Monkey See, Monkey Do
In a New York Times article, Children Learn by Monkey See, Monkey Do. Chimps Don’t, Carl Zimmer describes his daughter’s participation in some research about problem solving and imitation in chimpanzees and human children.
If chimpanzees watch someone open a box with a routine that includes several unnecessary steps, they just ignore those steps and open the box the obvious way. Human children imitate: They leave in the unnecessary steps even though they can easily see how to open the box the straightforward way.
Mr. Lyons sees his results as evidence that humans are hard-wired to learn by imitation, even when that is clearly not the best strategy. If he is right, this represents a big evolutionary change from our ape ancestors. Other primates are bad at imitation. When they watch another primate doing something, they seem to focus on what its goals are and ignore its actions.
As human ancestors began to make complicated tools, figuring out goals might not have been good enough anymore. Hominids needed a way to register automatically what other hominids did, even if they didn’t understand the intentions behind them. They needed to imitate.
Not long ago, many psychologists thought that imitation was a simple, primitive action compared with figuring out the intentions of others. But that is changing. “Maybe imitation is a lot more sophisticated than people thought,” Mr. Lyons said.
We don’t appreciate just how automatically we rely on imitation, because usually it serves us so well.
Let’s speculate: Usually, imitation and problem solving aren’t at odds: A child models an experienced adult who shows the proper way to do something. (You’ll find a discussion of the critical role of modeling in Grow With the Flow, on pages 158-160. That portion isn’t online yet.) Sometimes the task is one where the child would eventually develop the same strategy by problem solving. Sometimes the task is so complex that generations of problem solving have been involved, and the demonstration summarizes what has been learned. In that case, the reason for all the steps is unlikely to be either obvious or intuitive, so strict imitation is adaptive. As the skills involved in acculturation become more and more complex, modeling may more and more often avoid errors likely to develop from strictly individual problem solving.
We know that other primates imitate. We know they can be inspired problem solvers. We know that humans have pushed both of those cognitive abilities considerably forward. From this recent research, it appears that in human children at least, the default is to imitate when imitation and direct problem solving are in conflict, and that this is likely to lead to efficient learning and knowledge transmission.
Language adds another layer to the process. Our kids often simply watch us do something and imitate. Often we don’t notice they’re doing this – we’re getting the job done, they’re watching how we do it. But when we’re in our teaching mode, adults naturally combine language with demonstration, which is usually very efficient. There are times when modeling is primary: “Hold it this way.” There are times when language does what modeling can’t: “We hold it this way because…” But usually the process is a smoothly integrated one, with language and demonstration, modeling and questioning combining to give efficient learning.
Speculating on apace: Children’s frontal lobes, primary home to our executive functioning, aren’t fully developed until the late teens or early twenties. That is to say, the research described here involves kids who don’t yet have fully developed reasoning ability combined with an ability to direct that ability towards complex goals.
The middle years – when language is well developed, but the frontal lobes aren’t – is one of the great periods of knowledge acquisition. Brains in this era seem more like sponges than at any other point in our life span. During these years, we seem especially attuned to knowledge acquisition – both declarative knowledge (like the names of dinosaurs) and procedural ("how to") knowledge. That’s important, since these are the years when children develop much of the knowledge that they will need to function in their culture. It seems intuitively reasonable that imitation – modeling – would be a preferred mode in these years. It isn’t yet time to challenge, so much as to understand what and how.
Imagine yourself doing this same experiment: demonstrating how to open a box, where the person watching could easily figure out the simplest way to do the task, but where you add in unnecessary extra steps. We know from this recent research that a child would imitate with the extra steps. I can guess how a teenager would be likely to respond. ("That’s stupid.") But what about an adult? My guess: the adult would either use language ("Why did you add the extra steps?” or “Should I imitate your procedure?") or else open the box the sensible way. That would represent another level of human cognitive complexity. But that, in turn, makes me wonder: What would young chimps do? Wouldn’t it be a kick if durng their younger (acculturation) phases, chimpanzees imitated like human kids?
**********************
It seems to be Primate Research Speculation Week around here. I just posted a comment about girl bullies, amusing myself by wondering if they represent a remnant of primate dominance behavior, and whether that will still prove to be adaptive, or whether it may obtain short-term goals like status in junior high school at the expense of long-term goals like satisfying careers.
Comments
Add a comment
Sorry, the comment form is closed at this time.
I’ll bet this has something to do with why kids learn languages faster than adults: unencumbered by worries about doing it wrong, or a need to understand why the words go together the way they do, they imitate freely — sometimes with successful results, sometimes not — and communicate freely.
I’ll bet it also has something to do with early life experiences that become lifelong barriers to mathematics. Bad modeling of how to approach math, how to think about it, lays down patterns that undermine every subsequent experience. Math is abstract enough that counterproductive approaches learned early on through bad modeling never get corrected by logical observation of the “obvious way.”
I’m not so sure about the connection to math, precisely because math is so abstract. In my experience teaching math, teaching people how to think about math in general is quite different from teaching them specific mathematical methods, though the two overlap in the (extremely important) realm of explaining why the methods work. Understanding a method is essential to being able to see intuitively when and how to apply it, and knowing what its limitations might be. Often the biggest barrier for people who struggle with math is coming to terms with the idea that the techniques of math are not arbitrary, but rather work for very specific reasons, whose roots are in our most basic definitions of arithmetic.
Many people approach math as a collection of techniques to be memorized, and are surprised when I first ask them if they know why they can do something; it seems that many people have never really considered this an important question. But, invariably, when I give an explanation of some rule, people are better able to remember it, to know when to use it, and to see how it fits into larger problems of the sort which one actually wants to solve. It is certainly true that many people are turned away from math by bad early education, but I think the problem is more fundamental than being made to copy a bad model; the fundamental problem is that copying a model is not a useful way of approaching math even if the model is a good one. No matter how good the model one copies, copying a model is not a good way to approach math. Thus, the most important thing I try to do when I teach mathematical ideas is to communicate that the ideas can and should be understood, that the techniques of math are valid and useful for specific reasons. If someone has been taught to copy bad models, the surest way to introduce them to more efficient or useful ones is to first teach them to think analytically and critically rather than blindly accepting what they are given; Even if someone has had good models to copy, independent thought is essential to the mathematical process. Thus, the bad models themselves are not so problematic as the mindset created by copying models of any quality.
Paul, It has been one of the “things we’re sure we know” since the ’60s that our brains are highly structured and prepared to learn language. Whether that confidence extends to second languages, I’m not sure. We assume this ability to learn language(s?) with ease and competence is lost in the teens, presumably as brains trim down unused connections.
But whatever is going on with long-term wiring, it does seem like uninhibited modeling should make language acquisition easier – I’ve heard more than one adult say that they always speak much better French when they have a glasss of wine!
However, if it turns out that modeling and imitation are also developmental phases that fade in adolescence, the difficulty of learning a language would be increased by both changes – the wiring of adulthood would both inhibit imitation and make the language learning per se more difficult.
Andy, Agreed! I always remember the student from the École Poytechnique who said to me about math, “If you don’t understand it, it’s wrong.” His point was, that without the understanding, something would inevitably go wrong with the answer, probably when you most needed it.
I suspect that most math instruction in the U. S. suffers from “bad models” in more than one way:
* According to a hastily checked source, 69% of U. S. Math students are being taught by teachers who lack a major or certification in math. I know from my own experience that many teachers teach math essentially by rote because their own understanding of the process behind the algorithm is shaky. However good they are as teachers, how can they lead kids to an understanding of what they’re doing?
* Textbooks often feature brave prose about understanding. But they seem to a casual observer to be heavy on computational explanations, with interspersed “real-life examples,” which kids typically skip, and free of more than a passing explanation of the underlying logic of what is being done.
* Under the pressure of the accountability drummer, there’s no time to teach beyond what will be measured – on a multiple choice exam. And with loads and expectations the way they are, few teachers can spend time helping individual kids work to an understanding.
* Finally, as you note, the model is to blindly copy these inadequate models, not to think about what lies behind them.