Full house, as to be expected when Karen speaks. So first procedural thing, it's the first time that actually somebody, Dana Brooks, organized to have the talk filmed. So Karen is fine with it, but I think I need to ask everybody whether everybody is fine with it, because we might just get a little bit of your... Bold.
Hair. Hair. Hair.
You may or may not like that. So, is everybody fine with it? Even if we catch you on film? Does that mean we have to be way more polite than usual?
No. I mean, that actually has been a little bit my hesitation or fear to start filming because people have asked me to film that for two reasons. we start to feel a little bit constrained in our usual expressiveness. And the other thing that then people don't come anymore.
They just say they look at it at home and then we're here, an empty house. So, but anyway, it got organized today and you're fine with it. So, Karen. Welcome, Karen.
I've known Karen for a long time because there is a sort of a conceptual joint pedigree that we have. Which is, I guess, the Gibsonian route in education. So Karen did her bachelor's degree at the college that I couldn't remember.
Yes, Sarah Lawrence College, and then PhD from Emory University, following with a PhD at the Albert Einstein College of Medicine. Postdoc. I only have one PhD. Oh, did I say two?
Okay. One was plenty. Yeah. Importantly, rather than the institutions, are the people she worked with.
And in her PhD, her main advisor was Eleanor Gibson, the wife of J.J. Gibson, who became extremely famous in the psychological community. So did Eleanor, not quite as much, but I think I heard you say that she is by far at least as good as her husband. And I think Karen is the person who really takes on her legacy in developmental psychology and carries on the spirit of what she published originally by herself, and then Karen. So other than Karen, another influential person for Karen's work was Esther Thelen, again in the developmental psychology community, a very, very big name.
And lastly, Ulrich Neisser, again. a big name in psychology. So Karen has received a lot of awards and I'm not reading them all out believe me they're big awards where we amongst other things crossed paths was at the NIH study section for motor function speech and rehabilitation where Karen shared the session at one point and yeah so there we met also Karen. One thing that I totally envy you for, many other things too, but is the Merit Award from NIH. This is fabulous and speaks to Karen's achievement in developmental psychology.
Maybe just one or two words on what makes or characterizes her work is that she does very original work in creating new experimental paradigms that hopefully tease out or present scenarios where we gain... better insight into developmental features, emphasising the interaction with the environment in the true Gibsonian spirit. One other thing that came up several times, I need to mention to you, Karen organised an amazing concerted action by altogether 60-plus PIs under her own heading, an NIH grant. which is for something called data sharing, data library, data brick.
And I heard her saying that she set this up and did 20 plus subcontracts. I just did a grant with two subcontracts. It nearly killed me. 48 subcontracts.
48 subcontracts, okay. Well, she's still standing. So thank you.
You're welcome. So. I changed the title on you, but don't worry, it's still good.
And I'm going to tell you how researchers have typically, traditionally studied infant movements, infant motor development, and then I'm going to suggest some new ways for us to study motor development. And I think these new ways will give us a better picture of how it is that babies really learn to move and acquire the skills. that they need to get around in the world.
So one assumption that really needs to be rethought is that motor development is a universal process, like physical development, which actually also isn't all that universal, because it turns out that infants, healthy infants... all over the world will learn to sit and to reach and to walk and so on. But the forms that those actions take and the way that they learn to do those actions and what others say.
skills they learn. All of that depends on the child-rearing practices of their culture. So how parents raise their children, really, actually really everything about the physical and social environment affects opportunities for learning new skills. But we actually as researchers know very little about child-rearing practices or their effects. And that is because most of the world's population is simply not represented in any way in developmental science or behavioral science more broadly.
So basically what we've been doing as a field is we're studying only our own children and our own child-rearing practices, but it turns out that our babies and our ways don't represent the world. So some of you have babies, right? So I want to ask you, people who have experience with babies, how old would your baby have to be before you would like set your baby on the kitchen counter and walk away?
No, I was joking. I don't know. I'm thinking about it. No.
Not a two-year-old? Why? They're bound to fall. You think they would?
Do you know how old babies are when they learn to sit up in our culture? So, like, when a baby can sit in a stable way, you know, six months is the average in our culture. Yeah, but two-year-olds are unpredictable.
Yeah, they wouldn't jump. Okay, this is a typical five-month-old. hammer in. And so six to eight months in there would be the range when your own children would learn to sit up.
Sitting at four to five months is typical in this culture. and it's really amazing to see the sitting proficiency of babies in this culture but even more amazing I think are the caregivers expectations because the caregivers in this culture are so assured they're so sing when that their baby not only can sit up but can sit up well enough and smart enough that they could leave their babies on high places and walk away, walk out of arm's reach, even walk out of the room to do chores, and that their babies are not going to fall. Mother's willingness to leave their sitting infants on adult furniture like this was related to infants' sitting skills, so the more proficient the baby, the more likely the mom was to leave them alone like that.
And I'll tell you what, none of the babies ever did fall off of a high place, at least while we were filming. filming them. So here's another question.
How do you hold a newborn? Some of you have held a newborn. How do you hold? Okay, why are you doing that, Neville? What does that mean?
You have to support the head. How many would agree? It's not the righteous way to hold a newborn? Why do you think you have to support the head?
Like if you think if you let go of their head, their necks... just kind of crack and then it's like roll across the floor. They have big heads too. They do have big heads. This video is from an amazing film put together by Blundy and Will.
This is a typical bath and massage. and it involves exercise, passive stretching of infants' limbs. And this is in Mali, in traditional villages in Mali, where they do not think that you have to support a baby's head. So their babies are working against gravity to hold their heads up and the rest of their torso and their bodies up from the newborn period. And cross-cultural work by Brian Hopkins and others show that exercise...
Exercising baby's limbs accelerates the onset of skills like walking. And there's a dose response effect. So the more you exercise, the earlier, the younger they are.
when they start to sit or to walk. It's really cool to live in New York City because we have some of every culture in my town. So this is an Armenian doing a typical exercise routine. with the baby and his family because it is their expectation that you do not have to support the baby's head.
In fact, if you don't exercise your babies like this, they won't grow up to be normal. And in Cameroon and in Kenya and in Mali, they think that you have to train your babies to sit and to walk the same as we believe we have to train our children to use a toilet or to read. And this is in Iceland, which I'm just showing you because it's not like a black-white thing. It's not an ethnic thing. It's a cultural thing.
So if your culture believes that at four months you can stand and you can hold your breath underwater. and it's good to be swimming and doing underwater practice and so on, then that is what babies will do. And it's not just cross-cultural research. There's experimental data on the accelerative effects of exercise. So imagine white U.S. infants.
This wasn't one of them. This just happened to be a white U.S. infant who was getting exercise. Most famously, this was shown by Philip Zalazo in a famous science paper in 1972. But a few minutes of daily practice every day will keep infants retaining these sort of newborns.
stepping movements and it accelerates walk onset. There's also a cognitive issue because my daughter could run when she was nine months old. She would walk right off a visual cliff and a chicken won't do that.
That is a very good point and you need to hold on to that until I get to the visual cliff. So yeah, there are children in our culture who can... walk at nine months, running with a flight phase would be very rare, but there's children and some cultures who are running with a real flight phase at nine months. So these are examples of accelerating motor skill, but there's also examples of delaying motor skill.
So all over Central Asia, Central Asia is all the stands, Tajikistan, Afghanistan, Uzbekistan, etc. Kazakhstan. Caregivers use a special cradle called a Gabor cradle.
It's one way that they toilet their infants. So these are external catheters. That's for a girl, that's for a boy.
This is a little bowl that goes under this hole. This is where the baby's bottom goes and the bowl sits underneath to catch all the the waste. And so it's a way to toilet the babies to keep them safe and warm and dry. In a culture that started as nomadic it's got, well let me show you.
So This is what it looks like typically to get cradled. So the cradles have a handle that you can use to carry the babies around. That's putting on the catheter. It takes on average like 2.3 seconds to catheterize your baby.
Then they swaddle the lower limbs. And then because there's no sides on the cradle, they bind the lower limbs. This whole practice is highly routinized. So caregivers all over Tajikistan cradled their babies in the same way, with the same order, the same... accoutrements the same timing.
Swaddle the upper limbs, bind the upper limbs. These babies are bound neck to toe. They can turn their head.
They can wiggle fingers and toes. Central Asia has four seasons, so they cover them with blankets and they often cover the top as well. So these are the skill levels of cradled infants in Tajikistan. At zero and four months, babies didn't really have any problems. any postural or motor skills at eight months.
Forty percent of them still couldn't show any postural skills. Only 20% of them could sit independently and only, what is that, 16% could crawl. And in comparison in Western culture, in our culture, by eight months nearly all of our babies are sitting independently and two-thirds are crawling.
So that whole bar would be green and red. By 12 months, only 50% of the Tajik babies were crawling. None were walking. Walking is in black. At 12 months, 12 months is the average age at walk onset.
So half our babies crawl, half our babies walk at 12 months. And we didn't see walking until 16 months. That's the black bar. Most of the kids were walking at 16 months, only some of them, and it wasn't until 20 months out here that nearly all of the babies could walk.
What's considered a tripod? Tripod means it's a... It's a long sit with your legs out like that and your hands down in between to hold you up.
It doesn't qualify to murder people as independent sitting, but it's a little sit thing that babies can do. Cradle hours show a dose response effect on motor skill. So the more cradle hours, the lower their motor skill, even when you partial out the effects of age. And you should be... checking out like number of hours in the cradle.
So some of the babies are in the cradle for half to nearly all of the day. So it's not something you just do at night. It's not something you just do for naps. It's something you do to keep your baby clean. dry, warm, safe.
I can hold the books down to later, but does this, is there any catch-up effect? I mean, do these people have... Well, in the grant we're doing now, we're looking at preschoolers, and the reason we thought to look at...
Okay, that one too. Well, so I'm going to say the preschooler part. So, we're collecting all these data. Video is the primary source of the data, because they don't have electricity in Tijicasan, so we just took video cameras. And, um...
You really... at these kids who clearly looked delayed, like you know, like, ah, these babies, like anyone who does motor development, like, ah, what's wrong with these babies? Their siblings are racing through, you know, the back of the video screen and looked really good. So we brought another grant to ask about older children and to do a longitudinal study where we could follow up.
Preliminary data would suggest by preschool they look fine. And here's the things to keep in mind. So I thought about all these these things.
I'm like, are they like in the Olympics? Like, what do they do? Like, do they have sports there?
Like, what is this place? Everybody rides horses. So when I first heard that, I said, I'm like, what do you mean by everybody? Well, men and women. I'm like, women?
Children? Yeah. How else would you get around?
Okay, horses. So now I'm on the lookout. So there is, what is that called? You know, they have like that... No, it's the ribbon, and it's like gymnastic ribbon-y dancing with the ribbon.
So they're in the Olympics, and they ride horses. Everybody in the culture has a very, very flat head. So flat that they would all be wearing helmets if they were growing up in Boston or New York City. Seriously. That would need to be corrected, except it's not corrected in Central Asia.
And you know what? They walk, they talk, they wage war, they do all the things we do. Why do they do that?
Because they're lying on their back all day. And so the incidence of plagiocephaly and brachiocephaly, you know, flathead or side flatness, has increased in our own culture because we now have the back-to-sleep movement. And if you spend a lot of time lying on your back when your skull is still, you know, soft, it makes up for a flathead.
Very flatheaded culture. Does that have any impact on the development? Well, if it were your grandchild...
your grandchild would be wearing a helmet. If it were any child grown up or whoever in Tajikistan they think it looks really good and they don't think it has any effect. So you know like you can get positional torticollis by having your head turned to one side but they're like this.
Oh, it just meant age. Anyway, don't worry about it. Alright, so we can look to our own culture, so I referenced the back to sleep movement, but we toilet our babies with diapers, and it turns out that diapers affect infant walking.
So we compared babies... and three conditions, walking naked and thin disposable pampers and wearing a bulky cloth old-fashioned cloth diaper and we tested newly walking 13 month olds and experienced 19 month olds. And we tested their walking in a standard lab task.
So in the standard task, you encourage or coerce infants to walk in a straight line continuously. And then you can measure things about their gait. So in this case, the baby's walking over a pressure-sensitive mat.
And we're measuring the timing and placement of each footstep. And the short story is go naked because you can see. See the effects for yourself. These are little trails of footprints from that baby that I showed you.
So a typical infant walking over the gate carpet. When the babies are walking naked, their steps are longer. and they're closer together, and they're more consistent. And when they're walking in diapers, they have less mature gait. Their feet are further apart sideways, shorter steps.
So these are different individuals? Same. So within one.
Within subject design. Just look at the video. It looked like that infant was using a at the tarsals first, not heel first. Toe walking?
That's just common. That's just baby walking. Was there any difference across the... Not for diapers.
But babies almost never do a heel strike. So they either go toe to heel or they land flat footed. That's just baby walking.
It's not a diaper effect. Did it affect cadence? Huh?
Was there cadence effect? No, cadence effect. I know, I'm thinking, yes, their cadence would be affected because the speed and the timing were affected.
But I want to show you some data. So step width gets narrower over development. And so that means better walkers have a narrower step width.
Infants took the narrowest steps when they were walking naked and they took wider steps when they were walking in diapers. Step length is the front to back distance between feet. feet, feet, feet, and step length gets longer with development, so better walkers will take longer steps, and babies took the longest steps while they were naked or wearing disposables, shortest steps in bulky cloth diapers, and the dynamic base angle, which is the angle between three consecutive steps, it's close to 180 degrees in an adult, so bigger angles is better walking. you and babies had the largest space angles when they were walking naked, smallest when they were walking in diapers. Now, in every one of these is a significant effect, and in all these cases you can see that 19-month-olds walk better than 13-month-olds, which, okay, duh, you would expect, but there's no interaction.
There's absolutely no interaction. That means that wearing a diaper is equally disruptive to walking. When you first start walking is when babies have been walking.
for several months and they're actually asymptoting in terms of their skill level. And you may be thinking like, alright Karen, you told me they're significant and I guess I have to believe you, but I'm looking at the y-axis and it's just like really what's a couple of centimeters between friends? Is this a functional difference?
Does it even matter? And I want to argue that the cost of wearing diapers is not trivial. The cost of wearing a bulky cloth diaper on Stepwith, for example, is nearly two months of walking experience.
That means you put your baby in a cloth diaper and your baby walks as poorly as it would have done two months earlier while walking naked. Two months in motor development is like a lifetime. For a disposable diaper it's five weeks of cost. And it not only makes babies walk badly, hold on, it makes it harder for them to walk at all.
So while wearing diapers, babies... had more missteps and more falls compared to walking naked. Yes?
Do you know if they were to be able to walk at least, and you have someone that's been walking with the diaper on, and you remove the diaper so they're walking naked, if they would walk as well as someone that had not grown up without the diaper on at all? Uh huh. Is it something like two women and worse with the diaper on? I do not know that because we did this study in New York City, and we actually asked the parents how much experience does it take to have a child with a diaper on? does your baby have walking naked?
And I don't know, like, for Bostonians or wherever you guys come from, but I imagine you have, like, suburbs and backyards and grass and places where you would put an infant. Like, in New York City, you don't put your baby down. No one has a backyard.
You don't put your baby down, like, in public spaces because there's, like, syringes and poison everywhere. And so nobody's baby, like, I think it was, like, 19% or something like that had ever walked naked, you know. for even a few minutes in their home. So they did not have experience walking naked. But this isn't like a rearing experiment where we could randomly assign them to nakedness or diapers or something.
Have you looked into whether there's an after effect, such that when you take the diapers off, that the step widths and the step frequency remains affected? It does not. I mean, there's other things, like if you put...
It does not. If you put like a platform shoe on one of the baby's legs, so you elongate one leg, they have an after effect for like one sequence and then they fix it immediately. a baby with weeds and then take the weeds off, you know, it feels like it's almost immediate recalibration.
So there's no long-term damage as such? I don't know about long-term damage, like what would happen. I mean, what it means is that...
all of us and all of our babies are growing up under a condition that's harder than what they would necessarily need to do. Is that good or bad for development? I don't really know.
I would totally argue that growing up wearing different shoes and different socks, walking on different surfaces, sometimes wearing a snowsuit, wearing this kind of outfit, that kind of outfit, that is marvelous for a perception action development because babies are getting all those opportunities to learn about varied body environment relations. But in New York City, they always have that diaper on at least, and it's making it harder to walk, and so maybe it makes learning a little slower. I don't know, but that might be good in the long run.
Anyway, you get the point. So child rearing practices affects infants' bodies, and it affects the way that they learn to move. And child rearing can accelerate infants' motor development.
It can delay infants' motor development. and it can literally affect which skills they learn in the form that they take. So another kind of cool thing about Central Asia is that they don't have any furniture except for the Kevora. So everybody sits in a deep squat, which I will now demonstrate because I can do it. That's a deep squat when you can just sit back on your haunches.
They sit like that to eat, to go to the bathroom, to hang out, to rest, you know, to chat. I'm sure they can sit on chairs, they just don't have chairs. They just have coverings on the ground, and that's how everyone sits.
And some of you can sit like that, but a lot of you cannot sit like that. So the form of the movements are affected. Everything, the outcomes.
I mean, I could do a whole lecture just on cultural effects, but some of those vary. Here's another misconception about how children learn motor skills. The misconception is that milestones are meaningful in motor development.
And this is absolutely the most... most popular way to measure motor development since Arnold Gazelle popularized it in the 1920s. So milestones means a series of skills that are normed by age. And it turns out that this milestone metaphor is a terrible metaphor for development of anything in language development or motor development. Like a milestone is that thing on the road that tells you how far you've come.
calm, it's unidirectional, and it tells you how far you have to go to the next thing. This is not a good metaphor, and let me tell you why. So this figure is a very typical kind of chart of postural and motor milestones.
So they start with the baby's head on the carpet, they end with the baby running across the floor, and the skills are rank ordered from the least, you know, yuckiest skill to the most mature skill, and then age runs on. along the x-axis, and every skill has an average age at onset, and the red bars are the ranges for these typically developing children. But this notion is seriously flawed, and one problem is that it assumes that skills actually have an onset age, that that's a meaningful thing to even know or to ask, because what would that mean? An onset age means...
this time I can't do the thing and now here's a day and I can do it and now like a faucet now I have that thing in my repertoire and I can do that thing and that is only meaningful if in fact motor development is like a step function where you can't do it and then you know the faucet turned on so here's standing and this is data from one baby so this baby didn't stand until 420 days of age and then the baby stood and then every day after that the baby could stand. And this is what I'm talking about when I say this abrupt onset age, the assumption that there is something like this. There was for this particular skill, for this one baby, which I'm showing you for illustrative purposes, because it was totally rare.
This is what typically happens. So this is a different baby for standing and for 90% of skills measured, however you want to measure it, stands for one second, stands for three seconds, stands for 10 seconds, stands for 30. seconds, doesn't matter how strict the criterion is, and it didn't matter which postural or locomotor skill it was, most babies look like this. There's a period where they don't express the skill, then there's like a vacillating period where on some days they can do it and some days they cannot do it. And unless you think like they didn't really want to do it, I did this with my own daughter and I know how to elicit skills. Like they can't do it.
On some days you can do it, on some days you can't. cannot do it. So this baby shows 21 oscillations between expression and non-expression.
There's no on-off switch for motor skill acquisition. And another big problem with this whole idea of milestones is that it reifies particular skills, like crawling on hands and knees. So you probably think that when you hear crawling, you're probably thinking of that. Like, okay, that's good.
But every baby who can crawl hands and knees can also crawl on their own. on their hands and feet. About half of the babies who crawl hands and knees prior to that did some kind of crawl with their abdomen on the floor for some or all of the crawling cycle.
This kind of half crawl, half bum shuffle and bum shuffling have become far more popular now with the back to sleep movement. So that's another inadvertent effect of the change in our historical change in our child rearing practices in the West. Knee walking is also a popular thing. Log rolling is really popular. All of these things change historically depending on how cultures dress their children and, you know, what they think is cool or normal or not normal to do.
What about the sleep movement exactly? Ah, I'm sorry. That was so assuming.
So, back in the day, like when I was a baby and when you were a baby probably, our parents thought that we should be put to sleep on our stomach because we might... regurgitate milk and then aspirate it because it's all like the same pipe and wouldn't that be horrible so they put us to sleep on our stomach but then people found a correlation this is all true found a correlation between stomach sleeping and sudden infant death syndrome and so the American Academy of Pediatrics in the 90s started the back to sleep movement it's a pun because it means put your babies to sleep on their back so if you go to babies rs or something you're going to see these sleeper thingies like oh, your baby better not roll over in her sleep because she might get SIDS and die. And so all the Western countries and Australia and, you know, blah, blah, blah, like we all do this back to sleep thing. And then the pediatricians realize, holy moly, all the prone skills are getting delayed.
Well, why would that be? That is because when you put babies to sleep, whichever position you put them in, before they can roll on their own, they're going to stay in that position. And if you put them on their stomach and they're not quite asleep, they push up and they struggle and they turn. turn their head from side to side, or they wake up, you know, for a little bit, and you know, fuss a little, and then go back to sleep, and so they're exercising all of this stuff, right?
If you put them to sleep on their back, just, this is the easiest position in the world. They can't do anything, but you can like turn your head, and so like that's that. So then they discovered that all the prone skills, prone prop, you know, like propping up so you can reach for a toy, pressing up like this, like the cobra, getting on hands and knees, prone pivoting.
all of them delayed. So now they then they instituted tummy time. When your baby's awake you should put your baby just on its tummy so it can practice the prone skills.
Some people listen to pediatricians that would be educated Westerners. So if you had a college degree and your pediatrician says put your baby to sleep on his back you do and then they'll say give it tummy time but your baby cries and fuzzes because it's not used to doing it and so much work to be. So they don't don't do the time-to-time. If you're low SES or non-educated, you listen to your own generations of parents who are like, put them just like, you know, and then those kids.
So there's all these historical changes just since the 1990s because of this change in recommendations from pediatricians and which pockets of who it's affecting and so on. Interesting, right? It was evolutionary. You know, 200 years. is a nothing period of time.
So the babies were being born since prehistoric times and somehow human race survived. None of it matters. In the long run, if it's a healthy baby, none of it matters. It doesn't really, there's no evidence at all that I know of, and I think they would know about it, that if you crawled a month later or a month earlier that it's going to matter in the long run.
It probably doesn't even matter. I don't know for certain yet, but it probably doesn't matter. doesn't even matter that the cousin Tajikistan are off the chart, like delayed, for their long-term development.
Those things matter in the short term. If you can't crawl, you can't move. You can't go explore the space. You can't go see what's around the corner. And you can't do the things that those new skills would bring you.
But if you do them eventually, you're going to reap all the benefits that those skills would bring you. But the environmental factors of Tajikistan, you know, it's the pollution. The environmental degradation of those areas is so enormous.
The fat and... not where the babies are sleeping on their backs or in diapers or not would affect the baby so much more. There's no evidence that anyone has affected so much more or so much less on motor development. Yeah. The rest is still an open book.
Maybe having their face covered, you know, like their visual field covered for large parts of the day affects so much. aspects of development that's still under I mean that's still an open question we're like we're looking at it yeah so yeah this question may be a replacement is there any influence on cognitive development don't know don't don't know about the delay thing for Or exercise? Yes, there is.
It's immediate, and it's all things you would think are positive in terms of laboratory testing. But does it matter two months later? There's absolutely zero evidence that it does.
Nothing robust. Postural locomotor skills don't turn on and off like a faucet. They sputter into children's repertoires over time.
That means that there's no point. estimate for onset age. There's only an arbitrary approximation.
The items on milestone charts were selected by researchers. They were not selected by biology. There's cultures where children skip crawling altogether. It is not an obligatory skill, for example.
Walking is a good thing to do, but crawling is just totally optional. And you know, Gazelle, the guy who started all of this, he liked rolling, he liked hands-on and knees curling, but he wasn't wild about pivoting or say hands and feet curling or belly curling. He knew about them all.
He wrote tons of papers, literally dozens of papers about them, documented them in minute detail. He just didn't put them on the chart. And then like sheep, we just keep the same items that some guy chose in the 1920s for the charts that are hanging in your pediatrician's office. And then the last thing I really wanted to say is that norms don't represent the world. So the CDC norms...
example, that's our, what is that, the Center for Disease Control in Atlanta. They're normed only on US infants of legal citizens. The WHO, that's the World Health Organization, they don't even call their norms norms. They literally call them standards. They write in their papers these are standards for how development should be all around the world and their standards do not include cultures that exercise their babies.
They do not include cultures that restrain their babies. Well they don't include anyone from Central Asia. And so those kids are off the charts. Too early or too late. And so there's something I think kind of ironic, perverse, something about saying that someone's delayed or accelerated relative to us.
Anyway, traditionally in movement science, perception is treated as sort of a tag-on or an adjunct to skills like walking. And part of the problem is that, you know, most people who study things like walking, or only studying walking on treadmills or on uniform ground where perception really does have a minimal role. You could do that with your eyes shut.
And another part of the problem is the assumption that you can study those things separately. you know we're gonna figure out walking and then once we're done figuring that out we'll add in the perceptual information well okay not in my career but you know my daughter Lily could maybe figure that thing out but from a psychological view and from a developmental view actions are always perceptually guided and infants learn to move with perceptual guidance and so I'm going to give you a couple of examples so this is not a visual cliff, this is a real cliff. And it's not a visual cliff in the sense that it's just like nothing or a three-foot drop-off.
This has a hydraulic lift so it's adjustable in one centimeter increments from zero to 90 centimeters. So when it's a small drop-off it's like a little step, you know, like what functionally it's a step, but when it's a large drop-off it's a cliff and locomotion or walking at least, crawling and walking are impossible. So visual and haptic information are available to tell infants the relations between their body and the environment and infants can generate perceptual information by looking and touching so they can do something to get that information. information And experienced crawling infants, so these are 12 month old they've been crawling for several months They select their actions adaptively so they will easily crawl down drop-offs that are within their ability, but they do not try to crawl down a drop-off that's beyond their ability, even one centimeter beyond their ability.
They'll explore at the edge. They may try to figure out another way to get down, but they will not go. Novice walkers, in contrast, same age, 12-month-olds, they will walk repeatedly over drop-offs that are beyond their ability. Like, they will seriously walk over drop-offs that are beyond their ability, even a 90 centimeter. drop-off and babies do this on repeated trials they either know the limits of their ability or they behave as if they have no clue about the limits of their ability but six months later after babies have had several months of walking experience again they select their actions adaptively they can perceive precisely just as they could do when they were experienced crawlers within one centimeter of accuracy whether a drop-off is safe for walking and experienced walkers look like experienced crawlers.
When a drop-off is risky, they will stay put or find another way, an alternative strategy. Okay, so I'm going to show you some data. So zero here represents the largest drop-off that each baby could navigate.
So these data are normalized to each baby's individual ability, all right? Zero represents each baby's individual ability. The negative numbers are drop-offs that are smaller than their threshold, so these are safe by definition.
And the positive numbers are drop-offs that are larger than their threshold. These are risky. What's those BPM? Proportion.
Oh, thank you. Proportion of trials. So it's the average proportion of trials across babies.
But we get a lot of data per baby. So you can look at these little error bars. You can... could look at every individual baby's function, and they pretty much all look the same.
So this red curve here are the experienced 12-month-old crawlers. They look pretty perfect. These babies are probability matching, so they're matching the probability of attempting to the conditional probability of success.
And here are those 12-month-old novice walkers. They are vastly overestimating their abilities. They're attempting to walk on the risky side of things over and over and over again they're attempting a drop-off that's nine centimeters larger than their threshold on 75% of trials they're attempting that 90 centimeter drop-off on 50% of trials same age as the crawlers what they had failed is they had not learned to perceive body environment really they did not perceive affordances for locomotion and now I've added in the 18 month old experienced walkers and green and they're perceiving equally well as the experienced crawlers. So why should they? Why should they?
Because when you're, so when babies first start, well let me show you one more slide and then let me come back to that, okay? So what I think is happening is that infants are coordinating perception and action through experience with that perception action system. So crawling experience teaches infants to perceive possibilities for crawling, and walking experience teaches infants to perceive possibilities for walking. Wait, just wait one more slide, you guys, okay?
And I just want to show this to you longitudinally. So this is a typical infant in the first week of crawling, and so that would be coded as an error. So that's when we say they're attempting something beyond their ability.
It's coded as a failure. Ten weeks later, their errors have dropped to 50%, so half the time they'll do something like this, like correctly reject a slope that's too steep for crawling and find another way down or avoid going. By 20 weeks of crawling experience, babies have fast, efficient, exploratory movements.
They're generating exactly the right information. It's seamless. It looks like an adult, like just like I showed you in this cross-sectional data.
But when the same babies stand up the next week, now, as novice walkers, it's like all of that is gone. All of that is gone. are So they no longer are either generating or receiving the information.
They're attempting impossibly steep slopes or drop-offs all over again. Ten weeks later, they're correctly rejecting impossible slopes. slopes or drop-offs, 50% of the trials. So there are separate learning curves in development. It turns out there's a learning curve for crawling, there's a learning, sorry for sitting, learning curve for crawling, a learning curve for this is really hard.
Learning curve for cruising, that's when babies walk holding onto furniture sideways, and a learning curve for walking, and they're parallel learning curves. It doesn't get faster the next time they get a new perception-action system. Do you want to ask your question now, Neville? Did I kind of hammer it out of you?
Well, at the risk of seeming disputatious, I would have thought this evidence says that perception and action are not coupled, because it clearly acquired the action of walking. Okay, here's... Here's the thing. What does it mean to perceive affordances or to perceive body environment relations?
For crawling, that's a whole different action system that uses different parts of your body. And so the possibilities for crawling are not the same thing as the possibilities for walking. There's different systems for generating information.
There's different information that's relevant. There's different body parts for implementing the action. They're totally different skills.
So this is my answer to Dymar. Why should they? Why would we expect it?
The only reason, stop, the only reason we expect it is because as adults we're going like, it's like that Dr. Seuss thing, well, like, a four-foot drop-off, three-foot drop-off, if you're a baby, that's just risky. Crawling, walking, in the rain, on a train, in a tree, it's just risky. You know, like, you should know that.
No, you should only know that if they're learning, like, static facts about the environment or about their bodies. That would be totally maladaptive since their bodies and the environment are changing all the time. the time? Or if they were like cognizing something, if what you're perceiving is how to detect body environment relations online, why should you be learning that for a skill you can't even do yet?
Yeah, so just to spell it out, these 12-month crawlers, they've been crawling and interacting for a while. For four months on average. So they've developed this notion of bodies and affordance for them. They have developed the exploratory procedures for generating the information that tells them in this moment can I do that because if we experimentally change the crawlers body by loading them or you know putting like I have done any crazy thing you could think of you could do to a baby I've probably done it wicked knee pads whatever they said they they they recalibrate like wow I'm different now in this moment though to experience Walker with you know shoulder packs you know with lead or feather take them on and off from trial to trial Teflon on sold shoes on and off from trial to trial.
They totally recalibrate, just like an adult would do, with the same level of accuracy. What they cannot recalibrate to is a totally new perception action system. It's like, get on a unicycle, Neville.
You can figure it out. You should be able to, like, perceive affordances. No, you can't. It's a whole different body environment system. So is it worth, and, you know, you can put it off, trying to flesh out the idea of affordance.
Okay. But you're not saying they can't perceive it. It's slanted. Oh, they have depth perception.
nearly at birth. Of course they have depth perception, they're just not putting the functional meaning on it. So is it worth it, and there's a bunch of different ways of doing this, I'll just pick one. What's the reason for saying they don't have the affordance to say, you know what, they can't predict what that movement is going to do in that environment? I like both of those.
That's my two. Do they map? I think they do. It depends. At that level they do.
Just behaviorally they do. They're not predicting accurately. They really don't like following it.
into a precipice. And so they are not predicting accurately, they're not making the decisions, and I am inferring that that is like they're not perceiving what they need to perceive. Yeah, and how you think, I mean the prediction is one way of thinking the importance, but that would make some predictions about what they would try and what they would learn and what they wouldn't try and what they would do on the next project.
I think so, I think it would make the same predictions as if you used a foreign language. Do you have a question? Sure. So we're seeing that the motor skills are culturally, yes, but sequenced through age. So what I'm sort of wondering is, besides just using different sets of muscles and therefore different proprioceptors, what's making, say, walking more complex than falling, that it takes longer to learn it or it has to be learned separately?
Why aren't there transfer effects from learning? Okay, you're asking two things. So why are the milestone charts ordered?
the way they are. So one thing that's different about walking than crawling is it's a different balance control system. So it's not like crawling has no balance requirements.
It totally does and crawlers fall all the time just like walkers do. New crawlers, I mean, fall just crawling. flat ground the same way as new walkers will fall walking on flat ground.
I think they're different perception action systems. I want to show you like one more little bit of data. So that baby is wearing a head mounted eye tracker and maybe one reason way infants fail to transfer what they learned as a crawler to upright walking is that they get a new vantage point on the world.
So the transition from crawling to walking literally affects what babies see. So this bottom panel is just going to show you a baby crawling. The top panel is going to show you what that crawling baby sees while crawling.
And the crow's hair is the point of gaze. So this is what babies see while they're crawling. They see the floor right in front of their hands. That's the information that they're getting.
And this is what a walker sees while walking. Or, you know, so in an upright posture, um, infants see the whole room and all the stuff that's in the room. Um, and, yes, go ahead.
Sorry, no, I'm just gonna say like even at the same age or even the same baby Well that baby goes into a prone position to crawl the baby's seeing the floor in front of their hands Well when they stand upright to cruise or pull up or start walking or whatever the whole room sense of you. I come to it from vision point of view and it's very interesting that infant has to learn to calibrate their stereopsis because they have no there's no other absolute way stereopsis gives you a ratio and you have to know the distance of your eyes so the only way for them to calibrate is to reach okay and so they have experience with reaching earlier and they have experience with Yeah, with the promotion. So I was wondering, of course they can already tell the whole lyric of depth, but not the magnitude.
I don't, I mean, I'm not, I don't have at the tip of my tongue all the data on, like, the development of depth perception in babies, but there's so many depth cues, including stereo, and many of them are in place before the average baby starts to crawl, and certainly before they start to walk. But these are quite impoverished in some way, because you don't have any of the 3D cubes like Lousian and Parallax. What's impoverished?
What they see here on the cliff or on this kind of situation, it's very impoverished for 3D. It doesn't have the other cues. I don't think so. I mean, what you saw on the cliff, I did the whole checkerboard business. They're getting occlusion, they're getting motion parallax, they're getting texture gradients, they're getting, no?
The checkerboard doesn't have the occlusion, for example. And even motion parallax needs to be calibrated. Thank you.
So it's an interesting question. Right, well, so, yeah, I mean, I think you're right that there is a recalibration process. I think you're wrong that the infants don't have access to depth cues when they start to locomote. I think that what needs to be considered is that action is the only way to calibrate vision. That's my point.
I think I agree with you. I don't know how else they would do it. Yes?
Do infants try something walking that they're not successful? successfully capable of doing crawling. And the reason I ask this is because one potential explanation for this would be that you calibrate crawling and then you have just like a miss, you're overly confident that you can accomplish walking when you've been able to accomplish it.
Oh, I see what you're saying. Yeah, absolutely. So like for the crawler they can't crawl over a 90 centimeter dropout.
And then the same baby will do it. And I don't have it right here, but it's actually, the learning is so specific that babies will do it from trial to trial. So if you take a baby that's...
just learned to walk and put him down in his old crawling position, if he stays as a crawler, like stays down, they perceive affordances perfectly for crawling. Then the next, like a moment later, you stand him up at the top of the big drop off or the big slope or whatever, and then just walk off. And the terrible irony is that a little more than half the time, if you take a walker and go like, just be a crawler where you can perceive affordances, you know, when you put them in crawling, they're like, yeah, I'd rather be a walker.
Alright, so this cartoon shows all the results of that eye tracking study. The schematics are drawn to a scale. They're showing the average field of view. That's all grainy, but whatever.
For crawling compared to walking, so while babies are crawling they're seeing mostly the floor, while they're walking they're seeing the whole room. Why is that? That's because vision is embodied. So when babies are crawling their head points down and think about like the shape of your neck.
So when they're in a crawling posture, even if they crank their head up as high as they can go, they can still only see the bottom of the far wall. And we know that because we put motion tracking. markers on their heads and necks, but while you're walking, you can easily turn your head up to the ceiling or down to the floor. So learning to walk or any other kind of action is always perceptually guided.
What are babies learning? They're learning to perceive and to exploit affordances or possibilities for action, whatever term you like. And part of that learning includes discovering and honing the right exploratory actions so they can generate the necessary... information for perception.
So I think part of it is like actually doing something that generates motion parallax and knowing to do that thing. Learning from earlier developing skills doesn't transfer to later developing skills. Why not? Because new skills create new body environment relations. New skills create new information gathering systems.
Crawling and walking are different in terms of their vantage points relative to the ground. They're also different in terms of... haptic and proprioceptive information.
There's different forces involved. There's different body parts involved in moving. Excuse me.
Yes. Is it known at what age, it's about the previous slide, at what age the learning transfers into future skills? So far as I have found, no age. So at one point there was a grant, I wanted to do a grant, I was like, well what after walking? Like could I do something with older kids?
kids or even with adults and you know and it was like introspecting like you know we all transfer when you rent a car if it's kind of like but could you transfer to like a great big semi truck or what if it had more gears or like what is it that is or isn't you know like what where would that line be and I want so I thought well swimming could be that's not happening in New York City like no one has a pool but I thought the whole unicycle thing and I had a colleague who's a unicyclist and couldn't even juggle it and stuff. We tried it. It was just such a practical disaster of unicycle, people flying around the lab, not a good plan. Other things like roller skating really seemed to be just growing out the upright perception action system.
So I landed on brachiation. That's like when you go hand over hand on monkey bars. And you can do that with starting in like kindergartners and all the way through adults. And so it was totally cool and fun.
We had monkey bar Mondays, and so all the grad students and like physics professors and everyone on my floor would come and, you know, brachiate on the monkey bars. And it was also kind of a cool paradigm because it sort of has this weird parallel to walking. So like your five-year-olds just hang, which is kind of like, oh, I'm standing, you know, and then they can take like a little stuttery step and then they cruise sideways on the monkey bar apparatus before they can and they mark time. time like that before they could just alternate.
And the really good people have flight phases, so they can do ricochet abraciation, which is like running, like you thought your nine-month-old could do, but like, you know, they're like both hands in the air between bars. But so you studied that as a, as a task for, to test transfer from walking to something similar? Well, just, just, all right, you know, like Jagmar, go on the monkey bars. adjustable and you tell me you know like so we would give these different I just ran it the same way I would run anything else so we'd offer bars you know that were easily safe and then there'd be like a bar that was like you know oh so I'm just like it was it was yeah it's just like the same the same idea that you had to like the people had to decide could they get to the next monkey bar or not um same way as we were doing with the babies and adults and kids horrible like absolutely no no idea what they could do on monkey bars. And it doesn't seem to matter if, I mean, at some point, your kindergarten monkey bar experience probably lost to you.
I don't really know. But unless the kids were monkey barring, rakiating, couldn't do it. Yeah.
It's really a thought that has to do with the ability to accomplish dual tasks. So really early walking is going to be very common. And walking and perceiving would be really difficult to do.
I'm sort of smiling at you because that's like a very popular idea among psychologists who listen to me give a talk. I don't know what you are, but I've heard that a lot. And it also is sort of like...
this unfalsifiable idea like, oh, they actually did perceive affordances, but they just didn't have the extra capacity to inhibit or the extra capacity to also generate perceptual information or whatever. I mean the bottom line is as a new Walker Babies tend to keep their head up like Frankenstein So they're not getting more information for free the way a crawler is on the other hand colors are getting it for free Like okay, there's the floor. Oh, yeah, and over they go so Maybe like capacity is getting used up and they kiss can't process they didn't lose depth perception You know and they're not learning a static thing if you do did, then you would know it and it would transfer to every new skill and development. I mean, we've done BBs longitudinally. Sit them on the edge of the drop off as a new sitter.
Want this toy? Sure I do. Fall in, you know. A few months later, as you track them out, they totally, okay, outside my, you know, sway.
Now they're crawling. Or you do them in the same session, sitting to crawling. You know, like it just doesn't transfer.
Like they know something in one posture but not in the next. And then they relearn it again and relearn it again. So I think that speaks most strongly to what they could not have been learning in the first place. You can fit it in without screwing up the talk. Do you have results on how the things you've been talking about in the last few slides change cross-culturally?
Like so the kids who pick these things up early, how does it affect these later? I know. It's so cool.
We don't know. Yeah. Is there any sort of incentive not to jump off the cliff if you're a baby in these studies? Or even if you're doing the monkey bars in the lab, is there some sort of reward if you choose the right case?
Because I could see that if the baby just walks off the cliff... I get where you're going. And they don't get hurt? Stop. I get it.
Did you guys all get it? Well, like, they didn't die or get hurt. You caught them.
So there's no penalty for error. Maybe that's the dealie, right? Okay, so here's the answer to that.
I don't actually care. I didn't care. I mean, I've been doing this a long time now.
I care because I know all the answers. But in the beginning, I was like, that's a day to speak. You know, I didn't care.
So let me tell you the answer. And this made me think something else, and I forgot it. All right. Right, penalty for errors. So if you, alright, so first of all, like for drop-offs and slopes, something like that, if infants were just learning, that nice person's going to catch me.
And that could be true, because that happens all the time. time and you know like at home like I'm jumping off the counter mom you know like that and all that stuff um then you shouldn't see the babies getting more accurate over sessions like they should false alarms should stay high like they should continue to try the risky stuff. They shouldn't be gearing in to their own abilities. They shouldn't be saying no more and more often. Not when you test them longitudinally.
When you test the same baby in the same session in two different postures, blocked or interleaved. So an experienced posture and a new, unfamiliar posture. They try it and fall in the unfamiliar posture, but when you test them in the experience... posture, not going to do it. And it doesn't matter what they get first or whether you block the conditions or inherently the conditions.
All of that is a really strong argument against the penalty for error. There's two more little points about penalty for error. Visual cliff, beautiful design for testing rats who have vibriscy, you know, and are really sensitive to what is that called? Like the airflow and stuff like this. So that's why Eleanor Gibson had glass on both sides.
and then it got co-opted into this baby thing and I guess the original idea was to keep them safe. You can test babies one trial on the deep side of the cliff because it creeps them out but then they can feel that it's a solid surface and if you have a mom or a toy or snacks at the other side you get one chance on that baby because the next time you put them up they go across. So that's one reason I didn't do it.
I didn't know if this thing with having an experimenter touch them would be kind of like the safety glass. I was lucky. It turned out it's not at all like the safety glass.
We stand behind the baby, we don't talk to them, you know, blah, blah, blah. And they'll just do whatever they're going to do over and over and over again. Okay, satisfied on that?
Plus they do it outside the lab. That's why you don't leave a kid sitting on the counter. They can sit there all day. Apparently not in this culture, but in other ones.
Thank you. All right, so I think the biggest wrong thing that people have done in the study of infant walking, and really all of walking, is complaining. gait with walking because gait is rarely walking.
Natural walking is walking and I think natural locomotion is the phenomenon that we want to understand and explain. So before I showed you you know in that diaper thing like a baby walking down the pressure sensitive mat but that thing that's the standard way of testing babies. Once a baby can walk they don't they won't let you hold them on a motorized treadmill anymore.
So you test them on the straight path. test. But when you do that, a lot of the time this is what really happens.
So the babies weave around, they stagger, they fall, they do other things like go off the carpet or play with their diaper. And you know, even when they go to the end of the carpet, they don't always go there in a straight line. Babies really have other goals entirely. So this is a schematic of my lab.
This was the carpet at the time. And this is what a baby's actual path looks like in 10 minutes of spontaneous activity. This is the actual path of a 13-month-old novice walker in 10 minutes of just baby you can play.
Natural walking looks nothing like that straight path task because it is varied. So here's data from the traditional straight path. path tasks. Here's data that we also collected on the pressure sensitive floor and so is this and so is this.
These twisty, turny paths is what walking looks like when you just let babies walk the way that they want to walk, however, wherever they want to walk. And so with a large instrumented floor, we can record and quantify natural gait just like we would do in the traditional straight path task. So the baby's just play, if they happen to step on the floor.
If NIH will get its budget, then I won't instrument my whole floor. But anyway, babies just run around and they act like babies. And then we opportunistically can measure their gait during natural walking.
So you can measure the exact same parameters, only it's for real gait, natural walking, not that artificial gait during the straight path task. So I already told you that longer steps, narrower steps, for example. mean more mature walking.
These are typical data. Each dot is a different baby in the standard straight path test. These are natural gait data during free play.
So step length gets bigger, step width gets smaller, and these are the same babies. Okay, and now I superimpose what Neville wanted. The babies show the same developmental progression, so they tend to walk faster when you do the straight path.
path test, which explains the difference in step length, step width. There's actually no difference at all. But the point is, the developmental progression is the same. So if you're doing the straight path test because you want like an IQ test of the maturity of the baby's gait, you can just like let the baby walk around and you can get their gait the same way.
And gait measures during free play validate the gait measures in the straight path test. Well, phew, because I've at this point... point contributed more straight path data than, you know, Gazelle or any of the rest of them. And so, yay that individual differences in walking skill are stable across the two tasks.
So, this is critical. Nothing is lost from the straight path task if you just measure babies during free play. But lots is lost from natural walking if you only do the straight path task.
So... Much of natural walking is really about starting and stopping. It's not continually going.
So this is a one-step bow. I don't know if you're comfortable calling that walking, but it's something, and babies do a lot of it. This is another one-step bow.
And this is a two-step bow. And this is a three-step bow. And this is showing you that babies walk in little bursts of activity.
Their motion is distributed across time. So there's little bursts of walking separated by longer periods of rest. This is a roust or plutt.
Each row represents on some different baby. They're ordered from the least to the most experienced. And so you're seeing that every baby, no matter how well they walk or how much experience they've had walking, it's not like a death march. They like walk, they stop. They stop, they walk, they stop.
And those really short bouts of only one to three steps are common across development. It's not that it's like, I'm a bad walker, so oh, I can only take one step and I have to recover. They're doing that 10 months later when they're asymptoting in their walking skill.
So these babies can take longer steps, and they often do, but a lot of the time, they're only taking one or two or three steps. And here's the thing, those little, tiny, short bouts It's all about starting and stopping. It's not about continually going. Most FBABYs bouts contain omnidirectional steps.
So they include forward and backward and sideways steps. So here I'm playing it at half speed, and you can see what I mean. So sideways, then forward, forward, forward, and then we didn't know how to code it, and then backward and backward. And here's a 21 step bout that's only 50 steps forward.
50% forward steps. All the other stuff, it's not forward stepping. They're omnidirectional steps.
And these are babies that can walk. So only less than 20% of infants' bouts are all forward steps. Most of babies' bouts are a mixture of forward and not forward steps.
And 25% contain no forward steps. And that's true regardless of the bout length. Even short bouts, they're not all forward steps. It's true across development.
That means that omnidirectional steps is not a characteristic of poor walking. Oh no, I have to step backward. Babies are taking omnidirectional steps in their first week of walking and they're doing it ten months later.
Most of baby's paths are curved. They're not straight. Only 25% are straight paths.
75% are curved paths. That's true regardless of the length. And it's true across development.
development. And all the little belts of curvy, you know, omnidirectional walking adds up. Real walking is immense. So this is the accumulated number of steps across walking age that babies take per hour. The circles are babies tested in a lab playroom.
The squares are babies tested at home. They look the same. The average baby is taking around 2,400 steps an hour. Extrapolated out, that's 14,000 steps a day. and they travel immense distances.
So this is like if you strung all their steps together end to end, they're averaging 700 something meters per hour. Like we don't do meters. So how much is that?
That's eight football fields an hour or 46 football fields, American football fields in a day. So natural walking has really important features that are not captured in the straight path task. Natural walking is not a long, continuous string of steps.
Instead, there's lots of starting and stopping. Gate initiation, gate termination, that means that babies are having to learn and then do. We generate disequilibrium, maintain dynamic equilibrium, and regain dynamic equilibrium. It's not only forward steps. So they're generating disequilibrium in every direction.
and recovering it in every direction. And it's not only straight paths, and I love that, because if you do a curved path, it means the two sides of your body are doing different things, not the same thing. Like, it's really hard to have a CPG where, from the beginning, you're doing different things with the two sides of your body.
And their experience is immense. Like, their practice measurement is immense amounts of time-distributed variable practice. If you could set out to design a system based on anything we already know from motor learning, you'd be like, well, give them a whole.
lot of time distributed variable practice and that's what babies are getting and there's no change in that over development so how do babies learn to walk the traditional approach so I of course I totally get it but so then by allowing them to walk in their own way what more insights do you get about say the development what is how do you how to take all that information that these data give you and what conclusions do you draw about developmental processes or other things of interest? Okay, so I'm about to say that right here. So I think that's the right question and we're like right on top of it. The approach and the theory from this approach, literally from the last hundred years, because people have been studying baby walking really well since like the 1920s, has been all focused on alternating leg movements.
So at first babies can alternate their legs. Oh my goodness, they can do it as newborns. They can do it in utero.
And then they get out and they figure out how to keep balance. And then they walk in a straight line. And we call it periodic gait.
And then we have developed theories about alternating leg movements and are they in need? And I think that all of that reflects a problem that we've had throughout psychology, which is we invent a laboratory task. We turn it into a phenomenon. We reify it as something important. to be explained.
And then we invent theories to explain an artificial phenomenon and it wasn't even the thing you wanted to explain in the first place. Because that's how babies really learn to walk. There's nothing to do with a straight line. There's nothing to do with periodic gait.
And so any theories that we have have to take those characteristics that I described to you into account. There's actually lots more about the development of walking besides the gait aspect of it. I'm really interested in what makes them... move in the first place?
Like, I'm interested in what makes them want to take their first upright steps or their first quadrupedal steps. And then once you can move, why move at all? And are there things in the environment? Are there environments that make locomotor exploration more conducive or change the quality of locomotor exploration?
And NIH will get its budget. I really think I'm positioned, finally, to test what I think is the real hypothesis. that everyday environments, like everyday life, in Tajikistan or in New York City or wherever, is really well designed for babies to learn body environment relations.
And so I think we, you know, could do a training study where we give babies, like, really intense body environment relations and see if that's what is helping them to learn to perceive affordances, which I think is the whole point of learning to walk, is learning to get around and make decisions and modify your gaze. One more time, consistent with what you're showing. I mean, if you take what you're saying seriously, it holds for adult walk as well, right?
We're not walking in a straight line. And actually, I was reminded of some work in postural control, where half of the world in postural control looks at 30 seconds, still as quiet as possible, standing on the force plate, and then analyzes the sway. and one of my colleagues started to look at half an hour standing on a horse plate without any constraints other than don't step off it.
And it was revealing. And the interesting part there was that one of the results was that, say, for elderly you usually get, you know, there's more variable, but if you look at the half an hour sway, it was the other way around. So there were actually very interesting insights that were not at all anticipated.
by letting in the more natural behavior to come into your experimental data. So I couldn't agree more with you to allow that, but I'm still sort of pushing what exactly do we then learn from allowing the richness to come into play. Well, here's one example. So Peter Stone is a roboticist at UT Austin, and he uses robots to model the functional...
consequences of varied walking experience. So he builds real robots, he also builds simulated robots. So there's no human operator, you know, guiding the robot.
You put in the learning algorithm and that's what the roboticists are working on and then you just let it go. And the traditional learning algorithm in robotics is really like the traditional approach to how babies learn to walk. Get the robot to stand, get the robot to walk in a continuous straight line. let's get really complex and add in some turns and add in some omnidirectional steps and so on.
But you can see he trained his robots on curved and omnidirectional paths with lots of starts and stops from their very first walking step. And then how do they decide if the robot is good or not? Well, they duke it out in RoboCup, which is like robot soccer, and they have different divisions. And the UT Austin military...
team won the 2011 Robo With 24 wins and 0 losses, 136 goals, 0 goals conceded, so that was a revolution in RoboCup because they were the only team training robots to walk like babies learn to walk. And then they won 2012, 2013, 2014, 2015, 2016, 2017. And what I think is so fabulous about all of that is that the robots are evaluated in a functional context. So it's a variable environment that's filled with objects and surfaces and other agents.
and they have goals, like they literally have goals. Isn't that funny? And so, all right.
So we paired up with Peter's lab to use RoboCup to assess the functional consequences of varied experience in infant walking. So in this tournament number one, we asked whether training robots on geometrically shaped paths or on a baby, actual baby path, would be good or better for... winning RoboCup. And the short answer is that babies win. So across a thousand head-to-head matches between every pair of teams, babies show the largest average goal difference.
And in tournament two, we clustered infants' paths into groups based on the variability of their walking. So in terms of their path shape, the directions that they walked, the path length, and the number of starts and stops. And this purple team is the most varied on all men. So varied means sometimes straight, sometimes curved, sometimes forward, sometimes back.
Like you have to do some of everything to be on the purple team. And tournament two showed that more varied experience also has better functional consequences. And the purple team had more wins in RoboCup. So natural walking is walking.
And natural walking can reveal the same thing as the traditional approach with that straight path test. but plus much, much more. And I think that any theory of natural walking, either its development or in adults, has to take, you know, omnidirectional steps, starts and stops, turnings, blah, blah, blah, into account. Walking improves because infant's natural walking includes immense amounts of variable time distributed practice. And this is my last suggestion, and it's about video.
So there's lots of ways of recording movement. and its development. And for every kind of action, every kind of action you can think of, video provides a richness that is not captured by the other methods.
So video as an adjunct to these other methods adds things that those other methods cannot capture. And it provides face validity of other recording methods. And so for researchers who want to use the richness of video to inform their...
work. All the videos that I showed you plus tens of thousands of other videos are shared in this DataBury web-based library. The aim here is to promote open sharing and reuse of research videos among behavioral scientists and video is uniquely suited to document procedures.
So all those nuances and details that you can't possibly put in a method section because you're not allowed to but even if you were allowed like a textbook to describe your actual methods. Words don't do it, but a short video clip of your procedure could do it. Video is also uniquely suited for reuse. That means that people can ask questions beyond the scope of the original study. They can ask questions that the original researcher had never even imagined.
So if we have videos of the beautiful studies you were showing me, I might be scoring, or you know, like Lisa might be scoring their facial expressions when they're catching them. the ball and she doesn't really care about you know whether they caught what you know what I mean anyway and so by sharing and we using videos we can increase scientific transparency and speed progress and data Brary automatically transcodes videos into preservable formats into the indefinite whatever forever for the future so I'm offered these suggestions for new ways to study how infants learn to move We should recognize that it's not just human and biomechanics, it's in a social and cultural context. That we should be thinking more on the metaphor of scenic vistas, stopping points along the way rather than milestones. That perceptual guidance of movements, of actions, is integral right from the start.
That we should be trained to capture the natural phenomena that we want to study. Thank you. have the technology now.
...movement scientists to do that. And then finally we should use video to capture movements and we can share and use that data and that will make our science go faster and make us better. Thank you Karen.
Yes. Other things? Sorry, go ahead. In the robot experiments at the end, I'm assuming those were visually simulated robots. Those were simulated robots.
Were there any experiments done? You showed the transfer type of path that you trained them on. But were there any terms where then you incrementally altered the capacity of the robot? So what are you training?
you know, if you pull out a capacity, you know, when this, the variability of turns change how much it happens. So the learning, okay, so I thought you were going to ask about have we changed the robot's body and like make the robot develop? No.
But they can make the robot learn. And so its capacity is changing. So one thing we're doing right now is just testing different infant, different infant walks. I guess.
multi-joint coordination no like they fall less they get to their destination faster so their capacity is changing in each iteration of the what is that called? each epoch of training with the robot camera guy you had a question? I love Linda Smith's data set on object naming and currently with it do we have anything like that for like strategic either on vocalization of a parent maybe or yeah it's just such a cool data set. We do, like okay so language I don't, when I collaborate with Catherine Tamas- Amanda whose expertise is in language, we published several papers or I don't know more than several about things like that.
So one kind of cool finding is that when crawling infants want to interact with an object with their mother they just hold it up and the mother, the most common response is to ignore the baby, the next one is to do an affirmation. I'm like, thank you, that's right, okay, you're alive. Or the last is referential language. That's an orange ball, ducky, something like that. When a baby, could be a crawler or a walker, but it's just way more frequent in walkers, picks up an object, carries it over to their caregiver, shoves it in their face.
Now that's much less likely to ignore or just occur. And referential stays the same. I'm like, really?
increases are action directives, function words, stuff you can, you could throw that orange ball. Do you want to throw the ball with mommy? Let's read the book together. Shall I cover you with the blanket?
So they use now language about what you can do with the objects. So you get this rich informational reward from being able to walk over and hand it. Yeah, yeah, yeah, yeah.
And I love your question because Dagmar talked about the, you know, whatever it was, 58 subcontract grant. Yeah. And that's exactly what we're planning to do.
So some of those PIs do locomotion, and they're coding for whether the baby's movement. All of it's transcribed. Twelve of the PIs do language.
So we'll have it all, and it'll all be shared, and you could jump in. It feels like there's some folks at U of Rochester that are doing that too. Their names are something like, although the Connect got cheaper enough for everyone to be able to put that into their lab, I think. The Rochester people that were doing that have... now left Rochester but yeah and you know so Caitlin Fousey is part of this consortium and Ellica Berglason is doing stuff like that and you know yeah so yeah very cool yeah.
I don't know about toe walking, if that is a sporadic thing or not. I would guess, like if I had to just bet money. I'd bet like some decent money, like $500, something like that. That's butter? That's butter, yeah.
That it'll start out with most of, well, I don't think all of the steps are ever all toe walk, but you might get no heel strike, like heel toe walking. And then you'll start to get it, but it's not going to be like a step function, like, why was I walking flat-footed or toe-walking? But I think it might be true with adults as well, that depending on...
Well, depending on the age, if they get excited, they're often going to get... tell all the humans that the ritual healers. Yeah, yeah, that's a good point. So, you showed the cultural data with the time, time debated. without intense delaying certain development, certain skills.
I learned that with preterm infants in the NICU, many of them are swaddled, so that there is no movement possible. So given your observations, what leads, why are they swaddled, why do your observations or results not influence that they maybe not be swaddled? Because I... I'm stunned by that. There's been, since, I want to say 1940, I think, is the first Wayne Dennis study of the Hopi Indians swaddled on the cradle board. There's been like a constant, not huge, but constant stream of papers about swaddling.
And a recent one is a... actual experimental paper with babies randomly assigned to a swaddle-no-swaddle group, there are no, there's no robust evidence that swaddling per se is bad for development, including on the cradleboard. Why might that be the case?
all the natural just swaddling like the Hopi cradle or whatever, it titrates sort of right with the Western norms for age when you would do things. So being swaddled as a total neonate, like, eh, but then there's less swaddling every month as the baby gets older and less swaddling during the times when a baby would be moving. What's really unusual about the Gabor cradle, is that it's a little bit more of a swaddling. is that like a third of the babies are not out of the cradle at night. Like I think our intuition, my intuition anyway was like, well, if you're going to do it, you would do it when the baby's asleep.
But that's not the case. It's a way of minding the baby. It's like...
paper, playpen. It's all those, they're using it for all those functions at once. So they're restraining movement when in a baby that should be old enough to be sitting, crawling, walking, and when they're awake.
and could be doing those activities. But maybe I allow myself to add one more thing. Sorry. I think there is evidence, and this is partly what we are looking at, that there is an interaction or a relation between movement and other physiological functions, such as respiration.
You raised the idea of sleep. And so while there might not be any visible effects on learning to walk, walk, there might be an immediate interaction with other physiological function. Totally right.
And that might lead to infant death, so they don't even enter the long-term study or other more severe outcomes. And I'm just trying to grapple with... I know, I know. It's really, really interesting.
It's really complicated. It's really sensitive. So I'm really reticent about saying in print. that this is a bad thing, and if it's bad, in what way it's bad, or what that means.
And there's all this other stuff that's, like, in the Gavura that's confounded with it. So having, like, a heavy-wooled blanket over you, limiting visual input, but also breathing inside this little space in the cold months, and it's interesting. Like, I don't know what will happen, but it's very... very possible that we could do a true experiment in Tajikistan. Because since the Gavora cradling is so part and parcel of their, I mean we couldn't do that study here in the US, but we could offer parents, you know, like, I don't, I'm not, I'm not sure.
It seems like we may be able to do it with random assignment and say like if you're in the not Gavora group, we'll buy you diapers or whatever, you know, for the whole period and something. Yeah. So, obviously, baby infants have the plasticity to develop with regard to the environment and external assistance and instruction of whatever kind. What if social effects just are removed in a hyperpacific way?
People say she is a baby like a monkey, you know, who is removed from its social environment, cultural environment, whatever academic knowledge that goes into rearing. I really think they would not get the universal... skills.
I get your thought experiment, but I really believe they would not get the basic skills. There's some evidence that suggests that. There are wolf child children, Mowglies.
hard because it's not your experiment. Your experiment is don't confound it with you know all the other stuff just you know no. But there are some cases where so there's a family in rural rural Turkey. There were 17 children that the mom had.
15 of them alive when they were discovered. Some of them had cerebellar agenesis. you know, partial development or no development of the cerebellum. Some of them didn't.
I mean, some of them had a full cerebellum. And apparently the first kid who was born just was really, really like crawled for years, like three, four years before walking. But then she's pumping out the kids.
So then the next kid sees this kid. And I didn't meet any of these people, but I met the film crew that went to film them. And the whole thing is just this amazing story. They were first discovered by a Turkish researcher who, this family speaks Kurdish, so his initial reports were they can't even speak. Because, you know, they don't speak Turkish, but they can't even speak.
That was before the de-evolution. He named their syndrome with his own name. Neutron Syndrome. And then he said that it was like reverse evolution. You know, everything stupid that could be done and said, done and said.
So the BBC heard about it and went there with a film crew and spent some time with the family and interviewed them. And what on their reports and on their recordings what happened is the parents were like well so they were like you know because i think like if your child had cerebellar agenesis it would walk like children with cerebellar agenesis walk because what do we want our kids to walk talk and use a toilet like those are critical things and and these parents were like well it's the will of allah so they didn't walk that's good and so they're all these adult siblings just walking around on their hands and and feet crawling around like bear crawling as adults. The BBC I think felt guilty about the whole thing, it was like verging on exploitive or whatever, so they brought PT's in to like with stuff to help them get up right and you know they practiced but they were like really good at crawling around on handsome feet and basically did it.
One of the siblings like we microanalyze, well actually five of them we microanalyze their gait and one of them A male. He was just amazing. His whole body was changed from having spent his life crawling around like that. And he had, like, the most beautiful, fast, like, gallops that were just smooth. You know, it's, like, just amazing.
But also, like, he could have been walking. I don't know. The women were really troubled because they had to wear long gowns, and they were heavy, and they had, like, no bras and stuff.
So they had all these issues, like, biomechanically dealing with all this stuff. So. And your point about studying learning by having a rich natural environment.
But adult locomotion is particularly distinctive in. that it is good for endurance. So for example, humans probably have a lot of the persistent subjects.
And in endurance walking, it's not clear to me that you're doing a whole lot of all of that exploration. So if you run the New York Marathon, boy, I don't know. You don't.
No, no. But persistence hunters, they're running over rugged terrain. That's what they're doing.
The Tarahumara Indians are running over irregular terrain. I understand. So it's not like free exploration like a baby.
That's true. But it's not treadmill walking. Well, what I'm getting at is that, for example, with the babies exploring, it seems to me that... what the robotics people call capture point walking. What does it mean?
It's basically, you aim each step so that you're some percentage of where you'd be. If you did nothing else, you'd come back to upright balance. It's kind of like a sequence of steps rather than a rhythmic pattern.
On the other hand, my impression is that for endurance locomotion, there does seem to be rhythmic behavior. And so my question is, does the CPG play any role in walking? So based on the baby studies, you can say, oh, when you...
needs a CPG. On the other hand, if I go to persistence or endurance locomotion, that's so clear. What's your opinion?
So do we have CPG or not? No. I would say that it's a superfluous construct.
So it's the same position that James Gibson was in when there's this whole field all about sensation. It's a field, and it's all about sensation. Is it relevant for how perception really works?
His argument was probably not. It's just superfluous. It's kind of like masturbatory science.
That's where I would put the CPG. So despite the fact that we have clear evidence of CPGs in quadrupeds, I admit that there's some evidence of CPGs in humans, but it's really hard to find. So it seems like it's irrelevant.
Okay, but when the CPG construct becomes so flexible that it can allow for all the complexities of real walking, which is, you know, I mean, I don't claim to be a CPG expert. I'm totally up on the argument, but that was my take on it, is that... You know, okay, fine, a decelerate cat on a motorized treadmill, you know, whatever. But, so, you know, passive robot walkers, you know, whatever. That doesn't mean you have a CPG or whatever.
cat that can like know that his front leg stepped over something and now he lifts his hind legs even though his front legs are stepping over the second obstacle to have a cpg that's flexible enough to deal with that stuff i'm not sure that it any longer makes sense to like have the construct of the CPG like you already have all the other stuff like I'm not sure that I think you can dump the CPC so to pursue this my esteemed colleague seems to think that rhythmic behaviors are really important elements of normal behavior so I understand you're saying that it's probably not relevant no I just I think that there are rhythmic behaviors I think that sometimes walking is rhythmic you know I mean babies also do rhythmic spunting behaviors. But I don't know if it's like, it's a convenient and productive model for pursuing certain kinds of ideas. Is it going to provide us the deep insights into natural action?
Maybe not. But does everybody want to explain natural action? Maybe not.
I don't know. So how should I respond to that? I don't know. What's natural insight?
with natural insight does that data show about the, I mean it looks just like a random walk, it even looks it is, well it's a levee flight then we have a whole that's what our analysis the best, yes the best fit looks like a levee flight but I took it out because I said that at another meeting where Dagmar was at and people and it's not my work, that's all I know about levee flight, foraging animals, optimal ... I can describe it, but I didn't know the data well enough to... But it does.
It looks like a lowly flight. So what deep inside does that give us? It's just a random walk.
At that very early stage, you're learning to walk. You need as much experience as you need. I mean, that sounds very natural.
But does that explain how do we eventually learn how to walk like an experienced walker? So, it depends what you think learning to walk means. I think learning to walk or learning to reach or learning to move anything means learning to move your stuff in a way that's adaptive, that allows you to do...
the things you want done. And so to me, a better endpoint for learning to walk isn't like mature baby gait or whatever. It's like getting through an environment. So my tests would be some kind of test of perceiving affordances. or something like that, gate modifications in a complex environment, you know, efficient use of visual information or long distance source, something like that.
Like the John Matta stuff, I think that's a better endpoint for people studying. ...learning to walk. Or like the persistence hunting thing.
I think that would be, I mean, those are, like, why is the endpoint undergraduates on a treadmill? Or, you know, like, maybe that's 10 months of walking experience. Like, that's not enough.
endpoint we just made that up like those are just convenient that's just like white people in our intro class you know so like we're not the endpoint you know so like I think that would be really cool to study how you or load carriers like those load carriers that can do like 30% of their own body weight with no increased energetic cost they like totally change the biomechanics of walking well maybe we should study that as like learning to walk Some of us saw a great talk by Nsensi Kechang last week where he talked about certain invariants in coordinated muscle movement that appear in infants that's like adults and then it just appears during the infant learning to walk phase. and then it reappears adult-like. I don't know if I'm planning that correctly. I don't know about this. It sounds fascinating.
I think it's in prep, but yeah. I'd love to show you some of the slides. Okay. But yeah, synergies that only appear when a child is learning to walk. That sounds fascinating.
I don't know anything about it. Yeah. When you showed that milestones don't exist, and for the case where there was a lot of...
What happens to standing during that case? I saw it was like... hundred days of work of when you had learned so one is what do you know anything what happens on standing and second is you say that some days they just cannot walk and suppose we elicit them we make them stand they just fall or yes Yes, yes, yes.
So if you try, if you have a criterion for, you know, and we'd have multiple criteria for, you know, I think like 40 different skills. It doesn't, like none of it mattered. It was just really, really rare to get a step function. And I think the real takeaway on that story is that if you want to know what development looks like, you have to know what it looks like. you have to study what development looks like.
Where does development happen? It happens in an individual, not in a group. Development in a group is sociology.
If you're a psychologist, you have to study development in individuals. That means studying the same person over time. And we have assumptions about how development works.
Like, most people assume that growth, you know, baby growth, child growth, growth through adolescence, is continuous because that's how we see growth represented on a growth chart. That's a math. smoothening function you're looking at and worse it's cross-sectional data that have been mathematically smoothed. So growth it turns out is a whole lot like that vacillating thing. It's episodic so babies will go days or weeks on end with zero growth and then they shoot up in the course of a 24 hour period.
They can grow up to two centimeters in height a baby in a 24 hour period. It's huge, huge. And then and then it's flat again. And then, boom, it grows.
And then it's flat again. And so that's cool and funny and whatever and has implications for perception and action and blah, blah, blah. But I think the really critical take-home on these things is that, you know, until I learned about that work and got some knowledge about it, I just assumed that, like, you grow a teeny, eensy bit every day. And then during adolescence, you kind of grow, like, a bigger, eensy bit than during, you know, during childhood.
Do you know what I mean? Yeah. Bit.
The people that do this work, it changes the kinds of theories you would have about growth. Growth in long bones. It's totally a different theory now because they discovered that growth was episodic. And they would not have discovered that if they had not understood that development happens in an individual.
And so certain things that we do as a convenience, like using a growth chart to see if your baby is whatever, whatever, compared to whatever, continuing to grow or failing to thrive. That's fine, but don't get it confused with actual growth, right? And so, like, if you need to have, like, onset ages for something, they're onset ages, air quotes. It's not real. It's arbitrary.
It's like a, what's that? Like, okay. And those are useful things.
Like, you know, when you want to get services for children with disabilities, you want to say, like, it's 20 months and not walking and not growing up in Chichikistan. Get me services, you know? And that makes sense to me.
But don't confuse it with how roller skates enter and exit children's repertoire. I'm done with crawling. Over.
I don't think it's a butterfly, but it's a butterfly. When you look at the function, step size, or something, you find trajectory that optimize energy for each given speed. Does that happen the same way in babies? Yeah, I doubt it.
Yeah, it does not. Don't they maximize energy consumption? No, I didn't minimize it.
Oh, they maximize it. They're just blowing it. Yeah, yeah, yeah.
They're not. minimizing energy consumption. And neither do old people.
Neither do people who have backaches. Like the Max Donnelly. So I raised it with him that you know at some points in life or we're not We are maximizing information.
This is an extreme statement. So what are the data you refer to in old people? We have measured stuff. You know, shuffling is not maximizing energy.
It's balance. And those tiny baby steps are balance-related. Sometimes that's what we do.
You don't want to fall. And older people. And probably babies too.
You would assume the small steps. Yeah, I don't know. I didn't know.
I didn't show. I'm trying to think what I have on here. I didn't show you the fall data. Your average toddler falls 17 times an hour. Sure.
Like between 12 and 19. They're falling 34 times an hour when they first start, like in their first, well, at 12 months. So like when they're first starting to walk. They fall once for every five steps in their first day of walking. So not wanting to fall. Yeah, not really.
No. They don't have to be on their hips either. I know, and I love that too.
So we've been going back, and I used to just code falls, like fall check. And now we're looking at how they fall, like all the details of it. And babies fall like a kelp in the ocean. You know, like... And so usually they first fall into some fat body part, but if their trunk or head or something touches, like five things have...
impacted first you know and other people when they fought like you can almost hear like crunch crack break hospitalization pneumonia death you know you can just like oh it's like that