Your Baby Is Smarter Than You Think

http://www.nytimes.com/2009/08/16/opinion/16gopnik.html

by robg • 17 years ago

44 16 17 years ago

JabavuAdams2• 17 years ago

Not a great title, but a good article, nonetheless. If you're interested in machine intelligence, you need to spend some time with infants.

I have a nine month old, and it's been really interesting to watch her develop.

You can see where people get these ideas of multiple agents interacting. For a while, anytime her hand was within a certain distance from her face, it would "snap" to her mouth. She'd clearly be trying to do something else, but then the hand would get close to the mouth, and OMNOMNOM.

Pattern matching and mimicry. Most things we do, we do because that's what we've always done. /hand-waving

jerf2• 17 years ago

Yeah, I've noticed that with my kid, too. When he learned to sit up and crawl, he would be nearly asleep, presumably in that part of sleep where your brain is wandering but you're not quite asleep yet, then suddenly, bam, flip over and get on all fours to crawl, then start standing up on the bed. Once when I let him finish the task, he even visibly woke up, looked at me with some surprise, and then started crying again because he was tired, wanted to sleep, and clearly did not want to be standing up. Probably blamed me for it, too. No intentionality, it's just the "standing program" fired at an inopportune time.

Another funny one (from an AI perspective) dates from when he was about 4 months old; we were sticking our tongues out at him and he was starting to try to copy us, and I swear, you could see his leg jerk, then his left arm, then his neck, until finally he got his tongue out, just like there was a little man in his head pushing buttons until he found the right one.

On the parent's side: We have a dog and two cats, and our child just passed a year old. My wife was just telling me that she's caught herself being surprised a few times over the past few days, as my son does things like "puts the lid back on the container". She said she realized that up to this point she'd been subconsciously applying the "pet" template to our son, since that roughly matched his capabilities for the past few months (non-word vocalizations for needs, limited manipulation skills, understands only a few words, etc.), which she didn't even realize until suddenly he started to do a few things that didn't fit the template, like put a lid back on a container, or spontaneously use a keyring to mime locking the door (to the best of his ability). Kind of an interesting conceptual problem on the adult side.

andreyf2• 17 years ago

She said she realized that up to this point she'd been subconsciously applying the "pet" template to our son, since that roughly matched his capabilities for the past few months

This really surprised me, also. Especially so, because first "human" things my daughter did was to start offering to feed us when we fed her (~7 months). It would seem social reciprocation was her first social instinct - long before speech, feeling of ownership, etc.

Radix• 17 years ago

She said she realized that up to this point she'd been subconsciously applying the "pet" template to our son, since that roughly matched his capabilities for the past few months

Wait, where did guys meet your wives? Where do you meet someone who is aware of their templating of things in casual conversation?

andreyf2• 17 years ago

If you're interested in machine intelligence, you need to spend some time with infants.

Completely agreed - what blew my mind most was how little there is when children are first born. It was only a couple of months in that my daughter realized those things constantly around her were her arms and that she could control them. There was literally a moment when she stopped just flailing around, and stared intently at one hand as she moved it left and right in front of her.

mistermann• 17 years ago

One of the funniest things I noticed too, I loved watching when my daughter stared at her hand and slowly moved it around after finally discovering she controlled it. One of those stupid little things probably only a parent appreciates!

anatoly2• 17 years ago

What a disappointing article. Maybe there's more meat to the studies cited in it, but the way they're presented here makes them seem borderline trivial, and certainly much less interesting than claimed.

1. A baby sees a box with a lot of white balls with a few reds; then the baby is more surprised to see 4 reds with 1 white drawn out of the box than vice versa. Why does it "prove that babies could understand probabilities"? Why not suppose that the reds are more interesting to look at because they were more rare in the original set? Show a baby four identical bears and a giraffe; which toy will she later prefer to play with?

2. "In 2007, Laura Schulz and Elizabeth Baraff Bonawitz at M.I.T. demonstrated that when young children play, they are also exploring cause and effect."

In other related and equally exciting news, MIT researchers discovered that water appears to be wet. Come on! What parent of a young child hasn't noticed that they're exploring cause and effect, every single day, all the time, at least since they learn to manipulate small objects? And the study itself - gosh, couldn't it be simpler for a child to press one lever than two levers simultaneously, and therefore the group that didn't know what happens in that simple scenario spent more time with the toy?

3. "These children, who couldn’t yet add or subtract, were more likely to put the high-probability yellow block, rather than the blue one, on the machine."

But if the children learned the effect of each block from the first time they observed it in action, completely ignoring subsequent times and therefore unable to use any probabilistic reasoning, you would still expect the same effect to happen on average in a group of children. In other words, this setup doesn't even prove that the children can distinguish a mostly working block from a mostly non-working one (doesn't sound as exciting as "high-probability", does it?).

Given the way all these experiments are described, it also seems doubtful that the sample sizes were adequate, that there was a control group, that they were double-blind.

Basically all this seems like a bunch of wishful thinking, and a set of sloppy experiments drawn up to reach a foregone conclusion.

robg• 17 years ago

These seem like all good questions, but an op-ed in the NY Times isn't the place to address them. There just isn't room to specify all of the tested variables and their controls and how the results support the conclusions. Of course the NY Times should be linking to the original research reports.

Still, I don't know how this op-ed could lead someone to think the research was sloppy. For me to say that I'd have to, after I read the original reports, present alternative designs that eliminated confounding variables but that didn't introduce additional ones.

Here are the papers:

1. http://www.psych.ubc.ca/~fei/XuGarcia-PNAS.pdf

2. http://web.mit.edu/eccl/papers/bonawitzandschulzseriousfun.p...

3. http://www.alisongopnik.com/Papers/Kushnir%20DevPsych.pdf

I'm sure the researchers would be pleased to answer any questions.

anatoly2• 17 years ago

Thank you for tracking down the papers.

ahoyhere• 17 years ago

Maybe this aren't good examples but great ones are out there.

For example, even the littlest babies understand basic physics. Once they're old enough to focus, if you show them a video of a ball hitting a stack of blocks adn the blocks go flying, they don't care. But if you show even a few milliseconds' delay between the ball strike and the blocks flying, they get very, very interested indeed.

By the way, the studies no doubt have controls and reasons for the way that they determine the probability that babies are paying attention to red balls because of x than because of y. These types of studies have a long history in the field. But the op-ed author is not going to go into those details.

pbhj• 17 years ago

"The babies were more surprised, and looked longer and more intently at the experimenter when four red balls and one white ball out of the box — a possible, yet improbable outcome — than when four white balls and a red one were produced."

I'd warrant most people know that babies respond first to the contrast of black and white and next to bright red - if the experiment worked equally with other colours then just about any other colour combination would be more notable.

I do wonder if this is unique to light-pink skinned babies/parents as I suspect the reaction to red helps in locating the mothers nipple, though I'm not set on that hypothesis.

"When we say that preschoolers can’t pay attention, we really mean that they can’t not pay attention: they have trouble focusing on just one event and shutting out all the rest."

I disagree there too, preschoolers can focus intently on something they are interested in despite all effort to divert that attention. I suspect they don't have the ability to context switch nor really multi-task and so instead of momentarily diverting attention (multi-tasking) pending a context switch they attempt to complete the current process before interrupting.

"Parents and other caregivers teach young children by paying attention and interacting with them naturally and, most of all, by just allowing them to play."

What does "just allowing them to play" even mean. No toys? That's providing things, not "just". No scenery? That's providing a context. No interaction? I don't think kids really learn by just playing, they have to have experiences to play out, inspiration of their imagination, people to interact with and yes even things. The old favourite of playing with cardboard boxes is providing things (boxes) and relies on the child having been stimulated to have the experience of things to imagine the box as.

wglb• 17 years ago

Good, but old news.

This is a useful educational article, but I had a pediatrician in 1981 explain much the same facts to us about our just-born daughter.

The lesson is not to go off and overstimulate the child, but 1) observe closely what the child is paying attention to and 2) don't presume that the kid doesn't know significantly more about what is going on that you might think and 3) don't talk baby talk to the kid. Emulate as closely as you can what they say. This will give very positive feedback of the right kind.

i did an experiment with a friends just-born baby. I would get the kid's attention, then stick my tongue out. After a couple of repeats of this, the kid responded by sticking her tongue out, much to the astonishment of her mother. "There is a person in there".

So if we can follow this line of reasoning about adults, at least in some cultures, being off in understanding what kids understand just after birth, how far off has that perception slipped by the time they are six? teenagers?

Your Baby Is Smarter Than You Think

Related Stories