Monica Ellwood-Lowe is a PhD Candidate in the UC Berkeley Department of Psychology whose research focuses on differences between outcomes for students of different socioeconomic status, as well as the societal barriers that might hinder student success. Ellwood-Lowe tries to answer such questions as, what skills do children develop when they come from socioeconomically disadvantaged homes, even in the face of societal barriers to success? Do children’s brains simply adapt to their respective environments?
Ellwood-Lowe is co-mentored by Professors Mahesh Srinivasan and Silvia Bunge. She earned her bachelor’s degree from Stanford University. Monica’s work is supported by the NSF Graduate Research Fellowship Program, the UC Berkeley Chancellor’s Fellowship, and the Greater Good Science Center.
For this episode of the Matrix podcast, Matrix Content Curator Julia Sizek spoke with Ellwood-Lowe about her recent research on the topic of children’s cognitive performance, and how we might think about removing barriers to children’s success.
A transcript of the interview is included below (edited for length and clarity).
What is the achievement gap, and how has it typically been studied in psychology?
The idea of an “achievement gap” is the idea that kids who grow up in higher socioeconomic status homes, where their parents are more highly educated or have higher incomes, tend to do better in school than kids who grow up in lower socioeconomic status homes. I look at the achievement gap in terms of socioeconomic status (SES), but people also study it in terms of racial and ethnic differences. But it’s the idea that kids’ test scores, even by the time they enter kindergarten, are higher when they come from higher SES backgrounds.
What kinds of tests do researchers use to evaluate children’s cognitive performance?
Even before kindergarten, lots of researchers measure children’s vocabulary. They look at how many words kids understand and produce; starting as early as 18 months, you can get some indices of the number of words kids know. But one thing I really want to emphasize is that vocabulary doesn’t have to mean the same thing as achievement. One of the things that I think psychology does not do well, compared to other disciplines, is that when we think about these big issues like school performance or outcomes, we’re really focused on individual-level metrics, like vocabulary. When you think about how significant and longstanding these issues are, it’s really important to zoom out and think about the structural factors that are playing into all of this.
Beyond vocabulary, what are some of the other ways researchers can measure children’s cognitive function or school performance?
There are lots of different tests that measure children’s “executive function,” which is supposed to be an unbiased measure of children’s cognitive abilities. It’s gotten a lot of flack for not being unbiased for all sorts of different reasons. Typically, the people who have created it are white, upper-middle-class researchers, who have a certain idea of what cognitive performance looks like.
But normally, it ends up looking like kids playing games designed to tell us something about their cognition. They might be doing some kind of matching game, or there’s a measure where they see different sets of patterns, and they’re asked to fill in the missing set, as in, what completes this pattern? That’s called the matrix reasoning task. Those are the kinds of tests that we usually use.
What are the methods that psychologists use to explain the differences in cognitive function among the kids as they take these tests?
That’s been something psychologists have been really interested in. We administer these tests, and we find differences between kids from different backgrounds, and then psychologists come in and they want to know, why do we see these differences? When they’re happening even before school, it seems like it’s something that might be happening in the home.
One of the things many researchers have looked at is the amount of language that parents are directing toward their children. They’ll look at the very specific form of speech where parents are talking directly to their kid. This doesn’t include speech like parents talking to other siblings, or parents talking to other adults. That’s one of the things we have found really correlates with how many words kids end up knowing. But one of the things that’s really limiting about this is that children are perfectly capable of learning from those other forms of speech. We just think that kind of learning might be happening later, rather than earlier.
This is what people popularly call the “word gap.” You worked on a study about this concept and what it does for how we think about children’s performance. What did you learn in that study?
The concept of the “word gap” was popularized in 1995 by Hart and Risley, and they did a large study that led them to conclude that, by the time they are three years old, kids from higher SES homes have heard 30 million more words than kids from lower SES homes.
There are a lot of issues with this metric, and it has definitely come under fire. For one thing, we know that the gap can’t possibly be that big from more recent measures. For another thing, these were just numbers that were extrapolated from hours-long recordings in the home, and we now have better ways of quantifying kids’ all-day language environments. Third, this was again only looking at that very specific type of child-directed speech. When you zoom out and look at the entire language environment, that gap totally disappears.
That said, the general idea that higher SES parents talk more to their kids than lower SES parents has been replicated a lot. A lot of different researchers, even all around the world, have found this general phenomenon. That really led us to wonder why this is such a stable phenomenon. Lots of researchers have looked at individual-level mechanisms that might be promoting this. For example, maybe higher SES parents have more parenting knowledge, whatever that is, and that [knowledge] leads them to talk to their kids more. So maybe the solution is: let’s go into the home and train lower SES parents to talk more to their kids. But when you think about just how broad this problem is — it’s been documented since the 1950s in the US, it’s been documented all over the world, and in rural areas and urban areas — it doesn’t seem like these individual-level explanations can carry that much weight.
We were interested in zooming out to think about, structurally, what does it mean to be lower SES? When you think about socioeconomic status, it’s not a characteristic of an individual, but rather it has to do with their access to societal resources. So this was a first pass at looking at how structural barriers that lower SES parents are facing actually influence the amount that they can talk to their kid. We focus specifically for this study on financial strain, that is, maybe just having to think about their finances is actually quite taxing and leads parents to talk to their child less.
How did you go about measuring parents’ financial stress and how that might play into how they talk to their kids?
This was kind of a sneaky study on our part. What we were interested in is whether just the experience of being reminded of recent financial strain, or not having enough resources, would lead parents to talk less to their kids, regardless of their SES. We actually brought higher SES families into the lab, because these are the families that researchers in the past have said have the “parenting knowledge” to talk more to their kids, or they have whatever individual-level characteristics might lead them to talk more to their kid. We assigned half of these parents to fill out a survey about times when they didn’t have enough resources during the last week, or when resources were scarce. Some of them did talk about finances, but they talked about a range of things. And then we assigned the other half to fill out a control survey where they just reported on things they did in the last week.
After they filled out this survey, we left them in a room alone with their kid for 10 minutes under the guise of getting a second survey for them to fill out. We would say, we just realized the survey isn’t loaded, we’re going have to go to the other room and load it. So we would leave the parent and child alone in a room together for 10 minutes. And we gave them a fun puzzle box toy for the kid to play with, so the parents had the opportunity to engage in speech with their kid. They could narrate what was happening with the puzzle box toy, they could explain certain pieces of it. Or they could just sit quietly and sit on their phone and let the child play. We were interested in whether parents who had been thinking about their own experiences of scarcity would talk less during those 10 minutes than parents who just thought about things they had done over the last week.
How many people did you bring into the lab to ask these questions, and what did you find?
We brought in about 70 people to the lab. It’s a small sample, and it was our first pass at running the study, so I would call these very preliminary results. But what we found in general is that parents who thought about financial scarcity in particular talked less to their kid than parents who thought about all other forms of scarcity. And these parents didn’t differ in their income or in their education. They were all the same on these kinds of individual characteristics. But something about reflecting on financial scarcity might have led them to talk less to their kids.
How might one measure this outside of the lab setting?
There are a lot of different tools researchers have used to measure this. One is called a LENA recording device. It’s just a tiny little recording device, it sits in kids’ front pockets, and you turn it on at the start of the day, and then it records the entire day for 16 hours. For that full 16-hour recording, it quantifies the number of adult words spoken near the child, the number of child vocalizations, and the amount of back and forth between the adult and the child — what we call “conversational turns,” where maybe the kid says something and the adult responds.
Using LENA, how would you measure whether financial stress might be affecting the amount of language a child hears?
That’s what we were really interested in doing to follow up on this lab study. You can imagine that just bringing families into the lab and saying, “Okay, think about scarcity,” isn’t the most externally valid, meaning it doesn’t necessarily hold up in the real world.
What we really wanted to do next was make use of already available data and see if we could find any evidence for this phenomenon in the wild, so to speak. We used data from these LENA recording devices that other researchers around the country had already collected, and we used a few datasets where families had completed these LENA recorders multiple times over the course of a period of time. They ended up varying randomly in where in the month they fell. Some families recorded a couple times at the beginning of the month, and a couple times at the end of the month, in random order.
The reason we cared about that is because there’s a fair amount of research in economics showing that families feel more financial strain at the end of the month compared to the rest of the month. We thought that, if this was a real phenomenon, we should see dips in parents’ speech to their children at the end of the month, when they’re likely to be experiencing the most financial strain. What was really cool about this is that, because we had these multiple recordings for a single family, rather than comparing families to one another, we could really look within a family and see, do families talk less at times of the month that they’re experiencing more financial strain?
That’s a really amazing tool to have at your disposal. What did you find?
Again, I would call this pretty preliminary evidence. But we found some possible evidence that parents do indeed talk less to their kids at the end of the month. It looked like what was really affected was this specific form of child-directed speech or conversational turns. There were fewer conversational turns back and forth, vocalizations between parent and child, at the end of the month for a lot of these families. But things like the overall number of words adults were saying didn’t change. It seemed like it might be specific to child-directed speech.
They might be having conversations with other members of the family, but they aren’t thinking about talking to their kid.
Exactly. And I should say that many of the kids in this study, and in all of the studies that we’ve done, are really young. Think about kids in the first couple years of their lives. They’re not the most fun conversational partners, right? They don’t have that much to say. So it can take a bit more cognitive effort and energy to engage kids at that age.
And if you’re stressed out, that’s exactly the sort of thing that you wouldn’t have the capacity to do. This connects to the broader research you’ve been doing on other aspects of socioeconomic status and how it might affect children’s cognition. You conducted a study that uses fMRI imaging to look at how kids’ brains are working when we ask them questions. Tell us a bit about the methods used in that work.
We took the finding that there are some structural reasons why we see differences in kids’ early environments. We wanted to know how kids in lower SES environments then thrive. Because you’ll hear in the media that kids need to hear a certain number of words, or kids need to be exposed to lots of child-directed speech in the first three years of life. But really, when it comes to language development, that’s not actually true. We’re capable of learning new words throughout our lifetimes. Anybody who’s ever started a new job can identify times that they’ve learned words in later life. So we don’t think these kids are messed up if they’re not hearing lots of speech, but we want to know, what are the ways that they’re then succeeding? Because it might not be through the same mechanisms as higher SES kids.
So for the next study, we turned to the brain, using what’s called functional magnetic resonance imaging, or fMRI. We use resting state fMRI, which means kids sat in the scanner, and they didn’t do anything. They were instructed to look at a [neutral image] and do nothing else. And the brain never stops working. So during that time, the brain is activating; things are happening. We use what is happening in the brain during that time to make an inference about what their typical thought patterns are. What fMRI allows us to do is to look at what regions in the brain are activating in synchrony with one another. Where are neurons firing in the brain? And where are neurons typically firing at the same time as one another?
What are some of the ways that researchers typically have thought about how the neurons are firing in relation to having higher cognitive function?
One of the things we’ve learned about the brain is that there are a lot of different regions in the brain that perform really diverse tasks, but regions work together frequently. We have something called brain networks, which are made up of a whole bunch of regions that typically work together to carry out certain tasks.
One example of that is something called the frontoparietal brain network. This is a set of brain regions in the frontal and parietal parts of the brain, as the word would suggest; those regions are mostly along the forehead and the top of your head. These are regions that typically work together when we’re doing these externally demanding cognitive tasks. If you were filling out some kind of reasoning test, you would typically see a lot of activation from these regions in the frontoparietal brain network. That’s one that we look to a lot.
Another one that we think about is a different set of brain regions, which we call the “default mode network.” This is a set of brain regions that really work together at rest, so people thought maybe this was a default brain pattern, so that when you’re not doing anything, these are the regions that are activating. But we now know, they’re really involved in thinking about yourself, or thinking about things outside of the here and now, anything that’s really not external, but more internal. These are the brain regions that will typically work together to do those kinds of thinking patterns. These are the two brain networks, that frontal parietal network and the default mode network, that we investigated in our next study.
If we think about a child having more executive function, or cognitive ability, which parts of the brain do we think are doing that?
A pretty common finding in the literature is that as kids grow up, the connection between the frontal parietal network and the default mode network gets smaller. What this means is, say you are doing a really cognitively demanding task with intense reasoning, lots of researchers think you want the default mode network to shut down. You want thoughts about yourself to be really quiet, you want thoughts that have nothing to do with what’s going on to be as distant as possible. So you want less of a connection between the frontoparietal network and the default mode network. And so researchers have indeed found that a lack of connection develops with age. When kids are younger, the two networks work together more, and as they get older, they tend to separate more. They found that, at least among higher SES kids, the more separate those brain networks are, the better they do on cognitive tests. They found that all the way into adulthood.
Your research focused on the potential connection between these two networks that we wouldn’t expect for cognitively high performing kids. What was that connection?
What we were interested in is what’s going on for the kids in poverty who are doing really well on cognitive tests. When we think about things like the achievement gap, or kids’ test performance, we end up grouping kids off and saying “higher SES kids” and “lower SES kids.” But there are lots of lower SES kids who are living in poverty, and are still performing really highly on these cognitive tests. So we thought it would be interesting to see what’s going on for them. Are they achieving this high performance through the same mechanisms as higher SES kids? We went in looking at the connection between these two brain networks — the frontoparietal network and the default mode network. And we expected, based on all of the research that we had seen, that less of a connection between these two networks would be good for kids in poverty. We thought, maybe those kids in poverty who are doing really well have a lack of connection between these two networks. And what surprised us, and what we think is so cool, is that we actually found the opposite. We found this expected negative association for the higher SES kids, which all of the literature had shown before. But for the lower SES kids, we actually found that the kids whose two brain networks were more connected to each other were doing better.
Why might that be?
We don’t know yet. We’re still trying to figure out what the mechanism might be. But one of the things that we know is that those two brain networks, even in adults, do sometimes work together. They definitely work together for things like creative thinking. There are certain kinds of thinking where you want to be both engaging a lot of cognitive control, and thinking about things that are outside of the here and now.
You can think about designing something new. That’s the time when those two brain regions would be activated together. They would also be activated together if you were planning for the future. The future is not right in front of you, but you are planning it. We think that maybe the kids in poverty who are doing better on these cognitive tests are doing so because they’ve really had to adapt to a set of structural constraints that haven’t been set up for them to succeed. And maybe one of the ways that they’re doing that is by thinking outside of the box about how they can succeed, or planning for the future.
We actually found that this effect was strongest for kids who were living in more dangerous neighborhoods. It was strongest for kids who are Black relative to White. And we think that both of these things are evidence of structural barriers to success. It really points to kids having to adapt in creative ways to do well on these tests.
What do you think are the implications of this research, and what are the possibilities for future research in the same realm?
Whenever you read studies about brain development, it’s pretty likely that they were done with kids who are higher SES. If you have ever been in an MRI before, it’s a giant magnet. It’s not the most inviting machine. It requires a lot of time and patience. And it requires a lot of trust that the person who’s running the machine is not going to hurt you.
It just happens that higher SES families know more about this kind of research, and they’re more willing to participate, and they have more time to volunteer, whereas lower SES families don’t, often for good reasons. Typically, this ends up correlating with race and ethnicity. They don’t have a lot of trust in the research system to take care of them or to accurately report what’s going on for them. So a lot of our studies, and a lot of what we know about brain development, has come from this very specific set of kids whose parents are highly educated, wealthier, live near universities, and are excited about the idea of participating in research. And it has really limited the broader understanding of what healthy brain development really is.
That healthy brain development may not be one set of things. We can’t use universal measures to predict what is happening in someone’s resting brain state with how they’re going to perform on this cognitive test.
One thing we know for sure about the brain is that it’s really plastic, and it changes a lot throughout childhood. But it also continues to change in adulthood. And it’s built to adapt. Humans have lived in all sorts of different contexts and cultures very successfully for a very long time. We really think that one of the things that allows us to do that is the flexibility of the brain.
What do you think are the implications for thinking about the brain’s development beyond children and into adolescence and adulthood?
This last study was with 10 year olds, who are just entering adolescence. Adolescence is a really cool time. We think that some sensitive periods — times where the brain is very sensitive to certain kinds of environmental input — happen during adolescence. That might be a time when these kids are super receptive to new kinds of information. When you think about potential implications, if we were to make schools for adolescents and redesign them in a way that used the skill sets they already had, that’s one way of thinking about it. But it’s really just taking a broader view on what it means to be successful and how society can be restructured, and how it has been structured in the past.
One of the great potential applications for this research seems to be about that intersection of psychology and other disciplines. Have you made plans on how to work with other disciplines on this kind of research?
We have been working with economists right here at Berkeley. One of the things we’re doing, thinking back about the first study, is that we are going to give some families unconditional cash, and see whether that affects their speech to their kids. One very direct application is just giving parents more money, enough to change their behavior. There’s a big national study going on in that realm called the “Baby’s First Years” project. It’s an unconditional cash transfer study, as well. That’s one future direction.
I think it would be really cool to pair up with educators and people who are in the schools to think about what kinds of skills kids are developing from all different contexts — and how we can best measure and support that.
Listen to more episodes on the Matrix Podcast page, or listen on Apple Podcasts or Google Podcasts.