Inhibiting Left dlPFC Leaves Relational Evaluative Conditioning Unchanged: Evidence From Electrical Brain Stimulation

Inhibiting Left dlPFC Leaves Relational Evaluative Conditioning Unchanged: Evidence From Electrical Brain Stimulation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Inhibiting Left dlPFC Leaves Relational Evaluative Conditioning Unchanged: Evidence From Electrical Brain Stimulation Joanna Wąsowicz, Robert Balas, Katarzyna Gajos This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7252129/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Relational evaluative conditioning (EC) paradigms suggest that attitude formation involves both simple co-occurrence-based and relation-based propositional processes. However, the causal role of executive control systems in supporting this propositional learning remains debated. This highly-powered, pre-registered study (N = 76) aimed to causally test the role of the left dorsolateral prefrontal cortex (dlPFC), a key executive control hub, in this process. We replicated (Kukken et al., 2020 ) relational EC paradigm while applying inhibitory (cathodal) high-definition transcranial direct current stimulation (HD-tDCS) or sham stimulation over the left dlPFC during learning. A Stroop task served as a manipulation check, and Multinomial Processing Tree (MPT) modeling dissociated relational from co-occurrence processes. Results revealed a critical dissociation: while cathodal tDCS successfully impaired performance on the Stroop task, confirming effective neuromodulation, it had no effect on evaluative ratings. Crucially, MPT modeling confirmed that tDCS did not alter the parameters for either relational or co-occurrence processing. These findings challenge the hypothesis that the left dlPFC is indispensable for integrating relational information in EC, suggesting a more nuanced link between domain-general executive control and this fundamental form of attitude formation. evaluative conditioning attitude change transcranial direct current stimulation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 INTRODUCTION Evaluative conditioning (EC) is a change in the evaluation of a conditioned stimulus (CS) caused by its pairing with a valenced unconditioned stimulus (US) (De Houwer et al., 2001 ; Hofmann et al., 2010 ). Across procedures, repeated CS-US co-occurrences typically cause assimilative transfer, where the CS acquires the valence of the paired US, a pattern confirmed by meta-analyses (Hofmann et al., 2010 ; Moran et al., 2023 ). Early paradigms used simple, contiguous presentations of neutral CSs and affective USs without explicit instructions about their relationship (De Houwer et al., 2001 ; Hofmann et al., 2010 ; Levey & Martin, 1975 ). This procedural minimalism led to EC’s characterization as an automatic, stimulus-driven phenomenon dependent on spatiotemporal contiguity and requiring little conscious awareness or inference (Gawronski, Balas, et al., 2016; Gawronski, Brannon, et al., 2016; Gawronski et al., 2007 ; Gawronski & Bodenhausen, 2006 ; Hofmann et al., 2010 ). In this traditional view, theorists suggested automatic associative mechanisms drive attitude shifts even with minimal information, reinforcing EC’s conception as a basic pathway for attitude acquisition. Consequently, standard implementations often omit relational qualifiers, which isolates co-occurrence effects but presumes that co-occurrence alone is sufficient for evaluative change (Moran et al., 2016 ; Moran & Bar-Anan, 2020 ). However, a growing body of research shows that the stated relationship between a CS and US systematically modulates evaluative outcomes beyond mere co-occurrence (Förderer & Unkelbach, 2012 ; Hu et al., 2017 ; Hughes et al., 2019 ; Kukken et al., 2020 ; Moran & Bar-Anan, 2013 ). When given explicit relational qualifiers, the standard assimilative pattern can be altered. Similarity relations (e.g., causes, starts) typically maintain assimilation, while opposition relations (e.g., prevents, stops) often produce contrast effects (Hu et al., 2017 ; Moran & Bar-Anan, 2013 ). Self-reported evaluations consistently align with the propositional meaning of the relational verb, showing that participants actively integrate this semantic information (Hu et al., 2017 ; Kukken et al., 2020 ; Moran & Bar-Anan, 2013 ). Notably, the contrast effects from opposition relations are often weaker than the assimilative effects from similarity relations, an asymmetry suggesting the influence of countervailing processes (Moran et al., 2016 ). Relational framing thus produces different outcomes even with identical CS-US exposure, demonstrating that EC effects are not reducible to simple associative links (Moran & Bar-Anan, 2013 ). This conclusion holds across multiple verbs and stimulus domains (Hu et al., 2017 ; Kukken et al., 2020 ). These findings challenge a strictly relation-insensitive associative account, instead motivating a framework where propositional integration is a core determinant of EC outcomes. Paradigms that orthogonally manipulate US valence and relational qualifiers consistently reveal two influences: an assimilative pull from CS-US co-occurrence and a modulation based on the relational information (Gawronski & Brannon, 2021 ; Hu et al., 2017 ; Kukken et al., 2020 ; Moran & Bar-Anan, 2013 ). These influences align in similarity relations (e.g., CS starts positive US) but conflict in opposition relations (e.g., CS stops positive US). This conflict results in attenuated effects, where contrast effects from opposition relations are typically smaller than assimilative effects from similarity relations, which is a robust pattern replicated across verbs and stimuli (Gawronski & Brannon, 2021 ; Kukken et al., 2020 ; Moran et al., 2016 ). This asymmetry suggests two partially independent influences: a co-occurrence-based assimilative process, and a relation-based propositional process, whose simultaneous activation produces an outcome. Because observed evaluations reflect a composite of both, their relative strengths remain unclear, highlighting the need for formal separation of these contributions. The coexistence of assimilative and relationally qualified outcomes energizes the debate between dual- and single-process EC theories (Gawronski & Bodenhausen, 2006 , 2011b , 2018 ; Hütter, 2022 ). Dual-process models propose two distinct mechanisms: an associative system driven by spatiotemporal contiguity and a propositional system that evaluates relational statements (Gawronski & Bodenhausen, 2006 , 2011b , 2018 ). In contrast, single-process theories attribute all EC outcomes to propositional operations, suggesting that apparent “association-only” effects reflect incomplete encoding or retrieval of relational information. While findings like implicit-explicit dissociations are cited as evidence for parallel processes (Hu et al., 2017 ; Moran & Bar-Anan, 2013 ), they can also be explained by differential retrieval demands within a single system. Indeed, a recent review concluded that either framework can accommodate most data by adding auxiliary assumptions about processing stages (Hütter, 2022 ). To move beyond such interpretational flexibility, methods are needed that can analytically separate co-occurrence and relational contributions within a single measure (Gawronski & Brannon, 2021 ; Kukken et al., 2020 ; Moran et al., 2016 ). Net evaluative outcomes observed in relational EC paradigms can mask the concurrent operation of (a) an assimilative component driven by mere CS–US co-occurrence and (b) a relation-driven component that can either reinforce or oppose that assimilative pull (Kukken et al., 2020 ; Moran et al., 2016 ). For example, an opposition configuration (CS stops negative) simultaneously embeds exposure to a negative US (tending to decrease CS evaluation) and a positively valenced relational implication (stopping something bad is good), producing attenuated contrast effects whose absolute magnitudes are smaller than assimilative shifts under similarity relations (Moran et al., 2016 ; Moran & Bar-Anan, 2013 ). Likewise, similarity relations (CS starts positive vs. CS starts negative) align relational implication with co-occurrence, yielding larger unidirectional shifts (Hu et al., 2017 ; Moran & Bar-Anan, 2013 ). Reliance on task dissociation—contrasting implicit (speeded) and explicit (deliberative) measures—to infer separate influences has shown that explicit ratings predominantly track relational qualifiers while some indirect measures retain unqualified co-occurrence patterns (Hu et al., 2017 ; Moran & Bar-Anan, 2013 ). However, such cross-measure comparisons invite ambiguity because differences could arise from encoding, retrieval, or decision-stage constraints rather than distinct learning mechanisms (Gawronski & Brannon, 2021 ; Moran et al., 2016 ). Consequently, without analytic decomposition, observable mean evaluations in each relational cell represent underdetermined mixtures of latent co-occurrence and relational contributions, limiting theoretical leverage for adjudicating dual-process versus single-process predictions (Gawronski & Brannon, 2021 ; Kukken et al., 2020 ). A methodological solution requires within-task, process-partitioning approaches capable of estimating the independent strength of each influence to replace interpretively fragile indirect inferences from attenuated or asymmetrical net effect patterns (Kukken et al., 2020 ; Moran et al., 2016 ). Multinomial processing tree (MPT) modeling offers a formal, within-task solution by estimating the independent probabilities of a co-occurrence (assimilative) versus a relational (propositional) process driving evaluative responses (Gawronski & Brannon, 2021 ; Heycke & Gawronski, 2020 ). This approach models separate parameters for responses reflecting mere US valence versus those reflecting the propositional meaning of the relation (Heycke & Gawronski, 2020 ; Kukken et al., 2020 ). The functional separation of these processes is powerfully supported by findings of asymmetric malleability: intentional control selectively enhances the relational parameter without affecting the robust co-occurrence parameter (Gawronski & Brannon, 2021 ). This selective plasticity aligns with theories positing a controllable propositional process operating alongside a more automatic, co-occurrence-driven influence. Crucially, by providing distinct quantitative indices for each process, MPT modeling offers a clear criterion to test whether causal interventions, like brain stimulation, differentially impact relational versus co-occurrence components (Gawronski & Brannon, 2021 ; Kukken et al., 2020 ). While behavioral and MPT evidence establishes functionally separable co-occurrence and relational components (Gawronski & Brannon, 2021 ; Heycke & Gawronski, 2020 ; Kukken et al., 2020 ), this is indirect regarding the causal neurocognitive substrates distinguishing the two (Gawronski & Bodenhausen, 2006 , 2011a ; Gawronski & Brannon, 2021 ). For instance, the finding that intentional focus selectively enhances the relational parameter (Gawronski & Brannon, 2021 ) constrains theory but does not reveal whether this malleability stems from executive control systems or intrinsic representational properties (Heycke & Gawronski, 2020 ; Hütter, 2022 ). Discriminating between dual-process and single-process interpretations of these patterns requires a causal test. A targeted brain stimulation approach can directly probe whether suppressing neural systems thought to support propositional integration, such as prefrontal executive networks, preferentially impairs the relational component while sparing the co-occurrence effect. Such a test is a critical step to determine whether the relational component's flexibility depends on frontal control resources (Gawronski & Brannon, 2021 ; Kukken et al., 2020 ). There is growing evidence that the dlPFC is intimately involved in the reflective, propositional side of evaluative processing, particularly in the regulation and contextualization of emotional responses. Neuroimaging research has shown that the dlPFC acts as a key node in the brain’s corticolimbic regulatory circuit, exerting top-down influence to up- or down-regulate emotional reactions (Ochsner et al., 2004 ). Causally, brain stimulation studies have found that modulating the left dlPFC alters people’s ability to control or bias their evaluations. For instance, activating the left dlPFC via anodal tDCS can enhance cognitive control over emotional information, leading to reduced attentional bias toward threats and diminished negative affective reactivity. In one study, active (anodal) tDCS over left dlPFC attenuated participants’ physiological arousal in anticipation of negative social feedback, consistent with a facilitated emotion-regulation process (Allaert et al., 2022 ). Likewise, Clarke et al. ( 2021 ) reported that frontal tDCS over dlPFC increased individuals’ emotion regulation efficacy, as evidenced by lower self-reported distress in response to unpleasant images (relative to sham). These findings suggest that an active dlPFC can help implement goal-directed reappraisal or suppression of unwanted evaluative reactions. Conversely, inhibitory neuromodulation of the dlPFC tends to have the opposite effect. Mengarelli et al. ( 2015 ) found that applying cathodal tDCS to the left dlPFC impaired participants’ performance in an emotional interference task, as shown by stronger disruptive effects of emotional distractors on reaction times (and corresponding EEG markers of conflict). Within the context of relational EC, the dlPFC's top-down regulatory influence may be critical for inhibiting the default assimilative response to co-occurrence while actively maintaining and integrating the specific relational qualifier (e.g., 'stops') to form a correct propositional judgment. Indeed, a recent study showed that downregulating the left dlPFC with cathodal HD tDCS modulated attitude formation, suggesting that dlPFC-supported propositional processes play an important role in EC (Wąsowicz et al., under review). However, that study could not dissociate whether the impairment was specific to relational processing or also affected basic co-occurrence learning., or both. The present study is designed to resolve this ambiguity using MPT modeling to separately quantify the influence of dlPFC modulation on each process. Although research on neuromodulation in EC is still in its early stages, these converging observations align with the view that the dlPFC is a neural substrate of evaluative control – it helps reconcile new stimulus pairings with existing knowledge and goals. However, the precise relationship between executive resources and relational learning is debated. While it is often assumed that propositional integration is resource-dependent, recent evidence from cognitive load manipulations suggests a more complex picture. For instance, Gawronski ( 2022 ) found that high cognitive load can paradoxically enhance relational processing, possibly by triggering strategic shifts in resource allocation. This raises a critical question: Does the direct causal suppression of a key executive control hub like the dlPFC produce the predicted impairment of relational learning, or might it fail to do so, perhaps even triggering compensatory processes? By leveraging tDCS to directly manipulate dlPFC involvement, in contrast to indirect cognitive load manipulations, we can thus probe whether attitude acquisition in EC is causally sensitive to the availability of executive processing. The Present Study The present study integrates a relational EC paradigm with a dlPFC neuromodulation to test the causal role of executive control in attitude formation. We partially replicated and extended the design of (Kukken et al., 2020 ) by adding a between-subjects factor: participants received either cathodal HD-tDCS over the left dlPFC or a sham stimulation during the evaluative learning phase. Our key question is whether suppressing left dlPFC activity alters the balance of co-occurrence and relational influences. We hypothesized that if the dlPFC is critical for propositional integration, the cathodal tDCS group (vs. sham) would show a reduced impact of relational qualifiers, resulting in more assimilation-based attitudes. Conversely, if relational processing is independent of dlPFC control, tDCS should have no effect. We tested these competing predictions using both direct ratings and MPT modeling. Specifically, we predicted that cathodal stimulation would selectively reduce the MPT parameter for relational processing (m) while leaving the co-occurrence parameter (p) unaffected. In sum, the present study’s methodological contribution is the novel combination of a high-definition tDCS intervention with a relational EC paradigm, allowing a rigorous test of the dlPFC’s causal role in evaluative learning. This approach bridges social-cognitive theory and cognitive neuroscience, providing new insight into whether the control of affective learning (often theorized as “propositional” or reflective processing) is not only conceptually distinct but also neurobiologically dissociable from automatic association formation. Our findings will speak to the dual-process versus single-process debate by revealing whether disrupting a putative “control hub” in the brain selectively impacts the relational aspect of EC, thereby illuminating the mechanisms by which attitudes are formed and regulated. METHODS A priori power analysis was conducted using G*Power (Version 3.1; Faul et al., 2007 ) to determine the sample size required to detect our primary hypothesized interaction effect. We targeted the Stimulation Group × CS Type interaction, which represents the critical test of whether tDCS modulates relational evaluative conditioning. Based on prior research using neuromodulation to influence cognitive processes and The Smallest Effect Size of Interest, we aimed to power our study to detect minimal but theoretically important effect, which is a small-to-medium sized effect (ηp² = .04, equivalent to f = 0.20). The analysis for a mixed-design ANOVA (F-test: Repeated measures, within-between interaction) with two groups and four repeated measures (assuming a correlation of .5 among measures and α = .05) indicated that a minimum total sample of N = 36 would be required to achieve 80% power. To ensure robust and highly powered results, and to account for potential participant exclusions, we substantially exceeded this minimum and recruited a final sample of N = 80. A post hoc sensitivity analysis reveals that this sample size provided greater than 99% power to detect our target effect size, giving us excellent confidence in the stability of our findings and the interpretation of both significant and non-significant results. Participants Eighty Polish native speakers between the ages of 18 and 35 participated in the study. Participants were compensated with Pluxee cash vouchers corresponding to 200 PLN. They were recruited via social media, both the Institute official page and posts on Facebook groups. Participation invites were sent to the people who met all of the inclusion criteria in the study enrollment form. The form included, apart from other, questions regarding people's mental and physical well-being. Specific inclusion criteria are listed in the Supplementary Materials. Apart from that, participants were instructed to get sufficient sleep prior to the study, to refrain from consuming alcohol or caffeinated beverages, and to remove any jewelry from the facial and ear areas. They were informed about the potential risks and their right to withdraw from the experiment at any given moment. Right before each experiment, participants were asked once again to complete the study enrollment form to ensure their eligibility was up to date. The study took place between 10 March and 30 April 2025. The meetings were scheduled using the Calendly platform and/or direct email communication. The study was approved by the Research Ethics Committee of the Institute of Psychology, Polish Academy of Sciences. Procedure The study employed a 2 (US valence: positive vs. negative) × 2 (relation: start vs. stop) × 2 (stimulation manner: active vs. sham) mixed design, with the first two factors as within-subject variables. HD-tDCS stimulation was initiated at the beginning of the experiment and was followed by a battery of tasks. The specific tasks included in the experiment are presented in chronological order in Fig. 1 . and are described in detail later in this section. The Conditioning Procedure, CS Prerating, Dichotomous Evaluation Task, and Memory Task were implemented in accordance with the procedures described by (Kukken et al., 2020 ). The experiment script was written and launched in Millisecond Inquisit 6 Lab (Version 6.6.3 for Windows). The Stroop Task was additionally included for two reasons. Its primary and crucial purpose was to ensure that the electrical stimulation affected participants’ higher-order cognitive processing. Its secondary, supporting function was to fill the time required for the stimulation to end and to provide a delay before the evaluation tasks, allowing any potential after-effects of the stimulation to subside. The US Pre- and Postrating Tasks were also added to the original procedure to verify the affective value and relevance of the unconditioned stimuli for the Polish sample, and to examine potential habituation effects resulting from repeated exposure to the US. Two technical breaks were included in the procedure (see Fig. 1 .). After the first session of the Stroop Task, a short break was introduced to allow for the removal of the tDCS cap and the placement of headphones in preparation for the US Prerating Task. A longer break followed the second session of the Stroop Task, intended to allow potential after-effects of the stimulation to subside. The duration of this second break was calculated as 40 minutes minus the time elapsed since the start of the stimulation . This adjustment accounted for differences in task completion times and the duration of the first break across participants, ensuring sufficient recovery time without unnecessary waiting. As a result, each experimental session had a fixed duration. The second break took the form of a screen with a loading bar. Participants were informed of the duration of the waiting period. After completing the battery of tasks, participants filled out a brief questionnaire regarding their experience and sensations during stimulation, to assess whether the procedure was fully comfortable. The entire procedure, including tDCS preparation and the completion of consent and demographic forms, lasted approximately 90 minutes. Electrical stimulation. Cathodal stimulation at an intensity of 1 mA () was applied using high-definition transcranial direct current stimulation (HD-tDCS). The stimulation was assessed using Neuroelectrics Starstim wireless 8-channel hardware and administered with the use of Neuroelectrics Instrument Controller (NIC) software (version 2.1.3.11 for Windows). A circular saline-soaked sponge electrode (8 cm²) functioned as the cathode and was positioned over the left dorsolateral prefrontal cortex (dlPFC) at the F3 site, based on the international 10–20 EEG electrode placement system. Four additional electrodes acting as anodes were placed around the cathode at Fp1, Fz, C3, and F7. Active stimulation involved a 40-second ramp-up and ramp-down, with 15 minutes of stimulation (total duration: 16 minutes and 20 seconds). Prior to the session, participants were semi-randomly assigned to one of two stimulation groups – either active tDCS or sham – in a double-blind manner, consistent with protocols demonstrated to produce inhibitory effects on cortical excitability and modulate performance on executive function tasks (Friehs & Frings, 2019 ). CS Prerating Task. To ensure that the conditioned stimuli (CS) were neutral prior to the conditioning procedure, participants completed the CS Prerating Task. They rated 200 images of Pokémon pictures, in a random order, on a scale from − 99 (very negative) to 99 (very positive). Each rating was provided using a mouse and had to be confirmed to avoid accidental responses. The CS pictures were utilized from the original (Kukken et al., 2020 ) study. Based on these ratings, the 40 images closest to a neutral score were individually selected for each participant to be used as conditioned stimuli. The stimuli were randomly assigned to one of the four CS types: Starting Positive (SP), Starting Negative (SN), Ending Positive (EP), and Ending Negative (EN), resulting in 10 stimuli in each group. Stroop Task. A word-color version of the Stroop Task was utilized twice during the procedure. The task consisted of color words or rectangles displayed in one of four colors (red, green, blue, black). It included three types of trials: congruent (word matches the color), incongruent (word and color do not match), and control (a colored rectangle). During the first session, each type of trial was repeated 7 times in random order, resulting in 84 trials in total. The second session consisted of 40 repetitions per trial type, totaling 480 trials. Participants were instructed to indicate the color of the displayed stimulus by pressing the corresponding key on the keyboard: 'D' for red, 'F' for green, 'J' for blue, and 'K' for black. This key mapping was displayed at the top of the screen throughout the task as a reminder. Right before the start, participants were asked to place both hands on the keyboard, with the middle and index fingers of the left hand on 'D' and 'F', and the right-hand fingers on 'J' and 'K'. The task had no response time restrictions; however, participants were instructed to respond correctly as quickly as possible. If a response was incorrect, a red 'X' was displayed for 400 ms. Inter-trial intervals were set to 200 ms. US Prerating & Postrating. Positive and negative sounds were used as unconditioned stimuli (US), adapted from the study by (Kukken et al., 2020 ). The positive US consisted of a song fragment, while the negative US resembled a human scream. Each sound was played for 1750 ms. The Pre- and Postrating Tasks had the same structure. Each task included two trials – one rating per US – presented in random order. Participants were asked to rate each sound on a 199-point scale ranging from very negative to very positive. As in the CS Prerating Task, each response had to be confirmed. The sounds played automatically at the start of each trial and could not be repeated. Conditioning procedure. The conditioning procedure was similar to that described by (Kukken et al., 2020 ), with the exception that the duration of each stimulus exposure was fixed rather than randomly distributed. The procedure used a set of 40 conditioned stimuli (10 for each CS type: SP, SN, EP, EN), selected individually for each participant during the CS Prerating Task as the most neutral. Different presentation sequences were applied depending on whether a CS was used to start or end a US. The trial structure is illustrated in Fig. 2 . The conditioning phase included a total of 240 trials (10 stimuli × 4 CS types × 6 repetitions), presented in random order. Prior to the conditioning procedure, participants were informed about the different roles of the images and were instructed to try to memorize which image served which role. The exact instructions provided to participants are available in the Supplementary Materials. Dichotomous Evaluation Task. The Dichotomous Evaluation Task was implemented in a similar form to that used by (Kukken et al., 2020 ). Participants were asked to evaluate each CS image as either positive or negative by pressing the 'P' or 'N' key on the keyboard, respectively. They were instructed to respond based on their feelings toward the images, not their memory of them. In cases where they did not have a strong opinion, they were asked to choose the option that felt most suitable. All 40 images were presented one by one in random order. There was no time limit for providing a response. Memory Task. In the Memory Task, participants were asked to recall the meaning of each CS: a stimulus was considered positive if it started a positive sound or ended a negative one, and negative if it started a negative sound or ended a positive one. For CS with a positive meaning, participants were instructed to press the 'P' key on the keyboard, and the 'N' key for those with a negative meaning. If they could not remember the meaning of a given stimulus, they were asked to make a guess. The exact instructions for this task, as given to participants, are provided in the Supplementary Materials. As in the Evaluation Task, all responses were self-paced, and the CS were presented in random order. RESULTS Data Cleaning & Subject Exclusion Eighty participants took part in the study; however, two were excluded from the final dataset in advance. One participant was excluded due to technical issues that occurred during the experiment. The second was excluded due to an exceptional lack of attentiveness while performing the tasks, as observed by the experimenter. Additional data cleaning procedures were applied and are described below. In line with the data cleaning procedures used by (Kukken et al., 2020), we planned to exclude participants who provided the same response in more than 90% of the Memory Task trials. However, no participant met this criterion; the highest proportion of identical responses was 85%. Another exclusion criterion concerned participants whose mean ratings in the CS Prerating Task deviated from the overall sample mean (M = 0.91, SD = 9.92) by more than three standard deviations. One participant met this criterion and was excluded from the analyses. Additionally, individual trials were excluded if the CS prerating deviated from the participant's mean by more than three standard deviations. Among 3080 trials, there was only 1 such trial. In addition to the procedures used by (Kukken et al., 2020), further data cleaning steps were applied to tasks not included in the original study. For example, a US prerating task was administered to ensure that the subjective valence of the unconditioned stimuli (US) aligned with the intended categories. Participants who rated positive US as negative or negative US as positive were to be excluded. One participant met this criterion and was therefore excluded from all analyses. The Stroop task was administered in two separate sessions. Participants were to be excluded if their error rate exceeded 50% in the first session. No participants met this exclusion criterion; the highest observed error rate was 16%. Subsequently, within-subject trial-level exclusions were applied based on both Stroop sessions combined. A trial was considered invalid if it met at least one of the following criteria: (a) reaction time (RT) shorter than 200 ms, (b) RT exceeded 3,000 ms, or (c) RT exceeded three standard deviations from the participant’s mean RT. Based on these criteria, 988 trials were excluded from the analyses. The participant with the largest proportion of excluded trials had 8.5% of their trials removed. Metrics The analyses were conducted on data from 76 participants, out of which 54 were female and 22 were male. The mean age was 24.76 years (SD = 4.53). Participants belonged to one of two HD-tDCS stimulation groups: 39 were in the active stimulation group and 37 in the sham group. CS Prerating Does the mean CS prerating differ from the neutral midpoint? To test whether the mean rating of the 40 chosen as most neutral CS images in the Prerating task significantly differed from zero (neutral midpoint), a linear mixed-effects model (LMM) with a random intercept for participant was conducted to account for the dependency of observations within individuals. The model was estimated using restricted maximum likelihood (REML), and t -tests with Satterthwaite’s approximation for degrees of freedom were applied. The results did not reveal a significant difference from zero, b = 0.31, SE = 0.23, t (74.13) = 1.34, p = .18. Does the mean CS prerating differ based on the CS type or tDCS stimulation group? To test whether the mean image ratings in the prerating phase differed by tDCS group (active vs. sham), CS type, or their interaction, a linear mixed-effects model (LMM) with a random intercept for participant was used. Group, CS type, and their interaction were included as fixed effects, with participant as a random intercept. The model revealed no significant main effect of group ( b = − 1.32, SE = 0.71, t (107.39) = − 1.86, p = .07), nor of CS type compared to the reference category (StartingPositive; all p s > .10). Interaction effects between stimulation group and CS type were also non-significant (all p s > .05). Full model results are provided in the Supplementary Materials. US Pre- and Postrating Is there a difference between the mean US ratings before and after conditioning? A paired-samples t-test revealed no significant difference in the ratings of the positive US between the pre- and postrating phases, t(75) = − 0.22, p = .83, 95% CI [–9.29, 7.45]. A similar analysis showed a significant difference for the negative US, t(75) = 2.74, p = .008, 95% CI [1.56, 9.91]. The mean ratings for the positive and negative US in the pre- and postrating phases are presented in Fig. 3. Did tDCS stimulation mode influence US postrating scores? The mean postrating of the positive US in the active group was 49.85, compared to 52.05 in the sham group. An independent-samples t -test revealed no significant difference between the active and sham tDCS groups, t (64.30) = − 0.26, p = .80, 95% CI [–19.33, 14.92]. A similar analysis was conducted for the negative US. The mean postrating of the negative US was − 85.72 in the active group and − 83.30 in the sham group. Again, there was no significant difference between the groups, t (65.27) = − 0.51, p = .61, 95% CI [–11.95, 7.10]. Stroop Task Before examining the primary hypothesis, we first tested whether our cathodal tDCS manipulation successfully modulated higher-order cognitive processing as intended. For the purposes of the analysis, data from both Stroop Task sessions were combined into a single dataset. A Type III analysis of variance (ANOVA) based on a linear mixed-effects model (LMM) with a random intercept for participants was conducted to examine differences in reaction time (RT) as a function of Stroop Task trial type (congruent, incongruent, control) and tDCS stimulation mode (active vs. sham). The analysis revealed a significant main effect of trial type, F (2, 37,055.92) = 4.06, p < .001, with a small effect size, η²ₚ = .03, 90% CI [0.02, 1.00]. The main effect of stimulation group also reached significance, F (1, 73.48) = 4.06, p = .048, with an effect size of η²ₚ = .05, 90% CI [0.00, 1.00], indicating a small-to-medium effect. No significant interaction between trial type and stimulation group was found, F (2, 37,055.92) = 0.13, p = .874. Post hoc comparisons were conducted to further explore differences in RTs across trial types and stimulation groups. Contrast analysis with Tukey’s correction revealed significant differences between congruent and incongruent trials, estimate = − 125.5 ms, z = − 27.22, p < .0001, with a small effect size, d = 0.17, 95% CI [0.00, 0.34], as well as between control and incongruent trials, estimate = − 130.29 ms, z = − 28.20, p < .0001, Cohen’s d = 0.17, 95% CI [0.01, 0.34]. No significant difference was found between congruent and control trials, estimate = 4.79 ms, z = 1.05, p = .545, d = 0.16, 95% CI [–0.01, 0.33]. Stimulation mode significantly affected RTs for congruent trials (estimate = 61.13 ms, z = 2.01, p = .045) with a small effect size, d = 0.17, 95% CI [0.00, 0.34], and incongruent trials (estimate = 62.4 ms, z = 2.05, p = .041), Cohen’s d effect size was also small, d = 0.17, 95% CI [0.01, 0.34]. For control trials, the difference approached significance but did not reach it (estimate = 57.78 ms, z = 1.90, p = .058), Cohen’s d = 0.16, 95% CI [–0.01, 0.33]. As illustrated in Fig. 4, participants in the active tDCS group were significantly slower than the sham group. This result confirms that our neuromodulation protocol was effective in disrupting performance on a classic executive function task. Although we initially planned to include an analysis of error rates, a ceiling effect was observed in this measure, as participants made very few mistakes. As a result, error rates could not be considered a reliable indicator of between-subject differences. Dichotomous Evaluation Task Mean Evaluative Conditioning scores (EC effect) were calculated as the average of responses in the Dichotomous Evaluation Task, where positive responses were coded as 1 and negative responses as − 1. This allowed us to determine whether each type of conditioned stimulus was, on average, perceived more positively or negatively, and to compare with the results in Kukken et al. (2020). Is there an EC effect for each CS type? To determine whether an EC effect was present for each CS type, a series of one-sample t -tests was conducted to assess whether the mean EC scores differed from the neutral midpoint. The results revealed significant differences from the midpoint for all CS types ( p < .001 in all cases). Cohen’s d values ranged from − 0.51 for Ending Negative to 0.76 for Starting Positive, indicating medium-sized effects. Mean EC effect scores are presented in Fig. 5, and detailed results are provided in Table 1. Table 1 Results of one-sample t-tests assessing whether the mean EC effect for each CS type differed from the neutral midpoint. CS type EC effect SD 95% CI n df p value Cohen's d Cohen's d 95% CI Ending Negative − .22 .98 .07 760 75 p < .001 − .51 [-.75, − .27] Ending Positive .27 .96 .07 760 75 p < .001 .61 [.37, .86] Starting Negative − .29 .96 .07 760 75 p < .001 − .69 [-.94, − .44] Starting Positive .31 .95 .07 759 75 p < .001 .76 [.5, 1.01] Does the EC effect size differ based on different CS types or tDCS stimulation groups? To examine the effects of tDCS stimulation type (active vs. sham), CS type, and their interaction on the EC effect, a Type III analysis of variance (ANOVA) based on a linear mixed-effects model (LMM) with a random intercept for participant was conducted. The analysis revealed a significant main effect of CS type, F (3, 2956.8) = 87.10, p < .0001, indicating that the EC effect varied across CS types. Partial eta-squared effect size was small, η²ₚ = 0.08, 95% CI [0.07, 1.00]. There was no significant main effect of tDCS stimulation group, F (1, 73.8) = 0.10, p = .750, and no significant interaction between group and CS type, F (3, 2956.8) = 0.001, p = 1.000. These results suggest that tDCS stimulation did not influence CS ratings in the Evaluation Task and did not interact with CS type in modulating the EC effect. The main effect of CS type was further examined using post hoc contrast analysis with Tukey’s correction. The results showed no significant difference between the Starting vs Ending Negative conditions, Z = 1.37, p = .52, and no difference between the Starting vs Ending Positive conditions, Z = − 0.93, p = .79. All other pairwise comparisons were statistically significant ( p < .001) with medium effect size (Cohen’s d ranging from − 0.52 to − 0.64), and are described in detail in Table 2. Table 2 Post-hoc contrasts between the mean EC effect scores by CS type. contrast estimate diff. SE Z ratio p value Cohen’s d Cohen’s d 95% CI EN – EP − .49 .05 -10.22 p < .001 *** − .52 [-.63, − .42] EN – SN .07 .05 1.37 p = .52 .07 [-.03, .17] EN – SP − .54 .05 -11.15 p < .001 *** − .57 [-.67, − .47] EP – SN .56 .05 11.59 p < .001 *** .59 [.49, .70] EP – SP − .04 .05 -0.93 p = .79 − .05 [-.15, .05] SN – SP − .60 .05 -12.52 p < .001 *** − .64 [-.74, − .54] Memory Task Having established that cathodal tDCS successfully modulated executive control, we next examined our primary question: does this disruption of the left dlPFC function selectively impair the relational component of evaluative conditioning? Percentages of correct responses in the Memory Task were calculated for each CS type and are later referred to as the HIT Ratio. Following Kukken et al. (2020), correct implications were defined as follows: CS types SP and EN were associated with a positive meaning, whereas SN and EP – with a negative meaning. To test the accuracy of responses in the Memory Task, a series of one-sample t -tests was conducted, comparing the mean HIT Ratio for each CS type to the chance level of 50%. Mean HIT scores for each CS type are shown in Fig. 6. The results showed that for all CS types, the difference from the chance level was statistically significant, with a large effect for CS starting US and medium for those ending US. For CS that started a positive US, the HIT Ratio differed significantly from chance, t (75) = 9.21, p < .001, 95% CI [67.82, 77.65], with a large effect of Cohen’s d = 2.13, 95% CS [1.56, 2.69]. A similar result was found for CS that started a negative US, t (75) = 12.21, p < .001, 95% CI [72.35, 81.07], which as well showed a large effect, d = 2.82, 95% CI [2.18, 3.45]. CS that ended a positive US also differed significantly from chance, t (75) = − 2.93, p = .005, 95% CI [34.96, 47.15], with a medium effect of d = − 0.64, 95% CI [-1.10, -0.17], as did CS that ended a negative US, t (75) = − 2.77, p = .007, 95% CI [32.35, 47.12], also with a medium effect, d = − 0.68, 95% CI [-1.14, -0.21]. MPT model for Memory Task Data from Memory Task were analyzed using multinomial processing tree (MPT) modelling with the TreeBUGS package in R. Following (Kukken et al., 2020), response frequencies were modeled using latent-trait approach. The model was applied to individual response frequencies for each CS type. To estimate the potential effects of tDCS stimulation (active vs. sham) on the task performance, we added fixed group effects to the model on all parameters. The processing tree model used to represent participants’ responses was adapted from the original study. In the model, each node reflects a discrete cognitive state, and the associated parameters represent the probability of a response being guided by that state. The model includes three parameters: the probability of correctly recalling the meaning of the CS ( m parameter), the likelihood that the response was based on the CS–US association ( p parameter), and the probability of guessing correctly ( g parameter). The same tree structure was applied for modeling Memory and Evaluation Tasks’ responses. A detailed definition of the tree structure as well as the detailed results of the Evaluation Task MPT model are provided in the Supplementary Materials. The MPT model was estimated using four Markov chains with 50,000 iterations each. The adaptation (burn-in) phase was set to 50,000 iterations, and a thinning rate of 1 was applied, meaning that every posterior sample after burn-in was retained, thus maximizing the number of usable samples. Following Kukken et al. (2020), we tested for differences in response frequencies between participants. Pearson’s chi-square test revealed significant variability across participants, χ²(525) = 875.68, p < .001, suggesting meaningful individual differences in underlying parameter estimates. Posterior predictive checks indicated good model fit. The posterior predictive p-value (PPP) for the mean response structure (T1) was .466 (observed = 0.04; predicted = 0.04), while the PPP for the covariance structure (T2) was .343 (observed = 3.09; predicted = 2.44). Both values fall well within the acceptable range of [.05, .95], suggesting that the model accurately reproduced both response proportions and their covariation. All Markov chain Monte Carlo (MCMC) chains showed good convergence, with Gelman–Rubin statistics (R̂) below 1.04 for all parameters. The estimated parameters are provided on a probability scale. The m-parameter indicated that participants correctly remembered the meaning of the CS in 3.5% of trials. The p -parameter showed that in 29.2% of cases (when the meaning was not remembered) responses followed the paired US valence. The g -parameter reflects that when guessing occurred, participants showed no particular bias towards any answer. Estimated parameters as well as their correlations are presented in Table 3. No clear group-level effects of tDCS stimulation (active vs. sham) were observed on any of the parameters, as all 95% credible intervals for group effects overlapped with zero. For example, the group effect on the m -parameter (meaning) in a latent probit scale was Δ = 0.127, 95% CI [–0.33, 0.62]. All group effects are presented in Table 3. Table 3. Response frequencies, parameter estimates, estimated parameter correlations, and effects of group assignment in the MPT models for the Memory and Evaluation Tasks. Memory Task Evaluation Task Response frequencies SP 552 (73%) 499 (66%) EP 312 (41%) 277 (36%) SN 583 (77%) 489 (64%) EN 302 (40%) 296 (39%) Parameter estimates m .035 [.01, .09] .014 [.00, .04] p .292 [.15, .43] .209 [.12, .30] g .465 [.41, .52] .519 [.47, .57] Correlations m – p -.507 [–0.81, –0.13] .083 [-.58, .71] m – g .588 [0.16, 0.87] -.063 [-.73, .66] p – g -.462 [–0.79, –0.03] .012 [-.38, .39] active sham active sham Effects of factors (shift from overall mean) m .127 [-.33, .62] -.127 [-.62, .33] .120 [-.55, .86] -.120 [-.86, .55] p -.019 [-.37, .33] .019 [-.33, .37] .014 [-.25, .29] -.014 [-.29, .25] g .014 [-.11, .13] -.014 [-.13, .11] -.017 [-.13, .10] .017 [-.10, .13] Table notes Response frequencies for correct answers were aggregated across all participants. Correct responses were defined as those consistent with the implied meaning of the CS – that is, a positive response for SP and EN, and a negative response for SN and EP. Parameter estimates are presented on the probability scale and reflect the average likelihood of relying on each latent process across participants, irrespective of group assignment. Correlations between parameters, as well as effects of group factors, are estimated on the latent probit scale. Group assignment was coded as either sham or active tDCS stimulation. DISCUSSION The present study investigated the causal role of the left dorsolateral prefrontal cortex (dlPFC) in relational evaluative conditioning (EC). We successfully replicated the basic finding that relational qualifiers modulate attitude formation (Kukken et al., 2020 ), but our primary hypothesis was not supported. We found a critical dissociation: cathodal HD-tDCS successfully impaired performance on an executive control task (the Stroop task) but had no effect on attitude formation. This null effect of neuromodulation was consistent across both explicit evaluations and the cognitive processes of relational learning and co-occurrence tracking, as estimated by Multinomial Processing Tree (MPT) modeling. These findings, emerging from a highly-powered design, provide robust evidence that challenges the presumed role of the left dlPFC in this learning context. The absence of a tDCS effect on relational EC is a theoretically informative finding. We predicted that inhibitory stimulation of the left dlPFC would impair propositional processing, but this was not observed in either the overall ratings or the MPT model's relational parameter (m). Given our high statistical power and the successful manipulation check, this null result challenges the hypothesis that the executive functions subserved by the left dlPFC are indispensable for propositional operations in this EC paradigm (cf. Mengarelli et al., 2015 ; Ochsner et al., 2004 ). Our data suggest that the link between domain-general executive control and relational learning is less direct than previously assumed, a conclusion that aligns with recent complex findings from cognitive load studies (Gawronski, 2022 ). Our results may be interpreted as supporting the view that EC effects are not dependent on higher-order cognitive processing during the encoding phase, specifically in the left dlPFC. Although cathodal stimulation of this region has been shown to diminish preference change in a study involving higher-order processing (Mengarelli et al., 2015 ), it is possible that the processes crucial for the emergence of EC effects occur in a different brain region. The successful modulation of Stroop task performance is a key methodological strength of this study. Confirming that our tDCS protocol effectively disrupted reaction times on a classic executive function task provides evidence that the neuromodulation was indeed successful. We focused on reaction times because accuracy was near-ceiling, precluding a reliable analysis of error rates. The MPT model of our Memory Task data, when compared to Kukken et al. ( 2020 ), revealed a substantial procedural impact. Our relational knowledge parameter (m = .035) was drastically lower than in the original studies (m ≈ .21–.57), indicating that participants relied on explicit relational memory in only 3.5% of trials. This is best explained by the 23-minute delay in our procedure, which likely caused significant forgetting. This floor effect on relational knowledge provides context for our co-occurrence parameter (p = .292), which was slightly higher than in the original work. This pattern is consistent with the model's logic: with diminished relational knowledge, participants defaulted more often to the simpler co-occurrence heuristic (Gawronski & Brannon, 2021 ). Critically, this deeper analysis confirmed the primary null finding: tDCS did not credibly affect any MPT parameter. Behavioral data from the Memory Task further highlighted participants' cognitive challenges. While memory was accurate for CSs that started sounds, performance was systematically below chance for CSs that stopped sounds. This suggests participants were not guessing, but rather actively misremembering the implications. We interpret this as a cognitive conflict where the salient valence of the US overrode the more complex propositional meaning of the "stop" relation, leading to consistent errors. This pattern reveals the difficulty of overriding simple co-occurrence signals, even when performing an explicit memory task. Despite the study's strengths, including high statistical power and a successful manipulation check, several limitations must be acknowledged, each pointing toward important avenues for future research. The primary limitation stems from the procedural delay between the conditioning and testing phases. As revealed by the MPT analysis, this 23-minute interval, necessary for tDCS after-effects to subside, likely led to substantial forgetting of the specific CS-US relations, resulting in a floor effect on the relational knowledge parameter (m). This presents a significant challenge for interpreting our primary null finding. Cathodal tDCS was intended to impair a cognitive process, but it is plausible that there was very little active relational processing left to suppress by the time of the memory and evaluation tasks. In essence, one cannot detect a neuromodulatory impairment of a process that is already at its floor. Second, limitations inherent to HD-tDCS technology warrant consideration. Although we used a high-definition montage to improve spatial focality, the resolution is still diffuse compared to techniques like transcranial magnetic stimulation (TMS) (Chrysikou & Hamilton, 2011 ). Furthermore, while our protocol was effective in modulating Stroop performance, the optimal "dose" in terms of intensity and duration required to influence the specific neural computations of relational EC remains unknown. It is possible that the cognitive load imposed by the Stroop task is more susceptible to this form of broad neuromodulation than the processes involved in applying a learned relational rule. Finally, our design tested the causal role of the dlPFC exclusively during the encoding phase of learning. It is equally plausible that the critical contribution of the dlPFC is not in the initial formation of propositional links, but during their active retrieval and application at the time of judgment. The process of searching memory for the correct relational tag and using it to override a default assimilative response could be more executively demanding than the initial learning itself. Our study was not designed to test this retrieval hypothesis and therefore cannot speak to the dlPFC's role during the evaluation phase. The limitations of the present study directly inform several promising avenues for future research. First, to address the encoding versus retrieval question, a future study could apply neuromodulation not during the learning phase, but during the subsequent evaluation task. This would provide a direct test of whether the dlPFC is causally involved in the active retrieval and application of propositional knowledge, a process that may be more executively demanding than initial encoding (cf. Wąsowicz et al. under review). Second, to overcome the inherent limitations in the spatial resolution of tDCS, future work could employ more focal techniques like transcranial magnetic stimulation (TMS) to provide a more conclusive test of the left dlPFC's specific role. Third, to increase the cognitive demand and prevent the floor effects on relational knowledge that we observed, future studies should consider using less complex or demanding EC paradigms. For example, a task involving somewhat simpler relational rules would place less load on executive resources and allow for detecting the effects of neuromodulation. Future work should also investigate whether participants indeed attribute causal role to CSs as starting/stopping affective stimuli. For now it is only assumed by researchers that temporal configuration of CS-US pairs invokes relational interpretations. In conclusion, this study provides highly-powered evidence for a dissociation between the neural substrates of general executive control and those specifically recruited for relational evaluative conditioning. Our results demonstrate that while inhibitory stimulation of the left dlPFC successfully impairs performance on a classic executive control task, it does not alter the balance of co-occurrence-based and relation-based processes in attitude formation. This challenges the straightforward view that the left dlPFC is an indispensable substrate for integrating relational information in this context. Ultimately, our findings suggest that while the dlPFC is undoubtedly a key hub for executive function, its causal role in this fundamental form of attitude formation may be less direct or more task-dependent than previously assumed, paving the way for more nuanced neurocognitive models of how we learn our likes and dislikes Declarations Funding. This work was supported by the National Science Center grant no. 2019/33/B/HS6/02700. Author Contribution All authors contributed substantially to the study. J.W. and R.B. designed the experimental paradigm. J.W. programmed the experimental script, managed the necessary equipment, and coordinated participant recruitment. J.W. and K.G. conducted the experimental sessions and performed data analysis. J.W. and R.B. wrote the main manuscript, and J.W. prepared the figures. All authors reviewed and approved the final version of the manuscript. Data Availability The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request. References Allaert, J., De Raedt, R., Sanchez-Lopez, A., Baeken, C., & Vanderhasselt, M. A. (2022). Mind the social feedback: Effects of tDCS applied to the left DLPFC on psychophysiological responses during the anticipation and reception of social evaluations. Social Cognitive and Affective Neuroscience , 17 (1), 131–141. https://doi.org/10.1093/scan/nsaa066 Chrysikou, E. G., & Hamilton, R. H. (2011). Noninvasive brain stimulation in the treatment of aphasia: Exploring interhemispheric relationships and their implications for neurorehabilitation. Restorative Neurology and Neuroscience , 29 (6), 375–394. https://doi.org/10.3233/RNN-2011-0610 Clarke, P. J. F., Haridas, S. M. P., Van Bockstaele, B., Chen, N. T. M., Salemink, E., & Notebaert, L. (2021). Frontal tDCS and Emotional Reactivity to Negative Content: Examining the Roles of Biased Interpretation and Emotion Regulation. Cognitive Therapy and Research , 45 (1), 19–30. https://doi.org/10.1007/s10608-020-10162-9 De Houwer, J., Thomas, S., & Baeyens, F. (2001). Associative Learning of Likes and Dislikes: A Review of 25 Years of Research on Human Evaluative Conditioning. Psychological Bulletin , 127 (6), 853–869. https://doi.org/10.1037/0033-2909.127.6.853 Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods , 39 (2), 175–191. https://doi.org/10.3758/BF03193146 Förderer, S., & Unkelbach, C. (2012). Hating the cute kitten or loving the aggressive pit-bull: EC effects depend on CS-US relations. Cognition and Emotion , 26 (3), 534–540. https://doi.org/10.1080/02699931.2011.588687 Friehs, M. A., & Frings, C. (2019). Cathodal tDCS increases stop-signal reaction time. Cognitive Affective & Behavioral Neuroscience , 19 (5), 1129–1142. https://doi.org/10.3758/s13415-019-00740-0 Gawronski, B. (2022). Attitudinal Effects of Stimulus Co-occurrence and Stimulus Relations: Paradoxical Effects of Cognitive Load. Personality & Social Psychology Bulletin , 48 (10), 1438–1450. https://doi.org/10.1177/01461672211044322 Gawronski, B., Balas, R., & Hu, X. (2016). What do I know about it versus how do I feel about it? A theoretical analysis of response processes in evaluative conditioning. Social Cognition , 34 (5). https://doi.org/10.1521/soco.2016.34.5.357 Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin , 132 (5), 692–731. https://doi.org/10.1037/0033-2909.132.5.692 Gawronski, B., & Bodenhausen, G. V. (2011a). The Associative–Propositional Evaluation Model. In Advances in Experimental Social Psychology (Vol. 44, pp. 59–127). Elsevier. https://linkinghub.elsevier.com/retrieve/pii/B9780123855220000020 Gawronski, B., & Bodenhausen, G. V. (2011b). The associative-propositional evaluation model. Theory, evidence, and open questions. In Advances in Experimental Social Psychology (Vol. 44). Elsevier Inc. https://doi.org/10.1016/B978-0-12-385522-0.00002-0 Gawronski, B., & Bodenhausen, G. V. (2018). Evaluative Conditioning From the Perspective of the Associative-Propositional Evaluation Model. Social Psychological Bulletin , 13 (3), e28024. https://doi.org/10.5964/spb.v13i3.28024 Gawronski, B., Bodenhausen, G. V., & Becker, A. P. (2007). I like it, because I like myself: Associative self-anchoring and post-decisional change of implicit evaluations. Journal of Experimental Social Psychology , 43 (2), 221–232. https://doi.org/10.1016/j.jesp.2006.04.001 Gawronski, B., & Brannon, S. M. (2021). Attitudinal Effects of Stimulus Co-Occurrence and Stimulus Relations: Range and Limits of Intentional Control. Personality and Social Psychology Bulletin , 47 (12), 1654–1667. https://doi.org/10.1177/0146167220982906 Gawronski, B., Brannon, S. M., & Bodenhausen, G. V. (2016). The associative-propositional duality in the representation, formation, and expression of attitudes. Reflective and Impulsive Determinants of Human Behavior , 1 , 103–118. https://doi.org/10.4324/9781315523095 Heycke, T., & Gawronski, B. (2020). Co-occurrence and relational information in evaluative learning: A multinomial modeling approach. Journal of Experimental Psychology: General , 149 (1). https://doi.org/10.1037/xge0000620 Hofmann, W., De Houwer, J., Perugini, M., Baeyens, F., & Crombez, G. (2010). Evaluative Conditioning in Humans: A Meta-Analysis. Psychological Bulletin , 136 (3), 390–421. https://doi.org/10.1037/a0018916 Hu, X., Gawronski, B., & Balas, R. (2017). Propositional Versus Dual-Process Accounts of Evaluative Conditioning: I. The Effects of Co-Occurrence and Relational Information on Implicit and Explicit Evaluations. Personality and Social Psychology Bulletin , 43 (1), 17–32. https://doi.org/10.1177/0146167216673351 Hughes, S., Ye, Y., & De Houwer, J. (2019). Evaluative conditioning effects are modulated by the nature of contextual pairings. Cognition and Emotion , 33 (5). https://doi.org/10.1080/02699931.2018.1500882 Hütter, M. (2022). An integrative review of dual- and single-process accounts of evaluative conditioning. Nature Reviews Psychology . https://doi.org/10.1038/s44159-022-00102-7 Kukken, N., Hütter, M., & Holland, R. W. (2020). Are there two independent evaluative conditioning effects in relational paradigms? Dissociating the effects of CS-US pairings and their meaning. Cognition and Emotion , 34 (1), 170–187. https://doi.org/10.1080/02699931.2019.1617112 Levey, A. B., & Martin, I. (1975). Classical conditioning of human ‘evaluative’ responses. Behaviour Research and Therapy , 13 (4), 221–226. https://doi.org/10.1016/0005-7967(75)90026-1 Mengarelli, F., Spoglianti, S., Avenanti, A., & di Pellegrino, G. (2015). Cathodal tDCS Over the Left Prefrontal Cortex Diminishes Choice-Induced Preference Change. Cerebral Cortex , 25 (5), 1219–1227. https://doi.org/10.1093/cercor/bht314 Moran, T., & Bar-Anan, Y. (2013). The effect of object-valence relations on automatic evaluation. Cognition and Emotion , 27 (4), 743–752. https://doi.org/10.1080/02699931.2012.732040 Moran, T., & Bar-Anan, Y. (2020). The effect of co-occurrence and relational information on speeded evaluation. Cognition and Emotion , 34 (1), 144–155. https://doi.org/10.1080/02699931.2019.1604321 Moran, T., Bar-Anan, Y., & Nosek, B. A. (2016). The assimilative effect of co-occurrence on evaluation above and beyond the effect of relational qualifiers. Social Cognition , 34 (5). https://doi.org/10.1521/soco.2016.34.5.435 Moran, T., Nudler, Y., & Bar-Anan, Y. (2023). Evaluative Conditioning: Past, Present, and Future. Annual Review of Psychology , 74 , 245–269. https://doi.org/10.1146/annurev-psych-032420-031815 . Scopus. Ochsner, K. N., Ray, R. D., Cooper, J. C., Robertson, E. R., Chopra, S., Gabrieli, J. D. E., & Gross, J. J. (2004). For better or for worse: Neural systems supporting the cognitive down- and up-regulation of negative emotion. Neuroimage , 23 (2), 483–499. https://doi.org/10.1016/j.neuroimage.2004.06.030 Wąsowicz, J., Balas, R., Uram, P., & Okruszek, Ł. (under review). How do cognitive resource limitations influence evaluative conditioning? A tDCS study . https://doi.org/10.31234/osf.io/srtk4_v1 Footnotes The expected time between the end of the stimulation and the beginning of the Evaluation Task was approximately 23 minutes, which should have been sufficient for most of the stimulation effects to fade. If a participant took longer (or shorter) during the first technical break or required more (or less) time to complete the preceding tasks, the duration of the second break was automatically adjusted to ensure exactly 23 minutes between the end of stimulation and the Evaluation Task—no more, no less. However, this adjustment was implemented during the experimental phase of the project; therefore, the first five participants received a fixed 20-minute break instead. The Inquisit software crashed in the middle of the experimental session, and it was not possible to resume the procedure from the point at which it stopped. A Windows update had been performed shortly before the session, which is presumed to have caused the issue. No further updates were applied, and the problem did not occur again. Additional Declarations No competing interests reported. Supplementary Files ArticleRECSupplementaryMaterials.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7252129","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":500646647,"identity":"06cacf54-ac2a-4adc-a051-4f820baac38c","order_by":0,"name":"Joanna Wąsowicz","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDElEQVRIiWNgGAWjYBACxgYwJcHAwN4AIiFsMOAjqIXnAJoWNoL2SSQQqYV5RvLjDx/bLOTlZz5/eINxR11i/+wGtgc/dzDk4dLCOCPNTHJmm4Rh4+wcYwvGM4cTZ9w5wG7Ye4ahGKeW2QlmzLzbJBKYpXPYJBjbDiRukEhgk+BtY0hsw6kl/fNnkBY2yePPgFrqwFok/+LVkmMgDdLCI8FgBtTCDNYijdeW+W/KJGf+kzCcwQP0S+KZw8YzbiS2G8u2SeD0i2HP8c0fPpypk5dvP/7wxscddbL9M5KPPXzbZpPHj0tLAzIvsYHBsYGBEeQkYDThAPKozmxgsGeARiJOLaNgFIyCUTDiAAB5s1ZH+xvKhAAAAABJRU5ErkJggg==","orcid":"","institution":"Polish Academy of Sciences","correspondingAuthor":true,"prefix":"","firstName":"Joanna","middleName":"","lastName":"Wąsowicz","suffix":""},{"id":500646651,"identity":"2428bd30-6cd0-45a1-a0c9-6add62058d12","order_by":1,"name":"Robert Balas","email":"","orcid":"","institution":"Polish Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Robert","middleName":"","lastName":"Balas","suffix":""},{"id":500646652,"identity":"67b48bb9-4e79-45df-b31a-f57299abc215","order_by":2,"name":"Katarzyna Gajos","email":"","orcid":"","institution":"Polish Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Katarzyna","middleName":"","lastName":"Gajos","suffix":""}],"badges":[],"createdAt":"2025-07-30 11:08:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7252129/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7252129/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89603975,"identity":"e734425b-daf7-472c-9d34-cb2f56a25be8","added_by":"auto","created_at":"2025-08-21 19:14:37","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":106866,"visible":true,"origin":"","legend":"\u003cp\u003eThe experimental procedure, depicted in a chronological order. Stroop Task was utilized twice during the experiment.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/a29b0e25f81b1f99775f971d.jpeg"},{"id":89603977,"identity":"ef5c2b2f-b933-4b7b-a283-938d901111df","added_by":"auto","created_at":"2025-08-21 19:14:37","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":388903,"visible":true,"origin":"","legend":"\u003cp\u003eConditioning procedure. Trials within the conditioning procedure were divided into two types: starting and ending conditions. Horizontal arrows represent time within the trial in milliseconds. Both conditions started with a fixation cross displayed for 750 ms. In the starting condition, CS (of type SP or SN) displayed for 1250 ms and was followed by a fixation cross displayed for 1750 ms – till the end of trial. US – either positive for SP or negative for SN – started playing during the exposition of CS, played for 1750 ms, and stopped during the second fixation cross. In the ending condition, the fixation cross was followed by a black screen. After a while, US started playing (positive for EP or negative for EN) and continued for 1750 ms. During the US sound, appropriate CS was displayed and ended after 1250 ms, ending the sound. Black screen was displayed for 750 ms after that. The Inter-trial Interval (ITI) was set to 1000 ms.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/c4ef2d69a20f69e3d0c99754.png"},{"id":89603979,"identity":"87ee7e09-b99e-468d-a168-2b3ee1b0dae1","added_by":"auto","created_at":"2025-08-21 19:14:37","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":315108,"visible":true,"origin":"","legend":"\u003cp\u003eMean ratings of CS stimuli in the pre- and postrating phases. Ratings were given on a scale from –99 (very negative) to 99 (very positive). Error bars represent 95% confidence intervals. Selected between-group comparisons were added and calculated using Student’s t-tests.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/68774d1d90c5598b89ec21b8.png"},{"id":89603978,"identity":"ba02e5b7-9996-4b67-ac41-9c6fdc0e3b92","added_by":"auto","created_at":"2025-08-21 19:14:37","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":428278,"visible":true,"origin":"","legend":"\u003cp\u003eMean reaction times (RT) for each trial type in the Stroop Task (combined across both sessions), separated by tDCS stimulation mode. Error bars represent 95% confidence intervals. Selected post hoc comparison results are indicated above the corresponding pairs.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/dbe6970c698a230d209cc6ab.png"},{"id":89604149,"identity":"1c8cb5fb-4815-4564-b5b8-628067bcaab4","added_by":"auto","created_at":"2025-08-21 19:22:37","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":290549,"visible":true,"origin":"","legend":"\u003cp\u003eMean EC effect for each CS type: starting or stopping a positive or negative US. The pattern suggests that the valence of conditioned affect towards CS is rather influenced by the valence of the US than by its relation to the US. This pattern is generally consistent with the findings of Kukken et al. (2020), with the exception that the EC effect in the stopping condition reached statistical significance in the present study, which was likely not the case in the original article. Results from selected post hoc comparisons are depicted above the bars.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/9c632987298c338bd744fbed.png"},{"id":89604157,"identity":"f7485a6b-4573-43b1-bc6c-7c778a317d0f","added_by":"auto","created_at":"2025-08-21 19:22:37","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":324752,"visible":true,"origin":"","legend":"\u003cp\u003eFrequencies of providing a correct implication in the Memory Task by US valence and CSxUS relation. Given the binary nature of the task, the neutral midpoint was set at 50%.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/99d62cbc3ec29508275d0df3.png"},{"id":95318370,"identity":"3dd44ac7-32a7-4aea-b766-efc813ade87a","added_by":"auto","created_at":"2025-11-06 16:08:59","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2869715,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/73e7cae1-4e6e-4994-8797-7afc77cc7916.pdf"},{"id":89603982,"identity":"5f0a9771-23dc-4a47-bba2-6e39443cf6a7","added_by":"auto","created_at":"2025-08-21 19:14:37","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":2062392,"visible":true,"origin":"","legend":"","description":"","filename":"ArticleRECSupplementaryMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-7252129/v1/9b7c0e16d22b3ea92758a615.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Inhibiting Left dlPFC Leaves Relational Evaluative Conditioning Unchanged: Evidence From Electrical Brain Stimulation","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eEvaluative conditioning (EC) is a change in the evaluation of a conditioned stimulus (CS) caused by its pairing with a valenced unconditioned stimulus (US) (De Houwer et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Hofmann et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Across procedures, repeated CS-US co-occurrences typically cause assimilative transfer, where the CS acquires the valence of the paired US, a pattern confirmed by meta-analyses (Hofmann et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Early paradigms used simple, contiguous presentations of neutral CSs and affective USs without explicit instructions about their relationship (De Houwer et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Hofmann et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Levey \u0026amp; Martin, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e1975\u003c/span\u003e). This procedural minimalism led to EC\u0026rsquo;s characterization as an automatic, stimulus-driven phenomenon dependent on spatiotemporal contiguity and requiring little conscious awareness or inference (Gawronski, Balas, et al., 2016; Gawronski, Brannon, et al., 2016; Gawronski et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Gawronski \u0026amp; Bodenhausen, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Hofmann et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). In this traditional view, theorists suggested automatic associative mechanisms drive attitude shifts even with minimal information, reinforcing EC\u0026rsquo;s conception as a basic pathway for attitude acquisition. Consequently, standard implementations often omit relational qualifiers, which isolates co-occurrence effects but presumes that co-occurrence alone is sufficient for evaluative change (Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHowever, a growing body of research shows that the stated relationship between a CS and US systematically modulates evaluative outcomes beyond mere co-occurrence (F\u0026ouml;rderer \u0026amp; Unkelbach, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Hughes et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). When given explicit relational qualifiers, the standard assimilative pattern can be altered. Similarity relations (e.g., causes, starts) typically maintain assimilation, while opposition relations (e.g., prevents, stops) often produce contrast effects (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Self-reported evaluations consistently align with the propositional meaning of the relational verb, showing that participants actively integrate this semantic information (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Notably, the contrast effects from opposition relations are often weaker than the assimilative effects from similarity relations, an asymmetry suggesting the influence of countervailing processes (Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Relational framing thus produces different outcomes even with identical CS-US exposure, demonstrating that EC effects are not reducible to simple associative links (Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). This conclusion holds across multiple verbs and stimulus domains (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). These findings challenge a strictly relation-insensitive associative account, instead motivating a framework where propositional integration is a core determinant of EC outcomes.\u003c/p\u003e\u003cp\u003eParadigms that orthogonally manipulate US valence and relational qualifiers consistently reveal two influences: an assimilative pull from CS-US co-occurrence and a modulation based on the relational information (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). These influences align in similarity relations (e.g., CS starts positive US) but conflict in opposition relations (e.g., CS stops positive US). This conflict results in attenuated effects, where contrast effects from opposition relations are typically smaller than assimilative effects from similarity relations, which is a robust pattern replicated across verbs and stimuli (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This asymmetry suggests two partially independent influences: a co-occurrence-based assimilative process, and a relation-based propositional process, whose simultaneous activation produces an outcome. Because observed evaluations reflect a composite of both, their relative strengths remain unclear, highlighting the need for formal separation of these contributions.\u003c/p\u003e\u003cp\u003eThe coexistence of assimilative and relationally qualified outcomes energizes the debate between dual- and single-process EC theories (Gawronski \u0026amp; Bodenhausen, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2006\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2011b\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; H\u0026uuml;tter, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Dual-process models propose two distinct mechanisms: an associative system driven by spatiotemporal contiguity and a propositional system that evaluates relational statements (Gawronski \u0026amp; Bodenhausen, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2006\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2011b\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). In contrast, single-process theories attribute all EC outcomes to propositional operations, suggesting that apparent \u0026ldquo;association-only\u0026rdquo; effects reflect incomplete encoding or retrieval of relational information. While findings like implicit-explicit dissociations are cited as evidence for parallel processes (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), they can also be explained by differential retrieval demands within a single system. Indeed, a recent review concluded that either framework can accommodate most data by adding auxiliary assumptions about processing stages (H\u0026uuml;tter, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). To move beyond such interpretational flexibility, methods are needed that can analytically separate co-occurrence and relational contributions within a single measure (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eNet evaluative outcomes observed in relational EC paradigms can mask the concurrent operation of (a) an assimilative component driven by mere CS\u0026ndash;US co-occurrence and (b) a relation-driven component that can either reinforce or oppose that assimilative pull (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). For example, an opposition configuration (CS stops negative) simultaneously embeds exposure to a negative US (tending to decrease CS evaluation) and a positively valenced relational implication (stopping something bad is good), producing attenuated contrast effects whose absolute magnitudes are smaller than assimilative shifts under similarity relations (Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Likewise, similarity relations (CS starts positive vs. CS starts negative) align relational implication with co-occurrence, yielding larger unidirectional shifts (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Reliance on task dissociation\u0026mdash;contrasting implicit (speeded) and explicit (deliberative) measures\u0026mdash;to infer separate influences has shown that explicit ratings predominantly track relational qualifiers while some indirect measures retain unqualified co-occurrence patterns (Hu et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Moran \u0026amp; Bar-Anan, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). However, such cross-measure comparisons invite ambiguity because differences could arise from encoding, retrieval, or decision-stage constraints rather than distinct learning mechanisms (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Consequently, without analytic decomposition, observable mean evaluations in each relational cell represent underdetermined mixtures of latent co-occurrence and relational contributions, limiting theoretical leverage for adjudicating dual-process versus single-process predictions (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). A methodological solution requires within-task, process-partitioning approaches capable of estimating the independent strength of each influence to replace interpretively fragile indirect inferences from attenuated or asymmetrical net effect patterns (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Moran et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eMultinomial processing tree (MPT) modeling offers a formal, within-task solution by estimating the independent probabilities of a co-occurrence (assimilative) versus a relational (propositional) process driving evaluative responses (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Heycke \u0026amp; Gawronski, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). This approach models separate parameters for responses reflecting mere US valence versus those reflecting the propositional meaning of the relation (Heycke \u0026amp; Gawronski, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The functional separation of these processes is powerfully supported by findings of asymmetric malleability: intentional control selectively enhances the relational parameter without affecting the robust co-occurrence parameter (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This selective plasticity aligns with theories positing a controllable propositional process operating alongside a more automatic, co-occurrence-driven influence. Crucially, by providing distinct quantitative indices for each process, MPT modeling offers a clear criterion to test whether causal interventions, like brain stimulation, differentially impact relational versus co-occurrence components (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWhile behavioral and MPT evidence establishes functionally separable co-occurrence and relational components (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Heycke \u0026amp; Gawronski, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), this is indirect regarding the causal neurocognitive substrates distinguishing the two (Gawronski \u0026amp; Bodenhausen, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2006\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2011a\u003c/span\u003e; Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For instance, the finding that intentional focus selectively enhances the relational parameter (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) constrains theory but does not reveal whether this malleability stems from executive control systems or intrinsic representational properties (Heycke \u0026amp; Gawronski, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; H\u0026uuml;tter, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Discriminating between dual-process and single-process interpretations of these patterns requires a causal test. A targeted brain stimulation approach can directly probe whether suppressing neural systems thought to support propositional integration, such as prefrontal executive networks, preferentially impairs the relational component while sparing the co-occurrence effect. Such a test is a critical step to determine whether the relational component's flexibility depends on frontal control resources (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThere is growing evidence that the dlPFC is intimately involved in the reflective, propositional side of evaluative processing, particularly in the regulation and contextualization of emotional responses. Neuroimaging research has shown that the dlPFC acts as a key node in the brain\u0026rsquo;s corticolimbic regulatory circuit, exerting top-down influence to up- or down-regulate emotional reactions (Ochsner et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2004\u003c/span\u003e). Causally, brain stimulation studies have found that modulating the left dlPFC alters people\u0026rsquo;s ability to control or bias their evaluations. For instance, activating the left dlPFC via anodal tDCS can enhance cognitive control over emotional information, leading to reduced attentional bias toward threats and diminished negative affective reactivity. In one study, active (anodal) tDCS over left dlPFC attenuated participants\u0026rsquo; physiological arousal in anticipation of negative social feedback, consistent with a facilitated emotion-regulation process (Allaert et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Likewise, Clarke et al. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) reported that frontal tDCS over dlPFC increased individuals\u0026rsquo; emotion regulation efficacy, as evidenced by lower self-reported distress in response to unpleasant images (relative to sham). These findings suggest that an active dlPFC can help implement goal-directed reappraisal or suppression of unwanted evaluative reactions. Conversely, inhibitory neuromodulation of the dlPFC tends to have the opposite effect. Mengarelli et al. (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) found that applying cathodal tDCS to the left dlPFC impaired participants\u0026rsquo; performance in an emotional interference task, as shown by stronger disruptive effects of emotional distractors on reaction times (and corresponding EEG markers of conflict). Within the context of relational EC, the dlPFC's top-down regulatory influence may be critical for inhibiting the default assimilative response to co-occurrence while actively maintaining and integrating the specific relational qualifier (e.g., 'stops') to form a correct propositional judgment.\u003c/p\u003e\u003cp\u003eIndeed, a recent study showed that downregulating the left dlPFC with cathodal HD tDCS modulated attitude formation, suggesting that dlPFC-supported propositional processes play an important role in EC (Wąsowicz et al., under review). However, that study could not dissociate whether the impairment was specific to relational processing or also affected basic co-occurrence learning., or both. The present study is designed to resolve this ambiguity using MPT modeling to separately quantify the influence of dlPFC modulation on each process.\u003c/p\u003e\u003cp\u003eAlthough research on neuromodulation in EC is still in its early stages, these converging observations align with the view that the dlPFC is a neural substrate of evaluative control \u0026ndash; it helps reconcile new stimulus pairings with existing knowledge and goals. However, the precise relationship between executive resources and relational learning is debated. While it is often assumed that propositional integration is resource-dependent, recent evidence from cognitive load manipulations suggests a more complex picture. For instance, Gawronski (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) found that high cognitive load can paradoxically enhance relational processing, possibly by triggering strategic shifts in resource allocation. This raises a critical question: Does the direct causal suppression of a key executive control hub like the dlPFC produce the predicted impairment of relational learning, or might it fail to do so, perhaps even triggering compensatory processes? By leveraging tDCS to directly manipulate dlPFC involvement, in contrast to indirect cognitive load manipulations, we can thus probe whether attitude acquisition in EC is causally sensitive to the availability of executive processing.\u003c/p\u003e\u003cp\u003e\u003cb\u003eThe Present Study\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe present study integrates a relational EC paradigm with a dlPFC neuromodulation to test the causal role of executive control in attitude formation. We partially replicated and extended the design of (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) by adding a between-subjects factor: participants received either cathodal HD-tDCS over the left dlPFC or a sham stimulation during the evaluative learning phase. Our key question is whether suppressing left dlPFC activity alters the balance of co-occurrence and relational influences. We hypothesized that if the dlPFC is critical for propositional integration, the cathodal tDCS group (vs. sham) would show a reduced impact of relational qualifiers, resulting in more assimilation-based attitudes. Conversely, if relational processing is independent of dlPFC control, tDCS should have no effect. We tested these competing predictions using both direct ratings and MPT modeling. Specifically, we predicted that cathodal stimulation would selectively reduce the MPT parameter for relational processing (m) while leaving the co-occurrence parameter (p) unaffected.\u003c/p\u003e\u003cp\u003eIn sum, the present study\u0026rsquo;s methodological contribution is the novel combination of a high-definition tDCS intervention with a relational EC paradigm, allowing a rigorous test of the dlPFC\u0026rsquo;s causal role in evaluative learning. This approach bridges social-cognitive theory and cognitive neuroscience, providing new insight into whether the control of affective learning (often theorized as \u0026ldquo;propositional\u0026rdquo; or reflective processing) is not only conceptually distinct but also neurobiologically dissociable from automatic association formation. Our findings will speak to the dual-process versus single-process debate by revealing whether disrupting a putative \u0026ldquo;control hub\u0026rdquo; in the brain selectively impacts the relational aspect of EC, thereby illuminating the mechanisms by which attitudes are formed and regulated.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003eA priori power analysis was conducted using G*Power (Version 3.1; Faul et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2007\u003c/span\u003e) to determine the sample size required to detect our primary hypothesized interaction effect. We targeted the Stimulation Group \u0026times; CS Type interaction, which represents the critical test of whether tDCS modulates relational evaluative conditioning. Based on prior research using neuromodulation to influence cognitive processes and The Smallest Effect Size of Interest, we aimed to power our study to detect minimal but theoretically important effect, which is a small-to-medium sized effect (ηp\u0026sup2; = .04, equivalent to f\u0026thinsp;=\u0026thinsp;0.20).\u003c/p\u003e\u003cp\u003eThe analysis for a mixed-design ANOVA (F-test: Repeated measures, within-between interaction) with two groups and four repeated measures (assuming a correlation of .5 among measures and α\u0026thinsp;=\u0026thinsp;.05) indicated that a minimum total sample of N\u0026thinsp;=\u0026thinsp;36 would be required to achieve 80% power. To ensure robust and highly powered results, and to account for potential participant exclusions, we substantially exceeded this minimum and recruited a final sample of N\u0026thinsp;=\u0026thinsp;80. A post hoc sensitivity analysis reveals that this sample size provided greater than 99% power to detect our target effect size, giving us excellent confidence in the stability of our findings and the interpretation of both significant and non-significant results.\u003c/p\u003e\u003cp\u003e\u003cb\u003eParticipants\u003c/b\u003e\u003c/p\u003e\u003cp\u003eEighty Polish native speakers between the ages of 18 and 35 participated in the study. Participants were compensated with Pluxee cash vouchers corresponding to 200 PLN. They were recruited via social media, both the Institute official page and posts on Facebook groups. Participation invites were sent to the people who met all of the inclusion criteria in the study enrollment form. The form included, apart from other, questions regarding people's mental and physical well-being. Specific inclusion criteria are listed in the Supplementary Materials. Apart from that, participants were instructed to get sufficient sleep prior to the study, to refrain from consuming alcohol or caffeinated beverages, and to remove any jewelry from the facial and ear areas. They were informed about the potential risks and their right to withdraw from the experiment at any given moment. Right before each experiment, participants were asked once again to complete the study enrollment form to ensure their eligibility was up to date.\u003c/p\u003e\u003cp\u003eThe study took place between 10 March and 30 April 2025. The meetings were scheduled using the Calendly platform and/or direct email communication. The study was approved by the Research Ethics Committee of the Institute of Psychology, Polish Academy of Sciences.\u003c/p\u003e\u003cp\u003e\u003cb\u003eProcedure\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe study employed a 2 (US valence: positive vs. negative) \u0026times; 2 (relation: start vs. stop) \u0026times; 2 (stimulation manner: active vs. sham) mixed design, with the first two factors as within-subject variables.\u003c/p\u003e\u003cp\u003e HD-tDCS stimulation was initiated at the beginning of the experiment and was followed by a battery of tasks. The specific tasks included in the experiment are presented in chronological order in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. and are described in detail later in this section. The Conditioning Procedure, CS Prerating, Dichotomous Evaluation Task, and Memory Task were implemented in accordance with the procedures described by (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The experiment script was written and launched in Millisecond Inquisit 6 Lab (Version 6.6.3 for Windows).\u003c/p\u003e\u003cp\u003eThe Stroop Task was additionally included for two reasons. Its primary and crucial purpose was to ensure that the electrical stimulation affected participants\u0026rsquo; higher-order cognitive processing. Its secondary, supporting function was to fill the time required for the stimulation to end and to provide a delay before the evaluation tasks, allowing any potential after-effects of the stimulation to subside.\u003c/p\u003e\u003cp\u003eThe US Pre- and Postrating Tasks were also added to the original procedure to verify the affective value and relevance of the unconditioned stimuli for the Polish sample, and to examine potential habituation effects resulting from repeated exposure to the US.\u003c/p\u003e\u003cp\u003eTwo technical breaks were included in the procedure (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.). After the first session of the Stroop Task, a short break was introduced to allow for the removal of the tDCS cap and the placement of headphones in preparation for the US Prerating Task. A longer break followed the second session of the Stroop Task, intended to allow potential after-effects of the stimulation to subside. The duration of this second break was calculated as 40 minutes minus the time elapsed since the start of the stimulation\u003ca class=\"FNLink\" href=\"#Fn1\" id=\"#FNLinkFn1\"\u003e\u003c/a\u003e. This adjustment accounted for differences in task completion times and the duration of the first break across participants, ensuring sufficient recovery time without unnecessary waiting. As a result, each experimental session had a fixed duration. The second break took the form of a screen with a loading bar. Participants were informed of the duration of the waiting period.\u003c/p\u003e\u003cp\u003e After completing the battery of tasks, participants filled out a brief questionnaire regarding their experience and sensations during stimulation, to assess whether the procedure was fully comfortable. The entire procedure, including tDCS preparation and the completion of consent and demographic forms, lasted approximately 90 minutes.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eElectrical stimulation.\u003c/b\u003e Cathodal stimulation at an intensity of 1 mA () was applied using high-definition transcranial direct current stimulation (HD-tDCS). The stimulation was assessed using Neuroelectrics Starstim wireless 8-channel hardware and administered with the use of Neuroelectrics Instrument Controller (NIC) software (version 2.1.3.11 for Windows). A circular saline-soaked sponge electrode (8 cm\u0026sup2;) functioned as the cathode and was positioned over the left dorsolateral prefrontal cortex (dlPFC) at the F3 site, based on the international 10\u0026ndash;20 EEG electrode placement system. Four additional electrodes acting as anodes were placed around the cathode at Fp1, Fz, C3, and F7. Active stimulation involved a 40-second ramp-up and ramp-down, with 15 minutes of stimulation (total duration: 16 minutes and 20 seconds). Prior to the session, participants were semi-randomly assigned to one of two stimulation groups \u0026ndash; either active tDCS or sham \u0026ndash; in a double-blind manner, consistent with protocols demonstrated to produce inhibitory effects on cortical excitability and modulate performance on executive function tasks (Friehs \u0026amp; Frings, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eCS Prerating Task.\u003c/b\u003e To ensure that the conditioned stimuli (CS) were neutral prior to the conditioning procedure, participants completed the CS Prerating Task. They rated 200 images of Pok\u0026eacute;mon pictures, in a random order, on a scale from \u0026minus;\u0026thinsp;99 (very negative) to 99 (very positive). Each rating was provided using a mouse and had to be confirmed to avoid accidental responses. The CS pictures were utilized from the original (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) study. Based on these ratings, the 40 images closest to a neutral score were individually selected for each participant to be used as conditioned stimuli. The stimuli were randomly assigned to one of the four CS types: Starting Positive (SP), Starting Negative (SN), Ending Positive (EP), and Ending Negative (EN), resulting in 10 stimuli in each group.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStroop Task.\u003c/b\u003e A word-color version of the Stroop Task was utilized twice during the procedure. The task consisted of color words or rectangles displayed in one of four colors (red, green, blue, black). It included three types of trials: congruent (word matches the color), incongruent (word and color do not match), and control (a colored rectangle). During the first session, each type of trial was repeated 7 times in random order, resulting in 84 trials in total. The second session consisted of 40 repetitions per trial type, totaling 480 trials. Participants were instructed to indicate the color of the displayed stimulus by pressing the corresponding key on the keyboard: 'D' for red, 'F' for green, 'J' for blue, and 'K' for black. This key mapping was displayed at the top of the screen throughout the task as a reminder. Right before the start, participants were asked to place both hands on the keyboard, with the middle and index fingers of the left hand on 'D' and 'F', and the right-hand fingers on 'J' and 'K'. The task had no response time restrictions; however, participants were instructed to respond correctly as quickly as possible. If a response was incorrect, a red 'X' was displayed for 400 ms. Inter-trial intervals were set to 200 ms.\u003c/p\u003e\u003cp\u003e\u003cb\u003eUS Prerating \u0026amp; Postrating.\u003c/b\u003e Positive and negative sounds were used as unconditioned stimuli (US), adapted from the study by (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The positive US consisted of a song fragment, while the negative US resembled a human scream. Each sound was played for 1750 ms. The Pre- and Postrating Tasks had the same structure. Each task included two trials \u0026ndash; one rating per US \u0026ndash; presented in random order. Participants were asked to rate each sound on a 199-point scale ranging from very negative to very positive. As in the CS Prerating Task, each response had to be confirmed. The sounds played automatically at the start of each trial and could not be repeated.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConditioning procedure.\u003c/b\u003e The conditioning procedure was similar to that described by (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), with the exception that the duration of each stimulus exposure was fixed rather than randomly distributed. The procedure used a set of 40 conditioned stimuli (10 for each CS type: SP, SN, EP, EN), selected individually for each participant during the CS Prerating Task as the most neutral. Different presentation sequences were applied depending on whether a CS was used to start or end a US. The trial structure is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The conditioning phase included a total of 240 trials (10 stimuli \u0026times; 4 CS types \u0026times; 6 repetitions), presented in random order. Prior to the conditioning procedure, participants were informed about the different roles of the images and were instructed to try to memorize which image served which role. The exact instructions provided to participants are available in the Supplementary Materials.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eDichotomous Evaluation Task.\u003c/b\u003e The Dichotomous Evaluation Task was implemented in a similar form to that used by (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Participants were asked to evaluate each CS image as either positive or negative by pressing the 'P' or 'N' key on the keyboard, respectively. They were instructed to respond based on their feelings toward the images, not their memory of them. In cases where they did not have a strong opinion, they were asked to choose the option that felt most suitable. All 40 images were presented one by one in random order. There was no time limit for providing a response.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMemory Task.\u003c/b\u003e In the Memory Task, participants were asked to recall the meaning of each CS: a stimulus was considered positive if it started a positive sound or ended a negative one, and negative if it started a negative sound or ended a positive one. For CS with a positive meaning, participants were instructed to press the 'P' key on the keyboard, and the 'N' key for those with a negative meaning. If they could not remember the meaning of a given stimulus, they were asked to make a guess. The exact instructions for this task, as given to participants, are provided in the Supplementary Materials. As in the Evaluation Task, all responses were self-paced, and the CS were presented in random order.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cstrong\u003eData Cleaning \u0026amp; Subject Exclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEighty participants took part in the study; however, two were excluded from the final dataset in advance. One participant was excluded due to technical issues that occurred during the experiment. The second was excluded due to an exceptional lack of attentiveness while performing the tasks, as observed by the experimenter. Additional data cleaning procedures were applied and are described below.\u003c/p\u003e\n\u003cp\u003eIn line with the data cleaning procedures used by (Kukken et al., 2020), we planned to exclude participants who provided the same response in more than 90% of the Memory Task trials. However, no participant met this criterion; the highest proportion of identical responses was 85%. Another exclusion criterion concerned participants whose mean ratings in the CS Prerating Task deviated from the overall sample mean (M\u0026thinsp;=\u0026thinsp;0.91, SD\u0026thinsp;=\u0026thinsp;9.92) by more than three standard deviations. One participant met this criterion and was excluded from the analyses. Additionally, individual trials were excluded if the CS prerating deviated from the participant\u0026apos;s mean by more than three standard deviations. Among 3080 trials, there was only 1 such trial.\u003c/p\u003e\n\u003cp\u003eIn addition to the procedures used by (Kukken et al., 2020), further data cleaning steps were applied to tasks not included in the original study. For example, a US prerating task was administered to ensure that the subjective valence of the unconditioned stimuli (US) aligned with the intended categories. Participants who rated positive US as negative or negative US as positive were to be excluded. One participant met this criterion and was therefore excluded from all analyses.\u003c/p\u003e\n\u003cp\u003eThe Stroop task was administered in two separate sessions. Participants were to be excluded if their error rate exceeded 50% in the first session. No participants met this exclusion criterion; the highest observed error rate was 16%. Subsequently, within-subject trial-level exclusions were applied based on both Stroop sessions combined. A trial was considered invalid if it met at least one of the following criteria: (a) reaction time (RT) shorter than 200 ms, (b) RT exceeded 3,000 ms, or (c) RT exceeded three standard deviations from the participant\u0026rsquo;s mean RT. Based on these criteria, 988 trials were excluded from the analyses. The participant with the largest proportion of excluded trials had 8.5% of their trials removed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetrics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe analyses were conducted on data from 76 participants, out of which 54 were female and 22 were male. The mean age was 24.76 years (SD\u0026thinsp;=\u0026thinsp;4.53). Participants belonged to one of two HD-tDCS stimulation groups: 39 were in the active stimulation group and 37 in the sham group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCS Prerating\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDoes the mean CS prerating differ from the neutral midpoint?\u003c/strong\u003e To test whether the mean rating of the 40 chosen as most neutral CS images in the Prerating task significantly differed from zero (neutral midpoint), a linear mixed-effects model (LMM) with a random intercept for participant was conducted to account for the dependency of observations within individuals. The model was estimated using restricted maximum likelihood (REML), and \u003cem\u003et\u003c/em\u003e-tests with Satterthwaite\u0026rsquo;s approximation for degrees of freedom were applied. The results did not reveal a significant difference from zero, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.31, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.23, \u003cem\u003et\u003c/em\u003e(74.13)\u0026thinsp;=\u0026thinsp;1.34, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.18.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDoes the mean CS prerating differ based on the CS type or tDCS stimulation group?\u003c/strong\u003e To test whether the mean image ratings in the prerating phase differed by tDCS group (active vs. sham), CS type, or their interaction, a linear mixed-effects model (LMM) with a random intercept for participant was used. Group, CS type, and their interaction were included as fixed effects, with participant as a random intercept. The model revealed no significant main effect of group (\u003cem\u003eb\u003c/em\u003e = \u0026minus;\u0026thinsp;1.32, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.71, \u003cem\u003et\u003c/em\u003e(107.39) = \u0026minus;\u0026thinsp;1.86, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.07), nor of CS type compared to the reference category (StartingPositive; all \u003cem\u003ep\u003c/em\u003es\u0026thinsp;\u0026gt;\u0026thinsp;.10). Interaction effects between stimulation group and CS type were also non-significant (all \u003cem\u003ep\u003c/em\u003es\u0026thinsp;\u0026gt;\u0026thinsp;.05). Full model results are provided in the Supplementary Materials.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUS Pre- and Postrating\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIs there a difference between the mean US ratings before and after conditioning?\u003c/strong\u003e A paired-samples t-test revealed no significant difference in the ratings of the positive US between the pre- and postrating phases, t(75) = \u0026minus;\u0026thinsp;0.22, p\u0026thinsp;=\u0026thinsp;.83, 95% CI [\u0026ndash;9.29, 7.45]. A similar analysis showed a significant difference for the negative US, t(75)\u0026thinsp;=\u0026thinsp;2.74, p\u0026thinsp;=\u0026thinsp;.008, 95% CI [1.56, 9.91]. The mean ratings for the positive and negative US in the pre- and postrating phases are presented in Fig.\u0026nbsp;3.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDid tDCS stimulation mode influence US postrating scores?\u003c/strong\u003e The mean postrating of the positive US in the active group was 49.85, compared to 52.05 in the sham group. An independent-samples \u003cem\u003et\u003c/em\u003e-test revealed no significant difference between the active and sham tDCS groups, \u003cem\u003et\u003c/em\u003e(64.30) = \u0026minus;\u0026thinsp;0.26, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.80, 95% CI [\u0026ndash;19.33, 14.92]. A similar analysis was conducted for the negative US. The mean postrating of the negative US was \u0026minus;\u0026thinsp;85.72 in the active group and \u0026minus;\u0026thinsp;83.30 in the sham group. Again, there was no significant difference between the groups, \u003cem\u003et\u003c/em\u003e(65.27) = \u0026minus;\u0026thinsp;0.51, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.61, 95% CI [\u0026ndash;11.95, 7.10].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStroop Task\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBefore examining the primary hypothesis, we first tested whether our cathodal tDCS manipulation successfully modulated higher-order cognitive processing as intended. For the purposes of the analysis, data from both Stroop Task sessions were combined into a single dataset. A Type III analysis of variance (ANOVA) based on a linear mixed-effects model (LMM) with a random intercept for participants was conducted to examine differences in reaction time (RT) as a function of Stroop Task trial type (congruent, incongruent, control) and tDCS stimulation mode (active vs. sham). The analysis revealed a significant main effect of trial type, \u003cem\u003eF\u003c/em\u003e(2, 37,055.92)\u0026thinsp;=\u0026thinsp;4.06, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, with a small effect size, \u0026eta;\u0026sup2;ₚ = .03, 90% CI [0.02, 1.00]. The main effect of stimulation group also reached significance, \u003cem\u003eF\u003c/em\u003e(1, 73.48)\u0026thinsp;=\u0026thinsp;4.06, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.048, with an effect size of \u0026eta;\u0026sup2;ₚ = .05, 90% CI [0.00, 1.00], indicating a small-to-medium effect. No significant interaction between trial type and stimulation group was found, \u003cem\u003eF\u003c/em\u003e(2, 37,055.92)\u0026thinsp;=\u0026thinsp;0.13, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.874.\u003c/p\u003e\n\u003cp\u003ePost hoc comparisons were conducted to further explore differences in RTs across trial types and stimulation groups. Contrast analysis with Tukey\u0026rsquo;s correction revealed significant differences between congruent and incongruent trials, estimate = \u0026minus;\u0026thinsp;125.5 ms, z = \u0026minus;\u0026thinsp;27.22, p\u0026thinsp;\u0026lt;\u0026thinsp;.0001, with a small effect size, \u003cem\u003ed\u0026thinsp;=\u003c/em\u003e\u0026thinsp;0.17, 95% CI [0.00, 0.34], as well as between control and incongruent trials, estimate = \u0026minus;\u0026thinsp;130.29 ms, z = \u0026minus;\u0026thinsp;28.20, p\u0026thinsp;\u0026lt;\u0026thinsp;.0001, Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.17, 95% CI [0.01, 0.34]. No significant difference was found between congruent and control trials, estimate\u0026thinsp;=\u0026thinsp;4.79 ms, z\u0026thinsp;=\u0026thinsp;1.05, p\u0026thinsp;=\u0026thinsp;.545, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.16, 95% CI [\u0026ndash;0.01, 0.33].\u003c/p\u003e\n\u003cp\u003eStimulation mode significantly affected RTs for congruent trials (estimate\u0026thinsp;=\u0026thinsp;61.13 ms, z\u0026thinsp;=\u0026thinsp;2.01, p\u0026thinsp;=\u0026thinsp;.045) with a small effect size, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.17, 95% CI [0.00, 0.34], and incongruent trials (estimate\u0026thinsp;=\u0026thinsp;62.4 ms, z\u0026thinsp;=\u0026thinsp;2.05, p\u0026thinsp;=\u0026thinsp;.041), Cohen\u0026rsquo;s d effect size was also small, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.17, 95% CI [0.01, 0.34]. For control trials, the difference approached significance but did not reach it (estimate\u0026thinsp;=\u0026thinsp;57.78 ms, z\u0026thinsp;=\u0026thinsp;1.90, p\u0026thinsp;=\u0026thinsp;.058), Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.16, 95% CI [\u0026ndash;0.01, 0.33]. As illustrated in Fig.\u0026nbsp;4, participants in the active tDCS group were significantly slower than the sham group. This result confirms that our neuromodulation protocol was effective in disrupting performance on a classic executive function task.\u003c/p\u003e\n\u003cp\u003eAlthough we initially planned to include an analysis of error rates, a ceiling effect was observed in this measure, as participants made very few mistakes. As a result, error rates could not be considered a reliable indicator of between-subject differences.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDichotomous Evaluation Task\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMean Evaluative Conditioning scores (EC effect) were calculated as the average of responses in the Dichotomous Evaluation Task, where positive responses were coded as 1 and negative responses as \u0026minus;\u0026thinsp;1. This allowed us to determine whether each type of conditioned stimulus was, on average, perceived more positively or negatively, and to compare with the results in Kukken et al. (2020).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIs there an EC effect for each CS type?\u003c/strong\u003e To determine whether an EC effect was present for each CS type, a series of one-sample \u003cem\u003et\u003c/em\u003e-tests was conducted to assess whether the mean EC scores differed from the neutral midpoint. The results revealed significant differences from the midpoint for all CS types (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001 in all cases). Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e values ranged from \u0026minus;\u0026thinsp;0.51 for Ending Negative to 0.76 for Starting Positive, indicating medium-sized effects. Mean EC effect scores are presented in Fig.\u0026nbsp;5, and detailed results are provided in Table\u0026nbsp;1.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 1\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eResults of one-sample t-tests assessing whether the mean EC effect for each CS type differed from the neutral midpoint.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCS type\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eEC effect\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003edf\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCohen\u0026apos;s \u003cem\u003ed\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCohen\u0026apos;s \u003cem\u003ed\u003c/em\u003e 95% CI\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEnding Negative\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.75, \u0026minus;\u0026thinsp;.27]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEnding Positive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[.37, .86]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eStarting Negative\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.94, \u0026minus;\u0026thinsp;.44]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eStarting Positive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[.5, 1.01]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eDoes the EC effect size differ based on different CS types or tDCS stimulation groups?\u003c/strong\u003e To examine the effects of tDCS stimulation type (active vs. sham), CS type, and their interaction on the EC effect, a Type III analysis of variance (ANOVA) based on a linear mixed-effects model (LMM) with a random intercept for participant was conducted. The analysis revealed a significant main effect of CS type, \u003cem\u003eF\u003c/em\u003e(3, 2956.8)\u0026thinsp;=\u0026thinsp;87.10, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.0001, indicating that the EC effect varied across CS types. Partial eta-squared effect size was small, \u0026eta;\u0026sup2;ₚ = 0.08, 95% CI [0.07, 1.00]. There was no significant main effect of tDCS stimulation group, \u003cem\u003eF\u003c/em\u003e(1, 73.8)\u0026thinsp;=\u0026thinsp;0.10, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.750, and no significant interaction between group and CS type, \u003cem\u003eF\u003c/em\u003e(3, 2956.8)\u0026thinsp;=\u0026thinsp;0.001, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.000. These results suggest that tDCS stimulation did not influence CS ratings in the Evaluation Task and did not interact with CS type in modulating the EC effect. The main effect of CS type was further examined using post hoc contrast analysis with Tukey\u0026rsquo;s correction. The results showed no significant difference between the Starting vs Ending Negative conditions, \u003cem\u003eZ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.37, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.52, and no difference between the Starting vs Ending Positive conditions, \u003cem\u003eZ\u003c/em\u003e = \u0026minus;\u0026thinsp;0.93, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.79. All other pairwise comparisons were statistically significant (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001) with medium effect size (Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e ranging from \u0026minus;\u0026thinsp;0.52 to \u0026minus;\u0026thinsp;0.64), and are described in detail in Table\u0026nbsp;2.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 2\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003ePost-hoc contrasts between the mean EC effect scores by CS type.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003econtrast\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eestimate diff.\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eZ\u003c/em\u003e ratio\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e 95% CI\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEN \u0026ndash; EP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-10.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.63, \u0026minus;\u0026thinsp;.42]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEN \u0026ndash; SN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;=\u0026thinsp;.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.03, .17]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEN \u0026ndash; SP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-11.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.67, \u0026minus;\u0026thinsp;.47]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEP \u0026ndash; SN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[.49, .70]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEP \u0026ndash; SP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-0.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;=\u0026thinsp;.79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.15, .05]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSN \u0026ndash; SP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e-12.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026minus;\u0026thinsp;.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e[-.74, \u0026minus;\u0026thinsp;.54]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eMemory Task\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHaving established that cathodal tDCS successfully modulated executive control, we next examined our primary question: does this disruption of the left dlPFC function selectively impair the relational component of evaluative conditioning?\u003c/p\u003e\n\u003cp\u003ePercentages of correct responses in the Memory Task were calculated for each CS type and are later referred to as the HIT Ratio. Following Kukken et al. (2020), correct implications were defined as follows: CS types SP and EN were associated with a positive meaning, whereas SN and EP \u0026ndash; with a negative meaning. To test the accuracy of responses in the Memory Task, a series of one-sample \u003cem\u003et\u003c/em\u003e-tests was conducted, comparing the mean HIT Ratio for each CS type to the chance level of 50%. Mean HIT scores for each CS type are shown in Fig. 6. The results showed that for all CS types, the difference from the chance level was statistically significant, with a large effect for CS starting US and medium for those ending US. For CS that started a positive US, the HIT Ratio differed significantly from chance, \u003cem\u003et\u003c/em\u003e(75)\u0026thinsp;=\u0026thinsp;9.21, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [67.82, 77.65], with a large effect of Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.13, 95% CS [1.56, 2.69]. A similar result was found for CS that started a negative US, \u003cem\u003et\u003c/em\u003e(75)\u0026thinsp;=\u0026thinsp;12.21, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [72.35, 81.07], which as well showed a large effect, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.82, 95% CI [2.18, 3.45]. CS that ended a positive US also differed significantly from chance, \u003cem\u003et\u003c/em\u003e(75) = \u0026minus;\u0026thinsp;2.93, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.005, 95% CI [34.96, 47.15], with a medium effect of \u003cem\u003ed\u003c/em\u003e = \u0026minus;\u0026thinsp;0.64, 95% CI [-1.10, -0.17], as did CS that ended a negative US, \u003cem\u003et\u003c/em\u003e(75) = \u0026minus;\u0026thinsp;2.77, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.007, 95% CI [32.35, 47.12], also with a medium effect, \u003cem\u003ed\u003c/em\u003e = \u0026minus;\u0026thinsp;0.68, 95% CI [-1.14, -0.21].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMPT model for Memory Task\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData from Memory Task were analyzed using multinomial processing tree (MPT) modelling with the TreeBUGS package in R. Following (Kukken et al., 2020), response frequencies were modeled using latent-trait approach. The model was applied to individual response frequencies for each CS type. To estimate the potential effects of tDCS stimulation (active vs. sham) on the task performance, we added fixed group effects to the model on all parameters. The processing tree model used to represent participants\u0026rsquo; responses was adapted from the original study. In the model, each node reflects a discrete cognitive state, and the associated parameters represent the probability of a response being guided by that state. The model includes three parameters: the probability of correctly recalling the meaning of the CS (\u003cem\u003em\u003c/em\u003e parameter), the likelihood that the response was based on the CS\u0026ndash;US association (\u003cem\u003ep\u003c/em\u003e parameter), and the probability of guessing correctly (\u003cem\u003eg\u003c/em\u003e parameter). The same tree structure was applied for modeling Memory and Evaluation Tasks\u0026rsquo; responses. A detailed definition of the tree structure as well as the detailed results of the Evaluation Task MPT model are provided in the Supplementary Materials. The MPT model was estimated using four Markov chains with 50,000 iterations each. The adaptation (burn-in) phase was set to 50,000 iterations, and a thinning rate of 1 was applied, meaning that every posterior sample after burn-in was retained, thus maximizing the number of usable samples.\u003c/p\u003e\n\u003cp\u003eFollowing Kukken et al. (2020), we tested for differences in response frequencies between participants. Pearson\u0026rsquo;s chi-square test revealed significant variability across participants, \u0026chi;\u0026sup2;(525)\u0026thinsp;=\u0026thinsp;875.68, p\u0026thinsp;\u0026lt;\u0026thinsp;.001, suggesting meaningful individual differences in underlying parameter estimates.\u003c/p\u003e\n\u003cp\u003ePosterior predictive checks indicated good model fit. The posterior predictive p-value (PPP) for the mean response structure (T1) was .466 (observed\u0026thinsp;=\u0026thinsp;0.04; predicted\u0026thinsp;=\u0026thinsp;0.04), while the PPP for the covariance structure (T2) was .343 (observed\u0026thinsp;=\u0026thinsp;3.09; predicted\u0026thinsp;=\u0026thinsp;2.44). Both values fall well within the acceptable range of [.05, .95], suggesting that the model accurately reproduced both response proportions and their covariation. All Markov chain Monte Carlo (MCMC) chains showed good convergence, with Gelman\u0026ndash;Rubin statistics (R̂) below 1.04 for all parameters.\u003c/p\u003e\n\u003cp\u003eThe estimated parameters are provided on a probability scale. The m-parameter indicated that participants correctly remembered the meaning of the CS in 3.5% of trials. The \u003cem\u003ep\u003c/em\u003e-parameter showed that in 29.2% of cases (when the meaning was not remembered) responses followed the paired US valence. The \u003cem\u003eg\u003c/em\u003e-parameter reflects that when guessing occurred, participants showed no particular bias towards any answer. Estimated parameters as well as their correlations are presented in Table\u0026nbsp;3.\u003c/p\u003e\n\u003cp\u003eNo clear group-level effects of tDCS stimulation (active vs. sham) were observed on any of the parameters, as all 95% credible intervals for group effects overlapped with zero. For example, the group effect on the \u003cem\u003em\u003c/em\u003e-parameter (meaning) in a latent probit scale was \u0026Delta;\u0026thinsp;=\u0026thinsp;0.127, 95% CI [\u0026ndash;0.33, 0.62]. All group effects are presented in Table 3.\u003c/p\u003e\n\u003ctable width=\"603\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\" width=\"603\"\u003e\n\u003cp\u003eTable 3. Response frequencies, parameter estimates, estimated parameter correlations, and effects of group assignment in the MPT models for the Memory and Evaluation Tasks.\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e\u003cstrong\u003eMemory Task\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e\u003cstrong\u003eEvaluation Task\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\" width=\"603\"\u003e\n\u003cp\u003e\u003cem\u003eResponse frequencies\u003c/em\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eSP\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e552 (73%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e499 (66%)\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eEP\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e312 (41%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e277 (36%)\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eSN\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e583 (77%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e489 (64%)\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eEN\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e302 (40%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e296 (39%)\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\" width=\"603\"\u003e\n\u003cp\u003e\u003cem\u003eParameter estimates\u003c/em\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003em\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e.035 [.01, .09]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e.014 [.00, .04]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003ep\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e.292 [.15, .43]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e.209 [.12, .30]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eg\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e.465 [.41, .52]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e.519 [.47, .57]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\" width=\"603\"\u003e\n\u003cp\u003e\u003cem\u003eCorrelations\u003c/em\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003em \u0026ndash; p\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e-.507 [\u0026ndash;0.81, \u0026ndash;0.13]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e.083 [-.58, .71]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003em \u0026ndash; g\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e.588 [0.16, 0.87]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e-.063 [-.73, .66]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003ep \u0026ndash; g\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"276\"\u003e\n\u003cp\u003e-.462 [\u0026ndash;0.79, \u0026ndash;0.03]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd colspan=\"2\" width=\"249\"\u003e\n\u003cp\u003e.012 [-.38, .39]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"121\"\u003e\n\u003cp\u003e\u003cstrong\u003eactive\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"155\"\u003e\n\u003cp\u003e\u003cstrong\u003esham\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e\u003cstrong\u003eactive\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"127\"\u003e\n\u003cp\u003e\u003cstrong\u003esham\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\" width=\"603\"\u003e\n\u003cp\u003e\u003cem\u003eEffects of factors (shift from overall mean)\u003c/em\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003em\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"121\"\u003e\n\u003cp\u003e.127 [-.33, .62]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"155\"\u003e\n\u003cp\u003e-.127 [-.62, .33]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e.120 [-.55, .86]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"127\"\u003e\n\u003cp\u003e-.120 [-.86, .55]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003ep\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"121\"\u003e\n\u003cp\u003e-.019 [-.37, .33]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"155\"\u003e\n\u003cp\u003e.019 [-.33, .37]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e.014 [-.25, .29]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"127\"\u003e\n\u003cp\u003e-.014 [-.29, .25]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"78\"\u003e\n\u003cp\u003eg\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"121\"\u003e\n\u003cp\u003e.014 [-.11, .13]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"155\"\u003e\n\u003cp\u003e-.014 [-.13, .11]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e-.017 [-.13, .10]\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"127\"\u003e\n\u003cp\u003e.017 [-.10, .13]\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eTable notes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eResponse frequencies for correct answers were aggregated across all participants. Correct responses were defined as those consistent with the implied meaning of the CS \u0026ndash; that is, a positive response for SP and EN, and a negative response for SN and EP. Parameter estimates are presented on the probability scale and reflect the average likelihood of relying on each latent process across participants, irrespective of group assignment. Correlations between parameters, as well as effects of group factors, are estimated on the latent probit scale. Group assignment was coded as either sham or active tDCS stimulation.\u003c/em\u003e\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe present study investigated the causal role of the left dorsolateral prefrontal cortex (dlPFC) in relational evaluative conditioning (EC). We successfully replicated the basic finding that relational qualifiers modulate attitude formation (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), but our primary hypothesis was not supported. We found a critical dissociation: cathodal HD-tDCS successfully impaired performance on an executive control task (the Stroop task) but had no effect on attitude formation. This null effect of neuromodulation was consistent across both explicit evaluations and the cognitive processes of relational learning and co-occurrence tracking, as estimated by Multinomial Processing Tree (MPT) modeling. These findings, emerging from a highly-powered design, provide robust evidence that challenges the presumed role of the left dlPFC in this learning context.\u003c/p\u003e\u003cp\u003eThe absence of a tDCS effect on relational EC is a theoretically informative finding. We predicted that inhibitory stimulation of the left dlPFC would impair propositional processing, but this was not observed in either the overall ratings or the MPT model's relational parameter (m). Given our high statistical power and the successful manipulation check, this null result challenges the hypothesis that the executive functions subserved by the left dlPFC are indispensable for propositional operations in this EC paradigm (cf. Mengarelli et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Ochsner et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2004\u003c/span\u003e). Our data suggest that the link between domain-general executive control and relational learning is less direct than previously assumed, a conclusion that aligns with recent complex findings from cognitive load studies (Gawronski, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Our results may be interpreted as supporting the view that EC effects are not dependent on higher-order cognitive processing during the encoding phase, specifically in the left dlPFC. Although cathodal stimulation of this region has been shown to diminish preference change in a study involving higher-order processing (Mengarelli et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), it is possible that the processes crucial for the emergence of EC effects occur in a different brain region.\u003c/p\u003e\u003cp\u003eThe successful modulation of Stroop task performance is a key methodological strength of this study. Confirming that our tDCS protocol effectively disrupted reaction times on a classic executive function task provides evidence that the neuromodulation was indeed successful. We focused on reaction times because accuracy was near-ceiling, precluding a reliable analysis of error rates.\u003c/p\u003e\u003cp\u003eThe MPT model of our Memory Task data, when compared to Kukken et al. (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), revealed a substantial procedural impact. Our relational knowledge parameter (m\u0026thinsp;=\u0026thinsp;.035) was drastically lower than in the original studies (m\u0026thinsp;\u0026asymp;\u0026thinsp;.21\u0026ndash;.57), indicating that participants relied on explicit relational memory in only 3.5% of trials. This is best explained by the 23-minute delay in our procedure, which likely caused significant forgetting. This floor effect on relational knowledge provides context for our co-occurrence parameter (p\u0026thinsp;=\u0026thinsp;.292), which was slightly higher than in the original work. This pattern is consistent with the model's logic: with diminished relational knowledge, participants defaulted more often to the simpler co-occurrence heuristic (Gawronski \u0026amp; Brannon, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Critically, this deeper analysis confirmed the primary null finding: tDCS did not credibly affect any MPT parameter.\u003c/p\u003e\u003cp\u003e Behavioral data from the Memory Task further highlighted participants' cognitive challenges. While memory was accurate for CSs that started sounds, performance was systematically below chance for CSs that stopped sounds. This suggests participants were not guessing, but rather actively misremembering the implications. We interpret this as a cognitive conflict where the salient valence of the US overrode the more complex propositional meaning of the \"stop\" relation, leading to consistent errors. This pattern reveals the difficulty of overriding simple co-occurrence signals, even when performing an explicit memory task.\u003c/p\u003e\u003cp\u003eDespite the study's strengths, including high statistical power and a successful manipulation check, several limitations must be acknowledged, each pointing toward important avenues for future research. The primary limitation stems from the procedural delay between the conditioning and testing phases. As revealed by the MPT analysis, this 23-minute interval, necessary for tDCS after-effects to subside, likely led to substantial forgetting of the specific CS-US relations, resulting in a floor effect on the relational knowledge parameter (m). This presents a significant challenge for interpreting our primary null finding. Cathodal tDCS was intended to impair a cognitive process, but it is plausible that there was very little active relational processing left to suppress by the time of the memory and evaluation tasks. In essence, one cannot detect a neuromodulatory impairment of a process that is already at its floor.\u003c/p\u003e\u003cp\u003eSecond, limitations inherent to HD-tDCS technology warrant consideration. Although we used a high-definition montage to improve spatial focality, the resolution is still diffuse compared to techniques like transcranial magnetic stimulation (TMS) (Chrysikou \u0026amp; Hamilton, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Furthermore, while our protocol was effective in modulating Stroop performance, the optimal \"dose\" in terms of intensity and duration required to influence the specific neural computations of relational EC remains unknown. It is possible that the cognitive load imposed by the Stroop task is more susceptible to this form of broad neuromodulation than the processes involved in applying a learned relational rule.\u003c/p\u003e\u003cp\u003eFinally, our design tested the causal role of the dlPFC exclusively during the encoding phase of learning. It is equally plausible that the critical contribution of the dlPFC is not in the initial formation of propositional links, but during their active retrieval and application at the time of judgment. The process of searching memory for the correct relational tag and using it to override a default assimilative response could be more executively demanding than the initial learning itself. Our study was not designed to test this retrieval hypothesis and therefore cannot speak to the dlPFC's role during the evaluation phase.\u003c/p\u003e\u003cp\u003eThe limitations of the present study directly inform several promising avenues for future research. First, to address the encoding versus retrieval question, a future study could apply neuromodulation not during the learning phase, but during the subsequent evaluation task. This would provide a direct test of whether the dlPFC is causally involved in the active retrieval and application of propositional knowledge, a process that may be more executively demanding than initial encoding (cf. Wąsowicz et al. under review). Second, to overcome the inherent limitations in the spatial resolution of tDCS, future work could employ more focal techniques like transcranial magnetic stimulation (TMS) to provide a more conclusive test of the left dlPFC's specific role. Third, to increase the cognitive demand and prevent the floor effects on relational knowledge that we observed, future studies should consider using less complex or demanding EC paradigms. For example, a task involving somewhat simpler relational rules would place less load on executive resources and allow for detecting the effects of neuromodulation. Future work should also investigate whether participants indeed attribute causal role to CSs as starting/stopping affective stimuli. For now it is only assumed by researchers that temporal configuration of CS-US pairs invokes relational interpretations.\u003c/p\u003e\u003cp\u003eIn conclusion, this study provides highly-powered evidence for a dissociation between the neural substrates of general executive control and those specifically recruited for relational evaluative conditioning. Our results demonstrate that while inhibitory stimulation of the left dlPFC successfully impairs performance on a classic executive control task, it does not alter the balance of co-occurrence-based and relation-based processes in attitude formation. This challenges the straightforward view that the left dlPFC is an indispensable substrate for integrating relational information in this context. Ultimately, our findings suggest that while the dlPFC is undoubtedly a key hub for executive function, its causal role in this fundamental form of attitude formation may be less direct or more task-dependent than previously assumed, paving the way for more nuanced neurocognitive models of how we learn our likes and dislikes\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding.\u003c/h2\u003e\u003cp\u003eThis work was supported by the National Science Center grant no. 2019/33/B/HS6/02700.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAll authors contributed substantially to the study. J.W. and R.B. designed the experimental paradigm. J.W. programmed the experimental script, managed the necessary equipment, and coordinated participant recruitment. J.W. and K.G. conducted the experimental sessions and performed data analysis. J.W. and R.B. wrote the main manuscript, and J.W. prepared the figures. All authors reviewed and approved the final version of the manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAllaert, J., De Raedt, R., Sanchez-Lopez, A., Baeken, C., \u0026amp; Vanderhasselt, M. A. (2022). Mind the social feedback: Effects of tDCS applied to the left DLPFC on psychophysiological responses during the anticipation and reception of social evaluations. \u003cem\u003eSocial Cognitive and Affective Neuroscience\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(1), 131\u0026ndash;141. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/scan/nsaa066\u003c/span\u003e\u003cspan address=\"10.1093/scan/nsaa066\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChrysikou, E. G., \u0026amp; Hamilton, R. H. (2011). Noninvasive brain stimulation in the treatment of aphasia: Exploring interhemispheric relationships and their implications for neurorehabilitation. \u003cem\u003eRestorative Neurology and Neuroscience\u003c/em\u003e, \u003cem\u003e29\u003c/em\u003e(6), 375\u0026ndash;394. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3233/RNN-2011-0610\u003c/span\u003e\u003cspan address=\"10.3233/RNN-2011-0610\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eClarke, P. J. F., Haridas, S. M. P., Van Bockstaele, B., Chen, N. T. M., Salemink, E., \u0026amp; Notebaert, L. (2021). Frontal tDCS and Emotional Reactivity to Negative Content: Examining the Roles of Biased Interpretation and Emotion Regulation. \u003cem\u003eCognitive Therapy and Research\u003c/em\u003e, \u003cem\u003e45\u003c/em\u003e(1), 19\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10608-020-10162-9\u003c/span\u003e\u003cspan address=\"10.1007/s10608-020-10162-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDe Houwer, J., Thomas, S., \u0026amp; Baeyens, F. (2001). Associative Learning of Likes and Dislikes: A Review of 25 Years of Research on Human Evaluative Conditioning. \u003cem\u003ePsychological Bulletin\u003c/em\u003e, \u003cem\u003e127\u003c/em\u003e(6), 853\u0026ndash;869. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1037/0033-2909.127.6.853\u003c/span\u003e\u003cspan address=\"10.1037/0033-2909.127.6.853\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFaul, F., Erdfelder, E., Lang, A. G., \u0026amp; Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. \u003cem\u003eBehavior Research Methods\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e(2), 175\u0026ndash;191. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3758/BF03193146\u003c/span\u003e\u003cspan address=\"10.3758/BF03193146\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eF\u0026ouml;rderer, S., \u0026amp; Unkelbach, C. (2012). Hating the cute kitten or loving the aggressive pit-bull: EC effects depend on CS-US relations. \u003cem\u003eCognition and Emotion\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(3), 534\u0026ndash;540. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02699931.2011.588687\u003c/span\u003e\u003cspan address=\"10.1080/02699931.2011.588687\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFriehs, M. A., \u0026amp; Frings, C. (2019). Cathodal tDCS increases stop-signal reaction time. \u003cem\u003eCognitive Affective \u0026amp; Behavioral Neuroscience\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(5), 1129\u0026ndash;1142. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3758/s13415-019-00740-0\u003c/span\u003e\u003cspan address=\"10.3758/s13415-019-00740-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B. (2022). Attitudinal Effects of Stimulus Co-occurrence and Stimulus Relations: Paradoxical Effects of Cognitive Load. \u003cem\u003ePersonality \u0026amp; Social Psychology Bulletin\u003c/em\u003e, \u003cem\u003e48\u003c/em\u003e(10), 1438\u0026ndash;1450. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/01461672211044322\u003c/span\u003e\u003cspan address=\"10.1177/01461672211044322\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., Balas, R., \u0026amp; Hu, X. (2016). What do I know about it versus how do I feel about it? A theoretical analysis of response processes in evaluative conditioning. \u003cem\u003eSocial Cognition\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(5). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1521/soco.2016.34.5.357\u003c/span\u003e\u003cspan address=\"10.1521/soco.2016.34.5.357\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., \u0026amp; Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. \u003cem\u003ePsychological Bulletin\u003c/em\u003e, \u003cem\u003e132\u003c/em\u003e(5), 692\u0026ndash;731. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1037/0033-2909.132.5.692\u003c/span\u003e\u003cspan address=\"10.1037/0033-2909.132.5.692\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., \u0026amp; Bodenhausen, G. V. (2011a). The Associative\u0026ndash;Propositional Evaluation Model. In \u003cem\u003eAdvances in Experimental Social Psychology\u003c/em\u003e (Vol. 44, pp. 59\u0026ndash;127). Elsevier. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://linkinghub.elsevier.com/retrieve/pii/B9780123855220000020\u003c/span\u003e\u003cspan address=\"https://linkinghub.elsevier.com/retrieve/pii/B9780123855220000020\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., \u0026amp; Bodenhausen, G. V. (2011b). The associative-propositional evaluation model. Theory, evidence, and open questions. In \u003cem\u003eAdvances in Experimental Social Psychology\u003c/em\u003e (Vol. 44). Elsevier Inc. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/B978-0-12-385522-0.00002-0\u003c/span\u003e\u003cspan address=\"10.1016/B978-0-12-385522-0.00002-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., \u0026amp; Bodenhausen, G. V. (2018). Evaluative Conditioning From the Perspective of the Associative-Propositional Evaluation Model. \u003cem\u003eSocial Psychological Bulletin\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(3), e28024. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5964/spb.v13i3.28024\u003c/span\u003e\u003cspan address=\"10.5964/spb.v13i3.28024\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., Bodenhausen, G. V., \u0026amp; Becker, A. P. (2007). I like it, because I like myself: Associative self-anchoring and post-decisional change of implicit evaluations. \u003cem\u003eJournal of Experimental Social Psychology\u003c/em\u003e, \u003cem\u003e43\u003c/em\u003e(2), 221\u0026ndash;232. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jesp.2006.04.001\u003c/span\u003e\u003cspan address=\"10.1016/j.jesp.2006.04.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., \u0026amp; Brannon, S. M. (2021). Attitudinal Effects of Stimulus Co-Occurrence and Stimulus Relations: Range and Limits of Intentional Control. \u003cem\u003ePersonality and Social Psychology Bulletin\u003c/em\u003e, \u003cem\u003e47\u003c/em\u003e(12), 1654\u0026ndash;1667. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0146167220982906\u003c/span\u003e\u003cspan address=\"10.1177/0146167220982906\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGawronski, B., Brannon, S. M., \u0026amp; Bodenhausen, G. V. (2016). The associative-propositional duality in the representation, formation, and expression of attitudes. \u003cem\u003eReflective and Impulsive Determinants of Human Behavior\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e, 103\u0026ndash;118. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9781315523095\u003c/span\u003e\u003cspan address=\"10.4324/9781315523095\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHeycke, T., \u0026amp; Gawronski, B. (2020). Co-occurrence and relational information in evaluative learning: A multinomial modeling approach. \u003cem\u003eJournal of Experimental Psychology: General\u003c/em\u003e, \u003cem\u003e149\u003c/em\u003e(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1037/xge0000620\u003c/span\u003e\u003cspan address=\"10.1037/xge0000620\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHofmann, W., De Houwer, J., Perugini, M., Baeyens, F., \u0026amp; Crombez, G. (2010). Evaluative Conditioning in Humans: A Meta-Analysis. \u003cem\u003ePsychological Bulletin\u003c/em\u003e, \u003cem\u003e136\u003c/em\u003e(3), 390\u0026ndash;421. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1037/a0018916\u003c/span\u003e\u003cspan address=\"10.1037/a0018916\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu, X., Gawronski, B., \u0026amp; Balas, R. (2017). Propositional Versus Dual-Process Accounts of Evaluative Conditioning: I. The Effects of Co-Occurrence and Relational Information on Implicit and Explicit Evaluations. \u003cem\u003ePersonality and Social Psychology Bulletin\u003c/em\u003e, \u003cem\u003e43\u003c/em\u003e(1), 17\u0026ndash;32. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0146167216673351\u003c/span\u003e\u003cspan address=\"10.1177/0146167216673351\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHughes, S., Ye, Y., \u0026amp; De Houwer, J. (2019). Evaluative conditioning effects are modulated by the nature of contextual pairings. \u003cem\u003eCognition and Emotion\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(5). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02699931.2018.1500882\u003c/span\u003e\u003cspan address=\"10.1080/02699931.2018.1500882\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eH\u0026uuml;tter, M. (2022). An integrative review of dual- and single-process accounts of evaluative conditioning. \u003cem\u003eNature Reviews Psychology\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s44159-022-00102-7\u003c/span\u003e\u003cspan address=\"10.1038/s44159-022-00102-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKukken, N., H\u0026uuml;tter, M., \u0026amp; Holland, R. W. (2020). Are there two independent evaluative conditioning effects in relational paradigms? Dissociating the effects of CS-US pairings and their meaning. \u003cem\u003eCognition and Emotion\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(1), 170\u0026ndash;187. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02699931.2019.1617112\u003c/span\u003e\u003cspan address=\"10.1080/02699931.2019.1617112\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLevey, A. B., \u0026amp; Martin, I. (1975). Classical conditioning of human \u0026lsquo;evaluative\u0026rsquo; responses. \u003cem\u003eBehaviour Research and Therapy\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(4), 221\u0026ndash;226. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0005-7967(75)90026-1\u003c/span\u003e\u003cspan address=\"10.1016/0005-7967(75)90026-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMengarelli, F., Spoglianti, S., Avenanti, A., \u0026amp; di Pellegrino, G. (2015). Cathodal tDCS Over the Left Prefrontal Cortex Diminishes Choice-Induced Preference Change. \u003cem\u003eCerebral Cortex\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e(5), 1219\u0026ndash;1227. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/cercor/bht314\u003c/span\u003e\u003cspan address=\"10.1093/cercor/bht314\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoran, T., \u0026amp; Bar-Anan, Y. (2013). The effect of object-valence relations on automatic evaluation. \u003cem\u003eCognition and Emotion\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(4), 743\u0026ndash;752. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02699931.2012.732040\u003c/span\u003e\u003cspan address=\"10.1080/02699931.2012.732040\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoran, T., \u0026amp; Bar-Anan, Y. (2020). The effect of co-occurrence and relational information on speeded evaluation. \u003cem\u003eCognition and Emotion\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(1), 144\u0026ndash;155. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02699931.2019.1604321\u003c/span\u003e\u003cspan address=\"10.1080/02699931.2019.1604321\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoran, T., Bar-Anan, Y., \u0026amp; Nosek, B. A. (2016). The assimilative effect of co-occurrence on evaluation above and beyond the effect of relational qualifiers. \u003cem\u003eSocial Cognition\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(5). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1521/soco.2016.34.5.435\u003c/span\u003e\u003cspan address=\"10.1521/soco.2016.34.5.435\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoran, T., Nudler, Y., \u0026amp; Bar-Anan, Y. (2023). Evaluative Conditioning: Past, Present, and Future. \u003cem\u003eAnnual Review of Psychology\u003c/em\u003e, \u003cem\u003e74\u003c/em\u003e, 245\u0026ndash;269. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev-psych-032420-031815\u003c/span\u003e\u003cspan address=\"10.1146/annurev-psych-032420-031815\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Scopus.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOchsner, K. N., Ray, R. D., Cooper, J. C., Robertson, E. R., Chopra, S., Gabrieli, J. D. E., \u0026amp; Gross, J. J. (2004). For better or for worse: Neural systems supporting the cognitive down- and up-regulation of negative emotion. \u003cem\u003eNeuroimage\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(2), 483\u0026ndash;499. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.neuroimage.2004.06.030\u003c/span\u003e\u003cspan address=\"10.1016/j.neuroimage.2004.06.030\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWąsowicz, J., Balas, R., Uram, P., \u0026amp; Okruszek, Ł. (under review). \u003cem\u003eHow do cognitive resource limitations influence evaluative conditioning? A tDCS study\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.31234/osf.io/srtk4_v1\u003c/span\u003e\u003cspan address=\"10.31234/osf.io/srtk4_v1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e The expected time between the end of the stimulation and the beginning of the Evaluation Task was approximately 23 minutes, which should have been sufficient for most of the stimulation effects to fade. If a participant took longer (or shorter) during the first technical break or required more (or less) time to complete the preceding tasks, the duration of the second break was automatically adjusted to ensure exactly 23 minutes between the end of stimulation and the Evaluation Task\u0026mdash;no more, no less. However, this adjustment was implemented during the experimental phase of the project; therefore, the first five participants received a fixed 20-minute break instead.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e The Inquisit software crashed in the middle of the experimental session, and it was not possible to resume the procedure from the point at which it stopped. A Windows update had been performed shortly before the session, which is presumed to have caused the issue. No further updates were applied, and the problem did not occur again.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"evaluative conditioning, attitude change, transcranial direct current stimulation","lastPublishedDoi":"10.21203/rs.3.rs-7252129/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7252129/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRelational evaluative conditioning (EC) paradigms suggest that attitude formation involves both simple co-occurrence-based and relation-based propositional processes. However, the causal role of executive control systems in supporting this propositional learning remains debated. This highly-powered, pre-registered study (N\u0026thinsp;=\u0026thinsp;76) aimed to causally test the role of the left dorsolateral prefrontal cortex (dlPFC), a key executive control hub, in this process. We replicated (Kukken et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) relational EC paradigm while applying inhibitory (cathodal) high-definition transcranial direct current stimulation (HD-tDCS) or sham stimulation over the left dlPFC during learning. A Stroop task served as a manipulation check, and Multinomial Processing Tree (MPT) modeling dissociated relational from co-occurrence processes. Results revealed a critical dissociation: while cathodal tDCS successfully impaired performance on the Stroop task, confirming effective neuromodulation, it had no effect on evaluative ratings. Crucially, MPT modeling confirmed that tDCS did not alter the parameters for either relational or co-occurrence processing. These findings challenge the hypothesis that the left dlPFC is indispensable for integrating relational information in EC, suggesting a more nuanced link between domain-general executive control and this fundamental form of attitude formation.\u003c/p\u003e","manuscriptTitle":"Inhibiting Left dlPFC Leaves Relational Evaluative Conditioning Unchanged: Evidence From Electrical Brain Stimulation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-21 19:14:32","doi":"10.21203/rs.3.rs-7252129/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"03ae771f-5a43-479f-8043-1217d8454323","owner":[],"postedDate":"August 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-11-06T16:08:38+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-21 19:14:32","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7252129","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7252129","identity":"rs-7252129","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}