How People Learn, by the National Research Council
My detailed notes on an interesting primer on learning theory.
Every now and then I enjoy dusting off one of the books which started me down the path of summarizing. I devoured How People Learn back when I was helping an education technology company incorporate external research into their core learning model. Yes, it’s an academic book, but also one that has surprisingly relevant insights into the work we do.
For example, it’s not merely sufficient to be able to query for information in the moment you may need it. We need to have some basic threshold of memorized knowledge about the world to even know it’s possible to dig deeper.
How People Learn was put together by a research council in part to summarize the state of learning research in the year 2000 and to frame some of the more promising research areas for future study. As such, it draws heavily on examples from explicit learning environments — schools — as well as perspectives on K-12 education. Still, many of the concepts are intriguing to consider as we probe to uncover misunderstandings or discover the truth about any topic.
I’d say this summary pairs well with Thinking in Bets by Annie Duke, who teaches us how to engage in truthseeking regarding our beliefs about the world. It also pairs well with Surpassing Ourselves, which digs deeper into the key differences between people who actually develop expertise through experience versus those who plateau early and become “experienced non-experts.”
If you find the summary below interesting, I highly recommend reading the full study. You can buy it from Amazon or Bookshop.
PART I: INTRODUCTION
Chapter 1: Learning: From Speculation to Science
The science of learning has been heavily informed by cognitive science, which has come from our increasingly nuanced view of internal mental processes. This emerged from behaviorism, which considered only observable behaviors and the stimuli that controlled them. We’ve since learned:
Students come to the classroom with sophisticated preconceptions about how the world works. Starting from their initial understanding, children will move through normal paths of misunderstanding before accurate learning takes root. For example, children may imagine a “round” earth as a round pancake rather than a sphere. Formative assessments help bring preconceptions to the forefront.
Competence requires: a) have a deep foundation of factual knowledge; b) the context of a conceptual framework; and c) sufficient structure for easy retrieval and application. We cannot expect Google searches to replace fluid factual knowledge.
Students can take control of their own learning through explicit metacognitive strategies such as developing forms of internal dialogue (planning ahead, predicting outcomes, or explaining things to yourself.) [Definition: metacognition is defining your own learning goals and monitoring your progress in achieving them.]
Knowing how people learn provides us with a central starting point to organize environments and develop the best teaching practices. Successful learning environments must be::
Learner centered: accommodating cultural differences, students’ theories of fixed vs. growth mindsets, and students’ knowledge, skill levels, and interests to present challenges at the “just manageable” difficulty level.
Knowledge-centered: paying attention to what is taught (information, subject matter), why it is taught (understanding), and what competence or mastery looks like.
Informed by (learner friendly) formative assessments: to identify preconceptions, design instruction, and monitor students’ progress towards mastery.
Community-centered: with both a vibrant community within the classroom and school, and connected to the outside world.
PART II: LEARNERS AND LEARNING
Chapter 2: How Experts Differ from Novices
Experts are people who have successfully learned something to a very deep level. In general:
Experts more readily notice features and meaningful patterns of information (provided they are meaningful or realistic patterns). E.g. Expert teachers notice far more meaningful patterns in classroom observations, whereas novices tend to make superficial observations.
Experts have large quantities of content knowledge organized around the big ideas or core concepts of their field (as supports for understanding the facts). E.g. expert physicists tend to organize their knowledge around the core principles of mechanics whereas novices organize their knowledge around superficial elements of problems.
Experts have knowledge that reflects the various contexts in which it is applicable. For example, expert knowledge of physics tend to extend to the real-life circumstances in which forces are at work, whereas novice knowledge may be limited to the solving of mathematical equations and fail to transfer.
Experts can flexibly retrieve important aspects of their knowledge without attentional effort (although, interestingly, they tend to dwell longer with the problem before moving on to seeking the appropriate solutions).
Despite these strengths, experts are not necessarily good at teaching others. This is partly because they often lack the appropriate pedagogical content knowledge for effective teaching, specifically of the preconceptions they’ve long since struggled through. Experts also vary in the flexibility within novel situations (i.e. their process of expert learning may not transfer to other situations).
Chapter 3: Learning and Transfer
All learning involves some kind of transfer, we build on our prior knowledge as we tackle new topics. The following are factors that facilitate transfer of learning into other dimensions:
Mastery of the original subject (which provides the rich conceptual metaphors/analogies to facilitate transfer)
Learning with understanding rather than memorizing a set of facts
Having had sufficient time to learn the original subject (and struggle through issues of comprehension)
Engaging in deliberate practice, including the use of contrasting cases to build awareness of the various dimensions of what you’re learning
Motivation to learn (mastery motivation is better than performance motivation, and social motivations are also strong factors)
We can facilitate transfer by broadening the context in which understanding is created (create lots of “what if” scenarios!) We can also facilitate transfer through specific metacognitive strategies that facilitate independent transfer (e.g. asking “How does this relate to previous things I’ve seen?”).
Remember: transfer manifests not only in what you understand today, but also in the facility and speed with which you learn a new subject tomorrow. Developing a deep understanding of calculus will heavily influence how quickly you learn the concepts of physics.
Discrepancies between real-world experience and school learning also exacerbate issues of transfer. For example, a) most work is done collaboratively in the real-world, while it is independent within schools; b) most people rely heavily on tools within the real world, whereas we use ‘mental tools’ in school; and c) most people rely on contextualized reasoning in real life whereas we rely on abstract reasoning in school. There are promising approaches to solve these issues (e.g. problem-based learning).
Chapter 4: How Children Learn
Children are not tabula rasa. All children are active learners, with existing knowledge, who are able to set goals, plan, and revise. We recognize Vygotsky’s “Zone of Proximal Development” as the “bandwidth of competence that learners can navigate with support.” Aided by new methods for studying infants, we have improved our understanding of four major areas:
Early Predisposition: the privileged domains within which we are predisposed to learn (even as infants):
Physical concepts: understanding (physically) impossible events [example: box floating in mid air], early forms of causality [example: learning how to influence an inanimate object via a point of contact/ support, at 9 months].
Biological causality: distinguishing between animate (able to move itself) and inanimate (only moving with external force)
(Early) Number Concepts: the number of things, and having a sense when things are added or taken away.
(Early) Attention to Language: attending to words more than sounds, contextualizing the sounds of language [example: a request made while sitting in a high chair is interpreted as an invitation/request to eat eben before the word itself is explicitly understood]
Strategies and Metacognition: outside of the privileged domains, learners depend on will, ingenuity, and effort to enhance learning. All three increase with age.
We learn strategies to work more effectively with our limited memory. These include: rehearsal, elaboration, summarization, clustering, arithmetic, casual and scientific reasoning, spatial reasoning, referential communications, recall from memory, reading and spelling, and judgments of plausibility.
Instructional strategies have developed based on research on “strategies” based construction of understanding: e.g. reciprocal teaching, communities of learners, the ideal student, and Project Rightstart.
Theories of Mind: "children develop theories of what it means to learn and understand that profoundly influence how they situate themselves in settings that demand effortful and intentional learning."
Example of Dweck’s research on “fixed” vs. “growth” mindset (represented here as “entity theories” vs. “incremental theories” of intelligence).
Children are often motivated by pure satisfaction: they are problem solvers and problem generators. They can be achievement- or competence-motivated.
Children and Community: other people, settings, and tools play major roles in fostering the development of learning in children. In particular, parents are the first teachers.
Acquisition of language starts between children and caregivers working to accomplish everyday goals.
Scaffolding takes many forms: a) focusing attention (interest, sustained enthusiasm) b) regulating difficulty (reduce the number of steps, controlling frustration and risk); c) modeling the goal (highlighting discrepancies between current and target performance, demonstration).
Support also involves tweaking the environment or availability of materials, e.g. handing a child something (nesting cups) and saying “These are for you to play with”
Example: Lewis Carroll’s notes on how to “read” a picture book to a child … “if you look at this picture you’ll soon know what happens next…” Concentrate their attention on the picture, prod their curiosity by asking questions, engage the child in dialogue. Talk about the relations between objects in the picture.
It’s also important to consider cultural differences in communication (conversing vs. observing vs. eavesdropping) and in how questions are asked ("known answer" vs. questions used to initiate storytelling)
Chapter 5: Mind and Brain
The convergence of developments within research fields touching on the brain (from developmental psychology, cognitive psychology, neuroscience, and more) has clarified some of the neural mechanisms of learning. Specifically, learning changes the physical structure of the brain, and these changes extend to even the functional organization of the brain. Evidence of this comes from:
Studies of rats in enriched versus deprived environments, where they observed differences in astrocytes (cells that support neuron functioning by providing nutrients and removing waste) & comparative weight and thickness of the cerebral cortex.
Studies of acrobatic vs. exercising rats, where similar activity levels with the added element of learning adds synapses which exercise alone does not.
[The book continues with a brief overview of neuroscience, the nature of memory, and so on. Skipped for the purpose of this summary. These are useful elements to remember when designing educational experiences.]
PART III: TEACHERS AND TEACHING
Chapter 6: The Design of Learning Environments
“Physics constrains but does not dictate how to build a bridge.” The same is with learning theory and learning environments. It’s also important to note that our goals of learning and even working definitions of literacy have changed significantly over time (e.g. from being able to sign your name -> taking notation of oral messages -> memorizing familiar texts -> extracting useful information from novel texts). To meet today’s goals, environments should be:
Learner centered: accommodate the child’s culture as well as knowledge, skill levels, and interests to present challenges at the right level of accessible difficulty. This can be accomplished through “Diagnostic Teaching” (adapting to a child’s current knowledge through observation, questioning and conversation).
Knowledge-centered: pay attention to what is taught (information, subject matter), why it is taught (understanding), and what competence or mastery looks like. This can be accomplished by emphasizing sense-making (where you expect new information to make sense and ask for clarification when it doesn’t). And thorough progressive formalization (transforming initial informal knowledge into more formalized structures). Recognize that younger children can understand complex ideas if presented in an accessible way.
Assessment-centered: provide opportunities for meaningful feedback and revision, mainly through formative assessments with respect to one’s learning goals. Focus on making student thinking visible, and providing feedback in ways that allow the learner (or teacher) to adjust their strategy.
Example: presenting two different problems and asking if they would be solved with a similar approach (and why), or discussing & defending your work in a portfolio.
Theoretical framework for assessment (from Baxter and Glaser, 1997): Does the task present: a) “Lean -> Rich” demands for content knowledge outside of the assessment; and b) “Constrained -> Open” demands for process skills outside of the directions provided by the assessment.
Community-centered: cultivate social norms in the environment (classroom, school, family, or broader community) to support learning, and be aware of those that suppress learning. For example, American classrooms have an emphasis on being right and contributing by talking; some Japanese classrooms have developed a culture where students “are skilled at learning from one another, see analyzing errors as fruitful for learning, and value listening (and so learn from class discussion even if they don’t participate)."
Television plays an important role in child development and learning. The value of educational programs is most pronounced for younger children (even when taking various other factors into account). TV shows have a strong impact on beliefs and attitudes, either reinforcing stereotypes or helping to break them down.
Chapter 7: Effective Teaching: Examples in History, Mathematics, and Science
Pedagogical content knowledge is different from regular content knowledge. It is a popular - and dangerous - myth that teaching is a generic skill and a good teacher can teach any subject. Instead, teachers must develop the ability “to understand in a pedagogically reflective way; they must not only know their own way around a discipline, but must know the conceptual barriers likely to hinder others.” Some inspiring examples:
Barb Johnson, sixth-grade teacher at Monroe Middle School, starts the year by asking “What questions do you have about yourself?” and “What questions do you have about the world?” Students plan (individually and in groups) how they’re going to tackle the most interesting ones. Throughout the year, they see how their investigations relate to conventional subject-matter areas like language, mathematics, science, social studies, history, music, and art.
Bob Bain, ninth-grade history teacher in Beechwood, Ohio, emphasizes the central issue of history as separating what is important from what is peripheral (particularly when considering historical evidence), and has students create a time capsule of what they think are the most important artifacts of the past and justifying their choices.
Elizabeth Jensen, eleventh-grade history teacher, emphasizes debates from turning points in history to help students understand various perspectives, and grounds students’ examinations in the relevant philosophical context.
Magdalene Lampert, fourth-grade math teacher, emphasizes mathematics as "a culture of sense-making (based on mathematical principles)” and works hard to find example problems which students with varying levels of computational skill can still tackle successfully to learn the concepts.
Deborah Ball, third-grade math teacher, emphasizes developing a mathematical culture in which students conjecture, experiment, build arguments, and frame and solve problems (which she sees as the work of mathematicians).
Annie Keith, first- and second-grade teacher in Madison, WI), relies heavily on word problems, and has students discuss strategies with one another, in groups, and as a class, framed around concrete problems in the context of everyday first- and second-grade activities: sharing snacks, lunch count, attendance.
Minstrell (high school physics teacher) emphasizes classroom discussions in which students construct understanding by making sense of physics concepts with Minstrell participating as a coach and facilitator. Students share ideas (e.g. of the forces on a rock as it falls), and he asks “How do you know? How did you decide? Why do you believe that?”
Exemplary techniques in different subject areas lean into encouraging students to do legitimate work of the field of study. For example:
The process of doing history involves “sorting out competing claims and formulating reasoned interpretations between historical documents,” which is much more valuable than learning facts.
The process of writing involves enriching and improving our communication of ideas, which is more important overall than the remediation of factual/ stylistic elements.
The process of doing math involves computation as a tool of mathematics (along with problem solving or understanding structure and patterns), not its purpose.
Modeling of phenomena is a valuable way to apply math to examine questions within other disciplines. For example: the geometry of triangles “has an internal logic and predictive power for phenomena ranging from optics to wayfinding to laying floor tile.”
The process of doing science involves the dialogue-driven means for scientific sense making where you structure an appropriate hypothesis, see evidence as more than information you already know (e.g. data produced through observation or experimentation); and then debate conclusions that they derive from their evidence.
The process of doing science also involves a) describing a problem in detail before attempting to solve it; b) identifying the relevant information to solve it; and c) deciding what procedures to use. These methods are tacitly used by experts yet rarely by students.
Chapter 8: Teacher Learning
Practicing teachers learn from their own experience through monitoring and adjustment, which can be facilitated by “action research.” They also learn through interactions with other teachers (mentoring or informal inservice education) or from teacher educators (who are often tied to innovations in curriculum or pedagogy).
Teacher training programs are often of variable quality, though. The better programs apply the same principles identified above, and are:
Learner-centered: teachers with different levels of experience have vastly different needs
Knowledge-centered: pedagogical content knowledge is frequently ignored in favor of generic pedagogy like cooperative learning
Assessment-centered: the gap between theory and practice in education is often based on the conditionalized nature of expert knowledge: researchers can cite relevant theory and yet have challenges implementing their ideas in the vastly different environment of a classroom
Community-centered: studies of teacher collaboration demonstrate: a) the importance of shared experiences and discourse around texts and data about student learning; and b) a necessity for shared decisions.
Action research is an important way to set the stage for understanding the implications of new theories of how people learn. This is a promising domain of expertise, yet underfunded, and rife with disconnects between practitioner research and academic research.
Preservice education will play a big role in improving education in the future. Though it’s important to be aware of the four models that have dominated teacher education in the 20th century (most programs are a blend of several of these):
Academic tradition: emphasizes teachers’ knowledge of subject matter and their ability to transform that subject matter to promote student understanding
Social efficiency tradition: emphasizes teachers’ abilities to apply thoughtfully a “knowledge base” about teaching that has been generated through research on teaching
Developmentalist tradition: emphasizes teachers’ abilities to base their instruction on their direct knowledge of their students — their mental readiness for particular activities
Social reconstructionist tradition: emphasizes teachers’ abilities to analyze social contexts in terms of their contribution to greater equality, justice, and elevation of the human condition in schooling and society.
Chapter 9: Technology to Support Learning
Technology is usually wasted. When poorly implemented, it can even distract from or hinder learning. Technology must be used properly, particularly within the context of “old” but still useful technologies like books and chalkboards. Proper use of technology can come in a variety of forms:
New Curricula: bringing real-world problems into the classroom for students to explore and solve. Such as multimedia programs which present authentic problems (like banking simulations), or ways we connect students to scientists and enable them to do authentic research and work with real datasets and tools.
Scaffolds and Tools: helping students solve problems effectively. Such as Cognitive Apprenticeship, when the technology supports students working towards independent performance, or when technology can be applied to more easily visualize and organize information.
Feedback, Reflection, and Revision: making it easier for teachers to provide feedback and for students to revise their work. Such as help with diagnosis and formative assessment, simulations for rapid feedback,, and structured peer learning (CSILE project).
Connecting Classrooms to Community: creating semi-permeable boundaries for schools and communities. Such as enhanced communication with parents and linking students to expert networks (as long as the mentors are willing to assume new roles as an educator).
Note that successful communities share three characteristics: a) emphasis on group rather than one-to-one communication; b) well articulated goals or tasks; c) explicit efforts to facilitate group interaction and establish new social norms.
Teacher Learning: Enabling teachers to learn as well. Such as forums, professional development, high quality multimedia resources, and opportunities for classroom observation and feedback.
Note that when students know more about computers that teachers do, it dissolves the typical “epistemological authority” that teachers have in the classroom. Teachers may also get a chance to model learning something new in front of their students.
Ultimately, we must learn more about the influence technology will have over effective learning. “Learning research will need to become the constant companion of software development.”
PART IV: FUTURE DIRECTIONS FOR THE SCIENCE OF LEARNING
Chapter 10: Conclusions
We do not just engage in “accretion of information” but also conceptual reorganization as we learn.
Children learn most easily in the “privileged domains” related to actively making sense of their worlds.
Children are ignorant, but not stupid. They reason with the knowledge they understand.
Children are problem solvers and generate questions and problems through curiosity. They seek novel challenges and persist in solving them.
Their early capacities can be nurtured through adults providing catalysts and mediation.
Transfer of learning needs to be a primary goal of instruction. Successful transfer relies on:
Spending enough time on a problem to support transfer
Deliberate and thoughtful practice (not just ‘grinding through’)
Learning with understanding (not just memorizing facts)
Learning things within a variety of different contexts (i.e. interleaving practice so that performance is not too specifically context-bound)
Learning to extract underlying themes and principles rather than focusing on specifics (e.g. ascertaining the conditions of applicability for knowledge).
Actively engaging in the learning process, and linking our work to prior experiences.
Developing competence and expert performance is another primary goal of learning. Success here relies on:
Relevant structures for organizing knowledge (though this takes more effort early on)
Structured approaches for thinking in sophisticated “problem representations” beyond the information given for a task, which plays a major role in whether people see the same problem as easy, difficult, or impossible.
Regarding teachers and learning, we must first recognize that teachers are also learners, and the earlier principles apply to them. They must also:
Balance teaching the organizing principles of the subject matter with enough facts to aid in-depth understanding of the principles. It’s helpful to show students they already have relevant knowledge.
Develop both content knowledge (of the domain) and pedagogical content knowledge (how their students are likely to misapprehend the principles)
Assess their own effectiveness with their students and develop a model of their own professional development based on lifelong learning.
Regarding learning environments, we must:
Explore a more transformative role of technology in the classroom (see Chapter 9)
Develop the ability to bring effective assessment and feedback into the learning process (reflecting the quality of student thinking more than their recall of information)
Break down the borders between formal school settings and their broader communities and environments, supporting the family as an important learning environment.
Chapter 11: Next Steps for Research
Research rarely has a direct impact on classroom practice. Instead, it works through the mediating arenas of educational materials, pre-service and in-service education, polity, and the public. “Without clear communication of a research-based theory of learning and teaching, the operational theories held by various stakeholders are not aligned.”
We are better off focusing on researching areas where both the usefulness of and the desire for fundamental understanding is high. Some promising ideas (beyond the straightforward ones):
Expand the study of classroom practice to build a “cumulative knowledge base on learning and teaching” between research and practice.
Conduct research on formative assessment to formulate design principles that promote the development of coherent, well-organized knowledge and information presented in a way that promotes and informs growth.
Build a library of model lessons that exemplify the principles of the book, with annotations describing the practices and principles.
Research the key conceptual frameworks for disciplines and the related analytical methods they use to answer the question “How do we know…” in terms of the discipline. Rather than just classifying marine mammals, why not ask, “Why are there mammals in the sea?”
Research the key preconceptions (or misunderstandings) by each field, as well as instructional strategies to address them effectively.
Develop model pedagogical laboratories.
You might be interested in Stanislas Dehaene's work on learning. It's a little reductive but interesting.