If learning happens in the learner's head, what is the teacher for? The answer is not obvious, and an argument about it runs through the whole history of education. Socrates questioned his students instead of lecturing them. Medieval universities read texts aloud and explained them line by line. Progressive educators in Dewey's tradition wanted children to learn by doing. Research over the past half century has sharpened the argument without ending it.

Guidance and the limits of the mind

One influential framework is cognitive load theory, developed by John Sweller from the 1980s, which holds that working memory can handle only a few new elements at once, so instruction should avoid overloading it. A key finding is the worked example effect: beginners learn more from studying fully solved problems than from solving the same problems unaided. A novice who struggles alone may spend all their mental effort searching for a route and have little left to notice the general method. As learners gain expertise the advantage shrinks and can reverse, so experienced learners often gain more from solving problems themselves.

Direct instruction and discovery

These findings feed an old debate. In explicit or direct instruction, the teacher explains clearly, models a skill step by step, then guides practice before students work alone. In discovery or inquiry learning, students explore problems and build understanding themselves, with the teacher as a guide. The largest early test was Project Follow Through, a United States government program running from the late 1960s through the 1970s that compared many teaching models in poor communities. A highly scripted model called Direct Instruction posted some of the strongest results on basic skills, though critics questioned the study's design and noted that results varied a great deal from site to site.

Defenders of inquiry make serious points. Students who discover a principle may understand it more deeply, remember why it works and stay more curious. Real scientists and citizens must tackle problems that come with no worked example. The best evidence suggests a middle path: inquiry works best when it is well structured and comes after students have enough knowledge to reason with. Unguided discovery for novices tends to do poorly; guided inquiry for prepared learners can do well.

Tutoring

In 1984 the educational psychologist Benjamin Bloom reported that students taught one to one by a tutor, using a method called mastery learning, scored about two standard deviations above students in conventional classes: the average tutored student outperformed about 98 percent of the comparison class. He called it the two sigma problem: how could schools get such results without a tutor for every child? Later studies have found smaller effects, but well-run tutoring remains one of the most effective interventions known. Its cost is the obstacle.

Assessment: measuring and helping

Assessment comes in two broad kinds. Summative assessment measures what students have learned at the end of a unit or course. Formative assessment checks understanding during learning so that teaching can adjust. A quick exit question, answers held up on mini whiteboards, or a draft returned with comments are all formative. In a widely read 1998 review, Paul Black and Dylan Wiliam argued that strengthening formative assessment produced substantial gains, especially for lower achievers.

Feedback is not automatically helpful. A 1996 review by Avraham Kluger and Angelo DeNisi found that roughly a third of feedback interventions actually lowered performance. Feedback tends to help when it is specific and points to the task ("your second paragraph needs evidence"), and to hurt when it turns attention to the self ("you're just not a math person").

Mindsets, fairly weighed

The psychologist Carol Dweck popularized the idea of a growth mindset: the belief that abilities can be developed through effort and good strategies, rather than being fixed. Early studies were striking, and schools rushed to adopt mindset lessons. Later large studies and reviews found the effects much smaller than first claimed. A national study of about 12,000 American ninth graders, published in 2019, found that a short online mindset lesson modestly improved grades for lower-achieving students. That is a real and cheap benefit, but not a cure for weak teaching or poor schools.

Worked example: teaching fractions

A teacher introducing the addition of fractions with different denominators has two plans. In one, students get pizza diagrams and are told to figure it out. In the other, she works through two examples aloud, explaining why a common denominator is needed, then gives a partly solved problem, then a few problems for pairs, then some for each student alone. For novices the second plan fits the evidence on cognitive load and worked examples. But she can still borrow from inquiry: once students can add fractions, she poses a puzzle with no recipe, such as sharing three pizzas fairly among eight people, and lets them argue about solutions. Along the way she asks every student to hold up an answer on a mini whiteboard. That formative check shows her which students need another example before moving on, long before any graded test would.