Flanker Task: Measuring Attention and Distractor Interference

Author: Angela Bogdanova · Published: October 16, 2026 · Editorial Policy

Research cluster: Attention, Perception & Cognitive Control · Written by Artificial Sapiens

What Is the Flanker Task?

The flanker task is an experimental procedure for studying how task-irrelevant information influences the selection of a task-relevant response. On each trial, a participant identifies a designated target, usually located in the center of a small visual array, while attempting to disregard nearby stimuli called flankers. The flankers can signal the same response as the target or a competing response. Differences in speed and accuracy between these conditions reveal how information outside the current focus of attention can influence action selection. The classic paradigm was introduced by Barbara A. Eriksen and Charles W. Eriksen in 1974. Eriksen and Eriksen (1974).

The key measurement is the flanker compatibility effect, often called the flanker interference effect or congruency effect. In a typical two-choice task, participants respond more slowly, and sometimes less accurately, when the distracting flankers point toward the incorrect response than when flankers and target indicate the same response. This contrast makes the task an important tool for experimental research on selective attention, response competition, and cognitive control. The precise contribution of each mechanism depends on stimulus design, task instructions, and the analysis applied. Ridderinkhof et al. (2021).

A flanker trial is a controlled measurement occasion, not a miniature clinical examination. Observing an interference effect establishes a relation between a stimulus manipulation and performance under specified conditions. It does not, by itself, identify a disorder, reveal a person's general intelligence, or quantify an enduring capacity to concentrate in everyday life. Individual assessment requires evidence about reliability, validity, appropriate reference groups, and the intended interpretation of a score. Hedge et al. (2018).

How the Eriksen Flanker Task Works

The target and its distracting neighbors

A common version presents five arrows horizontally and instructs the participant to indicate the direction of the middle arrow. For example, in the array “→ → → → →,” the center target points right and the surrounding arrows support the right response. In “← ← → ← ←,” the center still points right, but the four neighboring arrows indicate left. Only the center arrow determines the correct answer in either case.

The first array is congruent because the flankers and target map onto the same response. The second is incongruent because the flankers activate a competing response. Depending on the protocol, investigators may also include neutral flankers, such as non-directional symbols, or a target-alone baseline. Neutral conditions help separate the cost of incompatible flankers from possible benefits of compatible ones, although a supposedly neutral stimulus can still change visual grouping, crowding, salience, and processing demands. Ridderinkhof et al. (2021).

Participants typically receive an instruction to be both fast and accurate. Each trial may begin with a fixation cross, followed by the stimulus array, the recorded keypress, and a brief interval. Many trials are required because a single keypress reflects motor preparation, attention, response policy, visual discrimination, and incidental variability as well as the influence of flankers. A valid inference depends on comparing predefined conditions across sufficiently many observations.

What congruent, incongruent, and neutral mean

Congruence refers to the relationship between the task-relevant target and the irrelevant context in a particular experimental design. In the standard arrow task, direction and response agreement are usually aligned. In letter variants, different letters may map to the same response key. For example, if both H and K require a left response, flankers made of K can be response-compatible with a central H even though the letters look different. This distinction separates physical identity, perceptual similarity, and response compatibility.

Neutral does not automatically mean the complete absence of distraction. A row of neutral symbols still contains visual energy and spatial structure, and it may alter target discrimination. Researchers therefore specify what the neutral stimulus controls, rather than assuming a universal ordering of neutral, congruent, and incongruent trials. The original experiments directly distinguished response-compatible noise, response-incompatible noise, and other kinds of neighboring letters. Eriksen and Eriksen (1974).

A single trial, step by step

The participant first establishes the instruction: respond only to the middle symbol. The array then appears. Early visual information may activate responses associated with more than one symbol, even though one response is formally relevant. The participant selects the appropriate target response and presses the mapped key. The experiment records reaction time and whether that choice was correct. The sequence repeats with target direction, congruence, and often other features varied across trials.

A hypothetical result makes the principle clear. Suppose mean correct reaction time is 470 milliseconds on congruent trials and 515 milliseconds on incongruent trials. The illustrative reaction-time interference effect is 45 milliseconds, calculated as 515 minus 470. These numbers are an explanation of the arithmetic, not a benchmark, normative score, or empirical estimate for a particular population.

Where the Task Came From: The Original Experiments

The letter paradigm of 1974

Barbara A. Eriksen and Charles W. Eriksen developed their paradigm within research on visual selection. Participants identified a target letter while surrounding letters varied in proximity and in whether they suggested the same or an alternative response. The important finding was that response-incompatible flankers interfered more than flankers assigned to the same response. Nearby irrelevant letters could therefore influence response selection despite explicit knowledge of which letter mattered. Eriksen and Eriksen (1974).

The researchers also varied the spacing of the letters. Generally, interference declined as the distance between target and irrelevant letters increased. This observation mattered because it linked the spatial organization of the display to the processing of information that participants were told to ignore. Subsequent work emphasized that perceptual processing can proceed sufficiently far to influence competing response tendencies. Eriksen and Schultz (1979).

Continuous flow and the emergence of arrow flankers

In a continuous-flow account, evidence about visual stimuli accumulates over time and can begin activating possible responses before visual identification is complete. Eriksen and Schultz developed this account partly to explain how irrelevant information affects response preparation rather than appearing only as an isolated early perceptual delay. Eriksen and Schultz (1979).

Later versions replaced letters with familiar directional arrows. Arrows offered a relatively intuitive stimulus–response relationship and allowed researchers to vary such factors as direction, timing, response mapping, and flankers independently. Historical reviews trace how the arrow task supported work on response activation, suppression, development, and the timing of action control. Ridderinkhof et al. (2021).

What the Flanker Effect Measures

The most familiar index is the average difference between incongruent and congruent reaction times on correct trials:

Flanker RT effect = mean correct RT on incongruent trials − mean correct RT on congruent trials.

Researchers may also compute an accuracy effect, such as error percentage on incongruent trials minus error percentage on congruent trials. When a neutral condition is available, they may estimate an incongruent cost relative to neutral and a congruent benefit relative to neutral. These contrasts answer different questions; authors should report their definitions explicitly.

A positive average reaction-time difference indicates slower correct responses in the incompatible condition. In tightly controlled experiments, that difference is evidence of an effect of flankers on performance. It is best interpreted as an observed behavioral contrast that may combine target selection, competing response activation, strategic adjustment, decision threshold, and motor execution. The label “inhibitory control score” alone cannot isolate any one of those mechanisms. Ridderinkhof (2002); Molloy et al. (2026).

Reaction time and accuracy must be read together

A participant can become faster by accepting a higher error rate or more accurate by responding cautiously. Consequently, interpreting only mean speed risks confusing improved target selection with a changed speed–accuracy policy. Analyses should show the number of trials, mean or distribution of correct latencies, error frequencies, excluded responses, and the handling of outliers or missed deadlines.

Very low error rates are not proof that flankers had no effect; interference may appear primarily in reaction time. Equally, a larger error difference need not imply the same kind of cognitive change as a larger reaction-time difference. Models of response selection use joint patterns of latency and accuracy to evaluate candidate explanations rather than reducing all observations to one subtraction. White et al. (2011).

Why one large interference score has several possible meanings

An effect can grow because incompatible distractors activate a competing response more strongly, because the relevant target is harder to discriminate, because response decisions are more cautious, or because the timing of relevant and irrelevant activation changes. Some of these possibilities predict distinct response-time distributions, error patterns, or effects of stimulus manipulations. Thus, an increased score cannot automatically be translated into “weaker attention” or “poorer self-control.”

This is a central question of construct validity: does the chosen experimental contrast reflect the psychological property that a researcher wants to infer? A result can be well established as a within-person experimental effect while still being unsuitable for ranking people by a stable trait. Hedge et al. (2018).

The Main Psychological Mechanisms

Selective attention and spatial competition

Selective attention prioritizes information relevant to the task. In the flanker task, knowing the target location does not always eliminate the processing of nearby stimuli. Stimulus proximity, relative size, contrast, and attentional focus can all influence how much irrelevant information reaches processes involved in choosing an answer. The original letter experiments and subsequent manipulations support the claim that the task measures consequences of incomplete selection under defined visual conditions. Eriksen and Eriksen (1974); Ridderinkhof et al. (2021).

This does not mean that attention is an all-or-none spotlight whose position can be read directly from an interference score. Spatial selection evolves over time, and observers adapt their priorities to target probability, cueing, expected distractors, and response demands. These influences have to be controlled or separately modeled if the goal is to study a specific attentional process.

Competing responses and partial activation

In incompatible trials, surrounding information may activate a response that contradicts the instructed target response. Experiments using measures beyond the final keypress have provided evidence of competing response activation. The concept of response competition explains why irrelevant flankers associated with a different response can have stronger effects than visually different flankers assigned to the correct key. Eriksen and Schultz (1979); Ridderinkhof et al. (2021).

Response competition need not involve a fully conscious temptation or an explicit decision to do the wrong thing. The relevant scientific claim concerns measurable activation of response tendencies and their timing in a speeded task. Researchers can investigate those tendencies through experimental manipulations and, in specialized studies, electrophysiological or muscular measures.

Response capture, suppression, and time course

One influential account distinguishes an initial tendency for irrelevant features to activate an incorrect response from subsequent selective suppression of that tendency. In some flanker experiments, particularly fast incongruent responses carry a higher risk of errors, consistent with early response capture. Slower responses can show a different interference profile, suggesting that the time available to resolve competing activation matters. Ridderinkhof (2002); Ridderinkhof et al. (2021).

Researchers examine this timing with conditional accuracy functions and distributional analyses rather than relying only on one grand mean. The interpretation of a particular shape remains model-dependent, and not every flanker variant has the same temporal signature. Activation–suppression is therefore a productive explanatory framework to test, with predictions that require appropriate data.

Conflict monitoring and adaptive adjustment

Some theories propose that detecting competition between possible responses changes the amount of control applied on subsequent trials. A participant may respond differently after an incongruent trial than after a congruent one; this sequential congruency effect has often been interpreted as an adjustment in conflict control. Gratton et al. (1992).

There is a substantial competing explanation: repetitions of stimuli, responses, and learned contingencies can produce apparent sequential adaptation even without a separate conflict-triggered executive adjustment. Experiments isolating such feature repetitions challenged simple interpretations of trial-to-trial congruency sequences. A strong design separates these explanations through counterbalancing, stimulus sets, and preregistered contrasts. Mayr et al. (2003).

Experimental Variations and What They Change

Letters, arrows, and non-directional stimuli

Letters require learned assignment of categories to response keys. Arrows capitalize on familiar directional associations. Picture, color, and numerical variants introduce still other perceptual and semantic properties. A task's stimulus material influences the automaticity of associated responses, target discriminability, possible stimulus repetitions, and how much visual crowding can occur. Findings should therefore be attributed to the tested version rather than presumed identical across every paradigm. Ridderinkhof et al. (2021).

Target–flanker spacing, onset, and visibility

Moving flankers farther from the target typically changes interference; so can changing their visibility, size, onset relative to the target, or duration. The same physical array can therefore yield different performance depending on how quickly information becomes available and how selectively the target can be processed. Such manipulations are valuable for locating when and how distractor information affects decisions. Eriksen and Eriksen (1974); Eriksen and Schultz (1979).

Probabilities, expectations, and block structure

The proportion of congruent trials may influence strategies and expectations. Blocking trials by condition versus mixing them changes what participants can anticipate. Repeated arrays can support learning about which stimuli predict which responses. Trial order, target balance, and response-key mapping are therefore components of the causal design and need transparent reporting. Sequential adaptation should be tested against repetition and contingency accounts rather than being assumed from a single interaction. Gratton et al. (1992); Mayr et al. (2003).

Attention Network Test and related paradigms

The Attention Network Test combines cueing manipulations with a flanker component to examine alerting, orienting, and executive conflict effects within one experimental protocol. Its executive network contrast draws on incongruent versus congruent flankers, while the other conditions target different functions. These network scores have specific experimental definitions, and their reliability and independence must be evaluated rather than inferred solely from their names. Fan et al. (2002).

Competing Models of Flanker Performance

A discrete-stage interpretation treats identification, response selection, and motor execution as relatively separable stages. Continuous-flow approaches emphasize that partial information about target and flankers can activate candidate responses in parallel before a final decision. Later dual-route accounts distinguish more direct, stimulus-triggered response activation from task-directed selection. Each model makes assumptions about which information becomes available when. Eriksen and Schultz (1979); Ridderinkhof et al. (2021).

Computational sequential-sampling models ask how evidence for possible choices accumulates until a decision criterion is reached. Models differ in whether selective attention changes abruptly or gradually and in how they represent time-varying distraction. A model can reproduce mean interference while failing to capture fast errors, full response-time distributions, or individual variability; adequacy requires confronting all of these features with data. White et al. (2011).

Recent joint-modeling research suggests that latent efficiency in extracting task-relevant evidence may explain meaningful between-person variance across conflict paradigms, while the contrast between congruent and incongruent mean reaction times can miss important components of that variation. These are model-based results that depend on assumptions and the populations and tasks sampled. They refine the interpretation of performance without turning a single mechanistic proposal into settled fact. Molloy et al. (2026).

How Researchers Analyze a Flanker Experiment

A defensible analysis starts by specifying the primary comparison and the unit of inference. Investigators should state whether the objective is the average within-participant congruency effect, changes produced by an intervention, or stable differences between individuals. Those goals demand different designs. Randomizing or counterbalancing target and distractor assignments, using enough trials per condition, and recording exclusions are foundational to each approach.

For mean comparisons, investigators commonly compute participant-specific condition summaries and use an appropriate repeated-measures or hierarchical model. Trial-level models can account for the fact that responses from the same person are related. Error analysis is separate and often uses proportions or models for binary accuracy rather than applying ordinary linear assumptions automatically to every outcome.

A careful report identifies how anticipatory responses, missed responses, incorrect trials, post-error trials, practice blocks, and extreme latencies are handled. It also states whether results change under reasonable alternative preprocessing choices. Reported confidence intervals or posterior uncertainty should match the intended inference. A difference that reaches statistical significance in a large sample need not have a large magnitude or strong measurement utility.

Reaction-time distributions offer information lost in simple means. Conditional accuracy functions, quantile analyses, and process models may help discriminate early response capture from delayed suppression or changed decision thresholds. However, adding a sophisticated model does not guarantee a reliable psychological trait; model fit, parameter recovery, prior assumptions, and independent replication remain important. White et al. (2011); Donzallaz et al. (2026).

Reliability, Validity, and the Reliability Paradox

A strong laboratory effect can make a weak individual test

The flanker task often produces a robust average interference effect across a group. Yet a measure intended to rank people by stable attentional control needs meaningful between-person differences that can be recovered consistently. The same task can satisfy the first criterion while falling short on the second. Hedge, Powell, and Sumner demonstrated this reliability paradox across several classic experimental paradigms, including the Eriksen flanker task. Hedge et al. (2018).

A difference score subtracts two estimates that each contain noise. When participants have similar true interference effects, measurement noise can dominate their ranking even while the average effect remains clear. The resulting test–retest reliability may be much lower than readers expect from a famous, repeatedly replicated experimental demonstration. Reliability is an empirical property of a particular score, protocol, sample, and testing context.

What can improve measurement

Increasing trial counts may reduce random measurement error, but benefits eventually diminish. More carefully calibrated stimuli and designs that magnify informative between-person variation can sometimes produce more reliable measures. Across five experiments, Kucina and colleagues found that specific modified conflict tasks, including a flanker variant, improved the reliability of individual-difference estimates relative to benchmark designs; the improvement depended on those modifications and does not certify every flanker implementation. Kucina et al. (2023).

Hierarchical models, latent factors, and cognitive process models can help when their assumptions fit the data and their parameters are estimable. A 2026 comparison of alternative models in Stroop datasets found that greater modeling sophistication did not consistently improve signal-to-noise ratios; those results are informative for conflict-task measurement generally but should not be misrepresented as direct flanker validation. Donzallaz et al. (2026).

Group effects, individual differences, and clinical inference

Three distinct statements require different evidence: incongruent flankers slow average responses in an experiment; a particular person's score is reliably different from other people's scores; and that score validly predicts functional impairment or supports a clinical judgment. The first does not establish the second or third. A flanker score is an experimental task score. Distractibility may be a symptom in some clinical contexts; a diagnosed disorder is established through appropriate clinical methods and criteria.

The task appears in studies of development, aging, neurological conditions, and psychiatric populations, but conclusions depend on the population, design, measurement properties, and the explanatory question. Differences between groups do not justify diagnosing any individual from a single reaction-time contrast. Ridderinkhof et al. (2021); Hedge et al. (2018).

Flanker Task vs. Stroop, Simon, Go/No-Go, and Stop-Signal Tasks

The flanker and Stroop tasks both study interference from task-irrelevant information. In a typical flanker trial, distractors are distinct neighboring objects; in the classic Stroop task, the irrelevant word and relevant ink color are dimensions of the same object. Their behavioral congruency effects can both reflect response competition, but differences in stimulus structure, practiced associations, and processing routes complicate direct equivalence. A detailed explanation of the Stroop phenomenon appears in The Stroop Effect: Why Conflicting Information Slows Responses.

The Simon task typically examines whether the irrelevant spatial location of a target corresponds to the required response location. Go/No-Go tasks examine withholding responses on designated trials, whereas stop-signal tasks investigate the cancellation of a response that has already been initiated after a later stop signal. The flanker task primarily manipulates competition between information mapped to different choices. These operational distinctions matter when a study uses the broad term inhibition.

Neither behavioral similarity nor differences in labels establish that all these tasks measure one unitary ability. Executive-function research emphasizes overlapping and distinguishable control processes, and cross-task studies show that associations among conflict effects depend on how performance is measured. Miyake and Friedman (2012); Molloy et al. (2026).

What Flanker Results Say About Everyday Attention

The experiment offers a highly controlled illustration of a familiar general problem: goal-relevant information must be selected while competing signals are present. That makes the task valuable for developing mechanisms and testing hypotheses about attentional selection and action. It does not mean that a participant's millisecond interference score directly measures success in driving, studying, working, or navigating digital distractions.

Everyday environments involve changing goals, motivation, multisensory input, interruptions, long time scales, and consequences for error. A brief visual two-choice paradigm samples only a restricted portion of that complexity. Generalization to practical functioning requires separate evidence using appropriate outcomes and ideally convergent methods. The broad executive-function framework is explained further in Executive Functions: Inhibition, Working Memory, and Flexibility.

Repeated practice on a flanker task can make a person more familiar with its stimuli and response requirements. Such improvement should be evaluated as learning on the trained procedure unless well-designed studies demonstrate transfer to untrained tasks or meaningful everyday functions. A claim of improved general attentional control needs direct evidence rather than an inference from faster practiced responses.

Important Limitations and Open Questions

One unresolved issue concerns which stage of processing produces interference in a given task version. Response competition is supported by classic experiments, but stimulus identification, visual grouping, temporal changes in attention, and choice thresholds can also contribute. Models that fit average congruency effects may disagree about the mechanisms behind latency distributions. Researchers continue to compare these accounts empirically. White et al. (2011); Ridderinkhof et al. (2021).

A second issue is the interpretation of sequential congruency effects. Apparent adaptation following conflict may reflect changes in control, perceptual learning, response repetition, or combinations of these. A convincing study distinguishes these predictions instead of equating trial history with an executive mechanism. Gratton et al. (1992); Mayr et al. (2003).

A third issue is measurement quality. Researchers increasingly distinguish experimentally robust manipulation effects from reliable individual differences and evaluate modified paradigms, hierarchical analyses, and latent mechanisms. This area is active, and a promising model or optimized task needs replication outside its development samples. Kucina et al. (2023); Donzallaz et al. (2026); Molloy et al. (2026).

Finally, experimental accessibility and sampling shape the evidence base. Visual acuity, familiarity with arrow symbols, response-device timing, motor differences, practice effects, age, fatigue, instructions, language, and cultural context can influence performance. Investigators should describe those conditions and validate task adaptations for populations in which the results will be interpreted.

Frequently Asked Questions

Is the flanker task a test of attention?

It is a research task used to examine consequences of selective attention and competing response activation. Whether a specific score validly measures stable attentional ability depends on the intended construct and the psychometric evidence for that implementation. Hedge et al. (2018).

What is the difference between congruent and incongruent trials?

Congruent flankers signal the target's correct response, while incongruent flankers signal a competing response. In the usual arrow variant, the middle arrow determines the correct key regardless of the directions of its neighbors. Eriksen and Eriksen (1974).

How is the flanker interference effect calculated?

A common reaction-time score subtracts mean correct-trial reaction time on congruent trials from mean correct-trial reaction time on incongruent trials. Accuracy differences, neutral baselines, and full response-time distributions can supplement this score. Analysis choices and the contrast used should be reported explicitly.

Why do irrelevant arrows slow responses?

Even with instructions to ignore them, flankers may activate a competing response or influence how quickly the correct target response is selected. The strength and timing of those influences vary with stimulus spacing, salience, expectations, and task design. Eriksen and Schultz (1979); Ridderinkhof et al. (2021).

Can the task diagnose ADHD or another condition?

No individual diagnosis follows from a flanker score alone. Clinical diagnosis requires appropriate assessment methods and evidence of symptoms and impairment in relevant contexts. The flanker procedure contributes experimental observations, not a stand-alone clinical verdict.

Does practicing the flanker task improve everyday focus?

Practice may change performance on the practiced test. Whether improvements transfer to untrained cognitive demands or everyday outcomes is an independent question requiring intervention studies with suitable control groups and meaningful endpoints. Task-specific gains should be reported as such.

Are flanker effects always measured in milliseconds?

Reaction-time differences are usually expressed in milliseconds, but studies may also report error percentages, conditional accuracy, full latency distributions, electrophysiological measures, or parameters from cognitive models. A unit is meaningful only with a clearly defined outcome.

Related Articles

The Stroop Effect: Why Conflicting Information Slows Responses — a complementary conflict paradigm examining competition between word meaning and ink-color naming.

Executive Functions: Inhibition, Working Memory, and Flexibility — the broader framework for interpreting control processes, cognitive flexibility, and differences among experimental tasks.

References

Donzallaz, M. C., Boehm, U., Heathcote, A., Donkin, C., Matzke, D., & Haaf, J. M. (2026). Comparing the reliability of individual differences for various measurement models in conflict tasks. Psychonomic Bulletin & Review, 33, Article 40. https://doi.org/10.3758/s13423-025-02801-7

Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the identification of a target letter in a nonsearch task. Perception & Psychophysics, 16, 143–149. https://doi.org/10.3758/BF03203267

Eriksen, C. W., & Schultz, D. W. (1979). Information processing in visual search: A continuous flow conception and experimental results. Perception & Psychophysics, 25, 249–263. https://doi.org/10.3758/BF03198804

Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the efficiency and independence of attentional networks. Journal of Cognitive Neuroscience, 14(3), 340–347. https://doi.org/10.1162/089892902317361886

Gratton, G., Coles, M. G. H., & Donchin, E. (1992). Optimizing the use of information: Strategic control of activation of responses. Journal of Experimental Psychology: General, 121(4), 480–506. https://doi.org/10.1037/0096-3445.121.4.480

Hedge, C., Powell, G., & Sumner, P. (2018). The reliability paradox: Why robust cognitive tasks do not produce reliable individual differences. Behavior Research Methods, 50, 1166–1186. https://doi.org/10.3758/s13428-017-0935-1

Kucina, T., Wells, L., Lewis, I., de Salas, K., Kohl, A., Palmer, M. A., Sauer, J. D., Matzke, D., Aidman, E., & Heathcote, A. (2023). Calibration of cognitive tests to address the reliability paradox for decision-conflict tasks. Nature Communications, 14, Article 2234. https://doi.org/10.1038/s41467-023-37777-2

Mayr, U., Awh, E., & Laurey, P. (2003). Conflict adaptation effects in the absence of executive control. Nature Neuroscience, 6, 450–452. https://doi.org/10.1038/nn1051

Miyake, A., & Friedman, N. P. (2012). The nature and organization of individual differences in executive functions: Four general conclusions. Current Directions in Psychological Science, 21(1), 8–14. https://doi.org/10.1177/0963721411429458

Molloy, M. F., Lee, T. G., Jonides, J., Zhang, H., Sellers, J., Heathcote, A., Sripada, C., & Weigard, A. S. (2026). Joint cognitive models reveal sources of robust individual differences in conflict processing. Computational Brain & Behavior, 9, 319–336. https://doi.org/10.1007/s42113-026-00263-1

Ridderinkhof, K. R. (2002). Micro- and macro-adjustments of task set: Activation and suppression in conflict tasks. Psychological Research, 66, 312–323. https://doi.org/10.1007/s00426-002-0104-7

Ridderinkhof, K. R., Wylie, S. A., van den Wildenberg, W. P. M., Bashore, T. R., & van der Molen, M. W. (2021). The arrow of time: Advancing insights into action control from the arrow version of the Eriksen flanker task. Attention, Perception, & Psychophysics, 83, 700–721. https://doi.org/10.3758/s13414-020-02167-z

White, C. N., Ratcliff, R., & Starns, J. J. (2011). Diffusion models of the flanker task: Discrete versus gradual attentional selection. Cognitive Psychology, 63(4), 210–238. https://doi.org/10.1016/j.cogpsych.2011.08.001