Interview between Speaker 1 (Meg) and Speaker 2 (Andrew Gordon)  



Welcome to Episode 31 with Dr. Andrew Gordon. We’re going to dive deep into the neuroscience of EF differences here.


Dr Gordon has spent his academic and professional career studying neuroscience. He worked on an Autism Center of Excellence grant to lead the neuroimaging component of the Specifying and Treating Anxiety in Autism Research (STAAR) study at the University of California, Davis, MIND Institute. Dr. Gordon now he works as the Academic Lead for Prolific, a startup specialising in providing fast, large-scale online data collection by connecting researchers with participants around the world.


If you’re new to the podcast, in nearly every episode I interview a professional in the field, autistic person, parent, or someone who crosses over between 2 or 3 of those categories, and we explore a topic in depth, relating it specifically and concretely to the work we do in our professional therapy practices.


This episode is going to be a little bit different because it’s sort of deep background on EF. In the next episode, I’ll speak with Oswin Latimer from the Foundation for Divergent Minds. We’ll get deep into how we can support autistic people with EF differences. But today, we’re going to go deep into neurobiology of EF, focusing on one specific study. If you’re someone whose eyes sort of glaze over when we go into brain structures and research studies, this may not be your favorite episode.  But if you’re someone who likes to deeply understand research, stick with me.


Being able to dive deep into things like neurology is one thing that makes SLPs and OTs different from behaviourists. We don’t study behaviour. We study people – holistically – in their real contexts. We strive to understand and figure out how to better support them meet their needs. And that doesn’t always start with observable behaviour. It starts with perspective taking, and cultivating a deep and complex understanding of who our clients are.


I’ll jump in throughout the interview to summarize some of Dr Gordon’s explanations, since research studies can be hard to grasp by just listening. But I think you’ll love where this goes. Okay without further adu, here’s the interview.

Meg 0:03:            Hi, Andrew! Welcome to the podcast.

Andrew 0:05:      Hi, thank you very much having me.

Meg 00:08:         Yeah, I’m really excited to have you here. I first came across your name when I was looking at research summarized on Science Daily. And there was a headline that said ‘Adolescents with autism may engage in neural control systems differently, study finds’ and this largely caught my attention because the quick one line summary underneath said, “Researchers studying executive control in adolescents and young adults with autism have published new research that suggests a unique approach rather than an impairment.” I wanted to know more about that, because many of our listeners work on skills and strategies to help autistic people who likely have executive function differences, and we know the executive function demands of this world are intense, that there’s a lot that’s hard for people with executive function differences. But very few of us really understand executive function on a neurological level, and we rarely see research that talks about differences instead of deficits. So, I want to go back and start with the basics before we dive into your research. Tell us what executive function even means and how it’s controlled.


Andrew 1:24:      Sure thing. So, I will start this by just sign-posting that I’m going to use the terms ‘executive function’‘executive control’, and ‘cognitive control’ interchangeably throughout this. They will mean the same thing. You’ll see researchers using any of those three terms to mean the same thing. I think the best way I have of kind of explaining what executive function is, is with an analogy. And the one I always like to use is I consider executive function something like an operating system on a laptop or computer, right. So, if we’re in your laptop or your desktop, there’s a whole bunch of different things going on. There’s wires, transistors, circuit boards, fans, all these kinds of things all acting by themselves, doing their own little job. But just the activation of all these things by themselves doesn’t lead to you having a good experience using a laptop. For that you need an operating system, right, like Microsoft or the Mac operating system. Executive function is kind of the same thing, right. All these areas of your brain, they’re all functioning to do slightly different things and they’re working in tandem with each other. But executive function is the thing that brings them all together into one kind of cohesive whole that basically enables your kind of higher-level cognition. A lot of people refer to it like the composer of an orchestra. The orchestra is playing different music, but it takes the composer to actually bring it all together and to make something understandable to us.

Generally, when we think about what cognitive control actually is, what makes it up, I tend to think of it as having kind of three major components. The first is cognitive flexibility; so, that’s the ability to kind of adapt your thinking or hold two things in mind at the same time. The second would be working memory; that’s updating information that’s in your working memory or maintaining information so you can keep responding in a certain way. And the third would be inhibition; and that’s the ability that we have to ignore irrelevant or distracting things in our environment. So, lots of researchers will say you can split that down into many, many, many more functions, but I think that those are kind of three key things that make it up.


And additionally, there’s another line of research says that executive function has two different forms: a proactive form and a reactive form. And that idea is that proactive executive control is essentially executive control engaged early before you need to do something. So, you actually prepare for an action ahead of time. And then reactive executive control is the executive control that you use when an action appears expectedly and you suddenly need very quick control to kind of respond to that action. That’s the kind of nutshell approach to how I understand executive control.


Meg 4:20:           That’s interesting. I’m thinking about how I break it down when I’m coming up with possible interventions for a child who isn’t doing something that they want or need to do, and I think how might executive function relate? And listening to you, I’m thinking I’m only talking about proactive executive function. Because we do things like clarify how to start the activity, help with initiation, how can I see the steps, when will it be finished; things around understanding time. Is that all proactive?


Andrew 4:52:      Yeah, yeah. I mean, essentially, it’s pro — I’d say proactive control really is the kind of control that I would say is more dominant. So yeah, those kinds of things would be good examples of proactive control. It’s essentially knowing what you have to do, having a context, and doing things in that context. For instance, if you know that in a minute, something’s going to pop up that you need to respond to in a certain way, you’ve mentally kind of prepared yourself to respond in that certain way. And there is some evidence — I wouldn’t say a lot of evidence — but there is some evidence that actually proactive control is quite efficient. That’s why people rely on it. Whereas reactive control, because it’s so late that it only occurs when an action is required, is actually prone to more error.


Meg 5:35:             Yeah, that makes a lot of sense, that we’d be better at planning than at reacting. And you described some concepts around cognitive flexibility too, which we know is a difference between neurotypical and neurodivergent people. Cognitive flexibility can be tough for autistic people and people with ADHD in ways that it’s not for neurotypicals. And I should add that neurotypical people are rigid in all sorts of ways, actually, that autistic people are much more flexibily. So, it’s not that rigidity is just this trait of autistic people. But it sounds like differences in cognitive flexibility do tie into executive function differences as well.


Andrew 6:16:      Yeah, certainly. I mean, I think cognitive flexibility is probably one of the — I mean, I said it was one of the kind of three main pillars, but probably one of the most important as well. And actually, as I’m sure we’ll get on to discuss with our paper, you know, cognitive flexibility is probably a pretty good — might be a pretty good explanation of what we actually found in that paper. So, it does seem to be quite important.


Meg 6:40:             Yeah, let’s, let’s talk about your paper. Tell us what exactly you did and what you found.


Andrew 6:46:      Okay. So, I’ll try and keep it short. Essentially, you’ve already kind of highlighted that there’s a lot of research that kind of shows what people generally refer to as ‘executive function deficits’ in people with autism. So, you tend to see that in behavioral data. So, when you give someone an executive control — someone with autism an executive control task — they may perform worse than a neurotypical individual on whatever measure is being used. But that kind of evidence for that at the neurological level is not really there. In fact, the research is pretty conflicting. It’s very conflated by the fact that people study individuals at different times of their lives, or small samples, and these types of things. Some people find that parts of the networks in the brain associated with control are impaired, some people don’t, but no one has currently looked at whether there’s any evidence for differences in proactive versus reactive control in autism at the neurological level, and whether that may help explain what we’re seeing in the behavioral data.


the behaviors and actions of autistic people. And autistic people report differences (or really, given the EF demands of our world, EF struggles) in their daily lives. But for people who study neurobiology, they want to see these differences at the neurological level. They want to understand the underlying mechanism that makes executive function unfold differently for autistic people. And so far, they haven’t really been able to find that.


Dr. Gordon explained that there are three networks in the brain that are responsible for executive control. Basically when you have a task that demands you use your EF skills, these three networks – the “task positive network” – go into action. You can see this on brain scans. The activity goes up in the task positive network.

But there’s also this fourth network called the default mode network. It’s different from the task positive network (the one that gets busy when you’re really thinking and planning). The default mode network goes into action when you’re daydreaming, relaxing, not doing anything too taxing. And when you’re working hard mentally, the default mode network goes offline.


So Dr. Gordon and his team wanted to look at these two networks during executive control tasks with autistic and non-autistic people. They put together a study with 141 kids, about half of whom were autistic, and put them in an fMRI machine while doing a cognitive control task. Here’s Dr. Gordon:

The task we gave them was something called the rapid preparing-to-overcome-prepotency task, which is a horrible mouthful, so we just called it rPOP. A lot easier. Basically, it’s what the task is, is they see a whole set of trials. And in each trial, they get a colored cue right at the start, which is a square. It can either be green or red. There’s a short gap, and then they see an arrow that points left or right. And the task is, if they see a green square, when they see the arrow, they need to press a button that’s on the same side as the arrow is pointing. So, if they see an arrow pointing right, they press a button on the right, and left on the left. It’s pretty easy, right? Because we’re kind of hardwired to be able to respond to congruent things really, really quite easily. Then the other type of trial is where they see a red cue instead. And if they see a red cue, that means that when the arrow pops up, they have to press a button on the opposite side from the way the arrow is pointing. Now that is a lot more difficult and it requires a lot more executive control, because you need to plan your response, you need to execute a response that’s not congruent. So, those are the harder trials, right? The idea is that this task gets at executive control and by using it in the scanner, we can look at what’s happening at the queue for proactive control and what’s happening the arrow probe for reactive control. That makes sense?


CLIP 3 I’ll be honest, when he asked this during the interview, I said, “Yeah…” But the truth was I needed to read it a few times to fully grasp it. So now that I’ve read over the transcript, I can recap this for you a little bit. The task is pretty simple. Green means do what makes sense, red means do the opposite. If you see green squares and an arrow pointing right, press the square on the right. But if you see red squares and an arrow pointing right, do the opposite. Press the square on the left. Green means do the logical thing, red means do the opposite. It definitely takes some thought and planning.


Andrew 11:36:    Okay, cool. So yeah, what we did, they completed this task while undergoing fMRI. And in terms of what we found, so the first thing we looked at was the behavioral data, because what we wanted to see is the actual evidence for the, air quotes, ‘deficit’ that people keep mentioning. And in terms of accuracy, we did find that individuals with autism were slightly less accurate, but not specifically so. But with reaction times, we actually found that individuals with autism was specifically slower on red trials compared to neurotypical controls. So, this is basically what you’d expect to see. This is the deficit that people keep saying about in behavioral data. So, we saw that which is great, because that means that we can then investigate what’s causing that with the fMRI data.


CLIP 4 So what they found, behaviorally, was that autistic people weren’t quite as good at following the “opposite” or red instructions. When the arrow points left and you touch the square on the right. And they wanted to understand why this was.

We then turn to the imaging, the neuroimaging analysis, and we did two types of analysis. The first was what we call functional recruitment analysis. So, in that we look at specific regions of the brain, and we say how much are they activating when the task is being completed. So, just the level of activity in that area of the brain. Then the second type of analysis is called a connectivity analysis, where we just — it’s a similar thing, it’s also looking at activity in the brain. But with the connectivity analysis, you’re looking for where one regions’ activity correlates with another regions’ activity. So, maybe this region goes up. And as this one goes up; this one goes down. That’s what you’re looking for. And the idea is, it gives you a better idea of the networks in the brain.
So yeah, there were quite a lot of imaging results. But I mean, the real kind of take-home messages are, first thing we did was look during the cue. So, when they see that red or green square, and this is looking for kind of — we’re looking for proactive control evidence here. And what we found was that both the individuals with autism and the neurotypical controls both activated these typical task positive networks on the red trials, which is kind of what you’d expect for executive control. But strangely, we saw that those with autism, it activated these networks much more on the easier green trials as well, whereas the neurotypical controls didn’t, which was a pretty surprising thing to see.


CLIP 5 Okay, remember here that the task positive networks are the EF networks in your brain that activate more the harder you are working from a cognitive control perspective. So on the hard task – the red blocks that tell you to do the opposite – both groups really activated their task positive networks. But the difference was, on the easier task – the green blocks – autistic people were using this network a lot more than the neurotypical kids. Basically – and this is my summary, not Dr. Gordon’s – the autistic kids were working a lot harder to do the same task.


Then we went to the probe part of the task, so when they see the arrow. And this is where we’re looking for reactive control. We found that those with autism had less activity in this default mode network, the one that reduces when you’re doing more executive control, than neurotypical controls. So, it’s on the red trials. It sounds like they’re doing a lot more executive control at the probe on the red trials.


CLIP 6 Okay this one took me a few reads too. When autistic people were doing the hardest part of the task and having to plan their response to the red blocks and arrows, the part of their brain that activates when they are cognitively relaxing was very low. They were not relaxing. They were working hard, but this time they were not in planning mode, they were in reaction mode – the other type of executive function. And the distinction here matters. Here’s Dr. Gordon.


Kind of taken together, that all sounds a little bit confusing, but what it means is that when in terms of proactive control, those with autism didn’t show any impairment at all. But they did show a different pattern of activity, in that they were actually doing a lot more executive control on the easier trials. They were preparing a lot more on those easier trials. Then when it got to the probe on the red trials, it’s seemed that they were doing more executive control at that point, possibly because they hadn’t prepared for those trials quite as much. So, they were having to then instigate this more late-stage reactive control on red trials. And because we know that reactive control can be a lot more prone to error, that is probably — well, I don’t want to say it specifically that it is — but it could be that that’s the cause of the behavioral difference you tend to see. So, it’s not really an impairment. They’re doing pretty much the same thing as typ, but they’re just doing it in a different way. And because they’re doing it in a different way, it’s leading to those behavioral differences you don’t see.


CLIP 7 Did you catch that? Autistic kids were preparing for the task the same way as non-autistic kids. They were upping their EF planning game. But on the red trials, this preparation wasn’t enough for the autistic participants, so they had to engage more reactive controls. And the reactive controls are more error prone. I think of it like memorizing your multiplication tables versus doing the math on the spot each time. Okay we are past the really dense part of this interview. From here, I’m just going to play the rest of our conversation for you.


Meg 15:39:         Yeah, that’s really interesting. And even if folks listening aren’t tracking exactly from this trial, to that result, to that pathway, I think that the implications that you say are a lot more clear. And here’s a quote from Marjorie Solomon, your co-lead on the paper from the Science Daily article. She said, when she was talking about the results, “This suggests that proactive control is not in fact paired —” She said, “This suggests that proactive control is not in fact impaired, but that those with autism implement it in a unique — and not necessarily maximally efficient — manner, because they employ proactive control to prepare for the easier versus harder trials.” And I know you just explained this, but give it to me one more time. What does this mean?


Andrew 16:37:    Okay, yeah. So, yeah. That’s exactly right. So, just — we’ll take it back a bit. I mean, generally, in neurotypical control — what you’d expect in this trial, in this paradigm — is that during the cue on the red hard trials, they will engage more proactive control because they know that it’s going to be harder to respond to the arrow later. But what we actually see is that they engage exactly the same level of — individuals with autism engaged exactly the same level of proactive control as those with typical development on those red trials. But they do a lot more on the green trials, which is not what typical development — what those with neurotypical development actually do. And because they’re preparing so much for the green trials, then on the red trials, when it comes to actually the time to make the action, they’re having to engage this late-stage reactive form of control, which isn’t quite as efficient as proactive control. And therefore, that may be why they are then struggling with those with those trials. Does that   —?


Meg 17:44:         Yeah, let’s use an example outside of the research if we can, and just sort of as a metaphor, almost, let’s say this is somebody trying to do a worksheet with subtraction problems, and then all of a sudden, there’s an addition problem. Is there an analogy we can make there?


Andrew 18:05:    Yeah, I think so. So, let me have a think. Yeah, so these results will suggest that — so, say they’re on this worksheet, they have multiple subtraction trials over and over again, and they’re in this pattern of preparing for a subtraction trial, right. So, they’re doing quite well on those trials, because they know, “I’m going to subtract, next. I’m going to subtract, next,” then, suddenly, this trial comes up, and it says, “You need to make an addition.” But they’ve already prepared to subtract at this point. So, when they finally see that addition, they think, “Oh, I need to change my behavior,” and then they try to do that kind of reactive control to that trial. And that reactive control is slightly less efficient, and therefore they are more likely to make a mistake on that trial.


Meg 18:56:         That makes a lot of sense. And there’s a lot of tie-ins potentially to other differences that we see with autistic people, right. We say once they’ve formed a routine, it can be harder or more stressful for them to change it. And it sounds like a lot of these things probably sort of tie together in the executive function component. But like you said, executive function is, it’s the executive, right? It’s linking together all of these other functions of the brain. It sounds like there’s a lot of relationships at play here, but with real implications on what we see for where an autistic child, or an otherwise neurodivergent child, let’s say with ADHD, might be successful and where they might need more support.


Andrew 19:46:    Right. Yes.


Meg 19:48:         As a researcher, what questions does this raise for future research?


Andrew 19:52:    So, I think — well, the first one was you’ve kind of hit on there already. I think the first most important question that comes out of this study is we see this different manner of completing an executive function task in autism, which is great. We see that evidence. But we have no idea of why they’re actually doing that. This study doesn’t answer that question. Now one of the answers to that may be cognitive flexibility, right. So, potentially, individuals with autism start doing this proactive control on green trials, but then fail to switch over the proactive control more to red trials, like a neurotypical individual would. And that kind of speaks to that rigidity that you just mentioned as well. So, potentially, that is an explanation for the findings we’re seeing. It’s just issues with cognitive flexibility.


And there are ways we can test that, right, we can do the same experiments but we could just present red trials for the first five minutes and then start adding in the green trials. We could see if the pattern reverses. Now that would be a great test. Other kinds of things that have popped up that this research has suggested, one interesting line of research that was done in individuals with schizophrenia was a attempt to actually teach more efficient proactive control strategies to them. And what they found was that after this teaching, not only did the kind of the deficits reduce, but also the activity in the brain became more like neurotypical individuals, which is interesting. That’s never been done with individuals with autism. So, I think that’s kind of a bit of low-hanging fruit there for future research. But I think the most important and the one that’s going to be hopefully answered by by our same group, even though I don’t actually work there anymore, is how does this — if we take our findings on face value, all we’ve seen as a snapshot of these individuals at about 17 years old, right? This is when they did the trial. But as part of this longitudinal study they’re a part of, they’re going to be scanned twice more about a year apart in each case, and they’ll do exactly the same task. So, there we can actually see how does this process evolve over time in these individuals. There’s a possibility that maybe this is just something that is a kind of delayed form of executive control, that as they get older, will become more neurotypical; or perhaps not, maybe they will retain this form of control throughout their entire lives. But we won’t know that until we essentially follow this up with more of the same research.


Meg 22:36:         That’s really interesting. I’ll keep my eyes out for future research that might have implications for how we can best support autistic kids in meeting the executive function demands of their daily life or changing those demands so that they can be successful given what their strengths are and what their challenges are. Thank you so much for your time today.


Andrew 22:58: No problem at all. Thank you very much for having me. 


I’m so grateful to Dr. Gordon for helping us dive deeper into the mechanisms at play in executive function differences, and the implications this may have for our autistic learners in executive control tasks. The takeaway here isn’t that we need to somehow change or remediate how autistic people approach executive function. This is just how they operate and there’s nothing fundamentally deficient about it. But understanding these differences can help us develop appropriate supports and accommodations when the executive function demands of a task are really a mismatch for how an autistic person learns.


I mean, do we have to have the subtraction problem in the middle of the addition worksheet? Are we trying to test reactive executive control? Or are we trying to see if they know how to subtract? When we know what we are trying to teach and are really crystal clear on that, it gives us the opportunity to scaffold, support, and accommodate all of the things around it that might make that task more challenging.


And if there are things we can’t change – say the stop light that is usually green, yellow, or red but is now suddenly flashing yellow – how can we best prepare our autistic clients to be successful in those moments?


If this peeks your interest, join us next episode when I interview autistic professional Oswin Latimer. In that conversation, we will really dive into how to support autistic people to be successful in this world, in light of their EF differences.