tasks used to measure working memory were very similar to and sometimes
indistinguishable from common reasoning tasks. This problem is apparent, e.g., in the
work of Kyllonen and Christal (1990), where the same type of task served as a workingmemory
task in one study and as a reasoning task in another. If tasks used to measure
working-memory capacity have many common features with those used to measure
reasoning, it is not clear which features are responsible for the high correlation obtained
between them.

[Pontus Granström]

The present study yielded three main results. (1) Working-memory capacity is highly
related to intelligence. The strongest relationship found was to reasoning ability, thereby
replicating results found by Kyllonen (1994a) and Kyllonen and Christal (1990). The
working-memory tasks in our test pool were selected so that overlap in cognitive
processes and strategies with the reasoning tasks was minimal.

[jttoto]

Present the evidence then. As far as my knowledge, there is no direct
study comparing the results of a standard WM training compared to n-
back or dual- n-back. A direct comparison must be made to draw such
conclusions, within the same parameters.

> > brain-trainin...@googlegroups.com<brain-training%2Bunsubscribe@go­oglegroups.com>
> > .

> > > For more options, visit this group athttp://
> > groups.google.com/group/brain-training?hl=en.- Hide quoted text -
>
> > > - Show quoted text -
>
> > --
> > 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<brain-training%2Bunsubscribe@go­oglegroups.com>
> > .

[Pontus Granström]

I rest my case........

[jttoto]

How does this wall of text disprove anything I said. It only
theorizes that WM may be multicomponental and is more complex than we
believe. #3 in facts supports what I've been saying. It says nothing
on comparing training exercises to each other. So again, show me a
study directly comparing n-back to other WM training games, where one
is the active control.

[Pontus Granström]

Well you said "The study used games that train and improve WM." While WM-tasks are as you point out complex and as stated in the article it is not necessarily those that are important to intelligence that are examined/trained. N-back however is one of those WMC processes that is.

[jttoto]

Ahh, I see, the two quotes are part of the same study. My mistake.

Still, that doesn't explain why a task that trains WM did not transfer
to most of the cognitive domains. We are talking about WM-training,
not what WM is. I am confused as to what you trying to prove with
that quote.

> > For more options, visit this group athttp://groups.google.com/group/brain-training?hl=en.-Hide quoted text -

[Pontus Granström]

All WM-training is not equal. You cannot arbitrary call things WM.

[jttoto]

Yes, but how does the game in the nature study break the standard
definition of WM? Is it comparable to say, the CogMed used on ADHD?
This is something I with the authors were more specific about, but it
seems reasonable, given the huge number of participants and obvious
knowledge of cognition, that they properly researched WM and chose a
game that would measure and improve it.

> ...
>
> read more »- Hide quoted text -

[Pontus Granström]

No it just shows that those WM-tasks did not lead to any improvements.

[Pontus Granström]

Because they do not tax the brain in the same way dnb does, and are not important underlying functions for solving Gf-problems.

[jttoto]

The researchers specifically said that the games used were standard
measures of WM, modified to get more difficult as the user improves.

Then please, explain how the game used is so different to n-back. It
requires the standard memorizing and updating information, does it
not?

[jttoto]

In other words, the games are simply standard measures of WM used
extensively in scientific literature. Are standard measures of WM not
taxing enough on the brain?

[Pontus Granström]

Once again WM is not a SINGLE thing. They used tasks that required something we might label as WM but not something that is equal to dnb from a neurological point of view, if it were they would have gotten the same results naturally.

[Pontus Granström]

Remembering information uses only 1/3 of the brain areas compared to dnb. The brain activity and oxygen supply to the brain is proportional to n-level.

[jttoto]

At one time, n-back was just another measure of WM like these games
were. And again, the games were based of scientifically recognized
measures of WM used extensively in scientific journals. Are you
telling me that you have indisputable proof that training in n-back
goes above and beyond other measures of WM. You don't, because no
study exists directly comparing two training methods.

"if it were they would have gotten the same
results naturally. "

Under the assumption that the Jaeggi study is irrefutable proof that
WM can be improved. You must be able to replicate results
consistently before you can say something has strong empirical
support. In fact, wasn't there a study that didn't replicate Jaeggi's
findings using the exact method? Not very encouraging.

[Pontus Granström]

I do not know which tasks that were used. Just because they are standard WM-tasks does not imply that they are the ones that indeed Gf-tests require. Since 0.6-0.8 of thev variation is explained by variation in WMC.

[jttoto]

I have to leave to work soon but I'm going to end with this. When
presenting with conflicting evidence one must sit back and objectively
look at the results with scrutiny. Quite frankly you are being very
close-minded about the null results. I personally believe that
intelligence can be improved, but one must look at the zero effects
closely to properly reduce what is behind the improvements. One must
also face the fact that maybe improvements could be partially due to
researcher bias. So again, since you never properly rebutted it:

- Dual n-back, and n-back in general, does not have strong empirical
support. Even some studies using the same experimental methods as
Jaeggi have not replicated the same results. The fact that
participants are young is void, since in theory children can improve
as well based on other studies.

- You still present no hard evidence that one method of training is
superior to another. Saying that one is inferior because it doesn't
produce results is a weak argument. One must directly compare within
the same study.



-

[Pontus Granström]

Besides they trained 6 tasks during 10 minutes three days a week. Meaning they spent in total 1.6 min on each task a day. No wonder they didnt improve. The WMC training were in total 5 minutes a week and 15 minutes for the whole period corrsponding to not even ONE training session with dnb. No wonder they didnt see any results....

[Pontus Granström]

I am sorry one group spent 2.5 minutes on short term memory tasks. I guess this study is joke.

[jttoto]

The study addressed that issue. Some spent more time. There was no
correlation with time spent and improvement. And again, you did not
address the points I brought up. Do I need to copy and paste?

On Apr 21, 10:41 am, Pontus Granström <lepon...@gmail.com> wrote:
> I am sorry one group spent 2.5 minutes on short term memory tasks. I guess
> this study is joke.
>

> On Wed, Apr 21, 2010 at 3:46 PM, Pontus Granström <lepon...@gmail.com>wrote:
>
>
>
> > Besides they trained 6 tasks during 10 minutes three days a week. Meaning
> > they spent in total 1.6 min on each task a day. No wonder they didnt
> > improve. The WMC training were in total 5 minutes a week and 15 minutes for
> > the whole period corrsponding to not even ONE training session with dnb. No
> > wonder they didnt see any results....
>

[jttoto]

Let me make this clear:

There is no hard evidence that dual n-back, a training method whose
results have not been consistently replicated, is a superior training
method to the games described in this study. There is only
speculation.

[Pontus Granström]

Well show me a evidence that a group spent 200 minutes on a Short Term Memory task. That all STM tasks are equal is a thing that is only speculation. N-back has a very special place due to it's load on the executive functions. The study is not equal to the Jaeggi study in that sense. N-back activates areas linked to IQ as seen by fmri scans, it increases oxygen to the brain (known to increase IQ), N-back requires extensive use of the updating executive function which has a 0.6 correlate with IQ and all other measures of intelligence. I would like to know what STM task that were used before that we can't say anything.

1. First no evidence that anyone spent 200 minutes on a STM-task.
2. The STM-task is probably inferior to dnb. (not just speculation)

[polar]

interesting study, I dont have time to go through it thouroughly now,
but I will. I just get the feeling they did not use n-back. And if
they didnt, that can a huge difference. Not because I'm a fan of it
(I'm even more fan of "truth"), but because n-back is subjectively
waaaaaaay more demanding that anything else in cognitive area. IMHO
you cant compare backward digit span training with n-back. And,
anything that transfers to raven (Kligberg, Jaeggi 2008, 2010) is
pretty much FAR transfer.

On 21. Apr, 10:01 h., Jonathan Toomim <jtoo...@jtoomim.org> wrote:
> http://www.nature.com/nature/journal/vnfv/ncurrent/pdf/nature09042.pdf
>
> Abstract:
>
>
>
>
>
> > ‘Brain training’, or the goal of improved cognitive function through  
> > the regular use of computerized tests, is a multimillion-pound  
> > industry, yet in our view scientific evidence to support its  
> > efficacy is lacking. Modest effects have been reported in some  
> > studies of older individuals and preschool children, and video-game  
> > players outperform non-players on some tests of visual attention5.  
> > However, the widely held belief that commercially available  
> > computerized brain-training programs improve general cognitive  
> > function in the wider population in our opinion lacks empirical  
> > support. The central question is not whether performance on  
> > cognitive tests can be improved by training, but rather, whether  
> > those benefits transfer to other untrained tasks or lead to any  
> > general improvement in the level of cognitive functioning. Here we  
> > report the results of a six-week online study in which 11,430  
> > participants trained several times each week on cognitive tasks  

> > designed to improve reasoning, memory, planning, visuospatial skills  

> > and attention. Although improvements were observed in every one of  
> > the cognitive tasks that were trained, no evidence was found for  
> > transfer effects to untrained tasks, even when those tasks were  
> > cognitively closely related.
>

> --
> 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 athttp://groups.google.com/group/brain-training?hl=en.

[Michael Campbell]

Pontus Granström wrote:
> 2. The STM-task is probably inferior to dnb. (not just speculation)

"probably" is pretty close to speculation.

[Pontus Granström]

Well true, but I do not think it's a coincident that they use n-back for so much neurological research when studying brain activity and so on.

[Pontus Granström]

Well I have found definite proof now, take a look in the files section "A comparison of laboratory and clinical working memory tests and their prediction of fluid intelligence". This proofs that n-back is superior to other WM-tasks when it comes to Gf. I rest my case.

Consistent
with previous research (Friedman et al., 2006; Gray et al.,
2003), n-back performance was also significantly correlated
to measures of fluid intelligence (rs range from .37–.40), and
tended to correlate more strongly across the gF measures
relative to the other WM tests used in the study. Overall, the
recall version of the n-back task used in the present study
proved to be a valid measure of WM function.