Transcript
[MUSIC PLAYING]
[Text on screen: What first got you excited about doing developmental psychology?]
[Text on screen: Linda B. Smith, Professor, Psychology and Cognitive Science]
LINDA SMITH
I want to understand why we’re as smart as we are, why we have the personalities we do, why we have the differences among us. We’re all our own unique little flower. And the only way to understand all that is to look at developmental process.
[Text on screen: Indiana University Bloomington. Cognitive Development Lab.]
LINDA SMITH
We really do, literally, get made in time. And early on, every experience is building, building, building, building the brain. And what happens, let’s say, when you’re in kindergarten very much is the result of the accrued experiences and brain changes that have occurred as a result of those experiences prior to that event.
[Text on screen: What was the thinking behind the Homeview experiment?]
A lot of my work and the real impetus for Homeview 1, which we began with, is concerned with how infants break into language. So infants say their first words at about a year. And then language gets much more sophisticated by 2 years. Then they’re really pretty good by 3 to 4. Even though they say their first words at a year, however, a lot of experiences up to that first birthday create the ability to do that.
So what we wanted to do was really try to understand, at the scale of everyday life, the baby’s role in creating their own experiences, both interacting with objects, playing with them, holding them, dropping them and how those behaviours elicited the naming of objects and talk about objects to the infant.
[Text on screen: How did you set up the Homeview experiment?]
Back in about 2013, I decided that what we needed to do was to actually try to measure everyday experience at the time scales which it happens and at the time scales which affect the brain and affect developmental process. And that’s fractions a second, OK. So that led to the idea of putting wearable sensors on babies – head cameras, head-mounted eye trackers, motion sensors – so that you could get a fine-grained detail of what they do, what happens, OK, what they personally experience from their point of view.
We put head cameras on babies in the home asking for day long recordings of their visual experiences and auditory experiences from their point of view. So you could really try to see the scale of experience, the scale of events that happen in a day.
[MUSIC PLAYING]
[Text on screen: How does understanding more about what babies see from their perspective tell us about language learning?]
Babies break into language by connecting words they hear to perceptual experiences. It could be they connect a word to a sound. They could be connecting a word to a seen object. It could be they connect a word to an emotion. All of those are potential meanings that the person feels that can co-occur with or a word could be about. About 50% of the words in almost all languages – there’s some interesting oddities – but almost all of the world’s languages, about 50% of the first words that babies learn are object names – cup, juice, banana, mama – our names for things, things that you see.
And the thing is that much of the other reason for looking at it is much of language learning of all kinds happens just listening to people talk about whatever is going on, being able to rapidly identify what is being talked about. When the mom says, ‘So what is Snooty the dog doing with your sippy cup?’, OK, they need to be able to find the dog and the sippy cup to even pull out the meaning of that in the moment, right. And that’s what everyday language is about. So being able to rapidly visually identify what’s being talked about is also a key skill in early language learning.
WOMAN
What does the puppy say? Puppies say woof, woof. Woof, woof.
[Text on screen: Why did you ask parents to put cameras on their infants' heads at home for this experiment?]
LINDA SMITH
When you are in a room and you, as an adult, are looking around and you are watching what a baby is doing, one, you see it through the lens of your own eye. You see the behaviour of the mother, the behaviour of the baby. You make inferences. Oh, the baby took that object because the mother showed it to him.
But when you look through the egocentric head camera, right between your eyes, of what the baby sees, baby never saw the mother show that object, OK. It wasn’t in the visual field. You find out that what babies are seeing – what you see at any moment in time – is highly determined on the spatial relation between your eyes, that is your head, and whatever it is you’re looking about.
So as you move your head around, right, you see different things. I can occlude the camera, guys, just by ducking down. You change the whole view. You need to see it from the baby’s point of view. The only visual information that is going to teach about visual objects, about the kinds of categories we talk about – about putting things in, on, under, over, throwing – all these things babies have to learn, the only thing that matters is the information that hits the baby’s eye, not your 3D view of that information, OK.
It’s an enormous insight. It has changed what we know about babies and babies’ learning remarkably. One thing … We don’t do this as adults, but if we want to selectively attend, OK, if I’m interested in something like this cup, you know, I can be sitting out there, and I can look at a distance. There’s my cup. Babies don’t do that. They bring it up. They block out all the distractors. They control their input. And that is really important to their learning, actually holding an object. I’m blocking you out, Jane. OK, you can’t distract me now. I’m like studying my cup.
They control the input – actually optimise it – in ways that we know help for the visual extraction of information and for the duration of time over which they are looking at that information. Gives you time to pick up the visual properties, get them represented robustly and strongly and in memory, and get a name attached to them or other information that a parent may talk about it.
This is really important for understanding, I think, typically developing children. But many developmental disorders – Down’s Syndrome, ADHD, autism – have their disruptions of the sensorimotor systems. And it used to be that they thought, oh, their motor abilities, that’s just some co-morbid thing on the side that’s not relevant. But because the baby’s own behaviour is creating the experiences, the data, for learning, it’s absolutely critical to understand those small bumps.
[Text on screen: Does this have any immediate real world application?]
I have a friend, Cole Galloway, he’s a professor of physical therapy at University of Delaware. And he studies babies with major muscle disorder, motor neuron disorders, major muscle disorders. And these children, you know, they develop very, very slowly. Their social skills are bad. Their communication skills are bad. Their understanding of relations in the world are bad. If you can think about the bumps they’ve got to get over in terms of the typical pathway to creating experiences of learning, he’s helped solve them.
He’s done two amazing things. He, one, makes exoskeletons for these babies, so they can sit stably. That means they can hold … they can make some progress in holding objects and sustaining attention, OK, early, so they can do that at the right time. Get over that bump on the road to, you know, breaking into language. He also has made little motor cars. You know, you can have 2-year-old children in these little, kind of, moving cars. They can move with minimal kinds of body movements. So they actually can be autonomous and interactive in the world.
MAN
You are looking around like I have never seen before.
LINDA SMITH
The whole point I’m trying to make is that, seeing it from the baby’s view, actually shows you where the potential problems are and the potential ways to get in and make a difference in the life of a child.
[Text on screen: Why have you now taken the Homeview experiment to India?]
We are also collecting data, or collected data, using the same procedures in Homeview 1 in a under-resourced fishing village outside Chennai.
And there are a number of reasons to do this. If I just study children in Bloomington, Indiana, I have both geography, spatial layout, context of here. And maybe it’s not universal. Maybe it’s not everybody. There may be many, many, many paths to becoming a successful human being. And Bloomington, Indiana, a college town in the Midwest, may not be, OK, optimal for understanding that.
One of the reasons we were interested in going to India was because we wanted a place, first of all, that was very, very dense in the number of people. And that had related to our work on early experiences of faces. What we had found in the studies in the US, here in Bloomington, was that 15 minutes out of every hour of a baby under 4 months of age visual experiences were the face of an adult close up, two eyes, and that that frequency declined pretty dramatically. So by the time you got to 12 months of age, faces were six, four minutes per hour, waking hour.
That’s really interesting because we know that early face experiences, if for some reason you don’t have them – either because you were born with cataracts that don’t get removed until you’re 4 or 5 months of age, or people hardly ever looked at you in your early months when you were cared for – that you have disruptions in visual perception and in face recognition throughout your life. So the … we have what the visual experiences are in the first three months of phases in babies in the US, and then they decline.
And the other interesting thing about the experiences of the US babies is that 80% of the faces they saw in the first three months were just three faces. Let’s say dad, mom, and grandma. And that’s very interesting because it suggests that the system for recognising any and all faces might be set up by just a few faces. You just need a lot of experiences, visual experiences, early of just three faces, and you’re good, good to go, OK.
But is that really true? Is it sufficient? But do … is it different elsewhere? So we went to this fishing village. It’s very, very dense. Houses are very, very small. Most of life is spent outdoors. And there is minimal electricity. So they have much more bigger light–dark cycles, OK. So the interesting thing is that we found in that study is that the babies … the young babies in that culture have visual experiences – the babies under 4 months – that are pretty much like the babies in Bloomington.
As you get older, it diverges, OK, because you’re spending time out in the street and in these dense communities. Then, the statistics change. But the first four months appear to be tightly constrained. That suggests to us that it’s really about … that what evolution has done is controlled the input of faces to babies.
And really, putting it in everyday terms, it’s done it through love, OK. So the love of the mom, the love of the dad, the love of the, you know, sister or grandma in the family. They just love looking at those babies, and they stick their faces in front of them because they can’t help it, OK. And so that is just really the primary people who are giving the optimal input in those early stages. And that’s probably just built into us, right.
[Text on screen: Do your studies tell us anything about parenting?]
You can measure parent behaviour in interacting with their infants. And what you’ll find is that parents who are what are called responsive – that’s a technical term – responsive to their infants have infants who actually learn better in the context. It predicts overall learning later. And it predicts success in school when you look at this, even in infants as young as 18 months of age, you can predict success at 5 years, OK.
But what makes a parent unresponsive? So should I just say to parents, ‘Be responsive’? No. What makes a parent unresponsive? Stress, anxiety over issues. A parent’s first job is to keep the kid fed, healthy and a roof over their head. And not everybody has those opportunities. Responsivity is deeply, deeply linked to socio-economic status. And any one of us, if we had a sick spouse, a sick child, a bill collector chasing us down, get evicted, we are not responsive in the moment. So I’m not into, like, lecturing parents. There’s other ways to fix those problems, I think.
[Text on screen: Have you had any particularly surprising results?]
There’s been a number of surprises. One big surprise is the issue that babies are directly controlling their visual input and optimising it. So if you look at results from laboratory experiments about what kind of images or video events are easiest for babies to learn, babies actually move their bodies, heads, hands, postures in ways that create what experimental psychologists would say are optimal streams for learning, OK. That’s amazing.
The second thing that is amazing, that I think we are playing a role in shifting in how people think about it, is that people have usually thought about learning in a kind of discrete trial model. That I might … Well, what do I mean by that? Well, experiments, if they wanted to examine how babies learn the word for objects, they would show a picture of an apple and say ‘apple’. They’d show a picture of banana and say ‘banana’. They show a picture of a dog and say ‘dog’.
And the assumption is it’s a co-occurrence of the visual experience with the heard name that matters. But the world doesn’t work like that, and neither do our memory systems. Think about getting dressed in the morning. Mom is trying to get your shoe on. You’re playing with the Cheerio. Your mom takes a Cheerio away because it’s three days old. She puts your other shoe on in an attempt to get you dressed. The dog runs through. Everybody’s distracted for a moment.
What does the naming of objects in that look like? It looks like a narrative. It looks like a story, OK, of events. And what we’ve also discovered is there’s these lots of predictive relations between what’s happening in the physical world, which you can see – the shoe on, the shoe off, the dog, the Cheerio – predictive relations between all those seeming wild little episodes of events in the visual world.
And then the words have relations to each other, predictive relations to each other. What’s happening in time has predictive relations to each other. And the words get very rapidly learned. You might only need one wonderful event with your first taste of pizza, 10 namings of pizza, one wonderful meal at 12 months, where somebody gives you that crust of pizza with a little bit of cheese on it, and you know pizza for the rest of your life. It doesn’t take days.