Deze video hoort bij een artikel dat net gepubliceerd werd in Science met daarin 5 aanbevelingen:

1. Because trying all educational options — more than 205 trillion — to find out what works best is impossible, research should focus on how different forms of instruction meet different functional needs, such as which methods are best for learning to remember facts, which are best for learning to induce general skills, and which are best for learning to make sense of concepts and principles.

2. More experiments are needed to determine how different instructional techniques enhance different learning functions. For example, the optimal way to memorize facts may be a poor way to learn to induce general skills.

3. Take advantage of educational technology to further understand how people learn and which instructional dimensions can or cannot be treated independently by conducting massive online studies, which use thousands of students and test hundreds of variations of instruction at the same time.

4. To understand impact, build a national data infrastructure in which data collected at a moment-by-moment basis (i.e., cognitive tutors tracking daily how a student learns algebra over a school year) can be linked with longer-term results, such as state exams and performances in a next class.

5. Create more permanent school and research partnerships to facilitate interaction between education, administration and researchers. For example, the PSLC, funded by the National Science Foundation (NSF), gives teachers immediate feedback and allows researchers to explore only relevant theories.