Following our annual tussle with hundreds of studies of merit, we’ve pared them down to 10 you shouldn’t miss—from what AI can (and can’t) do to the neuroscience of brain synchrony.
December 7, 2023For those of us hoping for a quiet, back-to-normal kind of year, the research coming out of 2023 might disappoint. A rising tide of teenage mental health issues sent researchers scurrying for answers, and the sudden ascendance of AI posed a new threat to codes of academic conduct and caused some educators to forecast the end of teaching as we know it (we’re here to dispel that myth).
There was plenty of good news in the mix—and fascinating news, too. Neuroscientists continued to push the envelope on mapping the human brain, using cutting-edge technology to get a sneak peek at the “brain synchrony” between students and teachers as they learn about complex topics, and a comprehensive review of social and emotional learning confirmed, once again, that there’s no substitute for caring, welcoming school environments.
Finally, we did our due diligence and unearthed classroom strategies that can make a big difference for students, from the use of math picture books to a better, more humane way to incorporate tests and games of knowledge into your classroom activities.
In case anyone thought the jury was still out on the Turing test, which proposes a hypothetical threshold at which humans and machines respond indistinguishably to a prompt—more evidence recently came in, and it’s becoming increasingly difficult to tell who’s testing who.
Researchers from the University of North Carolina set a “deep neural” AI model to work on a college-level anatomy and physiology textbook, after first training the software to recognize important information. The AI took stock, pondered in its fashion, and then dutifully produced 2,191 test questions tied to learning standards, which a panel of teachers judged to be “on par with human-generated questions in terms of their relevance to the learning objectives.” Remarkably, the instructors also said they’d consider adopting the machine-generated questions for their courses.
That’s spooky, but not without its silver linings. Test creation is time-consuming for teachers, and one knowledgeable educator who took AI for a test drive says that it performs well on other tasks like planning lessons, writing instructions, and even composing emails to parents. New AI-powered tools like Diffit, Curipod, and MagicSchool.ai, meanwhile, are starting to sound like revolutionary teaching aids.
Concern that the end of human teaching is one software release away is premature: Studies we’ve reviewed suggest that AI still requires a lot of fine-tuning, and in July of 2023, researchers concluded that without human intervention, AI is atrocious at mathematics, performing poorly on open-ended problems and routinely flubbing even simple math calculations. To be useful, it turns out, AI may need us more than we need it.
No one likes tests—except the three authors of a 2023 study, apparently. The trio, who have experience as teachers and researchers, sing the praises of virtually every kind of test, quiz, and knowledge game, asserting that such assessments should be frequent, low-stakes, highly engaging, and even communal. Their rationale: When properly designed and stripped of dread, tests and quizzes dramatically improve “long-term retention and the creation of more robust retrieval routes for future access,” a well-established phenomenon known as the testing effect.
The study is a fascinating, granular look at the mechanics of testing and its impacts on learning. Here are some of the highlights:
Mix it up: To maximize student engagement, quiz students frequently—but don’t let the format get stale. In their analysis, the authors endorse testing formats as varied as multiple choice, cued-recall tests, clickers, fill-in-the-blank, short answer, and contests of knowledge.
Be competitive: When designing multiple-choice or true-false tests, opt for “competitive alternatives” in your answers. For example, when asking “What is the hottest terrestrial planet?,” proffer Venus, Mars, and Mercury instead of Venus, Uranus, and Saturn—because “Uranus and Saturn aren’t terrestrial planets.” Competitive alternatives require students to scrutinize all options, the authors hypothesize, leading them to retrieve and consider more learned material.
Pretest: Quizzing students on material they haven’t yet learned improves long-term performance “even if [students] are not able to answer any of those questions correctly,” according to the researchers. Notably, pretesting can also lead to “a reduction in mind wandering” during subsequent lessons.
Get communal: Asking students to take tests in groups can improve retention and motivation while reducing anxiety. Consider focusing on specific rather than open-ended questions, the authors caution, since students can sometimes “recall and remember information less accurately” when working together.
Pass it on: Teach students to self-test by “summarizing the main points from a lecture… without looking at any notes,” or by meeting in “small study groups where the students practice testing one another—an activity that many students already report doing.”
Like the proverbial canary in the coal mine, subtle shifts in a teacher’s tone of voice—a sharp rise in volume or a sudden barrage of repeated instructions born of frustration—can be the first sign that something’s awry in the classroom, disturbing a fragile equilibrium and leading students to clam up or act out, a study published late in 2022 suggests.
Researchers observed as teens and preteens listened to instructions given by teachers—“I’m waiting for people to quiet down” or “It’s time to tidy up all of your belongings,” for example—delivered in warm, neutral, or controlling tones. While the effect was unintended, an authoritative tone often came off as confrontational, undermining students’ sense of competence and discouraging them from confiding in teachers. Warm, supportive tones, on the other hand, contributed to a classroom environment that reinforced learning across multiple social and academic dimensions like sense of belonging, autonomy, and enjoyment of the class.
It takes years to find the right tonal balance, says experienced middle school teacher Kristine Napper. “Neither high expectations nor kind hearts can do the job alone,” she coaches. Instead, teachers should strive for a warm, supportive tone and then draw on that “wellspring of trust to hold students to high standards of deep engagement with course content.”
In 2021, we reported that as students progressed through a computer science course, the learning material left neural fingerprints that mirrored brain activity in other students, the teacher, and experts in the field. “Students who failed to grasp the material,” we wrote, “exhibited neural signatures that were outliers; they were drifting.” But the brain patterns of students who performed well on a later test aligned strongly with other top performing students—and with the teacher and experts, too.
Intriguingly, even abstract concepts—those that lack any physical attributes—appeared to trigger similar mental representations in students’ minds, attesting to the remarkable cognitive flexibility underlying human communication and knowledge sharing.
A 2023 study using electroencephalography (EEG) largely confirms those findings. High school science teachers taught groups of young adults fitted with electrodes about science topics such as bipedalism, habitats, and lipids. Researchers found that stronger “brain synchrony” between peers—and between students and teachers—predicted better academic performance on follow-up tests, both immediately and a full week later.
Together, these studies underscore the importance of scholarly expertise and direct instruction, but also hint at the downstream power of peer-to-peer and social learning. As knowledge passes from teachers to learners to greater and lesser degrees—some students grasp material quickly, others more slowly—an opportunity to distribute the work of learning emerges. When advanced students are paired with struggling peers, assisted by nudges from the teacher, groups of students might eventually converge around an accurate, common understanding of the material.
The old adage that a picture is worth a thousand words—and two are worth two thousand—might be expressed, mathematically, as a simple multiplication formula. But can reading math picture books really multiply learning?
A 2023 review of 16 studies concluded that math books like Are We There Yet, Daddy? and Sir Cumference and the Dragon of Pi improved student engagement and attitudes toward math; strengthened kids’ grasp of math representations like graphs or physical models; and boosted performance on tasks like counting to 20, understanding place value, and calculating diameters. In early childhood, in particular, math picture books worked wonders—one study found that young students “tend to anticipate and guess what will happen next, resulting in high engagement, aroused interest in understanding the problems, and curiosity in finding solutions”—but even middle school students seemed mesmerized by math read-alouds.
Importantly, math picture books weren’t a substitute for procedural fluency or mathematical practice. Typically, the authors noted, teachers bracketed math units with picture books, introducing a mathematical concept “in order to prepare [students] for the upcoming practice and activities,” or, alternatively, used them to review material at the end of the lesson.
It’s hard to move the needle on student writing. Hours of close reading followed by the addition of dozens of edifying margin notes can swallow teacher weekends whole, but there’s no guarantee students know how to use the feedback productively.
In fact, without guidance, revisions tend to be superficial, a new study suggests—students might correct typos and grammatical mistakes, for example, or make cursory adjustments to a few ideas, but leave it at that. A promising, time-saving alternative is to deploy rubrics, mentor texts, and other clarifying writing guidelines.
In the study, high school students were graded on the clarity, sophistication, and thoroughness of their essays before being split into groups to test the effectiveness of various revision strategies. Students who consulted rubrics that spelled out the elements of an excellent essay—a clear central thesis, support for the claim, and cohesive overall structure, for example—improved their performance by a half-letter grade while kids who read mentor texts boosted scores by a third of a letter grade.
Rubrics and mentor texts are reusable, “increase teachers’ efficient use of time,” and “enhance self-feedback” in a way that can lead to better, more confident writers down the line, the new research suggests.
Parents, teachers, and medical professionals are wringing their hands over the alarming, decades-long rise in teenage mental health issues, including depression, feelings of “persistent hopelessness,” and drug addiction.
The root causes remain elusive—cell phones and social media are prime suspects—but a sprawling 2023 study offers another explanation that’s gaining traction: After scouring surveys, data sets, and cultural artifacts, researchers theorized that a primary cause is “a decline over decades in opportunities for children and teens to play, roam, and engage in other activities independent of direct oversight and control by adults.”
Scholarly reviews of historical articles, books, and advice columns on child rearing depict an era when young children “walked or biked to school alone,” and contributed to their “family’s well being” and “community life” through meaningful chores and jobs. If that all feels vaguely mythical, data collected over the last 50 years reveals a correlation: frank admissions by parents that their children play outdoors independently less than they did, and significant drops in the number of kids who walk, bike, or bus to school alone or are allowed to cross busy roads by themselves. In the U.S., for example, a government survey showed that 48 percent of K–8 students walked to school in 1969, but by 2009 only 13 percent did.
Risky play and unsupervised outdoor activities, meanwhile, which might “protect against the development of phobias” and reduce “future anxiety by increasing the person’s confidence that they can deal effectively with emergencies,” are often frowned upon. That last point is crucial, because dozens of studies suggest that happiness in childhood, and then later in adolescence, is driven by internal feelings of “autonomy, competence, and relatedness”—and independent play, purposeful work, and important roles in classrooms and families are vital, early forms of practice.
Whatever the causes, young children seem to sense that something’s off. In one 2017 study, kindergartners who viewed images of fun activities routinely struck pictures that included adults from the category of play, rejecting the role of grown-ups in a domain they clearly saw as their own.
It’s an often-fiery but ultimately dubious debate: Should teachers employ direct instruction, or opt for inquiry-based learning?
At its core, direct instruction often conveys information “by lecturing and by giving a leading role to the teacher,” researchers explain in a 2023 study examining the evidence supporting both approaches. Critics typically focus solely on its passive qualities, a straw-man argument that ignores activities such as note-taking, practice quizzes, and classroom discussions. Opponents of inquiry-based learning, meanwhile, characterize it as chaotic, akin to sending students on a wild goose chase and asking them to discover the laws of physics on their own—though it can actually unlock “deep learning processes such as elaboration, self-explanation, and metacognitive strategies,“ the researchers say.
Both sides misrepresent what teachers actually do in classrooms. Instructional models are “often combined in practice,” the researchers note, and inquiry-based learning is usually supported with direct instruction. Teachers might begin a lesson by leading a review of key concepts, for example, and then ask students to apply what they’re learning in unfamiliar contexts.
Let the debate rage on. Teachers already know that factual fluency and the need to struggle, flail, and even hit dead-ends are integral to learning. Teaching is fluid and complex and spools out in real time; it resists every effort to reduce it to a single strategy or program that works for all kids, in all contexts.
It’s déjà vu all over again. The researcher Joseph Durlak, who put social and emotional learning on the map with his 2011 study that concluded that SEL programs boosted academic performance by an impressive 11 percentile points, was back at it in 2023—working with an ambitious new team, led by Yale professor Christina Cipriano, on a similar mission.
The group just published a comprehensive meta-analysis that surveyed a whopping 424 studies involving over half a million K–12 students, scrutinizing school-based SEL programs and strategies such as mindfulness, interpersonal skills, classroom management, and emotional intelligence. The findings: Students who participated in such programs experienced “improved academic achievement, school climate, school functioning, social emotional skills, attitudes, and prosocial and civic behaviors,” the researchers concluded.
Intriguingly, SEL remained a powerful driver of better cultures and student outcomes into the middle and high school years, a reminder that there’s no cutoff point for building relationships, teaching empathy, and making schools inclusive and welcoming.
While politicians continue to stoke controversy on the topic, there’s actually widespread support for SEL, as long as it’s connected to better academic outcomes. A 2021 Thomas B. Fordham Institute survey revealed that parents reacted negatively to classroom instruction labeled “social and emotional learning,” but were favorably disposed when a single clause was added—calling it “social-emotional & academic learning” turned the tide and secured parental buy-in.
In the United States, the teaching of reading comprehension has ping-ponged between skills-based and knowledge-based approaches. In 2019, things appeared to come to a head: While reading programs continued to emphasize transferable skills like “finding the main idea” or “making inferences,” the author Natalie Wexler published The Knowledge Gap, an influential takedown of skills-based methods, and a large 2020 study from the Thomas B. Fordham Institute concurred, noting that “exposing kids to rich content in civics, history, and law” taught reading more effectively than skills-based approaches.
Now a pair of new, high-quality studies—featuring leading researchers and encompassing more than 5,000 students in 39 schools—appears to put the finishing touches on a decades-long effort to push background knowledge to the forefront of reading instruction.
In a Harvard study, 3,000 elementary students participated in a yearlong literacy program focused on the “knowledge rich” domains of social studies and science, exploring the methods used to study past events, for example, or investigating how animals evolve to survive in different habitats. Compared to their counterparts in business-as-usual classes, the “knowledge based” readers scored 18 percent higher on general reading comprehension. Background knowledge acts like a scaffold, the researchers explained, helping students “connect new learning to a general schema and transfer their knowledge to related topics.”
In the other study, a team of researchers, including leading experts David Grissmer, Daniel Willingham, and Chris Hulleman, examined the impact of the “Core Knowledge” program on 2,310 students in nine lottery-based Colorado charter schools from kindergarten to sixth grade. The approach improved reading scores by 16 percentile points, and if implemented nationally, the researchers calculated, might catapult U.S. students from 15th to fifth place on international reading tests.
The pendulum is swinging, but the researchers caution against overreach: There appear to be “two separate but complementary cognitive processes involved in development and learning: ‘skill building’ and ‘knowledge accumulation,’” they clarified. We may have the balance out of whack, but to develop proficient readers, you need both.