Background references: Dr. Anders Ericsson's faculty webpage; The Cambridge Handbook of Expert Performance; the 5/7/06 Sunday NYT Magazine article "A Star is Made" ; and the Sunday 3/14/07 NYT Play Magazine article "How to Grow a Super-Athlete".
Now that we have a handle on the kinds of practice that do and don't lead to continued improvement over long timescales, perhaps now we can go on to ask about the kinds of student-teacher relationships that seem reliably to foster high achievement. For example, if we take a look at experts and the teachers they've had over the course of their education and careers, what patterns can we see in their student-teacher relationships? Does any teaching arrangement other than intensive 1-on-1 tutelage reliably support the effective work that Ericsson calls "deliberate practice"? What is the feedback cycle we see happening between a teacher and a student engaging in deliberate practice? How does the teacher's own domain expertise play a role, and how does the answer to that question depend on the discipline of study?
I have a feeling these questions would become acutely important to the design of a school based on the concept of deliberate practice.
Just a couple of informal observations on the logistics of expert tutelage: In music, the elite teaching structure seems to be 1-on-1: you have a personal voice coach, or the piano teacher sits on the bench with you. The same is true of many sports, such as tennis and golf. Even in team sports, where there may be just one head coach for an entire team, it still seems to be the case that the actual teaching moments are 1-on-1. This is partly because there are different roles on any team. For example, in basketball you have the point guard, the center, the small forward, etc. The coach doesn't say, "Today we will practice shooting the basketball." Instead, somebody from the coaching staff will take the center aside and work on the drop step; somebody else will take the guard aside and work on defensive stance; etc. Maybe these moments are 1-on-2,3,4 instead of 1-on-1. But they are certainly not 1-on-10.
This line of thinking has actually led to a new idea for how to handle homework in my college classes, which I am experimenting with right now in my second-semester introductory physics course at Bennington. The problems I'm trying to address here involve the role of homework. For one thing grading homework is a huge hassle for me, and the feedback comes to the students too late, and they have little incentive to absorb the careful solutions. My colleagues in the humanities tell a similar story - some of them barely read the papers they assign. Everyone knows it's not working. In the research university setting, where you have introductory physics classes with hundreds of students enrolled, this problem is being addressed by putting all of the homework assignments on the web. The students submit their answers and get rapid feedback, without the professor even being involved. It is ironic that the professors are so grateful to be spared any contact with the homework cycle, when research shows that it's almost solely by the doing of problems that students actually learn physics.
Here's the system I'm trying out. The cycle begins on Fridays with me distributing a problem set in the usual way. The students have a week to work on it. The following Friday, instead of collecting the homework, I give out careful solutions, and the students have about a week to study them. Then, the following Thursday, I have each student in to my office for a fifteen minute 1-on-1 session, like a mini oral exam. I choose a homework problem at random from the assignment, perhaps with a small numerical change, and the student solves the problem at the board. I see how the student is thinking about it all. There are some follow-up questions to pinpoint trouble.
The student gets an instant pass-fail grade. To compute the student's final semester grade, I will subtract the number of Fails from the number of Passes, add 1-to-3 points for the final lab project, and then simply rank the students bottom to top. Cut points are introduced into the scale based on whatever the usual considerations are.
We just had our first day of oral exams, and it was amazing. The students were nervous, but most of them did great. It was clear that they had worked on the problems, studied the solutions, re-worked the problems, and practiced the solutions. It is amazing what can happen when the work gets done! On the way out of my office after the exam, several students told me how excited they were by this system. They love that the expectations are so clear. One said, "I actually thought about the homework, which I've never done in a science class before."
For this system to make sense, you can't treat homework problems like a commodity, the way the publishing industry and many instructors treat them. The problems have to be so valuable that you are comfortable making the entire accountability system focused on solving them. The problems also have to be, to use Gauss's phrase, "few but ripe," because the students can't master very many problems in a short time. People give lip service to reducing the pace of introductory courses, but I am actually doing it, and the results are fascinating so far.
When I unveiled this strategy to the students on the first day of class this term, I could tell they were scared of having to come in and perform at the board in front of me. I made sure to say, "Look, this system is not about punishment. It's really about sincerity. I sincerely want you to understand and be able to do these problems. So let's work with a system that won't let us fail in that." After Week 1, they are on board.