Two important words appear in a fast stream of letters. You catch the first. The second appears a fraction of a second later, and you miss it, even though it was on the screen and you were looking straight at it.
This laboratory pattern is called the attentional blink. No eyelid closes. The term describes a temporary reduction in the ability to report a second target after processing a first target in a rapid sequence.
What the attentional blink measures
In a common experiment, items appear one after another at the same location in a rapid serial visual presentation stream. Most are distractors. Two are designated targets, often called T1 and T2. The participant is asked to identify both. When T2 follows T1 after a short delay, report of T2 often falls compared with trials in which more time separates the targets.
Dux and Marois’s review describes the characteristic deficit when the second target appears roughly 200–500 milliseconds after the first, though the precise time course depends on the task. The outcome is usually examined as T2 accuracy conditional on correct T1 report. That condition matters: the effect concerns what happens after the first target was actually processed, not merely after an item happened to occupy the T1 position.
The experiment turns time into a controlled variable. It asks whether selecting one item for report changes the fate of another item arriving soon afterward. The answer is usually yes, but the mechanism remains an active topic of theory.
The original observation
Raymond, Shapiro and Arnell’s 1992 study asked participants to identify a partly specified letter in a rapid stream and then detect a later probe. Detection was impaired during a short interval after the target. The authors proposed that intervening items could interfere while identification of the first target was still being completed. Their title ended with a question mark: an attentional blink?
That question-mark history is useful. The phenomenon was named as a pattern requiring explanation, not as proof that attention literally switches off. Later experiments changed the targets, distractors, timing and instructions. They established a broad family of effects and also complicated the earliest account.
Many summaries say “the brain goes blind for half a second.” That is too strong. People can sometimes process aspects of a missed item, and the size and shape of the deficit vary with experimental conditions. Failure to report T2 is not identical to the absence of every perceptual or semantic process.
Why the first target matters
The first target creates work. The observer must select it from distractors, identify it and retain enough information to report it. Chun and Potter proposed a two-stage account in which rapid initial processing is followed by a more limited stage needed for stable identification. While T1 occupies that later stage, T2 may be vulnerable to interference from other items in the stream.
This model captures a central intuition: selection through time has constraints. But subsequent findings have challenged the idea that a single fixed capacity bottleneck explains every blink. Dux and Marois review several competing accounts involving target consolidation, distractor interference and control of the attentional set. Martens and Wyble likewise describe a theoretical landscape that changed as new experiments tested the boundaries of classic explanations.
The scientific job of the Learning Hub is to preserve the distinction between the robust experimental pattern and disputed mechanism. We can say a short-lag second-target report deficit occurs under many rapid-stream conditions. We should be more cautious about saying exactly what processing step “ran out.”
The surprise of lag-one sparing
A simple refractory-period account predicts that the closer T2 appears to T1, the worse T2 report should be. The data are more interesting. In many designs, a T2 that appears immediately after T1 is relatively well reported. This is called lag-one sparing. A deficit may be stronger at slightly later short lags.
That pattern is one reason the phrase “temporarily unavailable attention” can mislead. Immediate consecutive targets can sometimes be grouped or admitted into the same selection episode. The following distractor, and the structure of the stream, can affect whether the system continues selecting targets or changes state. Olivers, van der Stigchel and Hulleman’s experiments on “spreading the sparing” challenged a simple one-target-at-a-time resource account by showing that sparing can extend across consecutive targets under some conditions.
Lag-one sparing is not a universal escape hatch. It depends on the task and target sequence. Its value is theoretical: it forces models to explain why a very early second target can be easier than one arriving a little later.
Distractors are part of the mechanism
It is tempting to imagine T1 using all available resources and T2 arriving before those resources recover. But the items between targets matter. A post-T1 distractor can interrupt or mask target processing. Some experiments show that when a series of consecutive items all match the target category, the usual blink is altered or absent. That result supports the idea that losing control of a target-focused selection state, or allowing a distractor to enter at the wrong moment, may contribute.
This does not settle the debate in favor of one control theory. Limited processing and distractor interference can interact. The practical lesson for reading a study is to inspect the stream: What was shown after T1? How were targets defined? Was T2 masked? How long did each item appear? Did participants know in advance how many targets to expect?
Calling the effect an “attentional blink” without those details is like calling every missed message a memory failure. The name points toward a phenomenon; the design determines what can be concluded.
What a missed target can still do
Report is an important behavioral measure, but it does not capture all processing. Martens and Wyble review evidence that some blinked targets can influence later processing even when they are not reported accurately. Such findings complicate a simple all-or-nothing story about awareness. Their interpretation depends on the stimuli and methods, and they do not prove that all unnoticed items are fully understood.
This distinction matters beyond the laboratory because public discussions often slide from “not reported” to “not seen” to “not processed.” Those are different claims. A participant may have a weak, unstable representation of T2 that fails the experiment’s report criterion. Another may have no useful information about it. A careful account states the observed outcome before interpreting the hidden processing.
Nor does the attentional blink settle the general relationship between attention and consciousness. It gives a controlled way to study one boundary between rapid selection and reportable awareness. Broader claims require converging tasks and theories.
Measurement limits
A blink curve is a set of performance values across target separations. Its shape can change with presentation speed, target salience, distractor similarity, masking, practice, instructions and the requirement to report T1. Accuracy can also trade off between the two targets: a strategy that favors T1 may affect T2 performance.
When a study compares individuals, it needs enough trials to estimate each person’s curve reliably. A low T2 score in one short session is not a diagnosis of poor attention. Vision, reading familiarity, language, fatigue and task understanding may all affect performance. Even an apparently precise millisecond interval is meaningful only in the context of the stimulus sequence.
It is especially risky to use the attentional blink as a general “attention span” test. The paradigm concerns selection among two targets in a rapid stream, not how long someone can read, listen, work or care. Seconds of sustained focus and hundreds of milliseconds of temporal competition are different dimensions of attention.
What a stronger experiment would report
A transparent attentional-blink study reports more than one dramatic dip in a graph. It explains the presentation rate, the target-defining rule, the distractor set, the intervals tested and the number of trials at each interval. It distinguishes correct T1 identification from T2 detection and explains whether analyses of T2 are conditional on T1. It also reports the baseline against which a short-lag deficit is judged.
These details matter because a participant can use several strategies. If T1 is hard to identify, more effort may be devoted to it. If T2 has a distinctive color or category, it may be easier to detect despite short spacing. If every item after T1 looks target-like, the sequence may support continued selection rather than a conventional blink. The measured effect belongs to this configuration, not to time alone.
For replication or comparison across populations, researchers also need to show that the two groups understood the instructions and could perceive the stimuli under less demanding conditions. A group difference in blink magnitude can reflect different baseline performance, strategy or sensory conditions. It cannot be assigned automatically to a single neural bottleneck.
This is a broader lesson in cognitive science: a reliable experimental phenomenon can be useful precisely because its boundary conditions are informative. The cases where the blink weakens, shifts or disappears help distinguish theories. They should not be treated as exceptions to hide so the headline remains simple.
Is this the same as inattentional blindness?
No. Inattentional blindness generally concerns failure to notice an unexpected event while attention is engaged elsewhere in a task. The attentional blink concerns impaired report of a second target following a first target in a rapid sequence, often when both are task-relevant. Change blindness concerns another pattern: failure to detect a change across views under certain conditions.
These phenomena all challenge the intuition that looking guarantees seeing, but their procedures and explanations differ. Merging them into one claim that “humans are blind most of the time” loses the very distinctions that make the research useful.
Application without overclaiming
Interfaces sometimes present critical items in rapid succession: alerts, captions, symbols, instructions or dashboard changes. The attentional-blink literature gives designers a reason to test whether two consecutive targets can both be identified when timing is tight. It does not offer a universal minimum spacing rule for every interface, because the laboratory effect depends on target and distractor structure.
Consider an operator who must read an alarm and then notice a second change almost immediately. If the second event is missed, it may be tempting to blame the operator. A better design test would vary timing, salience, signal order and task load, and measure whether the operator can act on both events. The institute’s synthesis is that important information should not require a person to win an avoidable competition through time. That is a design hypothesis, not a direct result of a specific rapid-letter experiment.
Education provides another caution. A lecture slide that changes too quickly can leave students behind, but the attentional blink alone does not explain their learning. Comprehension, prior knowledge, note-taking and memory all matter. The paradigm can inspire timing questions without becoming a catch-all theory of teaching pace.
When the first target changes the question
The attentional blink is often described as a failure to see the second target, but the first target’s identity and task requirements can change what failure means. Naming a letter, deciding whether a word belongs to a category and remembering a target for later report impose different demands. A comparison across such tasks may alter the period of reduced second-target report without showing that one universal attentional gate has opened or closed.
A strong experiment therefore keeps the stream and report requirements explicit. It asks whether an apparent improvement came from a different allocation of processing, a less demanding first target, or a change in how responses were scored. The practical lesson is methodological: the shape of the blink is evidence about the designed sequence, and theory must explain that sequence rather than using the task name as its own explanation.
What the evidence does not show
The evidence does not show that attention completely shuts down after every important event. It does not show that any two real-world signals arriving within 500 milliseconds will be missed. It does not establish one universally accepted mechanism for the blink, and it does not turn a short RSVP task into a measure of general ability or character.
It also does not show that a missed T2 had no processing at all, that lag-one sparing always occurs, or that making all information slower will always improve performance. The effect belongs to a structured task. Its value lies in showing how selection across time depends on the relation among targets, distractors and reporting demands.
The second word
Return to the rapid stream. The first word matters, so you hold it. The second appears while the first is still being made reportable. Whether it survives depends on timing and the stream around it.
A fair comparison keeps the first target’s difficulty, the intervening items and the report rule visible. If a redesigned stream improves second-target reports, the next question is which ingredient changed. A longer interval, a weaker distractor and a less demanding first report could each improve performance for different reasons. Naming the ingredient is what makes the result useful beyond one demonstration.
The attentional blink gives a precise way to ask a human question: what can be lost when one important event arrives too close to another? The answer is not that attention is broken. It is that meaningful selection takes time, and the design of a sequence can decide which moment gets a fair chance to be noticed.
Research references
- Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visual processing in an RSVP task: an attentional blink? Journal of Experimental Psychology: Human Perception and Performance, 18(3), 849–860. View source
- Chun, M. M., & Potter, M. C. (1995). A two-stage model for multiple target detection in rapid serial visual presentation. Journal of Experimental Psychology: Human Perception and Performance, 21(1), 109–127. View source
- Olivers, C. N. L., van der Stigchel, S., & Hulleman, J. (2007). Spreading the sparing: Against a limited-capacity account of the attentional blink. Psychological Research, 71, 126–139. View source
- Dux, P. E., & Marois, R. (2009). How humans search for targets through time: A review of data and theory from the attentional blink. Attention, Perception, & Psychophysics, 71(8), 1683–1700. View source
- Martens, S., & Wyble, B. (2010). The attentional blink: Past, present, and future of a blind spot in perceptual awareness. Neuroscience & Biobehavioral Reviews, 34(6), 947–957. View source