From Speculation to Science: How Do People Learn?

Ask any American teacher to speak for thirty minutes on a subject or topic they teach in their work, and they’ll deliver. I’m speaking from my own experiences here. Of course, there will be systematic differences in the kinds of topics they select depending on grade level taught.

Elementary grade teachers may hone in on a special teaching interest—many have a thing for math, say, or writing, or science. Secondary teachers likely will choose a topic from their curriculum or from their undergraduate years.

Then ask them to speak for thirty minutes on how people learn. Elementary grade teachers will have a lot to say, laced with classroom anecdotes, likely heavy on personal narratives. Secondary teachers may show some discomfort at first and end their talk early.

Teachers all get subject matter preparation in undergraduate degree programs, though specific disciplinary degrees dominate the secondary level while elementary teachers get a smattering of disciplines organized as liberal studies. They then get coursework and supervised clinical experiences once admitted to a credential program, though in many states a candidate can get an undergraduate degree and a credential simultaneously.

Many will tell you they learned far more from their clinical experiences than from their coursework. Most will tell you, honestly, that they were never taught a science of learning.

Whether elementary or secondary, in America, teaching is an individual sport, and learning is what the teacher says it is. Each player has a unique style, and it can cause trouble administratively to try to harmonize their work. The one who follows the teacher’s manual religiously could be just as effective as the one who invents every lesson on their own. Who’s to know unless you look at standardized test scores, and even then who’s going to be convinced? Teaching can be a lonely profession.

Unlike preparation to practice medicine, teacher preparation is an eclectic business. Fifteen states don't require accreditation for teacher preparation programs. Many universities treat education departments as cash cows, funding them below other professional schools. Clinical experiences range from eight weeks to thirty. Some programs require deep study of child development; others require none.

A century after the Flexner Report standardized medical education around laboratory science, rigorous admissions, and tightly controlled clinical training, teaching still lacks an equivalent foundation. There simply is no transparent, consistent approach to making teachers.

In many states, alternative certification programs provide fast-track pathways into teaching and do not offer courses in child development. In some states that have cut back on training requirements, even traditional programs lack such courses. According to the National Council for the Accreditation of Teacher Education (NCATE), most teachers have not developed the capacity to provide developmentally oriented instruction. Without these understandings, the study of curriculum or teaching methods remains at the surface of professional practice (see Darling-Hammond et al. 2023).

Progress in Learning Science

There's a reason for this gap, and it isn't just negligence. Understanding the mind—and the thinking and learning that the mind does—remained an elusive quest for centuries, in part because researchers lacked the tools to study it.

Before the late nineteenth century, the study of how people learn was a matter for philosophers and theologians. The first systematic attempts to study the human mind through scientific methods began in 1879 in Wilhelm Wundt's Leipzig laboratory, where researchers tried to analyze consciousness through introspection, i.e., asking subjects to reflect on their own thought processes.

By the early twentieth century, behaviorism had mesmerized the field. John B. Watson declared in 1913 that consciousness was neither a definable nor a useable concept and that psychology must restrict itself to observable behaviors and the stimuli that control them. For the next half-century, learning was understood as forming connections between stimuli and responses, driven by external rewards and punishments.

In the late 1950s a new field emerged: cognitive science. From its inception, it approached learning from a multidisciplinary perspective that included anthropology, linguistics, philosophy, developmental psychology, computer science, neuroscience, and several branches of psychology.

New experimental tools and methodologies made it possible for scientists to begin serious study of mental functioning, to test theories experimentally rather than simply speculate about thinking and learning. In recent decades, researchers have also developed insights into the importance of social and cultural contexts, and today experimentalism has made room for ethnography and phenomenology as tools to understand subjective experiences in teaching and learning.

This research is now synthesized in two landmark reports from the National Academies of Sciences, Engineering, and Medicine: How People Learn (2000) and How People Learn II (2018). These aren’t policy documents. Nor are they prescriptions for teachers to apply on Monday morning. They’re consensus reports produced by the National Research Council through the same rigorous process the Academies use for climate science or vaccine safety.

We now know a thing or two about how people learn.

The Learning Sciences Today

Learning sciences as a field is housed at places like Northwestern, Stanford, Vanderbilt, Indiana, and Berkeley, bridging education schools and cognitive science departments. It emerged in the late 1980s and 1990s as researchers tried to connect laboratory findings about cognition to the realities of classrooms.

But the field was young when most of today’s teachers went through preparation programs. How People Learn appeared and then, eighteen years later, How People Learn II validated and extended the report. As research accumulates, we may get a How People Learn III, deepening and broadening the framework that by then could bring coherence to teacher preparation across the country.

Learning sciences is not yet a household term, and its findings have not yet filtered systematically into teacher preparation. Most American teachers entered classrooms without access to this knowledge base. They took on responsibility for developing young minds without the scientific foundation that we now have. Then they were often blamed when outcomes disappointed, measured by standardized tests that often reflect the very transmission model of teaching the most current research discredits.

The knowledge exists now. The question is whether we’ll use it and whether we’ll rebuild teacher preparation around what we’ve learned. After watching the rollout of LLMs in schools over the past three years, I conclude that professional development of teachers in service now has an urgent need to ground itself in the core principles of How People Learn.

The Common Sense Teacher

The good teacher today plans lessons anchored in a curriculum chart, performs classroom activities based in provided instructional materials, explains topics and concepts relying on subject matter preparation, past experience, and teacher’ manuals, manages according to institutional policies, and cares about student attendance and compliance.

This vision of teaching exposes the transmission model—knowledge flows from the teacher to the students—combined with crowd control and personal warmth. Content expertise matters above all; if you really understand calculus or American history, you can teach it. The rest is common sense and on-the-job learning. Some people are natural teachers. Others aren’t. Extended professional preparation is sometimes viewed as the credentialism that keeps talented people out of classrooms and protects mediocre education schools.

This view explains how Teach for America could build a multi-hundred-million-dollar organization on the premise that elite college graduates with five weeks of summer training could transform struggling schools. It’s why politicians routinely dismiss teacher preparation as an obstacle rather than a foundation. It’s why parents who would never accept a semi-trained surgeon operating on their child see nothing wrong with a semi-trained teacher shaping their child’s mind for a year.

Nothing in this common sense picture is exactly wrong. Clear explanation matters. Patience matters. Enthusiasm matters. Getting students to attend class on time and do their work matters. The problem is what’s missing and what assumptions the thinking rests on.

Asking questions and calling on raised hands means the teacher hears from students who already feel confident, who already know. The quiet ones—confused, uncertain, or processing differently—stay invisible. Call and response is a performance of knowing, not a diagnosis of understanding.

Writing on the board is fine as one tool among many. But it assumes learning happens through presentation, that the teacher’s job is to display knowledge clearly enough for students to absorb it.

Making lessons interesting is admirable, but “interesting” often means entertaining, hooking attention. That’s different from understanding why this particular content isn’t connecting with this particular child’s prior knowledge and experience.

Staying late to help a struggling student is genuinely caring. But it’s remediation after the fact. It assumes the lesson was fine and this kid just needs more of it—a slower re-explanation—rather than asking whether the instruction itself failed to connect with how this student learns.

The problem? All these behaviors are teacher-centered. They describe what the teacher does, not what the teacher understands about the learners. A teacher could do every one of them—explain clearly, engage students, stay late—and still have no idea what misconceptions a student holds, how that student’s home learning practices differ from school expectations, or what prior knowledge could scaffold new understanding.

The common sense view sees good teaching as experts in subject matter performing according to adopted procedures plus caring. The learning sciences sees it as assessment, adaptation, and connection-making.

From Speculation to Science

The title of Chapter 1 in the National Research Council’s landmark 2000 report How People Learn says it all: “Learning: From Speculation to Science.” The report opens with a striking claim: “The world is in the midst of an extraordinary outpouring of scientific work on the mind and brain, on the processes of thinking and learning, on the neural processes that occur during thought and learning, and on the development of competence.” A new theory of learning was coming into focus in 2000—one that “leads to very different approaches to the design of curriculum, teaching, and assessment than those often found in schools today.”

Three findings anchored the report, each with a solid research base and strong implications for teaching.

Preconceptions

First, students come to the classroom with preconceptions about how the world works. If their initial understanding is not surfaced, they may fail to grasp new concepts and information, or they may learn them for purposes of a test but revert to their preconceptions outside the classroom. If teachers spent more time activating learners’ prior knowledge and exploring their preconceptions and less time on presenting new information, classroom activity would require deep engagement with what learners think they know before they work on unlearning their misconceptions and learning more defensible ideas and connections .

A children’s book called Fish Is Fish, mentioned in the report, captures the problem precisely. A fish, curious about land, befriends a frog who eventually explores the world above water. When the frog returns to describe birds, cows, and people, the fish constructs mental images based on what he already knows. People become fish walking on their tailfins. Birds become fish with wings. The fish isn’t stupid; he’s doing exactly what learners do. New information gets filtered through existing mental models, and what emerges may bear little resemblance to what the teacher intended.

Consider attempts to teach children that the earth is round. Children who believe it’s flat picture a pancake. When told it’s round like a sphere, they imagine a flat surface sitting atop a sphere, with people standing on the pancake. Their mental model—which explains how they can walk on the earth’s surface—doesn’t accommodate spheres, so they assimilate new information into what they already believe.

The persistence of preconceptions is remarkable. Physics students from elite technical colleges, asked to play a game requiring Newton’s laws of motion, performed no better than elementary schoolchildren—both aimed directly at moving targets, failing to account for momentum. When researchers investigated one student who failed, they found she knew the relevant formulas perfectly well. But in context, she fell back on untrained intuitions about how the physical world works.

The phenomenon appears across subjects. Students persist in believing seasons result from the earth’s distance from the sun. Biology students—even those who have studied photosynthesis—continue believing soil is plants’ food. Children resist accepting that one-fourth is greater than one-eighth, because eight is bigger than four.

Simply telling students the correct answer leaves underlying mental models intact. They learn the right words for the test, then revert to intuitions everywhere else. Effective teaching requires surfacing what students already believe, designing experiences that reveal where those beliefs fall short, and building bridges from naive conceptions toward expert understanding. This is diagnostic work—and most teachers were never prepared for it.

Conditionalized Knowledge

Second, deep understanding requires that knowledge be organized around important concepts, connected, and “conditionalized” to specify when it applies. Usable knowledge is not the same as a list of disconnected facts. Experts don’t just know more than novices—their knowledge is structured differently.

Consider chess. When researchers showed chess masters and novices a mid-game board position for five seconds, then asked them to reconstruct it, masters placed sixteen pieces correctly while novices managed four. But when the pieces were arranged randomly—not conforming to any real game—masters performed no better than novices. The masters weren’t exhibiting superior memory; they were recognizing meaningful patterns. Their knowledge was organized around strategic configurations, allowing them to “chunk” multiple pieces into single meaningful units.

The same pattern appears in physics. When asked to sort problems, novices group by surface features: “inclined plane problems,” “pulley problems,” “spring problems.” Experts sort by underlying principles: “conservation of energy problems,” “Newton’s second law problems.” The novice sees an inclined plane; the expert sees a vector problem requiring force decomposition. Both are looking at the same problem, but they’re seeing different things entirely.

This is what “conditionalized” knowledge means. Expert knowledge isn’t just more extensive; it’s tagged with information about when and where it applies. A physics student might memorize Newton’s second law but not recognize when to use it. The expert’s knowledge fires automatically when the right conditions appear. Knowledge that lacks these conditional triggers remains “inert”—present but inaccessible when needed.

The implications for teaching are profound. Curricula that emphasize breadth over depth—what the Third International Mathematics and Science Survey called “a mile wide and an inch deep”—may actually prevent effective knowledge organization. There isn’t enough time to develop the deep, connected understanding that distinguishes expertise. And textbooks that present facts and formulas without attention to conditions of applicability leave students unable to retrieve what they’ve learned when they need it.

Metacognition

Third, metacognition matters. Students need to learn how to learn when learning is challenging. Learning isn’t a flash card affair. Learners must learn to monitor their own understanding, not simply their ability to recall details; to recognize when they’re confused; and to deploy strategies for making sense of difficult material.

The research is specific about what metacognition involves. It includes the ability to predict performance on tasks and adjust accordingly, to monitor current levels of understanding and repair misunderstanding, to recognize when additional information is needed in the case of non-understanding, and to assess whether new information is consistent with what one already knows. Experts do this automatically. When asked to think aloud while working, they continuously note gaps in their understanding and draw analogies to advance their thinking. But children don’t develop this capacity on their own.

Even when a task calls on a primitive memory for facts, metacognition is a natural reaction which can be further developed throughout formal schooling. Consider three- and four-year-olds asked to remember which of three cups hides a toy dog, a challenging task. Left alone during a delay, the children weren’t passive. Some looked at the target cup and nodded yes, looked at non-target cups and nodded no. Others rested a hand on the correct cup or moved it to a more visible position. These primitive strategies—precursors to mature rehearsal—worked. Children who prepared actively remembered better. Even preschoolers show dawning awareness that without strategic effort to connect information, forgetting occurs.

But sophisticated metacognition must be taught. Research demonstrates that children can learn to predict outcomes, explain to themselves in order to improve understanding, note failures to comprehend, activate background knowledge, plan ahead, and apportion time and memory. Reciprocal teaching, for instance, improves reading comprehension by helping students explicate, elaborate, and monitor their understanding as they read.

Reciprocal teaching, developed by Palincsar and Brown in 1984, illustrates what happens when metacognition is made explicit. The technique improves reading comprehension by teaching students four strategies that skilled readers use automatically: summarizing what they’ve read, generating questions about it, clarifying confusing passages, and predicting what comes next. As an aside, I see obvious relevance to the work of teachers looking for entry points for LLMs.

The teacher initially models these strategies aloud—making visible the internal dialogue that ordinarily remains hidden—then gradually transfers responsibility to students, who take turns leading discussions while the teacher fades into a coaching role.

The results have been replicated across multiple randomized controlled trials, earning positive ratings from the What Works Clearinghouse. Students don’t just comprehend better; they learn how to comprehend, a transferable capacity they carry into new texts and subjects.

The implications for teacher preparation are direct. If students benefit from having metacognitive strategies made explicit, teachers need to understand what those strategies are and how to model them. But modeling requires more than performance. It requires knowing why summarizing helps consolidation, why self-questioning reveals gaps in understanding, why prediction activates prior knowledge. Teachers need the science behind the technique, and most were never taught it.

Each finding demolishes a piece of the transmission model. If students arrive with preconceptions that filter everything they hear, then clear explanation isn’t enough; teachers must diagnose and engage existing beliefs. If knowledge must be organized and connected rather than accumulated piece by piece, then covering content isn’t the same as teaching it. If learning requires metacognitive awareness, then students need to develop capacities for self-monitoring.

Context and Culture

Two decades after How People Learn, the National Academy of Sciences released How People Learn II (2018), summarizing what had changed in the science. The central update: Culture moved from peripheral to central.

The original report had acknowledged culture as a variable educators should “take into account”—a factor affecting some learners in some contexts. The 2018 report reconceptualized it: Culture is constitutive of all learning for all people: “Learning does not happen in the same way for all people because cultural influences pervade development from the beginning of life.”

Every child arrives at school having already learned how to learn through participation in the particular practices of their family and community. A Mayan child develops keen observational learning through community life. An Oksapmin child in Papua New Guinea internalizes mathematical thinking through body-counting practices embedded in daily activity. An American child learns to expect individualized verbal instruction and immediate feedback.

These aren’t deficits or advantages. They’re different cultural tools for making sense of the world. Children don’t reinvent these tools; they inherit them across generations. The sociocultural tradition pioneered by Vygotsky established that cognitive growth happens through social interactions where children and more expert others work jointly to solve problems, learning to use their culture’s psychological and technical tools: number systems, writing systems, ways of reasoning about the natural world.

Integration as an Instructional Strategy

The classroom problem becomes clear: School itself is a cultural institution with embedded expectations about how language should be used, whether learning happens through observation or direct instruction, whether intelligence means book knowledge or socially responsible behavior. When these expectations align with home learning, school feels natural. When they clash, children struggle due to a cultural mismatch.

Research documented in How People Learn shows this mismatch in action, consonant with Flores and Rosa’s (2015) formulation of the White listening subject. Schools privilege certain discourse patterns. White middle-class children’s storytelling styles tend to be linear and chronological with a beginning, middle, and end. Teachers recognize this structure as coherent because it matches school expectations for how stories should be told.

Black children often use topic-associative styles—episodic, with implicit thematic connections rather than explicit linear progression. This cultural difference is neither better nor worse, but it is structured according to different conventions. Teachers unfamiliar with this style often misread it as disorganized or unfocused when it’s actually highly coherent by its own logic. The science says recognize these as different cultural strengths and build bridges. The non-scientist hears deficits.

Integration as a society-wide strategy instigated in 1954 for racial desegregation offered a partial solution to this mismatch, not by erasing cultural differences, but by creating conditions where they had to be navigated. The National Assessment of Educational Progress documented the results. During the 1970s and 1980s, as American schools reached their peak of racial integration, the achievement gap between Black and White students narrowed dramatically. Black students who attended integrated schools for all twelve years showed the largest gains.

The science of learning explains why. Integrated classrooms fostered belonging for students who had been systematically excluded. They reduced stereotype threat by normalizing Black students’ presence in rigorous academic settings. They diversified peer learning environments. And they forced at least some teachers to recognize that children who spoke and reasoned differently weren’t deficient, just different.

Then came resegregation. Since the late 1980s, as court oversight ended and schools re-separated by race and class, progress stalled. The NAEP data flatlined. The science predicts exactly this: Remove the conditions that support learning, and learning suffers.

Why Integration Worked

Belonging and motivation. HPL II (2018) emphasizes that motivation is fostered when learners perceive the learning environment as a place where they belong and where their sense of agency and purpose is promoted. Segregated schools—underfunded, stigmatized, marked as “other”—communicate the opposite. Integration, when done well, signals that all children belong in the same educational spaces. That signal affects how hard students work, how long they persist, and whether they see themselves as learners.

Reduced stereotype threat. Researchers document that marginalized students experience social identity threats through receiving messages that students holding certain identities are less capable or worthy. These threats induce stress and anxiety that undermine performance. Integrated classrooms, particularly when teachers hold high expectations for all students, can counteract stereotype threat by normalizing the presence and success of Black students in rigorous academic settings.

The sociocultural mechanism. Vygotsky’s framework, central to HPL II, holds that cognitive growth happens through social interactions. In segregated, high-poverty schools, students have fewer opportunities to interact with peers who have had access to enriched early learning environments, travel, books, and other cognitive resources. Integration diversifies the peer learning environment.

Access to knowledge-centered environments. HPL emphasizes that effective learning environments must be organized around deep understanding, not just fact accumulation. Segregated schools serving high-poverty populations of all races, ethnicities, and linguistic backgrounds often emphasize basic skills and test prep, while integrated schools are more likely to offer curricula that develop conceptual understanding.

Breaking the mismatch cycle. The science shows that school failure often results from mismatch between home learning practices and school expectations. In all-White schools, those expectations were designed around White, middle-class norms and no one noticed because all the students shared them. Integration forced at least some attention to the fact that children learn differently, creating pressure, however inadequate, to bridge home and school.

The sobering implication: resegregation reverses all of these mechanisms. When schools re-separate by race and class, belonging erodes, stereotype threat intensifies, peer learning narrows, teacher expectations calcify, curricula thin out, and cultural mismatch goes unaddressed. The science of learning predicts exactly what the NAEP data now show—stagnation and widening gaps since the late 1980s, precisely when desegregation efforts collapsed.

Integration worked not just because of funding or facilities, but because it changed the social, cultural, and motivational contexts in which learning happens. The science explains why. And the science explains why dismantling integration has cost children so much.

The LLM Disruption

The transmission model contains a hidden assumption: If students can produce correct, expected outputs, learning has occurred. The entire apparatus of schooling—lectures, textbooks, assignments, tests—is designed to generate and measure those outputs.

But production isn’t understanding. A student who writes “the earth is spherical” on a test may still picture a pancake with people standing on top. A student who correctly applies Newton’s second law on a problem set may still aim directly at a moving target when playing a video game. The HPL research documented that students who perform correctly on transmission-model school tasks reverted to naive conceptions the moment context changed.

The LLM exposes this gap by performing better than students ever could. It produces essays that hit every rubric criterion without understanding a single sentence. It solves physics problems without knowing what force means. It generates the five-paragraph response, the document-based question answer, the lab report conclusion—all the genres students learned to mimic.

And in doing so, it reveals that school was training mimicry all along. The student who learned to produce what teachers expected was doing exactly what the LLM does, i.e., pattern-matching to an anticipated output. The machine just matches faster.

This is why transmission-model teachers feel existential threat, not just irritation and inconvenience. The LLM makes visible that the work those teachers assigned didn’t require understanding in the first place—how could it have if a machine can do it—and that they had no way of knowing whether deep learning and understanding occurred. Their assessments measured performance. Their instruction optimized for performance. The LLM performs.

Teachers whose practice was already built around surfacing student thinking rather than privileging teacher thinking, diagnosing misconceptions, and creating cognitive challenge tend to have a different perspective on LLMs. They recognize risks—students outsourcing confusion instead of working through it, developing their metacognition. But the LLM doesn’t undermine their core work because their core work was never about producing correct outputs. It was about how learners are thinking while making knowledge. And knowledge is precisely what LLMs don’t have.

What Teachers Actually Need to Know

For more than three decades, Linda Darling-Hammond, our most prominent thought leader in reforming the institution of the school, has advanced a simple but radical proposition: The quality of teaching is the single most important school-based factor determining whether children learn. And the quality of teaching depends, first, on how teachers are prepared, and second, on how they develop from experience across their careers.

This argument cuts against the grain of American education. Politicians and philanthropists chase silver bullets—standardized testing, charter schools, merit pay, technology—while treating teacher preparation as an afterthought. Darling-Hammond’s response has been to insist that teaching is complex intellectual work deserving the same rigorous preparation we expect of doctors.

To professionalize teaching means to recognize that effective educators need deep knowledge of how children develop and learn in sociocultural contexts. Teachers need skills in curriculum design, instruction, inquiry, and reflection that produce what Darling-Hammond calls “adaptive expertise” or the capacity to make connections between children and content that enable children to transform misconceptions into accurate, reliable, well-organized, conditionalized knowledge. And teachers need dispositions supporting empathy, social-emotional capacity, cultural competence, and commitment to equity.

This preparation means professionals who respond to the specific children before them in all their complexity and possibility. Teachers need to understand how children’s prior knowledge, experiences, cognitive strategies, and motivation all shape what they can learn. They need to recognize that children actively make sense of their worlds, reason facilely with what they know, and persist because understanding is motivating. And they need skill in building bridges between home and school.

We don’t produce this kind of teacher through five-week summer institutes. We don’t produce this kind of teacher by assuming smart college graduates can figure it out on the job. We produce this kind of teacher through comprehensive professional preparation grounded in the learning sciences, preparation that integrates theory and practice, that provides extensive clinical experience with expert mentorship, that treats teaching as the complex intellectual work it actually is.

The Stakes

The science of learning tells us that teachers must understand each child’s prior knowledge, cultural tools, and ways of making meaning. This is the central challenge. The research shows that teachers must recognize children’s diverse oral styles, their home language practices, their culturally-rooted ways of reasoning, and help them connect meaningfully to academic content. The science says build on strengths.

School segregation levels are not quite at pre-Brown levels, but they are high and have been rising steadily since the late 1980s. In large school districts, White-Black segregation between schools increased 35 percent from 1991 to 2020. Segregation between poor and non-poor students increased by 47 percent over three decades. Black students remain considerably isolated in schools with higher poverty rates, in schools with less experienced teachers, higher turnover, fewer resources.

So we have a science of learning that demands teachers understand culture, context, and the child’s existing sense-making frameworks. We have a teaching force that largely lacks this preparation. We have a society where racialized listening practices systematically misperceive minoritized children’s linguistic and cognitive competence. We have schools more segregated by race and poverty than they were thirty years ago.

And we have LLMs.

The children most in need of teachers grounded in the science of learning, that is to say, teachers who can bridge home and school, recognize cultural strengths, diagnose actual understanding, are the children least likely to have such teachers.

The Election Ahead

The 2026 midterms will determine whether American education policy moves toward or away from the science of learning. The conservative agenda is clear: Vouchers and school choice that further stratify children by race and class, continued gutting of the federal Department of Education, elimination of civil rights enforcement that protects marginalized students. This is education policy designed by people who don’t know—or don’t care—how children actually learn.

But there is another path. We don’t let climate policy be dictated by people who reject atmospheric science. We don’t staff the CDC with people who dismiss epidemiology. We don’t hand economic policy to people who’ve never studied how markets function. The federal government maintains robust scientific infrastructure for agriculture, weather, public health, land management, and labor relations because we understand that governance requires expertise.

Why should children’s learning be different?

What voters need is to understand is that the science of learning is as real as the science of immunology or hydrology. It tells us things that common sense doesn’t know. It has accumulated over decades through rigorous research. And it has direct implications for how we prepare teachers, design schools, and allocate resources.

It deserves a featured spot in the Democratic party platform. No Race to the Top. No false accountability. To earn a place in the federal government, candidates must become experts in what expert teaching is.

A Department of Education led by a CEO who ran the World Wrestling Federation is not equipped to translate learning science into national policy. We need a robust federal center for education, not a hollowed-out bureaucracy awaiting abolition. We need a federal center working in genuine collaboration with state departments of education to ensure that every child, in every classroom, has access to teachers who understand how human beings learn.

The science exists. The question this November is whether we’ll elect leaders who are capable of turning the corner on the transmission model at the heart of the factory. Having American schools just as effective as those in, say, Finland are within reach—from preschool to the university.

No matter how long it might take, what we’ve seen from this painful blast of conservative regression beginning with No Child Left Behind, continuing through the Common Core, has balkanized public schooling. In 2026 voters can do something about it. Democratic politicians must put public school at the center of a new era, and they will if enough voters demand it.