Attention

Attention is the mind's way of deciding what to process and what to ignore.

11 frameworks, from 1879 to today · Cognitive Science · Social Sciences

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

Overview

Attention is the mind's way of deciding what to process and what to ignore1. At any moment the senses deliver far more information than the brain can fully analyse, and behaviour demands that some of it be prioritized: the voice you are listening to over the hum of the room, the brake light ahead over the billboards beside it. Cognitive science studies attention as a set of mechanisms—selection, filtering, orienting, sustaining, dividing, and switching—that allocate limited processing to some inputs, thoughts, or actions rather than others1. The field matters because almost every other mental capacity depends on it: perception, memory, learning, and skilled action all presuppose that something has been selected. Its history is a sequence of attempts to say precisely what is selected, at what stage, by what kind of mechanism, and under whose control.

Introspection and the First Experimental Psychology

Attention was a central topic of the introspective psychology that founded scientific psychology in the late nineteenth century. Wilhelm Wundt, whose laboratory at Leipzig opened in 1879, treated attention as the process by which contents of consciousness are apperceived—brought into clear focus—and distinguished passive from active attention. William James, in The Principles of Psychology (1890), gave the topic its most quoted formulation: everyone knows what attention is, he wrote, because it is "the taking possession by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or trains of thought." James distinguished sensory from intellectual attention and voluntary from involuntary attention, and he treated attention as a selection among competing streams of consciousness.

The method was introspection: trained observers reported the contents of their own experience under controlled conditions23. This produced rich descriptions but a problem the approach could not solve. Introspective reports were private, hard to compare across observers, and vulnerable to the very attentional processes being studied. When behaviourism swept Anglo-American psychology in the 1910s and 1920s, attention largely disappeared from serious research, since it was neither observable behaviour nor easily operationalized. It survived in Europe, notably in the work of the Dutch psychologist Franciscus Donders and later in Gestalt psychology, but the mainstream went quiet for several decades.

The Information-Processing Revolution and Filter Theory

Attention returned as a rigorous experimental topic with the information-processing revolution of the 1950s1. The immediate practical spur was wartime: radar operators and air traffic controllers had to monitor several channels at once, and engineers wanted to know the limits of human capacity. A key experimental tool was dichotic listening, in which different messages are played to each ear and the listener is asked to repeat one of them aloud—a task called shadowing1.

Colin Cherry, in the early 1950s, showed that listeners shadowing one message could report almost nothing about the unattended message beyond crude physical features such as whether it was speech or a tone, or whether the voice was male or female. In 1958 Donald Broadbent proposed the first major model. He argued that the senses feed a limited-capacity channel, and that a selective filter, operating on simple physical characteristics, admits only one input at a time to that channel145. The filter was conceived as an all-or-nothing gate. Broadbent's model explained why unattended speech was lost: it never reached the limited-capacity stage1. It was the first explicit statement of what came to be called filter theory, and it set the terms of debate for a generation.

Attenuation and Late Selection

Filter theory soon met evidence it could not accommodate. Neville Moray, in 1959, found that some listeners noticed their own name in the unattended ear—the "cocktail party" phenomenon. If the filter were an absolute gate, this should be impossible. Anne Treisman, working with Broadbent in the early 1960s, proposed attenuation theory: the filter does not block unattended inputs but turns their volume down. Attenuated messages can still be processed, especially if they are personally significant or if their content is primed, which is why a name can break through. Treisman later developed the idea that selection operates on features organized in a hierarchy, with some features requiring more processing than others.

A more radical challenge came from late selection theory, advanced by J. Anthony Deutsch and Diana Deutsch in 1963 and developed experimentally by Donald Norman in the late 1960s. On this view, all inputs are processed to the level of meaning, and selection happens only at the stage of response—what to report or act on15. The evidence included findings that unattended words could influence behaviour, for instance in priming tasks. The debate between early selection (selection before meaning is extracted) and late selection (selection after) became one of the defining controversies of the 1960s and 1970s. Its resolution was not a victory for either side but a recognition that selection can occur at multiple stages, and that the question is empirical: which stage dominates depends on the task and the load.

Capacity, Resources, and Effort

A parallel tradition asked not where selection happens but how much processing is available. Resource theory, associated above all with Daniel Kahneman's Attention and Effort (1973), treated attention as a limited pool of capacity that can be allocated flexibly among tasks. The pool is not a single undifferentiated quantity; Kahneman and others distinguished arousal from the allocation policy that distributes effort. The theory explained why difficult tasks interfere with one another more than easy ones, and why performance degrades under high demand. It also connected attention to effort and to physiological arousal.

Resource models were powerful but vague about what a "resource" is, and later work—notably by Christopher Wickens and by Navon and Gopher in the 1980s—proposed multiple, partly independent resources (for example, separate pools for verbal and spatial processing). The approach's lasting contribution was to make capacity and interference central experimental variables. Its weakness was that a resource pool is hard to observe directly, and the theory risked explaining any result after the fact by positing the right kind of resource.

Orienting, Control, and the Neural Turn

From the 1970s onward, attention research became increasingly physiological. Michael Posner's work on spatial orienting distinguished three operations—disengaging attention from a location, moving it, and engaging it at a new location—and used reaction-time tasks with spatial cues to measure them. Posner and colleagues linked these operations to distinct brain networks, and the covert orienting paradigm (attending to a location without moving the eyes) became a workhorse of the field. This line of work established attention as a set of control processes with identifiable neural substrates, and it connected the psychology of attention to the study of eye movements and to clinical conditions such as hemispatial neglect, in which patients ignore one side of space.

The neural turn accelerated with single-cell recording in monkeys. Robert Wurtz, Michael Goldberg, and others showed in the 1970s and 1980s that neurons in parietal and frontal areas respond more strongly to stimuli at attended locations, and that attention can enhance responses in visual cortex itself. The question of how attention modulates sensory processing became central, and it set up the theoretical developments that followed.

Selection for Action and Biased Competition

In the 1980s and 1990s, two related ideas reframed attention. The first, selection for action, argued that attention is fundamentally about guiding behaviour rather than about consciousness. On this view, selection is the process by which the perceptual system picks out the objects relevant to the actions an organism is prepared to perform6. The idea, developed by Odmar Neumann and others, dissolved some of the early-versus-late selection debate by shifting the question from "what reaches awareness" to "what gets used."

The second, biased competition, was formulated by Robert Desimone and John Duncan in the 1990s. It holds that stimuli compete for representation in the brain, and that attention biases this competition in favour of behaviourally relevant inputs6. The competition can be resolved by bottom-up salience or by top-down signals from frontal and parietal areas that enhance the responses of neurons representing the attended object and suppress their rivals. Biased competition unified psychological findings about selection with neural data on receptive fields and response modulation, and it remains one of the field's central frameworks.

Load Theory and the Return of Early Selection

In the 1990s and 2000s, Nilli Lavie's load theory offered a reconciliation of the early-versus-late selection controversy. Lavie argued that the stage at which selection occurs depends on perceptual load1. When a task imposes high perceptual load, spare capacity is exhausted and irrelevant stimuli are filtered out early; when load is low, spare capacity spills over and irrelevant stimuli are processed, so selection appears late1. The theory made testable predictions about distractor interference as a function of task difficulty, and it revived the idea that capacity limits, not a fixed filter, determine what gets through. It is a good example of how the field's older debates were not so much settled as reframed in terms of interacting variables.

Computational and Connectionist Models

Alongside the experimental work, computational modelling gave attention a formal vocabulary. Connectionist models of the 1980s and 1990s represented attention as the modulation of activation in networks of simple units, and they explored how top-down signals could bias processing without a central controller. Models of visual search, such as those developed by Treisman and by Jeremy Wolfe, formalized the distinction between parallel processing of simple features and serial processing of conjunctions of features. Later computational work, including Bayesian and predictive-processing accounts, treated attention as precision-weighting: a way of increasing the influence of reliable or relevant signals in inference. These models made attention a computational problem—how to allocate limited representational resources—rather than only a psychological one.

Predictive Processing and the Present

The most recent major framework is predictive processing, developed from the 2000s onward by Karl Friston, Andy Clark, and others. On this view, the brain is constantly generating predictions about its inputs and updating them in light of prediction errors7. Attention is reconceived as the process that weights prediction errors—determining which errors are trusted and therefore drive learning and perception. This gives attention a role in learning itself, not just in selecting among already-processed inputs, and it connects attention to Bayesian inference and to the broader computational turn in cognitive science.

Today the field is plural. Psychophysical and behavioural paradigms—cueing, search, dual-task, flanker tasks—remain central. Cognitive neuroscience uses fMRI, EEG, and single-cell recording to map attentional networks and their interactions. Computational models, from connectionist networks to Bayesian and predictive accounts, provide formal hypotheses. Clinical and applied work studies attention in neglect, attention-deficit conditions, ageing, and human-machine interaction. The old questions persist in new form: what exactly is selected, at which stage, by what mechanism, and how do bottom-up salience and top-down goals interact? The field has not settled them, but it has built an increasingly precise vocabulary—filtering, capacity, orienting, competition, load, precision—for asking them.

References

  1. Attention (introduction, 1950–1999, Filter, Selection, Load), Wikipedia
  2. Functional psychology (History), Wikipedia
  3. Structuralism (psychology), Wikipedia
  4. Perceptual load theory (History), Wikipedia
  5. Selective auditory attention (History), Wikipedia
  6. Biased competition theory, Wikipedia
  7. Predictive coding, Wikipedia

The frameworks of Attention

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

  1. 1879-1915

    Introspective Psychology of Attention

    How does the mind bring contents into clear focus?

    Fly to its tower in Nousfarer

  2. 1958-1975

    Broadbent's filter model of attention

    Where does the mind filter out unwanted input?

    Fly to its tower in Nousfarer

  3. 1960-1980

    Attenuation Theory

    Why do ignored messages sometimes break through?

    Fly to its tower in Nousfarer

  4. 1963-1980

    Late Selection Theory

    Is selection before or after meaning is extracted?

    Fly to its tower in Nousfarer

  5. 1973-1995

    Resource Theory

    How is limited mental capacity divided among tasks?

    Fly to its tower in Nousfarer

  6. 1980-Present

    Computational and Connectionist Models of Attention

    How can selection be implemented in a machine or network?

    Fly to its tower in Nousfarer

  7. 1980-Present
  8. 1987-2010

    Selection for Action

    How does attention guide what we do?

    Fly to its tower in Nousfarer

  9. 1995-Present

    Biased competition theory

    How do stimuli compete for representation in the brain?

    Fly to its tower in Nousfarer

  10. 1995-Present

    Perceptual load theory

    Does perceptual load decide when selection happens?

    Fly to its tower in Nousfarer

  11. 2005-Present

    Predictive Processing

    How does attention weight prediction errors in inference?

    Fly to its tower in Nousfarer

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.

Introspective Psychology of AttentionBroadbent's filter model of attentionAttenuation TheoryLate Selection TheoryResource TheoryComputational and Connectionist Model…Spatial Orienting and Attentional Net…Selection for ActionBiased competition theoryPerceptual load theoryPredictive Processing

Beyond the field

Where Attention's ideas came from in other fields, and what they led to there: the strongest 12 of 13 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.

Attention is a planet in Nousfarer

Its city has a tower for each of these 11 frameworks. Land, walk in, and climb one concept at a time.

Land on Attention

Related fields

Fields its ideas are linked to

More in Cognitive Science

See where Attention sits on the Map of Human Knowledge