A Blog by Carnegie Learning
Adaptive learning can go way beyond just right and wrong answers.
One of the buzzwords in education right now is “personalized instruction.” The problem with buzzwords, of course, is that you don’t know if you should get excited (or not) until you see the idea fully implemented. A weak version of personalized instruction, for example, is dividing a class into groups based on perceived student ability. Students are then given different homework assignments based on which group they’re in. A strong version might be that each student is given her own tailored instruction, based on her current knowledge level. The assignments that the student receives would then follow from the instruction that she received. However, instruction can be even more personalized that this. But how? To answer that, let’s take a detour.
If you’ve ever taken a computer programming class, one of the first concepts you learn is a “loop.” A loop is a set of programming instructions that repeat over and over again. To get a little more precise, there are three parts to a loop:
The fun thing about loops is that you can nest them inside each other. To make this a little more concrete, consider a music player like iTunes. In your musical library, you have Albums, which contain Songs. When you hit “play”, iTunes has a set of nested loops. The Inner Loop iterates (or loops over) the songs in an album, and the Outer Loop iterates over the albums in your collection (see Figure 1). The inner loop exits when it runs out of songs, and the outer loops stops when there are no more albums.
Figure 1. A high-level description of a piece of software that plays music.
An intelligent tutoring system (ITS) is a software package that is designed to teach students new concepts or to give them practice applying a new skill. One way to describe how intelligent tutoring systems are adaptive is by viewing them according to this two-loop model.1
MATHia is an example of an intelligent tutoring system because it provides personalized instruction in two ways. First, it provides personalized instruction at the level of individual problem-solving steps. Second, it tailors the problem set that a student sees. How is this accomplished?
First, MATHia has a nested structure of math problems. A workspace is composed of math problems, which can be decomposed into individual problem-solving steps. In other words, Problems contain Steps, much like Albums contain Songs (see Figure 2). Each problem has a slightly different characteristic.
When a student is using MATHia, the inner loop iterates over problem-solving steps, and the outer loop iterates over problems. This makes MATHia truly adaptive – it gives feedback not only on problems, but also on individual steps. This is critical because step-based feedback is what drives learning.2
Figure 2. A high-level description of an intelligent tutoring system.
It is secondarily adaptive because it intelligently selects the next problem. It does this by looking at the skills that the student has currently mastered, and it selects a problem that provides the student with the maximal opportunity to practice the unmastered skills.
Every student is unique, and as a result, they each solve math problems differently. By analyzing not only whether a student got a question right or wrong, but each step they took to solve the problem, MATHia helps students understand exactly why they got something right before they move on.
That’s the kind of personalized learning that lives up to the hype.
1 VanLehn, K. (2006). The Behavior of tutoring systems. International Journal of Artificial Intelligence in Education, 16(3), 227-265.
2 Anderson, J. R., Corbett, A. T., Koedinger, K. R., & Pelletier, R. (1995). Cognitive tutors: Lessons learned. The Journal of the Learning Sciences, 4(2), 167-207.
Dr. Bob joined Carnegie Learning in 2009 as a Cognitive Scientist. He received his PhD in Cognitive Psychology in 2005 from the University of Pittsburgh under the direction of Dr. Michelene T.H. Chi, and he received additional training at the Pittsburgh Science of Learning Center (PSLC) as a postdoctoral fellow with Dr. Kurt VanLehn and Dr. Timothy J. Nokes-Malach. In his spare time, Dr. Bob publishes a blog entitled Dr. Bob's Cog Blog, and is the author of the book Cognitive Science for Educators: Practical suggestions for an evidence-based classroom. The unifying theme that runs throughout all of these activities is a drive toward helping every student become an expert in a domain of her or his choice. When he isn’t thinking about cognitive science, which is rare, Dr. Bob enjoys long-distance running, mountain biking, and traveling with his wife.Explore more related to this author
Giving feedback not only on problems, but also on individual steps, is critical because step-based feedback is what drives learning.
Dr. Bob Hausmann, Learning Technology Scientist, Carnegie Learning