WEBVTT
00:00:09.150 --> 00:00:09.150
[Podcast jingle][ringtone] Hello?
00:00:09.150 --> 00:00:09.750
Oh, it is recording.
00:00:09.750 --> 00:00:10.679
I see the little figure.
00:00:10.740 --> 00:00:11.169
Okay, great.
00:00:11.369 --> 00:00:14.189
I will do my little spiel and then I'll introduce you.
00:00:14.310 --> 00:00:14.699
Nice.
00:00:14.910 --> 00:00:15.240
Okay.
00:00:15.630 --> 00:00:16.079
Here I go.
00:00:18.260 --> 00:00:18.649
Hello.
00:00:18.739 --> 00:00:23.210
I'm Sue from the Salvesen Mindroom Research Centre at the University of Edinburgh.
00:00:23.660 --> 00:00:34.340
We wanted to do something for all the practitioners and parents and students who are stuck at home, and who still wants to be developing their knowledge, even though it's a lockdown at the moment in the UK.
00:00:35.420 --> 00:00:41.929
Um, and also cause lots of people are homeschooling their kids at the moment and their usual family routines have been disrupted.
00:00:42.259 --> 00:00:47.060
It seems like developmental psychology is a really interesting thing to be talking about at the moment.
00:00:47.659 --> 00:00:53.990
So I'm doing these short phone calls with psychologists who study learning and development in children and young people.
00:00:54.649 --> 00:01:01.429
And today's PsychologiCALL is with my very dear friend, Duncan Astle from the University of Cambridge.
00:01:01.850 --> 00:01:08.269
And he's going to talk to me about some work he's done on executive function in a really interesting community sample.
00:01:09.049 --> 00:01:09.739
Hello Duncan!
00:01:10.969 --> 00:01:11.930
Hi Sue! Thanks for having me!
00:01:12.859 --> 00:01:16.609
[Laughs] Thank you for coming on my thing that's not quite a podcast.
00:01:16.049 --> 00:01:16.609
[Laughs]
00:01:19.909 --> 00:01:22.459
So, tell me why did you pick this paper?
00:01:22.460 --> 00:01:24.709
What did you discover that you want to talk about today?
00:01:25.939 --> 00:01:46.790
So this paper is all about executive function behaviours in children, and we found that parents ratings of executive function behaviours fall broadly into three different types, and those different types of behaviours don't really fit with the kinds of diagnoses that children often have.
00:01:48.680 --> 00:02:01.099
Um, that is very interesting! So are you, sorry, I'm jumping ahead to the conclusions that you concluding from that, that, um, parents don't have a good grasp on their kid's executive functions?
00:02:01.159 --> 00:02:06.349
Or that the way that we categorized things previously is maybe not as good as we thought it was?
00:02:07.069 --> 00:02:08.300
Option number two.
00:02:08.419 --> 00:02:28.039
So we infer from that the parents are actually good at rating, um, behaviours, but that the way that the kinds of categories that the children might come with, when we study them in our research studies, might not be that closely related to the actual behaviours that we were able to observe.
00:02:29.240 --> 00:02:29.569
Great.
00:02:30.590 --> 00:02:33.139
Um, so what, what motivated you to study this?
00:02:33.140 --> 00:02:35.599
Why did you decide this was an interesting thing to look at?
00:02:36.110 --> 00:02:36.409
So.
00:02:36.411 --> 00:02:37.099
Good question.
00:02:37.159 --> 00:02:57.199
So the background is that, um, in Cambridge, myself, Joni Holmes, Sue Gathercole, and a whole team of others has collected this really interesting community cohort of children who have been identified by practitioners as struggling in the areas of attention, learning, and memory.
00:02:57.200 --> 00:03:02.620
And they could have one diagnosis, multiple diagnoses, or no diagnosis at all.
00:03:02.889 --> 00:03:10.210
The key thing is that at some point some professional had said that this child is struggling and they had referred them to us study.
00:03:10.870 --> 00:03:20.979
And one thing that we observed quite early on is that the ratings of executive function related behaviour problems were particularly high in the cohort.
00:03:20.980 --> 00:03:35.289
So for instance, um, lots of, um, caregivers would say that the children struggled to maintain attention, um, symptoms like, um, hyper activity were quite common, um, and so that is kind of important context.
00:03:35.379 --> 00:03:37.750
So we've collected this large group of kids.
00:03:38.150 --> 00:03:44.050
We had lots of ratings of executive functioning behaviours, and they tended to be pretty high.
00:03:44.560 --> 00:03:50.110
And so we started to think about how we might look and analyze those kinds of behaviours.
00:03:49.661 --> 00:03:54.039
And that really kind of set the scene, um, for what we went on to do.
00:03:55.289 --> 00:03:58.139
And so tell me a bit then about the analysis.
00:03:58.740 --> 00:04:01.710
Um, did you use very complicated, clever techniques?
00:04:02.189 --> 00:04:03.990
Is it something you can describe to us?
00:04:04.889 --> 00:04:07.409
I think so.
00:04:07.411 --> 00:04:15.599
So, um, it's actually really all led by a post-doc who was working with me called Joe Bathelt, who now has his own lab at Royal Holloway, which is very exciting, anyway.
00:04:15.600 --> 00:04:20.970
And he came to me and he said"I think that we should do something called a network analysis".
00:04:20.971 --> 00:04:28.800
And this was about three or four years ago, when not many people were using this kind of technique and I didn't really know much about it.
00:04:28.800 --> 00:04:32.939
U m, and so he had to kind of explain it to me, and I will try and explain it to you now.
00:04:33.060 --> 00:04:39.209
So at the start I want you to imagine that you put a sack, a brown sack of some sort in your lap, and inside it, you've got fruits.
00:04:39.329 --> 00:04:39.930
Okay?
00:04:40.470 --> 00:04:42.990
I've got a...I've got a sack of fruit in my lap!
00:04:46.920 --> 00:04:46.920
Okay[laughs], and you've got a clear table in front of you...
00:04:46.920 --> 00:04:46.920
Yeah!
00:04:47.519 --> 00:04:47.790
Okay.
00:04:47.880 --> 00:04:51.329
You're going to draw the fruits from the sack one at a time.
00:04:51.420 --> 00:04:54.360
And let's say the first fruit you bring out is a lemon.
00:04:56.550 --> 00:04:56.670
Okay!
00:04:56.550 --> 00:04:56.670
Okay?
00:04:56.550 --> 00:04:58.589
And we placed that somewhere on the table.
00:04:59.430 --> 00:05:10.110
Now, as we draw out more fruit from the sack, what we going to try and do is sort it, according to how similar it is to the fruit that we've already got out on the table.
00:05:10.189 --> 00:05:10.319
All right.
00:05:10.321 --> 00:05:13.709
So let's say the next fruit you draw out is a mango.
00:05:14.189 --> 00:05:14.449
Okay?
00:05:14.480 --> 00:05:14.579
Right.
00:05:14.581 --> 00:05:24.329
So that's not got much in common with a lemon, you know, it's not citrus, it's not very kind of acidic, um, different types of seeds, that kind of thing, different size.
00:05:24.600 --> 00:05:28.050
So we might position that somewhere, very different on the table in front of us.
00:05:28.050 --> 00:05:28.050
Okay.
00:05:28.050 --> 00:05:28.050
My mango and lemon are far apart!
00:05:31.529 --> 00:05:31.920
Exactly.
00:05:32.040 --> 00:05:32.040
Right.
00:05:32.050 --> 00:05:35.009
You reached back into the bag that this time you pull out a lime.
00:05:35.339 --> 00:05:39.750
Okay, alright, so it's not exactly the same as the lemon, but it's quite similar, right?
00:05:40.110 --> 00:05:44.220
It's, you know, acidity, it's quite similar size, similar seeds,and so on.
00:05:44.221 --> 00:05:47.579
And so you might place that pretty close to the lemon.
00:05:47.581 --> 00:05:53.670
So you're kind of sorting them together, but you reach back into your sack this time you pull out a raspberry, right?
00:05:53.670 --> 00:05:58.939
Well, that's, um, neither much like citrus fruits nor like a mango.
00:05:58.940 --> 00:06:00.709
So you might position that somewhere else.
00:06:00.711 --> 00:06:11.449
And you imagine that you can keep this process going, until you've emptied the sack and you got all the fruits out in front of you, and they are kind of organized based upon how similar they are to each other.
00:06:11.899 --> 00:06:28.879
And then you can imagine, you could look at this display and then think"is there any organizational principle that I can easily detect?" So for instance, are their obvious gaps in the table where I've got no fruit, which would suggest that there's clear separation between two different groups of fruits, either sides of that barrier?
00:06:29.000 --> 00:06:36.529
Um, it's a bit of, kind of an odd way of describing it, but in a way, a network analysis does something very similar to that.
00:06:36.889 --> 00:06:44.149
It's it kind of sort information, or pieces of information, a little bit like the fruit, based upon how similar they are.
00:06:44.151 --> 00:07:15.769
So imagine that it's not fruit in the bag with, um, children[laughs] and executive functioning behaviours[laughs], and you kind of draw them out one at a time and you say, well, what's their profile of executive function behaviours on things like inattention, hyperactivity, um, problem-solving, and you organize them to see whether there are any little communities of children who have very similar profiles, a bit like the lemons and the limes, and they're different from say the children elsewhere, who are a bit more like the mangoes.
00:07:15.771 --> 00:07:26.360
And that in essence is how a network analysis works that uses profiles and information to sort individuals, um, in space.
00:07:26.360 --> 00:07:28.610
And that's essentially what we did.
00:07:28.610 --> 00:07:29.750
Does that make any sense whatsoever?
00:07:29.750 --> 00:07:30.139
[Laughs]
00:07:31.089 --> 00:07:39.459
Although it makes perfect sense, but I'm glad you've clarified that there were children in the sack, and it's not[laughs], and not um...
00:07:39.819 --> 00:07:44.410
Not sort of, um, uh, constructs within the measures, right?
00:07:44.411 --> 00:07:53.529
Cause I could imagine you could also do that kind of clustering or type analysis where you're looking at what scores on different tests go together, for example.
00:07:52.961 --> 00:07:53.529
Yeah, exactly, yeah.
00:07:55.240 --> 00:08:02.949
You're grouping the children together based on the information that you have from the parent ratings that you've got, is that right?
00:08:03.310 --> 00:08:04.079
Absolutely right.
00:08:04.129 --> 00:08:04.870
That's a really good point.
00:08:04.870 --> 00:08:08.920
So you can use this kind of analysis to look at all sorts of things.
00:08:09.029 --> 00:08:15.759
And you know, one thing you could look at is how measures group together, but we weren't doing that, we were looking at how children group together.
00:08:16.899 --> 00:08:19.209
And, and so one more question about it.
00:08:19.211 --> 00:08:23.379
This is a very, um, naive question, but does the, the...
00:08:23.860 --> 00:08:41.350
So in your metaphor you pull out a lemon first, does whoever gets pulled out of the sack sort of unduly influence all the patterns that follow, or does the process sort of, is it resilient to that or just less sequential than that, that metaphor in my head?
00:08:41.710 --> 00:08:41.980
Yeah.
00:08:42.009 --> 00:08:42.340
It's...
00:08:42.370 --> 00:08:45.340
The reality is less sequential than that metaphor.
00:08:45.700 --> 00:08:45.909
Right.
00:08:46.330 --> 00:08:51.940
So in a way, imagine it more like, um, it somehow knows some basic properties on the table by which to organise things.
00:08:52.059 --> 00:08:59.370
Um, but there's still, um, there's still a process that has to go through to check, to see whether there are any boundaries on the table.
00:08:59.370 --> 00:09:03.480
So you might be thinking about, there might be some gaps in the table where you've got no fruit.
00:09:03.629 --> 00:09:09.210
Um, like you might have some gaps in the data where there are no kids, implying that there's a kind of clean separation between two groups.
00:09:09.690 --> 00:09:26.250
Um, you can then, uh, in order to work out where those gaps are it does run an iterative process, and the order in which it does that is important, um, as it can lead to subtly different results each time as to where it puts the-, put-, you know, puts the boundary.
00:09:27.000 --> 00:09:40.080
So because of those sorts of things, what people do is they will run it sort of hundreds or thousands of times, and then they can integrate their findings across those, um, so that they're never relatively insensitive...
00:09:40.559 --> 00:09:40.679
Right.
00:09:40.559 --> 00:09:40.830
...to the order of things.
00:09:40.831 --> 00:09:41.340
But you're right.
00:09:41.341 --> 00:09:46.799
In theory it could do it but there's, there's very careful checks involved to make sure that it's not.
00:09:47.809 --> 00:09:48.169
Sure.
00:09:48.171 --> 00:09:51.230
So the fruit always ends up in the same group, basically.
00:09:52.820 --> 00:09:53.210
Yeah, exactly.
00:09:52.820 --> 00:09:53.210
Um...
00:09:53.690 --> 00:09:54.120
So...
00:09:54.590 --> 00:10:00.080
So, uh, going back to, um, your kind of headline finding, what-...
00:10:00.889 --> 00:10:05.480
remind us what that kind of big finding is, and especially what you think we can learn from that.
00:10:06.409 --> 00:10:15.049
Well, as we looked out across the table of children across the kind of the network analysis of children, what we found was that there were three groups that we could distinguish.
00:10:15.769 --> 00:10:25.429
That's not to say that the kids are all identical within each of those groups, but that the analysis was able to statistically distinguish these three different communities within the network.
00:10:25.460 --> 00:10:32.840
And there's one community of kids who have elevated inattention, hyperactivity, and impulsivity difficulties.
00:10:34.399 --> 00:10:41.929
And they are somewhat distinct from a group of children where their caregivers described them as having elevated learning difficulties.
00:10:42.769 --> 00:10:52.009
And they are relatively distinct from the group of children where their caregivers described them as having"elevated difficulties with aggressive behaviour and peer relationships".
00:10:53.509 --> 00:11:04.490
Um, and so that was the first kind of important finding, was that the, even though we haven't told the network that there'd be different communities, it was able to find these three different groups.
00:11:05.509 --> 00:11:18.470
Um, and there was one kind of more of your classic executive function difficulties, there was one more specific to learning difficulties in school, and then there was one more specific to executive function behaviours in social contexts.
00:11:19.549 --> 00:11:23.360
Um, and the network was able to distinguish those three different groups, statistically.
00:11:24.710 --> 00:11:27.200
And then who were those children then?
00:11:27.230 --> 00:11:35.720
So you hinted that the groups that the network detected didn't map onto the other ways that we might group those children.
00:11:35.721 --> 00:11:36.320
Is that correct?
00:11:37.129 --> 00:11:38.149
Yeah, that's correct.
00:11:38.600 --> 00:11:49.580
So children with an ADHD diagnosis were slightly more likely to appear in the first group, the children with elevated inattention, hyperactivity, and impulsivity difficulties.
00:11:48.980 --> 00:12:01.120
But kids with an ADHD diagnosis were very likely be found in the other groups as well![laughs] And there were lots of other kids with other types of diagnosis, like autism, who were also likely to be found in that first group.
00:12:01.450 --> 00:12:06.879
So knowing the diagnosis didn't really distinguish the kind of group that the children could be found in.
00:12:07.980 --> 00:12:09.809
And that's especially interesting, isn't it?
00:12:09.811 --> 00:12:11.669
Because it was parent report data.
00:12:11.730 --> 00:12:24.779
So these parents also know their children's diagnosis, and you might, if anything, expect that would lead them to rate their children's behaviour in a slightly more stereotypical way.
00:12:25.649 --> 00:12:35.309
Um, and yet their ratings aren't mapping onto the sort of, um, cliched or the, or the typical representations we have of those diagnostic groups.
00:12:35.909 --> 00:12:37.169
Yeah, I think that's because...
00:12:37.649 --> 00:12:52.679
I think parents are sort of, well they, you know, they live with these kids and so they're already wise to the, um, they're already wise to the fact that their children's behaviour may not fit into these sorts of neat, um, descriptions.
00:12:53.100 --> 00:13:01.830
And so they're, I think, just more willing just to describe exactly what life is like living with these kids, rather than sort of sticking to any kind of criteria.
00:13:02.759 --> 00:13:02.759
Hmm.
00:13:03.600 --> 00:13:06.570
And so what, what should we be doing with that information?
00:13:06.600 --> 00:13:15.120
What might, um, you know, a young person with a diagnosis of ADHD or who's struggling with their learning, do with that information?
00:13:15.120 --> 00:13:19.049
Or what might a teacher or a parent learn from this information, do you think?
00:13:20.340 --> 00:13:21.269
It's a very good question.
00:13:21.539 --> 00:13:48.120
I think that the first thing to say is that if you're a teacher, for example, and you had a kid arriving at your class and you, you knew that they had a diagnosis of some kind in advance, then sort of hold that diagnosis relatively lightly because the child that you actually encounter when they arrive could, could, um, just have all sorts of different kinds of, um, behavioural difficulties, and may not have the behavioural difficulties that you expect to come with that kind of diagnosis.
00:13:48.659 --> 00:13:58.919
So I guess that's one first important lesson, which is to hold the diagnosis relatively lightly because it may not do a good job of characterizing the child that you have in front of you.
00:14:00.210 --> 00:14:02.340
Um, I think in terms of, with...
00:14:02.039 --> 00:14:13.169
And I think probably something similar goes for parents, which I guess they probably know already, which is that at the point in which you receive the diagnosis, the mind probably races to, you know,"what can I expect for the future?".
00:14:13.740 --> 00:14:23.100
But the reality is that the-, the diagnosis that's been given often doesn't provide tremendously firm sort of prognostic or forward-thinking information.
00:14:24.120 --> 00:14:30.539
Um, and as a result, the kind of behaviours that you observe might not really fit the pattern that you were expecting.
00:14:30.809 --> 00:14:31.169
And...
00:14:31.830 --> 00:14:32.879
Is that a kind of thing you were thinking, Sue?
00:14:34.080 --> 00:14:35.909
Yeah, absolutely! I mean, I think it fits...
00:14:36.210 --> 00:14:41.639
This is a conversation I have a lot with people, is what are the value of diagnostic labels?
00:14:42.000 --> 00:14:47.370
You know, are they, did they primarily give you an advantage or can they become a burden and so on.
00:14:47.370 --> 00:14:57.139
And, and the thing I returned to is that the diagnostic can be very useful in terms of sort of gaining access to services, and as a...
00:14:57.490 --> 00:15:11.120
a sort of prominent badge that says"this child may need a little bit more of your support, or your patience, or your accommodation", but any further than that, you just have to know the individual child, right?
00:15:11.120 --> 00:15:18.889
The label's not going to take you the whole way, and it's not going to give you-, it doesn't come with a whole package of information and support recommendations.
00:15:18.919 --> 00:15:23.059
Would you say that sort of aligns with what you're finding in this study?
00:15:24.250 --> 00:15:24.580
I think so.
00:15:24.370 --> 00:15:25.570
And also that, um...
00:15:25.571 --> 00:15:50.080
I guess something we saw in the CALM cohort overall was that executive function behaviours were sort of elevated across the board as in pretty much most kids that we saw in that community sample, their parents and care givers described all of them as having elevated, um, elevated difficulties in one, in one area or another.
00:15:50.081 --> 00:15:58.870
And so I guess in a way t hey just seem to me, it was just a very common feature of lots of kids that come to the attention of practitioners.
00:16:01.360 --> 00:16:01.750
Mmh, mmh...
00:16:01.360 --> 00:16:12.399
Um, the other thing I think is interesting is, is what does it mean for researchers and scientists who are trying to think of sort of underlying mechanisms and so on.
00:16:12.700 --> 00:16:22.120
And something we have found is that these sorts of data-driven groupings, well that's not perfect by any means, but they do tend to be closer to underlying mechanisms.
00:16:22.120 --> 00:16:35.710
So for example, we use a lot of brain scanning in our lab, and we find that these sorts of data-driven groupings are much more closely aligned with underlying differences say brain structure, versus if we were to use the diagnostic groups, for example.
00:16:37.570 --> 00:16:37.750
Mmh...
00:16:37.570 --> 00:16:54.370
And that's presumably because using the data to do some kind of data-driven grouping in advance, um, reflects something about this closer to the individual profile, whereas the diagnostic categories are always going to be a much sort of coarser way of defining individuals.
00:16:55.529 --> 00:16:57.909
Well it sounds like you might be starting a revolution Duncan.
00:17:05.529 --> 00:17:05.920
[Laughs]
00:17:05.529 --> 00:17:29.170
Um, we should draw to a close, but before we finish our PsychologiCALL, um, I wondered if you had anything that you wanted to say to any early career researchers or students who might be listening, who are perhaps, um, a bit cut off from their normal learning and their normal research environment, um, any, uh, words of wisdom for them?
00:17:29.769 --> 00:17:45.940
Well, this whole analysis really came about because of the ideas of an early career researcher and my, in my experience, some of the most exciting things that I've been involved in has been ideas that came from early career researchers.
00:17:46.990 --> 00:17:49.119
Um, and they often think about things in different ways.
00:17:49.170 --> 00:18:04.230
And this study for-, the reason I chose it is because it sort of opened my way of thinking up to a kind of very different way of viewing data and thinking about how we can use the data to understand something about the individuals that you've seen.
00:18:05.039 --> 00:18:07.200
And once we kind of started down that path, we...
00:18:07.930 --> 00:18:13.349
In the end we used, we've developed a whole suite of different techniques to ask kind of similar questions.
00:18:14.369 --> 00:18:18.450
Um, but this was the first one and it all came from an idea from an early career researcher.
00:18:18.779 --> 00:18:32.250
So I guess my advice is sort of, don't be afraid to think somewhat outside the box, um, and to lay assumptions and suppose there's sort of, um, ways of doing things that you have to stick with.
00:18:32.609 --> 00:18:34.950
Don't be afraid of putting all those things to one side.
00:18:35.279 --> 00:18:42.720
Then maybe during this lockdown period, it's quite a nice time actually, to get a bit of time to think about how things could be done differently.
00:18:42.721 --> 00:18:48.569
If we didn't have to carry with us all these assumptions about how we do things, how we study children and how we categorize them.
00:18:48.570 --> 00:18:49.079
And so on.
00:18:51.019 --> 00:19:02.900
That's amazing! I think a very inspirational word and a big shout out to Joe Bathelt out for coming up with such a cool idea! Um, we should probably wrap it up there.
00:19:02.901 --> 00:19:05.420
Thank you very much for giving me your time Duncan!
00:19:05.539 --> 00:19:07.130
My pleasure! Thank you for asking!
00:19:08.329 --> 00:19:14.599
And thank you to everyone who is listening to this, and you can find out more about the work that Duncan's talked about.
00:19:15.079 --> 00:19:20.390
There will be links on the podcast page at ed.ac.uk/salvesen- research.
00:19:23.119 --> 00:19:31.819
Thank you very much then! Bye![ringtone] Okay we did it! I thought that went quite smoothly![Podcast jingle]
00:00:09.150 --> 00:00:09.150
[Podcast jingle][ringtone] Hello?
00:00:09.150 --> 00:00:09.750
Oh, it is recording.
00:00:09.750 --> 00:00:10.679
I see the little figure.
00:00:10.740 --> 00:00:11.169
Okay, great.
00:00:11.369 --> 00:00:14.189
I will do my little spiel and then I'll introduce you.
00:00:14.310 --> 00:00:14.699
Nice.
00:00:14.910 --> 00:00:15.240
Okay.
00:00:15.630 --> 00:00:16.079
Here I go.
00:00:18.260 --> 00:00:18.649
Hello.
00:00:18.739 --> 00:00:23.210
I'm Sue from the Salvesen Mindroom Research Centre at the University of Edinburgh.
00:00:23.660 --> 00:00:34.340
We wanted to do something for all the practitioners and parents and students who are stuck at home, and who still wants to be developing their knowledge, even though it's a lockdown at the moment in the UK.
00:00:35.420 --> 00:00:41.929
Um, and also cause lots of people are homeschooling their kids at the moment and their usual family routines have been disrupted.
00:00:42.259 --> 00:00:47.060
It seems like developmental psychology is a really interesting thing to be talking about at the moment.
00:00:47.659 --> 00:00:53.990
So I'm doing these short phone calls with psychologists who study learning and development in children and young people.
00:00:54.649 --> 00:01:01.429
And today's PsychologiCALL is with my very dear friend, Duncan Astle from the University of Cambridge.
00:01:01.850 --> 00:01:08.269
And he's going to talk to me about some work he's done on executive function in a really interesting community sample.
00:01:09.049 --> 00:01:09.739
Hello Duncan!
00:01:10.969 --> 00:01:11.930
Hi Sue! Thanks for having me!
00:01:12.859 --> 00:01:16.609
[Laughs] Thank you for coming on my thing that's not quite a podcast.
00:01:16.049 --> 00:01:16.609
[Laughs]
00:01:19.909 --> 00:01:22.459
So, tell me why did you pick this paper?
00:01:22.460 --> 00:01:24.709
What did you discover that you want to talk about today?
00:01:25.939 --> 00:01:46.790
So this paper is all about executive function behaviours in children, and we found that parents ratings of executive function behaviours fall broadly into three different types, and those different types of behaviours don't really fit with the kinds of diagnoses that children often have.
00:01:48.680 --> 00:02:01.099
Um, that is very interesting! So are you, sorry, I'm jumping ahead to the conclusions that you concluding from that, that, um, parents don't have a good grasp on their kid's executive functions?
00:02:01.159 --> 00:02:06.349
Or that the way that we categorized things previously is maybe not as good as we thought it was?
00:02:07.069 --> 00:02:08.300
Option number two.
00:02:08.419 --> 00:02:28.039
So we infer from that the parents are actually good at rating, um, behaviours, but that the way that the kinds of categories that the children might come with, when we study them in our research studies, might not be that closely related to the actual behaviours that we were able to observe.
00:02:29.240 --> 00:02:29.569
Great.
00:02:30.590 --> 00:02:33.139
Um, so what, what motivated you to study this?
00:02:33.140 --> 00:02:35.599
Why did you decide this was an interesting thing to look at?
00:02:36.110 --> 00:02:36.409
So.
00:02:36.411 --> 00:02:37.099
Good question.
00:02:37.159 --> 00:02:57.199
So the background is that, um, in Cambridge, myself, Joni Holmes, Sue Gathercole, and a whole team of others has collected this really interesting community cohort of children who have been identified by practitioners as struggling in the areas of attention, learning, and memory.
00:02:57.200 --> 00:03:02.620
And they could have one diagnosis, multiple diagnoses, or no diagnosis at all.
00:03:02.889 --> 00:03:10.210
The key thing is that at some point some professional had said that this child is struggling and they had referred them to us study.
00:03:10.870 --> 00:03:20.979
And one thing that we observed quite early on is that the ratings of executive function related behaviour problems were particularly high in the cohort.
00:03:20.980 --> 00:03:35.289
So for instance, um, lots of, um, caregivers would say that the children struggled to maintain attention, um, symptoms like, um, hyper activity were quite common, um, and so that is kind of important context.
00:03:35.379 --> 00:03:37.750
So we've collected this large group of kids.
00:03:38.150 --> 00:03:44.050
We had lots of ratings of executive functioning behaviours, and they tended to be pretty high.
00:03:44.560 --> 00:03:50.110
And so we started to think about how we might look and analyze those kinds of behaviours.
00:03:49.661 --> 00:03:54.039
And that really kind of set the scene, um, for what we went on to do.
00:03:55.289 --> 00:03:58.139
And so tell me a bit then about the analysis.
00:03:58.740 --> 00:04:01.710
Um, did you use very complicated, clever techniques?
00:04:02.189 --> 00:04:03.990
Is it something you can describe to us?
00:04:04.889 --> 00:04:07.409
I think so.
00:04:07.411 --> 00:04:15.599
So, um, it's actually really all led by a post-doc who was working with me called Joe Bathelt, who now has his own lab at Royal Holloway, which is very exciting, anyway.
00:04:15.600 --> 00:04:20.970
And he came to me and he said"I think that we should do something called a network analysis".
00:04:20.971 --> 00:04:28.800
And this was about three or four years ago, when not many people were using this kind of technique and I didn't really know much about it.
00:04:28.800 --> 00:04:32.939
U m, and so he had to kind of explain it to me, and I will try and explain it to you now.
00:04:33.060 --> 00:04:39.209
So at the start I want you to imagine that you put a sack, a brown sack of some sort in your lap, and inside it, you've got fruits.
00:04:39.329 --> 00:04:39.930
Okay?
00:04:40.470 --> 00:04:42.990
I've got a...I've got a sack of fruit in my lap!
00:04:46.920 --> 00:04:46.920
Okay[laughs], and you've got a clear table in front of you...
00:04:46.920 --> 00:04:46.920
Yeah!
00:04:47.519 --> 00:04:47.790
Okay.
00:04:47.880 --> 00:04:51.329
You're going to draw the fruits from the sack one at a time.
00:04:51.420 --> 00:04:54.360
And let's say the first fruit you bring out is a lemon.
00:04:56.550 --> 00:04:56.670
Okay!
00:04:56.550 --> 00:04:56.670
Okay?
00:04:56.550 --> 00:04:58.589
And we placed that somewhere on the table.
00:04:59.430 --> 00:05:10.110
Now, as we draw out more fruit from the sack, what we going to try and do is sort it, according to how similar it is to the fruit that we've already got out on the table.
00:05:10.189 --> 00:05:10.319
All right.
00:05:10.321 --> 00:05:13.709
So let's say the next fruit you draw out is a mango.
00:05:14.189 --> 00:05:14.449
Okay?
00:05:14.480 --> 00:05:14.579
Right.
00:05:14.581 --> 00:05:24.329
So that's not got much in common with a lemon, you know, it's not citrus, it's not very kind of acidic, um, different types of seeds, that kind of thing, different size.
00:05:24.600 --> 00:05:28.050
So we might position that somewhere, very different on the table in front of us.
00:05:28.050 --> 00:05:28.050
Okay.
00:05:28.050 --> 00:05:28.050
My mango and lemon are far apart!
00:05:31.529 --> 00:05:31.920
Exactly.
00:05:32.040 --> 00:05:32.040
Right.
00:05:32.050 --> 00:05:35.009
You reached back into the bag that this time you pull out a lime.
00:05:35.339 --> 00:05:39.750
Okay, alright, so it's not exactly the same as the lemon, but it's quite similar, right?
00:05:40.110 --> 00:05:44.220
It's, you know, acidity, it's quite similar size, similar seeds,and so on.
00:05:44.221 --> 00:05:47.579
And so you might place that pretty close to the lemon.
00:05:47.581 --> 00:05:53.670
So you're kind of sorting them together, but you reach back into your sack this time you pull out a raspberry, right?
00:05:53.670 --> 00:05:58.939
Well, that's, um, neither much like citrus fruits nor like a mango.
00:05:58.940 --> 00:06:00.709
So you might position that somewhere else.
00:06:00.711 --> 00:06:11.449
And you imagine that you can keep this process going, until you've emptied the sack and you got all the fruits out in front of you, and they are kind of organized based upon how similar they are to each other.
00:06:11.899 --> 00:06:28.879
And then you can imagine, you could look at this display and then think"is there any organizational principle that I can easily detect?" So for instance, are their obvious gaps in the table where I've got no fruit, which would suggest that there's clear separation between two different groups of fruits, either sides of that barrier?
00:06:29.000 --> 00:06:36.529
Um, it's a bit of, kind of an odd way of describing it, but in a way, a network analysis does something very similar to that.
00:06:36.889 --> 00:06:44.149
It's it kind of sort information, or pieces of information, a little bit like the fruit, based upon how similar they are.
00:06:44.151 --> 00:07:15.769
So imagine that it's not fruit in the bag with, um, children[laughs] and executive functioning behaviours[laughs], and you kind of draw them out one at a time and you say, well, what's their profile of executive function behaviours on things like inattention, hyperactivity, um, problem-solving, and you organize them to see whether there are any little communities of children who have very similar profiles, a bit like the lemons and the limes, and they're different from say the children elsewhere, who are a bit more like the mangoes.
00:07:15.771 --> 00:07:26.360
And that in essence is how a network analysis works that uses profiles and information to sort individuals, um, in space.
00:07:26.360 --> 00:07:28.610
And that's essentially what we did.
00:07:28.610 --> 00:07:29.750
Does that make any sense whatsoever?
00:07:29.750 --> 00:07:30.139
[Laughs]
00:07:31.089 --> 00:07:39.459
Although it makes perfect sense, but I'm glad you've clarified that there were children in the sack, and it's not[laughs], and not um...
00:07:39.819 --> 00:07:44.410
Not sort of, um, uh, constructs within the measures, right?
00:07:44.411 --> 00:07:53.529
Cause I could imagine you could also do that kind of clustering or type analysis where you're looking at what scores on different tests go together, for example.
00:07:52.961 --> 00:07:53.529
Yeah, exactly, yeah.
00:07:55.240 --> 00:08:02.949
You're grouping the children together based on the information that you have from the parent ratings that you've got, is that right?
00:08:03.310 --> 00:08:04.079
Absolutely right.
00:08:04.129 --> 00:08:04.870
That's a really good point.
00:08:04.870 --> 00:08:08.920
So you can use this kind of analysis to look at all sorts of things.
00:08:09.029 --> 00:08:15.759
And you know, one thing you could look at is how measures group together, but we weren't doing that, we were looking at how children group together.
00:08:16.899 --> 00:08:19.209
And, and so one more question about it.
00:08:19.211 --> 00:08:23.379
This is a very, um, naive question, but does the, the...
00:08:23.860 --> 00:08:41.350
So in your metaphor you pull out a lemon first, does whoever gets pulled out of the sack sort of unduly influence all the patterns that follow, or does the process sort of, is it resilient to that or just less sequential than that, that metaphor in my head?
00:08:41.710 --> 00:08:41.980
Yeah.
00:08:42.009 --> 00:08:42.340
It's...
00:08:42.370 --> 00:08:45.340
The reality is less sequential than that metaphor.
00:08:45.700 --> 00:08:45.909
Right.
00:08:46.330 --> 00:08:51.940
So in a way, imagine it more like, um, it somehow knows some basic properties on the table by which to organise things.
00:08:52.059 --> 00:08:59.370
Um, but there's still, um, there's still a process that has to go through to check, to see whether there are any boundaries on the table.
00:08:59.370 --> 00:09:03.480
So you might be thinking about, there might be some gaps in the table where you've got no fruit.
00:09:03.629 --> 00:09:09.210
Um, like you might have some gaps in the data where there are no kids, implying that there's a kind of clean separation between two groups.
00:09:09.690 --> 00:09:26.250
Um, you can then, uh, in order to work out where those gaps are it does run an iterative process, and the order in which it does that is important, um, as it can lead to subtly different results each time as to where it puts the-, put-, you know, puts the boundary.
00:09:27.000 --> 00:09:40.080
So because of those sorts of things, what people do is they will run it sort of hundreds or thousands of times, and then they can integrate their findings across those, um, so that they're never relatively insensitive...
00:09:40.559 --> 00:09:40.679
Right.
00:09:40.559 --> 00:09:40.830
...to the order of things.
00:09:40.831 --> 00:09:41.340
But you're right.
00:09:41.341 --> 00:09:46.799
In theory it could do it but there's, there's very careful checks involved to make sure that it's not.
00:09:47.809 --> 00:09:48.169
Sure.
00:09:48.171 --> 00:09:51.230
So the fruit always ends up in the same group, basically.
00:09:52.820 --> 00:09:53.210
Yeah, exactly.
00:09:52.820 --> 00:09:53.210
Um...
00:09:53.690 --> 00:09:54.120
So...
00:09:54.590 --> 00:10:00.080
So, uh, going back to, um, your kind of headline finding, what-...
00:10:00.889 --> 00:10:05.480
remind us what that kind of big finding is, and especially what you think we can learn from that.
00:10:06.409 --> 00:10:15.049
Well, as we looked out across the table of children across the kind of the network analysis of children, what we found was that there were three groups that we could distinguish.
00:10:15.769 --> 00:10:25.429
That's not to say that the kids are all identical within each of those groups, but that the analysis was able to statistically distinguish these three different communities within the network.
00:10:25.460 --> 00:10:32.840
And there's one community of kids who have elevated inattention, hyperactivity, and impulsivity difficulties.
00:10:34.399 --> 00:10:41.929
And they are somewhat distinct from a group of children where their caregivers described them as having elevated learning difficulties.
00:10:42.769 --> 00:10:52.009
And they are relatively distinct from the group of children where their caregivers described them as having"elevated difficulties with aggressive behaviour and peer relationships".
00:10:53.509 --> 00:11:04.490
Um, and so that was the first kind of important finding, was that the, even though we haven't told the network that there'd be different communities, it was able to find these three different groups.
00:11:05.509 --> 00:11:18.470
Um, and there was one kind of more of your classic executive function difficulties, there was one more specific to learning difficulties in school, and then there was one more specific to executive function behaviours in social contexts.
00:11:19.549 --> 00:11:23.360
Um, and the network was able to distinguish those three different groups, statistically.
00:11:24.710 --> 00:11:27.200
And then who were those children then?
00:11:27.230 --> 00:11:35.720
So you hinted that the groups that the network detected didn't map onto the other ways that we might group those children.
00:11:35.721 --> 00:11:36.320
Is that correct?
00:11:37.129 --> 00:11:38.149
Yeah, that's correct.
00:11:38.600 --> 00:11:49.580
So children with an ADHD diagnosis were slightly more likely to appear in the first group, the children with elevated inattention, hyperactivity, and impulsivity difficulties.
00:11:48.980 --> 00:12:01.120
But kids with an ADHD diagnosis were very likely be found in the other groups as well![laughs] And there were lots of other kids with other types of diagnosis, like autism, who were also likely to be found in that first group.
00:12:01.450 --> 00:12:06.879
So knowing the diagnosis didn't really distinguish the kind of group that the children could be found in.
00:12:07.980 --> 00:12:09.809
And that's especially interesting, isn't it?
00:12:09.811 --> 00:12:11.669
Because it was parent report data.
00:12:11.730 --> 00:12:24.779
So these parents also know their children's diagnosis, and you might, if anything, expect that would lead them to rate their children's behaviour in a slightly more stereotypical way.
00:12:25.649 --> 00:12:35.309
Um, and yet their ratings aren't mapping onto the sort of, um, cliched or the, or the typical representations we have of those diagnostic groups.
00:12:35.909 --> 00:12:37.169
Yeah, I think that's because...
00:12:37.649 --> 00:12:52.679
I think parents are sort of, well they, you know, they live with these kids and so they're already wise to the, um, they're already wise to the fact that their children's behaviour may not fit into these sorts of neat, um, descriptions.
00:12:53.100 --> 00:13:01.830
And so they're, I think, just more willing just to describe exactly what life is like living with these kids, rather than sort of sticking to any kind of criteria.
00:13:02.759 --> 00:13:02.759
Hmm.
00:13:03.600 --> 00:13:06.570
And so what, what should we be doing with that information?
00:13:06.600 --> 00:13:15.120
What might, um, you know, a young person with a diagnosis of ADHD or who's struggling with their learning, do with that information?
00:13:15.120 --> 00:13:19.049
Or what might a teacher or a parent learn from this information, do you think?
00:13:20.340 --> 00:13:21.269
It's a very good question.
00:13:21.539 --> 00:13:48.120
I think that the first thing to say is that if you're a teacher, for example, and you had a kid arriving at your class and you, you knew that they had a diagnosis of some kind in advance, then sort of hold that diagnosis relatively lightly because the child that you actually encounter when they arrive could, could, um, just have all sorts of different kinds of, um, behavioural difficulties, and may not have the behavioural difficulties that you expect to come with that kind of diagnosis.
00:13:48.659 --> 00:13:58.919
So I guess that's one first important lesson, which is to hold the diagnosis relatively lightly because it may not do a good job of characterizing the child that you have in front of you.
00:14:00.210 --> 00:14:02.340
Um, I think in terms of, with...
00:14:02.039 --> 00:14:13.169
And I think probably something similar goes for parents, which I guess they probably know already, which is that at the point in which you receive the diagnosis, the mind probably races to, you know,"what can I expect for the future?".
00:14:13.740 --> 00:14:23.100
But the reality is that the-, the diagnosis that's been given often doesn't provide tremendously firm sort of prognostic or forward-thinking information.
00:14:24.120 --> 00:14:30.539
Um, and as a result, the kind of behaviours that you observe might not really fit the pattern that you were expecting.
00:14:30.809 --> 00:14:31.169
And...
00:14:31.830 --> 00:14:32.879
Is that a kind of thing you were thinking, Sue?
00:14:34.080 --> 00:14:35.909
Yeah, absolutely! I mean, I think it fits...
00:14:36.210 --> 00:14:41.639
This is a conversation I have a lot with people, is what are the value of diagnostic labels?
00:14:42.000 --> 00:14:47.370
You know, are they, did they primarily give you an advantage or can they become a burden and so on.
00:14:47.370 --> 00:14:57.139
And, and the thing I returned to is that the diagnostic can be very useful in terms of sort of gaining access to services, and as a...
00:14:57.490 --> 00:15:11.120
a sort of prominent badge that says"this child may need a little bit more of your support, or your patience, or your accommodation", but any further than that, you just have to know the individual child, right?
00:15:11.120 --> 00:15:18.889
The label's not going to take you the whole way, and it's not going to give you-, it doesn't come with a whole package of information and support recommendations.
00:15:18.919 --> 00:15:23.059
Would you say that sort of aligns with what you're finding in this study?
00:15:24.250 --> 00:15:24.580
I think so.
00:15:24.370 --> 00:15:25.570
And also that, um...
00:15:25.571 --> 00:15:50.080
I guess something we saw in the CALM cohort overall was that executive function behaviours were sort of elevated across the board as in pretty much most kids that we saw in that community sample, their parents and care givers described all of them as having elevated, um, elevated difficulties in one, in one area or another.
00:15:50.081 --> 00:15:58.870
And so I guess in a way t hey just seem to me, it was just a very common feature of lots of kids that come to the attention of practitioners.
00:16:01.360 --> 00:16:01.750
Mmh, mmh...
00:16:01.360 --> 00:16:12.399
Um, the other thing I think is interesting is, is what does it mean for researchers and scientists who are trying to think of sort of underlying mechanisms and so on.
00:16:12.700 --> 00:16:22.120
And something we have found is that these sorts of data-driven groupings, well that's not perfect by any means, but they do tend to be closer to underlying mechanisms.
00:16:22.120 --> 00:16:35.710
So for example, we use a lot of brain scanning in our lab, and we find that these sorts of data-driven groupings are much more closely aligned with underlying differences say brain structure, versus if we were to use the diagnostic groups, for example.
00:16:37.570 --> 00:16:37.750
Mmh...
00:16:37.570 --> 00:16:54.370
And that's presumably because using the data to do some kind of data-driven grouping in advance, um, reflects something about this closer to the individual profile, whereas the diagnostic categories are always going to be a much sort of coarser way of defining individuals.
00:16:55.529 --> 00:16:57.909
Well it sounds like you might be starting a revolution Duncan.
00:17:05.529 --> 00:17:05.920
[Laughs]
00:17:05.529 --> 00:17:29.170
Um, we should draw to a close, but before we finish our PsychologiCALL, um, I wondered if you had anything that you wanted to say to any early career researchers or students who might be listening, who are perhaps, um, a bit cut off from their normal learning and their normal research environment, um, any, uh, words of wisdom for them?
00:17:29.769 --> 00:17:45.940
Well, this whole analysis really came about because of the ideas of an early career researcher and my, in my experience, some of the most exciting things that I've been involved in has been ideas that came from early career researchers.
00:17:46.990 --> 00:17:49.119
Um, and they often think about things in different ways.
00:17:49.170 --> 00:18:04.230
And this study for-, the reason I chose it is because it sort of opened my way of thinking up to a kind of very different way of viewing data and thinking about how we can use the data to understand something about the individuals that you've seen.
00:18:05.039 --> 00:18:07.200
And once we kind of started down that path, we...
00:18:07.930 --> 00:18:13.349
In the end we used, we've developed a whole suite of different techniques to ask kind of similar questions.
00:18:14.369 --> 00:18:18.450
Um, but this was the first one and it all came from an idea from an early career researcher.
00:18:18.779 --> 00:18:32.250
So I guess my advice is sort of, don't be afraid to think somewhat outside the box, um, and to lay assumptions and suppose there's sort of, um, ways of doing things that you have to stick with.
00:18:32.609 --> 00:18:34.950
Don't be afraid of putting all those things to one side.
00:18:35.279 --> 00:18:42.720
Then maybe during this lockdown period, it's quite a nice time actually, to get a bit of time to think about how things could be done differently.
00:18:42.721 --> 00:18:48.569
If we didn't have to carry with us all these assumptions about how we do things, how we study children and how we categorize them.
00:18:48.570 --> 00:18:49.079
And so on.
00:18:51.019 --> 00:19:02.900
That's amazing! I think a very inspirational word and a big shout out to Joe Bathelt out for coming up with such a cool idea! Um, we should probably wrap it up there.
00:19:02.901 --> 00:19:05.420
Thank you very much for giving me your time Duncan!
00:19:05.539 --> 00:19:07.130
My pleasure! Thank you for asking!
00:19:08.329 --> 00:19:14.599
And thank you to everyone who is listening to this, and you can find out more about the work that Duncan's talked about.
00:19:15.079 --> 00:19:20.390
There will be links on the podcast page at ed.ac.uk/salvesen- research.
00:19:23.119 --> 00:19:31.819
Thank you very much then! Bye![ringtone] Okay we did it! I thought that went quite smoothly![Podcast jingle]
