Educational Psychology

Abstract

Educational psychology is the branch of applied psychology that studies how people learn in instructional settings and how teaching can be arranged to make learning more effective. It emerged as a distinct discipline early in the twentieth century, when Edward Thorndike argued that education should rest on measurement and the experimental study of learning. This article traces the field through its central research programs: the cognitive architecture that constrains how much a learner can process at once, the motivational processes that determine whether effort is invested, the instructional methods that reliably raise achievement, and the retrieval-based techniques that turn study time into durable memory. It closes with the field's turn toward large-scale interventions and cumulative meta-analysis. Three interactive demonstrations let the reader overload a working memory, compare study strategies, and combine effect sizes across studies.

Keywords: educational psychology, learning, cognitive load, motivation, testing effect

Educational psychology is the field that applies the concepts, findings, and methods of psychology to teaching and learning. Its defining premise is that instruction can be studied scientifically: that the conditions under which people acquire knowledge and skill are lawful, measurable, and open to experiment, and that education should be designed on the basis of that evidence rather than on tradition or intuition. The field took its modern shape when Edward Thorndike, in the opening article of the first issue of the Journal of Educational Psychology, argued that psychology could give education the same rigorous, quantitative footing that the physical sciences had given engineering, and that the study of learning belonged in the laboratory (Thorndike, 1910). Over the following century the discipline absorbed the cognitive revolution, and its questions became as much about the mental architecture of the learner as about the arrangement of the classroom. Educational psychology today sits at the junction of cognitive science and instructional practice, and its questions are applied as often as they are theoretical.

What Educational Psychology Is

Educational psychology is best understood as the scientific study of learning in instructional contexts, together with the application of that science to the design of teaching, assessment, and educational environments. It overlaps with several neighboring fields whose boundaries are more institutional than real. It draws its theories of memory, attention, and reasoning from cognitive psychology, and it shares much of its subject matter with the psychology of learning, but it is distinguished by its applied purpose: it asks not only how learning works but how instruction should be arranged so that more of it happens. From its founding the field defined itself by method as much as by topic. Thorndike insisted that the contribution of psychology to education would come through measurement, the careful quantification of individual differences and of the effects of instruction, so that educational claims could be tested against data rather than asserted (Thorndike, 1910). That commitment to evidence remains the field's signature, and it explains why so much of educational psychology takes the form of controlled comparisons between instructional conditions and, increasingly, of meta-analyses that pool those comparisons across many studies.

Types of Educational Psychology

Psychology, Educational is a formal descriptor in the National Library of Medicine's Medical Subject Headings, filed under Psychology, Applied at tree position F02.784.629 and cross-listed under Psychology at F04.096.628.679. Beneath the descriptor MeSH hangs the narrower headings listed in Table 1. Two cautions apply. The list is an indexing classification built to organize the biomedical literature rather than a theory that carves the field at its joints, and its categories are not mutually exclusive: a study of a child's underachievement may equally concern aptitude, remedial teaching, and guidance at once. Only subtypes that are themselves live articles on this site are linked, and at present only Learning has a page of its own.

Cognitive Load and the Architecture of Learning

The single most consequential fact educational psychology has imported from cognitive science is that human working memory is severely limited. A learner can hold and manipulate only a few novel elements at once, and this bottleneck, rather than any shortage of long-term storage, is what makes learning hard. John Sweller built cognitive load theory on this observation, arguing that instruction must be designed around the limits of working memory because material that exceeds those limits simply cannot be processed into the schemas that constitute expertise (Sweller, 1988). The theory distinguishes the load intrinsic to the material, which depends on how many elements interact and cannot be reduced without changing what is learned, from the extraneous load imposed by poor instructional design, which can and should be eliminated. A worked example that shows a solution step by step imposes less extraneous load than an unguided problem, because it does not force the novice to search a vast problem space while also trying to learn its structure. Two decades of subsequent work refined the theory by tying it explicitly to the broader cognitive architecture, in which a capacious long-term memory built through practice is what ultimately relieves the working-memory bottleneck, since a retrieved schema counts as a single element however complex it is (Sweller et al., 2019). Richard Mayer applied the same logic to multimedia, showing that instructional materials combining words and pictures are learned better when they are designed to manage load, for instance by placing labels next to the graphics they describe rather than forcing the eye and mind to hold the two apart (Mayer, 2003). The demonstration below lets the reader push a limited working memory past its capacity and watch comprehension collapse as extraneous elements are added.

Motivation and the Will to Learn

Instruction can be perfectly designed and still fail if the learner does not invest effort, so educational psychology has always had to explain motivation as well as cognition. The field's most influential account begins with belief about the self. Albert Bandura argued that self-efficacy, a person's judgment of whether they can execute the actions a task requires, is a principal determinant of whether they attempt difficult work, how much effort they expend, and how long they persist in the face of failure, so that two learners of equal ability can differ sharply in outcome according to what they believe they can do (Bandura, 1977). A complementary tradition asks not only how much motivation a learner has but of what kind. Edward Deci and Richard Ryan's self-determination theory distinguishes intrinsic motivation, in which an activity is pursued for its own satisfaction, from extrinsic regulation, in which it is pursued for a separable reward, and holds that intrinsic motivation flourishes when the basic needs for autonomy, competence, and relatedness are met, which is why controlling rewards and pressures can paradoxically undermine the very engagement they are meant to produce (Deci & Ryan, 2000). A third strand concerns what learners believe about the nature of ability itself. Carol Dweck's research on implicit theories of intelligence showed that students who regard ability as fixed and those who regard it as capable of growth respond very differently to difficulty, the former tending to withdraw effort to protect their self-image and the latter treating a setback as information about where to work harder. Motivation, on this combined view, is not a fuel gauge but a structure of beliefs and needs, and instruction affects it by shaping what learners conclude about themselves and their prospects.

Instructional Methods and Their Effects

If educational psychology is to guide practice, it must be able to say which instructional methods work and by how much, and here the field's commitment to measurement has paid its clearest dividend. Benjamin Bloom framed the problem sharply with what he called the two-sigma problem: students taught one to one by a tutor performed about two standard deviations better than students taught conventionally in a class of thirty, an enormous gap, and Bloom posed the search for group methods that could approach that benefit as the central challenge for instructional research (Bloom, 1984). The modern answer to Bloom's challenge has come from the systematic accumulation of effect sizes across thousands of studies. John Hattie's synthesis of meta-analyses arranged hundreds of instructional influences on a common scale of effect size, allowing them to be compared directly and showing that while almost everything a teacher does has some positive effect, the influences differ by an order of magnitude, so that the practical question is not whether a method works but whether it works better than the alternatives competing for the same instructional time (Hattie & Donoghue, 2016). This meta-analytic turn transformed the field from a collection of favored theories into a cumulative science in which claims about instruction are weighed by pooled evidence. Table 2 reports illustrative effect sizes for several well-studied influences, expressed in the standardized units that make them comparable.

Because these influences differ so sharply in their return, the choice among study strategies is consequential. The demonstration below projects how much a fixed block of study time is retained over a delay under three common strategies, so that the long-run payoff of each can be compared directly rather than judged by how productive it feels in the moment.

Retrieval Practice and Durable Memory

Among the many findings educational psychology has produced, one stands out for the strength of its evidence and the clarity of its advice: testing is not merely a way to measure learning but a way to cause it. Henry Roediger and Jeffrey Karpicke gave learners material to study and then had some restudy it while others were tested on it, and found that although restudying felt more productive and produced better performance on an immediate test, the tested group retained far more when memory was assessed after a delay of days, a dissociation between how well a strategy feels and how well it works that has large practical consequences (Roediger & Karpicke, 2006). This testing effect, or retrieval-practice effect, is now among the most replicated results in the science of learning. A meta-analysis by Olusola Adesope and colleagues pooled hundreds of comparisons and confirmed that practice testing produces a moderate-to-large advantage over restudying across a wide range of materials, test formats, and learners, and identified the conditions, such as the use of open-ended rather than recognition formats and the provision of feedback, under which the benefit is largest (Adesope et al., 2017). The practical importance of such findings led to a broad effort to identify which study techniques actually deliver durable learning. John Dunlosky and colleagues reviewed the evidence on ten widely used techniques and concluded that the two most effective, practice testing and distributed practice, were also among the least used by students, who gravitate instead to rereading and highlighting, which feel efficient but produce little lasting benefit (Dunlosky et al., 2013). Figure 1 depicts the crossover that makes retrieval practice so counterintuitive: the strategy that produces the worse immediate performance produces the better long-term retention.

The strength of a claim like this rests on pooling evidence across many experiments rather than trusting any one. The demonstration below combines several study-level effect sizes into a single meta-analytic estimate, weighting each study by its precision, and shows how the pooled value and its confidence interval respond as the inputs change.

Worked Example

The claim that some instructional methods work far better than others is only as useful as the ability to combine evidence across studies, so it helps to see exactly how effect sizes are pooled. Suppose three small experiments each compare retrieval practice against restudying on a delayed test, reporting the standardized mean difference, Hedges's g, and its standard error. Study A reports g = 0.60 with standard error 0.20; Study B reports g = 0.50 with standard error 0.15; Study C reports g = 0.80 with standard error 0.25. A fixed-effect meta-analysis weights each study by the inverse of its squared standard error, so that more precise studies count for more. The weights are 1 divided by 0.20 squared, or 25.0, for Study A; 1 divided by 0.15 squared, or 44.4, for Study B; and 1 divided by 0.25 squared, or 16.0, for Study C. The pooled estimate is the weighted average: multiply each effect by its weight, add the products, and divide by the total weight. The products are 0.60 times 25.0, or 15.0; 0.50 times 44.4, or 22.2; and 0.80 times 16.0, or 12.8. Their sum is 50.0, and the total weight is 25.0 plus 44.4 plus 16.0, or 85.4. The pooled effect is therefore 50.0 divided by 85.4, or 0.585. The standard error of the pooled estimate is the square root of 1 divided by the total weight, the square root of 1 divided by 85.4, which is 0.108, giving an approximate 95 percent confidence interval of 0.585 plus or minus 1.96 times 0.108, or 0.37 to 0.80. The pooled effect of roughly 0.59 is both larger and far more precisely estimated than any single study, and its interval excludes zero, which is exactly why meta-analysis, and not any one experiment, is what lets educational psychology assert that retrieval practice reliably helps (Adesope et al., 2017). The demonstration above lets these inputs be varied; the same inverse-variance weighting produces the pooled value each time.

Discussion

Educational psychology has moved in a little over a century from a programmatic hope, that education might be placed on a scientific footing, to a body of cumulative evidence about how people learn and how instruction should be arranged. Cognitive load theory supplied an architecture, grounding instructional design in the hard limits of working memory; the motivation literature supplied an account of why learners engage or disengage; the meta-analytic synthesis of instructional effects supplied a way to rank methods by their measured benefit rather than their intuitive appeal; and the science of retrieval practice supplied a finding both robust and immediately actionable. The applied payoff has been substantial, since recognizing that testing causes learning, that study should be spaced, and that extraneous cognitive load should be stripped from materials gives teachers concrete, evidence-based levers. Real tensions remain. The field has been shaken by the wider replication crisis in psychology, and some once-celebrated findings, including aspects of the mindset and learning-styles literatures, have proved smaller or more conditional than early enthusiasm suggested. Effect sizes drawn from tightly controlled laboratory studies do not always survive the messiness of real classrooms, and the same intervention can help in one context and vanish in another. What is no longer in doubt is the central claim the field was built to defend: that learning is a lawful process, that instruction can be studied experimentally, and that some ways of teaching are demonstrably better than others.

Current Directions

The most visible recent development in educational psychology is the move to test its interventions at scale and with mechanistic precision. David Yeager and colleagues carried out a preregistered national experiment in which a short online growth-mindset intervention was delivered to a nationally representative sample of ninth-graders, and found that it produced a small but real improvement in the grades of lower-achieving students, an effect that appeared only where the school's peer norms supported the challenge-seeking behavior the intervention encouraged (Yeager et al., 2019). The study is notable as much for its method as its result: by using a representative sample, preregistration, and a heterogeneity analysis, it answered the replication-era demand that a psychological intervention specify not just whether it works but for whom and under what conditions. A second direction is theoretical consolidation of the cognitive architecture, as cognitive load theory has been reworked to align it with evolutionary accounts of how humans acquire biologically secondary knowledge, the culturally invented material such as reading and algebra that schools exist to teach and that the mind is not specially adapted to learn (Sweller et al., 2019). Running through both strands is the same commitment that has defined the field since Thorndike: that instructional claims must be settled by cumulative, quantitative evidence, now gathered with larger samples, sharper designs, and a franker acknowledgment of how much the effect of any method depends on the context in which it is deployed.

Common Misconceptions

Students learn best when taught in their preferred learning style.

The popular claim that matching instruction to a student's visual, auditory, or kinesthetic style improves learning has little experimental support; controlled tests rarely find the predicted interaction. What reliably helps is matching the format to the material, as multimedia research shows, not to a supposed trait of the learner (Mayer, 2003).

Rereading and highlighting are efficient ways to study.

These techniques feel productive because they make material fluent, but fluency is a poor guide to memory. When study techniques are compared directly, rereading and highlighting rank among the least effective, while practice testing and distributed practice, which feel harder, produce far more durable learning (Dunlosky et al., 2013).

Testing only measures learning; it does not create it.

The act of retrieving information from memory strengthens it. Learners who are tested on material retain more of it later than learners who spend the same time restudying, so a test is itself a learning event, not merely an assessment of one (Roediger & Karpicke, 2006).

Glossary

Achievement.

Demonstrated success in reaching an educational goal or attaining a level of proficiency, and the outcome most instructional research is designed to raise.

Aptitude.

The capacity to acquire a particular kind of skill or knowledge, distinguished from present attainment.

Cognitive load.

The demand an instructional task places on working memory, divided into the intrinsic load of the material and the extraneous load imposed by its presentation.

Distributed practice.

The spacing of study sessions over time rather than massing them together, which reliably improves long-term retention.

Effect size.

A standardized measure of the magnitude of a difference, such as Cohen's d or Hedges's g, that allows results from different studies to be compared and combined.

Extraneous load.

The portion of cognitive load created by how material is presented rather than by its inherent difficulty, and the target of good instructional design.

Feedback.

Information given to a learner about the correctness or quality of their performance, among the most powerful of common classroom influences on achievement.

Implicit theory of intelligence.

A learner's belief about whether ability is fixed or can grow, which shapes how they respond to difficulty and failure.

Intrinsic motivation.

Engagement in an activity for the satisfaction inherent in it, distinguished from extrinsic regulation driven by separable rewards.

Learning.

The relatively durable change in knowledge or behavior that results from experience, the core process educational psychology exists to support.

Meta-analysis.

A statistical synthesis that pools effect sizes across many studies, weighting each by its precision, to produce a more stable overall estimate.

Retrieval practice.

The act of recalling information from memory, which strengthens later retention more than restudying does; also called the testing effect.

Schema.

An organized mental structure that represents knowledge of a category or procedure, and that counts as a single element in working memory however complex it is.

Self-efficacy.

A person's judgment of their capability to carry out the actions a task requires, a principal determinant of effort and persistence.

Two-sigma problem.

Bloom's observation that individually tutored students outperform conventionally taught ones by about two standard deviations, and his challenge to find group methods as effective.

Working memory.

The limited-capacity system that holds and manipulates information over short intervals, whose narrow limits constrain how much can be learned at once.

Key Researchers

Albert Bandura (1925-2021). Psychologist at Stanford University whose social cognitive theory and concept of self-efficacy reframed motivation as grounded in beliefs about personal capability. Wikipedia - Wikidata

Benjamin S. Bloom (1913-1999). Educational psychologist at the University of Chicago; his taxonomy of educational objectives, work on mastery learning, and framing of the two-sigma problem shaped instructional research. Wikipedia - Wikidata

Carol S. Dweck (b. 1946). Psychologist at Stanford University whose research on implicit theories of intelligence established how beliefs about the malleability of ability affect responses to difficulty. Google Scholar - Faculty Page - Wikipedia - Wikidata

John A. C. Hattie (b. 1950). Educational researcher at the University of Melbourne whose Visible Learning synthesized meta-analyses of achievement into a common scale of effect sizes. ORCID - Faculty Page - Wikipedia - Wikidata

Richard E. Mayer (b. 1947). Psychologist at the University of California, Santa Barbara whose cognitive theory of multimedia learning applies working-memory limits to the design of instructional materials. ORCID - Google Scholar - Faculty Page - Wikipedia - Wikidata

Henry L. Roediger (b. 1947). Psychologist at Washington University in St. Louis whose experiments established the testing effect, showing that retrieval strengthens memory more than restudying. ORCID - Google Scholar - Faculty Page - Wikipedia - Wikidata

John Sweller (b. 1946). Educational psychologist and Emeritus Professor at the University of New South Wales who originated cognitive load theory, grounding instructional design in the limits of working memory. ORCID - Faculty Page - Wikipedia - Wikidata

Edward L. Thorndike (1874-1949). Psychologist at Teachers College, Columbia University, widely regarded as the founder of educational psychology; his law of effect and insistence on measurement gave the field its scientific footing. Wikipedia - Wikidata

David S. Yeager (living). Psychologist at the University of Texas at Austin whose national-scale, preregistered experiments test motivational interventions such as growth mindset with representative samples. ORCID - Google Scholar - Faculty Page

Frequently Asked Questions

What is educational psychology? Educational psychology is the branch of applied psychology that studies how people learn in instructional settings and how teaching, assessment, and educational environments can be designed to make learning more effective. It applies theories of memory, motivation, and reasoning to practical questions of instruction, and it rests on the premise that teaching can be studied experimentally (Thorndike, 1910).

How is educational psychology different from the psychology of learning? The two overlap heavily, and educational psychology draws its theories of learning directly from that literature. The difference is one of purpose: the psychology of learning asks how learning works as a basic process, while educational psychology asks how instruction should be arranged so that more learning happens in real educational settings.

What is cognitive load theory? Cognitive load theory holds that instruction must be designed around the narrow limits of working memory. It distinguishes the intrinsic load of the material from the extraneous load imposed by poor presentation, and argues that effective teaching minimizes the latter so that capacity is free for learning, for instance by using worked examples instead of unguided problem solving (Sweller, 1988).

Does the testing effect mean tests actually improve learning? Yes. Retrieving information from memory strengthens it, so learners who are tested on material remember more of it later than learners who spend the same time restudying (Roediger & Karpicke, 2006). A meta-analysis of hundreds of comparisons confirms a moderate-to-large advantage for practice testing across materials and formats (Adesope et al., 2017).

Which study techniques work best? When common study techniques are compared directly, practice testing and distributed practice are the most effective, while rereading and highlighting, though popular, produce little durable benefit (Dunlosky et al., 2013). The most effective techniques often feel harder, which is why students underuse them.

Do learning styles improve instruction? There is little experimental support for the idea that matching instruction to a student's preferred learning style improves outcomes. What reliably helps is matching the format to the demands of the material, as multimedia-learning research shows, rather than to a supposed trait of the learner (Mayer, 2003).

What did Bloom's two-sigma problem show? Bloom found that students tutored one to one performed about two standard deviations better than students taught conventionally in large classes, an enormous gap. He posed the search for group instructional methods that could approach that benefit as a central challenge for educational research (Bloom, 1984).

Does a growth mindset intervention raise achievement? A preregistered national experiment found that a short growth-mindset intervention produced a small but real improvement in the grades of lower-achieving students, but only in schools whose peer norms supported challenge-seeking (Yeager et al., 2019). The effect is real, modest, and conditional on context, which is typical of scaled educational interventions.

References