2 Selective Attention

What selective attention does

Selective attention helps people follow a conversation in a noisy room or find a particular object in a cluttered scene. It prioritizes relevant information and reduces interference from competing input, but it is not necessarily an all-or-none gate: unattended information can sometimes influence perception or behavior.

Where selection occurs

Theories differ chiefly in where selection occurs during information processing. They offer competing accounts of a processing bottleneck rather than simply claiming that attention either works or does not.

Selection before full understanding

Early-selection theory places the filter before input is fully identified or understood. In Broadbent’s classic model, incoming signals are briefly registered for basic physical properties, such as which ear a sound comes from. A limited-capacity filter then selects a channel for more complete processing. This helps explain why people shadowing one of two simultaneous speech streams typically remember little of the other.

A graded filter

Attenuation theory, proposed by Treisman, describes unattended input as weakened rather than completely blocked. Its physical features and some meaning can still be processed, especially when context makes a word or message important. For example, a listener may notice their name in an ignored conversation, although this does not happen reliably for everyone.

Selection after meaning analysis

Late-selection theory, proposed by Deutsch and Deutsch, holds that inputs can be analyzed for meaning before selection. Attention then determines which information guides conscious report, memory, or action. Unattended input may therefore be understood without becoming the basis of the person’s response.

Evidence that some unattended information is noticed challenges a strict early filter, while failures to report unattended content challenge the idea that all input is fully analyzed. Accounts that consider task demands help explain why selection can appear earlier in some situations and later in others.

Listening to competing speech

In dichotic listening, different speech streams are presented to the two ears. A participant may be asked to shadow one stream by repeating it aloud as it is heard. Cherry’s experiments showed that listeners could follow the attended stream but had limited access to the words and meaning of the competing stream. These findings became important evidence for selective filtering.

Unattended speech is not always completely excluded. Treisman found that listeners could sometimes detect a message switch across ears, and that the likelihood of repeating a word from the wrong stream depended on contextual predictability. These results fit attenuation: the ignored stream is reduced in strength but can still affect processing.

Listening experiments show both the usefulness and the limits of a perceptual filter. People can prioritize one stream, but salient or contextually supported information from another stream may still break through. Results also depend on what is measured: correctly shadowing a message, noticing a change, and later recalling its meaning test different outcomes.

Visual search and feature selection

Visual-search tasks examine how attention selects a target among distractors. In a feature search, the target differs from distractors by one salient feature, such as a red item among green items. The target can often be found quickly, with response time changing relatively little as the number of items grows. This pattern is commonly described as a “pop-out” effect.

In a conjunction search, the target is defined by a combination of features, such as a red vertical line among red horizontal lines and green vertical lines. Search often becomes slower as the display grows, consistent with attention needing to inspect items or bind their features together.

Treisman and Gelade’s feature-integration theory proposes that basic features are registered in parallel, while focused attention helps combine features into coherent object representations. These search results illustrate how attention supports perceptual organization, not just the choice between competing messages.

Task demands and flexible selection

Search patterns are evidence about selection, not simple proof that every task follows a fixed parallel or serial process. Target similarity, display arrangement, expectations, and task difficulty can all affect search.

Perceptual-load theory proposes that demanding relevant tasks leave less capacity for irrelevant stimuli, whereas easier tasks may allow more distractor processing. In Lavie’s experiments, distractor interference was more evident under low-load than high-load conditions.

Together, listening and visual-search findings suggest that selection is flexible and depends on stimulus properties and task demands. Attention is especially important when a task requires combining features, while information outside the attended focus may still affect processing under some conditions.