EEG Correlates of Suggestion-Induced Stroop Interference Reduction in High-Suggestible Individuals: A Systematic Review and Meta-Analysis

ABSTRACT

Studies have explored the impact of suggestion on the Stroop effect, aiming to understand how effective suggestion is in modulating this phenomenon. The suggestion effect has been replicated in multiple studies, supporting its robustness, but lacks systematic evaluation. We conducted a systematic review and meta-analysis of relevant English-language studies from PubMed, Web of Science, PsycINFO, Scopus, and ScienceDirect since databases inception until January 2023. Quality of included studies was evaluated using the Joanna Briggs Institute (JBI) appraisal checklist, and potential publication biases were assessed. Subgroup analyses were also performed, and effect sizes were estimated using Hedges’ g and analyzed using random effects model. The systematic review was comprised of 19 studies. For the meta-analysis, 14 studies examined the suggestion effect on Stroop interference effect (SIE), while six studies investigated suggestion effects on accuracy. Results have revealed significant overall effects of suggestion on Stroop performance in participants, as evidenced by SIE and accuracy. Subgroup analysis based on types of suggestion demonstrated a significant effect on SIE. Six EEG/ERP studies have also been discussed in the context of the review.

EEG-Korrelate der Suggestions-induzierten Stroop-Interferenz-Reduktion bei hochsuggestiven Personen: Eine systematische Überprüfung und Meta-Analyse

Aman Kumar Raturi, Sreelatha S. Narayanan, und S. P. K. Jena

Zusammenfssung: Studien haben den Einfluss von Suggestion auf den Stroop-Effekt untersucht, um zu verstehen, wie effektiv Suggestion dieses Phänomen moduliert. Der Suggestionseffekt wurde in mehreren Studien reproduziert, was seine Robustheit untermauert, aber es fehlt eine systematische Auswertung. Wir haben eine systematische Überprüfung und Meta-Analyse relevanter englischsprachiger Studien aus PubMed, Web of Science, PsycINFO, Scopus und ScienceDirect seit Beginn der Datenbanken bis Januar 2023 durchgeführt. Die Qualität der eingeschlossenen Studien wurde anhand der Beurteilungs-Checkliste des Joanna Briggs Institute (JBI) bewertet, und mögliche Publikationsfehler wurden beurteilt. Es wurden auch Subgruppenanalysen durchgeführt, und die Effektgrößen wurden mit dem Hedges’schen g geschätzt und mit dem Modell der zufälligen Effekte analysiert. Die systematische Überprüfung umfasste 19 Studien. Für die Meta-Analyse untersuchten 14 Studien die Wirkung von Suggestion auf den Stroop-Interferenz-Effekt (SIE), während sechs Studien die Wirkung von Suggestion auf die Genauigkeit untersuchten. Die Ergebnisse zeigten signifikante Gesamteffekte von Suggestion auf die Stroop-Leistung der Teilnehmer, wie sie durch SIE und Genauigkeit belegt werden. Eine Untergruppenanalyse nach Art der Suggestion zeigte einen signifikanten Effekt auf die SIE. Sechs EEG/ERP-Studien wurden ebenfalls im Rahmen der Überprüfung diskutiert.

Corrélats EEG de la réduction de l’interférence de Stroop induite par la suggestion chez les personnes très susceptibles : Une revue systématique et une méta-analyse

Aman Kumar Raturi, Sreelatha S. Narayanan et S. P. K. Jena

Résumé: Des études ont exploré l’impact de la suggestion sur l’effet Stroop, afin de comprendre l’efficacité de la suggestion dans la modulation de ce phénomène. L’effet de suggestion a été reproduit dans de nombreuses études, ce qui confirme sa robustesse, mais il n’a pas fait l’objet d’une évaluation systématique. Nous avons procédé à un examen systématique et à une méta-analyse des études pertinentes en langue anglaise tirées de PubMed, Web of Science, PsycINFO, Scopus et ScienceDirect depuis la création des bases de données jusqu’à janvier 2023. La qualité des études incluses a été évaluée à l’aide de la liste de contrôle du Joanna Briggs Institute (JBI), et les biais de publication potentiels ont été évalués. Des analyses de sous-groupes ont également été réalisées, et les tailles d’effet ont été estimées à l’aide de la méthode de Hedges et analysées à l’aide d’un modèle d’effets aléatoires. L’examen systématique comprenait 19 études. Pour la méta-analyse, 14 études ont examiné l’effet de la suggestion sur l’effet d’interférence de Stroop (EIS), tandis que six études ont examiné les effets de la suggestion sur la précision. Les résultats ont révélé des effets globaux significatifs de la suggestion sur les performances de Stroop chez les participants, comme en témoignent l’effet d’interférence de Stroop et la précision. L’analyse des sous-groupes basée sur les types de suggestion a démontré un effet significatif sur l’EIAS. Six études EEG/ERP ont également été discutées dans le cadre de l’examen.

Correlatos EEG de la reducción de la interferencia Stroop inducida por la sugestión en individuos altamente sugestionables: Una revisión sistemática y un metaanálisis

Aman Kumar Raturi, Sreelatha S. Narayanan y S. P. K. Jena

Resumen: Los estudios han explorado el impacto de la sugestión en el efecto Stroop, con el objetivo de comprender la eficacia de la sugestión en la modulación de este fenómeno. El efecto de la sugestión se ha replicado en múltiples estudios, lo que apoya su solidez, pero carece de una evaluación sistemática. Se realizó una revisión sistemática y un metanálisis de los estudios relevantes en inglés de PubMed, Web of Science, PsycINFO, Scopus y ScienceDirect desde el inicio de las bases de datos hasta enero de 2023. La calidad de los estudios incluidos se evaluó mediante la lista de verificación del Instituto Joanna Briggs (JBI) y se evaluaron los posibles sesgos de publicación. También se realizaron análisis de subgrupos, y los tamaños del efecto se estimaron mediante la g de Hedges y se analizaron mediante un modelo de efectos aleatorios. La revisión sistemática estaba compuesta por 19 estudios. Para el metanálisis, 14 estudios examinaron el efecto de la sugestión sobre el efecto de interferencia Stroop (SIE), mientras que seis estudios investigaron los efectos de la sugestión sobre la precisión. Los resultados han revelado efectos globales significativos de la sugestión en el rendimiento Stroop de los participantes, según lo evidenciado por el SIE y la precisión. El análisis de subgrupos basado en los tipos de sugestión demostró un efecto significativo sobre el SIE. También se han analizado seis estudios EEG/ERP en el contexto de la revisión.

Translation acknowledgments: The Spanish, French, and German translations were conducted using DeepL Translator (www.deepl.com/translator).

KEYWORDS:

• Automaticity
• electrophysiology
• hypnotizability
• Stroop
• suggestibility

Introduction

The relationship between suggestibility and hypnosis remains evident in contemporary hypnosis research. Current standardized measures of hypnotic susceptibility rely on suggestion-based items, and the likelihood of someone being hypnotized is primarily assessed through these tests. Influential researchers in the field, including Orne (1977), Spanos and Barber (1974), and Tellegen (1978) have primarily focused on understanding hypnosis in relation to suggestions and suggestibility. However, there are also researchers like Eysenck and Furneaux (1945), Hilgard and Tart (1966), and Kihlstrom (1985) who have expanded their exploration beyond suggestibility to consider the social context in which suggestions are given. The close connection between suggestibility and hypnosis has given rise to fundamental questions that demand further exploration. One such question is whether a relationship exists between “waking suggestibility” and hypnotic suggestibility (Eysenck & Furneaux, 1945; Hilgard, 1965; Stukat, 1958). Researchers have also sought to determine the extent to which a hypnotic induction procedure enhances suggestibility (Barber, 1969; Hilgard & Tart, 1966; Weitzexhoffer & Sjoberg, 1961). Furthermore, studies have been conducted to examine whether phenomena traditionally attributed to hypnosis can be solely explained by the influence of suggestion (Hartland, 1967). While some literature provides insights into these questions, limited attention has been given to studying suggestibility as an independent subject due to methodological concerns (Hartland, 1967).

To understand the relationship between hypnosis and suggestibility, the current literature presents three main approaches. The first approach involves studying suggestibility independent of the hypnotic context; this is done by examining research on hypnosis that limits the role of suggestibility as well as exploring suggestion independent of hypnosis or hypnosis independent of suggestibility. Early studies in the field revealed strong positive correlations between ideomotor and behavioral suggestions and hypnotic responsiveness (Eysenck & Furneaux, 1945; Hull, 1933; White, 1930). However, other evidence suggests that the relationship between nonhypnotic suggestibility and hypnotic responsiveness is weak, indicating that they are likely separate constructs (Bekerian & Bowers, 1983; De Pascalis, 1998; De Pascalis et al., 1989; Hilgard, 1973). Some researchers argue that there is little distinction between suggestibility and hypnotizability (Barber, 1969; Braffman & Kirsch, 1999; Kirsch, 1997; Wagstaff, 1991).

Re-Evaluating Hypnosis Scales: Recent Insights and Findings

The effectiveness of hypnosis scales, which traditionally measure “hypnotizability,” has recently been called into question. According to Kirsch and Braffman (2001) these scales primarily assess “imaginative suggestibility” (nonhypnotic suggestibility) rather than true hypnotic suggestibility or hypnotizability. Their empirical study found that individuals respond to suggestions even without being in a hypnotic state, and nonhypnotic suggestibility and hypnotic suggestibility are highly correlated. This suggests that the current hypnosis scales may not accurately measure hypnotizability. These standardized hypnosis scales serve a broader purpose beyond measuring hypnotizability or hypnotic suggestibility (Braffman & Kirsch, 1999). In Oakley et al. 2021 study, they discuss the use of scales for assessing hypnotizability and propose the term “direct verbal suggestibility (DVS)” to describe a trait where verbal suggestions lead to involuntary movements and cognitive changes. Kallio’s (2021) commentary on Oakley’s work emphasizes DVS, which can occur with or without hypnotic induction. Kallio also proposes reevaluating the concept of suggestibility to better understand how individuals from diverse backgrounds respond to suggestions. Several studies, including Lifshitz and Raz (2015) and Zhang et al. (2018), provide evidence indicating that the effects of posthypnotic suggestions and nonhypnotic suggestions exhibit similarities in influencing specific cognitive tasks.

Utilizing Hypnotic Suggestion in Cognitive Psychology: Investigating Automatic Processes

Over the past two decades, suggestions have been employed as a tool in studying automatic cognitive processes (Oakley & Halligan, 2013). Within the framework of limited-capacity models of attention and information processing, cognitive psychologists have sought to develop a precise definition of automaticity (LaBerge & Samuels, 1974; Posner & Snyder, 1975; Schneider & Shiffrin, 1977). According to this perspective, automatic processes are triggered by specific stimuli in the environment and, once activated, they are executed in a reflexive manner. Automatic processes require minimal cognitive resources, operate in parallel to avoid interference with other cognitive processes, and share similarities with involuntary reflexes and instincts. The concepts of automatic and controlled processing are integral to the understanding of executive function (Shallice, 1982; Stuss & Benson, 1984). Some modern theories of automaticity challenge strict dichotomies, suggesting that automatic processes can develop gradually with practice and may also depend on the context in which they are assessed (D’Angelo et al., 2013; Dehaene & Naccache, 2001; Neumann, 1990).

In the exploration of how hypnotizability or suggestibility impacts the modulation of automatic processes, various perspectives have emerged (Egner & Raz, 2007). The first viewpoint suggests that individuals with high susceptibility to hypnotic suggestibility possess strong attentional focusing abilities, and the hypnotic state is characterized by a heightened state of focused attention (Barber, 1965; Spiegel, 1994; Tellegen & Atkinson, 1974). On the other hand, an alternative perspective argues that high susceptible individuals may exhibit impaired attentional control following hypnosis (Crawford & Gruzelier, 1992; Gruzelier, 1998; Hilgard, 1977; Hilgard & Tart, 1966; Jamieson & Sheehan, 2004; Woody & Bowers, 1994). Evidence reviewed by Egner and Raz (2007) overwhelmingly supports the view that hypnosis leads to impaired attention rather than enhanced focused attention. This lends support to models proposing that the hypnotic state involves inhibition of frontal lobe cognitive control functions (Crawford & Gruzelier, 1992; Gruzelier, 1998) or dissociation. The control mechanisms responsible for initiating actions become detached from the elements that trigger those actions (Woody & Bowers, 1994). In contrast, some researchers have put forth the argument and shown that this inclination for dissociation is predominantly restricted to a small subset of high susceptible individuals (Carlson & Putnam, 1989; Perry, 1986; Zamansky & Bartis, 1984). Neurophysiological evidence supports the impaired frontal view (Crawford & Gruzelier, 1992; Gruzelier, 1998). Other perspectives diverge from the impaired frontal view, placing greater emphasis on socio-cognitive factors (Brown, 1999; Brown & Oakley, 2004; Oakley, 1999; Spanos, 1986; Spanos & Chaves, 1989) response expectancy (Kirsch, 1985), and metacognitive processes – such as not being aware of one’s intentions (Dienes & Perner, 2007).

Suggestibility and Stroop Interference

The Stroop task and the Eriksen flanker task, both traditional selective attention tasks, have been widely used to assess executive or cognitive control (Eriksen & Eriksen, 1974; MacLeod, 1991; Stroop, 1935). These tasks require individuals to focus on one stimulus dimension (the “target” dimension) while ignoring another stimulus dimension (the “distracter” dimension). The Stroop effect has sparked our interest in understanding the role of attentional systems in suggestibility. It has been suggested that the Stroop task can serve as a useful mediator between different models of cognitive control processes in relation to suggestibility and hypnotizability (Kirsch & Lynn, 1998), and variations of this paradigm have been extensively used to study attentional control in hypnosis (Iani et al., 2006; Raz et al., 2002). Studies investigating the Stroop task in the context of hypnosis suggest that hypnotic suggestions can be utilized to reduce lexical automaticity. These studies indicate that under hypnosis, words can be delexicalized by altering perception, leading to a reduction in Stroop interference (Raz et al., 2002, 2003, 2006, 2007). Raz et al. (2005) attributed this reduction to the effect of hypnotic suggestion, which narrows attentional focus and modulates the processing of input words in high hypnotizable individuals. Their research revealed that during hypnosis decreased Stroop interference was associated with reduced activity in the anterior cingulate gyrus. In a subsequent study, Raz et al. (2006) found that the reduction in Stroop interference was primarily due to suggestion rather than hypnotic induction, supporting the claims made by Kirsch and Braffman (2001). However, other studies have reported contrasting findings, suggesting that hypnotic suggestion does not affect performance on the semantic Stroop effect (e.g., a color-associated word like “sky” printed in the incompatible color “green;” Augustinova & Ferrand, 2012; Augustinova et al., 2010). These studies provide evidence that semantic activation in the Stroop task is automatic, and suggest that hypnotic suggestion may impact response competition rather than de-automatizing word reading, as proposed by Raz and his colleagues (Raz & Campbell, 2011; Raz et al., 2005). In a review, Parris et al. (2022) argue that the current literature does not clearly distinguish between conflicting and facilitating representations at various levels (phonological, semantic, and response levels); however, there is strong evidence for distinguishing task conflict from informational conflict in the Stroop task, suggesting at least two loci of attentional selection processes. This view may necessitate a reevaluation of the Stroop task’s conduction and administration. This shift in perspective has the potential to introduce novel avenues of research and exploration in this field.

The study of the Stroop effect presents a challenge in pinpointing the stage at which interference occurs, whether during stimulus encoding or response production. While response time (RT) provides an objective measure, the duration of stimulus encoding remains elusive. Electroencephalogram (EEG) signals, which measure brain activity, can provide insights into different brain states. Typically, EEG reflects cortical function at a network level, with some limited insight into specific brain activities (Niedermeyer & Lopes da Silva, 2004). In the context of hypnosis research, a certain hypothesis associates hypnosis with alpha brain rhythms. This idea emerged from early findings that linked meditation practices to alpha oscillations (Kasamatsu & Hirai, 1966) and from the perceived similarities between hypnotic and meditative states. Some initial research reported increased alpha activity in individuals experiencing hypnosis, particularly those highly responsive to hypnotic suggestions (De Pascalis & Palumbo, 1986; Graffin et al., 1995; Macleod-Morgan, 1979; Morgan et al., 1974). It is important to note, however, that not all studies have consistently found this increase in alpha activity during hypnosis, and some have even reported variations (Kihlstrom, 2013; Ray, 1997; Sabourin et al., 1990). The relationship between gamma brainwave activity and hypnosis is complex and, at times, contradictory in research (De Pascalis, 2007; Jensen et al., 2015). Some studies have shown that high hypnotizable individuals exhibit higher baseline gamma activity (Akpinar et al., 1971; De Pascalis, 1993; De Pascalis et al., 2004; Schnyer & Allen, 1995) and increased gamma activity during hypnotic responding (De Pascalis, 1993), while other studies have reported the opposite, with lower gamma power in high hypnotizable individuals (De Pascalis et al., 1987) and decreased gamma activity in response to hypnotic analgesia suggestions (De Pascalis et al., 2004). Hypnosis can influence gamma activity, but the direction of this influence varies across studies (Jensen et al., 2015). On the other hand, there is more reliable evidence connecting hypnosis to theta oscillations. Individuals with high hypnotizability tend to exhibit elevated baseline theta activity (Freeman et al., 2000; Galbraith et al., 1970; Kirenskaya et al., 2011; Montgomery et al., 2000; Sabourin et al., 1990; Tebēcis et al., 1975). When people undergo hypnotic inductions and suggestions, particularly those highly responsive to hypnosis, they often exhibit increased theta activity (De Pascalis & Perrone, 1996; Jensen et al., 2013; Sabourin et al., 1990; Williams & Gruzelier, 2001). It is worth noting that this effect may not be universal across all individuals and hypnotic procedures. Theta power has been highlighted in studies involving cognitive load (Nigbur et al., 2011) and the utilization of executive functions (Sauseng et al., 2005), such as in the Stroop task. Moreover, elevations in beta power, particularly in the frontal region, have been associated with heightened levels of selective attention and enhanced cognitive control (Clayton et al., 2015; Coelli et al., 2015; Stoll et al., 2016). In addition to power, the analysis of EEG coherence – reflecting the synchronization of brain activity across different regions – has been employed to assess cognitive and memory load. Previous research suggested that during tasks with varying cognitive and memory demands, changes in theta coherence, particularly in frontoparietal regions, might signify the use of executive functions (Sauseng et al., 2005). Some researchers have also proposed that hypnosis might lead to a separation of conflict monitoring and cognitive control processes (Egner & Hirsch, 2005).

Event-related potentials (ERPs), particularly the P300 wave, serve as a valuable tool by reflecting stimulus-evaluation duration independently of RT. Notably, across ERP studies, a consistent finding emerges: the latency of the P300 component remains unaltered in response to incongruent stimuli (Atkinson et al., 2003; Duncan-Johnson & Kopell, 1981). However, certain studies have reported smaller amplitudes for P3a and P3b in response to incongruent stimuli (Zurrón et al., 2009). Inconsistencies in the reported amplitudes of P300 waves are observed among different studies (Bekçi & Karakaş, 2009; Ilan & Polich, 1999; Szücs & Soltész, 2010; West & Alain, 1999). The non-discrepant nature of the P300 component suggests that it remains unaffected by incongruent stimuli, implying that the Stroop effect on RT occurs after stimulus evaluation, with facilitation and interference manifesting in later stages of response production following P300 elicitation (Sahinoglu & Dogan, 2016). Additionally, some studies have reported the presence of a prolonged medial dorsal negativity between 350–500 ms post-stimulus (peak at 410 ms) in response to incongruent stimuli and a prolonged positivity developed between 500–800 ms post-stimulus over left superior temporo-parietal scalp (Liotti et al., 2000), an N400 peak (Bekçi & Karakaş, 2009), and an N450 (Ergen et al., 2014). This late negative component around 400 ms has been associated with semantic conflict or conflict detection (Badzakova-Trajkov et al., 2009; Liotti et al., 2000; Sahinoglu & Dogan, 2016; Szücs & Soltész, 2010). Furthermore, a handful of studies have found differences in N1 and P1, suggesting modulation in early attention-related components (Atkinson et al., 2003).

In alignment with the comprehensive analysis undertaken by (Lifshitz et al., 2013), which delved into the capacity of top-down influences, such as suggestion, to regulate the automaticity of cognitive processes across a spectrum of tasks (like the Stroop, Flanker and McGurk tasks), our current analysis focuses on the specific modulation of these processes within the context of the Stroop task. We adopt a systematic and quantitative methodology, incorporating a meta-analytical review. The current analysis additionally includes a neurobiological dimension by highlighting the electrophysiological correlates, as evidenced in EEG and ERP studies.

Methods

This systematic review was conducted in accordance with the PRISMA-P 2020 statement for systematic review and meta-analysis protocols (Figure 1; Page et al., 2021) and has been registered in the international prospective register of systematic reviews (PROSPERO) under the registration number: CRD42023390048.

Figure 1. PRISMA Flowchart of Study Selection Process

Eligibility Criteria

The inclusion criteria for this study consisted of four main components: (1) the use of either a hypnotic or non-hypnotic suggestion, (2) a control condition without any active intervention, (3) the utilization of a Stroop paradigm task, (4) the measurement of hypnotic suggestibility using established scales such as HGSHS-A, SHSS-C, WSGC-C, and SWASH, and (5) for a separate analysis of the electrophysiological basis, only studies that have utilized EEG/ERP in both the suggestion and no suggestion conditions will be further analysed. While there is no universally agreed-upon definition of imaginative suggestibility, we adopted the conceptualization proposed by Kirsch and Braffman (2001). The method of suggestion employed to induce alterations in perceptual experiences in another person was primarily based on the framework introduced by Raz et al. (2002). Exclusion criteria included (1) use of suggestions with other interventions such as meditation, (2) pharmacologically induced suggestions (e.g. oxytocin) and (3) participants with major psychiatric and neurological disorders.

Search Strategy

PubMed, SCOPUS, PsycINFO, SCIENCEDIRECT and Web of Science databases were searched independently by two reviewers for potentially eligible studies indexed from database inception until Jan 2022. The search string consisted of three elements related to Suggestibility AND Stroop AND Hypnosis. Searches were applied to all database fields where possible, or title/abstract/keywords where this restriction was imposed by the database. Results were limited a posteriori to “human studies” and searches were augmented through manual searches of reference lists of included articles and reviews.

Study Selection

Titles and abstracts of articles returned by initial searches were independently screened by two reviewers who rejected articles not meeting eligibility criteria. The full-text of remaining articles was independently examined by the same reviewers to reach a final list of articles. Disagreements at either screening stage were resolved through discussion with a third reviewer.

Study Outcomes

The outcome variables assessed in this study included: (1) Stroop interference effect (SIE): The reaction time discrepancy observed between incongruent stimuli and congruent or neutral stimuli.; (2) accuracy: The percentage of accuracy or error rates in task performance were measured in both experimental conditions.

Study Quality

Two raters conducted independent assessments of each study’s methodological quality using a validity scale consisting of 15 items. These items were designed to evaluate methodological rigor, selection bias, and reporting bias, following the criteria outlined by PRISMA.

Data Extraction

Data extraction and coding were conducted by two authors using a standardized template. To ensure accuracy, all data entries were reviewed by another reviewer. The following information was extracted from the studies: (1) Study outcomes: The data included measures of the SIE and accuracy in task performance; (2) sample characteristics: Information regarding the participants’ age, gender, and hypnotic suggestibility was collected; (3) study characteristics: This encompassed details about the study design, types of suggestion employed, nature of suggestion (hypnotic or non-hypnotic), specific suggestibility scales used, and the Stroop task employed; (4) brain recording technique: If applicable, the data included the type of brain imaging technique utilized in the study, such as EEG/ERP, analysis and result.

Imaginative Suggestibility

Imaginative suggestions refer to requests to engage in an imagined scenario as if it were real (Kirsch & Braffman, 2001). These suggestions are distinct from other types of suggestions, such as the placebo effect, sensory suggestions, and the misinformation effect, as evidenced by their relatively weak correlations with these various types of suggestion.

Imaginative suggestions can be administered with or without the induction of hypnosis. In a hypnotic context, they can be delivered by a designated or perceived hypnosis interventionist, or they can be self-administered, in which case it is referred to as “self-hypnosis.” Individuals who exhibit responsiveness to the imaginative suggestions commonly associated with hypnosis are often referred to as high hypnotizable. However, the strong correlation between responsiveness to these suggestions in both hypnosis and non-hypnosis conditions (r = .67 for behavioral scores; r = .82 for subjective scores; Braffman & Kirsch, 1999) suggests that the term “high suggestible” is a more accurate characterization of these individuals. We posit that the ability to respond to imaginative suggestions relies on the capacity to translate or experience the suggested sensations and mental imagery as credible and compelling subjective experiences and actions. The range of hypnotic suggestibility among individuals varies from almost complete responsiveness to none at all. Recently, there has been scrutiny regarding the effectiveness of so-called hypnosis scales, which primarily assess hypnotizability. According to Kirsch and Braffman (2001), these hypnosis tests actually measure imaginative suggestibility (nonhypnotic suggestibility) rather than hypnotic suggestibility or hypnotizability. Study samples were classified as low, medium or high in hypnotic suggestibility if scores on standardized measures fell within the following ranges: (a) Harvard Group Scale of Hypnotic Suggestibility: Form A (HGSHS:A) and Stanford Hypnotic Suggestibility Scale (SHSS) forms A and C: low (0–4), medium (5–7), high (8–12; Shor & Orne, 1963; Weitzenhoffer & Hilgard, 1962); (b) Carleton University Responsiveness to Suggestion Scale: Objective dimension (CURSS:O): low (0–2), medium (3–4), high (5–7; Spanos et al., 1983); (c) Sussex-Waterloo Scale of Hypnotizability: low (0–2), medium (3–5), high (5–10) (SWASH; Lush et al., 2018): and (d) Waterloo-Stanford Group C Scale of Hypnotic Susceptibility (WSGC; Bowers, 1993): low (0–3), medium (4–7), high (8–11). Hypnosis research often selects participants based on their level of suggestibility, which can vary on a continuum. Some studies have focused on high and low suggestible individuals, omitting those with medium suggestibility levels to maximize resources (Jensen et al., 2017). In recent neuroscience studies on hypnosis, even low suggestible groups have been dropped, with only high suggestible individuals used, acting as their own controls (Oakley & Halligan, 2010). However, without a group of medium suggestible individuals, assessing these differences becomes challenging. It is worth noting that extreme lows and highs in suggestibility are atypical, with most individuals falling within the medium range (Jensen et al., 2017; Lynn et al., 2007; Perri et al., 2020). Moreover, these scales rely on the construct of suggestibility, which has become increasingly questioned in the context of hypnotizability (Facco et al., 2017; Kirsch & Braffman, 2001; Tasso & Pérez, 2008; Testoni et al., 2020; Wagstaff et al., 2008; Weitzenhoffer, 2002).

Post Hypnotic Suggestions and Non-Hypnotic Suggestions

The suggestions were categorized as “posthypnotic” when they were accompanied by a standard hypnotic induction before being given; whereas the suggestions were classified as “non-hypnotic” when they were given without a hypnotic induction. Out of the total number of studies conducted, 18 studies employed posthypnotic suggestions prior to the task, while the remaining 4 studies utilized non-hypnotic suggestions. One study included both types of suggestions, incorporating both posthypnotic suggestions non-hypnotic suggestions.” Two studies used suggestions called “hypnotic alexia” and “perceptual/semantic” suggestion, while 15 studies used the “word blindness” suggestion. The suggestions were categorized as word blindness suggestions if they followed the format used by Raz et al. (2002), which can be described as follows:

“Very soon you will be playing the computer game. When I clap my hands, meaningless symbols will appear in the middle of the screen. They will feel like characters of a foreign language that you do not know, and you will not attempt to attribute any meaning to them. This gibberish will be printed in one of four ink colors: red, blue, green, or yellow. Although you will only be able to attend to the symbols ink color, you will look straight at the scrambled signs and crisply see all of them. Your job is to quickly and accurately depress the key that corresponds to the ink color shown. You will find that you can play this game easily and effortlessly.”

Results

Study Inclusion

A total of 108 studies were identified through database searches, with one additional record obtained through manual searching. Screening of titles/abstracts resulted in 54 potentially eligible articles, and after full-text review, a final list of 20 eligible studies was compiled. The key characteristics of these studies can be found in Table 1. Five of these studies provided additional EEG/ERP data, which is presented in Table 2.

Table 1. Summary Characteristics of Included Studies

Authors	Study Location	N	N (High suggestible)	Mean age(S.d/range)	Gender (M/F)	Scale	Task	Type of Suggestion	Suggestion	Medium	Design
Sheehan et al. (1988)	Australia	26	13	NA	3/23	HGSHS- A	Modified Stroop task	Posthypnotic Suggestions.	Strategy Manipulation	Verbally	MFD
Nordby et al. (1999)	Norway	10	5	(19–30)	NA	SHSS-C	Norwegian Stroop task (Vocal output.)	Posthypnotic Suggestions.	Active Blocking	Verbally	MFD
Mallard and Bryant (2001)	Australia	85	32	19.92 (5.01)	25/60	HGSHS- A; SHSS-C	Reverse Stroop naming task	Posthypnotic Suggestions.	Color blindness	Verbally	RM
Raz et al. (2002)	USA	32	16	24.3	17/15	SHSS-C	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	MFD
Raz et al. (2005)	USA	16	8	27 (20–35)	8/8	HGSHS- A; SHSS-C	Classical Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	MFD
Raz et al. (2006)	USA	25	25	NA	10/15	CURSS: C	Modified Stroop task	Posthypnotic Suggestions & Non- hypnotic suggestions.	Word blindness	Tape recorded	RM-C
Raz et al. (2007)	USA	49	49	27 (20–35)	25/24	HGSHS- A; SHSS-C	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Casiglia et al. (2010)	Italy	12	6	28.6 (3.8)	4/8	SHSS-C	Modified Stroop task	Posthypnotic Suggestions.	Hypnotic Alexia	Verbally	MFD
Raz and Campbell (2011)	Canada	83	49	29 (20–41)	9/38	HGSHS- A; SHSS-C	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	MFD
Augustinova and Ferrand (2012) (Exp 1)	France	28	28	NA	NA	HGSHS- A; SHSS-C	French Stroop Task	Non- hypnotic suggestions	Word blindness	Tape recorded	RCT
Augustinova and Ferrand (2012) (Exp 2)	France	15	15	NA	NA	HGSHS- A; SHSS-C	French Stroop Task	Non- hypnotic suggestions	Word blindness	Tape recorded	RM-C
Parris et al. (2012)	UK	19	19	21.6 (2.4)	6/13	WSGC- C	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Parris and Dienes (2013)	UK	35	11	25.5 (4.02)	10/25	WSGC- C	Modified Stroop task	Non- hypnotic suggestions	Word blindness	Verbally	MFD
Zahedi et al. (2017)	Germany	12	12	24.4 (3)	84	HGSHS- A	Persian Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Zahedi et al. (2019)	Germany	16	16	31.1 (14.4)	5/11	HGSHS- A	German Stroop Task.	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Bollinger et al. (2020)	USA	33	11	50.7 (18–65)	15/18	SHSS-C	Classic Stroop task; Smoking Stroop task.	Posthypnotic Suggestions.	Word blindness	Verbally	MFD
Perri et al. (2021)	Italy	17	6	23.5 (6.7)	4/13	HGSHS- A	Italian Stroop task	Posthypnotic Suggestions.	Perceptual Suggestion; Semantic Suggestion	Verbally	MFD
Palfi et al. (2021) (Exp 1)	UK	33	33	NA	NA	WSGC- C	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Palfi et al. (2021) (Exp 2)	UK	48	48	20.32 (4.07)	12/36	SWASH	Modified Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Parris et al. (2021)	UK	16	10	22.24 (6.40)	1/15	SHSS-C	Classical Stroop task	Posthypnotic Suggestions.	Word blindness	Verbally	RM-C
Palfi et al. (2022) (Exp 1)	UK	57	10	19.61 (1.47)	6/51	SWASH	Modified Stroop task	Non- hypnotic suggestions	Looking-away; blurring; single-letter focus; goal-maintenance	Verbally	MFD
Brunel et al. (2023) (Exp 1)	France	27	27	20.79 (2.88)	4/23	HGSHS- A	Emotional Stroop Task	Posthypnotic Suggestions.	Emotional Intensity	Tape recorded	RM-C

Note: MFD- Mixed Factorial Design; RM-C – Repeated Measure design with counterbalancing; RM- Repeated Measure design; RCT-Randomized controlled design; NA - Not available.

Table 2. EEG/ERP Studies

Authors	Tools	Electrodes	Locations Analysed	Reference	Sampling rate	Impedance	Filtering	Baseline	Time Window	Analysis	Results
Nordby et al. (1999)	ERP	11 channels	Fp1, Fp2, F7, F8, T3, T4, T5, T6	Ears	200 Hz	<2 kOhms	.01 to 30 Hz	150 ms	N2 - 250-450 ms P3- 350-500 ms	ERP: P3 and N2 components were analyzed.	Event-related potentials indicated that the high hypnotizable subjects showed a reduced P3a amplitude and a decreased N2b latency to the visual stimuli in both waking and hypnotic conditions, suggesting a lack of orienting to or disengagement peripherally occurring stimuli.
Raz et al. (2005)	fMRI/ERP	128 channels	Electrode No 6 (Midfrontal) and 76 (Midoccipital)	Vertex 26	250 Hz	NR	15 Hz	200 ms	120–250 ms	ERP: N100 and P100 components were analyzed.	P100 and N100, showing both a shift and a reduction in amplitude after suggestions. Brain activity in posterior and anterior regions was delayed and largely reduced, supporting the notion that the posthypnotic suggestion affected processing of the entire visual stream and was not specific to visual words.
Casiglia et al. (2010)	EEG/ERP	19 channels	P3, Pz, P4, C3, Cz, C4, F3, Fz and F4	Nose	NR	NR	0.05–20 Hz	NR	ERP:300–700 ms LORETA: 300 to 500 ms	ERP: Late positive complex was analyzed; LORETA Analysis.	ERP: LPC amplitude was greater for the congruent than for the incongruent conditions (p < .03).The late fronto-parietal potential (Fp1, Fp2) detected during incongruence increased significantly from −0.6 to 8.3 μV (p < .01) when post-hypnotic conditioning was active. LORETA: Bilateral differences of activation between congruent and incongruent conditions are detected bilaterally in medial prefrontal gyrus, in the anterior cingulate gyrus, in the cingulate gyrus and in the posterior parietal cortex. During post-hypnotic alexia, differences of activation between congruent and incongruent conditions are particularly detected in the left hemisphere and more in detail in medial prefrontal gyrus and in the inferior parietal cortex.
Zahedi et al. (2017)	EEG	32 channels	Fp1, Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, T6, O1 and O2	Mastoids reference (A1-A2)	NR	<10 kOhms	0.1 to 40 Hz	NR	NA	Resting state and task-related EEG periods were analyzed separately with two methods:- Spectral Analysis were used to compute mean power and Coherence Analysis (2 short-range coherence & 1 long-range coherence)	EEG recorded during the Stroop task showed a significant increase in both frontal theta and frontal beta power when participants were under the impact of post-hypnotic suggestions, in comparison to the two other sessions. Coherence analysis in theta band of frontal channels revealed a significant main effect of instruction. Decrease of coherence in the theta band in the hypnosis-plus session in comparison to the two others, although no-hypnosis and hypnosis session were not significantly different (none of coherences of fronto-parietal and parietal channels in theta band were significant).
Zahedi et al. (2019)	EEG/ERP	60 channels	Fz, CPz,; FC1, FCz, FC2, C1, Cz, C2, CP1, CPz, and CP2; P3, P1, Pz, P2, P4, CP3, CP1, CPz, CP2 and CP4; P5, P3, P1, Pz, P2, P4, P6, PO3, POz, PO4	Mastoids reference (A1-A2)	250 Hz	<5 kOhms	0.05–40 Hz	NR	N1:120–160 ms N2:200–30 ms P3: 300–400 ms N400: 420–520 ms	ERP: N 400, N1, N2, and P3 components were analyzed	ERPs showed task- invariant Stroop effects in N400 amplitudes. PHS increased the N1 and decreased the N2 components in ERPs. Stroop effects in the N400 were reduced by PHS
Perri et al. (2021)	ERP	32 channels	Prefrontal F8 and F7 (selected based on collapsed localizer procedure)	Left Mastoid & FCz.	250 Hz	<5 kOhms	Butterworth cutoff frequency 30 Hz, slope 12 dB/octave	200 ms	Prestimulus & poststimulus ERPs	Prestimulus : pN poststimulus : N1, P180, N300 and LPP were analyzed.	Compared to baseline, the hypnosis conditions enhanced the anterior pN1 and reduced the visual N1 component. For the specific effect of the semantic suggestion, it emerged through a huge reduction in P180 activity over the left frontal cortex.

Participants Characteristics

A total of 20 studies with two additional independent experiments (Total = 22) have provided data for 694 participants (high suggestible n = 449, medium suggestible = 92, low suggestible = 153). The mixed factorial design was used primarily. The mean study age (reported by k = 15 of 22 studies) was 27.7 years. The mean gender composition (k = 18) was 70% female.

Study Characteristics

The study designs used included a mixed factorial design with suggestibility (high/medium/low) as a between-subjects factor and suggestion (suggestion/no suggestion) as a within-subjects factor (k = 10), a repeated measure design counterbalanced with suggestion as a within-subjects factor (k = 10) and without counterbalancing (k = 1), and a randomized controlled design (k = 1). Additionally, data was collected from diverse study locations, including the USA (k = 5), UK (k = 6), France (k = 3), Italy (k = 2), Germany (k = 2), Canada (k = 1), Australia (k = 2), and Norway (k = 1).

Suggestibility Tests and Stroop Task

Suggestibility was assessed using HGSHS: A (k = 5), SHSS-C (k = 5), HGSHS:A+ SHSS-C (k = 6), CURSS (k = 1), SWASH (k = 2), and WSGC (k = 3) scales. Different language versions of Stroop tasks were used including English versions (k = 16), French (k = 2), German (k = 1), Persian (k = 1), Italian (k = 1), and Norwegian (k = 1). All of the Stroop tasks included congruent, incongruent and neutral words except for two studies which did not include neutral words in the task.

EEG/ERP

A total of five studies conducted EEG/ERP analysis. Three of these studies focused on ERP analysis, while one study utilized both ERP and Low Resolution Brain Electromagnetic Tomography (LORETA) analysis. Another study examined EEG oscillations and mean power using power spectral and coherence analysis. Although the electrode placements differed across the studies, the majority of them analyzed the frontal and parietal regions, along with other locations. The studies examined various ERP components such as N1, N2, N400, P3, and late positive components. Additionally, two studies included LORETA & Neuroelectric Source Imaging (the minimum-norm method [MNM]) analysis implemented in BrainStorm to localize the sources. One study examined the resting state and task-related EEG periods. Further important details can be found in Table 2. It is crucial to acknowledge that, across the studies that employed EEG, none of them incorporated non-hypnotic suggestions into their experimental design. This omission means that the findings and interpretations drawn from these EEG studies should be limited primarily to posthypnotic suggestibility, as the influence of non-hypnotic suggestions remains unexplored.

Study Validity Criteria

Study ratings for each validity criteria are shown in Appendix. Although most study criteria were well met, perhaps most importantly, 20 of 21 experiments of the studies explicitly reported counterbalancing, and one between group design experiment reported randomization.

Quality Assessment of the Studies

The Joanna Briggs Institute (JBI) critical appraisal checklist for quasi-experimental studies was used to assess the quality of the studies by two reviewers.

Meta-analysis

Meta-analysis Model

A meta-analysis was conducted to examine the impact of suggestion on high suggestible participants. The analysis considered data from studies focusing on this specific group and aimed to account for expected heterogeneity arising from variations in study methodology. To address this, a random-effects model was utilized, acknowledging that effect sizes are likely to differ beyond sampling error alone. To compute the effect size, the Meta-Essential workbook for dependent groups was utilized. In order to ensure comparability across studies, Hedges’ g was computed as the effect size measure. Hedges’ g (Hedges & Olkin, 1985) is a standardized mean difference that adjusts for potential biases in small sample sizes. As few studies typically report multiple effect size data (e.g. from the same subjects across multiple time points/outcomes), we used a data aggregation method (Borenstein, 2009) to account for within-study dependency of effect sizes. We conducted sensitivity analysis using lower (r = .25) and higher (r = .75) correlations to examine the effect on parameter estimates.

Effect Size

In instances where the direct value of Cohen’s d was available, it was utilized; otherwise, calculations were made using mean and standard deviation, F ratio, or t ratio based on the available data. These calculations were performed using Meta-Essential, a statistical workbook. The effect size between the suggestion and control groups was determined using Hedges’ g formula. In accordance with J. Cohen (1988), effect sizes of 0.20, 0.50, and 0.80 were generally considered small, medium, and large, respectively.

Publication Bias

Various methods were used to assess publication bias in the study. The funnel plot indicated potential publication bias for Stroop interference (Figure 2), while showing symmetry for accuracy (Figure 3). Additionally, the Trim and Fill analysis revealed two imputed data points for both outcomes. Egger’s regression (Egger et al., 1997) test showed a significant effect of publication bias for Stroop interference (β0E = 2.41, t = 2.42, p = .029, CI₉₅ [0.29, 4.53]), but not for accuracy (β0E = 2.41, t = 1.61, p = .167, CI₉₅ [−1.24, 6.06]). Similarly, Begg & Mazumdar’s (Begg & Mazumdar, 1994) test demonstrated significance for Stroop interference (z = 2.16, p = .031), but not for accuracy (z = 1.05, p = .293). The Fail-Safe index indicated the number of studies needed to offset the SIE and accuracy (Rosenthal = 614; 88), while Gleser & Olkin (Gleser & Olkin, 1996) indicated the number needed for SIE and accuracy (37; 34). Orwin index (Orwin, 1983) did not identify any studies for both effects (Orwin = 0;0), and Fisher index (Fisher, 1925) suggested unpublished studies for Stroop interference and accuracy (193; 69). Overall, these assessments shed light on the potential impact of publication bias on the meta-analysis results, emphasizing the need for caution in interpreting the findings.

Figure 2. Funnel Plot of Hedges’ g Using Trim and Fill Effect for Stroop Interference Effect

Sixteen Stroop Interference experiments (filled circles) and 2 potentially missing studies due to publication bias imputed using the trim and fill method (empty circles).

Figure 3. Funnel Plot of Hedges’ g Using the Trim and Fill Method for Accuracy

Seven accuracy experiments (filled circles) and 2 potentially missing studies due to publication bias imputed using the trim and fill method (empty circles).

Outliers

The standardized residual histogram identified outliers or influential studies at values of ±3.25 for SIE and ±3.15 for accuracy, suggesting their potential impact on the results. Standard residuals above and below ±3.25 and ±3.15 from initial meta-analysis suggested one potential outlier for SIE (+5.18; Raz et al., 2007) and one for accuracy (+3.54; Raz et al., 2005). These cases were conservatively removed to prevent potential distortion of results. After removing these cases a reduction in I² has been observed from 74.06% to 57% for SIE and 68.17% to 36.58%. Although no obvious reason for these outlying values could be identified from further scrutiny of these papers, as these were all high positive values, removal resulted in marginally reduced rather than inflated effect sizes.

Stroop Interference Effect (SIE)

Out of 22 studies only 15 studies provided sufficient data for meta-analysis. One study was deemed an outlier and was therefore excluded. Two studies conducted additional experiments on different participants, and the results of these experiments were reported separately. Five studies were not included in the analysis due to insufficient data. Meta-analysis of 14 studies (16 effect sizes) of SIE across 318 participants found suggestions to result in reduction of SIE, SMD = 0.65, CI₉₅[0.41, 0.89], z = 5.77, p = .000 (Figure 4), which is classifiable as a large effect (J. Cohen, 1988). Positive effect sizes were found in all studies. Table 3 includes Hedges’ g values, CI, and weight distribution of each studies. Significant moderate heterogeneity was observed (I² = 57%, Q = 34.88, pQ = 0.003, T² = 0.08, T = 0.28).

Figure 4. Forest Plot of Hedges’ g (With 95% Confidence Intervals) and Study Weights for 16 Experiments for Stroop Interference Effect

SMD = 0.65, CI₉₅[0.41, 0.89], z = 5.77, p = .000); I² = 57%, Q = 34.88, pQ = 0.003, T² = 0.08, T = 0.28.

Table 3. Meta-Analysis for Stroop Effect

	Study name	Hedges’ g	CI Lower limit	CI Upper limit	Weight
1	Raz et al. (2002)	1.28	0.58	1.98	5.13%
2	Raz et al. (2005)	2.22	0.72	3.72	1.98%
3	Raz et al. (2006)	0.89	0.42	1.37	7.21%
4	Casiglia et al. (2010)	0.66	−0.31	1.63	4.30%
5	Raz and Campbell (2011)	0.79	0.46	1.11	9.09%
6	Augustinova et al. (2010) (Exp 1)	0.40	−0.14	0.94	6.66%
7	Augustinova et al. (2010) (Exp 2)	0.52	0.12	0.92	8.12%
8	Parris et al. (2012)	0.45	−0.04	0.93	7.21%
9	Zahedi et al. (2017)	0.85	0.15	1.56	5.29%
10	Zahedi et al. (2019)	1.96	1.06	2.85	3.75%
11	Bollinger et al. (2020)	0.20	−0.42	0.82	6.09%
12	Perri et al. (2021)	0.19	−0.66	1.04	5.07%
13	Palfi et al. (2021)(Exp 1)	0.35	−0.01	0.71	8.70%
14	Palfi et al. (2021)(Exp 2)	0.34	0.05	0.64	9.51%
15	Parris et al. (2021)	0.64	0.08	1.20	6.46%
16	Palfi et al. (2022) (Exp 1)	0.54	−0.16	1.24	5.45%
	Combined Effect size	0.65	0.41	0.89

Accuracy

In the meta-analysis of 6 studies (7 effect sizes), one study was deemed an outlier and was therefore excluded, resulting in 6 effect sizes. A total of 185 participants revealed reduction in error percentage in suggestion condition, SMD = 0.47, CI₉₅[0.22, 0.71], z = 4.72 p < .000 (Figure 5). Moderate heterogeneity in effect size was observed (I² = 36.58%, Q = 9.46, pQ = 0.149, T² = 0.02, T = 0.16), although positive effect sizes were observed for all studies. Table 4 includes Hedges’ g values, CI, and weight distribution of each studies.

Figure 5. Forest Plot of Hedges’ g (With 95% Confidence Intervals) and Study Weights for 7 Experiments for Accuracy

SMD = 0.40, CI₉₅[0.22, 0.71], p < .000; I² = 36.58%, Q = 9.46, pQ = 0.149, T² = 0.02, T = 0.16.

Table 4. Meta-analysis for Accuracy

	Study name	Hedges’ g	CI lower limit	CI upper limit	Weight
1	Raz et al. (2002)	0.80	0.21	1.39	9.94%
2	Raz et al. (2007)	0.19	−0.10	0.47	22.35%
3	Raz and Campbell (2011)	0.43	0.14	0.73	21.61%
4	Augustinova and Ferrand (2012) (Exp 1)	0.85	0.23	1.47	9.28%
5	Augustinova and Ferrand (2012) (Exp 2)	0.73	0.30	1.16	14.75%
6	Zahedi et al. (2017)	0.29	−0.31	0.89	10.10%
7	Zahedi et al. (2019)	0.31	−0.21	0.83	11.98%
	Combined effect size	0.47	0.22	0.71

Subgroup Analysis

Subgroup analysis was performed based on the type of suggestion for SIE. The groups are divided into two groups based on the type of suggestions received, whether post-hypnotic suggestions (PHS) or non-hypnotic suggestions (NHS). The PHS group showed SMD = 0.68, CI₉₅ [0.38, 0.98] and PI [−0.06, 1.42] with significant heterogeneity (I² = 66.07%, Q = 32.42, pQ = 0.001, T² = 0.09, T = 0.30), whereas the NHS group showed SMD = 0.59, CI₉₅ [0.38, 0.81] and PI [−0.43, 1.61] with no heterogeneity (I² = 0%, Q = 2.45, pQ = 0.484, T² = 0.09, T = 0.30). The overall subgroup effect was significant (p = .002; Table 5, Figure 6).

Table 5. Subgroup Analysis

	Study name/Subgroup name	Hedges’ g	CI lower limit	CI upper limit	Weight
1	Raz et al. (2006)	0.89	0.42	1.37	26.20%
2	Augustinova and Ferrand (2012) (Exp 1)	0.52	0.12	0.92	29.23%
3	Augustinova and Ferrand (2012) (Exp 2)	0.40	−0.14	0.94	24.35%
4	Palfi et al. (2022) (Exp 1)	0.54	−0.16	1.24	20.23%
	NHS	0.59	0.38	0.81	66.41%
5	Raz et al. (2002)	1.28	0.58	1.98	7.20%
6	Raz et al. (2005)	2.22	0.72	3.72	2.88%
7	Casiglia et al. (2010)	0.6631	−0.31	1.63	6.10%
8	Raz and Campbell (2011)	0.79	0.46	1.11	12.20%
11	Parris et al. (2012)	0.45	−0.04	0.93	9.89%
12	Zahedi et al. (2017)	0.85	0.15	1.56	7.41%
13	Zahedi et al. (2019)	1.96	1.06	2.85	5.35%
14	Bollinger et al. (2020)	0.20	−0.42	0.82	8.45%
15	Perri et al. (2021)	0.19	−0.66	1.04	7.13%
16	Palfi et al. (2021) (Exp 1)	0.35	−0.01	0.71	11.74%
17	Palfi et al. (2021) (Exp 2)	0.34	0.05	0.64	12.71%
18	Parris et al., 2021	0.64	0.08	1.20	8.93%
	PHS	0.68	0.38	0.98	33.59%
	Combined effect size	0.62	0.54	0.71

Note. Posthypnotic suggestion (PHS); Non-Hypnotic suggestions (NHS).

Figure 6. Forest Plot Showing Subgroup Analysis

Discussion

In the present systematic review and meta-analysis, our objective was to address important research gaps by examining empirical evidence in the field. Specifically, there were doubts raised regarding the effectiveness of suggestions in manipulating automatic processes in high suggestible individuals. For instance, researchers questioned whether word blind suggestions were truly effective in reducing Stroop interference, as some studies had suggested, or if there were conflicting findings in the literature. Moreover, the inquiry extended to exploring whether the manipulation of automatic processes was solely restricted to the context of hypnosis, particularly in hypnotically induced effects, or if it could be achieved through other means as well. This question aimed to assess the generalizability of the findings across different contexts and conditions. Through a comprehensive analysis of the available studies, this review systematically addressed these important questions. It synthesized the existing evidence to provide a clearer understanding of the effectiveness of suggestions and their impact on automatic processes in high suggestible individuals. Prior to conducting the meta-analysis, we thoroughly analyzed the results, revealing interesting findings.

Among the studies that included both high and low suggestible participants, 70% (7 of 10) of the studies indicated that high suggestible individuals exhibited a significant difference – predominantly a reduction – in the Stroop effect as compared to low suggestible individuals after receiving suggestions. Additionally, 80% (4 of 5) of the studies demonstrated a significant difference in errors – mainly a reduction – among high suggestible participants compared to low suggestible individuals. Studies solely focusing on high suggestible participants, 91% (11 of 12) of the studies revealed a significant difference in Stroop interference after suggestion, primarily resulting in a reduction of the Stroop effect. However, only 50% (2 of 4) of the studies reported a significant difference in errors. More importantly, among the studies that utilized non-hypnotic suggestion, 100% (4 of 4) of the studies revealed that suggestion had a substantial impact on reducing the Stroop effect, specifically in high suggestible individuals. In 3 of these studies, high suggestible participants exhibited a significant reduction in the Stroop effect following the non-hypnotic suggestion (Augustinova & Ferrand, 2012; Augustinova et al., 2010; Raz et al., 2006). In one study, which compared both high and low suggestible participants, this reduction was significant in high suggestible as compared to the low suggestible participants (Palfi et al., 2022).

The findings of this review underscore that suggestions, regardless of their specific nature, consistently produced a positive effect size, indicating their effectiveness and robustness in manipulating automatic processes, particularly in high suggestible individuals. This underscores the presence of a robust phenomenon often referred to as “word blindness” or “hypnotic alexia” among those with high suggestibility. This finding aligns with the perspective that high suggestible individuals employ distinct mechanisms for processing cognitive conflict, involving different neural processes (Cojan et al., 2015; Crawford & Gruzelier, 1992; Gruzelier, 1998). However, it is important to note that some studies did not observe this phenomenon when utilizing the semantic variant of the Stroop task (Augustinova & Ferrand, 2012), which challenges (Raz et al., 2006) assertion that suggestions do not de-automatize or impede reading but rather influence response competition. Additionally, in a study by Perri et al. (2021) incorporating different types of suggestions (perceptual and semantic) and encompassing both medium and high suggestible participants, this effect was shown to be particularly applicable to the extremely high suggestible subset of participants. Importantly, this effect was observed irrespective of whether the suggestions were administered within a hypnotic context or outside of it, aligning with the view that these effects are more closely linked to the construct of suggestibility rather than hypnosis itself (Facco et al., 2017; Kirsch & Braffman, 2001; Tasso & Pérez, 2008; Testoni et al., 2020; Wagstaff et al., 2008; Weitzenhoffer, 2002). This insight challenges the notion that the influence of suggestions is solely dependent on hypnosis and highlights the broader applicability of suggestion-based interventions.

The observation of reduced heterogeneity in our subgroup analysis, which was based on the type of suggestions given, offers valuable insights. Specifically, the analysis suggests that studies utilizing non-hypnotic suggestions showed more consistent results compared to those employing hypnotic suggestions. This finding implies that non-hypnotic suggestions may have a more predictable and standardized impact on the phenomenon being studied. In contrast, the use of hypnotic suggestions might introduce greater variability, which could be attributed to various factors, including differences in hypnotic procedures, induction methods, and other potential differences. This underscores the importance of considering the specific context and techniques used in research involving hypnosis, as they can significantly influence the consistency and reliability of study outcomes. This systematic review and meta-analysis supports the idea that suggestions can be a powerful tool for manipulating automatic processes in high suggestible individuals, regardless of the specific conditions or techniques employed. This comprehensive analysis enhances our understanding of the mechanisms underlying suggestion and provides valuable insights for future research and practical applications in the field. The potential mechanisms are discussed in the next section.

The second portion of the review involved studies that utilized EEG/ERP techniques to gain neurological insights and complement the overall findings. A resting EEG study by Zahedi et al. (2017) revealed increased frontal theta and frontal beta power under the influence of suggestions, which is consistent with the literature on high hypnotizability individuals who exhibit elevated baseline theta activity (Freeman et al., 2000; Galbraith et al., 1970). When highly responsive individuals undergo hypnotic inductions and suggestions, elevated theta activity is often observed, suggesting cognitive load and executive function engagement (De Pascalis & Perrone, 1996; Jensen et al., 2013; Williams & Gruzelier, 2001). Moreover, heightened beta power in the frontal region has been linked to increased selective attention and enhanced cognitive control (Clayton et al., 2015; Coelli et al., 2015; Stoll et al., 2016). Coherence analysis in the theta band revealed reduced coherence during the hypnosis-plus suggestion session, particularly in the frontal channels, aligning with the modulation of brain oscillatory activity and coherence by hypnosis and posthypnotic suggestions, especially in regions associated with attention and cognitive processes (Egner & Hirsch, 2005; Sauseng et al., 2005). When examining ERP studies, which can be instrumental in identifying the origin of interference commonly referred to as the “Locus of the Stroop effect” (e.g., Dyer, 1973; Logan & Zbrodoff, 1998; Luo, 1999; MacLeod, 1991; Parris et al., 2022; Parris, Augustinova, et al., 2019; Scheibe et al., 1967; Seymour, 1977; Wheeler, 1977), it becomes evident that several studies have shown changes in early ERP components following suggestion. For instance, Raz et al. (2005) observed a decrease in P100 and N100 components after suggestion, while Nordby et al. (1999) found reductions in P3a and N2b components in both waking and hypnotic conditions. Moreover, Zahedi et al. (2019) reported an increase in the N1 component and a decrease in N2 ERPs after post-hypnotic conditions, and Perri et al. (2021) noted an enhancement in the anterior pN1 component and a reduction in the visual N1 component during hypnosis. These shifts and reductions in the amplitudes of early ERPs following suggestions indicate alterations in early attention-related components (Atkinson et al., 2003). In a study conducted by Perri et al. (2021), a significant reduction in P180 activity was observed over the left frontal cortex. This reduction can be interpreted in terms of semantic processing, aligning with evidence that the semantic interpretation of words typically occurs around 160 ms (Hauk et al., 2006; Szücs & Soltész, 2010). Additionally, the reduction in the N400 component observed in Zahedi et al. (2019) also suggests the involvement of semantic conflict or conflict detection (Badzakova-Trajkov et al., 2009; Liotti et al., 2000; Sahinoglu & Dogan, 2016; Szücs & Soltész, 2010). In the study conducted by Casiglia et al. (2010), it was found that the amplitude of the late positive component was higher for the congruent conditions compared to the incongruent conditions, with a statistical significance level of p < .03. Additionally, they observed a significant increase in the late fronto-parietal potential (specifically at Fp1 and Fp2) during incongruent conditions, which rose from −0.6 μV to 8.3 μV (p < .01). This increase in the fronto-parietal potential was particularly notable when post-hypnotic conditioning was active and is also associated with semantic detection, as previously discussed by Liotti et al. (2000). In their LORETA analysis, activation differences between congruent and incongruent conditions are bilaterally detected in the medial prefrontal gyrus, anterior cingulate gyrus, cingulate gyrus, and posterior parietal cortex. Yet, during post-hypnotic alexia, the left hemisphere, particularly the medial prefrontal gyrus and inferior parietal cortex, exhibits more pronounced activation differences which are associated with conflict related activities (Botvinick et al., 1999; Carter et al., 2000; Casey et al., 2000; Richeson et al., 2003).

A meta-analysis of 16 effect sizes for stroop interference effect and 6 effect sizes for accuracy, involving a total of 318 and 185 participants respectively, provided support for the effectiveness of suggestion in reducing stroop interference and enhancing accuracy. The key findings were as follows: (1) Suggestions led to a significant overall reduction in both outcomes for high suggestible participants, with a moderate to large effect size; (2) the type of suggestion (hypnotic or non-hypnotic) was found to be an important factor determining the effectiveness, with non-hypnotic suggestions producing a much larger effect size; and (3) although some evidence of publication bias was detected, its impact on the effect sizes was minimal.

Mechanism

While the exact mechanisms behind the suggestion effect are not fully understood, various theories have been proposed. One key question is where this interference occurs, known as the “locus of the Stroop effect.” (e.g., Dyer, 1973; Logan & Zbrodoff, 1998; Luo, 1999; MacLeod, 1991; Parris, Augustinova, et al., 2019; Scheibe et al., 1967; Seymour, 1977; Wheeler, 1977). The Stroop effect suggests that when responding to color, interference can happen at different stages, such as task set activation, semantic processing, and word form response. Previous research has often focused on response-level conflict (J. D. Cohen et al., 1990; Roelofs, 2003), but recent studies indicate that the Stroop effect has multiple possible loci (e.g., Augustinova et al., 2018; Entel & Tzelgov, 2018; Ferrand et al., 2019; Hasshim et al., 2019; Hershman & Henik, 2019; Kalanthroff et al., 2018; Levin & Tzelgov, 2016; Parris, 2014; Parris, Sharma, et al., 2019). Parris et al. (2022) examined the level of Stroop interference, highlighting two key contributors: informational conflict, driven by semantic and response-related information from irrelevant words, and task conflict, resulting from the competition between the task sets for color identification and word reading. The simultaneous activation of these task sets creates conflict before specific Stroop dimensions are processed. Parris et al. (2022) have also delved into other forms of informational (phonological processing & word Frequency) and task conflicts, offering valuable insights into the complex nature of Stroop interference and its underlying cognitive mechanisms.

In the context of suggestibility or hypnosis this tendency to deal with such interference have been studied by number of researchers. The prevailing perspective in the field asserts that suggestions operate through a top-down mechanism that modifies the processing of input words through a means not voluntarily available. This modification of processing involuntary input words may aid high suggestible individuals in reducing Stroop interference (Raz et al., 2002). This phenomenon has been referred to as the de-automatization of cognitive processes, blurring the traditional distinction between controlled and automatic processing (Lifshitz et al., 2013; Parris et al., 2012; Verbruggen & Logan, 2009). This perspective argues that hypnotic manipulation obstructs the activation of irrelevant words in the internal lexicon, while another perspective suggests that activation does occur but is subsequently inhibited and prevented from running its full course (Raz et al., 2003).

However, despite these mechanisms, the semantically based Stroop effect remains unresolved, leading to two further explanations. First, single-letter coloring and cuing may limit semantic processing. Second, they may help separate informational sources (e.g., color and word), thereby reducing interference, while semantic analysis remains unaffected. Additionally, the examination of how imagination influences behavior and cognitive processes has been a crucial area of research. Imagination, such as imagining the Stroop words as meaningless, may be a key strategy through which individuals reset top-down cognitive processing to align with the suggestion and reduce Stroop interference (Palfi et al., 2022). This imagination hypothesis aligns with findings indicating that suggestion reduces response conflict while leaving the visual input stream unaffected, as imagination is not limited to visual or mental imagery (Goldie, 2004). High suggestible individuals experience a reduction in the Stroop effect when given a suggestion to perceive words as meaningless scribbles. Remarkably, this effect is not contingent upon the formal induction of hypnosis (Augustinova et al., 2010; Parris & Dienes, 2013; Raz et al., 2005).

Additional factors explored include the dependence of suggestion application on conditions allowing sustained activation (Parris et al., 2012), substances like oxytocin (Parris & Dienes, 2013), suggestion with targeted processing such as aversion suggestion (Li et al., 2017), modulation of emotional intensity (Brunel et al., 2023), and smoking cessation (Bollinger et al., 2020). It appears that suggestion triggers a general change in early attentional processing rather than a language-specific filter, leading to a reduced Stroop effect. This is supported by evidence of diminished event-related potentials (ERPs) under suggestion, suggesting a significant modulation of activity in the early components. These altered visual processing patterns likely influence subsequent cognitive processes, including activation in the anterior cingulate cortex (ACC; Casiglia et al., 2010; Raz et al., 2005) or by enhancing proactive executive control in managing conflicts related to word meaning and semantic information, as well as increasing the allocation of additional cognitive control resources (Zahedi et al., 2017, 2019). This understanding may serve as a foundation for uncovering the neural correlates of other interventions based on suggestion (Raz et al., 2005).

Implications

The review’s findings have far-reaching implications that extend beyond its immediate scope. One notable implication is the recognition of suggestions as an effective tool for investigating the top-down impact of various factors, such as anticipation and placebo effects. By employing suggestion-based methodologies, researchers can gain deeper insights into these psychological processes, shedding more light on their mechanisms and effects. Furthermore, the review’s findings hold great significance for clinical interventions. The identified findings provide a fundamental basis for developing effective treatments and interventions in a clinical setting. Understanding the impact of automatic processes and volitional control on our behavior and consciousness can guide professionals in designing more targeted and efficient therapeutic approaches. However, it is important to note that while suggestions can be a valuable tool in clinical practice, they may not be a panacea, and the clinical implications should be interpreted with caution. Clinical interventions are multifaceted, and the effectiveness of suggestion-based approaches may vary depending on the specific conditions and individuals involved.

Importantly, these implications are not limited to the field of psychology alone. The insights gleaned from this review can have broader applications in areas such as organizational settings and advertising. Recognizing the influence of suggestion and its impact on behavior and decision-making can assist in creating more effective strategies for managing and motivating individuals within organizations. Likewise, understanding the role of suggestion in advertising can lead to more persuasive and impactful marketing campaigns.

In summary, this review’s implications go beyond its immediate research domain. They highlight the effectiveness of suggestion in studying higher-level processes, provide a foundation for clinical interventions, and offer valuable insights for fields such as organizational settings and advertising. The neuropsychological implications of suggestion provide a more comprehensive picture of its effects on our behavior and cognition. This enables researchers to link the observed behavioral effects of suggestion to specific neural correlates, thereby enhancing our understanding of the complex interplay between suggestion, cognitive processes, and the brain.

Limitations

There are several limitations to consider in the current findings. Firstly, the review only focuses on studies that utilized tasks related to cognitive control such as the Stroop task, and does not include research using other tasks like the Flanker or Simon tasks. This narrow selection of tasks limits the generalizability of the findings to the modulation of various automatic processes. Secondly, most of the studies included in the meta-analysis do not provide detailed summary data such as means or standard deviations. Instead, the meta-analysis primarily relies on summary statistics such as t-tests or F-tests to calculate Hedges’ g, which may introduce bias and impact the outcomes of the analysis. Thirdly, the low average age of the participants in the studies (27.7 years) raises questions about the applicability of the findings to older populations. The age restriction limits our understanding of how the effects observed in the meta-analysis may differ across different age groups. Lastly, the meta-analysis focuses exclusively on high suggestible participants, and does not compare them with individuals classified as medium or low suggestible. None of the studies compared hypnotic and non-hypnotic suggestions within the same groups, hindering the ability to definitively attribute findings to hypnotizability or hypnosis itself. It is essential to note that none of the EEG studies included non-hypnotic suggestions, which underscores the need to limit EEG study interpretations to post-hypnotic suggestibility. This limitation emphasizes the necessity for further research that directly compares the neural mechanisms underlying both hypnotic and non-hypnotic suggestions within the same groups of participants to more comprehensively understand their distinct effects.

Future Studies

While the investigation into the impact of suggestions on mitigating Stroop interference among high suggestible participants has received considerable attention, it is important to acknowledge that the existing body of research is hampered by a scarcity of high-quality data due to small sample sizes and a multitude of design biases. Consequently, drawing dependable conclusions from the current literature proves to be a challenge. Hence, there exists a distinct need for further meticulously controlled studies to ensure more reliable insights. Furthermore, it is advisable to expand research efforts toward exploring alternative automatic processes in conjunction with suggestions, as this avenue holds promise for uncovering valuable new perspectives. In forthcoming studies, a noteworthy emphasis should be placed on delving into the psycho-social implications associated with these phenomena. Moreover, it would be beneficial to standardize the employment of electrophysiological evidence while complementing it with other psychophysiological measurements. This integrated approach would contribute to a more comprehensive understanding of the subject matter and enhance the robustness of the research findings.

Conclusion

This study represents the first systematic review and meta-analysis on the topic, delving into the efficacy of suggestibility – both in hypnosis and non-hypnosis contexts – as a means of reducing the Stroop effect. By examining data from 20 different studies, the evidence gathered provides strong support for the notion that suggestions, regardless of their hypnotic nature, have a notable impact on modulating Stroop interference specifically in high suggestible individuals. These findings have significant implications, suggesting that suggestions could serve as an effective tool in influencing automatic cognitive processes among high suggestible individuals. Consequently, this could have practical applications in various domains such as clinical interventions, placebo studies, decision-making processes, and other important areas of research.

Open Scholarship

This article has earned the Center for Open Science badge for Preregistered. The materials are openly accessible at https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023390048.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.