Last reviewed on October 2, 2026.
The field of cognitive science spans many disciplines, and it comes with its own terminology. Use this glossary to look up basic definitions and refresh your memory on important concepts. The definitions are intentionally brief and listed alphabetically; each term has its own anchor, and many link to a longer article on the site.
Jump to: A · B · C · D · E · F · H · I · L · M · N · O · P · Q · R · S · T · W · Z
- ACT-R
- A cognitive architecture developed at Carnegie Mellon University that models cognition as a set of interacting modules for perception, memory and action.
- Aphasia
- A language disorder, usually caused by brain injury, that affects production, comprehension or both.
- Artificial Intelligence
- Computer systems designed to perform tasks requiring human‑like intelligence.
- Attention
- The process of selectively focusing on certain stimuli while ignoring others. See attention and perception.
- Backpropagation
- A learning algorithm for multi-layer neural networks that adjusts weights by propagating the output error backwards through the network; popularised by Rumelhart, Hinton and Williams in 1986.
- Bayesian Inference
- A statistical approach in which beliefs are updated by combining prior probabilities with new evidence; used as a model of perception, learning and decision-making.
- Behaviourism
- The school of psychology, dominant in the US from roughly the 1910s to the 1950s, that limited psychology to observable behaviour and rejected talk of internal mental states. Its decline set the stage for the cognitive revolution.
- Bounded Rationality
- Herbert Simon’s idea that real decision makers are limited by time, information and computational capacity, so they satisfice (choose an option that is good enough) rather than optimise.
- Broca's Area
- A region of the left inferior frontal gyrus associated with speech production and aspects of grammatical processing.
- Change Blindness
- Failing to notice large changes to a visual scene when the change coincides with a brief interruption such as a blink, cut or flicker.
- Chunking
- Grouping items into meaningful units so that more information fits in working memory, as when a phone number is remembered as three groups of digits.
- Cocktail Party Effect
- The ability to follow one voice among many while still noticing salient unattended input, such as your own name, studied by Colin Cherry (1953).
- Cognition
- Mental processes involved in acquiring, storing, transforming and using knowledge — including perception, attention, memory, language, reasoning and decision making.
- Cognitive Architecture
- A unified theory or model describing the fixed structures and processes of cognition.
- Cognitive Bias
- A systematic deviation from rational judgement or decision making, usually produced by a heuristic applied outside the situations it suits. See why cognitive biases exist.
- Cognitive Load
- The demand a task places on limited working memory; cognitive load theory uses it to design instruction that avoids overloading learners.
- Cognitive Neuroscience
- The study of the neural mechanisms underlying cognitive processes.
- Cognitive Revolution
- The shift in the 1950s and 1960s from behaviourism to the scientific study of internal mental processes such as memory, attention, language and problem solving, driven by work from Chomsky, Miller, Newell, Simon and others. See the cognitive revolution.
- Cognitive Science
- The interdisciplinary study of mind and intelligence, drawing on psychology, neuroscience, artificial intelligence, linguistics, philosophy and anthropology. See what is cognitive science and its disciplines.
- Connectionism
- An approach to modelling cognition using networks of simple units analogous to neurons, where knowledge is stored in connection strengths. See connectionism explained.
- Consolidation
- The process by which new, fragile memories are stabilised over time, involving the hippocampus and, during sleep, reactivation of recent experiences.
- Deep Learning
- A subfield of machine learning that uses neural networks with many layers to learn representations from data.
- Dual-Process Theory
- The proposal that judgement and reasoning rely on two systems: a fast, automatic process and a slow, deliberate one.
- Electroencephalography (EEG)
- A method that records electrical activity from the scalp to study brain function.
- Embodied Cognition
- A view that cognition depends on the body’s interactions with the environment. See embodied cognition.
- Embodiment
- The view that the body shapes cognition through its sensory and motor interactions with the world; often used interchangeably with embodied cognition.
- Encoding
- The process of transforming sensory input into a form that can be stored in memory.
- Episodic Memory
- Memory for specific personally experienced events, including their time and place.
- Executive Function
- A set of high-level processes that control attention, working memory and goal-directed behaviour.
- Extended Mind
- The thesis of Andy Clark and David Chalmers (1998) that tools such as notebooks can, under certain conditions, form part of a person’s cognitive system.
- Framing Effect
- A bias in which the way a choice is described changes which option a person prefers, even when the underlying outcomes are equivalent.
- Functional Magnetic Resonance Imaging (fMRI)
- A brain imaging technique that measures changes in blood flow to infer neural activity.
- Hard Problem of Consciousness
- David Chalmers’ term for the question of why and how physical processes give rise to subjective experience, as opposed to the “easy problems” of explaining cognitive functions. See consciousness.
- Heuristic
- A simple rule or strategy used to make judgements or decisions quickly.
- Inattentional Blindness
- Failing to notice a clearly visible but unexpected object when attention is engaged elsewhere, famously shown in the “invisible gorilla” study (Simons & Chabris, 1999).
- Information Processing
- The framework, central to the cognitive revolution, that treats the mind as a system that takes in information, transforms and stores it, and uses it to guide behaviour.
- Language Acquisition
- The process by which humans learn to understand and produce language.
- Large Language Model (LLM)
- A very large neural network trained to predict text, which can generate fluent language. Cognitive scientists study LLMs both as tools and as possible models of human language processing. See AI and cognitive science.
- Learning
- The acquisition of knowledge or skills through experience, study or teaching.
- Long‑Term Memory
- Stores information over extended periods, including semantic, episodic and procedural memory.
- Magnetoencephalography (MEG)
- A non-invasive imaging method that measures magnetic fields produced by neural activity, with high temporal resolution.
- Marr's Levels of Analysis
- David Marr’s proposal (1982) that a cognitive system can be explained at three levels: computational (what problem is solved and why), algorithmic (which representations and procedures are used) and implementational (how it is realised physically).
- Memory
- The capacity to encode, store and retrieve information. See how memory works.
- Mental Model
- An internal, simplified representation of how something works or of a situation described, used to reason, predict and understand.
- Metacognition
- Awareness and regulation of one’s own thinking and learning processes.
- N400
- An event-related brain potential peaking about 400 ms after a word, larger when the word is semantically unexpected (“He spread his bread with socks”).
- Neural Network
- A computational model composed of interconnected units that process information in parallel.
- Neuroplasticity
- The brain’s capacity to change its structure and function in response to experience, learning and injury.
- Object Permanence
- The understanding that objects continue to exist when they cannot be seen. Piaget placed it in the sensorimotor stage; later studies suggest infants show signs of it earlier than he thought.
- Perception
- The interpretation of sensory information to form a coherent picture of the world.
- Perceptron
- An early single-layer neural network introduced by Frank Rosenblatt in 1958 that learns to classify inputs by adjusting connection weights.
- Phonology
- The study of the sound systems of languages.
- Piaget's Stages
- Jean Piaget’s four stages of cognitive development: sensorimotor (0–2 years), preoperational (2–7), concrete operational (7–11) and formal operational (11+). See cognitive development in children.
- Prediction Error
- The difference between what the brain predicted and the input it actually receives; in predictive processing it is the signal used to update beliefs.
- Predictive Processing
- A framework in which the brain continually generates predictions about incoming input and updates them based on prediction error. See predictive processing.
- Procedural Memory
- Memory for how to perform skills and actions, often expressed without conscious recollection.
- Psycholinguistics
- The study of how humans comprehend, produce and acquire language. See language processing in the brain.
- Qualia
- The subjective, felt qualities of experience, such as what it is like to see red or taste coffee.
- Reading Comprehension
- The construction of meaning from written text, combining word recognition, language comprehension, background knowledge and a mental model of the situation described.
- Reinforcement Learning
- Learning to choose actions that maximise reward, used both as a psychological theory and as a class of machine-learning algorithms.
- Representation
- A mental symbol or model that stands for information about the world.
- Retrieval
- The process of accessing stored information when it is needed.
- Retrieval Practice
- Learning by actively recalling information rather than re-reading it; one of the most reliable ways to improve long-term retention (the testing effect).
- Salience
- The property of a stimulus that makes it stand out from its surroundings and capture attention.
- Satisficing
- Choosing the first option that meets an acceptable threshold instead of searching for the best one; a term coined by Herbert Simon.
- Schema
- An organised mental structure that summarises what a person knows about a category of objects, events or relations.
- Selective Attention
- Focusing processing on one source of information while filtering out others, as when following one conversation in a noisy room (the cocktail party effect).
- Semantic Memory
- Memory for general facts and meanings, independent of when or where they were learned.
- Sensation
- The detection of physical energy (light, sound, pressure, chemicals) by sensory receptors. Perception is the interpretation that follows.
- Spacing Effect
- The finding that study sessions spread out over time produce better long-term retention than the same amount of study massed together.
- Symbolic AI
- An approach to artificial intelligence that represents knowledge using discrete symbols and explicit rules.
- Syntactic Parsing
- The process of analysing the grammatical structure of a sentence.
- Theory of Mind
- The ability to attribute mental states — beliefs, desires, intentions — to oneself and others. It is usually tested with false-belief tasks, which most children pass at around four to five years. See cognitive development.
- Top-Down Processing
- Perception guided by prior knowledge, expectations and goals, as opposed to bottom-up processing driven by the stimulus itself.
- Transcranial Magnetic Stimulation (TMS)
- A technique that uses magnetic fields to temporarily disrupt neural activity in a targeted region, used to test the causal role of that region in a task.
- Turing Test
- Alan Turing’s 1950 proposal to replace the question “Can machines think?” with a test of whether a machine’s conversation can be distinguished from a human’s.
- Wernicke's Area
- A region of the left posterior superior temporal cortex associated with language comprehension.
- Working Memory
- A limited capacity system for temporarily holding information available for processing.
- Zone of Proximal Development
- Lev Vygotsky’s term for the range of tasks a learner cannot yet do alone but can do with guidance from a more skilled person.
How to read these definitions
The entries above are intentionally short — a definition is meant to anchor a term, not to substitute for an explanation. The notes below give a little more context on the clusters of terms that newcomers find most slippery, and link to the longer articles on the site that go further.
Memory is not one thing
It is tempting to treat memory as a single capacity, but the field has long distinguished several stores and several processes. Working memory is the small, fast store that holds the contents of the current task — a phone number you are about to dial, the start of a sentence you are still reading. Long-term memory is much larger and includes semantic memory (facts and meanings), episodic memory (specific experiences) and procedural memory (skills). Each kind of memory depends on partly different brain systems, and each is supported by different encoding and retrieval processes. The article on memory and learning goes into how each type develops and decays.
Attention is not one thing either
Like memory, attention is a family of related processes. Selective attention filters the world; divided attention coordinates several streams; sustained attention keeps a focus over time; alternating attention shifts deliberately between tasks. The control of attention can be top-down (goal-driven) or bottom-up (driven by salience). Saying that someone "has trouble with attention" is rarely specific enough to be useful — the more interesting question is which kind of attention. The article on attention and perception works through the distinctions.
Representation, the bridge concept
Most theories in cognitive science share a working assumption: minds carry representations — internal stand-ins for things in the world. The disagreements are about what kind of representations they are. Symbolic AI treats representations as discrete tokens that are combined according to rules. Connectionism treats them as patterns of activation distributed across many simple units. Embodied cognition argues that some representations are not abstract at all but are grounded in sensory and motor experience. Different traditions answer different questions well; reading across them is more rewarding than choosing a side early.
Methods come in pairs
The methods in this glossary tend to trade off two things: spatial resolution (where in the brain something happens) against temporal resolution (when it happens). fMRI shows where with millimetre precision but lags behind neural activity by seconds. EEG and MEG follow neural activity in milliseconds but blur the location. TMS and lesion studies, by contrast, are not just measurement tools — they let researchers test whether a region is causally necessary for a function, not merely correlated with it. The article on research methods explains how studies often combine techniques to get the best of more than one.
Bias, heuristic and rationality
The terms heuristic, cognitive bias and dual-process theory are often run together. They are related but not the same thing. A heuristic is a strategy — a shortcut. A bias is a systematic departure from a normative standard, often produced by a heuristic. Dual-process theories propose that biases arise when fast, automatic reasoning takes over from slow, deliberate reasoning. None of this implies that intuition is always wrong: in many real-world settings the heuristic is well-calibrated to the environment. The article on decision making and biases walks through the most common ones and the conditions under which they bite.
This list is not exhaustive, and the field is constantly evolving. For more depth on any term, follow the linked articles or browse the blog.