Automatic vs Controlled Processing: Effort, Practice, and Attention

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

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

Automatic and controlled processing describe how cognitive operations are initiated, coordinated, and regulated as people perceive information, select responses, learn skills, and pursue goals. Automatic processing typically relies on well-established associations or practiced procedures that can operate with relatively little moment-to-moment supervision. Controlled processing recruits flexible, goal-sensitive coordination when a task is unfamiliar, ambiguous, conflicting, or requires an adjustment to what would otherwise occur. The distinction is useful for explaining why familiar actions often feel easy, why incompatible information creates interference, and why practice can change both speed and attentional demands. Schneider and Shiffrin (1977); Moors and De Houwer (2006).

These terms describe properties of processing, not two mutually exclusive kinds of person or two sealed compartments of the mind. A practiced routine can be influenced by a current goal, and a deliberate decision can incorporate rapidly retrieved information. How a process unfolds depends on learning history, task instructions, stimulus–response consistency, attention, competing demands, and environmental context. Contemporary accounts therefore evaluate automaticity along several dimensions rather than treating it as an all-or-nothing label. Cohen, Dunbar, and McClelland (1990); Melnikoff and Bargh (2018).

What Is Automatic Processing?

Automatic processing is a cognitive operation that, under particular learned or contextual conditions, can be triggered and carried out with reduced need for active supervisory control. An experienced reader may extract the meaning of a common printed word rapidly even while attempting to concentrate on its ink color. A skilled typist may execute familiar letter sequences without consciously planning each finger movement. These observations motivate the concept, although neither speed nor subjective ease alone establishes that an operation is automatic. Shiffrin and Schneider (1977); Logan (1988).

The word automatic can refer to several partially independent properties: an operation may begin without a deliberate decision, be difficult to stop after initiation, remain efficient under competing task demands, proceed rapidly, or occur without conscious access to some of its intermediate steps. The absence of any one characteristic does not prove the absence of the others. In particular, a process may be automatic in its initiation yet remain sensitive to goals, attention, and context. Moors and De Houwer (2006).

Automatic processing is also task-relative. Reading is highly practiced for a literate adult, but extracting meaning from unfamiliar technical notation may demand careful analysis. The same individual can process a stimulus efficiently in one situation and with substantial effort in another. The relevant unit of analysis is the operation performed under specified conditions, not the entire activity labeled 'reading,' 'driving,' or 'thinking.' Cohen, Dunbar, and McClelland (1990).

What Is Controlled Processing?

Controlled processing refers to goal-directed regulation that configures a task, selects relevant information, coordinates alternative actions, monitors conflict or errors, and modifies behavior when current circumstances require it. It is especially visible when instructions are new, a learned response is inappropriate, or multiple operations compete for limited processing resources. Controlled operations often impose measurable time costs and can be sensitive to concurrent tasks, although the amount of effort needed varies with expertise and task structure. Schneider and Shiffrin (1977); Kiesel et al. (2010).

A person who has learned to touch-type may still use controlled processing to decide what sentence to write, correct an unexpected mistake, or adapt to a different keyboard layout. Control is therefore compatible with highly automatic component operations. Likewise, controlled processing is not defined by a feeling of strain: some deliberate judgments are made efficiently, while an unwanted learned response can be unusually persistent. Hommel (2019).

In experimental cognitive psychology, cognitive control is inferred from patterns of task performance rather than directly observed as a single mental substance. Researchers change the rules, introduce conflict, manipulate preparation time, or add secondary tasks, and then examine the resulting response times, errors, and adaptation. Each manipulation supports a more specific inference than a general claim that a participant 'used control.' Kiesel et al. (2010).

Automatic vs Controlled Processing: The Core Differences

Initiation and intention

A controlled operation commonly begins because the person adopts a specific goal or rule, such as sorting cards according to a newly announced category. An automatic operation may be elicited by familiar stimulus features without a fresh intention to perform that particular component. Yet goal states can alter which learned response becomes likely, and an intentional task can contain automatic stages. Intentionality is best measured at the level of a defined operation and occasion. Moors and De Houwer (2006).

Effort, capacity, and competition

Automatic processes are often described as economical because extensive practice can reduce interference from a second task. Controlled coordination is more likely to produce dual-task costs when two activities draw on overlapping operations or response-selection requirements. The size of those costs depends on the pairing of tasks, their timing, their modalities, and the stage of practice. Efficiency cannot be inferred from the mere fact that two actions happen close together. Hazeltine, Ruthruff, and Remington (2006).

Speed and accuracy

Well-practiced responses are frequently faster and less variable. Rapid responses, however, can arise from simple sensory conditions, response expectations, or a strategy that sacrifices accuracy. Conversely, deliberately controlled operations may be fast when the rule is clear and well supported. Researchers jointly examine reaction time and accuracy, and often assess performance across many trials, to avoid equating speed with automaticity. Logan (1988); Schneider and Shiffrin (1977).

Awareness and subjective experience

A person may become aware of the product of an automatic operation without recognizing every processing step that produced it. Similarly, consciously noticing a stimulus does not guarantee voluntary control of the response it activates. Research on word processing and Stroop interference challenges the simple equation of automatic with unconscious. Claims about unconscious processing require methods that assess awareness independently of task effects. Tzelgov, Porat, and Henik (1997); Moors and De Houwer (2006).

Flexibility and context dependence

Controlled processing is particularly important when a rule changes or a familiar association must be overridden. Learned processing can nevertheless remain context-sensitive: environmental cues, current goals, and recent experience affect what is retrieved and how competing pathways interact. Even highly practiced behavior may require control when errors, exceptions, or hazards arise. Cohen, Dunbar, and McClelland (1990); Wood and Rünger (2016).

The Historical Origins of the Distinction

The systematic study of automatic and controlled processes grew out of questions about selective attention and visual search. In influential experiments, Walter Schneider and Richard Shiffrin asked participants to detect targets among distractors while varying the relationship between stimuli and required responses. With consistent mapping, a stimulus category was reliably a target or a distractor across trials. With varied mapping, an item could play different roles depending on the trial. Extensive practice under consistent mapping produced detection patterns that the authors interpreted as the development of automatic processing, whereas varied mapping more often maintained demanding controlled search. Schneider and Shiffrin (1977); Shiffrin and Schneider (1977).

These experiments were theoretically decisive because they studied a change in processing within a learning environment. A task's attentional demands could not be read directly from the stimulus; they depended on consistent practice, the mapping of categories, and the broader sequence of trials. The original theory provided a productive contrast between learned automatic activation and temporary capacity-limited control, while later accounts revised the assumption that the two are sharply separated. Schneider and Shiffrin (1977); Cohen, Dunbar, and McClelland (1990).

The well-known Stroop experiments preceded this framework. In the color–word task, participants name the ink color while attempting to disregard an incongruent written color word. Reading a familiar word is strongly practiced, and the meaning of that word can compete with the goal of naming its ink. This asymmetry offers a clear example of how a learned operation may interfere with goal-directed selection. The Stroop effect has multiple contributing processes, so it should not be interpreted as a pure measurement of one automatic reading mechanism. Stroop (1935); MacLeod (1991).

How Practice Changes Processing

Practice generally improves performance by changing the information and operations available to the learner. Early in acquiring a new procedure, a person may need to maintain instructions, evaluate each step, correct errors, and explicitly select the next action. As mappings become stable and experience accumulates, relevant responses can be retrieved or coordinated with less active reconstruction. Faster performance may therefore reflect both greater knowledge and a different organization of processing. Logan (1988); Chein and Schneider (2005).

Practice does not guarantee unrestricted transfer. A highly efficient response to one stable set of inputs may be poorly suited to a new mapping, an unusual exception, or another task that shares only superficial features. The conditions of learning influence what becomes accessible later; consistent practice produces different opportunities from constantly changing stimulus–response relationships. An improvement on a trained laboratory task is therefore evidence for learning in that task before it is evidence for a broadly improved attentional capacity. Schneider and Shiffrin (1977); Logan (1988).

Consistent mappings and learned expectations

When the same stimulus repeatedly calls for the same response, the cognitive system can strengthen an efficient association. In consistent-mapping visual search, items reliably retain their target or distractor roles, and performance can become less dependent on display size after training under the relevant conditions. Varied mapping prevents a comparably simple, stable response association and can preserve the need to apply a rule on the current trial. The exact pattern depends on the stimuli and experimental procedure. Schneider and Shiffrin (1977).

Instance retrieval and memory-based automatization

Gordon Logan's instance theory explains a transition from algorithmic processing to retrieval from memory. Each encounter with a task can leave an episodic instance; with repetition, retrieving a stored answer can compete successfully with computing an answer anew. The theory predicts characteristic improvements in response speed and variability under suitable practice conditions. It also shows why acquired automaticity can be specific to particular materials and stimulus combinations. Logan (1988).

This account concerns the mechanisms of skill learning and retrieval. A separate question, owned by memory research, concerns the broader organization and forgetting of stored representations. For automatic versus controlled processing, the central issue is how retrieval alters the attentional organization of present performance.

Graded pathways and parallel distributed processing

Cohen, Dunbar, and McClelland proposed that processing pathways acquire different strengths through experience. A strongly practiced pathway, such as word reading, may become influential even when a task goal prioritizes another response. Attentional control can bias processing toward the goal-relevant pathway. Under this model, automaticity emerges gradually, and automatic and controlled influences interact within the same network. Cohen, Dunbar, and McClelland (1990).

The account is valuable because it explains why practice changes the competition between operations without requiring a complete switch from one processing system to another. It also clarifies that controlling a learned response and eliminating that response's activation are different accomplishments. A person can select the correct ink color while still showing measurable interference from the word.

What Experiments Reveal About Attention and Automaticity

Visual search: consistent versus varied mapping

Visual search tasks allow researchers to manipulate the number and identity of distractors, the target rule, response mapping, and practice schedule. Early automaticity experiments found that practice with consistent mappings could substantially change target detection, whereas varied mappings often demanded more capacity-limited search. Researchers infer processing characteristics from changes in speed, accuracy, display-size functions, and sensitivity to other task demands. A relatively flat search slope is informative within a defined paradigm, although by itself it does not prove that attention is completely absent. Schneider and Shiffrin (1977); Shiffrin and Schneider (1977).

The Stroop effect: practiced response and goal conflict

In a classic Stroop trial, a participant may see the word GREEN written in red ink and be instructed to say 'red.' Word meaning can influence response selection even when reading is irrelevant to the task. The resulting conflict is a major reason researchers study the Stroop effect in relation to automatic processing. Yet interference varies with task demands, stimulus proportion, attention, and other methodological factors. Modern research treats Stroop performance as the result of interacting attentional and processing mechanisms rather than a transparent gauge of an unchangeable automatic reflex. Stroop (1935); MacLeod (1991); Cohen, Dunbar, and McClelland (1990).

Dual-task performance: how much capacity is occupied?

To test whether a practiced operation places less demand on attention, investigators may ask participants to perform it alongside a second task. Researchers compare single-task and dual-task response times and accuracy, sometimes manipulating the time between stimulus onsets. A decrease in interference after practice supports an efficiency claim for the tested combination. Residual costs, however, depend on what tasks are paired and which processing stages overlap. Hazeltine, Ruthruff, and Remington (2006).

One influential finding is that changing the input and output modality pairings of two tasks can strongly alter interference even after substantial practice. This complicates a simple formula in which every controlled operation consumes a single universal store of attention. Some task combinations remain difficult because their content and response-selection operations interfere with one another. Automaticity therefore needs an operational definition within a precise experimental arrangement. Hazeltine, Ruthruff, and Remington (2006).

Task switching: preparation, interference, and reconfiguration

Switching between tasks requires coordination of the currently relevant rule and management of competing task settings. Researchers often observe a switch cost: performance on a switch trial is slower or less accurate than on a repetition trial. The size and interpretation of this cost depend on preparation, cueing, stimulus–response overlap, residual activation, and other design choices. Practice can make some operations more efficient while leaving specific switch-related costs. These findings demonstrate a persistent interaction between learned associations and flexible control. Kiesel et al. (2010).

Task switching in controlled laboratory experiments is distinct from the wider social and technological question of frequent switching between digital media. The latter involves goals, platforms, habits, notifications, and ecological context. The two domains inform one another, but they should not be treated as interchangeable evidence.

Neuroimaging of practice-related change

Functional neuroimaging research has examined how neural activity changes as people practice cognitive tasks. A combined experimental and meta-analytic investigation by Chein and Schneider evaluated whether extensive practice reduces engagement of broad control-related networks and whether distinct networks emerge. This work offers a neural perspective on changing task demands, while regional activation differences do not independently identify a process as automatic or controlled. Interpretation depends on the study design, task contrasts, and model of learning. Chein and Schneider (2005).

Is Automatic Processing Always Unconscious or Uncontrollable?

Automaticity has often been associated with lack of intention, lack of awareness, limited control, and efficiency. The conceptual review by Moors and De Houwer demonstrates why these properties must be separated. A process can start without deliberate intention but still be modulated by current goals. It may be efficient yet consciously experienced. It can also be difficult to interrupt after initiation while remaining sensitive to attention before initiation. Moors and De Houwer (2006).

Experimental studies on word processing likewise complicate the assumption that learned reading proceeds identically regardless of awareness. Tzelgov and colleagues studied relations between reports of word meaning and Stroop interference, illustrating that conclusions about consciousness depend on how awareness is measured. This is evidence about a specific paradigm and theoretical interpretation; it is not a general verdict on all unconscious processing. Tzelgov, Porat, and Henik (1997).

The modern debate over dual-process taxonomies reinforces this caution. Melnikoff and Bargh argued that grouping all processes into two categories defined by paired attributes can obscure the many combinations that actually occur. Hommel similarly emphasized the intertwined contributions of learned habits and intentional action planning. The most informative scientific question is which particular operation exhibits which properties, under which conditions, and according to which independent tests. Melnikoff and Bargh (2018); Hommel (2019).

Automatic Processing, Habits, Skills, and Executive Control

Automaticity, skill, and habit overlap, but each answers a different research question. Automaticity describes how an operation is initiated and executed with particular attentional or control demands. Skill refers to acquired proficiency in accomplishing a task, which may include automatic components and strategic control. Habit research often focuses on repeated context–response patterns and their relationship to goals. These constructs can be studied together without assuming they are identical. Wood and Rünger (2016); Logan (1988).

Executive control is involved in maintaining goals, managing competing task rules, and altering responses to changing demands. Its contribution can increase when a familiar routine encounters an unexpected exception. A well-learned sequence of driving movements, for example, may be executed efficiently while the driver simultaneously monitors the road and can redirect action when an obstacle appears. This example illustrates a possible organization of control, not a claim that driving safely can be delegated to fully automatic behavior. Kiesel et al. (2010).

Habits may support intended goals in a stable setting and interfere with those goals when circumstances change. Reviews of habitual behavior describe its dependence on recurring contexts and learned cue–response associations, alongside interaction with deliberate action. The relevant distinction is whether the present context supports a useful response or elicits a response that needs reconsideration. Wood and Rünger (2016).

Mental Effort: What It Can and Cannot Tell Us

The experience of effort can signal that a task is demanding, unfamiliar, or difficult to coordinate. Yet effort is a subjective report influenced by task meaning, motivation, expectations, and individual experience. A person can report substantial effort while performing accurately, or little effort while overlooking errors. Controlled processing and mental effort therefore overlap without being identical psychological measurements. In cognitive experiments, they should be investigated using converging behavioral and task-demand evidence. Moors and De Houwer (2006); Kiesel et al. (2010).

Effort and cognitive load are related but have different roles in explanation. Cognitive load concerns demands placed on limited processing resources under specified conditions; automaticity concerns the practiced organization and control properties of component operations. Learning a stable procedure may reduce one source of workload, while understanding a genuinely difficult new problem can still require extensive controlled reasoning. Reducing felt strain is therefore not a universal measure of cognitive improvement.

How Scientists Measure Automatic and Controlled Processing

No single test directly reveals a complete internal 'automaticity score.' Researchers build operational definitions around a mechanism and evaluate predictions across converging tasks. Measures commonly include reaction time and its distribution, error rates, interference effects, display-size slopes, practice curves, dual-task costs, task-switch costs, and changes after a rule reversal. Each measure can have alternative explanations. Schneider and Shiffrin (1977); Logan (1988).

Reaction time and learning curves

Repeated measurement across training can reveal progressive changes in mean speed and response-time variability. Logan's instance theory offers specific predictions about practice-related changes that go beyond merely observing quicker responding. Researchers must also examine errors, response strategies, fatigue, stimulus frequencies, and whether the improvement persists when materials or mappings change. A learning curve demonstrates acquired performance under measured conditions; broad claims require additional tests. Logan (1988).

Interference and congruency effects

In tasks such as Stroop, the difference between congruent and incongruent conditions reveals interference under a specific rule. A larger interference effect can reflect conflict, relative pathway strength, response competition, or other features of the design. It cannot be translated automatically into a general psychological diagnosis or a person's fixed capacity for self-control. Interpretations become stronger when studies isolate plausible alternatives and replicate effects under clearly specified conditions. MacLeod (1991); Cohen, Dunbar, and McClelland (1990).

Transfer, dual-task costs, and context changes

Strong tests of acquired automaticity examine what happens when a second task is added, the stimulus–response mapping changes, the task context shifts, or a novel item is introduced. A response may be highly efficient on trained items but less efficient elsewhere. Dual-task costs are also influenced by perceptual and response modalities. Researchers therefore need appropriate comparison conditions before attributing performance changes to a reduced need for attention. Hazeltine, Ruthruff, and Remington (2006); Schneider and Shiffrin (1977).

Reliability, ecological validity, and reproducibility

Reaction times depend on equipment, instructions, trial exclusions, participant preparation, and speed–accuracy strategies. A substantial group-level effect can coexist with uncertain interpretation of an individual's score. Laboratory paradigms isolate useful processes by simplifying tasks; daily behavior adds changing goals, social demands, emotional relevance, and environmental variability. Research conclusions should identify which outcomes were measured and which real-world inferences remain untested. Kiesel et al. (2010); Moors and De Houwer (2006).

Individual Differences, Expertise, and Development

Automaticity develops through exposure and practice, but people differ in prior knowledge, opportunities for training, task strategies, and the specific contexts in which skills are used. The same set of instructions can consequently be easy for one participant and demanding for another. Within-person comparisons over practice help isolate learning, while between-person contrasts require attention to differing histories of familiarity. Logan (1988).

Expertise often involves fluent component operations alongside sophisticated monitoring and adaptation. A musician may execute a familiar motor sequence while directing attention to expression or coordinating with an ensemble. When the pattern changes unexpectedly, deliberate correction can become important. The example illustrates how expertise redistributes attentional demands rather than abolishing the need for control. Practice-related neural and behavioral changes are compatible with this mixed organization. Chein and Schneider (2005).

Age, education, language experience, disability-related access, and cultural learning environments can affect which operations become familiar and what laboratory tasks mean to participants. Such differences do not license ranking groups by an abstract amount of automaticity. Appropriate research designs establish whether the task, instructions, stimulus familiarity, and response modes support comparable measurement across groups.

What the Evidence Supports in Everyday Life

The strongest practical inference concerns the value of consistent, appropriately structured practice for specific skills. Repeating a stable mapping can make its component response more accessible and reduce the coordination it demands. However, practicing a rigid routine may also strengthen a response that is less useful when circumstances change. Training should therefore reflect whether the goal is reliable execution under stable conditions, flexible adaptation, or both. Schneider and Shiffrin (1977); Logan (1988).

A second inference concerns environments that minimize unnecessary competition. When a task requires complex judgment, reducing irrelevant interruptions can preserve opportunities to apply the current rule. When routine components are well learned, a person may have more capacity to coordinate other relevant operations, but simultaneous performance remains constrained by the particular tasks being combined. The dual-task evidence does not warrant a general promise that practice makes unrestricted multitasking safe or effective. Hazeltine, Ruthruff, and Remington (2006).

A third implication is that deliberate review remains valuable when practiced patterns encounter unusual conditions. Familiarity should support efficient action while the person retains access to error correction, supervision, and alternative strategies. This follows from the interaction of habitual and goal-directed components found in research on task switching and habits. Kiesel et al. (2010); Wood and Rünger (2016).

Automaticity research does not establish that commercial brain-training programs generally improve attention across unrelated tasks. Evidence for improvement in one practiced paradigm must be distinguished from evidence of far transfer to different activities, enduring everyday benefit, or clinical improvement. Similarly, unusual distractibility or a low score on a laboratory control task is an observation requiring context, rather than a diagnosis of attention-deficit/hyperactivity disorder or another condition.

Frequently Asked Questions

Is automatic processing the same as a habit?

They overlap in learning and repeated responses. Automatic processing characterizes how a specific cognitive operation runs and how much active control it requires. Habit research typically investigates learned behavior linked to recurring contexts and its relation to goals. A habit may contain multiple automatic and controlled components. Wood and Rünger (2016).

Can a process become automatic through practice?

Yes, repeated experience under suitable consistent conditions can make a response faster, more reliable, and less demanding in particular tasks. What improves, which representations are learned, and whether the advantage transfers depend on the practice conditions and mechanism. Schneider and Shiffrin (1977); Logan (1988).

Does automatic mean that attention is entirely absent?

No. An operation can exhibit efficient or unintentional features while remaining influenced by goal settings or attention. Experiments assess which attentional demands remain under controlled variations in the task rather than presuming zero attention from subjective ease. Moors and De Houwer (2006); Cohen, Dunbar, and McClelland (1990).

Why does the Stroop effect demonstrate a conflict between processing tendencies?

The written color word activates a practiced response tendency while the task instruction demands ink-color naming. The resulting interference is informative about competition and goal regulation, although multiple processes contribute to the observed reaction-time difference. Stroop (1935); MacLeod (1991).

Is controlled processing always slower or more effortful?

It often becomes more demanding when rules are new or conflict is high, but speed and effort are task-dependent. A well-prepared controlled response may be efficient, while an automatic tendency can interfere and delay the correct action. An experimental comparison must specify the rule, practice history, competing operations, and outcome measure. Kiesel et al. (2010).

Can automatic and controlled processing happen together?

Yes. Current goals can bias learned pathways, while practiced component operations contribute to deliberative activity. Models of graded pathway strength and contemporary analyses of action control both provide ways to understand their interaction. Cohen, Dunbar, and McClelland (1990); Hommel (2019).

Scientific Conclusion

Automatic and controlled processing provide a framework for explaining how learning, attention, and goal regulation shape behavior. Extensive practice can strengthen accessible responses and reduce some demands of task execution. Control helps establish new rules, manage competing information, adjust to exceptions, and coordinate learned operations toward current goals. Research from visual search, Stroop interference, dual-task performance, and task switching shows both the explanatory value of the distinction and the importance of describing automaticity through separate measurable properties. A full account of skilled cognition therefore asks what is learned, what remains sensitive to attention, and when flexible control is recruited. Moors and De Houwer (2006); Cohen, Dunbar, and McClelland (1990).

Related Articles

The Stroop Effect: Why Conflicting Information Slows Responses examines how a highly practiced tendency to read can interfere with a goal-directed color-naming response.

Cognitive Load: How Mental Demands Affect Performance explains how task demands interact with available cognitive resources, extending the discussion of effort and capacity.

Task Switching and Digital Multitasking: Why Constant Switching Drains Focus considers the distinct applied problem of repeated digital task changes.

References

Chein, J. M., & Schneider, W. (2005). Neuroimaging studies of practice-related change: fMRI and meta-analytic evidence of a domain-general control network for learning. Cognitive Brain Research, 25(3), 607–623. https://doi.org/10.1016/j.cogbrainres.2005.08.013

Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the control of automatic processes: A parallel distributed processing account of the Stroop effect. Psychological Review, 97(3), 332–361. https://doi.org/10.1037/0033-295X.97.3.332

Hazeltine, E., Ruthruff, E., & Remington, R. W. (2006). The role of input and output modality pairings in dual-task performance: Evidence for content-dependent central interference. Cognitive Psychology, 52(4), 291–345. https://doi.org/10.1016/j.cogpsych.2005.11.001

Hommel, B. (2019). Binary theorizing does not account for action control. Frontiers in Psychology, 10, 2542. https://doi.org/10.3389/fpsyg.2019.02542

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

Logan, G. D. (1988). Toward an instance theory of automatization. Psychological Review, 95(4), 492–527. https://doi.org/10.1037/0033-295X.95.4.492

MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109(2), 163–203. https://doi.org/10.1037/0033-2909.109.2.163

Melnikoff, D. E., & Bargh, J. A. (2018). The mythical number two. Trends in Cognitive Sciences, 22(4), 280–293. https://doi.org/10.1016/j.tics.2018.02.001

Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and conceptual analysis. Psychological Bulletin, 132(2), 297–326. https://doi.org/10.1037/0033-2909.132.2.297

Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information processing: I. Detection, search, and attention. Psychological Review, 84(1), 1–66. https://doi.org/10.1037/0033-295X.84.1.1

Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84(2), 127–190. https://doi.org/10.1037/0033-295X.84.2.127

Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643–662. https://doi.org/10.1037/h0054651

Tzelgov, J., Porat, Z., & Henik, A. (1997). Automaticity and consciousness: Is perceiving the word necessary for reading it? The American Journal of Psychology, 110(3), 429–448. https://pubmed.ncbi.nlm.nih.gov/9339539/

Wood, W., & Rünger, D. (2016). Psychology of habit. Annual Review of Psychology, 67, 289–314. https://doi.org/10.1146/annurev-psych-122414-033417