Question:

There's something intuitively familiar about an affinity between children and artificial intelligence, if only as a trope from old TV shows like 'Lost in Space,' where a boy befriended an endearing robot, a robot that could almost be viewed as his alter ego.

Answer:

There has been a big shift in how people in artificial intelligence look at learning, and that new picture of artificial intelligence happens to fit with much of what we are learning about children.  At the new Center for Developmental Cognitive Science, we’re going to put these two bodies of work together and try to find out what looking at young children, who are after all the best learners in the universe, can tell us about designing the next generation of artificial intelligence.

Question:

What are some of the advances that you're building on?

Answer:

Bayesian models give us a tool to study the way young children use statistics and probability to test hypotheses. It turns out that even the youngest babies and children think more rationally than we ever realized. 

We’re talking about these babies being little scientists. And one of the things they’re most interested in is other people. So we’re also thinking of them as little psychologists. 

Question:

You've written that for babies 'every wobbly step is skydiving, every game of hide-and-seek is Einstein in 1905.' If children have such an intense experience of life, why do we have so few memories of early childhood?

Answer:

We don’t know why we have infant amnesia, what we call episodic autobiographical memory. Knowing we are part of a single story may have something to do with brain development. 

Question:

We're struck by the fact that young children have more neural connections than adults. With experience, some connections are strengthened, while others disappear. The pre-frontal cortex, the part of the brain responsible for inhibition, develops last; and its adult configuration has a lot to do with childhood experience. What are the implications for Western philosophy's notion of 'Self?' Are we nothing but glorified neural editors?

Answer:

One of the things that has motivated our work is that looking at babies can help with this debate.  Descartes thought that the self was innate and incontrovertible. David Hume and the Buddhists think that’s actually a bit of an illusion. I’d say developmental psychology is on the side of the Buddhists. 

Question:

You've written that: 'Children are both profound and puzzling, and this combination is the classic territory of philosophy. Yet you could read 2,500 years of philosophy and find nothing about children.' Is it belaboring the obvious to ask if this omission has something to do with the fact that nearly all philosophers were men?

Answer:

Yeah, of course. In the case of philosophy, not only were they men, but they tended to be celibates and clergymen. There is not a very good record of great philosophers as human parents. What philosophers think is important tends to be determined by what they’ve experienced. 

When I talk to people about the new studies in babies’ cognitive development, people who know babies well — caregivers, moms or teachers — nod and say, “Of course.”  But you don’t get a good feeling for babies unless you pay a lot of attention to them, figure out what their body language is saying and put the pieces together. We’ve had this brilliant miracle going on under our noses, but it was a miracle that was easy to miss if you just looked at babies superficially.

Question:

Have people dismissed your work as 'women's work?'

Answer:

I have this experience all the time. When I go to an important scientific meeting, people say, “Oh, you should talk to my wife, she’s a pre-school teacher.” The fact is, I probably would like to talk their wives — but the new work in developmental cognitive science is as scientifically rigorous and philosophically significant as any in science. 

Question:

In terms of getting past those preconceptions about gender, Berkeley was one of the first places to teach child development in an academic setting, wasn't it?

Answer:

Seventy-five years ago, the Institute for Human Development was one of the first places to study children’s learning. In those days, when people thought about children, they considered them learning animals in a Skinner box, reacting to stimuli and responses. We’re realizing that the kind of learning babies do goes beyond this, and we’ve seen exactly how that process can work mathematically. Babies are still much, much better learners than the best computers. 

Question:

Are many researchers looking at the relationship between computers and early childhood learning?

Answer:

I think it’s quite clear that Berkeley is in the lead in this whole area, relating children’s learning to computation and big philosophical issues. We have a whole cadre of people in different disciplines that make us one of the top centers for this kind of multidisciplinary study. But our new center will not only involve people from Berkeley. There are Stanford researchers, and the project will be interactive with MIT and other universities. 

Question:

Artificial intelligence has brought up challenging questions in regard to human evolution. Ray Kurzweil and Bill Joy, among others, suggest that humans may fuse with robots, become robots or become extinct by robots. You come from another angle, talking about how evolution plays a role in learning. Can you discuss that?

Answer:

When we’re thinking about the evolution of human cognition, one of the striking things is that we have this sophisticated learning mechanism that takes time to develop and requires a bigger parental investment. Grandmoms seem to be part of that project, since human women stick around for much longer than other animals. So, when we’re no longer fertile, we have a role to play. Now, I think grandmoms are the center of the universe. As I’m talking to you I have my beautiful grandson just waking up. I have a scientist right here.