RESEARCH DOSSIER
Cognitive Control and Task Switching
Abstract. Cognitive control enables goal-sensitive coordination of perceptual, mnemonic, and action processes; task switching offers experimental contrasts for examining its organization. This fundamental research monograph traces the historical formation of task-switching paradigms and critically synthesizes research on task sets, switch and mixing costs, advance preparation, cue-encoding confounds, response congruency, task-set reconfiguration, episodic priming, backward inhibition, voluntary task choice, context learning, and neural implementation. It evaluates methodological and psychometric constraints, age-related differences, clinical interpretation, intervention transfer, and possible applications to human–artificial work environments. Original experiments, major theoretical reviews, meta-analyses, and results through 2025 are compared with explicit attention to competing explanations and the distinction between group-level experimental effects and reliable individual-difference measurement. The synthesis supports a context-dependent, multi-process account of flexible task coordination and identifies unresolved questions concerning process decomposition, ecological validity, and computational and neural model testing. This is a critical narrative research synthesis without original experimental data or a protocol-registered systematic-review claim.
Keywords: cognitive control; task switching; task set; switch cost; mixing cost; task-set reconfiguration; executive functions; preparation interval; cue priming; backward inhibition; response congruency; proactive control; reactive control; cognitive flexibility; construct validity; neuroscience of cognitive control
Definition, Scope, and Scientific Questions
Cognitive control is the family of processes through which an organism organizes perception, memory, and action according to a currently relevant goal. Task switching is a particularly informative experimental domain because participants alternate between competing task rules while the experimenter measures response latency, errors, and effects of recent task history. An operational definition specifies the rule to be used, how it is signaled, which stimuli afford multiple responses, and the environmental conditions under which the rule changes. Cognitive control includes sustained task focus, rule retrieval, response selection, monitoring, and flexible adjustment; a switch between rules makes several of these functions observable within one trial. Monsell (2003) and Egner (2023) provide complementary syntheses of task-switching findings and the relation between stability and flexibility.
The scientific object of this dossier is goal-sensitive coordination of task sets, including task-rule representation, advance preparation, switch and mixing costs, interference from competing rules, episodic retrieval, control learning, neural implementation, and the validity of observed switching measures. It also examines age differences, interventions, and functional consequences where sufficiently documented. It remains distinct from response inhibition as a stand-alone mechanism: suppressing a prepotent button press may occur within a task switch, yet a task can change without the particular inhibition demands characteristic of stop-signal or go/no-go procedures. Conversely, response conflict may persist when the task rule is unchanged. The distinctions matter because a common reaction-time cost does not identify the operation that produced it. Koch et al. (2010) reviews inhibitory mechanisms in switching, while Miyake et al. (2000) illustrates why executive components require multiple indicators.
The principal questions concern what a task set represents, why task alternation is costly, when preparation reduces that cost, how learned associations and response conflicts participate, and what task-switching scores mean beyond an individual experiment. This work is a critical narrative synthesis of original behavioral experiments, theoretical models, reviews, meta-analyses, and selected neural research. It presents no original participants, new effect-size estimates, or preregistered systematic-review protocol. Claims are separated by evidence level: an experimental contrast establishes a difference under specified conditions; a mechanism is inferred by comparing alternative predictions; a broad individual-difference conclusion requires reliable measurement and replication. Kiesel et al. (2010) and Vandierendonck et al. (2010) demonstrate why interpreting the same behavioral signature requires multiple convergent manipulations.
Historical Formation of the Task-Switching Problem
The experimental study of flexible control emerged from broader traditions of attention, mental set, stimulus–response compatibility, and interference. In the modern task-switching literature, a decisive methodological advance was to alternate well-learned tasks within an ordered sequence and compare performance immediately after a change with performance after a repetition. Rogers & Monsell (1995) investigated predictable switches between simple cognitive tasks and showed that alternation remains consequential even when the order can be anticipated. Such experiments transformed a broad notion of mental flexibility into identifiable contrasts among preparation, task demands, and trial history. Historical importance belongs to the methodological articulation of the contrast rather than to the claim that one simple numerical cost measures a unique mental faculty.
Meiran (1996) further analyzed reconfiguration before task performance, placing preparation interval and processing mode at the center of the experimental question. Rubinstein et al. (2001) examined executive control in task switching with an emphasis on component processes. These contributions established a research vocabulary for rule activation, reconfiguration, execution, and residual costs. However, the vocabulary can mislead when a component name is treated as a directly observed event. Reaction time is the end product of several operations; a difference between two experimental cells constrains theories without automatically selecting one. Later methodological critiques specifically challenged whether apparent switching costs could arise partly from changes in the cue rather than changes in the underlying task.
By the early 2000s, competing programs were well developed. Reconfiguration accounts emphasized task-specific preparation, whereas interference and priming accounts emphasized carryover from recent events. Studies of backward inhibition and voluntary switching extended the temporal and strategic dimensions of the field. Mayr & Keele (2000) supplied a prominent backward-inhibition paradigm; Logan & Bundesen (2003) questioned the attribution of cued switching costs to an endogenous controller; Arrington & Logan (2004) examined voluntary choices between tasks. Reviews such as Monsell (2003) and Kiesel et al. (2010) subsequently organized these disagreements into experimentally testable contrasts. The field's enduring lesson is methodological: task switching is a controlled window into adaptive coordination, and its explanation demands attention to every event separating the cue from the measured response.
Task Sets, Goals, and Rule Representations
A task set is an organized configuration of instructions and processing priorities that permits a stimulus to be interpreted as a particular kind of problem and mapped to an appropriate response. In a number classification task, the same displayed digit might demand an odd–even judgment on one trial and a greater-than-five judgment on the next. Relevant information, response categories, and evaluation criteria shift with the rule. The task set can therefore include stimulus attention, categorization criteria, decision thresholds, response mappings, and representations of expected feedback. Sakai (2008) surveys evidence linking task-set organization to prefrontal neural systems, while Egner (2023) treats task focus and switching readiness as related but distinguishable problems of adaptive control.
Task sets are experimentally inferred rather than located as self-evident internal objects. A cue can retrieve an instruction from memory, prime a response mapping, bias attention to one stimulus dimension, or activate an episodic trace of an earlier trial. Several pathways may generate the same apparent rule activation. For example, a color cue might lead participants to attend to color before the target appears, and a familiar cue–response association might simultaneously promote a specific motor tendency. Schneider & Logan (2005) demonstrates that short-term priming can reproduce aspects of explicitly cued task-switching performance without positing a separate switching operation for every observed cost. Any strong claim about task-set reconfiguration must therefore show why alternative cue and retrieval explanations predict a different empirical pattern.
Goal maintenance and goal change interact across timescales. Maintaining one task set may require resistance to distractors over seconds; adapting to environmental shifts may require learning that yesterday's mapping has become obsolete. Short-term task-switching experiments isolate precise moments of transition, but habitual control policies can reflect longer experience. The unity-and-diversity perspective Miyake et al. (2000) links shifting to broader executive functioning while retaining differentiated components, and Friedman & Miyake (2017) refines the distinction between common executive ability and particular shifting demands. This theoretical context clarifies why cognitive control is usefully understood as an organization of processes, not as a single all-purpose capacity inferred from the speed of one keypress.
Cognitive Stability, Flexibility, and Control Demands
Cognitive stability permits persistence with a relevant rule in the presence of distraction, competing task associations, and changing superficial stimulus features. Flexibility permits the individual to abandon or revise the currently prioritized rule when a new goal becomes relevant. Both functions have costs and benefits that depend on environmental regularity. A highly stable configuration can prevent errors in repetitive work but make transitions slower; a highly flexible configuration can accommodate change while leaving the current task more vulnerable to intrusion. The framework developed by Egner (2023) treats focus and switch readiness as potentially supported by distinct mechanisms whose settings are adjusted through experience. Dreisbach & Fröber (2019) examines experimental circumstances that modulate this stability–flexibility balance.
A stability–flexibility trade-off should be stated as a contingent empirical relationship rather than a universal law. A participant can improve both focus and transition performance when instructions become clearer, cues more predictive, or working-memory demands lighter. Conversely, a manipulation designed to increase switching readiness may raise distraction costs only under certain stimulus and timing conditions. Experimental conclusions therefore depend on measuring both repeat-trial performance and switch-trial performance, rather than rewarding a small switch cost in isolation. If both types of trial slow equally, the switch cost may shrink while overall effectiveness deteriorates. Vandierendonck et al. (2010) emphasizes the joint influence of configuration and interference, providing a basis for interpreting such patterns.
Flexible control also includes selecting when to invest in preparation. In a stable environment, maintaining a strong task rule may be efficient; in a volatile environment, keeping alternatives accessible may save time at the next change. These are predictions about the allocation of control, not a declaration that flexibility is always desirable. Braver (2012) distinguishes proactive from reactive control and connects resource investment to when relevant information becomes available. Egner (2023) relates control settings to recent task demands and context-dependent retrieval. Together these accounts support a research program that measures conditions, strategies, and outcomes instead of interpreting a single average switching score as psychological adaptability in general.
Experimental Paradigms and Operational Controls
A standard task-switching experiment presents at least two tasks, assigns each a stimulus–response rule, and intermixes trials under predictable or unpredictable sequences. Participants may alternate between judging shape and judging color, or between classifying the parity and magnitude of a number. The investigator manipulates whether the rule repeats or changes, how the rule is cued, and the interval available for preparation. Outcomes include reaction times for correct responses, error probabilities, omissions, and sometimes distributional measures. The contrast between switch and repeat trials is meaningful only after specifying the trial sequence and response demands. Rogers & Monsell (1995), Meiran (1996), and Monsell (2003) provide foundational demonstrations of these design principles.
Predictable alternation makes the upcoming rule knowable from sequence position, while explicit cueing places a task identifier before the target. Voluntary procedures allow participants to choose the next task under instructions intended to balance unpredictability and task frequencies. Each design addresses a different explanatory question. Predictable sequences can reveal the benefits and limits of temporal preparation; cueing permits manipulation of cue–target intervals; voluntary protocols address strategic choice, although selection and execution remain intertwined. Arrington & Logan (2004) shows how voluntary switching produces measurable costs and preferences. Logan & Bundesen (2003) and Schneider & Logan (2005) emphasize that explicit cueing introduces its own encoding and priming processes, which experimental designs must analyze explicitly.
A rigorous design balances task frequencies, transition probabilities, response mappings, target congruence, and cue repetitions. It accounts for the fact that consecutive trials are statistically dependent: a target category, response hand, or cue might repeat even when the nominal task changes. Exclusion criteria and trimming of long latencies can modify observed means; accuracy must be evaluated alongside speed to detect strategic trade-offs. Wylie & Allport (2000) interrogates what is actually measured by task-switch costs, and Kiesel et al. (2010) reviews how design details influence their interpretation. Clear operational reporting makes it possible to compare laboratories, reproduce findings, and decide whether an observed cost reflects the task-rule transition or a more local feature of the experimental episode.
Switch Costs and Their Measurement
The switch cost is ordinarily estimated as the difference in response time or error rate between task-switch trials and appropriately matched task-repeat trials. A conventional latency measure subtracts mean correct-trial reaction time for repetitions from mean correct-trial reaction time for switches. Its simplicity makes it useful for detecting experimental effects, yet it combines processes unfolding before the target, during response selection, and after stimulus evaluation. Rogers & Monsell (1995) and Monsell (2003) established the characteristic finding that alternation can slow performance under many conditions. The critical interpretive task is to ask what the comparison holds constant: stimuli, cues, response probabilities, temporal predictability, and recent actions can all differ between switching and repeating.
A large switch cost can indicate interference from a prior task, effortful reconfiguration, cue-decoding difficulty, response competition, or a combination. It can also depend on the baseline speed of repeat trials; especially fast repeats enlarge the difference without necessarily indicating impaired transition ability. Accordingly, raw switch costs, proportional costs, accuracy costs, and full latency distributions answer partly different questions. Wylie & Allport (2000) showed why the construct must be tied to the precise experimental contrast. Vandierendonck et al. (2010) and Kiesel et al. (2010) compare reconfiguration and interference explanations and document the need to distinguish process models using factorial manipulations rather than assuming a single mechanism from the direction of the cost.
Task-switch costs are frequently described as robust at the group level because their typical experimental contrast can be reproduced. That robustness does not establish that the difference reliably ranks individuals. Measurement error grows when two correlated reaction-time conditions are subtracted, and stable participant-specific variation may be small compared with the average within-person effect. Hedge et al. (2018) describes this reliability paradox for experimental tasks used to assess individual differences. If the purpose is to measure a trait or predict outcomes outside the laboratory, investigators need explicit test–retest evidence, enough trials, multiple indicators, and a defensible model of what personal difference the observed contrast represents. Experimental sensitivity and psychometric validity are separate properties.
Mixing Costs and Sustained Control
A mixing cost compares repeat trials performed in a mixed block with performance on the same task in a pure block containing no need to alternate rules. This contrast differs from the local switch cost because a mixed-block repetition does not follow a task change. It may reflect maintaining multiple task rules, monitoring which rule applies, uncertainty about upcoming demands, or a strategic speed adjustment throughout the block. Monsell (2003) distinguishes local transition effects from broader costs of mixed-task contexts. Both measures are needed to understand adaptive control because a participant might execute actual switches efficiently while remaining slow on every mixed-block trial due to sustained monitoring.
Pure and mixed blocks differ in several ways beyond the number of active tasks. The proportion of target types, surprise, average cue frequency, task predictability, and attentional orientation can change across block structures. If a pure block omits cues while a mixed block displays them, the mixing cost may partly be the time needed to interpret cues. If the investigator equates task cues, stimuli, and responses, some of these sources can be reduced; residual differences require further explanation. Kiesel et al. (2010) reviews how task-set management and interference operate across trial and block levels. A full account should identify whether the estimated effect concerns sustained goal maintenance, episodic history, or the local act of changing a rule.
Mixing and switch costs can change differently with age, practice, and experimental instructions. Thus, improvement in one component is not evidence of improvement in the other. Wasylyshyn et al. (2011) meta-analyzed task switching and aging, emphasizing the importance of distinguishing global and local measures when interpreting age-related differences. Such differentiation also matters in training studies, where practice may optimize a known sequence or reduce time spent monitoring cues without producing broad transfer. An evidence-centered report states the block design, baseline condition, error treatment, and whether improvement represents absolute performance or a cost computed from two conditions. This preserves the scientific meaning of the measure.
Advance Preparation and Residual Costs
In predictable alternation and cued task switching, advance information may allow a participant to retrieve the impending rule before the target appears. Experiments manipulate preparation time through response–stimulus intervals, cue–target intervals, or both. When longer preparation reduces the switch cost, the finding is consistent with the possibility that at least some switching-related processing can occur in advance. Meiran (1996) investigated advance reconfiguration, and Monsell (2003) reviewed the frequent observation that measurable costs may remain even after substantial preparation. The term residual switch cost designates the remaining difference under a specified long preparation interval; it is a descriptive result that requires interpretation.
A residual cost has sometimes been presented as evidence that task-set reconfiguration cannot be entirely completed before target onset. Other accounts point to incomplete cue processing, lingering interference, task-specific stimulus associations, or constraints that require the actual stimulus to resolve. The experimental contrast alone does not decide among these alternatives. Altmann (2004) analyzed what work occurs during advance preparation, and Schneider & Logan (2005) offered a cue-priming perspective that generates competing predictions. To isolate active preparation, a study should manipulate task identity, cue identity, available time, and the probability that preparation will pay off. It should also check whether longer intervals permit additional distraction or decay.
Recent experiments clarify the temporal interpretation of task preparation. Swainson et al. (2024) demonstrated that preparing a task can influence switching costs on a later trial even when the earlier trial did not include target performance. In a direct seven-experiment investigation, Yamaguchi & Swainson (2025) found that costs following cue-only trials depended on procedure and preparation interval; effects after cue-only events were more short-lived than effects after completed target-response trials. These findings support multiple temporally distinguishable sources of cost. They show why a single residual-cost value cannot be assigned exclusively to one hidden process and why cue-only and completed trials must be matched as closely as feasible before a mechanism is inferred.
Task Cues, Cue Repetition, and Perceptual Priming
Explicit cues provide information about which task to perform, yet the acts of identifying the cue and retrieving its meaning can themselves affect latency. Suppose each rule is represented by only one visual symbol. A task repetition necessarily produces a cue repetition, and a task switch necessarily produces a cue change. The observed latency difference then conflates repetition of the instruction signal with repetition of the task rule. Logan & Bundesen (2003) challenged a strong endogenous-control interpretation of explicit cueing by demonstrating the explanatory force of cue-encoding operations. The appropriate experimental question is whether the switching disadvantage persists when cue repetition and task repetition can be varied independently.
Using at least two cues for each task permits comparisons among cue repetitions, cue changes without task changes, and genuine task changes. This design makes it possible to estimate how much of a standard cost belongs to cue encoding and how much remains associated with task identity. However, even a changed cue within the same task might retrieve a different episode or create distinct expectancy effects. Schneider & Logan (2005) developed a short-term priming account of cued task switching that highlights this issue. Model comparison should assess predicted interactions with cue familiarity, preparation interval, and prior event history rather than equating a statistical contrast with an underlying executive mechanism.
Cue meaning and sensory appearance can also influence preparation strategies. A short arbitrary label may require a learned translation into a rule, while an informative word or stable spatial position may directly guide attention. Changes in cue discriminability can reduce task-retrieval time while leaving stimulus–response interference unchanged. An investigator who interprets such improvement as enhanced cognitive control risks confusing interface design with psychological capacity. Kiesel et al. (2010) surveys cue-related effects, and Yamaguchi & Swainson (2025) shows that even the presence or absence of a target on the preceding trial changes the time course of transition costs. These results support reporting stimulus and cue specifications in sufficient detail for direct replication.
Response Congruency and Cross-Task Interference
Bivalent stimuli are capable of supporting more than one task judgment; a numeral can be categorized by parity or magnitude, and a colored shape can be judged by color or shape. Some stimuli imply the same response under both currently learned rules, while others imply different responses. The distinction defines cross-task response congruency. Incongruent stimuli can generate conflict even when the active task is correctly identified, because the irrelevant task can activate a competing response. Kiesel et al. (2010) reviews congruency effects and their interpretation within the task-switching literature. Congruency manipulations are especially informative because they reveal that knowledge of an inactive rule can influence current performance without a conscious decision to perform that rule.
The size of a congruency effect can depend on whether the task was just repeated or changed, how strongly each task has been practiced, and whether target features repeat. A larger conflict after a switch could arise from imperfect shielding of the new rule, lingering activation of the previous one, or a response-mapping event that was recently bound to a similar stimulus. Vandierendonck et al. (2010) emphasizes the interplay of interference and reconfiguration, while Koch et al. (2010) analyzes inhibitory accounts. A decisive experimental interpretation requires checking whether the same pattern appears with univalent stimuli, alternative mappings, and response repetitions held constant. The researcher should also inspect errors, since some participants may slow deliberately to protect accuracy.
Interference research gives task switching its explanatory breadth. Control involves regulating the relative influence of simultaneously available interpretations and actions, and rule selection is therefore never wholly detached from stimulus content. Still, a congruency effect alone cannot identify whether competition occurred at the perceptual, conceptual, task-rule, or motor stage. Models need stage-specific predictions and designs that alter one source without inadvertently altering others. Rubinstein et al. (2001) considered separable executive processes, and Monsell (2003) connects response-level findings to wider task-set regulation. The strongest conclusion is that task focus and transition costs reflect interacting levels of representation; quantifying their individual contributions remains an experimental problem.
Task-Set Reconfiguration and Its Alternatives
Task-set reconfiguration theories propose that an impending change requires modification of the processing configuration needed for successful performance. Such modification may include enabling a different categorization criterion, retrieving new response mappings, redirecting attention, and changing the expected relevance of stimulus dimensions. Rogers & Monsell (1995) and Meiran (1996) made preparation a central test of this idea, and Rubinstein et al. (2001) analyzed executive control in terms of task-switching processes. A reconfiguration account becomes empirically meaningful when it specifies which operations occur, when they can start, and how their duration changes with preparation interval, sequence predictability, and cue validity. A generic claim that the mind must reset after every switch lacks the specificity needed to distinguish explanations.
Different reconfiguration accounts make different assumptions about whether preparation is complete before the target arrives. If an operation depends on target identity, increasing the cue–target interval may leave a residual cost. If it depends only on knowing which rule applies, informative cues may permit earlier completion. Yet slower switching under both conditions can also reflect residual competition from earlier events. Monsell (2003) surveys the evidence for preparatory and residual components, whereas Vandierendonck et al. (2010) explicitly integrates interference with reconfiguration. Thus, model evaluation requires experiments that distinguish prospective task preparation from retrospective effects of previous trials. The timing of cue encoding, target classification, response selection, and learning must each be considered.
A productive theoretical strategy uses selective interventions to rule out alternatives. Separate cues can remove cue-repetition confounds, congruency conditions can probe response competition, and cue-only trials can isolate some preparatory effects. Yamaguchi & Swainson (2025) found that preparation-only and completed-task history can generate different temporal profiles of subsequent costs. The finding constrains pure one-component accounts but is compatible with several multi-process interpretations. Explanatory adequacy therefore depends on convergent patterns across experimental designs, mathematical model comparisons, and independent replication. Reconfiguration remains a useful theoretical concept because it organizes testable predictions; the evidence does not justify identifying the entire switch cost with one unobserved act of executive resetting.
Priming, Episodic Retrieval, and Learning
Recent stimulus, cue, response, and task events leave traces that can influence the next trial. Repeating an instruction signal may speed cue identification; encountering a familiar target may reinstate a previously associated task response; and a response that was rewarded may acquire temporary priority. Such history effects can mimic some of the latency differences conventionally attributed to top-down executive control. Logan & Bundesen (2003) demonstrated why explicit cueing must account for cue repetition, and Schneider & Logan (2005) formalized a short-term priming alternative to a purely switch-specific process. These models invite the experimenter to vary the overlap between consecutive episodes, not merely the label describing whether a task switched.
Episodic accounts treat a prior trial as a bound event that includes task identity, stimuli, response, and context. Reencountering part of that event may retrieve compatible or incompatible elements. If an old response is retrieved on a trial demanding a different response, performance can slow; if the retrieved event matches current requirements, it can facilitate performance. This perspective places memory retrieval at the center of task control without claiming that all flexible behavior is passive repetition. Egner (2023) proposes context-sensitive episodic reinstatement alongside incremental adjustment based on recent experience. The relative contribution of these mechanisms can be tested using recurrence of context, manipulation of stimulus repetition, and changes in transition probability.
Event-history explanations and intentional control can interact. A participant who knows that a rule is likely to repeat may deliberately prepare it, while recent repetitions also make its retrieval easier. Both pathways improve repeat performance, and separating them requires manipulating instruction and learned statistics independently. Kiesel et al. (2010) reviews the mixture of intentional and automatic contributions to switching. Dreisbach & Fröber (2019) describes how stability and flexibility change with task conditions. The scientific implication is that a switch cost cannot be interpreted as the amount of conscious effort spent changing tasks; it is the behavioral consequence of interacting goals, learned mappings, and recent episodes. Stronger inference requires designs in which model predictions diverge rather than verbal attribution after the experiment.
Backward Inhibition and Task Sequence
Backward inhibition describes a pattern in which returning to a recently abandoned task may be slower than switching to a task that was not used as recently. In a three-task design, the critical comparison is typically a return sequence A–B–A against a sequence C–B–A, with care taken to balance other factors. Mayr & Keele (2000) reported evidence that changing internal constraints on action could involve inhibition of a recently relevant task. The resulting n−2 repetition cost became an influential operational signature for task-level inhibition. The interpretation matters because it proposes that overcoming a previously activated rule can carry a future cost even when that rule becomes appropriate again.
The return cost is not synonymous with an experimentally isolated inhibitory process. A–B–A and C–B–A sequences can differ in episodic similarity, response repetition, expectancy, target repetition, and the opportunity to reactivate an earlier event. Rehearsal and task-selection policies may further affect the outcome. Koch et al. (2010) reviews task inhibition and notes the importance of disentangling it from other sources of interference. Kiesel et al. (2010) places backward inhibition within a wider review of switching mechanisms. Consequently, a credible test should balance transition frequencies and sensory details, avoid unexamined sequence contingencies, and seek independent evidence that an unavailable task set was actively suppressed rather than simply becoming less accessible.
Inhibitory proposals can also explain some asymmetric switching costs, in which returning to a better-practiced task is unexpectedly more costly than switching to a weaker one. However, task strength, response associations, and preparation differences complicate this interpretation. An asymmetry does not independently establish suppression of the dominant task. Vandierendonck et al. (2010) reviews the interplay of mechanisms and shows why a unitary inhibition story is insufficient for all switching observations. A careful synthesis retains the distinction between a measured sequence effect and the process posited to explain it. This approach prevents circular reasoning: inhibition should be supported by manipulations that uniquely affect predicted suppression, not simply renamed whenever performance is slow after a sequence change.
Voluntary Switching and Strategic Choice
Voluntary task switching asks participants to choose between available tasks while following constraints such as approximately equal task frequencies and unpredictable ordering. The paradigm adds an important dimension: the decision to change tasks can itself be measured, rather than determined entirely by an external cue. Arrington & Logan (2004) examined voluntary switching and found a tendency toward task repetition together with measurable switch-related performance costs. The result suggests that task selection and task execution have partially distinct determinants. Choosing a task may require evaluating availability, anticipated effort, or recent history, while executing the chosen task still requires classifying the target and producing an appropriate response.
Freedom of task choice is experimentally constrained. Participants may adopt hidden sequences, favor easier tasks, alternate according to a simple strategy, or choose based on stimulus expectations. As a consequence, selection frequency and reaction-time differences do not transparently reveal preferences for flexibility. The investigator should analyze task choice separately from task-performance time, report achieved frequencies and sequential regularities, and test whether externally imposed conditions reproduce the supposed voluntary advantage. Arrington & Logan (2004) provides a foundational procedure, while Egner (2023) frames strategic control as the regulation of task focus and switching readiness under current demands. A voluntary cost can thus arise from deliberate policy, learned behavior, or selection-based sampling of different trial types.
The voluntary paradigm also clarifies that cognitive control involves deciding when change is useful. In everyday settings, a person may choose to switch goals because the expected value of the alternative rises, interruptions become salient, or the current activity ceases to be productive. Laboratory voluntary switching captures only a restricted analogue of this situation. It offers precise choices between predefined rules, usually with prescribed frequency constraints and trivial rewards. Extrapolating to complex life decisions requires evidence that the same mechanisms generalize across environments. Braver (2012) supplies a broader framework for strategic control allocation, and Dreisbach & Fröber (2019) examines how experimental contexts tune flexibility. These sources support a functional interpretation without collapsing ecological decision-making into simple task-alternation frequency.
Temporal Dynamics, Speed–Accuracy Trade-offs, and Model Testing
Mean reaction time summarizes an entire distribution of trials and can conceal meaningful process differences. A manipulation might selectively slow the longest responses, shift the whole latency distribution, or change error frequency. These outcomes support different mechanistic possibilities even when their mean switch costs are equal. Researchers can examine quantile plots, conditional accuracy functions, trial-by-trial dependence, and model-based estimates of decision processes. Rubinstein et al. (2001) explicitly analyzed components of executive control during switching, while Vandierendonck et al. (2010) compared accounts grounded in reconfiguration and interference. The appropriate analysis depends on what the proposed model predicts: differences in preparatory latency, evidence accumulation, retrieval time, or response caution should produce distinguishable patterns.
Response speed and accuracy must be interpreted together. An instruction emphasizing caution may decrease errors at the price of longer reaction times; a participant who anticipates a switch may deliberately slow all responses. In such a case, a reduced switch cost can be produced by an inflated repeat-trial baseline. Error trials may also occur disproportionately in incongruent stimulus conditions, making reaction-time averages conditional on selective survival of correct responses. An informative report includes accuracy by condition, handling of anticipations and outliers, and sensitivity analyses for trimming choices. Wylie & Allport (2000) makes the measurement problem explicit, and Hedge et al. (2018) shows how apparently strong experimental contrasts can behave differently as individual-difference scores.
A model comparison becomes more decisive when it predicts multiple observables jointly. A reconfiguration model may predict preparation-dependent latency patterns; a priming model may predict cue and stimulus repetition interactions; an inhibition model may predict sequence-dependent return costs. Any one account might fit one average, but a model that simultaneously predicts accuracy, history dependence, and timing across independent paradigms is more constrained. Schneider & Logan (2005) illustrates a detailed priming explanation; Yamaguchi & Swainson (2025) demonstrates how experimentally disentangling cue-only from completed events restricts time-course hypotheses. The aim is an evidentiary architecture in which proposed processes face discriminating tests, with model complexity and uncertainty made explicit rather than hidden behind a convenient psychological label.
Proactive and Reactive Cognitive Control
Proactive control describes the anticipatory maintenance of task-relevant information before competing demands arise. Reactive control describes the recruitment or adjustment of goal information when conflict, uncertainty, or an unexpected event demands intervention. Braver (2012) developed a dual-mechanisms framework in which the balance between these modes depends on context, motivation, capacity, and the opportunities for preparation. In task-switching experiments, informative cues may encourage proactive activation of a new rule, whereas an unexpected target or invalid cue can require reactive adjustment. These categories characterize timing and regulation, not discrete anatomical systems or mutually exclusive psychological types. The same individual may use both forms across different moments of a trial.
Experimental tests must separate the possibility of advance preparation from evidence that preparation actually occurred. A participant may have enough time to prepare but elect not to invest effort when the next task is uncertain, cues are unreliable, or the cost of waiting is low. Conversely, strong stimulus priming may produce fast performance despite little sustained advance maintenance. Meiran (1996) and Altmann (2004) studied preparation processes, while Yamaguchi & Swainson (2025) demonstrated distinct persistence of costs following cue-only and completed trials. Combining cue validity, probability, delay, and performance measures can reveal when anticipatory representations change subsequent behavior. Reports should distinguish an observed preparation benefit from a theoretical inference about proactive control.
The proactive–reactive distinction has implications for learning and intervention. Strategies that maintain rules may help when task demands are stable and predictable; responsiveness to late-arriving evidence becomes critical when the environment changes abruptly. However, training participants to favor one control mode in a laboratory does not ensure useful performance in other contexts. Braver (2012) links control allocation to changing conditions, and Egner (2023) examines how the same control system can adjust focus and readiness from recent experience and contextual cues. The most informative research asks when each policy is advantageous, what resources it requires, and which observable behaviors reveal that participants adopted it. Functional success is evaluated in context rather than by choosing one mode as universally superior.
Context, Expectation, and Adaptive Learning of Control
Task transitions unfold in environments containing regularities. If a task repeats frequently, a participant can learn that persistence is usually rewarded; if changes are common, maintaining readiness for alternatives becomes more useful. The adaptation of control to such transition statistics offers an explanation for why identical stimuli and rules can produce different costs across contexts. Egner (2023) proposes that control settings can be adjusted through incremental experience with recent demands and through retrieval of previously encountered contexts. This perspective integrates control with learning while leaving open the question of which neural and computational pathways implement each adjustment. Empirical tests manipulate environmental contingencies rather than assuming that every trial is independent.
Context-sensitive control can arise at several levels. A location might signal that one task is usually relevant, a color cue might predict a demanding switch, or a sequence pattern might make a particular transition likely. Some effects can be explained by learned attention to cues; others require participants to maintain an explicit goal or infer an underlying state. The distinction is tested by changing context, reversing contingencies, and asking whether the learned policy follows surface features or abstract rules. Egner (2023) connects recency-based adjustments with episodic reinstatement, and Dreisbach & Fröber (2019) treats stability and flexibility as settings shaped by experience. Both sources underscore that adaptive change is measurable only with a specified comparison between alternative environmental histories.
A special challenge arises when experiments make transition probabilities unequal but analyze switch trials as though their baseline expectancy were identical. Switch costs can incorporate surprise and expectancy violation in addition to the operations of changing a rule. Trial-level models should therefore include recent switches, task frequencies, cue probabilities, and the distribution of preparation intervals. An observed probability effect becomes stronger evidence for learned control when it generalizes beyond identical stimuli and responses or persists after suitable history controls. Kiesel et al. (2010) reviews contextual influences on switching, while Egner (2023) articulates candidate learning principles. This approach enables a functional interpretation of control policies while maintaining the distinction between predictive regularity and causal explanation.
Neural Systems of Task Rules and Flexible Coordination
Cognitive control engages distributed neural systems that participate in goal representation, contextual integration, action preparation, and evaluation of outcomes. Sakai (2008) reviews the role of prefrontal cortex in task-set organization, including evidence from task-rule maintenance and changes in task demands. A task cue can evoke neural patterns related to the currently relevant rule before the target arrives, offering a way to connect behavioral preparation with measured brain activity. Yet a regional response during a switch is not direct proof that the region performs a unique reconfiguration operation. Neural activation can reflect increased difficulty, attention to a cue, response uncertainty, memory retrieval, or monitoring as well as control itself.
Functional imaging, electrophysiology, lesion studies, and multivariate decoding provide different kinds of constraint. Imaging can identify networks whose activity covaries with task contexts; timing-sensitive recordings can help separate cue-period from target-period operations; causal perturbation or focal neurological injury can test whether particular systems are required under specified conditions. Each approach has limits, including correlational interpretation, temporal and spatial resolution, compensatory strategies, and task-specific confounds. Sakai (2008) describes how prefrontal representations organize task behavior. Chuikova et al. (2025) synthesized multiple cognitive-flexibility paradigms and reported convergent and distinct neural correlates, highlighting the danger of assuming that every laboratory flexibility measure recruits an identical neural circuit.
Neural evidence is most informative when theory specifies representations or computations rather than simply predicting greater activation on more difficult trials. For example, an account can predict that informative cues permit earlier decoding of the next rule, that interference strengthens competition among rule-sensitive patterns, or that context recurrence reinstates an earlier configuration. Independent neural and behavioral measurements can then test whether putative preparatory representations predict subsequent transition performance. Egner (2023) discusses the potential algorithmic and neural implementations of focus and switching readiness, and Sakai (2008) provides a foundational survey of task sets. The current evidence supports distributed coordination of cognitive control, with specific mechanisms varying by paradigm and interpretation.
Development, Aging, and Population Variation
Task switching changes across development and adulthood as participants acquire task knowledge, improve the use of instructions, and experience changes in processing speed, memory, and control demands. Age comparisons require care because an older participant may be slower in both switch and repeat conditions, and a child may differ from an adult in understanding or remembering the rules. Wasylyshyn et al. (2011) meta-analyzed aging and task switching, examining how conclusions vary with the distinction between global and local costs. These findings encourage researchers to report absolute latency, accuracy, baseline differences, and the reliability of each comparison. Age is a descriptor of a population, not a mechanistic explanation for a measured performance difference.
Developmental and aging studies frequently involve different stimulus familiarity, strategy selection, and motivation. If a task relies heavily on rapid reading, sensory discrimination, or motor speed, group differences in those components can influence the estimated switching cost. The same nominal protocol can therefore recruit unequal processes across populations. Dajani & Uddin (2015) reviews cognitive flexibility in developmental and clinical neuroscience, emphasizing a multidimensional construct. Karbach & Kray (2009) investigated age differences in the transfer of task-switching training, providing a controlled setting in which the distinction between practiced performance and broader executive change could be examined. Careful interpretation requires demonstrating that the groups encounter comparable instructions, error contingencies, and task demands.
Population variation also includes the influence of education, language, familiarity with digital interfaces, fatigue, and testing environment. Many frequently cited laboratory studies rely on restricted convenience samples, so generalization to other cultural and occupational contexts requires additional evidence. This is especially important when switching performance is used to infer everyday adaptability or a person's ability to manage complex work. Friedman & Miyake (2017) discusses individual differences in executive functions, and Hedge et al. (2018) warns that robust experimental effects need not translate into reliable individual rankings. A defensible developmental synthesis integrates behavioral design, population characteristics, and measurement invariance rather than treating a universal switch-cost formula as equally diagnostic across all ages and contexts.
Neuropsychological and Clinical Interpretation
Clinical and neuropsychological research sometimes employs rule-switching, set-shifting, and flexibility tasks to characterize difficulties in adapting behavior to changing instructions. Such tasks may be informative in the assessment of cognitive functioning, but their scores are influenced by many processes: comprehension, working memory, sensory discrimination, motor speed, practice, motivation, and error monitoring. Dajani & Uddin (2015) discusses the relevance of flexibility to developmental and clinical neuroscience, while Sakai (2008) connects task sets to prefrontal neural systems. A laboratory switching cost is therefore one task-derived measure among others, and its clinical meaning depends on normative evidence, differential interpretation, and the person's functional context.
Laboratory task switching is not identical to every form of clinical set shifting. Neuropsychological instruments may demand discovery of a hidden classification rule, reversal after feedback, or sequential connection of visual symbols. Explicitly cued experiments often provide the next rule directly, whereas reversal-learning procedures require inference from outcomes. These paradigms can involve overlapping cognitive resources while remaining functionally distinct. Chuikova et al. (2025) found that different flexibility paradigms share some neural associations and differ in others, reinforcing the need for careful operational comparison. Miyake et al. (2000) similarly demonstrates the value of separating correlated executive functions. A description that treats all these tasks as interchangeable measures of one deficit loses theoretically relevant variation.
An observed performance difference cannot, on its own, diagnose a disorder, localize a lesion, or justify an intervention. Scientific and clinical interpretation requires validated instruments, appropriate population norms, information about concurrent symptoms and medication, and, when applicable, assessment by qualified professionals. The research reviewed here informs the interpretation of task-switching paradigms rather than individualized clinical management. Dajani & Uddin (2015) supports studying flexibility through developmental and clinical mechanisms, and Hedge et al. (2018) underscores the psychometric limitations of difference scores. Future clinical work benefits from preregistered hypotheses, convergent tests, longitudinal measurement, and outcomes tied to everyday function. The relevant goal is evidence that a measure adds valid information beyond simpler indicators of processing speed or general task performance.
Executive Functions, Individual Differences, and Construct Validity
Executive functioning is commonly discussed in terms of related capacities such as shifting, updating, and inhibition. Miyake et al. (2000) used multiple tasks and latent-variable analysis to examine both shared and distinguishable components. The method is consequential for task-switching research because a single observed score reflects its specific instructions, stimuli, and response demands as well as the intended construct. Friedman & Miyake (2017) reviewed subsequent evidence about unity and diversity, highlighting a common executive component alongside process-specific variation. Thus, the fact that task switching is a familiar executive-function paradigm does not license the inference that its numerical switch cost measures executive functioning as a whole.
Construct validity demands an explicit account of how task measures relate to a hypothetical ability and to one another. If two tasks both require flexible rule use but differ in stimulus modality, cueing, interference, and scoring, weak correlations may indicate measurement error, distinct mechanisms, or both. A sound psychometric program uses sufficiently reliable scores, repeated sessions, and multiple task indicators to estimate shared variance. Hedge et al. (2018) shows that group-level experimental reliability of effects can coexist with poor reliability for ranking individual participants. Consequently, dramatic and reproducible average switch costs can have limited usefulness as individual predictors. Difference scores require especially careful evaluation because subtracting two measurements can amplify error relative to the variance of interest.
The relationship between executive tests and real-world behavior also requires independent validation. Predicting an individual's performance when planning complex work, changing strategies after failure, or managing competing responsibilities calls for longitudinal and ecological criteria, not a rhetorical analogy to a cue-switching trial. The strength of the laboratory paradigm is its controlled comparison; its limitation is the small subset of natural goals and environmental contingencies it represents. Miyake et al. (2000) and Friedman & Miyake (2017) supply models for separating components through latent measurement, while Dajani & Uddin (2015) illustrates the broader conceptual scope of flexibility. The most useful result is a carefully defined relationship between operational tasks and a theoretical construct, reported with uncertainty and independent validation.
Training, Practice, and the Problem of Transfer
Repeated task-switching practice often improves performance on the trained activities. Participants can learn cues, remember response mappings more efficiently, anticipate common transitions, and develop strategies that reduce unnecessary processing. These improvements are real within the practiced task and should be measured separately from claims about generalized cognitive flexibility. Karbach & Kray (2009) investigated near and far transfer of executive-control training across age groups, contributing to the debate about whether practice can generalize beyond its immediate setting. Karbach et al. (2010) examined the role of verbal processes in transfer of task-switching training among older adults. Theoretical interpretation depends on whether improvements appear on structurally new tasks, persist over time, and exceed appropriate active-control conditions.
Near transfer refers to improvement on activities sharing substantial operations with training; far transfer concerns benefits in meaningfully different abilities or outcomes. A task-switching intervention that teaches participants to alternate familiar rules more quickly may alter practiced cue translation without changing capacity to manage complex external interruptions. Strong tests require preregistered outcomes, adequate randomized controls, blinded scoring when feasible, delayed follow-up, and analysis of baseline differences. Karbach & Kray (2009) provides an influential experimental example, while Hedge et al. (2018) warns that low reliability of individual-difference scores can make intervention effects difficult to interpret. The reader should distinguish task-specific learning, improvements in near-transfer probes, and broad behavioral benefits rather than combining them under the label of better executive control.
Training mechanisms may involve learned task sequences, refined attentional priorities, reduced uncertainty, or greater familiarity with testing procedures. Demonstrating the mechanism calls for transfer patterns that differ across these possibilities. If performance gains vanish when cues, stimuli, or task contingencies change, an account based on narrow procedural learning becomes more plausible. If gains extend across independent paradigms with minimal feature overlap, a broader control explanation gains support. Braver (2012) offers hypotheses about proactive and reactive control strategies that training might influence, and Egner (2023) describes learning principles governing control settings. The literature supports careful investigation of transfer conditions, while broad practical promises should rest on replicated changes in independently measured outcomes.
Everyday Multitasking and Human–Artificial Work Environments
Everyday multitasking encompasses interruptions, resumption of unfinished goals, navigation among interfaces, and competition for attention across several ongoing activities. Laboratory task switching contributes precise concepts to this domain: rule uncertainty, preparation time, interference, and the cost of reestablishing a goal. However, switching between two binary classifications differs substantially from resuming a complex written document after an interruption. Natural tasks have nested goals, long memory demands, external reminders, social priorities, and variable rewards. Monsell (2003) supplies the experimental foundations, while Egner (2023) explains why focus and switch readiness may require distinct regulatory settings. Their principles suggest questions about interface design, though direct application requires field evidence.
A digital tool can reduce some control demands by making the current task explicit, preserving the state of an unfinished activity, and minimizing irrelevant cues. Conversely, frequent notifications or ambiguous workflow states can create repeated shifts in goal and attentional orientation. Such design hypotheses must be evaluated through controlled usability studies, interruption-resumption experiments, and behavioral outcomes such as error rate and completion time. A low laboratory switch cost cannot establish that an individual manages real-world digital work effectively, and a highly automated interface may improve performance simply by lowering memory demand. Hedge et al. (2018) cautions against weak individual-level inference from experimental contrasts; Braver (2012) provides a framework for asking when proactive task maintenance becomes advantageous.
Human–artificial collaboration introduces additional possibilities, including delegated monitoring, automated sequencing, and external maintenance of instructions. The psychological question concerns how such arrangements alter a human participant's task demands, control allocation, and ability to resume or revise plans. A system that remembers context may reduce the need to reconstruct a task set; an unreliable agent may increase supervision and interruption costs. These are experimentally testable hypotheses rather than established universal consequences of artificial agents. Egner (2023) identifies control principles relevant to such questions, while Friedman & Miyake (2017) underscores the importance of distinguishing executive constructs from task-specific performance. Research should evaluate actual interaction histories, reliability, user goals, and population differences before drawing claims about long-term cognitive transformation.
Evidence Integration, Methodological Limits, and Open Problems
The strongest established conclusion across decades of task-switching research is that changing a currently relevant task rule often affects speed and accuracy, and that the magnitude and form of the effect depend on preparation, cues, recent events, and interference. Rogers & Monsell (1995), Meiran (1996), and Monsell (2003) established foundational demonstrations; Kiesel et al. (2010) and Vandierendonck et al. (2010) synthesized multiple competing mechanisms; Yamaguchi & Swainson (2025) refined the causal analysis of preparation using cue-only trials. This combination of original experiments, theoretical confrontation, and newer experimental replications supports a multi-process interpretation. It does not supply a universally fixed decomposition of every switch cost into one percentage attributable to reconfiguration, another to inhibition, and another to priming.
Several methodological uncertainties remain central. Task impurity complicates the interpretation of behavioral differences, cue and stimulus repetitions can produce hidden confounds, and laboratory designs vary widely in transition probabilities and preparation schedules. Hedge et al. (2018) establishes a separate challenge for individual-difference measurement: a strong within-person effect may yield an unstable between-person score. Neural studies face additional questions about reverse inference and the specificity of regions activated by difficult switching conditions. Sakai (2008) reviews neural task-set accounts, and Chuikova et al. (2025) shows both shared and distinct associations across flexibility tasks. Reviews can organize the evidence, but a quantitative consensus requires explicit inclusion criteria, harmonized contrasts, and consideration of publication bias.
Priority research questions concern how task-set representations are encoded, how episodic retrieval and forward-looking preparation are weighted, which mechanisms explain sequence-sensitive inhibition, and how control policies adapt across stable and volatile environments. The emergence of richer neural recordings and temporally precise analyses offers opportunities to test stage-specific predictions. Equally important is the development of reliable multi-task measurement capable of predicting functioning outside tightly constrained laboratory procedures. Egner (2023) proposes learning-based coordination of focus and switching readiness; Friedman & Miyake (2017) outlines the rationale for latent executive constructs; Yamaguchi & Swainson (2025) indicates that even apparently basic switching phenomena may contain multiple time courses. Scientific progress follows from convergent tests that force alternatives to make different observable predictions.
Conclusion
Cognitive control organizes the use of task-relevant information so that behavior remains responsive to goals and changing conditions. Task switching makes this organization experimentally accessible through precisely specified changes in rules, cues, stimuli, responses, and temporal context. The central body of evidence demonstrates reliable effects of task alternation within many paradigms while revealing multiple contributing processes: preparation, task-set adjustment, retrieval of recent events, interference between competing mappings, response selection, and learned expectations. Monsell (2003), Kiesel et al. (2010), and Egner (2023) place these findings in an integrated explanatory landscape. The interpretive task is to identify which operations a particular experimental contrast supports rather than turn one observable latency difference into a complete psychological theory.
The history of the field shows a productive progression from the discovery of switching costs to the decomposition of their sources. Reconfiguration models prompted studies of advance preparation; priming accounts demanded better cue controls; inhibition proposals made sequence history experimentally visible; voluntary task choice and context learning expanded the meaning of flexible control. Logan & Bundesen (2003), Mayr & Keele (2000), Arrington & Logan (2004), and Schneider & Logan (2005) illustrate these distinct research directions. Contemporary results using cue-only trials confirm the value of designs that isolate preparation from completed action Yamaguchi & Swainson (2025). Neural and developmental studies broaden the picture while preserving the need to examine task specificity, measurement, and alternative explanations.
The resulting scientific synthesis treats cognitive control as a context-sensitive coordination problem supported by multiple interacting processes whose contributions vary with the task and participant. Its future progress depends on direct replication, explicit model comparison, reliable measurement across tasks, careful population sampling, and evidence of transfer to consequential everyday behavior. Hedge et al. (2018) identifies why validity must be tested at the intended level of inference, and Friedman & Miyake (2017) explains the value of separating common and distinctive executive components. The most defensible application of task-switching research is to specify conditions that help people maintain appropriate goals, recognize when change is required, and accomplish transitions with measurable accuracy and efficiency. This monograph thereby situates a well-developed experimental field within the broader scientific study of goal-directed psychological functioning.
References
- Altmann, E. M. (2004). Advance preparation in task switching: What work is being done? Psychological Science, 15(9), 616–622. https://doi.org/10.1111/j.0956-7976.2004.00729.x
- Arrington, C. M., & Logan, G. D. (2004). The cost of a voluntary task switch. Psychological Science, 15(9), 610–615. https://doi.org/10.1111/j.0956-7976.2004.00728.x
- Braver, T. S. (2012). The variable nature of cognitive control: A dual mechanisms framework. Trends in Cognitive Sciences, 16(2), 106–113. https://doi.org/10.1016/j.tics.2011.12.010
- Chuikova, Z. V., Filatov, A. A., Faber, A. Y., & Arsalidou, M. (2025). Mapping common and distinct brain correlates among cognitive flexibility tasks: Concordant evidence from meta-analyses. Brain Imaging and Behavior, 19(1), 50–71. https://doi.org/10.1007/s11682-024-00921-7
- Dajani, D. R., & Uddin, L. Q. (2015). Demystifying cognitive flexibility: Implications for clinical and developmental neuroscience. Trends in Neurosciences, 38(9), 571–578. https://doi.org/10.1016/j.tins.2015.07.003
- Dreisbach, G., & Fröber, K. (2019). On how to be flexible (or not): Modulation of the stability-flexibility balance. Current Directions in Psychological Science, 28(1), 3–9. https://doi.org/10.1177/0963721418800030
- Egner, T. (2023). Principles of cognitive control over task focus and task switching. Nature Reviews Psychology, 2(11), 702–714. https://doi.org/10.1038/s44159-023-00234-4
- Friedman, N. P., & Miyake, A. (2017). Unity and diversity of executive functions: Individual differences as a window on cognitive structure. Cortex, 86, 186–204. https://doi.org/10.1016/j.cortex.2016.04.023
- Hedge, C., Powell, G., & Sumner, P. (2018). The reliability paradox: Why robust cognitive tasks do not produce reliable individual differences. Behavior Research Methods, 50(3), 1166–1186. https://doi.org/10.3758/s13428-017-0935-1
- Karbach, J., & Kray, J. (2009). How useful is executive control training? Age differences in near and far transfer of task-switching training. Developmental Science, 12(6), 978–990. https://doi.org/10.1111/j.1467-7687.2009.00846.x
- Karbach, J., Mang, S., & Kray, J. (2010). Transfer of task-switching training in older age: The role of verbal processes. Psychology and Aging, 25(3), 677–683. https://doi.org/10.1037/a0019845
- Kiesel, A., Steinhauser, M., Wendt, M., Falkenstein, M., Jost, K., Philipp, A. M., & Koch, I. (2010). Control and interference in task switching—A review. Psychological Bulletin, 136(5), 849–874. https://doi.org/10.1037/a0019842
- Koch, I., Gade, M., Schuch, S., & Philipp, A. M. (2010). The role of inhibition in task switching: A review. Psychonomic Bulletin & Review, 17(1), 1–14. https://doi.org/10.3758/PBR.17.1.1
- Logan, G. D., & Bundesen, C. (2003). Clever homunculus: Is there an endogenous act of control in the explicit task-cuing procedure? Journal of Experimental Psychology: Human Perception and Performance, 29(3), 575–599. https://doi.org/10.1037/0096-1523.29.3.575
- Mayr, U., & Keele, S. W. (2000). Changing internal constraints on action: The role of backward inhibition. Journal of Experimental Psychology: General, 129(1), 4–26. https://doi.org/10.1037/0096-3445.129.1.4
- Meiran, N. (1996). Reconfiguration of processing mode prior to task performance. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22(6), 1423–1442. https://doi.org/10.1037/0278-7393.22.6.1423
- Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex frontal lobe tasks: A latent variable analysis. Cognitive Psychology, 41(1), 49–100. https://doi.org/10.1006/cogp.1999.0734
- Monsell, S. (2003). Task switching. Trends in Cognitive Sciences, 7(3), 134–140. https://doi.org/10.1016/S1364-6613(03)00028-7
- Rogers, R. D., & Monsell, S. (1995). Costs of a predictable switch between simple cognitive tasks. Journal of Experimental Psychology: General, 124(2), 207–231. https://doi.org/10.1037/0096-3445.124.2.207
- Rubinstein, J. S., Meyer, D. E., & Evans, J. E. (2001). Executive control of cognitive processes in task switching. Journal of Experimental Psychology: Human Perception and Performance, 27(4), 763–797. https://doi.org/10.1037/0096-1523.27.4.763
- Sakai, K. (2008). Task set and prefrontal cortex. Annual Review of Neuroscience, 31, 219–245. https://doi.org/10.1146/annurev.neuro.31.060407.125642
- Schneider, D. W., & Logan, G. D. (2005). Modeling task switching without switching tasks: A short-term priming account of explicitly cued performance. Journal of Experimental Psychology: General, 134(3), 343–367. https://doi.org/10.1037/0096-3445.134.3.343
- Swainson, R., Prosser, L. J., & Yamaguchi, M. (2024). Preparing a task is sufficient to generate a subsequent task-switch cost affecting task performance. Journal of Experimental Psychology: Learning, Memory, and Cognition, 50(1), 39–51. https://doi.org/10.1037/xlm0001277
- Vandierendonck, A., Liefooghe, B., & Verbruggen, F. (2010). Task switching: Interplay of reconfiguration and interference control. Psychological Bulletin, 136(4), 601–626. https://doi.org/10.1037/a0019791
- Wasylyshyn, C., Verhaeghen, P., & Sliwinski, M. J. (2011). Aging and task switching: A meta-analysis. Psychology and Aging, 26(1), 15–20. https://doi.org/10.1037/a0020912
- Wylie, G., & Allport, A. (2000). Task switching and the measurement of switch costs. Psychological Research, 63(3–4), 212–233. https://doi.org/10.1007/s004269900003
- Yamaguchi, M., & Swainson, R. (2025). Does preparation generate the cost of task switching? A recipe for a switch cost after cue-only trials. Psychological Research, 89, Article 72. https://doi.org/10.1007/s00426-025-02107-2