PsychologiCALL

On theta waves and information processing, with Emily Jones

April 02, 2021 SalvesenResearch Season 2 Episode 9
PsychologiCALL
On theta waves and information processing, with Emily Jones
Show Notes Transcript

Emily is a developmental psychologist who works at the Birkbeck Babylab in London and specialises in understanding early neurodevelopmental pathways to conditions such as autism and ADHD. During this podcast she chats to Sue about a piece of work looking at how changes in early brain activity may predict later cognitive skills in neurodiverse cohorts.

You can find out more about Emily's work by checking out the BONDS project pages and the Birkbeck Babylab site, and you can follow her on twitter here.

The paper discussed in this episode is:
Jones, E. J. H., Goodwin, A., Orekhova, E., Charman, T., Dawson, G., Webb, S. J., & Johnson, M. H. (2020). Infant EEG theta modulation predicts childhood intelligence. Scientific reports, 10(1), 1-10.

Many thanks to Naomi Meiksin for editing the transcript for this episode. 

Intro:

Oh, it is recording. I see the little figure. Okay, great . I will do my little spiel and then I'll introduce you. Nice . Okay.

Sue:

Hi everyone. I am Sue from the Salvesen Mindroom Research Center at the University of Edinburgh, and I'm recording another episode of our podcast, PsychologiCALL. Um, we are trying to make a little bit of an evidence-based contribution to the conversations that lots of people are having about child and adolescent wellbeing and development and learning at the moment. And today I'm absolutely thrilled to be talking to Emily Jones from the Center for Brain and Cognitive Development at Birkbeck University of London, and she is going to be talking to me about a paper. We will have to unpick the title of this a bit, which is called Infant EEG theta modulation predicts childhood intelligence. Hello, Emily.

Emily:

Hi, Sue.

Sue:

How are you doing?

Emily:

Good - well, surviving.

Sue:

[laughs] Thank you for coming on the, on the podcast. Um, so before we kind of analyse the , some of the words in that title a bit, could you just start by telling me what you discovered when you did this piece of research?

Emily:

Yeah, so we found that individual differences in how the child's brain responds to new social and non-social videos predict their later cognitive development in childhood.

Sue:

Amazing. Oh, so snappy Emily [inaudible].

Emily:

[Laughs].

Sue:

Okay. So let's just go back into , um, uh, the kind of , um , details about it a little bit. Right. So , um, you w - describe to me what you mean by EEG theta modulation and what that kind of represents, if you could please.

Emily:

Sure. So what we're doing here is measuring brain activity using EEG. Um, so networks of sensors that measure the coordinated activity of lots of neurons together , um, and brain signals are , uh , generally these sort of complex wave forms, and we can divide those up into different frequencies, so different sort of speeds of , of communication. Um, and theta's one of those, one of those frequency ranges. Um , and people, you know , have looked at it a lot and we think that theta has some role in , in learning attention and memory. And here, what we were doing was looking at how activity in that theta band changes in response to viewing naturalistic things. So babies watch videos of women telling nursery rhymes or toys moving, and we look at how their theta power changes as they watch that video. So the degree to which , um, theta power changes in that individual child .

Sue:

And do you, is there a kind of , um , precise , uh, interpretation, you know, like an increased theta power means that that child is more engaged with what they're looking at or, or is, I don't know, working harder to process what they're looking at? Can we , can we take it that far or is that really conjecture?

Emily:

Yeah, well, so that's a great question. So I, I think those are two hypotheses for what's going on, but yeah , we don't exactly know. Yeah. One possibility is that its increase in theta means that they're engaging with , uh , um, the material. Um, you know, we find that even at the teen age , the kids who had a bigger increase in theta were the ones who had stronger cognitive skills. So it could be about engaging with the material or, yeah , it could be about working harder to process it, which again, could be indicating that they're processing it at a deeper level. We're not sure. And that's , that's the part that's difficult to address because, you know, with infants, you can't do the kind of manipulation studies that you might do in a , in an animal model. So we can't manipulate theta to causally look at what it's actually doing, so we have to put a try to infer from patterns of association with other tasks and things.

Sue:

But I suppose, given this link to , um, uh, a later kind of measure of IQ, you know, you can , I guess, rule out the possibility that , um, it's not just adaptal processing in the sense of someone who's actually really struggling with the material, right? Because this is something I think about with eye-tracking data, you know, are you looking at something more because you're just so fascinated by what you're looking at, or are you looking at it more because you're puzzled by it and can't really work it out. And so that link to IQ is quite revealing, isn't it?

Emily:

Yeah, I think there's, I think, yeah, you can certainly probably rule out that it's to do with struggling. It could be, that they're making more effort still, but they're making more effort and that leads to more learning and you know, why the kids have positive development. But yeah, you could probably argue that they're not making more effort because they're finding the material difficult [inaudible], or they're not understanding the rides , or, you know, they're not being able to process them even as a toy or something like that. And these are relatively, you know, the toy videos are like little balls going down a shoot , you know, that they're probably not, for a 12 month old, they're probably not very complicated to understand, but I think it probably is much more likely that it's a sort of attention engagement measure. Um, but yeah , we can't say for certain.

Sue:

Yes, yes. Extracting lots of, of information and learning from the content that they're getting.

Emily:

Exactly. Yeah. If you see something new and you engage more with it and you learn more from it, you know, over time you might imagine that that accumulates into, you know, a more, sort of, positive cognitive developmental trajectory.

Sue:

Um , so gosh, we , we, we're getting ahead of ourselves though , because I really should be asking you, I suppose, how you came into asking this question in the first place. Could you tell us a bit more about the kind of wider context of the study? Um, was this part of one of your longitudinal cohorts?

Emily:

Yeah, so we, we run a lot of longitudinal studies of both typically developing babies, but also babies who have , um, family members with neuro-developmental conditions like autism or ADHD. And we're interested in what in early development might predict later outcomes. So both cognitive outcomes, but also social outcomes, you know , for two reasons, partly, partly from a sort of more clinical idea of identifying children early, who might need some extra support, but also scientifically trying to understand y'know what individual differences in early development matter for later skills and which ones don't. You know , and so can we identify some of these paths through which you might get, you know, influences on learning? So , you know, what, what types of attention matter for what the child accumulates uh and what don't. So it came out of a broader, broader set of work on that. And here we're particularly looking at the, yeah the cognitive development side.

Sue:

And was this a particular group of infants who took part in this study?

Emily:

Yeah, this study actually integrated data across a number of cohorts . So the first is a group of typical babies where they were all seen at 12 months. So we were looking concurrently at these change in , um, cognitive development. And then we looked at whether we could replicate and extend it in a couple of other cohorts. One that I worked with in the US when I was a post- doc there, So those are babies with older siblings with autism. And so in that cohort, we saw that this theta change measure at 12 months predicted cognitive development by two, and then, in the UK, the basis study that I now work with, we then looked at that study and we could show that this theta change measure at 12 months , predicts their cognitive outcomes both at three and age seven. Um , so it was nice because we were able to pull together these existing data sets to look at this one measure and sort of consistency across multiple different , um, different developmental timeframes.

Sue:

That's very cool. And so impressive, especially in infant research, right? Where, where any kind of replicability is so challenging to achieve. Um, so yeah, well done. That's amazing, Emily .

Emily:

Thank you. And then we, so we did then go on with, with , um, Ellie , one of my Masters students, she then looked at , um, whether or not we could replicate the same thing in a preregistered design , um , for that exact reason that you were saying that, you know, in terms of replicability, it's really important to think about whether or not we can then, you know , pre-specify and that other paper gave her the nice hypothesis. So then we looked at , um, a different cohort that was Karla Holmboe's cohort of babies. And again, we showed that theta change , um, predicted later cognitive development - this time between six and nine months. Um, but we effectively confirmed our preregistered hypothesis . So we think this is , uh , you know, at least a relatively robust result. Um, even if we don't understand, y'know fully.

Sue:

Yeah. Well, I might try and push you on that in a moment or two, but I suppose first I'm just curious about the process of data collection, right? I think people listening will be interested to hear how you do something like EEG with babies. So could you tell us a wee bit about that?

Emily:

So - with difficulty. So we use these, these sensor nets that , um , they get soaked in water and there's like 128 little, little sensors with sponging on them. So they're designed to be infant friendly , um, and you know, once the hats are made , they tend to forget about it and then we play them. And this is partly why we use these kinds of videos because they're engaging for the baby and enjoy them. So, you know , we use the kinds of videos that infants might like watch. So yeah, here they're videos of nursery rhymes or toys moving. Um , and then we do some other tasks. [inaudible] babies have snacks if they need them we blow bubbles, lots of [ laughs] lots things to try to get them to , to stay calm and engage with us. But , um, broadly speaking we're fairly successful.

Sue:

Yeah, do you think, you know, because I I've done some infant research as well, and that, it's something I often think about, you know, there are babies who come in that you don't manage to get data from. Right? You know, how much do you feel like that sort of random, you know, the baby on the day, not really in the mood and how much do you worry that we're sort of systematically missing some kids who maybe are developing atypically or, you know, and it was sort of not complying with our data collection efforts?

Emily:

No , it's a great question. We looked at it in a few studies. There was one EEG study we did with, with kids with autism, where we had a very high dropout rate and we looked at y'know, but the kids were all extremely well-characterized in a ton of different measures. And the only individual difference measures that related to whether they did EEG or not was tactile sensitivity, which you might imagine. But all of their cognitive skills, social skills, like none of those , um , actually related to whether or not they did the EEG . And we find a similar thing in the baby studies but whenever we looked at it, a) it doesn't seem to be particularly reliable: so, you know, a kid'll do the EEG at five months and then won't at 10 months and they will again at fourteen months... Y'know it doesn't seem to be a particularly stable individual difference. Um , and we've been doing some more test retest reliability studies recently where we see a similar thing. So I think that actually probably in infancy, certainly, it's actually a bit more random, just is it a good day for that child or not, with , uh , with a bit perhaps of , uh , sort of tactile issues thrown in there. So I think if particularly studying tactile sensitivity, then it might be more of a worry. But interestingly enough actually, we don't seem to find that systematically it drives anything out, at least in what we've done.

Sue:

Oh, that's so reassuring.

Emily:

Yeah, and surprising somewhat because you sort of, and it's interesting cause you get , you know, when you talk to mums beforehand often they'll say "oh, he doesn't like hats, he won't wear hats" and again, that doesn't seem to necessarily predict whether or they will or won't do it. And so, yeah.

Sue:

Mm. And so tell me a bit about the kind of analysis that you were doing. You know, you've got all of these different cohorts , um, I guess you've got different timeframes. You mentioned, I think the first one was a cross-sectional data set and then you're looking more longitudinally, you know, how, what were some of the kind of challenges that you were dealing with with the analysis?

Emily:

Yeah, I mean, you know, there's obviously with longitudinal studies, we have a lot of issues with things like missing data, you know, thinking about imputation methods or using models that can take that into account. And the other issue always with hte data is the sort of potential for multiple comparisons. And so that's why actually in his paper we did, in the first, the 12 month concurrent cohort, we looked at lots of different things. But then, because what we found was this associating with frontal theta, we just looked at in the other two cohorts . So we were able to use the fact that we have multiple cohorts and take a sort of discovery and pseudo-replication. I mean, then it's not a direct replication because, as you say, we looked at slightly different ages in the other two, but we were able to take this , um, this approach of then narrowing down and just looking at one thing, which, I think gave us more confidence in the findings. Um , and then the other thing I would say that, the reason we were able to do that is because , um, those videos that we designed and used across multiple cohorts were...originated from the collaboration between Mark and Jerry and [inaudible] a long time ago. And we all sort of committed to using the same videos in multiple different studies. And that's allowed us now to start reaping the benefits of that, where we can look at at least 70 replication cohorts to get more confidence in things. And I think that that data sharing is really important, particularly in developmental science.

Sue:

Yeah, absolutely completely agree. And it's such a kind of inspiration, I think the kind of group down at Birkbeck has been such leaders in this. So , um, yeah, I completely agree. And it's so nice to see it yielding results because I bet you put all of that in place, you know, years of years ago, right?

Emily:

Yeah, years ago. That was like 2005 probably, yeah . Right.

Sue:

Yeah. So it's a , it's a very patient approach to science that you're adopting.

Emily:

It's very slow science . Right . [Laughs]

Sue:

So what do you think, what do you think we should be learning from this finding? I suppose I'm interested in , um, the, you know, is theta, is this theta signal that you're picking up a kind of driver of something? I know, I mean, I know obviously, as you said, we can't really manipulate an instance of theta waves. Right. But is it , is it, is it a signal, is it a marker or something, or is it a driver of something? And can we tell the difference, I suppose?

Emily:

Yeah, I mean, that's a really great question. So I mean, we can start to get towards it. So, Katarina Vegas has done some really nice stuff looking at , um, you know, if you give children, individual toys and you look at which toy they show a theta response to, you can subsequently show they're more likely to learn or remember it, so, you know, differentiating in , in between different objects that the child interacted with and using that, theta signal to predict immediate learning. Um, something that we're trying to do at the moment is, is set up these sort of real-time closed-leaf approaches where we measure theta in real time and then try to find what it is that engages theta responses best in that particular baby. And then again, we can look at is that also then the thing that they would learn and remember better? So we can get more towards at least whether, you know, theta modulation associates with learning and memory on a smaller timescale, but it still doesn't fully address t hat sort of causality question. Is it the theta that's sort of 'causing it', or is the theta just an index for something else that's happening. I n human infants, y ou'd h ave, you know, I d on't k now if y ou can do it. I think in animal models, y ou k now, obviously there's hippocampal theta, you can look at t hose kinds of rhythms, and hippocampal theta is very important in learning and memory, and you can look at it causally b e there, but in human development, I don't know, I'm sure somebody will invent a c lever method. I mean, you can drive theta, right? You can show babies, you know, visual stimuli that flicker in the theta range and increase their theta power, and then if that does improve memory for that thing they were looking at. So that might be one way of doing it. I don't know if anyone's done that yet, probably.

Sue:

Well, maybe that's the next step. But I mean , we don't necessarily want to end up with, you know , classrooms full of ... um , [ inaudible]. I guess I'm curious as well, whether it's really interesting to hear you talking about, you know, there's studies comparing the theta response to different individual toys or objects, right? And then relating that to learning. I suppose I'm curious as well, whether you'd see that, maybe at a domain level. You know, so I mean, I'm risk - this is a bit risky because I don't want to invoke anything like learning styles here, right? Listeners - learning styles are nonsense. But, you know , definitely people have different areas of strength, right? You might, you know, you might have really advanced language development, you might have really good motor control or whatever, you know, and I just wonder if there would be any relationship there in terms of individual differences and kind of profiles of , of cognitive strengths at a more domain level rather than , you know, individual things you've interacted with.

Emily:

Yeah, and I think, that, so that's kind of where we want to go with this real time stuff because in a lot of the work we do with say kids with autism, you know, we , we show often what we do is show them stimuli that typical kids like to look and then look at whether they don't like them as much. But actually, what we want to know, right, is what they are interested in: what would engage them? What are they learning about? Then we can build off of that and this sort of closed loop idea where you might use something like a theta signal, exactly, to look at different domains or different sort of types of experience or ways in which information is presented and you can identify for individual kids. Yeah, what, what, which way of presenting information is optimal for them or which yeah, types of types of domain they may be interested in through using feedback from their brain, you know, that could potentially be much more generalizable then.

Sue:

Um , yeah, well that is really exciting.

Emily:

Yeah!

Sue:

Um , well that is amazing. Emily, thank you so much for telling me all about it. And before we finish, we do have to wrap up the call, I keep calling this podcast bitesize, but I'm sorry, I'm trying to deliver on that promise, but , um, I wondered if you had any advice for any kind of early career researchers or students who are listening, if there was anything that you would want to share with them from your , um , perspective?

Emily:

Yeah. I mean, I always struggle with this you know and I think the only thing I always say is like , I don't really have any advice, and I think the main thing I would say is maybe don't listen to most of the advice that's out there. I think everybody has a very different path and I, you know, it's an interesting thing, because it's sort of the extent to which you can derive things or , you know, what , what other [inaudible] in your research career, I dunno but, I would just try and enjoy it, but w and trying not to look at other people too much and think I should be doing that, I should be doing that. Because honestly, sometimes that's just a path to misery, maybe that's just me .

Sue:

Oh, no, I think that's really fantastic and quite liberating for people. And then problem, isn't there with kind of survivor bias, you know, so you can look back on your, on your career pathway and see a sort of causality, you know, 'I did this and that's how I got to where I am' and yeah. Not necessarily.

Emily:

No, exactly. I mean, I think my tea is at McDonald's , I would say that was an incredibly formative experience, but other than that , yeah, I don't know.

Sue:

This is great. This is a true longitudinal developmental researcher, right ?

Emily:

Exactly, your research analysis: your own career path. Exactly, correlation, not causation.

Sue:

Fantastic, Emily. Thank you so much for your time . Um, for anyone who is listening, you will be able to find out more about Emily's work by following the links in the podcast description in your podcast app, by tapping on the podcast details or on the Buzzsprout page where the podcast is listed. Um, thank you so much, Emily. It was fantastic to talk to you. Thank you for giving me some time in the middle of lockdown. I really appreciate it.

Emily:

No worries, thank you!

Sue:

Good luck with whatever comes next.

Outro:

Okay. We did it. I thought that went quite smoothly.