Dramatic example of discovery learning

Here is a rather dramatic example of unguided (or self-guided) discovery learning. See also this PBS video feature on the same story. A man in India put a computer with net access in a wall next to an alley so that poor children could use it. They developed quite a fluency in its use. The man nudged them once, showing them that the computer could play music, too. But for classroom purposes, I was also interested in another little experiment he did with classroom students. The part quoted from the article is in italics below:

Well, I tried another experiment. I went to a middle-class school and chose some ninth graders, two girls and two boys. I called their physics teacher in and asked him, “What are you going to teach these children next year at this time?” He mentioned viscosity. I asked him to write down five possible exam questions on the subject. I then took the four children and said, “Look here guys. I have a little problem for you.” They read the questions and said they didn’t understand them, it was Greek to them. So I said, “Here’s a terminal. I’ll give you two hours to find the answers.”

Then I did my usual thing: I closed the door and went off somewhere else.

They answered all five questions in two hours. The physics teacher checked the answers, and they were correct. That, of itself, doesn’t mean much. But I said to him, “Talk to the children and find out if they really learned something about this subject.” So he spent half an hour talking to them. He came out and said, “They don’t know everything about this subject or everything I would teach them. But they do know one hell of a lot about it. And they know a couple of things about it I didn’t know.”

That’s not a wow for the children, it’s a wow for the Internet. It shows you what it’s capable of. The slum children don’t have physics teachers. But if I could make them curious enough, then all the content they need is out there. The greatest expert on earth on viscosity probably has his papers up there on the Web somewhere. Creating content is not what’s important. What is important is infrastructure and access … The teacher’s job is very simple. It’s to help the children ask the right questions.

And I would mention now, based on the earlier papers I cited in my last post, if the teacher followed up with the children on a lesson about viscosity, they would learn even more. That is what has been variously termed a teachable moment, or a time for telling. The students explored on their own and learned a great deal, but still had many questions and areas they did not understand, plus areas they may have understood incorrectly. The teacher at that point could help fill in the gaps, so to speak.

Links + direct instruction vs. discovery learning

• Here’s a post at IGN about the use of videogames in education.
• Abdulaziz Ghuloum has written a tutorial about writing your own compiler. The difference is that it is a progressive tutorial rather than incremental. You don’t have to wait until the end before you can actually try the compiler out.

Richard Hake, one of the creators of the well-known Force Concept Inventory for assessing physics understanding, posts a lot of very informative material to the physlrnr, phys-l, and math-teach lists. Lately there was a re-hash again over a research article by Klahr & Nigem about direct instruction vs. discovery learning. In a typical unfairly designed psych experiment, Klahr compared direct instruction to a group in which students were left to fend completely on their own (unguided discovery). Students fared better in the direct instruction condition (although actually they had a discovery activity before the instruction began as well). People in the government and people who favor traditional schooling approaches jumped on that article as proof that direct instruction is best.

Here is Hake’s response to it from back in 2004. See the archives of the above lists for more recent discussions. I would also add some other references that counter the mis-interpretation of Klahr’s study, and clarify when and how direct instruction can be useful. There is an article entitled “A time for telling” from 1998 by John Bransford in the journal Cognition & Instruction. Also another 1991 article showed that giving students a discovery-learning experience before a lecture made for much better learning gains:

Brant, G., Hooper, E., & Sugrue, B. (1991). Which comes first: The simulation or the lecture? Journal of Educational Computing Research, 7(4), 469-481.

Here’s a related paragraph from a review I wrote a few years ago:

Simulation before instruction. A number of studies have found advantages to using simulations in sequence with other forms of instruction such as a lecture, but not necessarily during a lecture as earlier examples demonstrated (Thomas & Hooper, 1991). This makes sense when again realizing that simulations are environments that do not provide information to students so much as they require information from students to be used effectively. Particularly effective is allowing students to use a simulation before a lecture or other activity (Brant, Hooper, & Sugrue, 1991). Even though students may or may not fully understand the model underlying the simulation they use, after exploring the environment they may have formed questions or strategies for learning about the domain. So even when a simulation is not completely understandable to a student, it may engender a preparedness for future learning, and the students may attend differently to a subsequent lecture or other learning activity.

Here are some more general paragraphs from earlier in the review:

In a large-scale analysis of mathematics classrooms, Harold Wenglinsky found higher student achievement in classrooms using computers for simulation and data exploration activities. In contrast, achievement scores actually declined by six tenths of a year in eighth-grade classrooms in which computers were used for decontextualized drill activities instead of simulations (Wenglinsky, 1998).

Students learn more than just facts when using computer simulations. Njoo and de Jong (1993) have found that the educational benefits of simulations are revealed more in tests of intuitive knowledge, such as reasoning about “what if” scenarios, rather than tests of more factual knowledge. Similarly, Thomas and Hooper (1991) concluded in a review of research that the use of simulations is more effective when the educational goal is for students to transfer and apply knowledge to real-world problems rather than memorize facts or procedures. This is not as surprising when considering that the primary design principle of simulations, fidelity, is designed to lower the amount of work needed to transfer one’s knowledge to the actual system it is modeling.