Dual-Task Interference: Why Performing Two Tasks Slows Us Down

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

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

When two demands arrive close together, the mind may take longer to choose a response, commit more errors, or sacrifice accuracy in one activity to protect another. These effects can occur even when the tasks appear straightforward and involve different senses or muscles. The experimental phenomenon is called dual-task interference. Its importance lies in a precise scientific question: which operations can proceed at the same time, which compete for limited processing capacity, and how does the nervous system coordinate the competing demands?

Dual-task research has produced some of cognitive psychology's clearest evidence that the timing and organization of mental operations matter. It has also produced competing accounts that fit many of the same findings. A rigorous explanation therefore needs to distinguish measurable performance costs from the theoretical mechanisms inferred to produce them. Pashler's foundational review and a later integrative review by Koch and colleagues provide the historical framework.

What Is Dual-Task Interference?

Dual-task interference is a change in performance caused by having to perform, prepare, or coordinate two tasks under overlapping demands, compared with appropriate single-task or less-overlapping conditions. Interference often appears as increased response time, reduced accuracy, slower movement, altered movement variability, or a change in the balance of performance between tasks. Researchers sometimes use the broader term dual-task costs for the numerical differences measured under those conditions.

The defining feature is the experimental comparison. A person who is slow at two tasks individually has not thereby demonstrated dual-task interference. The phenomenon concerns what happens to performance when those tasks compete or overlap. Conversely, interference can be present even when both tasks are completed correctly: a longer reaction time is itself a measurable cost.

Dual-task interference, divided attention, and multitasking

Divided attention describes the allocation or coordination of attention across more than one source or goal. Multitasking is an everyday category for managing multiple activities. Dual-task interference is a specific empirical finding about performance when two demands overlap. These terms are related, but they answer different questions. Divided attention asks how attention is distributed; dual-task research tests the conditions under which performance changes. The broader framework is developed in Divided Attention: Why Multitasking Has Limits.

Sequential task switching is another neighboring phenomenon. A task-switching experiment might compare trials on which the required task changes with trials on which it repeats, creating switch costs or mixing costs. A psychological refractory period experiment manipulates the delay between the starts of two overlapping tasks and typically emphasizes the response to the second. Coordination and control mechanisms can connect these paradigms, but their performance measures and causal interpretations are not interchangeable (Koch et al., 2018).

An ordinary example and its limits

Imagine identifying a spoken tone while selecting a key in response to a visual symbol. The tone and image can be processed through different sensory channels, and the responses may use different hands. Nevertheless, the response to the later event may be delayed. This example illustrates why the popular claim that interference occurs only when two tasks use the same sense or body part is too narrow. Some perceptual and motor operations can overlap while central decisions or coordination remain costly (Pashler, 1984).

How Scientists Demonstrate the Effect

The psychological refractory period paradigm

The psychological refractory period, or PRP, paradigm is the best-known laboratory approach to closely overlapping tasks. Participants respond to stimulus 1 (S1) and stimulus 2 (S2). The interval between the stimulus onsets is called stimulus onset asynchrony (SOA). A short SOA means the second stimulus appears soon after the first; a long SOA gives more time for initial processing of the first task.

Researchers normally measure the reaction time to each event separately: RT1 for the first stimulus-response task and RT2 for the second. A characteristic PRP effect occurs when RT2 increases as SOA becomes shorter. Under standard instructions to prioritize the first task, its response may be relatively protected while the second response absorbs much of the delay. That response asymmetry is especially informative, because it helps locate a possible processing constraint rather than merely showing that doing more work takes longer (Pashler, 1994).

Consider an illustrative pattern, not a measured result from a particular study: the same person's second response might take 520 milliseconds with a long SOA and 730 milliseconds with a short SOA. The extra 210 milliseconds is a short-SOA RT2 cost under the chosen conditions. It does not establish that the individual has only 210 milliseconds of available attention, nor does it identify one brain region as the source of the delay.

Why task order and priority matter

Many PRP studies explicitly instruct participants to respond to task 1 first or give it priority. This reduces ambiguity about which task should be processed first but also shapes the observed pattern. Under different instructions, participants may allocate processing differently, delay the first response, or attempt more overlapping selection. Priority is therefore part of the experimental design and not a nuisance detail that can be ignored (Schumacher et al., 2001; Tombu & Jolicœur, 2003).

Dual-task versus single-task blocks

A second approach compares the same activity alone and concurrently with another activity. For example, participants may perform an auditory classification task by itself and while monitoring a visual display. In cognitive-motor studies, they may walk at a comfortable pace and then walk while counting backward. Each task needs a clearly defined single-task baseline and a compatible dual-task performance metric.

The comparison is conceptually simple, but the interpretation becomes more complicated when participants change their strategy. A person may deliberately walk more slowly to preserve counting accuracy, or maintain walking speed while making more counting mistakes. Examining only one outcome can conceal this tradeoff. The systematic review by Al-Yahya and colleagues describes the heterogeneity of walking dual-task protocols and the importance of task choice.

Simultaneous, overlapping, and interleaved activities

Two tasks can start at exactly the same moment, overlap for a portion of time, or alternate rapidly. Each arrangement imposes distinct requirements. In a PRP design, stimuli can arrive close together without the responses necessarily being produced simultaneously. In continuous driving or conversation, both tasks may extend across seconds or minutes. In task-switching paradigms, the key manipulation often concerns consecutive task sets rather than two actively overlapping response selections.

This distinction helps prevent an overgeneralization: a PRP effect in brief button-press trials demonstrates a constraint under that protocol, whereas the impact of a continuous real-world activity requires additional evidence about goals, stimuli, environment, and time scale (Koch et al., 2018).

The Central Bottleneck Account

The central bottleneck account proposes that particular processing operations cannot be executed for two tasks at exactly the same time. The relevant constraint is often associated with central response selection: choosing which action is appropriate for a perceived event. When task 1 occupies the constrained stage, task 2 can complete some earlier perceptual operations but must wait before completing the stage requiring exclusive access.

The model is a claim about a functional stage of information processing. A bottleneck should not be imagined as a literal narrow anatomical tube or as proof that all cognition occurs serially. Different components of the tasks may run in parallel; a bottleneck concerns particular operations at particular times. Pashler's 1984 experiments tested processing-stage predictions associated with that view.

A stage-by-stage illustration

Suppose task 1 requires detecting a tone, identifying whether it is high or low, selecting the appropriate key, and producing the keypress. Task 2 requires detecting a visual letter, identifying its category, selecting another key, and producing that response. A bottleneck model permits parts of sensory identification and some motor execution to overlap. If the two response-selection stages require an exclusive central process, the second stage begins only after the first releases it. The response to the second stimulus consequently starts later than it would in isolation.

Longer gaps between S1 and S2 make overlap less likely, so the second stage more often arrives after the first has finished. Shorter gaps increase the probability or amount of conflict. This relationship generates a systematic RT2-by-SOA pattern without asserting that the eyes, ears, or hands themselves cannot process two streams at once.

The locus-of-slack logic

One influential method for investigating processing stages manipulates how long a component of task 2 takes while also varying SOA. If task 2 must wait at a central bottleneck, making an early perceptual stage somewhat slower may have little additional effect on RT2 at a short SOA: some extra perceptual processing is absorbed into the waiting time, or cognitive slack. At a longer SOA, the same manipulation may have a clearer effect because little waiting time is available to absorb it.

A manipulation of a central or later stage is often predicted to add to the second response time instead of being absorbed. This interaction pattern, commonly described as underadditivity for early task-2 factors, is useful evidence about processing order. It depends on assumptions that the manipulation selectively influences the intended processing stage and that participants have not changed strategy. It is therefore an inferential technique, not a direct photograph of a mental bottleneck (Pashler, 1984; Tombu & Jolicœur, 2003).

What the evidence supports

PRP studies establish a robust relation between temporal overlap and delayed responses in many simple choice tasks. Central postponement models provide a coherent explanation for important timing interactions. But the same broad reaction-time patterns can be generated by models that distribute limited central capacity rather than enforcing strictly exclusive central selection. The behavior constrains theories; it does not settle all details of a unique internal architecture (Pashler, 1994; Tombu & Jolicœur, 2003).

Capacity Sharing and Multiple-Resource Models

Central capacity sharing

Capacity-sharing models propose that tasks may use a shared, limited processing resource, so both can proceed with reduced efficiency when demand exceeds available capacity. A task that receives less of the resource can take longer to complete even while it remains active. This is a different interpretation of waiting and slowing from a model that prohibits two specific central stages from operating concurrently.

Tombu and Jolicœur developed a formal central capacity-sharing model capable of reproducing several classic PRP signatures, including characteristic SOA effects and processing-stage interactions. Their work is important because it demonstrates a problem of theoretical identification: a behavioral pattern that strongly supports the existence of limited processing does not always establish an exclusively serial bottleneck. Under different parameters, alternative mechanistic models may make closely related predictions (Tombu & Jolicœur, 2003).

Multiple-resource theory

Multiple-resource approaches ask whether different tasks draw on separable capacities. Interference may depend on whether the activities use auditory or visual processing, verbal or spatial representations, and overlapping response or processing resources. Two tasks that differ in one dimension may interfere less than two tasks that overlap strongly on several dimensions. Nonetheless, central coordination can remain costly even for different modalities.

Wickens's model of multiple resources and mental workload developed an influential framework for predicting conditions of overload in applied settings. It is useful for designing displays, controls, and workloads, but a resource label is a theoretical explanation: it requires operational definitions and experimental tests. A simple declaration that attention has been 'used up' is not itself a measurement of a finite quantity.

An economy of processing resources

Earlier work by Navon and Gopher developed a resource-allocation perspective in which the costs of concurrent performance depend on task demands and the allocation of a processing system's capacities. Such approaches helped move the field away from assuming that one universal limit would explain every situation. The remaining challenge is to specify what the resource is, how allocations vary, and which outcomes distinguish a capacity theory from scheduling or structural alternatives.

Contemporary evidence and continuing disagreement

A 2024 experimental study by Wu and Wang compared automatic and controlled task conditions and interpreted persistent interference as support for resource constraints. Its experimental results add to the debate but do not establish that resource sharing universally replaces bottleneck mechanisms. Task instructions, stimulus-response mappings, preparatory states, and other model assumptions affect the competing predictions. The appropriate synthesis is that multiple mechanisms can contribute and particular experiments must discriminate between them.

Executive Coordination and Strategic Scheduling

Concurrent performance involves more than stimulus recognition and response execution. A participant needs to maintain task instructions, establish priorities, coordinate order, monitor progress, and sometimes avoid responding to the wrong stimulus. Executive-control accounts give these functions a central role in the observed cost.

Meyer and Kieras proposed the executive-process interactive control architecture, or EPIC, to explain coordination of perceptual, cognitive, and motor operations. This line of modeling examines how strategic scheduling of component processes can yield task interference even without assuming an inflexible universal bottleneck for every operation. It provides explicit computational commitments against which behavior can be compared (Meyer & Kieras, 1997).

Logan and Gordon's theory similarly illustrates the value of specifying executive control over perceptual attention under dual-task requirements. The major theoretical question is whether delays arise from structural processing exclusivity, divisible central capacity, deliberate coordination policies, or combinations of these factors. The field's strongest conclusions concern tested relationships among tasks, timing, and instructions, rather than one universal explanation for every concurrent activity.

Interference Can Occur at Several Processing Stages

Perceptual overlap

Two visual tasks may compete for spatial attention, object selection, or discrimination, and two speech-related tasks may place demands on related auditory or linguistic processing. If participants must detect multiple weak or brief signals, accuracy costs can arise even before any button-press decision. Perceptual interference therefore cannot be reduced to the same central response-selection bottleneck invoked to explain certain PRP effects.

Central decisions and memory retrieval

Even with distinct perceptual inputs, tasks may conflict while selecting a response, interpreting a rule, or retrieving an appropriate action. This is one reason different sensory modalities do not guarantee interference-free performance. However, individual studies differ in the degree to which their manipulations separate perception, decision, memory, and response coordination (Pashler, 1994).

Motor execution and action compatibility

The response devices also matter. Two actions involving the same hand, conflicting movements, incompatible spatial mappings, or constrained postures can interfere for reasons partly independent of central decision selection. A credible dual-task experiment therefore describes both input modalities and output modalities, response mappings, task order, and practice. Otherwise, an apparent cognitive interference effect may include peripheral motor competition.

Cross-talk between task rules

Interference also occurs when features of one task accidentally activate a response or rule relevant to the other. Sharing the same stimuli, response categories, or action mappings may introduce response conflict. Separating central capacity limits from task-set cross-talk often requires manipulating mappings or comparing conditions with similar central demand but different opportunities for response confusion (Koch et al., 2018).

How Dual-Task Costs Are Measured

Reaction time and accuracy

Reaction time is usually recorded in milliseconds from the presentation of a stimulus to the registered response. Researchers compare RT1 and RT2 across SOAs, single-task baselines, task-order conditions, and stimulus or response mappings. Accuracy is reported alongside speed, because a participant who responds faster by guessing or omitting hard trials may display a misleading improvement in mean reaction time.

An adequate analysis specifies exclusions, anticipatory responses, missed responses, and the treatment of extreme reaction times. It also distinguishes within-person experimental contrasts from stable differences between people. A task can produce a reproducible average experimental effect while being a less reliable individual-differences instrument.

Absolute and proportional costs

In tasks where higher values indicate slower performance, one common measure is absolute cost: dual-task reaction time minus single-task reaction time. A proportional cost expresses that difference relative to a single-task baseline, often as a percentage: 100 × (dual-task time − single-task time) / single-task time. For a speed measure where lower values indicate impairment, analysts may reverse the sign or use a different formula. The sign convention and selected denominator must always be stated.

For illustration, if a person takes 400 milliseconds alone and 480 milliseconds with a competing task, the absolute slowing is 80 milliseconds and the proportional increase is 20%. This is a constructed example of calculation, not an empirical population estimate. A 20% reaction-time increase also has a different practical meaning from a 20% decrease in walking speed.

Assessing both tasks, not just one

A full dual-task analysis considers the first and second tasks together. Improving performance on one activity by deprioritizing the other is not an overall gain in concurrent capacity. Researchers can display each task's performance under different instructions or difficulty settings, producing an empirical picture of tradeoffs. This is especially important when participants have the option to prioritize balance, speed, or accuracy.

Reliability and construct validity

The existence of a mean experimental cost does not make the cost a precise diagnostic score. A 2023 meta-analysis by Pike and colleagues found notably different test-retest reliability for gait-related and cognitive dual-task cost metrics, together with limitations in predictive and concurrent validity. The review reported higher average reliability for locomotor costs than cognitive costs in the studies evaluated and cautioned against broad clinical application without stronger measurement evidence.

Reliability concerns how consistently a measure distinguishes people or repeats across occasions. Validity concerns whether it measures the intended construct and supports a proposed use. Both depend on task selection, baseline procedures, outcome scaling, sample characteristics, and measurement error. Neither should be inferred from the fact that a mean cost is statistically significant.

Why Results Change Across Tasks and People

Difficulty, similarity, and stimulus timing

Harder discrimination, less compatible mappings, shorter SOAs, increased memory demands, and overlapping response channels may increase interference, but not always through the same mechanism. A ceiling on accuracy or floor on response time can obscure differences. Some ostensibly easy tasks require substantial central response selection, while some practiced skills become highly efficient under specific task combinations.

Preparation and expectations

If the second task is predictable, participants may prepare its stimulus-response mapping. If the arrival of a stimulus is uncertain, monitoring the event may itself require attention. Experimental instructions about priority, preparation, or permitted response order can shift cost patterns. Consequently, a dual-task effect should be interpreted within the operational conditions that produced it.

Individual differences and age

People differ in practiced skills, sensory acuity, motor performance, processing speed, and task strategies. Some studies of cognitive-motor dual-tasking observe larger performance costs in older groups, although estimates depend on the chosen tasks, outcomes, and populations. The systematic review by Al-Yahya et al. reported associations involving age and reduced gait speed in dual-task conditions. Such group-level evidence does not justify classifying an individual as impaired on the basis of a single dual-task score.

Culture and learning history

Task familiarity, language, literacy, numeracy, and experience with computerized responses can all influence performance. Counting backward, classifying words, and responding to symbolic keys do not impose identical demands across participants. Researchers should treat these variables as design and interpretation considerations rather than assuming laboratory task equivalence merely because the instructions use the same words.

What Training Can and Cannot Change

Practice can reduce interference in particular learned task combinations. When perceptual identification, response mapping, or task coordination become more efficient, participants may spend less time competing for the same operation or adopt better scheduling. Schumacher and colleagues showed that some participants attained virtually perfect time sharing after practice on carefully arranged choice-reaction tasks. This finding demonstrates substantial plasticity under favorable conditions; it does not prove that arbitrary real-world tasks can always become simultaneous without a cost.

Strobach and colleagues studied improvement and transfer of dual-task coordination skills, supporting the possibility of learning beyond the exact practiced trial configuration under some circumstances. Yet transfer has to be tested, not assumed. A reduction in one experimental cost may reflect a strategy, familiarity with the stimuli, or a changed speed-accuracy balance rather than a general increase in available cognitive capacity.

A related warning comes from broader cognitive-training research. In a randomized placebo-controlled trial, Redick and colleagues observed improvement on practiced training tasks without evidence of transfer to the tested broader cognitive abilities. That study examined working-memory training rather than PRP interference directly; its relevance is methodological. It illustrates why near-task improvement and generalized enhancement must be treated as different claims.

Dual-Task Interference in Everyday Activities

Driving and conversation

Driving requires monitoring a changing scene, selecting actions, anticipating other road users, and responding rapidly to hazards. In a simulated-driving study, Levy, Pashler, and Boer reported that shortening the temporal interval between a choice task and a braking event increased brake reaction time in a manner consistent with PRP interference. This is a specific demonstration that a well-practiced response such as braking can be delayed by a competing task.

Strayer, Drews, and Johnston also found that hands-free phone conversation impaired performance and visual attention in simulated driving. The critical distinction is that freeing the hands does not remove the cognitive demands of conversation. Such findings support care with divided demands while driving, but laboratory or simulator effect sizes should not be converted directly into an individual's on-road crash probability without additional epidemiologic evidence.

Walking while thinking or speaking

Walking and cognition can interfere, particularly when gait must be adapted to changing conditions or a cognitive task requires active monitoring, calculation, or response selection. A systematic review and meta-analysis of walking studies documented effects on gait speed and noted differences associated with cognitive task requirements. These findings show that even activities often regarded as automatic can recruit attention depending on context.

Applied walking tests should be understood as multidimensional assessments. A change in gait speed under concurrent counting may reflect response prioritization, sensory or motor limitations, worry about stability, or a particular task combination. It is not, on its own, evidence of ADHD, dementia, or a generic 'multitasking disorder.' The Pike et al. meta-analysis highlights unresolved validation problems in such measurement.

Work, learning, and safety-critical settings

In work or learning environments, two activities compete most strongly when each requires timely decisions, effortful selection, or accuracy-sensitive processing. A notification that requires reading and choosing a response can interrupt a demanding reasoning task, but the exact consequences depend on overlap, task goals, and delay. The everyday phrase 'multitasking makes everyone inefficient' goes beyond the evidence: some combinations overlap effectively, others interfere substantially, and training or prioritization can change the outcome.

In safety-critical design, the scientific implication is to reduce incompatible simultaneous demands, allow adequate time between urgent response selections where feasible, use interfaces that make priorities clear, and test realistic task combinations. Wickens (2008) discusses multiple-resource models as a tool for anticipating mental workload under overlapping tasks.

Neural Evidence and Its Limits

Neuroimaging studies can ask which neural systems are more engaged when two tasks must be coordinated. For example, Szameitat and colleagues (2002) compared separate and concurrent auditory-visual choice-reaction tasks and investigated brain activation associated with dual-task control. Their results link experimental interference to executive-control networks, but activation differences do not identify a unique causal bottleneck or show that one brain area alone contains a limited pool of attention.

Reaction-time experiments establish temporal patterns; imaging adds spatial and network-level observations; computational models articulate candidate mechanisms. Strong theory develops when these approaches converge on predictions that distinguish competing explanations. A correlation between an activation signal and a task cost cannot by itself establish which cognitive stage caused that cost.

Established Findings, Model Debates, and Open Questions

The strongest established result is that closely overlapping tasks often slow responses or alter accuracy, and these costs depend systematically on stimulus timing, instructions, and task demands. The PRP effect has been replicated across many experimental configurations and remains a central paradigm for the study of cognitive limitations (Pashler, 1994; Koch et al., 2018).

The competing interpretation of that interference is a live theoretical issue. Central bottleneck models, central capacity-sharing models, multiple-resource frameworks, and executive-scheduling models attribute different portions of the same costs to structural stages, divisible capacity, resource overlap, or control policies. No single broad label automatically determines the mechanism of a given experiment (Meyer & Kieras, 1997; Tombu & Jolicœur, 2003; Wu & Wang, 2024).

Evidence about far transfer, stable individual differences, and clinical prediction is more limited and varies with task and outcome. Practice effects are real under some laboratory conditions, but an observed decrease in task-specific interference does not establish universal cognitive enhancement. Similarly, performance on one laboratory dual-task protocol should not be treated as an independent medical diagnosis (Schumacher et al., 2001; Pike et al., 2023).

Several questions remain scientifically productive: Which processes are best described as exclusive bottlenecks, and which as capacity sharing? How much apparent limitation is a consequence of learned coordination strategies? What features make gains transfer to unfamiliar tasks? How should task-specific performance, ecological validity, and individual measurement reliability be combined? These questions require experiments that manipulate theoretically diagnostic variables rather than relying solely on everyday impressions of busyness.

Frequently Asked Questions

Does the brain literally process only one thing at a time?

No universal one-task-at-a-time rule follows from dual-task research. Perceptual and motor processes can overlap, and practiced tasks may be coordinated efficiently. Specific operations, including response selection in certain paradigms, often show strong interference under close temporal overlap. The empirical question is which operations can overlap under defined conditions (Pashler, 1994).

Why does the second task often slow down more than the first?

Classic PRP instructions tend to prioritize the first task. Under a bottleneck account, the second task reaches a central stage while the first is still using it, creating postponement. Capacity-sharing and scheduling models can also produce characteristic asymmetries. The experimental instructions matter as much as the timing measure (Tombu & Jolicœur, 2003).

Does using different senses eliminate dual-task interference?

Different sensory channels can reduce some forms of competition, particularly when tasks otherwise require overlapping perceptual resources. They do not guarantee cost-free simultaneous central decisions or coordination. Auditory and visual tasks can still interfere in experiments (Szameitat et al., 2002; Wickens, 2008).

Is a dual-task cost the same as a task-switching cost?

No. A dual-task cost normally compares concurrent or closely overlapping demands with an appropriate baseline. A switch cost typically compares switching to a different task with repeating the task across sequential trials. Both inform theories of control, and their mechanisms partly intersect (Koch et al., 2018).

Can people get better at performing two tasks at once?

Yes, performance can improve with practice for some combinations, and some laboratory tasks permit unusually efficient time sharing. The size and transfer of improvement depend on task structure, training, and evaluation design. It would be unwarranted to promise that any particular training exercise will make all multitasking effortless (Schumacher et al., 2001; Strobach et al., 2012).

Does a high dual-task cost indicate ADHD or dementia?

A single dual-task cost is an experimental performance result. It may depend on sensory, motor, linguistic, motivational, strategic, and timing factors, as well as measurement error. Clinical conditions require their own diagnostic criteria and assessments. Even in applied gait research, the validity of cost measures for predictive uses remains an active question (Pike et al., 2023).

Is talking while driving a dual-task problem if both hands are free?

Yes, it can be. Controlled driving-simulation studies show slowing or reduced attention during concurrent cognitive conversation, including hands-free conditions. These experimental findings support treating cognitive distraction as relevant to driving safety without pretending that a simulator predicts every real-world outcome (Levy et al., 2006; Strayer et al., 2003).

Scientific Conclusion

Dual-task interference reveals the organization of cognitive processing through measurable competition between tasks. When task demands overlap, the resulting slowing and performance tradeoffs provide evidence about response selection, limited capacity, task-set coordination, perceptual conflict, and motor requirements. The core experimental achievement is the ability to manipulate overlap and test which model predictions follow. The broader practical lesson is conditional: coordinating two tasks depends on the processing operations they require, their timing, their priority, and the participant's experience. Accurate interpretation requires both a theory of the mechanism and a careful account of what was actually measured.

Related Articles

Divided Attention: Why Multitasking Has Limits examines the broader psychology of allocating attention between activities. Task Switching: What Switching Costs Reveal About Mental Control explains sequential task-switching costs and their experimental interpretation. Attention: Types, Mechanisms, and Limits places interference within the larger organization of attention. These linked scientific articles cover related but separately defined questions.

References

Al-Yahya, E., Dawes, H., Smith, L., Dennis, A., Howells, K., & Cockburn, J. (2011). Cognitive motor interference while walking: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews, 35(3), 715–728. https://doi.org/10.1016/j.neubiorev.2010.08.008

Koch, I., Poljac, E., Müller, H., & Kiesel, A. (2018). Cognitive structure, flexibility, and plasticity in human multitasking: An integrative review of dual-task and task-switching research. Psychological Bulletin, 144(6), 557–583. https://doi.org/10.1037/bul0000144

Levy, J., Pashler, H., & Boer, E. (2006). Central interference in driving: Is there any stopping the psychological refractory period? Psychological Science, 17(3), 228–235. https://doi.org/10.1111/j.1467-9280.2006.01690.x

Logan, G. D., & Gordon, R. D. (2001). Executive control of visual attention in dual-task situations. Psychological Review, 108(2), 393–434. https://doi.org/10.1037/0033-295X.108.2.393

Meyer, D. E., & Kieras, D. E. (1997). A computational theory of executive cognitive processes and multiple-task performance: Part 1. Basic mechanisms. Psychological Review, 104(1), 3–65. https://doi.org/10.1037/0033-295X.104.1.3

Navon, D., & Gopher, D. (1979). On the economy of the human-processing system. Psychological Review, 86(3), 214–255. https://doi.org/10.1037/0033-295X.86.3.214

Pashler, H. (1984). Processing stages in overlapping tasks: Evidence for a central bottleneck. Journal of Experimental Psychology: Human Perception and Performance, 10(3), 358–377. https://doi.org/10.1037/0096-1523.10.3.358

Pashler, H. (1994). Dual-task interference in simple tasks: Data and theory. Psychological Bulletin, 116(2), 220–244. https://doi.org/10.1037/0033-2909.116.2.220

Pike, A., McGuckian, T. B., Steenbergen, B., Cole, M. H., & Wilson, P. H. (2023). How reliable and valid are dual-task cost metrics? A meta-analysis of locomotor-cognitive dual-task paradigms. Archives of Physical Medicine and Rehabilitation, 104(2), 302–314. https://doi.org/10.1016/j.apmr.2022.07.014

Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z., Kane, M. J., & Engle, R. W. (2013). No evidence of intelligence improvement after working memory training: A randomized, placebo-controlled study. Journal of Experimental Psychology: General, 142(2), 359–379. https://doi.org/10.1037/a0029082

Schumacher, E. H., Seymour, T. L., Glass, J. M., Fencsik, D. E., Lauber, E. J., Kieras, D. E., & Meyer, D. E. (2001). Virtually perfect time sharing in dual-task performance: Uncorking the central cognitive bottleneck. Psychological Science, 12(2), 101–108. https://doi.org/10.1111/1467-9280.00318

Strayer, D. L., Drews, F. A., & Johnston, W. A. (2003). Cell phone-induced failures of visual attention during simulated driving. Journal of Experimental Psychology: Applied, 9(1), 23–32. https://doi.org/10.1037/1076-898X.9.1.23

Strobach, T., Frensch, P. A., Soutschek, A., & Schubert, T. (2012). Investigation on the improvement and transfer of dual-task coordination skills. Psychological Research, 76(6), 794–811. https://doi.org/10.1007/s00426-011-0381-0

Szameitat, A. J., Schubert, T., Müller, K., & von Cramon, D. Y. (2002). Localization of executive functions in dual-task performance with fMRI. Journal of Cognitive Neuroscience, 14(8), 1184–1199. https://doi.org/10.1162/089892902760807195

Tombu, M., & Jolicœur, P. (2003). A central capacity sharing model of dual-task performance. Journal of Experimental Psychology: Human Perception and Performance, 29(1), 3–18. https://doi.org/10.1037/0096-1523.29.1.3

Wickens, C. D. (2008). Multiple resources and mental workload. Human Factors, 50(3), 449–455. https://doi.org/10.1518/001872008X288394

Wu, Y., & Wang, Q. (2024). Dual-task interference: Bottleneck constraint or capacity sharing? Evidence from automatic and controlled processes. Attention, Perception, & Psychophysics, 86, 815–827. https://doi.org/10.3758/s13414-024-02854-1