On this page
Barbara Oakley and Terrence Sejnowski describe Learning How to Learn as one of the world's most popular online courses (Oakley and Sejnowski, 2019). Its Coursera page listed more than four million enrolled learners when we checked in October 2026. Oakley is a professor of engineering; Terrence Sejnowski co-teaches it. This post separates the parts backed by strong research from the parts that are useful metaphors.
What does the course teach?
The course description says it covers how the brain uses two very different learning modes, how it "chunks" information, illusions of learning, memory techniques and procrastination. Its first module includes a video titled "Introduction to the Focused and Diffuse Modes".
In a 2019 paper in *npj Science of Learning*, Oakley and Sejnowski describe how they built the course. They write that it begins by describing the difference between the brain's task-positive and task-negative ("default mode") networks, and that it also covers interleaved practice, deliberate practice, frequent low-stakes testing, chunking, procrastination, exercise and sleep. The same paper reports nearly 2.5 million registered learners in the course's first four years.
Are "focused" and "diffuse" modes real?
Treat them as a teaching frame, not a measurement. The course places them alongside brain-network research, but you cannot check which mode you are in, and the study techniques later in this post have much more direct evidence behind them.
The closest tested idea is incubation: setting a problem aside before trying again. A meta-analysis by Sio and Ormerod (2009) found a positive incubation effect overall, with more benefit for open-ended, creative tasks than for some insight puzzles. Two details matter for students. Longer preparation before the break gave a bigger effect, so the break helps after real effort, not instead of it. And filling the break with a demanding task reduced the effect, while light activity worked well.
Is chunking well supported?
Yes, under other names. In Chi, Feltovich and Glaser's 1981 study, physics experts sorted problems by the underlying principle, such as conservation of energy, while novices sorted them by surface features like "inclined plane". Sweller (1988) argued that such domain-specific schemas are the main thing that separates experts from novices. A chunk, in the course's language, is roughly this: a pattern you can recognise and use as a single unit.
What about the study techniques?
These are on the firmest ground. A major review by Dunlosky and colleagues (2013) rated practice testing and distributed practice as high-utility techniques, because they helped learners of different ages and abilities across many kinds of tests. Interleaved practice and self-explanation were rated moderate utility, promising but with less evidence.
A 90-minute session built on the evidence
Here is one way to put it together for a JEE or NEET-UG chapter, say electrochemistry.
Recall first (10 minutes). Before opening notes, write everything you remember from the last session. This is practice testing.
Focused block (35 minutes). Work problems with your phone in another room. If you are stuck on one for 10 minutes, mark it and move on.
Light break (10 minutes). Walk, stretch, refill water. Avoid scrolling or starting another hard task, since demanding activities reduced incubation benefits.
Return to the marked problems (20 minutes). Try them fresh. If still stuck, study a worked solution and explain each step aloud.
Build the chunk (10 minutes). For each problem type, write a two-line pattern card: "When I see X, I use Y because Z."
Plan the next spacing (5 minutes). Schedule a short recall test on this chapter for three days from now.
For more on related ideas, read does the Pomodoro technique work?, why rereading feels like learning, and building a study routine that lasts.
If you sit down and do not know where to start, Fermi's AI tutor works out what you already know and builds a plan for the session, so you begin with a clear next step.
