Learning Glossary

Active recall is the practice of deliberately retrieving information from memory – through self-testing or flashcards – rather than simply rereading it. The act of retrieval itself strengthens the memory trace.

An AI hallucination is a statement produced by an AI language model that is factually wrong or entirely made up, yet sounds fluent and confident. It happens because the model predicts the most probable next word rather than checking the truth.

Chunking is the grouping of individual pieces of information into larger, meaningful units ("chunks"). Because working memory can hold only a few units at once, chunking increases how much you can effectively remember.

Cognitive load is the amount of information the working memory has to process at once while learning. Because that capacity is tightly limited, the level of load largely determines how well learning succeeds.

Dual coding is the pairing of words and images in learning. According to Allan Paivio's theory, the brain processes verbal and visual information in two separate systems; using both improves retention.

The Ebbinghaus forgetting curve describes how newly learned material is forgotten over time without review – rapidly at first, then more slowly. It goes back to the self-experiments of Hermann Ebbinghaus published in 1885.

The Feynman Technique is a learning method in which you explain a concept in plain language, as if teaching a child. Points where the explanation breaks down reveal gaps to study. It is named after the physicist Richard Feynman.

Flashcards are two-sided study cards with a question or cue on the front and the answer on the back. They force active recall and are ideal for spaced repetition.

Implementation intentions are specific if-then plans of the form "If situation X occurs, then I will do Y." By linking a planned action to a concrete cue, they make people more likely to follow through on their goals.

Interleaving means mixing different types of problems or topics within a single practice session instead of practising them in separate blocks. It makes practice harder but often improves later retention.

A large language model (LLM) is an AI system trained on huge amounts of text to process and generate language. It works by predicting, token by token, the most likely next piece of text given everything before it.

Learning styles refers to the widespread belief that people learn better when material is presented in their preferred sensory channel (e.g. visual or auditory). This meshing hypothesis is not supported by scientific evidence.

The Leitner system is a flashcard method using several boxes: correctly answered cards move to boxes reviewed less often, while mistakes go back to box one — combining active recall with spaced repetition.

Long-term memory stores knowledge and experience durably — from seconds to a lifetime. Its capacity is considered effectively unlimited, and it is organised into several specialised systems.

Metacognition is awareness and control of one's own thinking and learning — the ability to plan, monitor and evaluate how you learn. The term was coined by psychologist John Flavell in 1979.

The method of loci (memory palace) is a mnemonic technique in which you attach information to fixed locations along a familiar route and recall it by walking that route in your mind.

Microlearning is an instructional approach that delivers content in small, tightly focused units of usually just a few minutes. Studies point to better retention and lower cognitive load, though no universally agreed definition exists.

A mnemonic is a memory aid that links hard-to-remember information to easily recalled images, rhymes or patterns, making facts quicker to encode and retrieve.

NotebookLM is Google's AI-powered research and note-taking tool that builds answers only from sources you upload. It is best known for "Audio Overviews", which turn documents into a podcast-like AI conversation.

OCR (optical character recognition) is the machine conversion of images of printed or written text into machine-readable, searchable text. It turns scans, photos and image-only PDFs into editable content.

The Pomodoro Technique is a time-management method: you work in focused intervals of usually 25 minutes (a "pomodoro"), separated by short breaks. It was developed in the late 1980s by Francesco Cirillo.

Procrastination is the voluntary delay of an intended action despite expecting to be worse off for it. It is regarded as the quintessential failure of self-regulation and hits learning especially hard.

A prompt is the input — usually text — you give a generative AI such as a language model to make it perform a task. It can range from a short question to a detailed instruction with context and examples.

Retrieval-augmented generation (RAG) is a technique in which a language model retrieves relevant information from an external knowledge source before answering and incorporates it into the response, making answers more current and factual.

Spaced repetition is a learning method in which material is reviewed at expanding time intervals – ideally just before it would be forgotten. It fixes knowledge in memory more efficiently than cramming everything at once.

The spacing effect is the finding that learning distributed over time produces better long-term retention than the same amount of study massed together. It is one of the most robust results in memory research.

Test anxiety is a situation-specific form of anxiety triggered by evaluative situations such as exams and tests. It has a cognitive side (worry, rumination) and a physical side (tension) that can impair performance.

The testing effect (also called the retrieval-practice effect) is the finding that trying to recall information on a test strengthens long-term retention more than rereading the same material. Testing is not only measurement – it is learning.

Text-to-speech (TTS, speech synthesis) is software that automatically converts written text into spoken audio. Modern neural systems produce voices that trained listeners can barely distinguish from human recordings.

Working memory is the cognitive system that holds information active for a few seconds while manipulating it. Its capacity is severely limited — to roughly four units — making it the bottleneck of every learning process.