Learning Sciences

Learning sciences is the interdisciplinary study of how people learn and how to design better learning environments.

14 frameworks, from 1920 to today · Education · Social Sciences

still a working view ended · each leads to its entry below

Overview

Learning sciences is the interdisciplinary study of how people learn and how to design better learning environments1. It brings together psychology, education, computer science, anthropology, linguistics, and neuroscience, and it is defined less by a single theory than by a commitment: to understand learning as it happens in real settings—classrooms, museums, workplaces, online communities—and to use that understanding to build environments that help people learn1. What is at stake is practical and theoretical at once. If we can explain why some instruction works and most fails, we can design schools, software, and workplaces that do better. The field's history is a story of successive attempts to make learning research both scientifically rigorous and useful in the messy world where learning actually occurs.

From Behaviorism to the Cognitive Revolution

For much of the early twentieth century, the dominant psychology of learning in the United States was behaviorism. Its central claim was that learning could be studied through observable stimulus and response, without appeal to mental states2. B. F. Skinner's operant conditioning—reinforcing desired behaviors—underpinned teaching machines and programmed instruction in the 1950s and 1960s, in which learners worked through small steps and received immediate feedback. This was a genuine advance: it made instruction systematic and testable. But it could not explain what learners did with information, why experts and novices solved the same problem differently, or why some students understood while others merely memorized.

The cognitive revolution of the 1950s and 1960s changed the questions2. Researchers began treating the mind as an information-processing system and asking how knowledge is represented, stored, and retrieved3. George Miller's 1956 paper on the limits of short-term memory, Jerome Bruner's work on how children represent and organize knowledge, and the broader rise of cognitive psychology reframed learning as the construction of mental structures rather than the accumulation of responses. This shift made it possible to ask what understanding actually consists of—and that question became the seed of the learning sciences.

Expertise and the Structure of Understanding

A decisive line of research asked what distinguishes experts from novices. Studies of chess players in the 1960s and 1970s, and later of physicists solving problems, showed that experts did not simply think faster or remember more. They perceived problems differently, chunking information into meaningful patterns and recognizing deep structure where novices saw surface features. Novices solving physics problems often sorted them by what the problem looked like (pulleys, inclined planes); experts sorted by the underlying principle (conservation of energy). This research, associated with Herbert Simon, John Anderson, and others, established that expertise is organized knowledge, not raw capacity—and that instruction should therefore target the structure of understanding, not just the amount of practice.

Cognitive Tutors and Intelligent Tutoring Systems

If expertise is structured knowledge, could a computer model that knowledge and diagnose a learner's misconceptions? This question produced one of the field's most concrete achievements. John Anderson and colleagues at Carnegie Mellon developed cognitive tutors in the 1980s and 1990s, built on his ACT theory of cognition. The tutors modeled the steps a student should take to solve a problem—say, solving equations—and tracked the student's path, offering hints when the student deviated. The approach was called model-tracing: the system compares the learner's actions against an expert model and responds to the gap.

Intelligent tutoring systems (ITS) generalized this idea4. They combine a domain model, a student model, and a pedagogical model to adapt instruction to the individual. Early systems in the 1970s and 1980s, such as those for geometry and programming, demonstrated that one-to-one tutoring could be approximated by software. Evaluations of cognitive tutors in real schools showed measurable gains over conventional instruction, and the approach spread into commercial and school products. The lesson was double-edged: the systems worked, but only where the domain could be modeled precisely, and they tended to teach procedural skill more reliably than deep conceptual understanding.

Constructionism and Learning by Making

A different tradition, rooted in Jean Piaget's constructivism—the view that learners build knowledge through active assimilation and accommodation—took a more open approach5. Seymour Papert, who worked with Piaget in Geneva, argued in the 1980s that learners construct knowledge most powerfully when they construct something shareable: a program, a machine, a project. He called this constructionism6. Papert's LOGO programming language, developed in the late 1960s, let children direct a "turtle" on screen and discover mathematical ideas by making it move. The claim was not that all learning should be discovery, but that making artifacts gives learners a concrete object to think with and to debug. Constructionism shaped the design of programming environments, robotics kits, and later maker spaces, and it established a durable tension in the field between structured tutoring and open-ended construction.

Situated Cognition and Learning in Context

In the late 1980s and early 1990s, a challenge to the cognitive mainstream emerged from anthropology and social psychology. Research on everyday cognition—such as Jean Lave's studies of grocery shoppers doing arithmetic—showed that people who performed calculations fluently in the supermarket often failed on equivalent school problems. The explanation was that learning is situated: knowledge is bound to the activity, context, and culture in which it is acquired7. Abstract, decontextualized instruction, on this view, produces knowledge that does not transfer.

This insight generated influential designs. Cognitive apprenticeship, articulated by Allan Collins, John Seely Brown, and Susan Newman in the late 1980s, proposed teaching by making expert processes visible—modeling, coaching, scaffolding, and gradually fading support—much as a craft apprentice learns from a master. The approach was applied to reading comprehension, writing, and mathematics. Related work on distributed cognition, associated with Edwin Hutchins, argued that thinking is often spread across people, tools, and representations rather than confined to one head; his studies of navigation aboard ships showed how a crew plus instruments constituted a cognitive system. Communities of practice, developed by Lave and Etienne Wenger, described learning as legitimate peripheral participation—newcomers gradually moving from the edge of a community's activity toward full membership. These ideas pushed the field toward studying learning as a social and cultural process, not only an individual one.

Design-Based Research and the Turn to the Classroom

By the 1990s, a methodological problem had become acute. Controlled experiments in laboratories produced clean findings that often failed in classrooms, while classroom innovations were rarely studied rigorously enough to yield general knowledge. Design-based research (DBR) emerged as a response: researchers design an intervention in a real setting, study it iteratively, and refine both the design and the theory together8. The approach was codified in the early 2000s by Ann Brown and Allan Collins, among others, who argued that educational research needed to be conducted in the complexity of actual practice. DBR made the field's central method explicit—design as a form of inquiry—and it tied the learning sciences to the improvement of real instruction rather than to laboratory generalization alone.

Computer-Supported Collaborative Learning

A parallel strand asked what happens when learners work together, especially with technology. Computer-supported collaborative learning (CSCL) grew through the 1990s and 2000s, studying how groups construct shared understanding through discussion, argumentation, and joint problem-solving, and designing tools—shared workspaces, discussion forums, collaborative scripts—to support it. The field drew on sociocultural theory, particularly the idea that learning is mediated by tools and language. CSCL research produced both theory (how groups negotiate meaning) and practical designs (how to structure productive collaboration), and it reinforced the learning sciences' commitment to studying learning as interaction rather than as an isolated mental event.

Complex Systems and Learning Analytics

More recent decades brought two further developments. Complex systems approaches treat learning as an emergent phenomenon in a system of interacting agents—students, teachers, resources, and environments—and use methods from network science and dynamical systems to study how patterns arise. This has been applied to understanding classroom discourse, the spread of ideas, and the dynamics of collaborative groups.

Learning analytics and educational data mining emerged in the 2000s and 2010s as digital learning environments generated unprecedented amounts of data. These fields analyze logs, clickstreams, and assessment records to model learners, predict outcomes, and provide feedback. They descend from the cognitive tutor tradition—which already collected detailed step-level data—but scale it to massive online courses and learning platforms. The promise is personalized support at scale; the open questions concern privacy, the gap between prediction and genuine understanding, and whether data-driven models capture what matters.

The Field Today

The learning sciences today is a recognized interdisciplinary field with its own journals, graduate programs, and professional society, the International Society of the Learning Sciences, founded in 2002. Its main approaches descend directly from the history above: cognitive modeling and intelligent tutoring; constructionist design; situated and sociocultural studies of learning in context; collaborative learning research; design-based research as the field's signature method; and data-intensive analytics. These traditions coexist and often combine—a modern learning environment may embed a cognitive tutor, support collaborative discourse, and generate analytics, all studied through iterative design in a real classroom.

The questions still open are the ones the field was founded to answer. How does conceptual understanding differ from procedural skill, and how can instruction build the former? How does knowledge transfer across contexts, and what designs promote it? How can technology support learning without reducing it to what is easily measured? And how can research conducted in the complexity of real settings produce knowledge that is both rigorous and useful? The learning sciences began by insisting that these questions be asked together, and that insistence remains its defining commitment.

References

  1. Learning sciences (introduction, Domain definition), Wikipedia
  2. Cognitive revolution (Historical background, introduction), Wikipedia
  3. Cognitivism (psychology), Wikipedia
  4. Intelligent tutoring system, Wikipedia
  5. Constructivism (philosophy of education) (introduction, Background), Wikipedia
  6. Constructionism (learning theory), Wikipedia
  7. Situated cognition, Wikipedia
  8. Design-based research, Wikipedia

The frameworks of Learning Sciences

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1920-1970

    Behaviorism

    How do stimuli and reinforcement shape behavior?

    Fly to its tower in Nousfarer

  2. 1956-Present

    Cognitivism

    How is knowledge represented and processed in the mind?

    Fly to its tower in Nousfarer

  3. 1965-Present

    Expertise and Novice Studies

    What distinguishes expert from novice understanding?

    Fly to its tower in Nousfarer

  4. 1968-Present

    Constructionism

    How does making artifacts build knowledge?

    Fly to its tower in Nousfarer

  5. 1970-Present

    Intelligent tutoring system

    Can software adapt instruction to each learner?

    Fly to its tower in Nousfarer

  6. 1985-Present

    Cognitive Tutors

    Can a computer model expert steps and diagnose errors?

    About Cognitive Tutors
    How can a computer tutor adapt to each student's unique thinking, not just their right or wrong answers? That question drove the development of Cognitive Tutors, a framework that built intelligent tutoring systems on a foundation of cognitive psychology. Emerging in the mid-1980s, Cognitive Tutors were among the first learning technologies to model the mental steps a student takes while solving problems—and to use that model to deliver real-time, individualized instruction. Unlike earlier computer-assisted instruction that simply checked answers, Cognitive Tutors aimed to understand the student's cognitive processes, diagnose misconceptions, and guide learning through targeted feedback and practice. This framework sits at the intersection of cognitive science and education, offering a concrete way to put theories of learning into computational practice.
    At the heart of Cognitive Tutors lies a cognitive architecture called ACT-R Theory (Adaptive Control of Thought—Rational), which describes how the mind organizes knowledge into Production Rules—condition-action pairs that capture procedural skills. A Cognitive Tutor begins with a Cognitive Task Analysis, breaking down a domain like algebra or programming into the specific knowledge components students need. These components are classified as Declarative and Procedural Knowledge: facts versus skills. The tutor then uses two key algorithms. Model Tracing compares each student action to a cognitive model of correct and incorrect solution paths, allowing the system to follow the student's reasoning step by step. Knowledge Tracing estimates the probability that a student has learned each knowledge component, updating the Student Model over time. This model enables Cognitive Diagnosis—identifying what the student knows and where they struggle. The instructional principles are straightforward: Learning by Doing (students actively solve problems rather than passively receive information), Immediate Corrective Feedback (errors are addressed right away to prevent practice of incorrect procedures), and Mastery Learning (students progress only after demonstrating competence on each component). To support learning, tutors provide Hint Sequences—graduated prompts that guide the student toward the correct next step without giving away the answer.
    The intellectual history of Cognitive Tutors begins with the rise of Information Processing Theory and Expert-Novice Studies in the 1970s, which showed that expertise involves not just more knowledge but better-organized, proceduralized knowledge. John R. Anderson, building on his ACT-R theory of cognition, saw that these ideas could be turned into a teaching machine. In the early 1980s, Anderson and his colleagues at Carnegie Mellon University began developing tutors for high school geometry and algebra. A seminal work was Anderson's 1983 book The Architecture of Cognition, which laid out the ACT-R framework. The first Cognitive Tutor for algebra, developed with Kenneth Koedinger and Albert Corbett, appeared in the mid-1980s and was tested in Pittsburgh schools. A key paper, "Cognitive Tutors: Lessons Learned" (Anderson, Corbett, Koedinger, & Pelletier, 1995), codified the approach and reported significant learning gains. The framework drew on precursors like Schema Theory and Behaviorism only indirectly—its real roots were in cognitive science's shift from behaviorist stimulus-response to mental models. Cognitive Tutors coexisted with and influenced neighboring frameworks: Cognitive Apprenticeship (which emphasized social context and modeling) and Design-Based Research (which used iterative design of learning environments). While later frameworks like Learning Analytics and Computer-Supported Collaborative Learning expanded the focus to data mining and group interaction, Cognitive Tutors remain a landmark demonstration that cognitive theory can directly inform effective educational technology. Their legacy includes widespread use in mathematics classrooms and the development of platforms like Carnegie Learning's MATHia.
    The 12 concepts its tower teaches
    • ACT-R Theory
    • Cognitive Diagnosis
    • Cognitive Task Analysis
    • Declarative and Procedural Knowledge
    • Hint Sequences
    • Immediate Corrective Feedback
    • Knowledge Tracing
    • Learning by Doing
    • Mastery Learning
    • Model Tracing
    • Production Rules
    • Student Model

    Fly to its tower in Nousfarer

  7. 1987-Present

    Cognitive Apprenticeship

    How can expert processes be made visible to learners?

    Fly to its tower in Nousfarer

  8. 1987-Present

    Situated Cognition

    How is knowledge bound to context and activity?

    Fly to its tower in Nousfarer

  9. 1990-Present

    Distributed Cognition

    How is thinking spread across people and tools?

    Fly to its tower in Nousfarer

  10. 1991-Present

    Community of practice

    How do newcomers become full members of a group?

    Fly to its tower in Nousfarer

  11. 1992-Present

    Design-Based Research

    How can research and design improve real classrooms together?

    Fly to its tower in Nousfarer

  12. 1995-Present

    Computer-Supported Collaborative Learning

    How do groups build shared understanding with technology?

    Fly to its tower in Nousfarer

  13. 2000-Present

    Complex Systems Approaches

    How does learning emerge from interacting agents?

    Fly to its tower in Nousfarer

  14. 2005-Present

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

BehaviorismCognitivismExpertise and Novice StudiesConstructionismIntelligent tutoring systemCognitive TutorsCognitive ApprenticeshipSituated CognitionDistributed CognitionCommunity of practiceDesign-Based ResearchComputer-Supported Collaborative Lear…Complex Systems ApproachesLearning Analytics and Educational Da…

Beyond the field

Where Learning Sciences's ideas came from in other fields, and what they led to there: the strongest 12 of 77 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Learning Sciences is a planet in Nousfarer

Its city has a tower for each of these 14 frameworks. Land, walk in, and climb one concept at a time.

Land on Learning Sciences

Related fields

Fields its ideas are linked to

More in Education

See where Learning Sciences sits on the Map of Human Knowledge