Using gam­i­fied train­ing to improve cog­ni­tive control

Niko Stein­beis, UCL
@LabDcp

YouTube

By load­ing the video, you agree to YouTube’s pri­va­cy pol­i­cy.
Learn more

Load video

Full Tran­script:

Dr Niko Stein­beis:
Okay, here you go. Thank you so much, Jo, for the won­der­ful invi­ta­tion to come here. Are there you go, start my video. To present here today. It’s a real­ly won­der­ful oppor­tu­ni­ty and it’s actu­al­ly pret­ty much one of the first times that we’re going to be pre­sent­ing data from the games that we have designed togeth­er with Caul­dron. So the top­ic of the talk is Train­ing Cog­ni­tive Con­trol Using a Gam­i­fied Approach, and I’ll take you through step by step what exact­ly it is that we mean by that. In terms of the back­ground, then what exact­ly it is that we mean by cog­ni­tive con­trol? So this is very much to do with some­thing that we are try­ing to achieve each day, that is reach and man­age goals on a short term and a longterm basis. So we go through­out our day try­ing to achieve those goals, be it mak­ing break­fast, or get­ting some work done in the short-term, or slight­ly more long-term and more ambi­tious goals, such as try­ing to lose weight, or climb­ing the Mat­ter­horn in Switzer­land or even pub­lish­ing in nature.

Dr Niko Stein­beis:
So we’re try­ing to nav­i­gate our day-to-day in achiev­ing these goals and to be able to do that, we need to keep focus on the goal. We need to keep it in mind. And more impor­tant­ly, we need to try and resist temp­ta­tion on the way of achiev­ing that goal. And that requires us to inhib­it impuls­es that would stop us from achiev­ing those long-term goals. So cog­ni­tive con­trol then in terms of a more for­mal def­i­n­i­tion, sup­ports flex­i­ble and adap­tive respons­es to envi­ron­men­tal changes in the pur­suit of goals that we have in mind. And it’s been shown to be incred­i­bly impor­tant for a range of oth­er domains and real life out­comes. So it sup­ports qual­i­ty of rela­tion­ships. It is very pre­dic­tive of aca­d­e­m­ic attain­ment. It’s strong­ly relat­ed to well­be­ing, as well as men­tal and phys­i­cal health. And even more impor­tant­ly, the extent or abil­i­ties of cog­ni­tive con­trol dur­ing child­hood, are high­ly, high­ly pre­dic­tive of cog­ni­tive con­trol lat­er life and all those oth­er out­comes and aspects relates to cog­ni­tive con­trol lat­er in life.

Dr Niko Stein­beis:
So as such, it’s a real­ly foun­da­tion­al cog­ni­tive skill that’s impor­tant for a range of domains, which are real­ly impor­tant for lat­er life suc­cess and well­be­ing. Now, giv­en that cog­ni­tive con­trol plays such a core role in our every­day lives, there’s been a lot of dis­cus­sion and attempts to try and sup­port cog­ni­tive con­trol and improve it through ded­i­cat­ed train­ings. So I’ll talk a lit­tle bit about plas­tic­i­ty and the lit­er­a­ture so far on plas­tic­i­ty of cog­ni­tive con­trol. So it’s a real­ly, real­ly impor­tant ques­tion giv­en how impor­tant exec­u­tive func­tions and cog­ni­tive con­trol are for lat­er life. Well-being now so far today, most of the work and attempts to try and approve con­tent con­trol have focused on work­ing mem­o­ry. So that’s the abil­i­ty to keep infor­ma­tion in mind. So if we relate this to keep­ing goals in mind, work­ing mem­o­ry is very, very impor­tant. If we keep for­get­ting what it is that we want­ed to do, then we would­n’t be able to achieve it in the first place that work­ing mem­o­ry plays a real­ly impor­tant role in all of this.

Dr Niko Stein­beis:
But the evi­dence has been quite lim­it­ed in terms of stud­ies on train­ing cog­ni­tive con­trol. So it’s been pri­mar­i­ly focused on work­ing mem­o­ry. And the basic log­ic behind try­ing to improve cog­ni­tive con­trol of work­ing mem­o­ry in this par­tic­u­lar case is this, that we see a very strong rela­tion­ship in chil­dren for instance, between indi­vid­ual dif­fer­ences or abil­i­ties in work­ing mem­o­ry, so the extent to which we can keep infor­ma­tion in mind and aca­d­e­m­ic achieve­ments such as math or read­ing or IQ. So there’s a very well estab­lished rela­tion­ship between the two. And the idea is then that if we try and improve or train work­ing mem­o­ry, we also ought to see asso­ci­at­ed improve­ment in these relat­ed domains, such as aca­d­e­m­ic achieve­ment and IQ, but we don’t actu­al­ly see that. So by now there’s a lot of evi­dence that these so-called far trans­fer­ence and effects are extreme­ly elu­sive and very, very dif­fi­cult to get.

Dr Niko Stein­beis:
And this is baf­fling for a range of dif­fer­ent rea­sons because we know that expe­ri­ences gen­er­al­ize in long-last­ing ways and nat­ur­al learn­ing con­text. So just tak­ing edu­ca­tion for instance. And we also know that exec­u­tive func­tions or cog­ni­tive con­trol, it can be per­sis­tent­ly affect­ed by neg­a­tive events. So stress for instance, that can real­ly impact cog­ni­tive con­trol in long last­ing way. So why is it so hard to real­ly improve and train cog­ni­tive con­trol in long-last­ing ways? So the main ques­tion that we’ve pur­sued in our lab is whether improv­ing cog­ni­tive con­trol trans­fers to oth­er domains, and if it does, whether it’s also durable?

Dr Niko Stein­beis:
So I’ll talk a lit­tle bit about the train­ing and go to the actu­al approach that we use to do this, which is where Caul­dron came in and has been incred­i­bly cre­ative and, yeah, sup­port­ive try­ing to get this off the ground. So the basic idea is that at the very core of all of this is a very sim­ple cog­ni­tive con­trol task. This is called a Stop Sig­nal Reac­tion Time Task, where par­tic­i­pants are asked to respond to the pre­sen­ta­tion of a visu­al stim­u­lus on a screen. Each time they see a cir­cle for instance, they’re meant to press a but­ton. On some tri­als, 25% of the tri­als, they hear an audi­to­ry stim­u­lus. And when they hear the audi­to­ry stim­u­lus, which is played with a very, very short delay after the visu­al stim­u­lus, they have to inhib­it. And what’s inter­est­ing is that the delay between the go sig­nal and the stop sig­nal wax­es and wanes with indi­vid­ual per­for­mance. And the short­er the delay is the eas­i­er it is to inhib­it. And the longer the delay is the hard­er it is to inhibit.

Dr Niko Stein­beis:
So this is a very sim­ple test, which allows us to look at indi­vid­ual dif­fer­ences in cog­ni­tive con­trol abil­i­ties, but also the fact that there is such a thing as a Stop Sig­nal Delay, which is what this is called. It allows us to use this to actu­al­ly train inhi­bi­tion in a real­ly nice and well con­trolled and adap­tive fash­ion. So that’s exact­ly what we did. We lever­aged this par­tic­u­lar very, very sim­ple cog­ni­tive task to train inhi­bi­tion. So we have two groups, two groups that trained, one was the cog­ni­tive con­trol train­ing or what we call the exper­i­men­tal group. They were asked to respond to the go sig­nal and to inhib­it to the stop sig­nal. That was the exper­i­men­tal group. And then we had a con­trol group, which just, a trained response speed. So every time they saw the visu­al stim­u­lus, they pressed a but­ton as fast as they could.

Dr Niko Stein­beis:
What was real­ly impor­tant is that the task got eas­i­er and hard­er depend­ing on how good indi­vid­ual par­tic­i­pants were over indi­vid­ual train­ings. Now, one of the core chal­lenges in all of this is that we need lots and lots of train­ing to see durable effects. So what we went for, we designed a train­ing that last­ed eight weeks in total and where our par­tic­i­pants were asked to train four times a week. And to make sure that they actu­al­ly do this train­ing as well, we had to imple­ment a cou­ple of fea­tures. And again, that’s where the cre­ativ­i­ty of Caul­dron came in. So first off we opt­ed to mak­ing this gam­i­fied. We pre­sent­ed par­tic­i­pants with this task. They would get bored very, very quick­ly. And in our par­tic­u­lar case, we were look­ing at chil­dren to sort of increase the impact of all of this. And if we did­n’t make this in a gam­i­fied fash­ion, inter­est­ing fash­ion, then they get very, very bored and drop out very quickly.

Dr Niko Stein­beis:
So the gam­i­fi­ca­tion was absolute­ly key. And I’ll give you a cou­ple of exam­ples, how we did this. We tried to make it very vari­able through­out. And we made it adap­tive. So as I said, the train­ing got eas­i­er and hard­er, depend­ing on how good our par­tic­i­pants were. We made it com­plex and diverse. So I’ll show you exact­ly how we did that. So we pre­sent­ed to them with an over­ar­ch­ing nar­ra­tive by telling them that they had crashed a plane. They were a pilot and they’d crashed a plane in the desert. And to fix their plane, they need­ed to find the Sage. The Sage was able to give them spare parts, and to get the spare parts from the Sage they need­ed to col­lect trea­sure along the way. And to get the trea­sure, they nav­i­gat­ed these dif­fer­ent worlds. And in these worlds they were pre­sent­ed with tasks that gave them trea­sure, but also these tasks were used to train cog­ni­tive control.

Dr Niko Stein­beis:
So I’ll just give a short video here. So this is just a visu­al pre­sen­ta­tion here. This is where our par­tic­i­pants were at any giv­en point in the worlds which they nav­i­gat­ed. So here they are just about to enter a world. And we gave them a cou­ple of choic­es in terms of the kind of world that they’re about to enter. Before each game they were giv­en instruc­tions and here is one of the sets of instruc­tions. So here they were meant to press a but­ton each time they see a rock. But some­times there would be a gem dis­played in the rock. And when that hap­pens, there you go they weren’t meant to press the but­tons. So that’s exact­ly an indi­ca­tion of go tri­al and the stop tri­al. So you had eight dif­fer­ent games designed like this, and we changed the set­ting, we changed the con­text in ways to try and make this inter­est­ing and engaging.

Dr Niko Stein­beis:
So what I’m going to present to you now… Isn’t it fun? I love these games. So what I’m going to present to you now, a lit­tle bit more of the meth­ods and then some of the data in terms of how chil­dren actu­al­ly engage with these games. So we use this train­ing to address what the effect of cog­ni­tive con­trol train­ing is on var­i­ous out­come mea­sures. And whether these effects hold up after one year fol­lowup. We looked at this in 209 chil­dren from the Greater Lon­don Area. The mean age was around nine years. And here’s the basic design, we had pre-train­ing of a par­tic­u­lar set of trans­fer domains or train­ing domains. We looked at exec­u­tive func­tions, deci­sion-mak­ing tasks, men­tal health, and aca­d­e­m­ic achieve­ment, as well as brain struc­ture and brain func­tion. Then chil­dren engaged in eight weeks of the train­ing. And we looked at whether there was any change as a func­tion of the train­ing imme­di­ate­ly after­wards. And then one year afterwards.

Dr Niko Stein­beis:
Now we all know what hap­pened a year and a half ago, less than a year and a half ago, COVID struck. So we’re a lit­tle bit flum­moxed by this, but again, because all of this was done online, we could nav­i­gate this real­ly, real­ly beau­ti­ful­ly. And that was a huge plus of hav­ing designed all of this in the way that we had with, with Caul­dron. So I’m not going to present you any of the trans­fer. What I’ll show you now is just how exact­ly chil­dren engage with the train­ing. Because this is so incred­i­bly excit­ing and encour­ag­ing for a range of dif­fer­ent rea­sons. So first I’m going to look at the actu­al amount of train­ing [inaudi­ble 00:11:03] total. There were 32 ses­sions that chil­dren could have done over the eight weeks. And what we see here, here’s the exper­i­men­tal group, here’s the con­trol group. We see real­ly nice nor­mal dis­tri­b­u­tion. And we see that over­all chil­dren par­tic­i­pat­ing around 18 train­ing sessions.

Dr Niko Stein­beis:
Each train­ing ses­sion lasts around 15 min­utes and some even made it right through up until the 32. So that it’s real­ly, real­ly nice. What’s par­tic­u­lar­ly nice is that chil­dren in both groups trained approx­i­mate­ly equal­ly. So there are no dif­fer­ences between the groups and that makes it real­ly nice because we can then gen­uine­ly com­pare any effects pre-post due to the nature of the train­ing, as opposed to the amount of train­ing that chil­dren had engaged in. Anoth­er fea­ture that we imple­ment just to engage moti­va­tion, is we had a cou­ple of bonus games after each train­ing ses­sion, chil­dren train­ing. So what we have here is the per­cent of bonus games that chil­dren engaged in. And what we see around 30% of the bonus games were actu­al­ly tak­en up on. Which is again, quite a nice fea­ture [inaudi­ble 00:12:07] where we moti­vat­ed to do more of the train­ing. Then we had actu­al­ly asked them to. And again, this is com­pa­ra­ble between the two groups.

Dr Niko Stein­beis:
Now it gets bet­ter. So we also asked chil­dren… So each week we did the train­ing at school, and at school, we gave them a lit­tle ques­tion­naire to ask them how much they enjoyed the train­ing, how much they liked it, whether they felt they’d got­ten bet­ter at what­ev­er it is that they were train­ing. So what we see here, the top score is 40. So we sum these over sev­en ques­tions. The top score is 40. And here we’ve got the means. The means are sort of rea­son­able, I think chil­dren were pret­ty engaged through­out and it was sim­i­lar for both groups. And now what’s real­ly, real­ly strik­ing is here we see the moti­va­tion change over the weeks of train­ing that hap­pened. What we actu­al­ly see is that both groups seem to increase in terms of their motivation.

Dr Niko Stein­beis:
So even though the train­ing itself was a sig­nif­i­cant por­tion of time that they did each week, but chil­dren actu­al­ly became more moti­vat­ed doing the train­ing. And that was absolute­ly incred­i­ble and real­ly, real­ly sur­pris­ing because usu­al­ly what you’d see, the more you do some­thing, unless it’s actu­al­ly incred­i­bly enjoy­able, you’ll see a drop in moti­va­tion. But we don’t see that here, we actu­al­ly see an increase in moti­va­tion. That’s real­ly, real­ly love­ly. So again, won­der­ful. And it’s com­pa­ra­ble across the two groups. And now what I’m going to show you is just the extent to which the chil­dren improved on the tasks that they actu­al­ly trained dur­ing the train­ing. So here we have the inhi­bi­tion mea­sure, and this is the group of chil­dren that did the inhi­bi­tion, and here we see the abil­i­ty to inhib­it across the dif­fer­ent sessions.

Dr Niko Stein­beis:
And what’s absolute­ly amaz­ing is that vir­tu­al­ly all chil­dren improve on the core tasks that we want­ed to train with inhi­bi­tion. And we saw exact­ly the same for our con­trol group. So the chil­dren that trained inhi­bi­tion improved on inhi­bi­tion, the chil­dren that trained on response speed, improved on response speed. So with these data, we real­ly are in the best pos­si­ble posi­tion to actu­al­ly look at our effects and our trans­fer effects as gen­uine. It does­n’t real­ly mat­ter almost at this point, whether we see any­thing or not, the train­ing itself has worked extreme­ly well in terms of engage­ment, in terms of moti­va­tion. And in terms of improving-

Jo:
Sor­ry about that. We don’t know what hap­pened for some rea­son. The ses­sion just end­ed. You had got to your slide where you were show­ing us the impact of the train­ing schools on the SSRT and in your con­trol con­di­tion. Can you go back to that slide?

Dr Niko Stein­beis:
Yeah.

Jo:
In fact, there’s a ques­tion from one of the par­tic­i­pants is did the con­trol par­tic­i­pants improve on the SSRT?

Dr Niko Stein­beis:
No, the con­trol con­di­tion their aim was to just get faster basi­cal­ly at respond­ing. So the improve­ment for the SSRT was only seen for the group that trained SSRT and the con­trol par­tic­i­pants only improved on reac­tion time. So they got faster.

Jo:
Bril­liant. Now, why don’t you con­tin­ue from where you left off?

Dr Niko Stein­beis:
Sure. I mean, it’s pret­ty much done actu­al­ly, I’m so sor­ry. So just want to con­clude, the train­ing uptake was real­ly excel­lent. Par­tic­i­pants were very moti­vat­ed, which even increased over time. Key cog­ni­tive func­tions that were meant to be trained improved. So we’re real­ly in an absolute ide­al posi­tion from which to exam­ine near and far trans­fer effect. So to ref­er­ence the Sage, games are pret­ty close to per­fect. Did you cog­ni­tive train­ing stud­ies with chil­dren. Which is absolute­ly won­der­ful. So yeah, just thank­ing the lab, the fun­ders, you for your atten­tion and obvi­ous­ly Caul­dron and for the won­der­ful, won­der­ful, extreme­ly accom­mo­dat­ing and patient col­lab­o­ra­tion. It was real­ly ter­rif­ic. And I think these results absolute­ly speak for themselves.

Jo:
Yeah. Well, you’re more than wel­come Niko, the work you and your team do is real­ly extra­or­di­nary. Now I was say­ing this before you joined in, so many researchers have been try­ing to mea­sure trans­fer effects in the lab and have failed to do this for years and years and years. And you have suc­ceed­ed, or at least we’ve got the begin­nings of a glimpse that you are suc­ceed­ing. What impact do you think that’s going to have on your area of research and relat­ed areas of research? What does this open up as for possibilities?

Dr Niko Stein­beis:
I mean, I guess the jury is poten­tial­ly still a lit­tle bit out on whether we actu­al­ly have trans­fer effects. I think we’re in the best pos­si­ble posi­tion to real­ly exam­ine whether we do. And I think the amount of data that we have col­lect­ed gives us a huge amount of lever­age in terms of being able to under­stand why the train­ing would be effec­tive. Because we record­ed every sin­gle ses­sion of a par­tic­i­pant. We can look at the wax­ing and wan­ing across ses­sions. We can look at the fluc­tu­a­tions and that gives an enor­mous scope to real­ly explore what it is that poten­tial­ly leads to and explain trans­fer effects.

Dr Niko Stein­beis:
So, I mean, I think if we do end up demon­strat­ing that this is pos­si­ble, I think that’d be just huge­ly excit­ing because on the one hand show­ing that chil­dren are very moti­vat­ed to do this. I think there would be a lot of scope to try and col­lab­o­rate with schools in terms of imple­ment­ing this more in the cur­ricu­lum. Both on the one hand, look­ing at pos­i­tive effects on men­tal health, but also look­ing at attain­ment. But also look­ing at social abil­i­ties and qual­i­ty of social rela­tion­ships. I think these are aspects that mat­ter mas­sive­ly to child devel­op­ment. And I think if we do real­ly gen­uine­ly show some trans­fer effects, then the poten­tial for the appli­ca­tion is absolute­ly enor­mous, absolute­ly enor­mous. Yeah.

 

Get on the Registration List

BeOnline is the conference to learn all about online behavioral research. It's the ideal place to discover the challenges and benefits of online research and to learn from pioneers. If that sounds interesting to you, then click the button below to register for the 2023 conference on Thursday July 6th. You will be the first to know when we release new content and timings for BeOnline 2023.

With thanks to our sponsors!