Open Question: Training Methods And Transfer To Fluid Intelligence



When my infant son woke me at 4:30am this morning, what question crept into my head? Can we figure out the optimal training approach, or optimal training approaches, for generating transfer to fluid intelligence; what do we know, what can we deduce, and how can we measure our theories?

I don’t want to assume just one answer. The answer may differ from person to person or from day-to-day. We’ve discussed some of the differences between taking breaks and training straight through, between canceling a block and not, between meditating between blocks and not, and between training frequently or less frequently. We’ve also discussed how we might consistently measure working-memory and transfer to fluid intelligence.

I thought it might be helpful to state some of the principles at work, as I understand them, and begin an open discussion of this topic. If we develop some theories, we may be able to test them out. (This might ultimately involve an on-line study along the lines that Shaun framed out, but with the added dimension that we could compare different training strategies.)

Or we might simply provide food for thought so that people can try out different approaches and find what works best for them.

The principles:

1. Working-memory gains transfer to fluid intelligence gains. The Jaeggi, Buschkuehl research hypothesized, tested, and demonstrated the validity of this principle; we can use it to help us think through training approaches and measurement strategies. (We might accept that measuring increases in working-memory is as relevant as measuring increases in fluid intelligence…)

2. Progressive training of working-memory increases working-memory focus and span. The dual n-back training method exemplifies this principle, requiring us to use visual and aural working-memory simultaneously.

So, in considering different ways to approach the dual n-back training sessions, we must try to determine, if we can, which of the strategies will most effectively lead to working-memory increases.

3. Maximum short-term working-memory capacity does not necessarily reflect effective working-memory capacity. By this I mean that we can increase our short term performance on the n-back task and on a working-memory span test by meditation, exercise, etc. but that this doesn’t necessarily reflect an increase in our day-to-day or long term working-memory capacity. (I increased my n-back average to 7.0 when I was using block cancellation, breaks, meditation, etc., but when I train straight through without extensive use of these techniques, my n-back score hasn’t been higher than 6.5.)

We must be careful to consider this in any measurement methodology.

4. Neurogenesis and plastic change require focused attention on a task, as well as a sense of satisfaction, pleasure, or achievement from the task. (Through the release of acetylcholine and dopamine respectively.)

These are the core principles as I understand them.

We also have two other findings that we should consider:

5. Sleep consolidates gains. Open question – could some form of meditation or deep relaxation substitute for sleep and might it be used to help boost long term working-memory?

6. Exercise vs. no exercise boosts short-term performance.

Discussion:

So, an effective dual n-back training approach must be rewarding, focused, and demanding.

Some tricky questions:

- Strategies that help boost short-term n-back scores (training after exercise, taking breaks, canceling blocks, etc.) can lead to increased satisfaction by boosting n-back score. But this increase in satisfaction may come at the expense of demanding more of our prolonged focus and attention.

- The straight-through strategy can lead to FLOW and the satisfaction of training more frequently. Does it demand less of us in the work of stretching working-memory span? (Example: Since I’ve been going straight through I’ve been spending more time at n=5 and 6 and less at n=7 and 8.)

Alright, I’ve probably said enough. Others please chime in…

Martin

PS. I’m going pin this post to the blog home page so that it’s easy to find.

Tags: , , , , , , ,

16 Responses to “Open Question: Training Methods And Transfer To Fluid Intelligence”

  1. Tim says:

    You emphasise the point the for plastic change, you need a sense of satisfaction, pleasure or achievement from the task. The “brain training” games you get for Nintendo and other consoles often have little sequences when you have done well, for example playing a fanfare, animating smiling people congratulating you, and awarding you a “medal”. I had always thought of that aspect of these games was just a gimmick, but thinking harder, perhaps this kind of reinforcement will be more effective at giving you satisfaction than just printing up a score on a screen. Maybe these sequences actually improve learning. Perhaps another one for future research!

  2. Shaun Luttin says:

    Martin Walker writes:

    3. Maximum short-term working-memory capacity does not necessarily reflect effective working-memory capacity. By this I mean that we can increase our short term performance on the n-back task and on a working-memory span test by meditation, exercise, etc. but that this doesn’t necessarily reflect an increase in our day-to-day or long term working-memory capacity. (I increased my n-back average to 7.0 when I was using block cancellation, breaks, meditation, etc., but when I train straight through without extensive use of these techniques, my n-back score hasn’t been higher than 6.5.)

    We must be careful to consider this in any measurement methodology.

    >>>>>

    Shaun Luttin replies:

    The statement that you made, that is quoted above, means that what we do before and during working memory or intelligence tests has at least a small impact on performance. For instance, doing exercise and meditation before testing working memory would likely lead to a better result than not exercising or meditating before testing working memory.

    To test the results of n-back training, it’s important to do pre-test and post-test measures of either intelligence or working memory. To make these tests valid, it will be important to make the conditions of the pre- and post-tests similar. That would mean controlling for those variables that, in addition to n-back training, also create variance in working memory and / or intelligence scores.

    Other variables do seem to impact on working memory score. These include might include amount of sleep before testing, exercise before testing, meditation before testing, good nutrition before testing, mood during testing, and other factors. Controlling for these other variables will be important for isolating how much variance in working memory scores, or intelligence scores, is causes by n-back training. These is a name in psychological research for variables, like exercise or sleep in this case, that get in the way. They are called something like confounding variables, or nuisance variables, or something along those lines.

    Controlling for the nuisance or confounding variables will be important when trying to determine the amount of variance in working memory or intelligence that is predicted by BFP Training.

    Warm regards,
    Shaun.

  3. Shaun Luttin says:

    Martin Walker writes:

    So, an effective dual n-back training approach must be rewarding, focused, and demanding.

    Shaun Luttin replies:

    Rewarding: The need for reward seems to follow from the need for neurogenesis to be rewarding, be pleasurable, or lead to a sense of achievement.

    Focused: The need for focus seems to follow from the vital importance of paying attention.

    Demanding: The need for demanding training seems to follow from the finding that progressive training for working memory leads to improvement.

    I offer another aspect that might be a requirement:

    Repeated Inputs: The need for repeated training, and repeated training of a repetitive type, seems to be required. For instance, the n-back training has a high degree of repetition. Repeated inputs may be classed as a core aspect of the training.

    As a result, the core principles could be expanded to: An effective dual n-back training approach must be repeated, rewarding, focused, and demanding.

    Warm regards,
    Shaun.

  4. martin says:

    Yes, good catch. Repeated training is an important core principle.

  5. Shaun Luttin says:

    Martin writes:

    Maximum short-term working-memory capacity does not necessarily reflect effective working-memory capacity.

    Shaun replies:

    While at the University of British Columbia, I took a course called Motor Learning and Control. Part of the course explained the measurement of learning. Learning can be measured by observing a person’s behavior. But here is where the measurement of learning becomes tricky.

    There are three types of important observations to make when measuring learning. These are the observation of performance, of retention, and of transfer. It turns out that observing retention and especially transfer are the best indicators of learning.

    Lets look at n-back training, for example. The BPF Training involves 19 30 minute sessions over a period of four to five weeks. Using the vocabulary mentioned above, here is how the training could be conceptualized:

    a) During the 19 days of training working memory, we could observe the persons PERFORMANCE on the n-back training.

    b) After the 19 days of training, we could do a RETENTION test. The retention test could occur one week (or so) after the last of the 19 days of training. Importantly, the retention test would simply be a 30 minute session of BFP Training.

    c) In addition to testing RETENTION, we could also test TRANSFER. This transfer test could also occur one week (or so) after the last of the 19 days of training. Importantly, the transfer test would be something different than BFP Training. For instance, the transfer test might involve a span test, an IQ test, or perhaps some other tasks the involves fluid intelligence.

    So those are the three observations that someone might look at when they are trying to determine how much someone has learned: performance, retention, and transfer.

    The most important point, that I learned from my class, is that performance does not indicate learning. That is to say that if a person performs very well on the BFP Training during the 19 days of training, that does not indicate learning. For instance, if my final n-back average at the end of the 19 days were 7.6, that alone would be a poor indicator of what I learned. Indeed, motor control and learning research shows that poor performance scores can sometimes correlate with very high levels of learning.

    That leads me to the next point. Retention does indicate learning. Doing a retention test, especially after a period of absence, shows what the person has retained from the training. Further, transfer tests represent an even better indicator of learning.

    The main conclusion here is the how a person performs on the BFP Training is a poor indicator of how much he or she is learning, growing, or changing. Actually, what I have written serves to bolster what Martin wrote above: “Maximum short-term working-memory capacity does not necessarily reflect effective working-memory capacity.” In other words, maximum performance working-memory scores do not necessarily reflect growth or change in working-memory capacity. To show learning, growth, or change in working memory, a retention or transfer test do the trick.

    Warm regards,
    Shaun.

  6. Will W says:

    I’m going to respond to this post not in the general audience way but more to people here who are at Mensa level or higher….which means, I guess, questioning what fluid intelligence is, and how it impacts the higher order stuff of thought…bear in mind this post does go a little off track from the specific question at hand but relates more to the problem of fluid IQ and the person entering training with a high IQ (which seems to be a lot of the people who post here – Martin, Shaun, Erin, sundry Mensans).

    So, relevant to this group.

    Perhaps I’m going off topic, but it’s worth noting because it gets to why we are all here, and we’re here for different reasons. I think, given that a lot of the people here are of high IQ, it’s important to ask the question of how working memory, processing speed, and fluid intelligence informs higher level problem solving, requiring extremely abstract intelligence. It seems
    that WM and processing speed might serve as a bottleneck for abstract thinking, parallel processing, and so forth, but it seems only up to a point. At higher levels some other processes takes over, a process which is not understood very well and which probably has little to with WM, processing speed, or fluid intelligence (and yet, everything to do with those things as a necessary condition). Note: leading researchers in IQ do not go into the
    differences of IQ above 125, since its big time point of diminishing returns as special abilities, creativity, persistence, personality, et al, take over.

    However, I’ll offer some very amateur guesswork on IQ differences among the very bright based more on wild insight than any solid data… but at this point, the data of the high IQ set is not very solid anyway…so

    At its lowest level, fluid intelligence is speed of information processing — reaction time. However, abstract reasoning ability exists within the higher form of “fluid intellence” — tapped by tests like the raven’s, similarities on the Wechsler — along with wm and visual processing. And I tend to think that what happens at the highest level of problem solving is that ability to handle abstractions is what matters most and will separate a stratosphere IQ from one that is just extremely high. That is, wm and processing speed
    lose importance and simply become necessary but insufficient conditions to
    solve problems of a very abstract nature…some other process takes over which we might just call ability for extraodinary abstraction/logic, usually found with those in 160+ IQ range and becomes the sine qua non of those in the 180+ (I don’t mean Hollingworth’s ratio IQ kids — I mean deviation 180+ adults — C. Langan, M. Vos Savant, etc).

    I realize personal anecdotes are not to be taken as stastically meaningful but I’ll use myself as an example. In 2004 I took the Mega test and scored 145 (sd 16); a test of mainly chrystallized abiliity, the Mega is an untimed test for gifted adults. Since then, I have very much sharpened my WM & processing, so much so that I feel like a different person. In the last few months, I have been playing the ECT’s over at mybraintrainer and have reached brainmaster +8 levels and ultra +8, which is a very high ranking — and if ECT/IQ rankings were correlated all the way to the top of the IQ scale, then I’d have a IQ level somewhere in the mid 150′s. Yet, I can tell you, I do not have an IQ ANYwhere in the 150s….and by that I mean on a test with sufficient ceiling…if I took the Mega test today, I am pretty sure I’d hit very near to where I scored back in 2004. In other words, my IQ is
    better, sharper, faster, svelter but I do not think any “higher” as far as the
    ability to solve extremely hard problems go.

    The most important aspect of training, imo, defies jumping up to a new level on the psychometric ladder and has more to do with maximizing functioning on a daily basis. The limits set on very high IQ (Mensa level or higher) are imo best set by tests like the Mega Test — where someone like a Feynman or Einstein would crush the average 140ish IQ Mensan…even if the standardized WAIS scores or other standardized scores were roughly the same (Feynman scored 124 IQ in high school, probably the Stanford Binet. However, his true IQ was clearly much higher; purportedly Jensen himself tested Feynman in the high 150′s, using, I believe, the CMT).

    I suppose the point I’m going for is, while I might outscore someone on a “lower order” brainspeed or WM task, there are those who have much higher IQ’s than I do, no matter how fast my brain moves compared to theirs. Likewise I might then outscore someone on higher order IQ test who has faster reaction time than I do, i.e. a racecar driver. (However, my simple RT is very fast, about 170 ms as tested by cogfun. The IQ/RT correlation hovers at about .2, so, while this is me bragging — I’m not bragging about my RT as correlating to an IQ).

    Finally, in my case expanded WM helps “manage” my creativity…this is very important that I do this, since divergent thinking can get easily lost in the neural shuffle…nothing to do with IQ, ipso facto, but very valuable in a an “out of the box” thinking type of way, not easily tracked by tests, etc.

    I think the dual n back is clearly a great thing to do, especially if one is a high IQ person that’s low functiong (as I was!) before I started training
    …but I think the gains made are probably getting one up to where biology wants than going beyond…beyond that, there’s room to improve WM and fluid IQ, but I don’t know if getting to an 8, 9, 10 back level is going to improve the very bright person to the very highest levels of abstraction and logic that seem to involve some other process other than WM and that seems to separate a 160 from a 150 and 170 from a 160 and so on. All this is of course compounded by the highly speculative assessments of IQ in this range…

    Granted, I’m making only a semi-argument here, more just offering up observations (that can be refuted)…ultimately, I would say, though, that the dual n back is the best — most grueling — brain training one can find and definitely worth doing, apart from IQ gains.

  7. martin says:

    Interesting thoughts, Will.

    I particularly like the way you separate out the concept of a training gain or transfer that leads to a “better, sharper, faster, svelter” thinking mode, or IQ, from that of a training gain that might improve one’s “ability to solve extremely hard problems.”

    And, as you point out, the concept of ongoing brain training to meet the first goal is by no means an unimportant or unworthy goal.

    Your comment prompts several thoughts on my part:

    - First of all, well done for posing a provocative perspective.
    - Secondly, I agree that there is more to the process of solving problems than the brute force of working-memory and processing speed.

    Here are the buts…

    - I’m convinced that the training leads to neurogenesis and plastic change. In which case, this plastic change in the brain’s ability to hold and process short term information for visual and aural stimuli might in fact have a bearing on one’s ability to solve really hard problems.

    This might be a matter of degree. What if someone had the creative intelligence to solve the really hard problems, but often came unstuck by overloading his or her working memory? And why do so many brilliant minds produce their best output when they’re young? Could it be that the normal deterioration of working memory over time is enough to blunt the edge of their intellectual power?

    These are maybes, I realize, but the idea of plastic change makes me cautious about writing off a change in a person’s ability to solve the really hard problems.

    - The second but has to do with another unanswerable: How do the brains of the Einsteins and the Feynmans develop their profound abilities to tackle the extremely hard problems? Through innate structure and through haphazard environmental influences. What’s to say that an adult brain, through exposure to certain influences, can’t be reorganized so as to promote a significantly higher level of functioning?

    I freely admit that I am an optimist and that this colors my own perspective!

    I guess the only way to answer this would be to persuade you to take the Mega test again… ; )

    Martin

  8. Shaun Luttin says:

    Hi there Will and Martin:

    While I don’t have any comments at this point, I am chiming in to let you two know that I have read the posts. After reading them, I feel enthusiastic about the training, and excited to post some of my own thoughts and feelings.

    Interestingly, I scored >99 percentile on the similarities part of the WAIS-III, but I scored in the 92 percentile in the working memory part of the WAIS-III. That might imply that working memory was indeed limiting my ability to use my potential.

    Interestingly, I have had major depressive disorder, which is now well managed in my life. But one of the last aspects of the psych to recover after depression is cognitive functioning. In particular, memory and concentration take a very long time to recover after a profound depression. If any of you are interested, I can post research article links that speak to this.

    Warm regards,
    Shaun.

  9. Will W says:

    “This might be a matter of degree. What if someone had the creative intelligence to solve the really hard problems, but often came unstuck by overloading his or her working memory? And why do so many brilliant minds produce their best output when they’re young? Could it be that the normal deterioration of working memory over time is enough to blunt the edge of their intellectual power?”

    Excellent points, Martin, and it would certainly make sense.

    In the case of Einstein, he certainly did his major work when young. And he reportedly was not very great at math, but was far more visual — this goes against Gardner’s typing him as a ‘logical-mathematical’ type. So it could be, though, that Einstein’s thought experiments required a great deal of visual memory in addition to pure visualization. Eidetic memory tends to fade with age — e.g. the novelist Vladimir Nabokov had perfect visual recall for numbers as a child but lost the gift later in life…(but obviously retained general recall ability as his memoir Speak Memory, attests to).

    I am very curious to how G affects higher level abilities…why some people can do certain things (such as high level math) while others cannot, given roughly the same level of G. The idea behind the Mega and tests such as that is that G might be extended to a higher order processes beyond wm and processing speed…yet beyond that there is really only chrystallized IQ (learning) and special talents; so perhaps no other aspect of general intelligence takes over space in the high range. The thing is no one really knows what goes on….but in the end, I agree with you. There can be no Genius in the machine; higher order process can only be informed by the lower ones…which is probably what got Galton so obsessed with lower level cognitive processes…however, what the layman associates with high IQ is being able to juggle complexity and engaging in deep abstractions, which seems to defy the notion that such complexity could be built on so simple of constructs as WM and processing speed…but may very well be so.

  10. Gman says:

    Interesting discussion thread- not entirely sure what would be the optimum training regime- but instinct tells me that it will vary from individual to individual. I have been through the training as suggested- 1 session almost every day for 19 days and achieved an n-back of 3.95. I wouldn’t say this is an impressive score given the reports of 5, 6 and 7 from some people. Maybe the score is about right for me…i would describe myself as ‘smart enough’. Anyway…now that i have wrestled with the software (it was a fight to begin with) i am ready to go for an intense approach. My plan is to go through 2 complete sessions in one month or less…approximately 2 sessions a day monday to friday for 4 weeks. I feel that i am the type of person who is all or nothing…this approach should work for me e.g. i have been known to go from completely unfit to superman fit within one month…whenever the mood takes me. I have reasoned that i might boost mental fitness with a similar approach…wish me luck…will post the results at the end…if i am still alive!

  11. martin says:

    Hello Gman.

    I like your plan! Please do let us know what happens with your intensive training.

    Martin

  12. Aitor says:

    hi all,first of all Im sorry because my english is not very good ,Im an spanish guy who is interested in dual n back.I think it is very interesting your discussion,it makes me wonder why some persons who have a higher IQ than Einstein can not manage what he did.Perhaps this people might have many hobbies …chess..programming…hacking….dual n back…..but Einstein was obssesed with physics and was thinking about that the whole day.His mind,his brain specialized just in that.However,if he had to play chess against me I could easily beat him cause I play chess since I was a child.If Kashparov(IQ 190)have to solve a problem about physics he could not do it better than Einstein because his knowlege about this subject is lower.However,if kasparov had to study the physics degree at the university he could do it “easily” cause his high IQ.Kashparov’s processing speed is higher but he cant solve problems about physics because his cristallized intelligence in physics is low.The same thing would happen with Einstein and chess.I think you can play dual n back the whole day but if you dont have a good cristallized intelligence about the subject or problem that you want to solve you won’t get it.In my opinion the intelligence is the capacity to learn quickly.

  13. martin says:

    Hello, Aitor.

    Many thanks for your contribution. I think you make excellent points. Our life experience has a tremendous impact on the paths our intelligence follows and the skills and knowledge we develop. Some of this is truly crystallized knowledge (e.g., the rules of chess or physics) and some of this is pseudo-crystallized or semi-crystallized, if you will, in that repeated experience with similar sets of data or patterns of thinking leads us to be better able to manipulate those problem sets. (As Kasparov saw thousands of chess openings and how they developed, his brain responded by becoming able to more effectively process all of the variables required to interpret the impact of a new opening.)

    I would raise two points that extend this beyond a discussion of IQ and ability:

    1. Special skills and creativity don’t necessarily correlate to IQ. One can have an extremely high IQ and not much imagination. Einstein became so lauded as a scientist in large part because he was able to imagine solutions that others could not. (Curved space time, for instance.) Einstein himself is quoted as attributing his success to imagination.

    2. We can be drawn to certain activities or paths in life because we have an aptitude or interest in them. Einstein was drawn to physics, not driven to it. (I don’t know about Kasparov.)

    The great thing about dual n-back training seems to be that it benefits us in whatever intellectual or cognitive-heavy pursuit we give our energies to. It works on strengthening core cognitive functions that have a general utility in every conscious mental activity – and perhaps some unconscious ones too.

    Martin

  14. Charles says:

    If I said in’sight’ was the key to the highest order thinking. You
    would understand where I am going. Visualization is about extracting the
    similarities — a table looked at from different directions looks
    different (excluding contrived environments). Auditory processing is
    discreet and linear ‘g’ is not like ‘d’, try going to a church
    and use the similarities and you will offend. We define
    intelligence as the ability to understand the new; this is easier if you can
    ‘see’ how what is before you is similar to what you have seen
    before.
    From what I can gather brain functions build on lower level
    abilities.
    In the brain itself, neurons devoted to visual processing number
    in the hundreds of millions and take up about 30 percent of the
    cortex, as compared with 8 percent for touch and just 3 percent
    for hearing – Discovery magazine.
    Seeing the vast difference it is easy to understand how valuable
    hijacking some of this to solve problems has been.
    I’ve read that some pioneers of fields feel the would not have
    made it though the rote teaching of the subject they created.
    I think the reason that people only make great discoveries when
    they are young is that there tends to be a process we go through
    when we first approach a subject, that in some ways mirrors or
    lives. Those that are curious just play trying things and seeing
    what happens, gradually will buildup a model of what is
    happening, then use the model refining it as they go. Those with
    a goal tend to adopt what they see working. I would contend that
    the curiosity needed to make radical breakthroughs tends to occur
    at young ages the persistence and goal setting needed to make use
    of insight tends to come later. Once that goal nature has
    developed it becomes more difficult for most to aimlessly pursue
    there curiosity. I have heard it said that people have made other
    breakthroughs in different fields in older age but not in the
    same the reason ventured was that they had such a good ingrained
    model that they lost the ability to look at the field with fresh
    eyes and see the weakness in their view.
    The fact that Dual N Back increases auditory and visual memory is
    of interest to me as mine have been in the bottom 40% (and started
    much lower) for my age and have defiantly hindered me.
    Charles

  15. Neil says:

    In terms of overall brain training though has anyone ever utilised any other mode of potential cognitive enhancement? I happened to stumble across a website called The Complete Guide to Genius and it claims to be able to increase an individuals IQ up to 180. I was wondering if anyone has tried that? I have not done so myself, but it would be interesting to know if any of the methods they claim to use actually have any benefits in increasing ones IQ.

Leave a Reply

You must be logged in to post a comment.