My Child the Scientist

I Want my Jetpack!

Dave Sobel — Sat, 14 Nov 2009 16:21:09 +0000

My lab has a new website. The URL is

http://www.cog.brown.edu/research/causalitylab/index.html

Most of the design was done by my grad students and undergrads, but as part of the redesign, I wanted all of my grad students to write a short paragraph about their research. One of my students wrote the following:

"I play games with preschoolers, and then try to make computer programs that are as smart as preschoolers. The children usually win."

I thought this was interesting. It reminded me of a conversation I had when I was a graduate student 12 years earlier. I was in my fifth year of grad school, almost finished, and I was talking with a new student about what she wanted to work on for her first year project. Her response was that she wanted to build computer programs of children's behavior (she actually wound up doing something quite different).

It also reminded me of my own undergraduate thesis, written in 1991, in which I described a (pretty flawed) computer program that was designed to predict human behavior on a learning task. It was based on a model of animal cognition that was built ten years previously, which in turn, were based on a set of models designed in the early '70's.

The goal of building computer programs that model human behavior extends even earlier than that. In my University's Psychology Department, one of our distinguished professors has been working on a model of timing (in non-human animals) since the ‘50's. I heard him describe his research once (last year): He's still working on the model.

So, here we are: 50+ years of modeling. One of the promises of the "Cognitive Revolution" was the "Computational Model of the Mind" - the idea that the brain might act like a computer, and thinking about this metaphor might offer us insight into how the brain produces behavior and thought. In graduate school, I learned this as "Mind is to Brain as Software is the Hardware", a mantra that undergraduate students in the Introductory Cognitive Science class I TAed were forced to recite.

In cognitive development, various kinds of models have bubbled up to explain children's behavior - symbolic AI, connectionism, dynamic systems, causal models, Bayesian inference (which is now all the rage) - and each has advantages and contributions. But, I wonder whether the promise of building these models for developmental processes is akin to the promise of jetpacks when I was a kid: a nice idea, but more wish-fulfillment than actual science. A "wouldn't it be cool if we had these" idea, as opposed to a realistic endeavor?

Yes, this is a straw man (and yes, I'll knock it down in a few paragraphs). But, I've been thinking recently it's important to think about why this argument is a straw man.

My favorite computational model comes from the animal cognition literature. It's the Rescorla-Wagner model, first published in 1972. Basically, it was designed to explain a phenomenon in conditioning called blocking, which it did quite nicely. The model itself is flawed, and there have been numerous (and I do mean numerous) variations and recreations of the ideas they present in that original paper.

Here's what I like about the model. First, it does make some really counterintuitive predictions, which turn out to be true. Rescorla and his students went on to demonstrate several phenomena that emerge just from sitting around calculating possible inputs to the model. But more importantly, it also suggested a set of paradigms that emerged from trying to break the model. Almost immediately after its publication, Wagner and his students authored a few papers that demonstrated the model did not account for all of animal conditioning, and many of those phenomena came again from thinking about the predictions made by the model.

It's not the model that I like (although it's quite elegant) - it's that almost 40 years after it was published, researchers (including myself) are still thinking about what it does and does not predict, and using those predictions to design new experiments that explain human (adult and child), as well as animal behavior.

This is certainly part of the promise of the Cognitive Revolutions - models should act as formal systems - allowing developmental psychologists (and psychologists in general) with yet another way of thinking about the relation between theory and data. A good model doesn't just explain behavior, it makes predictions about what other behaviors we should expect to observe.

I suspect that the previous sentence I wrote is widely agreed-upon within cognitive science, and certainly among computational modelers. But, what strikes me about the field right now is that so few people actually bother to write it down (or even say it out loud). Even worse, I'm often struck how few papers actually practice this. A challenge faced by cognitive scientists in general, and researchers interested in modeling children's behavior is to stick to this principle. Otherwise, aren't we just pretending to build jetpacks?

A note about blankets

Dave Sobel — Tue, 20 Oct 2009 16:21:03 +0000

Just a short update from my last entry: I wrote that entry on Saturday morning. That afternoon, I dropped Lisa and Paulina off at a local Specialty-Baby-Store-That-Must-Not-Be-Named for a class. When I came to pick them up, I found that the class was running a little long, so I got to look around the store. In among the overpriced books, toys and flashcards, which all claimed to promote development, I discovered a blanket manufactured by Company-That-Must-Not-Be-Named. The packaging on the blanket came so close to stating that that wrapping your child in this product would promote healthy development and learning, that I felt compelled to mention it here. So, the update is that now, I'm not making the part up about the blanket in my last entry. It's real. And for $30, you can own the blanket too (Seriously).

Confessions of an Average Driver

Dave Sobel — Sat, 17 Oct 2009 14:39:34 +0000

I've been driving for a while now, and I feel like I have a good sense of my abilities. I'm an average driver. I'm not a particularly bad driver. I don't cause accidents. I don't try to cut other drivers off. That said, I also don't think I'm that good of a driver. I have twice backed into stationary poles (in my defense, the poles did come out of nowhere). I'm cruising along somewhere in the middle lane of the driving-ability highway.

My point isn't that it's OK to be content with average. My wife is a much safer driver than I am, and I would like to be as safe as her (or at least, stop backing into poles). With Paulina here, I've make a concerted effort to drive safer, and to think about my driving a lot more. It's great to strive towards being better than we are.

My point is that average has become a dirty word, particularly when it comes to children. No one wants their child to be average. And so, we spend a lot of money on stuff - stuff that we think will prevent our child from being average. I've now walked through my local Giant-Baby-Store-That-Shall-Not-Be-Named thoroughly, and what I've seen is a lot of "Child Development" toys, videos, books, and blankets.

OK, so I made up the part about the blankets. But, you can see it, can't you. "Buy this "Snoggie" Blanket because it promotes learning." Let's ignore the fact that there's no possible mechanism by which a particular blanket allows children to learn better. You can see a company making this claim, can't you? Just as an aside, whenever you see the claim that a particular product "promotes child development," ask yourself how it does this.

Anyway, here we are, new parents, and we buy stuff. And why do we buy stuff? We see our friends with children buying similar kinds of stuff, and talking to each other about their stuff, and we see advertising for the stuff, and something deep inside of us feels the need to have the stuff. If all of our friends have the stuff, and we don't, then we feel our children will fall behind. This is a fallacy, of course, but it's hard to make anyone believe that. For instance, at a recent family outing, I overheard Lisa talking with another parent, who was singing the praises of a particular kind of jumper because it had all of these bells and whistles, which "kept her child busy," but also "helped her child learn."

Learn what exactly?

Here's the thing. It's what developmental psychologists have been telling parents for a really long time now. It's not about stuff. It's about time. It's about interaction. Parents can buy all the stuff they want, but if they don't spend time with their babies, showing them how to interact with the stuff - playing, pretending, and just basically hanging out, focused on their kids - the stuff doesn't really do that much. And really, the stuff is just gravy - you are more interesting to your baby than any object, even if it lights up, plays music, or is named after a famous scientist.

Heard this before? Good. It's a lesson worth repeating. But there is a flip side to this rant, and I don't think developmental psychologists talk about it much.

When Paulina was born, Lisa and I bought a Book-That-Shall-Not-Be-Named about communicating with infants. I liked it because it was written by developmental psychologists I respect, and it taught a system that requires parents to spend time with their children. But, here's the thing: I read (adult books) only when Paulina sleeps (I also only write this blog when she sleeps). Thus, my time to read is at a premium. I was stunned to see that the first chapter of the book was nothing more than glorified praise for the system. It contained no useful information. I spent the time reading this for no reason whatsoever.

OK, I get it Authors/Scientists-Who-Shall-Not-Be-Named, you're proud of your system, which has done a lot of good for parents and babies, and it's nice to illustrate that. But we're new parents, which is an exercise in sleep deprivation, patience, tolerance for bodily functions, and laundry. We need to know what we need to know and not extra stuff that just sells more books or the accompanying DVD that can be purchased for another $12.99. During the time it took me to read that first chapter, I could have been sleeping, or buying diapers online, which is my new hobby, or just goofing off, which would have been more productive. I want that time back. Authors Who Shall Not Be Named, get to the point!

Science is hard. Communicating scientific findings to the public is even harder (it's why almost all scientists I've ever met find Science on national news to be a joke). But Developmental psychologists specifically have the obligation to communicate to parents in a way that is not only clear, but also concise and relevant. This is unfortunately, not easy. It's an art form, and I've met only a handful of scientists who are good at it. I don't think I'm one of them. I think I'm about average. If only there were flashcards to help me.

This Entry is not about The Sexual Paradox

Dave Sobel — Tue, 29 Sep 2009 14:32:53 +0000

I'm not going into my office that much this semester. When I do, I always check my mailbox, just to make sure things don't pile up. Last week, I found myself there on Friday. In my box was an envelope with a Simon and Schuster mailing label (actually from Scribner publishing, which is now owned by S&S, I guess). I opened it to reveal a book - Susan Pinker's The Sexual Paradox - and a single page, which contained (on both sides) a number of positive reviews of the book. There was no letter, no contact person, and no explanation for why I received this book.

A brief investigation of the cover (and the double-sided page of praise) suggested the book is about gender differences, and is focused in particular on gender inequality in the workplace. I don't study gender differences, either in development or adult cognition. In my 14+ years studying cognitive development and adult cognition, I've found exactly two significant gender differences in studies. In both cases, my colleagues and I couldn't replicate them, and we dismissed them as statistical anomalies.

I also noticed that this wasn't a department-wide or university-wide mailing; I asked some of my colleagues, and none of them received the same package. I was the only one who received this package.

A more detailed investigation of the envelope found a single word written above the return address label: "Maxx" A search of the Simon and Schuster website for this word didn't reveal much (just an author - Maxx Ardman - who contributed to a volume called Take my Advice). Seemed like a dead end.

Finally, I'd like to think that this wasn't a message sent to me anonymously saying that I need to think more about mentoring or supervising one gender over another. This seems a little elaborate, and I suppose if this were the case, an instructional letter or word of advice might have accompanied it, not the page of praise. Also, I've always thought I have a pretty good track record of mentoring women - 17 of the 20 students I've sent to graduate school from Brown are women. I've also been my department's Affirmative Action representative for the past two years.

Confused yet? Me too. I honestly have no idea why I received this book. I've never received a book outside my area before, and certainly never without a letter of explanation. This is not to say that I've never gotten a book in the mail randomly - but usually it's a new textbook in Cognition or Development, with a letter from the publisher asking me to consider adopting it for one of my classes.

When I told my wife this story, she actually came up with the same hypothesis that I did. Maybe someone at the publishing company knew that I was writing this blog, and figured they would send me a copy of the book under the impression that I would write about the book - free publicity for them, and a blog entry (and free book) for me.

OK. This can't be right. Just absolutely can't be right. Not in a million years right. Yet, why did both my wife and I think of it? I mentioned it to my colleagues as well, and they all thought it was a possible (but not probable) explanation. Now, I'm sure Pinker's book is a nice piece of scholarship, but I'm a developmental psychologist who doesn't study gender differences, and I haven't read her book. I just received it under mysterious circumstances.

Why is my (and my wife's and colleagues') explanation plausible? I think Bruce Hood, a professor at the University of Bristol, has the answer. In his book Supersense, he examines why it is that human beings hold various forms of supernatural beliefs. His argument is that we seek out explanations of human experience in the course of our interacting with the world. We notice patterns among events where none might exist, and we infer mechanisms between actions (particularly our own actions) and outcomes that might be nonexistent. This is the "supersense" on which the book is based.

I had a lot of fun reading this book. It's well-written and filled with examples that should resonate with both the scientist and layperson. For example, Hood writes that most rituals emerges from this supersense - repetitive behaviors that have no mechanistic relation to an outcome often emerge from successful action (his example is John McEnroe, who never stepped on the lines when serving a tennis ball - clearly a learned behavior from successful actions). But more than our own ritualistic inventions, we collectively believe in supernatural mechanisms just from association. For example, Hood argues that most people wouldn't wear a murderer's sweater (he apparently brings a cardigan with him when he speaks publically, and claims it was owned by a convicted killer. He asks members of the audience if they would come up and put it on, and most refuse). Why wouldn't you wear a killer's cardigan? Hood argues that we believe that the evil associated with the act of murder will somehow be transmitted to us, simply by putting on the sweater.

Hood also argues that many of these beliefs begin in childhood. I certainly agree with this idea. Tamar Kushnir and Alison Gopnik published an article in Psychological Science in 2005 that supports this idea. They found that preschoolers are biased by the results of their own actions. They showed children a machine that lit up and played music when objects were placed on it. In one condition, an experimenter showed 4-year-olds two different wooded blocks (which I'll call A and B). The experimenter placed Block A on the machine three times, and the machine activated twice. She then placed Block B on the machine three times, and the machine activated once. Children were then asked which block was more likely to make the machine go, and the majority of them picked Block A (more so than chance).

In their other condition, the same blocks and machine were used, but now the child had an opportunity during the demonstration to act on the machine. The experimenter first put Block A on the machine twice, which activated both times. Then she handed Block A to the child, who put it on the machine, which did not activate. The experimenter then put Block B on the machine twice, which failed to activate, but when the child put it on the machine, it did so. Note that the frequency with which the blocks activate the machine is same as in the previous condition - Block A makes the machine go 2 out of 3 times while Block B does so 1 out of 3 times. What differs is how the child's own action interacts with the efficacy. For the block that activates the machine more overall, the one time it doesn't is when the child puts it on the machine (and opposite for Block B). Children were heavily influenced by their own action. In this condition, they chose Block B as the one that was more likely to make the machine go, more so than in the previous condition.

What these data suggest is that preschoolers think that regardless of the overall probabilistic outcomes, their own actions affect their belief structure. This resonates well with Hood's argument about how rituals emerge. If we think our own actions are more important than observations of others' actions, then we are bound to repeat those actions when we observe success.

I also really enjoyed reading about Hood's work with Paul Bloom (which appears in a 2008 issue of Cognition). They introduced children to a "Copy Machine" - a machine that makes exact duplicates of objects (through a mechanism unbeknownst to the child, but that is actually demonstrated through a slight of hand - in reality, it is just a magic trick, and children are shown this at the end of the experiment). Hood and Bloom found that children were willing to accept duplicates of many kinds of familiar objects, except one specific kind - attachment objects like their security blanket or a special stuffed animal (actually, they almost never would allow these objects to be duplicated in the first place). Children recognize that an object's experience is as critical to its identity as its physical appearance. A security blanket makes a child feel safe because he's had it for a long time, and thus it's worth more than a similar looking, but novel blanket. In this way, some of us would pay huge amounts of money for collectibles, not because we want to use JFK's golf clubs, but because we want to admire the fact that we're associated with someone we admire.

One of the things I find so interesting about this work is that my former student Claire Cook and I have a manuscript under review right now about children's intuitions about the possibility of machines in the real world. We find that preschoolers deny the possibility of machines that violate real-world causal structure. What's interesting to me is that before seeing a "Copy Machine," I would bet that children would flatly deny its existence (we don't use this example, but many similar to it). But, they can reason about it (in quite subtle ways) after just seeing it live. This strikes me as evidence that children can learn about technology incredibly easily, which is potentially why each generation seems more technologically sophisticated than the next.

Finally, Hood also argues that one's supersense could be broken - he argues that Capgras Syndrome, in which individuals believe that people in their lives have been replaced with duplicates, might be evidence for impaired supersense. I'm not sure what I think about this argument, but it resonated with me, as the next piece of fiction to read is Richard Power's The Echo Maker, which is also about Capgras Syndrome. I haven't really thought about this deeply, but found the connection interesting.

That said, I will mention that some of the examples in Supersense are not for the faint of heart (the chapter on receiving a heart transplant from a murderer is definitely rated R). But even if some of the book's examples are a little ghoulish, they reflect profound aspects of human nature. The scholarship is impressive, and Hood nicely describes how supernatural thinking fits into our ordinary human experience.

So, what made me think of Hood's book upon receiving The Sexual Paradox. Well, given the mysterious way in which I received this book, and given that my wife, my colleagues, and I could only come up with the explanation that we did, I'm believe that my (and my wife's and colleagues') supersenses are working just fine. We all come up with explanations to explain events, some that involve real-world causality, others that don't. Here's an explanation that doesn't: Last April, at a conference, I had a brief conversation with Hood, and mentioned that I was starting this blog. Isn't it possible that Hood sent me a copy of The Sexual Paradox in this mysterious manner so that it would encourage me to review his book? Of course it isn't, and obviously this is my supersense working overtime.

The Biggest Lie I was Told about Becoming a Parent

Dave Sobel — Thu, 17 Sep 2009 20:44:51 +0000

The biggest lie I was told about becoming a parent is that baby's poop doesn't really smell for the first few months. Seriously. I can't imagine anything further from the truth. Paulina's poop smells, well, like poop. She obviously eats a lot (she was 99th percentile for weight at her 2 month checkup), but I didn't think that would cause anything out of the ordinary (in all honesty, though, I have no idea - I'm happy to hear thoughts on this - well, not exactly happy, but I always welcome dialogue).

I was thinking about this when I saw a trailer for a movie coming out in October called "The Invention of Lying." Here's a quote, describing it from Yahoo Movie's Coming Attractions:

"In an alternate reality, lying -- even the concept of a lie -- does not even exist. Everyone -- from politicians to advertisers to the man and woman on the street -- speaks the truth and nothing but the truth with no thought of the consequences. But when a down-on-his-luck loser named Mark suddenly develops the ability to lie, he finds that dishonesty has its rewards."

Link to Yahoo Movies on The Invention of Lying

What this reminded me of what it must be like to be a child, and develop a concept of lying.

How might this work? Answering this question would take a while - there's a lot of research on how children deceive and understand deception. One of my favorite papers on this topic is work by Beate Sodian and her colleagues from 1991. They found that 4-year-olds recognize when to deceive another person (in a competitive situation) and when not to deceive (in a cooperative one). Three-year-olds, in contrast, deceived indiscriminately.

Link to Sodian et al. (1991) in Child Development

This finding makes sense. In order to deceive another person, you have to represent what they are thinking, and how your actions will give them a false belief. Children's understanding of false belief typically develops between the ages of 3 and 4.

But there is some recent research that challenges this assumption. In 2005, Kristine Onishi and Renée Baillargeon published an article in Science suggesting that much younger children understood when another person had a false belief. In their experiment, 15-month-olds watched an actor place a desirable object in one location. Then, unbeknownst to the actor, the object moved to another location (both of the locations were containers, so you couldn't tell where the object was just by looking at it). Some infants then observed the actor return and look in the original location; others saw the actor look where the object really was. Infants stared at the display longer when the actor looked where the object actually was, as opposed to where the actor thought the object was; that is, they showed surprise when the actor didn't act according to his false belief.
Importantly, they didn't do this when the actor had a true belief about the world - when she witnessed the object move

Link to Onishi and Bailargeon (2005) in Science

There's an aside here - I don't really think the longer looking time has anything to do with surprise, but that's an easy way to think about this paradigm.

Onishi and Baillargeon's paper has now inspired a whole field of researchers to examine infants' understanding of others' mental states (and notably infants' understanding of belief). At the 2009 meeting of the Society for Research in Child Development, I saw numerous talks and posters dedicated to examining infants' early comprehension of mental states using Onishi and Baillargeon's techniques (as another aside, I'll mention that two of my students got really interested in this, and are starting data collection on their own experiment soon - if things go well, I'll mention their findings sometime in the future). There is genuine interest in this topic in the field right now.

But there's also a big question: Why do 15-month-olds behave as if they understand others' mental states in these experiments, but 3-year-olds do not when asked to deceive (or for that matter, when asked to recognize that someone has a false belief using any task that involves simply asking them). It could be that the linguistic aspect of these tasks adds cognitive demands for the older children, which causes them to fail. On this view, children have the cognitive wherewithal to understand others' mental states, but fail because of the way experimenters ask the question. A similar possibility is that the tasks used on older children all require the child to inhibit some kind of prepotent response, and such inhibitory control develops around the same time (and indeed, researchers like Stephanie Carlson and her colleagues have found that success on measures of inhibitory control that have nothing to do with mental states predict children's understanding of others' false beliefs). Again, on this view, there's nothing about mental states that develops between ages 3-4; these findings are about children developing the capacity to demonstrate success - they have the underlying cognitive abilities.

I'm a little skeptical of these explanations. In 1991, Alison Gopnik and Virginia Slaughter found that when the inhibitory demands were equated, children appeared to understand another's desires earlier than their beliefs. Some of my own work has looked at the role of inhibition in pretending, and while there is a benefit to reducing inhibitory demands or other demand characteristics within an experiment, it's not the whole story.

Link to Gopnik and Slaughter (1991) in Child Development

So what is the story about children's developing understanding of belief? Ian Apperly and Stephen Butterfill recently argued in a paper in Psychological Review that there are two systems for understanding belief - one that is innate (or acquired very early) and that is potentially shared with non-human animals - and one that develops from this system. Put simply, Onishi and Baillargeon's findings indicate the presence first system, while the development observed by Sodian and others indicates the latter.

Link to Apperly and Butterfill (in press) in Psychological Review

This is a good explanation. But, I want to speculate on another. Consider the phenomenon of infant amnesia. It's somewhat reassuring to know that when Paulina has bouts of gas (that occasionally are a precursor to the aforementioned poop), she's not going to remember how traumatized she was as a preschooler or an adult. Why not? Why are our first memories as children usually from the preschool (or rarely, the toddler) years? There are a lot of theories, but the one that I like the best is that our memories go through a reorganization that is related to learning language. Language gives us a narrative structure, which helps us to remember what happened to us. Why can't the same be true of our semantic memory - our understanding of rules and concepts, such as when another has a false belief? It might be that Onishi and Baillargeon's data shows early competence, but sometime afterwards, we lose that understanding, only to reacquire it as part of learning various aspects of language and linguistic structure.

I've got no evidence for this - it's not exactly clear how this would function or what aspects of language truly affect this cognitive system. But, this does explain a nagging finding in the literature. Right at the third birthday (before children show any understanding of others' false beliefs on standard tasks, or on tasks of deception like Sodian's), they generate contrastives about belief in their linguistic utterances. Three-year-olds say things like "I thought it was a car, but it was really a truck." Karen Bartsch and Henry Wellman's 1995 book, Children Talk about the Mind dedicates a chapter to this phenomenon. If children have the cognitive capacity to understand others' false belief as infants, but must relearn it when they learn language wouldn't it make sense that this understanding should first appear in their own linguistic utterances?

Renée Baillargeon is coming to my department to give a talk at the beginning of November. I'm excited to hear what she has to say. Stay tuned.

The Philosophical Baby Daddy

Dave Sobel — Sun, 06 Sep 2009 16:37:19 +0000

I'm writing this on the Sunday morning of Labor Day weekend, 2009. Paulina is just shy of three months old. But this entry starts with a story about me:

I went to graduate school at the University of California. My first act as a graduate student was to be oriented. This meant learning my way around campus, picking up an ID card and email address, and meeting with my advisor to discuss classes and research. The meeting took place on the Friday before Labor Day weekend, 1995.

Deciding classes was easy (statistics, the required first-year "bonding" class, a seminar in cognitive science, and a research credit - all required). The majority of the meeting was about research. I came to Berkeley from a small liberal arts college. I had performed research in computer science, and had run some experiments on adults, but my graduate advisor was a developmental psychologist, and the expectation was that I would now run experiments on children and study cognitive development. Although I spent a year in college volunteering in a kindergarten classroom, I had no real experience thinking about how children developed, or what contribution I would be expected to make. I also had never studied development, having spent my college career studying artificial intelligence and adult (and some non-human animal) cognition.

In response to my concerns, my advisor handed me a thick stack of pages, held together by two rubber bands. "This is a draft of my book," she told me. "I'm still working on it with my co-author, but it's a good starting point. Read this, and we'll meet next week."

The book was called Words, thoughts, and theories. My advisor was Alison Gopnik.

I read the manuscript over the long weekend (we had scheduled a meeting for first thing Tuesday morning, and I was tasked in that meeting for coming up with ideas for a project). I was hooked. That was when I knew I wanted to figure out how children understood the world and develop that knowledge. I'm still asking research questions based on that first reading. I'm sure that during that Tuesday meeting I rattled off several incoherent ideas, but they led the way to what's now been a 14+ year research career. (I even feel like I've answered some of those questions, so that's a start).

I was reminded of this story by two events this past week. The first was the orientation meeting I had last Friday with a new graduate student who has come to work in my lab. We talked about classes, which was easy, and about research, which was more in depth. We're meeting again on Tuesday, and I'll see what ideas she comes up with. But the second was a package I received from Alison the day before, which I had only opened that morning. It contained a copy of her new book, The Philosophical Baby, which I read over the weekend (with the help of some long naps taken by Paulina).

The book represents scientific writing at its best - warm and accessible while at the same time deep and thought-provoking. I'm obviously a fan (if you haven't figured out by now, I borrowed liberally from Alison's last book with Andy Meltzoff and Pat Kuhl, The Scientist in the Crib to title this blog). As someone who studied with Alison for six years, my insider contribution is that, yes, she really does love Christmas (you have to read to the end to get the reference - I'll just say that for the lab Christmas party, she would make a Bûche de Noël, which would sit on the table for most of the night admired by her graduate students as a work of art. Eventually, on her insistence, one of us would cut into it - usually the most senior student).

I won't do a thorough review of the book here. I can't be unbiased. I'm sure, though, that I'll mention some of the ideas in future entries. I will only say now that I highly recommend it as an introduction to why the study of development is important for every student of psychology, but more importantly, anyone with the aspiration to be a parent or the awareness that they were once a child.

That said, the short section on awe at the end of the book resonated with me much in the same way that I was struck by the ideas I was introduced to 14 years ago. Gopnik suggests that most scientists share the belief that there is a profound significance of their work - discovery for its own sake expands our human potential because it reminds us that universal truths are large and inspiring. Her point is that this is true for children as well - watch a 4-year-old's reaction to a magic show and you see looks of awe. I've see in it Paulina's eyes as she looks intently at her hand, only to bring it to her mouth and suck. I've found something to put in my mouth that doesn't rely on one of these big creatures who are like me to stick something in there! (Incidentally, she's done this with her toes too, but at just shy of 3 months, she doesn't yet have the coordination to bring her toes to her mouth. It'll come. There are other rewards besides discovery).

More generally, what I'm left with is that the task my daughter faces is awesome - to learn about the world. As she wakes up from her nap now (which indicates the end of this entry), I'm reminded that the road of any learner - whether mid-career professor, new graduate student, or new baby - is full of wonder. But we're also guided along the way. My hope is to guide my daughter the best I can.

Expertise and Scientific Thinking

Dave Sobel — Sun, 30 Aug 2009 17:17:07 +0000

Sadly, the summer is coming to an end. On Monday, my wife goes back to work (she's also a professor and has to teach full time in the Fall). I'm not teaching in the Fall, but I knew that I would have some work responsibilities and time commitments, so we decided to hire a nanny to help out with Paulina a few hours a week.

I have to admit that the process of hiring a nanny was a little strange. We used some websites to post a job ad, as well as contacted potential nannies (on some of the same sites). Because most of this was done over the internet, I got a chance to observe how people communicate (particularly when they want a job). This certainly played a role in our decision making. It was important to both my wife and I that our nanny had good communication skills. For example, we chose not to interview the woman who mentioned that she had a lot of experience with children with Asperger's Syndrome, but spelled it "Assburgers." Beyond lapses in spelling, however, we were looking for someone who would understand our goals and views about parenting.

We hired someone last week, and so far so good. But the experience got me thinking. In my last entry, I talked about children's ability to trust others as sources of information when learning the meaning of words. I also mentioned that children can easily recognize when someone is a reliable or unreliable source of knowledge, and integrate that information into their reasoning. Further, children recognize their parents as reliable, based on the nature of the relationship they have with them. But what about other people? And what about learning information other than the meaning of words?

In a 2006 article in Child Development, Paul Harris and Melissa Koenig point out that trust in others is important for learning about domains beyond words - they focus on a few different kinds of socially constructed information. My favorite example from their paper is science and scientific entities. We never directly observe germs - we're just told (and we tell our kids!) to wash our hands and trust that doing so will eliminate germs. In general, scientific entities are pretty interesting from the point of view of developmental and educational psychology. Most (all?) of us never actually see a germ, or an atom, or a quark (we may see representations of them, but that's different). We rely on other people to tell us that they're there, and what role they play in causal relations (e.g., germs make us sick, washing hands eliminate the germs).

In a response to my last post, Sarah D. asked whether it matters that parents aren't always reliable sources of information. My favorite example of hers was "the rides are sleeping today" (presumably applicable in the circumstance where a child wants to go on a ride, which is closed, or more generally, when the child wants to go to the amusement park, but the parent doesn't). The short answer is that it does. Here's a case where the adult isn't technically "wrong" (like when the confederates in Koenig & Harris' experiments who label a shoe a "truck"). Rather, the parents are inventing explanations and rationales that suit their purposes (which they believe the child can't verify). I'd like to suggest that young children are often sensitive to this information as well.

To illustrate, consider an experiment that Jessica Sommerville and I ran (presented in the January, 2009, issue of Cognitive Development). We showed 4-year-olds a puzzle box. On it, there were different-colored lights, each of which was activated by a unique buttons, all in clear view of the child. In the puzzles, some lights made other lights go (so, for example, when you pressed the red button, the red and blue lights would illuminate, but when the blue button was pressed, only blue activated - thus, red made blue go). After we trained children about the nature of the box, we introduced them to a set of puzzles. In each puzzle, we first presented them with some ambiguous data. We pointed out the ambiguity, and then showed them the information that would resolve the ambiguity.

What we manipulated was what we said to the children when we showed them the critical information that disambiguated the puzzle. For one group of kids, we told them a fairly nondescript rationale for this action that was related to learning the puzzle. For another, we told them a rationale that was related to the experimenter's personal aesthetic (that he liked the color of one of the lights), which was unrelated to puzzle learning. Finally, in a third condition, we said nothing to the children. What we found was that children who heard the appropriate rationale could better reconstruct the way the puzzles worked than children in the other two groups (who showed equivalent levels of learning).

A good question is why this information helps children learn. One possibility is that we simply engaged the children more in this condition. We didn't think this was too likely, because that hypothesis would also suggest that an inappropriate rationale would hurt children's learning (it didn't). We discuss this more in the paper itself (I've linked to it below).

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6W47-4T...

We were more convinced by the idea that appropriate rationales allow children to plan prospective events, and then evaluate whether what they observe was commiserate with what they thought would happen. This suggests that the same person could be reliable about one puzzle and unreliable elsewhere. We haven't investigated this, but it makes sense in light of the comment about parents not always being reliable. Parents don't have all the answers, and certainly when it comes to conflicting desires, unverifiable explanations might be treated by young children as reliable. The point is that when children can verify things, unreliable or irrelevant rationales might influence their learning and behavior.

What does this have to do with hiring a nanny? Well, Paulina's a little young now, but in a few years, I know that she'll be learning things from her nanny (or any adult for that matter). It makes being sure how well the nanny communicates might have been a good move.

A broader point of the research Sommerville and I did together was that the rationales that children hear for why they're seeing something matter. Other researchers, such as Patrick Shafto at the University of Louisville, have begun to write about children's "intuitive pedagogy" - that children are sensitive to why they are seeing particular material in learning environments. I agree with this hypothesis, and suggest that a real challenge in early science education classrooms and children's museums that foster scientific thinking is to provide children with the appropriate rationales when children need them the most. There's a bigger question here though - how exactly do we do this? I'll try writing about this in a later post.

Parents and Experts

Dave Sobel — Mon, 24 Aug 2009 11:12:24 +0000

The heating system in our house is not so great. My wife and I knew this when we bought the house, and this summer, we decided that we would do something about it. We've had three heating contractors come to the house. Each said something different. The first wanted to install a heat pump. The second thought a heat pump wouldn't be effective and suggested a particularly kind of boiler to use with our existing radiators. The third suggested ripping out the radiators and replacing them with baseboard heating. Each seemed convinced of his assessment, and talked my wife and I through the advantages and disadvantages, and the price (which were all surprisingly similar).

My wife and I also recently took our daughter for her two month doctor visit. Paulina received her first vaccinations. Before the visit, we had talked with the doctor about what schedule of vaccinations she would be on and what combinations of vaccines were available. We have a friend who's also a pediatrician (with a daughter born on the same day as ours), and we talked about this issue with her. She gave us exactly the same advice our doctor gave us.

Let me start by saying that I'm not an expert in heating, nor am I a medical doctor. That said, the doctors inspired trust; the heating contractors, less so.

There's a lot of research on how children learn to trust the information they learn from others. Paul Harris and Melissa Koenig brought this issue to the field's attention in a set of articles over the past few years. They argued that most information in the world is not directly observable, and that you need other people and social interaction to function. In order to learn the meaning of words, for example, children need to know that other people supply them with information. So, Harris and Koenig argued that a serious question in cognitive development is when do children recognize that some people are more or less reliable sources of information, and do they treat the information generated by these individuals differently?

As an example, Koenig and Harris (2005) introduced preschoolers to two adult confederates. An experimenter showed the child and the two adults a set of familiar objects, one at a time. Each confederate was asked to label the objects. One always generated the reliable label (e.g., she called a shoe a "shoe"). The other always generated an unreliable label (e.g., she called the same shoe a "horse"). After being trained with a few familiar objects, the experimenter brought out a novel object - an object that children would not be able to reliably label. Each confederate referred to this object with a novel label (e.g., one called it a dax, the other called it a wug). Children were then asked what they thought the object was called. They usually went with the reliable individual's label.

I'd be remiss here if I didn't mention the work of my former undergraduate student Kathleen Corriveau. Kathleen, who is now a graduate student working with Harris, just published a paper in Child Development, which I think is particularly important for the interrelation between cognitive development and parenthood. Working with colleagues who study parent-child relationships, she took a group of preschoolers and gave them procedures similar to the set up that Koenig and Harris used. In one of these tasks, children and two confederates saw pictures of animals that were made to be ambiguous - they were altered to appear like 50% of one animal and 50% of another. I attached a link to the paper, which has some examples:

http://www.people.fas.harvard.edu/~kcorriv/Papers/corriveau%20et%...

Unlike Koenig and Harris's procedure, where the two confederates were strangers, here, one of the confederates was the child's mother (the other confederate was a stranger). Five-year-olds saw these 50-50 pictures, and the experimenter asked the mother and the stranger what they thought this animal was called. The mother generated one interpretation of the picture; the stranger generated the other.

You might think that children would naturally go with the mother's interpretation, and indeed, that's what the majority of children did. However, it depended on the nature of the relationship children had with their mothers. These children were taken from a longitudinal study of attachment - at 15 months old, they had been given the classic "strange situation" task, which measured their attachment style with their mother. Children with secure relations with their mother (i.e., children who were securely attached) endorsed their mother's label about two-thirds of the time. Children whose attachment style was insecure-avoidant were less likely to respond this way (they responded more at chance levels), and children who were insecure-resistant were more likely to respond this way. Corriveau and colleagues cite Mary Ainsworth, who pioneered the study of attachment, as describing avoidant children as children who "typically explore the environment independently and avoid interaction with the caregiver" and resistant children as being "preoccupied with the caregiver to the detriment of independent and collaborative exploration of the environment" (p. 753). Seen in this light, these findings are not surprising - the avoidant children appear to trust their mothers less, and the resistant children are over-reliant on their caregiver.

This study suggests that the relation children have with their parents influences the extent to which they view their parents are reliable sources of information. This is important because children probably rely on their parents as sources of information for more than just the meaning of words (which typically coincide with others' utterances). Any coherent set of beliefs about convention or unobservable events must be acquired from others. For example, children are told to wash our hands to get rid of germs that never actually see. They're also told that they (and everything for that matter) are made of tiny little particles that we theorize (with good reason) are there. But even more than that, one has to wonder what the foundations of political, economic, and religious beliefs are if not our parents.

A degree of skepticism is healthy, and probably worth fostering in children - otherwise, children might never seek out information on their own. But it's also the case that children's belief in the reliability of individuals - particularly the ones supplying them with the most information - is critical.

Also, there's an important lesson for cognitive development researchers here - researchers tend to take a group of children and present their average behavior as what they can do. Corriveau's study suggests that this isn't always the right approach - there are contextual factors that explain (some of) the noise in children's responses, particularly on ambiguous or difficult tasks. Sometimes variance in children's responses is just that - but sometimes it's explained by other cognitive or social factors that influence development. This is worth exploring as part of a mechanism for cognitive development, and is often not given careful consideration.

For the record, we did vaccinate our daughter (there wasn't really any question about our doing this), and we're going with the baseboard guy. The consensus associated with the former was nice (it turns out that Corriveau has also performed studies that suggest consensus is critical for learning from others). Because there wasn't consensus with the heating contractors, I'll just say that baseboard was the option my wife and I had thought about prior to having the contractors visit. But, how we (and children) make decisions based on the interaction between existing beliefs and present data is another topic.

75 Days Later

Dave Sobel — Mon, 17 Aug 2009 01:49:31 +0000

So, my intention was to start this blog with approximately 75 days before my wife gave birth to our first child. That didn't happen. Instead, I'm writing this approximately 75 days after my daughter was born (stuff happens). Anyway, for purposes of introduction, I'm Dave. My wife Lisa gave birth to our daughter Paulina in June. I've been a professor in the Department of Cognitive and Linguistic Sciences at Brown University for the last eight years. I teach cognitive development and cognitive psychology and my research focuses on causal reasoning, social cognition, and word learning in infants and preschoolers. My goal in writing this blog is to think about issues in cognitive development as they relate to fatherhood.

When Lisa and I moved in together, I became a dad to Bunny, her cat. A little later on, Mousy, our second cat, joined the family (yes, we name our cats after other animals - it will probably confuse the kid, but I'm not going to write about that today). Lisa and I find a lot of humor in the way our cats interact with the world, and I often emphasize that humor by both talking to and for the cats (basically narrating their actions). Unsurprisingly, the voice I use to talk to the cats is not the same voice I use to talk to Lisa (or other adults) - it's happier, higher pitched, and has lots of modulation.

None of this seemed irregular to me until Lisa became pregnant and I started having conversations with her belly (and then later with Paulina). What Paulina understood while she was not yet born is an open question (perhaps one I'll write about later). What's important is that both Lisa and I noticed that I talked to Paulina in pretty much the same voice that I use to talk to the cats.

The idea that adults talk to infants in a different way from other adults is not novel. There's a long literature on the importance of infant-directed speech and how such speech might benefit infants' language learning. But there's also a set of studies describing how infant-directed speech differs from pet-directed speech. My favorite example is a study by Hirsh-Pasek and Treimen in 1982. They asked mothers with dogs to talk to their infants and their dogs separately and analyzed the speech patterns. Basically, they found some differences between the two patterns of speech.

A more recent study by Burnham, Kitamura, and Vollmer-Colla (2002) recorded mothers talking to their infants, pet cat or dog, and to another adult. (Incidentally, I like this study a little more because they investigated both cats and dogs - we have cats in the house. Hirsh-Pasek and Treiman called their study an investigation of "doggerel", and I was glad to learn there wasn't a difference between talking to your cat or your dog). Burnham et al. found that infant-directed speech was similar to pet-directed speech with regards to pitch (and both were different from adult-directed speech). But they also found that infant-directed and pet-directed speech differed in affect. When mothers talked to their children, they used greater affect than when they talked to their pet (unsurprisingly, both kinds of speech had more affect than adult-directed speech). There are some other differences between infant and pet-directed speech as well (basically about how vowels are pronounced).

So these studies are nice, but they raise some questions to me. First of all, do we talk to all of our pets this way? I run a developmental lab that works with a lot of preschoolers, so we kept fish in the lab for many years. I can't recall ever talking to the fish in anything but adult-directed speech (of course, I can only recall talking to the fish once). But, for those out there with pet lizards, hamsters, or snakes, is there a difference? Might "pet directed speech" get more like adult-directed speech as your pet goes down the evolutionary ladder? I believe there's a study out there that found no difference between the pet directed speech directed to a live dog and a robot dog (I don't remember the citation off the top of my head - and if I'm making this up, it's a nice study to contemplate). But participants are probably pretending that the robot dog is real, so this might not tell us much about the evolutionary scale.

Moreover, investigations of pet directed speech made me wonder about the nature of infant directed speech. I've always thought that IDS was natural - something human beings just did in the presence of an infant (there are other examples of this - a few years ago, I saw a conference presentation by Fernandez-Duque who with his colleagues found that undergraduates with no experience feeding babies opened their mouths to elicit that behavior from the infant). So, this may be true, but investigating PDS made me wonder if the IDS we generate changes over time. Has anyone ever longitudinally investigated a group of parents generating IDS to their infant? Does the nature of the IDS change over the course of the child's development? Similarly, are their differences in how the same parent talks in IDS to their first and second child? I doubt there are, but my wife and I have a ways to go before we decide whether we want to test this firsthand.

But here's the question I thought of first: Do parents of infants with pets in the house generate IDS with the same acoustic features as parents of infants who have never had pets? My guess is yes, that they do, and that the infant directed speech you generate is independent of whether you own a pet. That said, I wonder if anyone has checked, and it would be nice to know the results. Incidentally, two branches of NIH just issued the following call for funding:

http://grants.nih.gov/grants/guide/rfa-files/RFA-HD-09-031.html

in which they are investigating "The Role of Human-Animal Interaction in Child Health and Development." I doubt this is the kind of project they have in mind. Either way, with all the stresses about being a new parent, it's nice to know that I (and all new fathers) have at least one thing going for me - I can talk to my baby.