A researcher asks volunteers to find one target among several shapes. On some trials, an irrelevant item has a color that stands out. The researcher records how quickly people respond when that conspicuous item is present and when it is absent. A slower response may suggest distraction, but the explanation depends on what the participants were looking for, what they had learned from earlier trials and how the display was arranged.
Jan Theeuwes’s 2025 review, Attentional Capture and Control, offers a way to read that experiment. This article is a critical guide to the paper and its place in a live scientific debate. It is distinct from the Library’s general attentional-capture entry: its subject is what this specific review argues, which evidence it integrates and where readers should hold the conclusion open.
The paper in its research setting
Published in the Annual Review of Psychology, Theeuwes’s article synthesizes work on visual attention rather than reporting one new experiment. It centers on the competition among current goals, physical salience and selection history. The abstract argues that proactive control of capture can arise mainly through implicit statistical learning associated with selection history. The review also acknowledges continuing controversy about when capture can and cannot be prevented.
A review can be authoritative in scope while still presenting an interpretation. It selects studies, organizes terminology and weighs competing explanations. Readers should therefore distinguish the paper’s framework from a single decisive test. The review is particularly useful because the older language of top-down versus bottom-up control can conceal what past experience contributes.
The paper concerns laboratory visual-search and attention tasks. It does not directly measure whether a person can resist a phone, choose a political view or remain attentive at work. Those applications may motivate research, but the route from a colored singleton in an experiment to a real digital environment must be argued and tested separately.
What counts as capture?
An object can be visually distinctive yet irrelevant to the task. “Attentional capture” names a claim that the object receives priority despite the observer’s current goal. Researchers can infer that priority from reaction-time costs, eye movements, probe performance or neural signals, but none is a transparent readout of attention in every circumstance. Different measures can answer different questions about timing and selection.
Imagine the target is a unique shape and the distractor is a unique color. If the color distractor slows responses, one account says its salience drew attention away. Another asks whether the participant was searching for any unique item, a strategy that would make the color relevant to the search procedure. Still another asks whether prior trials taught the observer that a particular location was likely to contain a distractor. The trial difference alone does not decide among these accounts.
Theeuwes’s review is most useful when read as a guide to those conditions. It does not say that every bright item always wins, nor that a deliberate goal can always shut salience out. Capture is a conditional result whose interpretation requires details of display, task and learning history.
Three sources of priority
The tripartite account distinguishes goal-driven influences, stimulus salience and selection history. A current goal can favor the feature or location a person seeks. A physically distinctive item can compete for selection. Past selections, rewards and statistical regularities can bias attention even when they are not part of the person’s present instruction. These influences can overlap on one trial.
Awh, Belopolsky and Theeuwes argued in 2012 that the simple top-down versus bottom-up dichotomy fails to account for a broad class of history effects. The 2025 review develops that line in the capture-and-control debate. A reader should not treat “history” as a synonym for conscious intention. Learning can alter priority without a person deciding in advance to ignore a particular spot.
The three-part language is a framework, not a set of three boxes into which every experiment fits neatly. A participant can carry a goal learned from practice, and a salient feature can also be task-relevant. The value of the framework is that it makes the experimenter ask what was held constant and what changed.
A classic disagreement about the observer’s set
Folk, Remington and Johnston’s 1992 experiments supported contingent attentional capture: an abrupt cue draws attention according to its relationship to the participant’s attentional control settings. That result challenged the idea that physical novelty automatically captures attention regardless of goal. Theeuwes’s earlier visual-search work emphasized conditions under which an irrelevant unique feature interferes despite a target task.
The disagreement is scientifically productive because the paradigms differ. Spatial cueing, feature search, singleton detection and the timing of a distractor can invite different strategies. If two papers use the same word “capture” while measuring different steps, their conclusions may appear more opposed than the data warrant. Conversely, a claim that all results can be explained by strategy risks becoming unfalsifiable unless the strategy is measured or constrained.
A good reading of the 2025 review asks which results are hard for each account. Can goal settings prevent selection of a salient but irrelevant item under some conditions? Can salience still produce a measurable cost under a specific feature-search strategy? Does a learned spatial regularity reduce that cost without awareness? The paper offers a map of these questions rather than a universal yes-or-no answer.
Suppression is part of the debate
Gaspelin and Luck’s review of inhibition argues that a salient distractor may generate a signal to attend while suppression prevents it from being selected. That proposal helps explain some experiments in which an item is distinctive but appears not to capture attention. It also changes what a null reaction-time cost can mean: no cost might reflect successful control, not a stimulus that never competed.
Suppression is inferred through converging methods, including probe tasks and electrophysiological components. These methods have their own assumptions. A lower probability of reporting a probe at a distractor location could indicate reduced priority, but researchers must consider how the probe interacts with the search task. A neural component can support a timing account without proving that every individual’s attention followed one exact route.
Theeuwes’s emphasis on selection history provides another explanation for proactive control. Repeated distractor locations can be down-weighted over trials. The two perspectives need not be mutually exclusive: learned statistics may shape a suppressive priority pattern. Their empirical disagreement concerns when, how and with which evidence suppression is established.
Statistical learning without a spoken plan
Wang and Theeuwes showed that statistical regularities about distractor location can modulate capture even when participants use a feature-search strategy. Their study supports the idea that repeated experience changes the competition for attention. It does not mean that a person can intentionally train away every unwanted stimulus in daily life.
Experimental designs make some regularities reliable and repeated. Real environments change. A phone alert may vary in sound, timing and social significance. Learning a stable laboratory location has no simple equivalent in a stream of messages that a person sometimes needs to answer. Translation requires specifying what regularity exists and what the user has an opportunity to learn.
The paper guide’s practical question is narrower than “how do we avoid distraction?” It is: what did repeated exposure teach the participant before the critical trial, and was that learning accounted for? Without that question, a result may be attributed to deliberate control or raw salience when history played a substantial role.
There is a particularly useful tension in the Wang and Theeuwes experiments. In one condition, the high-probability distractor location produced less interference than less frequent distractor locations. In a harder feature-search condition, a conventional capture cost was absent, yet target selection was slower at the location usually occupied by the distractor. The pattern supports learned location-specific suppression while showing why a single “distractor present” contrast can miss it. A reader should still inspect the paper’s trial exclusions, task demands and alternative explanation for faster responses on some distractor-present trials. The finding is informative precisely because its authors analyze those complications rather than treating one average latency as a direct picture of the mind.
Reading the evidence table
When evaluating a capture study cited by the review, start with the target definition. Is the participant searching for a known feature or for any item that differs from its neighbors? Next inspect the distractor: does it share a target feature, appear abruptly, occur at a predictable place or carry reward history? Then inspect the measure and comparison: what is slower or less accurate than what?
Look for trials without a distractor, neutral baselines and conditions that separate selection from later disengagement. A cost after a distractor appears could reflect initial capture, difficulty leaving it or an effect on decision or response. Eye tracking may show where gaze went; it does not reveal all covert shifts of attention. A reaction-time difference can be reliable while its mechanism remains disputed.
Finally, ask whether participants had enough trials to learn a regularity. An early-versus-late comparison may matter. So may awareness checks and individual variation. One average curve can hide different strategies. A transparent review states where the evidence supports a robust effect and where the mechanism depends on interpretation.
What the review changes
Theeuwes’s paper challenges two easy stories. One says bright stimuli seize attention regardless of our aims or experience. The other says a clear goal can prevent every distraction. The review presents a more constrained account: salience, goals and prior selection interact, and proactive control may be learned implicitly under some conditions. That is a substantive shift in how a researcher designs a test.
For educators and designers, the review is a caution against slogans about “attention hacks.” A feature that captures attention in one visual-search task may fail in another. A repeated cue may become less distracting as its location is learned, or remain consequential because it signals something important. The relevant behavior must be measured in the intended setting.
The paper also suggests an institutional lesson, which is the Institute’s synthesis rather than the author’s empirical claim: design environments where repeated experience helps people predict and manage interruptions. Consistent placement, meaningful alerts and control over irrelevant cues may reduce unnecessary competition. Whether those choices improve real work requires field testing.
Limits and unresolved questions
A review’s organization can make a field look more settled than its controversies are. Theeuwes explicitly notes lingering disputes about capture and control. Some concern the interpretation of suppression, the role of search mode and the temporal stage at which a distractor affects performance. New methods may refine the relative contribution of the three influences.
Publication bias and task selection also matter. Visual-search experiments often use simple arrays because they permit control. The simplicity strengthens causal inference about a tested display but limits generalization to rich scenes and personal priorities. The paper does not assign one “capture susceptibility” to every person across contexts.
Nor is every attentional shift a failure. A sudden sound can signal danger; a conspicuous item may be exactly what a person should notice. The ethical question of whether an interface ought to attract attention is separate from the empirical question of whether it can. This paper primarily addresses the latter.
What the evidence does not show
The review does not prove that attention is fully trainable, fully involuntary or governed by one universal mechanism. Laboratory capture effects do not establish a clinical diagnosis, a measure of willpower or a policy rule for all digital notifications. Its tripartite framework organizes evidence, but the weight of goal, salience and history must be estimated in a particular task.
A source guide should not turn the author’s conclusion into a fact about a person’s character. Someone distracted by an alert may have encountered a strong cue in a demanding context. Another may have learned to disregard a predictable irrelevant signal. Both patterns are compatible with a conditional science of attention. Neither licenses blame or a promise of effortless control.
Return to the search display
The researcher runs a second condition. The irrelevant color now appears often in one location, and participants gain experience with that pattern. The cost of the distractor may change. The change does not close every theoretical argument, but it shows why the first result could not be read from salience alone. History has entered the experiment.
Theeuwes’s 2025 review is valuable because it asks readers to keep all three forces in view. A distinctive item, a current goal and a learned pattern can compete within one act of looking. The careful reader leaves with better questions for the next experiment, not a slogan that attention is either captured or commanded.
Research references
- Theeuwes, J. (2025). Attentional capture and control. Annual Review of Psychology, 76, 251–273. View source
- Awh, E., Belopolsky, A. V., & Theeuwes, J. (2012). Top-down versus bottom-up attentional control: A failed theoretical dichotomy. Trends in Cognitive Sciences, 16(8), 437–443. View source
- Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception and Performance, 18(4), 1030–1044. View source
- Theeuwes, J. (1992). Perceptual selectivity for color and form. Perception & Psychophysics, 51(6), 599–606. View source
- Gaspelin, N., & Luck, S. J. (2018). The role of inhibition in avoiding distraction by salient stimuli. Trends in Cognitive Sciences, 22(1), 79–92. View source
- Wang, B., & Theeuwes, J. (2018). Statistical regularities modulate attentional capture independent of search strategy. Attention, Perception, & Psychophysics, 80, 1763–1774. View source