The following is an automatically generated (and unchecked) transcription of this talk:
So hi, I’m Sam. And the title of this is understanding early self regulation, when does self regulation become dysregulation? So how are we defining self regulation? So we’re taking the definition from neg regulation as the ongoing dynamic and adaptive modulation of internal state and emotional cognition, or behavior mediated by central and peripheral physiology. And what are we regulating? So we’re talking about the behavioral regulation of autonomic arousal, defined from a composite of heart rate, and RSA. So that’s respiratory sinus arrhythmia, and movement. And now there’s a ton of stuff to say about that. There are some people that think of the autonomic nervous system very much in terms of differentiable fractional subsystems, whereas we’re treating it as a one dimensional construct, hear tons to say about that more detail on these reviews, link to the bottom of this page, or just get contacted if you’ve got any questions. But for now, we’re averaging heart rate, research science, a with me and movement for a single one dimensional composite measure of autonomic arousal. Okay? So self regulation in various forms, is related to lots of different psycho pathologies. I’m not going to be talking about that in more detail here. But just to flag it aren’t these references are also from Nick. And so today, I’m going to be talking about two findings I’ve described as weird. One is looking at a baby’s actually self regulate in real world settings, and second to self regulation and get worse between six months and 12 months. So two separate projects, two separate chunks of data that I’m looking at, that, for me, are some really quite fundamental questions about, you know, what it is that we really know about self regulation during early life already at the moment. Okay. So remember this definition that we started with. So I’ve just been writing a paper, saying that really in quite a fundamental way, the whilst I agree with this definition, I think it’s a very good definition. I think most of the ways that we in my lab, as well as lots of other people measure self regulation doesn’t really do a very good job of measuring it in the sense that it’s defined here. Yeah. So almost all experiments and tests of self regulation in infancy take the same basic approach. So you’ve got your baseline, and then you give at a time determined by the experiments, you administer a stressor, so something like the still face Protocol, or the toy removal or arm restraint as a whole variety of different ones. And then after a set period, which again, is determined by the experimenter, you stop the stressor, and you look at their recovery. Okay? So, there are lots of ways in which this model is a very poor fit for how we actually experienced stress in real life. And so in real life, the environment doesn’t deliver an isolated stimulus, and then patiently wait for one to admit an isolated response. That’s the Vivian de or put it 2006. In real life, or the response determines the stimulus, just as truly as vice versa. Yeah. So that’s from Julian, he noted that even the way back in 1896. So just to just to explain what I mean by these ideas, and what these other people are meaning by these quotes, let’s compare two examples of Ettore removal. Yep. So the first is, as I just described it in a lab battery, where the experimenter in a controlled setting, the experimenter administers this one single stressor. Yeah, wait for two minutes. And then from a stopwatch removes the stressor after two minutes. Yeah, let’s compare that with a real world version. So say, I’m with my son, Freddie. And he’s and we’re out shopping. And he as as often is the case for children when they’re out shopping, and gets a bit overwhelmed, a bit overstimulated, a bit uncomfortable, the fact that he’s getting dragged from shot to shot without really knowing you know what the aim is, because I’m not, you know, telling him what I’m doing, you know, I’m just chopping. And rather than doing what he would normally do, which is negotiate, and these exogenous stresses mean that he’s more inflexible in his cognition, and it will pick up a toy and say, I want it Yep, I’m out shopping with him, I’m
often in a hurry. And I would say, No, you can’t have it, you know, talk to him more strongly than I normally would, which then increases their stress or further and he becomes even more set in his ways, you know, I would attempt physically to remove the toy from him, which will lead to a tug of war, which increases his stressed or further, he might, you know, lose the tug of war, sit down with a bump, make a noise hurt his bottom, and other people might turn around in the shop, to look, you know, the toy might get broken and through this exchange, and we basically get a chain reaction of events where one thing leads to another leads to another. Yeah. So there are a variety of different ways in which this real world toy removal differs from the experimental version that I’ve mentioned yet. One The last version is passive. Yeah, the child isn’t in control of when the events take place. Yeah, the experimenter determines what happens. Yeah. But in the real world, you stress has only happened when your behavior causes a stressor to happen. Yeah. So there’s nothing things aren’t just flashing on and off externally out of a out of the darkness. Yeah, in a way that experimenters love to do. The child has an active agent in the world. Yeah. And secondly, there’s not just one stressor. Yeah. And then we wait. And then it disappears. Yeah. One thing causes another causes and other causes and others. So you get this self feeding, self replicating cascade. Yep. So these are some really, really important kind of differences between the two. So that’s what I was talking about a lot in these paper in this paper, I talking a lot about other people who’ve pointed out similar differences in different contexts, like Edelman, you know, really, really nice kind of writing about this. Yeah. So but we like the definition. Yeah, I think this is a good definition. Yeah. But I think it’s a legitimate question, based on how little workers looked at this, to ask this question. Well, do we actually do self regulation? In this sense? Yeah, during early life, do we actually modulate our behavior contingent on our arousal state? And if we do the mechanism for how we do this will be a very simple one. Yeah. And it’s Alice stasis. And the predictions from Nick are very clear, in response to an increase in arousal I would modulate my way here is to decrease my arousal, yep. And in response to decreases in arousal, and modulate behaviors to increase my arousal, yeah, so very simple predictions for what self regulation should look like in real world settings. So we tested this. So we measured the autonomic arousal in a 12 month old infant. So this is arousal determined, defined, as I mentioned at the beginning, from a Composite Measure. So an average measure of heart rate, respiratory sinus arrhythmia, and movement. Yep. I’ve actually looking at the course of the whole day, but only during segments when these 12 month old infants are at home and awake. Yeah. And that’s mainly just because at times when they’re outside, they’re there quite often strapped in, in a buggy. And we didn’t want that those types of factors to be kind of getting in the way of our data. And so we just looking at times when they’re at home and awake. So I’ve got my arousal data during the course of the whole day here. Yeah. And then this is basically five oh, sorry, nine one minute epochs from this course of the day. Yep. And basically, this is an adapted version of a quanta Ray plot. So
those of you who know that plot and from heart rate studies, yep. And basically So here, I’m at arousal been three at time t. And then the next bit I’m also in arousal pin three. Yep. So that goes here. So arousal been three here. And then arousal been time t plus one here. Also three, so that could go here. Yeah. And this dot here I’m at arousal been for at time t. And at time t plus one I’m at arousal been two? Yep. So that would go around with him t for arousal pin tema plus one, two. So that would go here in this square here. Yep. So we basically populate the graph epoch by epoch in this way. Yeah. So the predictions from Alice stasis are very simple. Yep. If I’m at a high arousal state at time t, yeah, I should be down regulating my arousal state to the next state. Yep. If I’m at a low arousal state at time two, I should be increasing my arousal to the next day. Yeah. So. So we basically tested this and whether whether the data actually fits this. Yep. So our prediction, remember is if I’m in a high arousal state, to the likelihood that my arousal will go down to the next thing is high. Yeah. So the higher my arousal states, the more the likelihood the my arousal should go down. Yeah. And the lower my arousal state, the more likely or the my arousal state should go up. Yeah. So we compared the prediction with some control data, which was just randomly shuffled, so we’ve just destroyed so we’ve just randomly shuffled these time series to destroy the temporal interdependencies? Yep. And this basically, is regression to the mean. So this graph shows regression to the mean. Yeah. And we were looking at is our data follow this prediction? Yeah. In fact, we found that opposite. Yep. So we found that the higher your arousal, the less likely you are to decrease arousal. Yeah. And vice versa, for low arousal states. Yeah. Another way that we looked at it was just looking at the density of the dots on this line of where they stay at the same arousal state from being one to the next. Yeah. Our prediction based on our stasis was the intermediate arousal states should be the most stable, and that high and low arousal state should be less stable because we’re correcting for them. Yep. So this was our prediction based on our stasis yet. The real data we found exactly the opposite. The real data, we found that high and low arousal states last longer than intermediate arousal states. So the way that we described in this paper is high and low arousal states are sticky. Yeah, they last longer than you’d expect them at chance. Yeah. We did that. Based on across we sound the same pattern across multiple timescales. Yeah, when we downsampled our data into one second bins, to 2/62 bins, three to 902nd bins, we found exactly the same pattern across multiple timescales. So this is the finding that extreme high and low arousal states are sticky. So there’s a variety of reasons why this could be. It could be because high and low arousal states have intrinsically higher histories. Yeah, so that’s a term from physics, meaning this stickiness of a state. So we know that sleep, for example, is a sticky arousal state. Yeah, when I’m in that state, I’m likely to stay in it for a long time. Remember, we excluded a sleeping state from this data, but it could be that naturally occurring low and high arousal states have high histories in the same way. Another possibility is that high and low arousal states cause self reinforcing stress cycles. Yeah. Similar to that example, I was giving you the toy removal. Yeah, where one thing, one thing that can be outside of the baby causes a change in behavior. Yeah. Which causes then indirectly further increases in child arousal. Yeah,
there’s another talk that I’ve just recorded, looking at the self reinforcing stress recycles across a dyad. Yet how they can happen across a diet. But the same thing can also happen within an individual as in the toy removal task example that I’ve given you. Okay. So yeah, so those were ages from earlier the dysregulation example, in another tool, and the toy removal example I’ve just given you here. So this is this difference between self regulation? Yeah, how an increase in infant arousal can cause Allostatic corrective mechanisms, which causes then a decrease in infant arousal, so correcting for it. Yeah. So that’s what we think of when we think about self regulation. But there’s much less research into dysregulation so dynamical approaches to understanding dysregulation. So this is how an increase in infant arousal may cause these types of metastatic dynamic dis regulatory mechanisms. Yeah, seminar. And again, we talk about this a lot in this paper similar to mechanisms that have been shown in, for example, cognitive maintenance models of anxiety, operating via changes in how the infant interacts with the environment, physical or social. We don’t know this yet. We haven’t shown this, all that we’ve shown is that high and low arousal states last longer than you’d expect by chance. Yeah. So high and lows, lie, low arousal states are sticky. And it could be because there are these types of dynamic dis regulatory mechanisms, whereby increases and decreases in arousal become amplified over time. Yeah, that’s the opposite of Alice stasis. Okay. So, future questions. So we think now, based on this data, it’s a legitimate question to us. Do we self regulate in real world settings? Yeah, we’ve got a load of data from this from experimental paradigms, but experimental paradigms are very different to the real world. Yep. So do we dynamically and adaptively modulate our behaviors to correct for increases in decreases in arousal? Yeah. And in 12 month olds, the data suggests, you know, if anything, there are active dis regulatory properties, or that just this idea of history says, okay, so yeah. Okay. So part two of the talk is looking at a different aspect of dysregulation. And this is from some data from my PhD student Zeynep. On the second supervisor, Edmund, Sonuga-Barke is her first supervisor. And what Zeynep did for her PhD was she tested about a typical infants longitudinally, the same infants at six and 12 months. And she did a version of the toy removal task that I was just talking about. So this is Lab version, bearing in mind, all the caveats about lab measures that I’ve been talking about already 15 seconds, a baseline 15 seconds of toy removal repeats times three with a baseline at the end. And she video coded what we think of as traditional self regulatory behaviors like gaze aversion and self soothing, as well as reactivity. So that’s facial and vocal ethics. Okay, so this is a bit of a busy graph, and I talk you through it bit by bit. So this is time in seconds, these boxes are the times when the toy was attracted, and the toy present or the times or the baseline periods. Yeah. So the firt, the thick red line shows the baby’s reactivity at six months, I should say this is still all in prep. These graphs are subject to change, but we’re pretty confident that this finding is going to stay. Yeah. So this the thick red last line shows the reactivity at six months increasing during the toy removing and decreasing thereafter. Yeah, as you would expect. The dotted red line shows the charge reactivity defined from the same children using exactly the same way in exactly exactly the same setting at 12 months. And you can see that the dotted red line goes much much higher than the thick red line. Okay, so that is reactivity. So that’s facial and emotional reactivity, when we do the same thing, looking at regulation. So this is the, the thick blue line. Yeah. And this is the dotted blue line, you get less of a change. So regulatory behaviors don’t seem to change a lot between six months and 12 months, but reactivity. So that’s facial and vocal effect does seem to go up at 12 months compared to six months. Yeah. And when you break this down by different types of self regulatory behavior, so this is distraction. This is looks from other looks to experimental approach. Yep. You see that across the board? There’s no real significant changes between six months and 12 months. Yeah. And when you break this down by different independent reactivity measures, yep. So this is negative effect, negative vocal effect and negative behavioral effects, then you see, particularly for negative facial effects, and negative OSHA vocal effect, yeah, at six months, you get them. But you get them much, much more at 12 months, compared to six months. Yeah. And this is highly significant as a pattern in the data. Okay.
So this is this finding that when we first thought it, I thought, Well, I haven’t read this before, but it must be quite a well known finding, it’s quite an easy experiment to do. And there must be lots of people out there who’ve measured it. And we really haven’t been able to find very much in the data about this, this idea that we activity, but not regulation increases between six months and 12 months. Yeah. And obviously, given that this is this exact measure that we used is a clinical screener, and all of those citations that I started with about how excess regulatory behaviors, sorry, how poor self regulation, which often includes these reactivity measures is a marker of later psychopathology. We think it’s really important to understand why this is, theoretically, you know, someone who works at the boundary between attention control and effective control. Yep, attention control, we’re reasonably we think, at the moment kind of increases, not necessarily linearly, but it increases continuously between about naught and 10 years. Yeah. And if it is the question that affective control defined in this way is emotion reactivity to challenge increase gets worse during the course? first few years of life? Yeah, before it starts to get better. Yeah. Then I, you know, would really say at the moment, we really don’t have much theoretical insight into why this is, yeah. Why does emotion reactivity to challenge increased during the first few years of life? And why does it show different trajectory to attention control? Okay. So one possibility, yeah. Which is quite a common sense. One in a way is, it might be to do with the difference between feeling an emotion and showing an emotion. Yeah, between six months and 12 months, they just get used, they just realize that when I have a toy taken off me, if I show a lot of negative facial effects, in response, I’m more likely to get the story given back to me. Yeah. So that’s one thing, which as I say, isn’t just a low level point because it impacts on these behavioral measures, which from which, you know, big conclusions being drawn. Okay. So this is something we’re looking at. So we’re looking at the regulation between autonomic arousal, a vocalization type, effect and intensity. And we’re looking at that with respect to things like children growing up in chaotic households, later attachment outcomes, and so on and so forth. But we think this is another potential driver of these types of dysregulated dynamics that I’ve been talking about. Okay. So just to say what I’ve said. So we’ve got some evidence that it certainly at 12 month old data, in naturalistic settings, it looks like we can’t see much evidence that children are actually self regulating. In fact, if anything, the opposite, high and low states of arousal appear sticky. Yeah, they lay us longer than you’d expect by chance. This might be because of these types of short term dynamic dis regulatory cascades. Yeah, that can happen either between two people, or in terms of how one person how they interact with the environment. Yeah. And we also talked about this other area, which I think is quite problematic for our current approaches to understanding self regulation, which is how and why does self regulation get worse before it gets better? Yeah, is it to do with the difference between showing an image feeling and emotion and showing it? Okay, so two of the papers that I’ve talked about here, are already out. These are the references. The work from Zeynep that I talked about in the in, in part two of the talk is still in prep. And as I said, that might still change and but we’re confident in the basic findings by now so get in contact with me if you want to hear more about that. But for now, thank you very much to our funders, to colleagues and most importantly to our participants.