Dual Process Theory: The Effect of Alignment and Interference between Automatic and Controlled Processing on Decision-Making during the Stroop Test

Abstract

Objectives: Thinking and decision-making depend on automatic and controlled processing, as highlighted by the Dual-Process Theory. The present study aims to identify the effect of alignment and interference between these two types of processing, by utilising a variation of the Stroop Colour and Word Test.

Methods: A repeated-measures design was applied for the purposes of the investigation, along with counter-balancing of the Stroop test conditions (Congruent/Incongruent List). The Incongruent List aimed to trigger interference between the two systems of thinking, while the Congruent List would permit alignment. The instructions and the test itself were standardised. Participants' decision-making abilities were assessed indirectly, by recording the number of errors and the time taken to complete each condition.

Results: The mean time taken for all participants (n = 30) to complete the Incongruent condition was greater compared to the Congruent condition (t = 100.30  vs t = 52.32). A parametric related t-test was applied due to the experiment’s repeated measures design, the presence of ratio-level variables, the large sample size with no anomalies, and the normal distribution of data, verified by the Shapiro-Wilk test (W(30) = 0.95, p = 0.186 and W(30) = 0.95, p = 0.131). The t-test confirmed that the difference observed in decision-making abilities is statistically significant (p < 0.00001).

Conclusion: Interference between automatic and controlled processing impedes decision-making, while alignment increases both speed and accuracy. Hence, the study demonstrates the effect of alignment and interference between automatic and controlled processing on decision-making during the Stroop test.

KEYWORDS

Stroop Effect, Dual-Process Theory, Decision-making, Cognitive control, Automatic and Controlled Processing, System I and II Thinking, Stroop Colour and Word Test

---

Introduction

Dual-Process Theory (DPT) explains the processing involved in decision-making and thinking [1]. Kahneman (2003) proposed the existence of two systems of processing. System I (SI): automatic, unconscious, and reliant on heuristics; System II (SII): slower, logical, and conscious [2]. Examples of functions carried out by SI include walking and typing, whereas mathematical calculations and logical reasoning are examples of SII functions. While SI may be efficient and essential for daily functioning, it is often unreliable and error-prone. SII, on the other hand, is resource-intensive and easily disrupted by cognitive load. Consequently, decision-making and thinking ultimately rely on the coaction of the two. DPT has many real-life applications, including the development of medical models for decision-making, efficient group-work in education, and decision-making in economic behaviour [3],[4],[5]. Moreover, the DPT explains many observed cognitive behaviours, including the Stroop Effect.

The Stroop Effect (SE) refers to the delay in reaction time for congruent and incongruent stimuli [6]. This was first identified by Stroop (1935). The original paper is one of the most cited studies in experimental psychology [7]. This research remains relevant today, with SE being used in neuropsychological tests, including the Stroop Colour and Word Test, the Emotional Stroop Test, and the Stroop Neuropsychological Screening Test [8],[9],[10]. In these, SE is used to evaluate executive functions, by measuring selective attention, cognitive flexibility and processing speed [11],[12].

Amongst the three experiments conducted by Stroop (1935), SE was observed in Experiment-II. The aim was to investigate the effect of interfering word stimuli upon naming words serially. The sample included 100 university students. Participants were given two lists: (i) “Naming-Colour test” (NC); (ii) Naming Colours where colour of print and word differs (NCWd) (e.g. “green” printed in purple). The time taken for participants to identify the ink-colour of all items in each condition was measured. Results demonstrated that the mean reaction time was greater in NCWd, compared to NC. This may be explained by DPT. The task in NCWd urged participants to reach a decision regarding the ink-colour. Participants, however, automatically read words first before identifying ink-colours, as reading is an automatic process (SI), whereas identifying ink-colours is a type of controlled processing (SII). Hence, SII must overcome and correct SI, resulting in the observed delay. From a neuroanatomical perspective, the distinction between SI and SII involves the anterior cingulate cortex (ACC) and dorsolateral prefrontal cortex (DLPFC) [13],[14]. The DLPFC counteracts biases (the recognition of the word’s semantic perception), and separates relevant from irrelevant information, to focus on the actual ink-colour [15]. The ACC detects interference and evaluates responses, sending feedback back to the DLPFC [16],[17]. The increased activation of the DLPFC and ACC during the incongruent condition, as observed in studies using brain scans, represents the conflict between SI and SII processing [18],[19],[20],[21].

In the present investigation, Stroop’s (1935) Experiment-II is replicated. The aim is to assess the effect of alignment and interference between SI and SII on decision-making.

There are some variations to the original experiment. The sample includes Year-12 students, instead of university students, to determine whether younger generations are more efficient in dual-processing. Furthermore, the congruent condition originally contained different coloured shapes, as the study was evaluating interference theory (two conditions used, with and without interference). In our investigation, the SE is used to test DPT instead. Therefore, the congruent list contains colour-words with matching ink-colour, instead of shapes. Apart from maintaining consistent stimuli, this ensures equal engagement of lexical processing (SI). This alteration provides insight on the effect of alignment versus conflict between SI and SII (rather than the activation/inactivation of SI), for further applications or variations of SE and DPT across fields, including process dissociation paradigms in cognitive neuroscience [22].

Independent Variable: The type of list containing 100 colour-words printed in different ink-colours; Congruent List (CL), where ink-colour matches the colour-word (e.g. “red” printed red), versus Incongruent List (IL), where ink-colour contrasts the colour-word (e.g. “red” printed blue).

Dependent Variable: Mean Reaction Time per Hundred Reactions for the participant to correctly identify ink-colour (seconds).

Null Hypothesis: There will be no statistically significant difference between the Mean Reaction Time per a Hundred Reactions of the participant in the IL and the CL.

Research Hypothesis (One-Tailed): The Mean Reaction Time per a Hundred Reactions will exhibit a statistically significant increase for the participant’s performance in the IL, compared to the CL.

Methods

Design, Sampling, and Controlled Variables

A repeated-measures design was used; the same group of participants were exposed to both CL and IL, eliminating the influence of participant variables (reading proficiency, or neurodevelopmental conditions e.g. Attention Deficit Hyperactivity Disorder) [23],[24]. This ensured that the effect of congruency was measured rather than individual differences. Counterbalancing was applied to eliminate the order-effect, equating practice and fatigue per trial. Half of the participants began with CL, the rest with IL. This is a modified method compared to the one used in the original experiment (two instead of four lists). This was chosen due to possible cognitive fatigue which could affect participant’s performance.

Opportunity sampling was utilised due to time-restrictions in student availability. The sample included 30 Greek students, aged 16-17. It was ensured that all participants studied English at Proficiency-level. Language constitutes an extraneous variable, so this homogeneity controlled for its effect on results (all participants affected equally). Note that age and education are also variables that influence individuals’ performance in the Stroop Test [25],[26]. Consequently, all selected participants belonged to the same age group and attended the same school and curriculum. This was done to reduce the variability of results.

Regarding confounding variables, standardised instructions were employed. This preserved internal validity and reliability by minimising differences in participants’ understanding/motivation, that could potentially conflict with results. Similarly, to ensure that participants interpret colours identically, a sample trial was applied, where participants identified ink-colours. Note that the same lists (IL/CL) and number of items per list (100) were utilised for equal difficulty across trials. Moreover, both lists had words printed in the same font (Times New Roman, lower-case) and size (20pt). Lists were divided into rows, five items each. Similar to the original study, three colours were chosen from the Woodworth-Wells colour-sheet: red, blue, green (#FF0000,#0000FF,#00FF00). Yellow was rejected due to visibility, and black due to the inactivation of SII processing [27]. Brown and Purple were used instead (#9900FF,#945B06). Careful consideration was taken into the arrangement of items, avoiding repetition of words/colours in columns/rows, with each word and ink appearing once per row. Furthermore, the same location, time (9:30-11:30am) and artificial lightning were used, keeping environmental variables constant. Lastly, to eliminate the effects of researchers’ reaction time on measurements, three stopwatches were used, and a mean was recorded in each trial.

Ethical Considerations

Informed consent was obtained from all participants to ensure voluntary participation. Deception was not applied; the aim of the study was made clear to participants from the beginning. Any questions participants had were answered. In certain cases, IL caused visible frustration. To protect participants from emotional distress, discussions were held after the study and debriefing sheets were provided, ensuring that participants comprehended the nature of SE.

Results

The mean reaction time per 100 reactions was calculated for IL and CL, for all participants (n = 30). Results are shown in Table 1.

Table 1: Difference (Δ) in Mean time per 100 reactions, and Standard Deviation (SD), between Incongruent List (IL) and Congruent List (CL).

The dependent variable of this investigation was quantitative, had a true zero (initiation of timer), and was thus a ratio-level variable. From inspection of quantile-quantile plots, data were approximately normally distributed, with no anomalies (see Figure 1).

Figure 1: Quantile-quantile plots for (a) the Incongruent List; and, (b) the Congruent List. From visual inspection of the data, we deduce that there is approximately a normal distribution for both sets. These observations are confirmed with a Shapiro-Wilk Test.

Consequently, mean and standard deviation (SD) were chosen as measures of central tendency and dispersion, respectively. SD in the two lists was similar (indicating relative homogeneity in the distribution of scores within each condition), likely due to the repeated measures design. The mean of IL was greater than that of CL, as predicted by the research hypothesis. Figure 2 provides a visual illustration of the processed data.

Figure 2: Mean Time per 100 Reactions in seconds for the Incongruent List (IL) and the Congruent List (CL).

A parametric related t-test was applied because of the experiment’s repeated measures design, the presence of ratio-level variables, the large sample size with no anomalies, and the normal distribution of data, verified by the Shapiro-Wilk test (for IL, W(30) = 0.95, p = 0.186; for CL, W(30) = 0.95, p = 0.131).

The critical value for one-tailed hypothesis (p-value < α), at α = 0.05 level of significance (as appropriate for psychological research) and 29 degrees of freedom (for n = 30), is t = 1.70. The calculated statistic from the raw data exceeded this value (t = 16.26), and thus the results of the investigation are statistically significant (p < 0.00001). Hence, we accept the research hypothesis and reject the null hypothesis.

Discussion

The results of the investigation support DPT. In the IL, participants had to override the automaticity of reading (SI) and engage SII to produce the correct response (naming the ink-color which conflicted the word). Consequently, the mean reaction time per 100 reactions was greater compared to the CL, where SI and SII aligned. This observed difference between CL and IL was confirmed as statistically significant (p < 0.00001). This reaffirms that it is the interference between SI and SII that produces the difference in performance. Another observation was that participants made errors more frequently in the IL compared to the CL. Hence, the research hypothesis is accepted, while the null hypothesis is rejected.

Despite the variation in the CL from the original study (where SI was entirely disengaged), performance in the IL remained inferior both in terms of accuracy, as more mistakes were made, and speed. This demonstrates that it is not the activation of SI that impedes performance; rather, it is whether SI aligns or not with SII. The variation in the sample further confirms the presence of SE even in participants of younger age-groups.

The repeated-measures design utilised means that the same participants were exposed to both IL and CL. Therefore, the design was a strength of this experiment, as it eliminated participant variables from interfering with the results. Participant variables nevertheless introduce variance in the results, as exemplified by the large SD. This variance could be further addressed by preliminary tests in which participants could be further divided into groups depending on their skill-level and analyse their overall performance within subgroups.

Counter-balancing was also used to eliminate order-effect, equating practice and fatigue for each condition. Note that participants were exposed to two lists, as opposed to four in the original study. Participants reported some fatigue towards the end of the experiment, and their performance, in terms of accuracy and speed, tended to show minor decline. Therefore, this variation proved beneficial as severe fatigue was avoided.

Opportunity sampling was utilised. This saved time that alternatively would have been spent on encouraging possible participants to sign up. Nevertheless, this method of sampling may have introduced bias. Given that the majority of participants were acquaintances, some viewed the study as a competition. Few participants misinterpreted the task and focused mainly on speed rather than accuracy (despite instructions highlighting that no mistakes should be left uncorrected). Random sampling could be used to overcome this issue, to avoid possible acquaintances and allow participants to focus on their individual performance.

Additionally, all participants were non-native English speakers. This homogeneity was a strength, as all participants were impacted equally from language, an extraneous variable. To further limit the effect of language, it was ensured that all participants acquired English proficiency. However, several studies on bilingualism have suggested that the SE differs in an individual’s dominant and non-dominant language [28]. For bilingual individuals, reading may not be fully automatic (SI) and may require SII engagement. Consequently, performance in IL reflects the cognitive load of SII multitasking, rather than the intended interference between SI and SII. To overcome this limitation, native English-speakers should be chosen to participate instead.

To maintain high internal validity, the procedure utilised in the experiment was standardised to eliminate confounding variables. Participants were tested in the same environment and at the same time of day. Standardised instructions were used, while the IL and CL used were identical for all participants. Nevertheless, through this procedure, it was not possible to investigate to what extent participants were engaged. For this reason, multiple studies have begun utilising brain scans, such as fMRI, to assess to what extent SI and SII are active during the experiment. As this experiment was conducted in an educational setting, performing brain scans was indeed impossible. Nevertheless, an interesting extension of the study would be to replicate the experiment and further measure interference between SI and SII from a neuroanatomical viewpoint.

Conclusion

Interference between SI and SII impedes decision-making, resulting in a decrease in performance during the Stroop test. Contrastingly, alignment between SI and SII increases both speed and accuracy. Hence, the study successfully demonstrates the effect of alignment and interference between automatic and controlled processing on decision-making during the Stroop test.

Data Availability Statement

All raw data can be obtained by contacting the correspondance author.

Conflict of Interest Disclosure

The author declares no competing interests.

Ethics Approval Statement

This research protocol was reviewed and approved by Doukas School, Athens, Greece. All procedures performed in studies involving human participants were conducted in accordance with the standards of the Ethical Principles of Psychologists and Code of Conduct, and its later amendments, as published by the American Psychological Association (APA).

Participant Consent Statement

Informed consent was obtained from all individual participants included in the study. Participants were also informed of their right to withdraw.

Bibliography

[1] Bellini-Leite, S. C. (2022). Dual process theory: Embodied and predictive; symbolic and classical. Frontiers in Psychology, 13. https://doi.org/10.3389/fpsyg.2022.805386

[2] Kahneman, D. (2003). A perspective on judgment and choice: Mapping bounded rationality. American Psychologist, 58(9), 697–720. https://doi.org/10.1037/0003-066X.58.9.697

[3] Djulbegovic, B., Hozo, I., Beckstead, J., Tsalatsanis, A., & Pauker, S. G. (2012). Dual processing model of medical decision-making. BMC Medical Informatics and Decision Making, 12(1), 94. https://doi.org/10.1186/1472-6947-12-94

[4] Vermette, P. J., & Kline, C. L. (2017). Group work that works: Student collaboration for 21st century success (1st ed.). Routledge. https://doi.org/10.4324/9781315618364

[5] Alós-Ferrer, C., & Strack, F. (2014). From dual processes to multiple selves: implications for economic behavior. Journal of Economic Psychology, 41, 1–11. https://doi.org/10.1016/j.joep.2013.12.005

[6] Gluud, L. L., Jeyaraj, R., & Morgan, M. Y. (2019). Outcomes in clinical trials evaluating interventions for the prevention and treatment of hepatic encephalopathy. Journal of Clinical and Experimental Hepatology, 9(3), 354–361. https://doi.org/10.1016/j.jceh.2019.02.004

[7] 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

[8] Scarpina, F., & Tagini, S. (2017). The Stroop color and word test. Frontiers in Psychology, 8, 557. https://doi.org/10.3389/fpsyg.2017.00557

[9] Markela-Lerenc, J., Kaiser, S., Fiedler, P., Weisbrod, M., & Mundt, C. (2006). Stroop performance in depressive patients: A preliminary report. Journal of Affective Disorders, 94(1–3), 261–267. https://doi.org/10.1016/j.jad.2006.04.011

[10] Zalonis, I., Christidi, F., Bonakis, A., Kararizou, E., Triantafyllou, N. I., Paraskevas, G., Kapaki, E., & Vasilopoulos, D. (2009). The Stroop effect in Greek healthy population: Normative data for the Stroop Neuropsychological Screening Test. Archives of Clinical Neuropsychology: The Official Journal of the National Academy of Neuropsychologists, 24(1), 81–88. https://doi.org/10.1093/arclin/acp011

[11] Lezak, M. D. (2004). Neuropsychological assessment. Oxford University Press.

[12] Strauss, E., Sherman, E. M. S., & Spreen, O. (2006). A compendium of neuropsychological tests: Administration, norms, and commentary. Oxford University Press.

[13] Ehlis, A.-C., Zarantonello, L., Haeussinger, F. B., Rohe, T., Rosenbaum, D., Fallgatter, A. J., & Maier, M. J. (2024). The DLPFC is centrally involved in resolving Stroop conflicts, suppressing distracting sensory input within the auditory and visual system. Frontiers in Psychology, 15, 1427455. https://doi.org/10.3389/fpsyg.2024.1427455

[14] Roelofs, A., Van Turennout, M., & Coles, M. G. H. (2006). Anterior cingulate cortex activity can be independent of response conflict in Stroop-like tasks. Proceedings of the National Academy of Sciences, 103(37), 13884–13889. https://doi.org/10.1073/pnas.0606265103

[15] Baraniuk, J. N., Thapaliya, K., Inderyas, M., Shan, Z. Y., & Barnden, L. R. (2024). Stroop task and practice effects demonstrate cognitive dysfunction in long COVID and myalgic encephalomyelitis / chronic fatigue syndrome. Scientific Reports, 14(1), 26796. https://doi.org/10.1038/s41598-024-75651-3

[16] Heidlmayr, K., Kihlstedt, M., & Isel, F. (2020). A review on the electroencephalography markers of Stroop executive control processes. Brain and Cognition, 146, 105637. https://doi.org/10.1016/j.bandc.2020.105637

[17] Banich, M. T. (2019). The Stroop effect occurs at multiple points along a cascade of control: Evidence from cognitive neuroscience approaches. Frontiers in Psychology, 10, 2164. https://doi.org/10.3389/fpsyg.2019.02164

[18] Kerns, J. G. (2006). Anterior cingulate and prefrontal cortex activity in an FMRI study of trial-to-trial adjustments on the Simon task. NeuroImage, 33(1), 399–405. https://doi.org/10.1016/j.neuroimage.2006.06.012

[19] MacDonald, A. W., Cohen, J. D., Stenger, V. A., & Carter, C. S. (2000). Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in cognitive control. Science (New York, N.Y.), 288(5472), 1835–1838. https://doi.org/10.1126/science.288.5472.1835

[20] Grandjean, J., D’Ostilio, K., Phillips, C., Balteau, E., Degueldre, C., Luxen, A., Maquet, P., Salmon, E., & Collette, F. (2012). Modulation of brain activity during a stroop inhibitory task by the kind of cognitive control required. PLOS ONE, 7(7), e41513. https://doi.org/10.1371/journal.pone.0041513

[21] Barkley-Levenson, E., Xue, F., Droutman, V., Miller, L. C., Smith, B. J., Jeong, D., Lu, Z.-L., Bechara, A., & Read, S. J. (2018). Prefrontal cortical activity during the Stroop task: New insights into the why and the who of real-world risky sexual behavior. Annals of Behavioral Medicine: A Publication of the Society of Behavioral Medicine, 52(5), 367–379. https://doi.org/10.1093/abm/kax019

[22] Lindsay, D. S., & Jacoby, L. L. (1994). Stroop process dissociations: The relationship between facilitation and interference. Journal of Experimental Psychology: Human Perception and Performance, 20(2), 219–234. https://doi.org/10.1037/0096-1523.20.2.219

[23] Protopapas, A., Archonti, A., & Skaloumbakas, C. (2007). Reading ability is negatively related to Stroop interference. Cognitive Psychology, 54(3), 251–282. https://doi.org/10.1016/j.cogpsych.2006.07.003

[24] Schwartz, K., & Verhaeghen, P. (2008). ADHD and Stroop interference from age 9 to age 41 years: A meta-analysis of developmental effects. Psychological Medicine, 38(11), 1607–1616. https://doi.org/10.1017/S003329170700267X

[25] Zurrón, M., Lindín, M., Galdo-Alvarez, S., & Díaz, F. (2014). Age-related effects on event-related brain potentials in a congruence/incongruence judgment color-word Stroop task. Frontiers in Aging Neuroscience, 6. https://doi.org/10.3389/fnagi.2014.00128

[26] Imbrosciano, A., & Berlach, R. G. (2005). The Stroop test and its relationship to academic performance and general behaviour of young students. Teacher Development, 9(1), 131–144. https://doi.org/10.1080/13664530500200234

[27] 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

[28] Ning, R. (2021). How language proficiency influences stroop effect and reverse-stroop effect: A functional magnetic resonance imaging study. Journal of Neurolinguistics, 60, 101027. https://doi.org/10.1016/j.jneuroling.2021.101027