Can Intelligence Be Trained? Martin Buschkuehl shows how

Today I had a great con­ver­sa­tion with Mar­tin Buschkuehl, one of the Uni­ver­si­ty of Michi­gan Cog­ni­tive Neu­roimag­ing Lab researchers  involved in the cog­ni­tive train­ing study that has received much media atten­tion (New York Times, Wired, Sci­ence News…) since late April, when the study was pub­lished at the Pro­ceed­ings of the Nation­al Acad­e­my of Sciences.

Ref­er­ence: Jaeg­gi, S. M., Buschkuehl, M., Jonides, J., & Per­rig, W. J. (2008). Improv­ing Flu­id Intel­li­gence With Train­ing on Work­ing Mem­o­ry. Pro­ceed­ings of the Nation­al Acad­e­my of Sci­ences of the Unit­ed States of Amer­i­ca, 105(19), 6829–6833 (You can read it here, with subscription).

Before you keep read­ing, let me clar­i­fy a cou­ple of terms:

- “Work­ing Mem­o­ry” is the abil­i­ty to hold sev­er­al units of infor­ma­tion in our minds and manip­u­late them in real time. For exam­ple, imag­ine I ask you to remem­ber, and then say back­wards, the 7 dig­its of my phone number.

- “Flu­id intel­li­gence” can be described as the abil­i­ty to deal with new chal­lenges and new prob­lems, those that we encounter for the first time.

Dr. Buschkuehl, nice to talk to you. Can you first pro­vide us with some con­text on your research?

My col­lab­o­ra­tor Susanne Jaeg­gi and I start­ed our train­ing work four years ago in the Lab of Prof. Wal­ter Per­rig at the Uni­ver­si­ty of Bern, Switzer­land. Now we are both Post Docs in Prof. John Join­des Lab at the Uni­ver­si­ty of Michi­gan. We devel­oped a com­plex com­put­er­ized task and have tried it in a num­ber of stud­ies. We report­ed our results in two unpub­lished dis­ser­ta­tions, but this is the first time it has been pub­lished in a peer-reviewed journal.

Could you please explain the train­ing involved in this par­tic­u­lar study? 

We recruit­ed 70 stu­dents aged around 26 years and set half of them on a chal­leng­ing com­put­er-based cog­ni­tive train­ing reg­i­men, based on the so-called “n‑back task.” This is a very com­plex work­ing mem­o­ry task that involves the simul­ta­ne­ous pre­sen­ta­tion of visu­al and audi­to­ry stim­uli. The exper­i­men­tal group watched a series of screens on their com­put­ers, where a blue square appeared in var­i­ous posi­tions on a black back­ground. Each screen appeared for half a sec­ond, with a 2.5 sec­ond gap before the next one appeared. While this hap­pened, the trainees also heard a series of let­ters that were read out at the same rate. 

At first, stu­dents had to say if either the screen or the let­ter matched those that popped up two cycles ago. The num­ber of cycles increased or decreased depend­ing on how well the stu­dents per­formed the task. The stu­dents sat through about twen­ty-five min­utes of train­ing per day for either 8, 12, 17 or 19 days, and were test­ed on their flu­id intel­li­gence before and after the reg­i­men using the Bochumer-Matrizen Test (this is a prob­lem-solv­ing task based on the same prin­ci­ple as the very well known Raven’s Advanced Pro­gres­sive Matri­ces. How­ev­er, it is more dif­fi­cult and there­fore espe­cial­ly suit­ed for aca­d­e­m­ic samples).

What were the results?

Par­tic­i­pants in the exper­i­men­tal group did sig­nif­i­cant­ly bet­ter on the flu­id intel­li­gence test (which was not direct­ly trained) than par­tic­i­pants in the con­trol group. Those in the con­trol group had not gone through any train­ing. The con­trol group did improve slight­ly, but real “trainees” out­per­formed them (see Fig­ure Xa). Fur­ther­more, we found that the improve­ment was dose-depen­dent: the more they trained, the larg­er the gain on flu­id intelligence.

Images: PNAS.

We just pub­lished a mar­ket report to cov­er the grow­ing brain fit­ness soft­ware mar­ket. A com­mon ques­tion we get is, “How are com­put­er­ized pro­grams like the one you used fun­da­men­tal­ly dif­fer­ent from, say, sim­ply doing many cross­word puzzles?

First, thank you for send­ing the report along. Fas­ci­nat­ing to see what is start­ing to hap­pen in this field.

In terms of why our pro­gram worked, I could say that the pro­gram has some inher­ent prop­er­ties that are at least in this com­bi­na­tion unique to our train­ing approach. Our pro­gram is:

• Ful­ly adap­tive in real-time: The per­son using the pro­gram is tru­ly pushed to his or her peak lev­el all the time, there­by “stretch­ing” the tar­get­ed ability.
• Com­plex: We present a very com­plex task, mix­ing dif­fer­ent forms of stim­uli (audi­to­ry, visu­al) under time pressure.
• Designed for Trans­fer­abil­i­ty: The tasks can be designed in a way that do not allow for the devel­op­ment of task-spe­cif­ic “strate­gies” to beat the game. One needs to tru­ly expand capac­i­ty, and this helps ensure the trans­fer of to non-trained tasks.

This is very dif­fer­ent from enhanc­ing task-spe­cif­ic capac­i­ties, such as mem­o­riz­ing lists of 100 num­bers, which have been shown not to nec­es­sar­i­ly trans­fer to relat­ed domains.

Can you give an exam­ple of the lack of trans­fer­abil­i­ty of oth­er train­ing methods?

In Eric­sson’s clas­sic paper (Eric­s­son, K. A., & Delaney, P. F. (1998). Work­ing mem­o­ry and expert per­for­mance. In R. H. Logie & K. J. Gilhooly (Eds.), Work­ing Mem­o­ry and Think­ing (pp. 93–114). Hills­dale, NJ: Erl­baum), peo­ple who could mem­o­rize 100 num­bers, using a vari­ety of mnemotec­nic tech­niques, could not get even close to 100 let­ters. Remem­ber­ing num­bers did­n’t trans­late into remem­ber­ing oth­er things, so it was­n’t a gen­er­al mem­o­ry capac­i­ty that had been improved.

What are the par­tic­u­lar aspects of the Uni­ver­si­ty of Michi­gan study that sur­prised you the most?

First, the clear trans­fer into flu­id intel­li­gence, that many researchers and psy­chol­o­gists take as fixed.

Sec­ond, I was sur­prised to see that the more train­ing the bet­ter the out­come. The improve­ments did not seem to peak early.

Third, that all trained groups improved, no mat­ter their respec­tive start­ing points. In fact, stu­dents with low­est flu­id intel­li­gence seemed to improve the most. But that was not the main focus of our study, so we can not say much more about it.

How did par­tic­i­pants describe the expe­ri­ence, and their benefits? 

Many liked the train­ing. They saw the chal­lenge, and tried hard to push them­selves through the train­ing to see how far they could go.

We did not ana­lyze how the flu­id intel­li­gence gains trans­ferred into real life. But from an anec­do­tal point of view, many par­tic­i­pants have shared sto­ries of how they per­ceive a major ben­e­fit. Now they can fol­low lec­tures more eas­i­ly, under­stand math bet­ter etc.

There is a degree of arti­fi­cial con­tro­ver­sy these days in the media and the sci­en­tif­ic com­mu­ni­ty on the respec­tive ben­e­fits of phys­i­cal or men­tal exer­cise. Your thoughts?

We obvi­ous­ly need both. Phys­i­cal exer­cise keeps the body in a good shape but espe­cial­ly in old­er peo­ple also leads to cog­ni­tive ben­e­fits. Men­tal exer­cise, like the one we used, can enhance impor­tant abil­i­ties and is most like­ly the most effi­cient way to improve a spe­cif­ic cog­ni­tive process but also gen­er­al­izes to a broad­er range of skills, as we showed.

Research will need to help clar­i­fy who needs what type of exer­cise more. Some peo­ple may get enough men­tal exer­cise through very com­plex jobs and what they need is phys­i­cal exer­cise. For oth­ers, it may be the opposite.

What are your plans now?

First, to con­duct fol­low-up research to ana­lyze the neur­al basis of the improve­ment via neu­roimag­ing stud­ies and try to mea­sure ben­e­fits in real life.

But our main hope is to be able to inves­ti­gate and devel­op appli­ca­tions for peo­ple who need it most: chil­dren with devel­op­ment prob­lems, stroke/ TBI rehab, and old­er adults.

Also, let me note that there is a cross-plat­form appli­ca­tion avail­able (Note: Here), that allows to train with the dual n‑back task and sev­er­al oth­er train­ing tasks that we devel­oped for oth­er stud­ies. Although the appli­ca­tion is avail­able in Eng­lish, the Man­u­al and the Brain­Twister Web­site are not at the moment. We are about to release an Eng­lish ver­sion, but unfor­tu­nate­ly I can­not give you a release date right now. If the train­ing pro­gram is used for research (i.e. a train­ing study), it is pro­vid­ed free of charge.

Mar­tin, many thanks for shar­ing your time and insights with us. Please keep us informed of new developments.

My plea­sure. We will.

—————————-

Ref­er­ence: Jaeg­gi, S. M., Buschkuehl, M., Jonides, J., & Per­rig, W. J. (2008). Improv­ing Flu­id Intel­li­gence With Train­ing on Work­ing Mem­o­ry. Pro­ceed­ings of the Nation­al Acad­e­my of Sci­ences of the Unit­ed States of Amer­i­ca, 105(19), 6829–6833 (You can read it here, with subscription).

For relat­ed inter­views on work­ing mem­o­ry train­ing, see

Mem­o­ry train­ing and atten­tion deficits: inter­view with Notre Dame’s Bradley Gibson

Work­ing Mem­o­ry Train­ing: Inter­view with Dr. Torkel Klingberg

Work­ing Mem­o­ry Train­ing from a pedi­a­tri­cian perspective

And, if you want to try the task your­self before the offi­cial web­site men­tioned above is ready (and we’ll keep you updat­ed), you can do so Here.