The phone is not on the table.
The hand reaches toward its usual place anyway.
Nothing vibrated. No notification appeared. The current goal is elsewhere.
For a fraction of a second, history has entered attention before the present has supplied a reason.
Selection history refers to the influence of previous attentional selections and learning on future attentional priority. What has been selected, rewarded, ignored, suppressed or repeatedly encountered can change what attention is likely to prioritize later.
This idea transformed modern attention theory because it explained effects that the traditional top-down/bottom-up distinction could not.
The missing third source
For decades, attentional control was often organized around two categories.
Top-down attention followed current goals.
Bottom-up attention followed stimulus salience.
Awh, Belopolsky and Theeuwes argued in 2012 that this dichotomy was insufficient. Some selection biases are neither explained by what a person currently wants nor by what is physically most conspicuous.
Past selection can become a force in its own right.
Their proposal helped establish a tripartite framework: current goals, physical salience and selection history.
By 2021, Anderson and colleagues could describe the previous decade as a shift from dichotomy to trichotomy and offer a detailed review of the many forms selection history can take.
This is not a decorative addition to the theory.
It changes what attentional behavior means.
Reward can teach attention what to notice
Value-driven attentional capture is one of the clearest examples.
Anderson, Laurent and Yantis trained participants so that particular visual features were associated with reward. Later, those features could capture attention even when the reward was gone and the features were irrelevant to the new task.
The effect demonstrates a form of learned priority.
The stimulus does not need to be the current goal. It does not need to be the most physically salient object.
Its history matters.
This is why attention cannot be understood from a snapshot.
The visible scene is only part of the causal field.
History can teach suppression too
Selection history is not only about learned attraction.
People can also learn where distractors tend to occur and become better at suppressing those locations. Repeated statistical structure can guide attention without requiring explicit conscious knowledge of the regularity.
Work by Wang and Theeuwes, among others, has shown that learned distractor suppression can arise from statistical regularities.
This complicates the idea that suppression is always an active top-down act.
The system can learn where not to spend attention.
In practical terms, history can create both grooves and guardrails.
Attentional habits are not identical to behavioral habits
It is tempting to translate selection history directly into the language of habit.
There is overlap, but the terms should not be collapsed.
A behavioral habit usually refers to learned action tendencies elicited by contexts, often discussed in relation to reinforcement and automaticity. Selection history concerns learned biases in attentional priority.
The two can interact.
A notification cue may capture attention, which makes checking behavior more likely. Repeated checking may strengthen associations with the cue. But the attentional and behavioral components are not the same mechanism.
Maintaining this distinction helps the Knowledge Library connect attention research with habit science without pretending they are one field.
The moral significance of history
Selection history changes the language of personal responsibility.
If attention were governed only by current goals, distraction would be easy to interpret as weak intention.
But a person can sincerely hold one goal while previous reward history pulls attention elsewhere.
This does not erase agency.
It makes agency temporal.
People act with systems that have been trained by previous environments.
The question becomes not only "What do I want to attend to now?" but "What has repeatedly been teaching my attention what to prioritize?"
That is a more humane and more scientifically defensible question.
Digital systems can participate in selection history
Digital environments create repeated attentional contingencies.
A badge predicts social information. A sound predicts a message. A feed rewards the next swipe unpredictably. A work platform trains urgency around certain visual markers. A personalized system learns which cues reliably produce re-engagement.
It would be an overclaim to say that every digital cue produces a durable selection-history effect equivalent to a laboratory reward paradigm.
The contexts differ enormously.
But the evidence makes one general question legitimate: repeated selection and reward can alter future priority, so design histories deserve attention.
The unit of concern is not one notification.
It is the repeated relationship between cue, selection and consequence.
Selection history in expertise
History is not merely a source of distraction.
Experts also develop useful attentional biases.
A radiologist learns what features deserve inspection. A musician hears timing deviations that a novice misses. A craftsperson sees material irregularities before they can fully explain them.
These abilities involve many mechanisms beyond selection history, including perceptual learning, conceptual knowledge and motor expertise.
Still, the broad principle holds: repeated selection changes future selection.
The same learning architecture that can make a reward cue difficult to ignore can make subtle domain-relevant information easier to detect.
The ethical question is not whether attention should have history.
It inevitably does.
The question is what histories we are constructing.
Selection history and institutions
Institutions create repeated patterns of attention.
A company teaches employees which dashboard indicators matter by repeatedly tying them to evaluation and consequence. A school teaches students which kinds of questions receive reward. A newsroom trains sensitivity to particular forms of novelty. A bureaucracy teaches which fields on a form become actionable and which lived realities remain administratively invisible.
Calling this "selection history" at institutional scale is an analogy, not a claim that organizations have a unitary human cognitive mechanism.
The useful synthesis is that repeated reward and selection structures shape the attentional habits of the people inside them.
Culture becomes partly visible in what people learn to notice without being told each time.
AI creates adaptive histories
AI systems add a new layer because the environment can now adapt to the user while the user is adapting to the environment.
A recommender learns which content receives attention. The user learns what kinds of cues predict reward. The system then changes what it presents.
This is a coupled learning loop.
Again, laboratory selection-history findings do not by themselves establish the long-term cognitive effects of recommender systems.
But they provide a disciplined reason to study the interaction rather than treating each exposure as independent.
An AI-mediated environment can have memory.
So can attention.
Can selection history be rewritten?
Learned biases can change.
Reward-driven capture can extinguish or weaken under some conditions. Distractor suppression can be learned. New task demands can create new priority structures.
But change is not always immediate.
Some learned attentional biases persist after their original usefulness has disappeared.
This is one reason changing an environment can feel easier than changing the attention that was trained inside it. Remove the cue, and the hand may still reach toward where it used to be.
Practice architecture should therefore include time.
A new intention is not always enough to erase an old selection history.
Measurement problems
Selection history is inferred through experimental manipulations of reward, repeated target selection, distractor probability, priming and statistical regularity.
Different forms of history may rely on different mechanisms.
Reward learning is not identical to repetition priming. Learned distractor suppression is not identical to target-value associations. Some effects are explicit; others emerge without reportable awareness.
The umbrella term is useful because it identifies a class of influences that are not reducible to current goals or physical salience.
It should not imply that every historical effect is one process.
What the evidence does not show
Selection-history research does not show that past experience rigidly determines present attention.
It does not show that learned biases cannot be changed.
It does not show that every repeated digital behavior produces measurable attentional capture.
It does not show that reward history is the only form of selection history.
It does not show that selection-history effects are always unconscious.
It does not show that current goals become irrelevant.
And it does not show that an attentional bias is inherently harmful. Learned priority is central to expertise, navigation and adaptive behavior.
Selection history is not one kind of history
The umbrella includes several distinct routes by which the past can bias attention.
Reward history can increase priority for features associated with value. Repetition can prime previously selected targets. Statistical learning can make likely target locations easier to prioritize or likely distractor locations easier to suppress. Search history can alter how efficiently a current display is scanned. Learned associations can connect otherwise neutral cues with significance.
These effects can differ in duration, awareness and flexibility.
Some are short-lived carryover from the immediately previous trial. Others persist across many trials or into a new task context. Some depend on explicit knowledge. Others appear even when participants cannot report the regularity that influenced them.
This diversity is why selection history should not become a new catch-all explanation.
The phrase identifies where to look for cause: in previous episodes of selection and learning. It does not specify the mechanism until the history itself is defined.
History can conflict with values without being pathology
A person may repeatedly orient toward information they no longer endorse.
The former job's email icon still raises vigilance after leaving the job. A phrase associated with an old relationship catches attention years later. A notification sound produces anticipatory checking even after the app has ceased to matter.
These examples are plausible illustrations, not diagnoses and not direct demonstrations that a specific laboratory selection-history mechanism is responsible.
They are useful because they show the human meaning of temporal mismatch.
Learning can outlast the context that made it adaptive.
This is common across many systems. Stress responses, habits, expectations and motor skills can all persist after circumstances change. Selection history is the attentional version of a broader developmental truth: plasticity does not guarantee immediate updating.
That gives practice an ethical tone different from self-correction.
The aim is not to shame the old bias for surviving.
It is to create enough repeated new evidence that the system can learn the present.
Forgetting is part of attentional health
Research often asks how learning creates bias. Human functioning also depends on letting obsolete biases weaken.
Extinction, changing contingencies, new environments and repeated non-reward can alter priority. But old learning can sometimes reappear under stress or context change.
A complete science of selection history therefore needs both acquisition and updating.
For technology design, this raises a practical question rarely asked: if a system spends years training a cue-response relationship, what responsibility does it have to support disengagement when the user wants to leave?
That question is normative. The research does not already provide the answer.
But it follows directly from taking learning history seriously.
Selection history and individual difference
People do not share the same attentional history.
Two users can encounter the same interface and experience different pull because their previous rewards, habits, expertise and associations differ. A cue that is neutral to one person may be highly charged for another.
This makes population averages useful but incomplete.
Designers should be cautious about assuming that one cue has one attentional effect. Researchers should distinguish group-level effects from prediction about individuals.
For Inner Technology, the diversity of selection history is a reason to prefer adaptable practice architectures over universal prescriptions.
The update problem
The most important capacity may not be eliminating history but updating it.
A learning system needs continuity or expertise would never accumulate. It also needs flexibility or obsolete priorities would dominate indefinitely.
Attention therefore participates in a larger balance between stability and plasticity.
Too little stability and nothing becomes skilled. Too little updating and yesterday's importance becomes today's automatic distraction.
The science of selection history gives that balance an empirical foothold without pretending that it can explain every form of psychological persistence.
Attention has a biography
The empty patch of table still receives the hand for a moment.
That movement is small enough to miss.
But it carries a large correction to how human attention is often imagined.
Attention is not only a beam aimed from the present.
It has a biography.
What has repeatedly mattered leaves traces in what can matter quickly again.
For human development, that is both vulnerability and possibility.
We can be trained by environments we did not choose.
We can also build environments, practices and histories that make better forms of noticing easier to inhabit.
Selection history is the science of that accumulated leaning.
Research references
- Awh E, Belopolsky AV, Theeuwes J. Top-down versus bottom-up attentional control: a failed theoretical dichotomy. Trends in Cognitive Sciences. 2012;16(8):437-443. https://doi.org/10.1016/j.tics.2012.06.010
- Anderson BA, Kim H, Kim AJ, Liao MR, Mrkonja L, Clement A, Grégoire L. The past, present, and future of selection history. Neuroscience & Biobehavioral Reviews. 2021;130:326-350. https://doi.org/10.1016/j.neubiorev.2021.09.004
- Anderson BA, Laurent PA, Yantis S. Value-driven attentional capture. Proceedings of the National Academy of Sciences. 2011;108(25):10367-10371. https://doi.org/10.1073/pnas.1104047108
- Anderson BA, Yantis S. Persistence of Value-Driven Attentional Capture. Journal of Experimental Psychology: Human Perception and Performance. 2013;39(1):6-9. https://doi.org/10.1037/a0030860
- Wang B, Theeuwes J. How to inhibit a distractor location? Statistical learning versus active, top-down suppression. Attention, Perception, & Psychophysics. 2018;80(4):860-870. https://doi.org/10.3758/s13414-018-1493-z
- Theeuwes J. Attentional Capture and Control. Annual Review of Psychology. 2025;76:251-273. https://doi.org/10.1146/annurev-psych-011624-025340