Neural dynamics of perceived agreement and disagreement with peer and expert opinions: An MEG study

doi:10.21203/rs.3.rs-5109976/v1

Download PDF

Article

Neural dynamics of perceived agreement and disagreement with peer and expert opinions: An MEG study

https://doi.org/10.21203/rs.3.rs-5109976/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Individuals change their opinions under the influence of others' opinions; however, the extent and nature of this influence critically depend on their attitudes toward those exerting the influence. In this study, we compare two sources of influence that drive conformity behavior: an expert group, and a peer group. Furthermore, we investigate the underlying neural dynamics using magnetoencephalography to determine whether the processing of these two influences shares their neural mechanisms. Twenty-two participants performed a task in a fashion choice context and received feedback from a peer and an expert group. When participants re-evaluated the clothing after a delay, we found that participants' opinions changed in line with disagreement feedback when feedback was lower than the participant's first rating – without distinct conformity to the social sources. On the neural level, however, there was a difference between conflict with peer and expert groups, with a stronger response for peers in 170–590 ms time window in gradiometer channels. Furthermore, agreement evoked stronger neural responses than conflict, in 590–960 ms time window in magnetometer channels. Taken together, our findings suggest that conflicting feedback from peers and experts regarding clothing preferences elicits distinct temporal dynamics. However, conformity behavior is influenced solely by the feedback valence.

Biological sciences/Neuroscience/Social neuroscience

Biological sciences/Psychology/Human behaviour

Social influence

fashion context

expert opinion

peer opinion

magnetoencephalography

Social conformity—the modification of opinions and decisions to match those of others—is a fundamental aspect of human behavior and decision-making¹. This is especially significant in today's online environment, where others’ opinions are ubiquitous in the form of recommendations, ratings, and reviews about products, services, etc.². Conformity is driven by different underlying motivations, such as the desire to form precise perceptions of reality or to be “right”, and the desire to create and preserve relationships with others or to be “liked”³. These motivations often coexist, particularly when employing majority opinions, as the rule of the majority not only highlights the unpleasantness of standing alone but also underscores a rational foundation for this compromise: a collective group is statistically more likely to make an accurate decision than an individual^4–6.

However, when the source of influence and the nature of the relationship between the individual and the social source are specified, the underlying motivation for influence can be driven by one of the motivations, depending on the context and the source of influence. In a study by Klucharev and colleagues⁷, pairing an expert celebrity with a product that matched their expertise produced better product recognition and a positive attitude toward the product in the later day’s assessment, compared to when the expert celebrity was linked to a product that did not match their reputation. Similarly, more recent research demonstrated that participants who listened to a health expert's narrative about the risks of sugar consumption significantly decreased their willingness to pay for sugar-containing foods compared to the control group⁸. Research by Engelmann and colleagues⁹ further supports this perspective by showing that risk-averse advice from a financial expert on a certainty equivalent task influenced the decision-making process in adults and adolescents. In a task requiring particular knowledge (price information of rental apartments), expert advice compared to non-expert advice significantly influenced the subject’s opinions and enhanced their judgment accuracy¹⁰. In the mentioned studies, perceived credibility and reliability of the expert sources were key drivers of conformity and positive attitude, especially when compared to scenarios where expertise was irrelevant or when expert sources were compared to control and novice groups.

Individuals rely on peer group judgments to achieve accurate knowledge¹¹, however, they are frequently motivated to align with their peer group's norms and expectations to gain approval or avoid negative consequences such as social exclusion^12–15. Empirical evidence demonstrates that during adolescence, individuals are more prone to engage in risky behaviors when accompanied by peers compared to when they are alone or with adults ^16,17. This indicates that their motivation is not due to a lack of knowledge about appropriate actions, but rather a strong desire to be liked and accepted. Neuroimaging studies further corroborate this notion, revealing that conformity to peer norms shares neural activities associated with social exclusion ¹², and experiencing social exclusion predicts subsequent conformity to peer norms ¹³.

While the studies mentioned earlier have examined the effect of peer and expert separately, a study with two social sources¹⁸ demonstrated that in an investment decision task, participants followed the advice of a financial expert significantly longer than the peer advice, despite identical and randomized suggestions. Similarly, another study¹⁹ with a probabilistic learning task and a within-subject design revealed that participants favored stimuli endorsed by experienced adults over those recommended by peers. A further study² observed that when expert and peer recommendations conflicted in a profile picture selection task, participants were more influenced by expert endorsements, especially when these endorsements were positive. In these studies, social influence hinges on the reliability of the social source and quality of the advice rather than being driven by peer norms. Indeed, when the accuracy of an opinion is the main objective, people value and follow seemingly reliable advice and recommendations rather than following peer opinion ^20,21. However, in a subjective evaluative judgment task where feedback on performance was provided by peers and experts rather than advice or recommendations for decision-making, adolescent participants conformed similarly to peer and expert groups²². This suggests that the context and task demands influence the effectiveness of peer and expert opinions. Specifically, expert influence may dominate in contexts requiring accuracy and informed decision-making and peer influence may prevail when there is an explicit identity component or a need to fit in.

A large body of neuroimaging studies has identified conflict monitoring as a crucial component within the neurocognitive network that underlies social conformity when there is a discrepancy between an individual’s opinion and peer group ^4,23,24. One explanation for this conflict detection mechanism is that social conformity is based on reinforcement learning, suggesting that when an individual’s judgment contradicts the group consensus, a negative prediction error is triggered in the medial frontal cortex (MFC) and ventral striatum (VS). Conversely, when an individual’s judgment aligns with the group, reward signaling in the brain is amplified, reinforcing consistent opinions ⁴. In addition to the reinforcement learning framework of social conformity, Berns et al demonstrated in their study a significant positive correlation between specific neural activation (anterior cingulate cortex (ACC) and anterior insula (AI)) and adolescents' propensity to conform to peer group norms in the context of evaluating music popularity¹². The areas involved in this type of conformity to peers also exhibited similar neural activity during social exclusion, suggesting that such activation may reflect the perceived threat of social rejection and the consequent need for behavioral adjustment to maintain group harmony¹³. In a study with two experts as social sources, agreement with the experts on music preference was reflected in the activity of the brain regions associated with reward processing ²⁵. In the study by Klucharev and colleagues, pairing an expert celebrity with a product that matched their expertise led to better product recognition and a positive attitude toward the product in the later assessment⁷. This expertise-congruent pairing was accompanied by increased activation in brain regions involved in reward processing and memory formation.

In studies comparing brain responses to two different social sources, where two differentially credible sources, a financial expert versus a peer source were compared. Participants who were exposed to the expert advice exhibited greater activation in the ACC and superior frontal gyrus regions when they chose to ignore the expert’s advice compared to those who disregarded the peer advice¹⁸. The activation of the ACC when ignoring the expert’s advice was considered an indicator of error detection and conflict monitoring, signaling the need to adjust future behavior based on the expert’s advice. The ACC activation was accompanied by superior frontal gyrus activation which is linked to certainty-related processing, meaning participants in the expert group may have required more certainty that disobeying the advice was the rational decision to make before choosing that action¹⁸. In the study comparing expert advice to novice advice, expert advice was used more by participants compared to the novices’ and judgment accuracy was higher in the expert group compared to the non-expert, and this preference correlated with activity in the ventral striatum and orbitofrontal cortex (OFC), brain regions associated with reward processing¹⁰.

The main body of studies utilizing electroencephalography (EEG) to investigate social influence with a peer group has demonstrated that a discrepancy between an individual's opinion and the peer group elicits a fronto-central voltage deviation in the event-related potential (ERP) component known as feedback-related negativity (FRN). The FRN occurs between 200 and 400 milliseconds post-onset of conflicting group feedback, often recorded at the Fz electrode, and is localized to the MFC. It has been proposed that the FRN reflects a neural response akin to a negative prediction error, similar to the activation of the MFC and deactivation of the ventral striatum observed in reinforcement learning models ^5,26–28. In magnetoencephalography (MEG) studies of social influence with group opinion electromagnetic brain responses to the disagreement between an individual and group opinion were similar to FRN component with larger negative deflection for conflict trials^6,15. A mismatch between individual and group opinions evoked an N400-like component instead of FRN, predominantly involved in conceptual processing and violation of semantic expectations in language studies^29,30. In a social influence context, a larger N400 might reflect getting feedback that contradicts an individual's knowledge or beliefs about the world²⁹. Irani and colleagues' study²² with peer and expert feedback showed that conflict with peers compared to conflict with experts in adolescent populations evoked a stronger response in the FRN time window probably related to violation of peer group norms, as the FRN component reflects the automatic detection of errors with sensitivity to the significance of the negative outcome.

Peer and expert opinions affect individuals through distinct underlying motivations. However, when comparing the effects of these influences, it remains unclear which will have a greater impact on the individual. Based on prior evidence, we assumed that expertise could outweigh peer influence when the context involves specialized knowledge. Conversely, peer influence is expected to be more compelling in situations involving social norms, identity, and the need for acceptance. Here, in the domain of fashion preferences, where peer influence is typically strong³¹, we hypothesize that perceived conflicts between individual and peer group opinions will lead to greater opinion change toward that group’s opinion compared to conflicts with expert opinions. This means that positive and negative discrepancies with peer group opinions are expected to cause an individual’s opinion to shift in a more positive or negative direction, respectively.

Furthermore, we hypothesized that a conflict between an individual's preference and the peer group's opinion would evoke stronger FRN-like neural responses compared to a conflict with the expert opinion as automatic detection of errors and deviation from group norm is often reflected in the FRN component^5,6,22,27. In addition to evoked responses, it has been observed that when there is a conflict between an individual’s opinion and the group’s opinion, the associated cognitive and emotional responses are reflected in modulations of neuronal oscillatory activity, specifically theta (4–8 Hz) synchronization and beta (13–30 Hz) desynchronization ^6,15. Thus, we predicted that conflict with peer group opinion compared to expert group opinion would increase the power of the theta band and decrease the power of the beta band oscillations.

Behavioral Results

Experimental manipulation test

The five items used to measure normative influence showed strong internal consistency in ratings from both groups (peer group alpha = 0.74, expert group alpha = 0.60). There was a highly significant difference in normative impression between the two groups (mean difference = 1.57 [mean: expert = 5.09, SD = 0.75, peer = 3.55, SD = 1.10], t(21) = 6.77, p = 0.0001). The four items used to assess Cronbach’s α for informational influence were found to be below 0.40, so we did not combine the single paired-sample t-tests. For the informational subscale, we found a significant mean difference between peer and expert in 2 pairs (“I found peer/expert rating accurate”: Mean difference = 1.22 [mean: expert = 4.77, peer = 3.55], t = 4.07, p = 0.001; and “I trusted peer/expert ratings”, mean difference = 0.90 [mean: expert = 4.55, peer = 3.64], t = 2.44, p = 0.023). The scale was 1 = totally true, 7 = totally untrue.

Conformity

After categorizing trials based on the direction of the group feedback (positive, negative, and no-conflict, see Fig. 4, we tested whether participants’ opinions changed toward the groups’ opinions by comparing the 1st and 2nd ratings. We converted the conditions to dummy variables, and positive and negative conflicts were compared to the no-conflict condition. In our multilevel model, the difference between negative and no-conflict feedback (the second random slope, S2) was significant (mean difference = -0.156, CI = [-0.296, -0.025]; Fig. 1), but no significant difference between positive and no-conflict feedback (random slope S1) was observed (mean difference = -0.024, CI = [-0.129, 0.088]). The influence of the initial rating (first session) on subsequent rating change due to RtM was used as a covariate (S3). The initial rating was found to affect rating change (mean effect= -0.244, CI (-0.299, -0.188) in line with previous similar studies^32,33. The main effect of feedback groups and interaction between feedback groups and feedback were not significant and were therefore omitted from the model.

We did not find any significant correlations between the behavioral results and the personality traits using the seven questionnaires and tests.

Memory task

We found that the participants gave 416 correct responses out of 880 trials in the memory test. The overall hit rate (HR) was 0.47. We compared the actual HR to the HRs from 10,000 random permutations. By chance, the minimum and maximum number of correct guesses were 266 and 372, respectively. We found that the actual HR was higher than the permutation HRs for all three categories: no-conflict, positive, and negative. The actual HRs for each category were 0.45, 0.46, and 0.52, respectively. The permutation HRs for each category ranged from 148 to 202 for no-conflict, from 80 to 130 for positive, and from 22 to 58 for negative. The actual HR was highest for the negative category.

MEG results

The evoked response, locked to the presentation of groups’ ratings, is depicted in a butterfly plot in Fig. 2 for the grand average of both conflict and no-conflict in magnetometer channels (Fig. 2, upper right panel) and for peer conflict and expert conflict, in gradiometer channels with magnetic fields mapped for the highest peaks (Fig. 2, upper left panel). A peak is seen between 200 to 300 ms in the frontal regions. Statistical analysis of the magnetometer data revealed a single cluster where the evoked response to conflicting feedback differed from no-conflict feedback (Fig. 2, lower right panel). This late positive complex-like activity appeared in right hemisphere sensors and left posterior at 590–960 ms after feedback, showing a higher amplitude for the no-conflict than the conflict (p = 0.03). Analysis of the gradiometer channels also revealed a difference between processing of conflicting peer and expert feedbacks at 170–590 ms in right temporo-parietal channels, with expert conflict eliciting higher amplitude at the early stage and peer conflict eliciting higher amplitude at the later stage (Fig. 2, lower left panel).

In the present study, we investigated the influence of peer and expert groups’ feedback on individual opinion changes in a fashion preference task, and we examined the underlying neural mechanisms of this influence. Our results showed that participants only changed their opinions when they received feedback that was lower than their ratings, regardless of the source of feedback. At the neural level, we found that agreement (no-conflict feedback) between individuals and groups evoked stronger activity at the late (590–960 ms) time window in the right hemisphere and left posterior magnetometer channels, compared to conflicting feedback. Furthermore, we found a difference between conflict with peer and expert groups in an earlier (170–590 ms) time window in the right temporo-parietal gradiometer channels, with a stronger response to peer feedback compared to expert feedback.

Contrary to our hypothesis, peer and expert feedback did not influence opinion change differently. Nevertheless, our experimental manipulation test confirmed that the participants distinguished between the two feedback groups. The manipulation results indicated that peer feedback exerted a stronger normative influence on participants. Surprisingly, participants also perceived peer feedback as having a greater informational impact than expert feedback, particularly regarding perceived trust and accuracy. However, despite reporting the normative and informational influences of the peer group, participants did not exhibit behavioral conformity to peers. This lack of conformity to peers aligns with the previous study²², where feedback from peers and experts on a social perception task elicited similar conformity effects in adolescents. In Irani et al²², the normative impressions of peer feedback dominated the expert feedback, and the informational impression of the expert feedback was stronger than the peer feedback. However, in the current study with adults, the informational impression from peers was stronger than that of experts. This discrepancy may be attributed to the fashion domain utilized in the current study, where peers often serve as the primary reference group—a group whose attitudes and values an individual appropriates as a basis for his/her behavior ³¹. In this context, participants reported that they identified more strongly with peers and trusted them in possessing more socially appropriate and accurate knowledge compared to experts. Nevertheless, they did not conform to peers more than to experts. The specific experiment here may not have been conducive to observing behavioral conformity. For instance, the peer group identity might be narrower, comprising like-minded individuals from one's immediate social milieu. Consequently, the peer group used in the experiment might not have exerted a distinct conformity compared to the expert group ^11,34. Moreover, dominant social categorizations can be overridden by self-categorizations that better fit the context and individual’s interests³⁵.

We expected both positive and negative conflict to induce conformity behavior, but our results failed to show conformity to the positive conflict. This finding aligns with prior research indicating that conflicts in a negative direction have a more substantial impact than those in a positive direction^4–6,22. Based on Nohlen and colleagues negative judgments are easier and may be considered the ‘default’ action when experiencing evaluative conflict, meaning conflict generates negative value, and this negative value of conflict has also been suggested to extend to evaluations³⁶. In our study, the participants who either did not recall the feedback or only recalled that it conflicted with their judgment, tended to judge the conflict as more negative than positive. In addition to the design effect noted by Nohlen and colleagues, previous studies on social influence related to consumer behavior have shown that negatively framed group opinions are often perceived as more informative or diagnostic than positively framed opinions ^15,37,38. This suggests that negativity may have a stronger impact in consumer contexts, especially when financial outcomes are more prominent. The conformity to the negative direction is also in line with our memory test, where negative feedback was slightly remembered better than positive and no-conflict feedback. Thus, conformity here might be based on the explicit recollection of the feedback and the deliberate adjustment of the ratings to match the negative feedback.

Contrary to the behavioral response, there was a difference between peers and experts in their neural responses to conflicting feedback. However, this result did not replicate the results from the previous work with adolescents showing larger FRN-like responses to conflict with peers than with experts²². In the present study, we observed a different pattern of ERF responses to peer and expert feedback in a more extended time window (170–590 ms)(Fig. 2, left lower panel), exhibiting a spatial distribution uncharacteristic of the FRN component which has a more frontal distribution and typically reaches its maximal amplitude 250 ms after feedback presentation. The timing and topography of activation here might reflect simultaneously occurring processes rather than a particular information processing, e.g., error detection or expectation violation reflected in FRN response. These differences in evoked responses of adults and adolescents align with the findings from the manipulation test, which demonstrated a stronger normative influence from peers in adolescents, the neural response to differences between peer and expert feedback was observed within the FRN time window, typically associated with norm violation and expectation. In contrast, in adults, both normative and informational influences are more pronounced for peers, with neural responses to peer-expert differences occurring over a more extended time window, likely reflecting a combination of normative and informational processing.

Additionally, the neural responses indicated a sensitivity to peer feedback that did not immediately translate into behavioral changes. This is likely because behavioral conformity may require more time to manifest, as it involves higher-order decision-making processes^39–41. However, it is important to acknowledge that cluster-based permutation has limitations in accurately estimating specific time points and cluster topographies⁴².

Beyond the observed evoked responses to group differences, our study revealed distinct neural activity in response to conflict versus no-conflict feedback, diverging from previous findings. Notably, the FRN-like responses identified by many studies^5,6,26,27. The N400 component reported by Huang and colleagues²⁹, and the long latency responses observed by Irani and colleagues, generally exhibit stronger responses to conflicting social feedback¹⁵. These evoked responses, characterized by longer latencies, are typically indicative of higher-level cognitive processing and are less focal in location, potentially reflecting network-level patterns ⁴³. Our findings demonstrated the emergence of a late positive complex (LPC)-like activity with stronger activation for no-conflict feedback at 600–1000 ms after stimulus onset. The LPC is an ERP component commonly recorded as mean amplitude at the CPz electrode between 400–1000 ms and is known to be larger for motivationally salient stimuli and high arousal levels ^44,45. Moreover, positive-going LPC amplitudes are associated with enhanced attention, elaborative evaluation of ongoing events, and improved subsequent memory performance⁴⁶. Previous research has indicated that LPC responses to social feedback tend to be larger for acceptance feedback compared to rejection feedback⁴⁷. In our study, agreement feedback elicited stronger neural responses than disagreement feedback, regardless of the social source. This may indicate sustained cognitive processing of agreement feedback, occurring at a later time window compared to disagreement feedback processing. The potential significance of the sustained agreement processing may be the individual perceived self-similarity to a social group and reinforcing the perceived correctness of participants' initial opinions.

In conclusion, we observed that young adults exhibited similar levels of conformity to peers and experts in the context of fashion preference, despite the higher normative and informational impressions from peers. The distinct neural early-stage temporal dynamics in response to conflicting feedback from peers versus experts might reflect this perceived distinction between peers and experts, followed by a more elaborate evaluation of the feedback regardless of social source in a later time window. Furthermore, our findings indicate that behavioral conformity was present only in response to negative feedback, supported by memory test results suggesting explicit memory and intentional conformity might play a role. Future studies should replicate these findings with larger sample sizes to validate our results and further explore the mechanisms underlying social influence and conformity.

Participants

Twenty-two students (18 females, mean age 23.40 years, SD = 2.19, 21 right-handed) from the University of Jyväskylä took part in the study. None of the participants reported a history of psychiatric or neurological illness, or drug and alcohol abuse. All participants had normal or corrected to normal eyesight. Subjects received a 20-euro gift card as compensation for their participation in the experiment. The ethical committee of the University of Jyväskylä approved the study, and all participants signed an informed consent form.

Stimuli and task

The stimuli were images of 300 unisex clothing and accessories. Before the start of the experiment, subjects were explained that they were participating in an experiment on fashion opinions and how their brains function while evaluating the clothing. They were informed they would be shown the average ratings of two different groups about the same items that have been evaluated already to keep their interest during the experiment. The subjects were told that two groups consisted of hundreds of people: one group of students from Jyväskylä University, and another group of top experts in fashion businesses including designers, stylists, and textile designers. Therefore, participants were unaware of the real aim of the experiment. During the MEG measurement, subjects were first presented with the 300 items (first rating session, session duration 45 min). Subjects were instructed to rate each item on an 8-point scale, based on the claim ‘how much you like the item’ (1 = totally dislike; 8 = totally like). Two nonmagnetic, 4-button devices (Cambridge Research Systems, Ltd., UK) were used to select the ratings. At the beginning of the experiment, subjects rated 10 practice trials with unrelated stimuli to get familiar with the use of the response buttons and the task (including group cues and feedback symbols). The structure and timing of the trials are illustrated in Fig. 3.

In each experimental trial, a clothing item was displayed on the screen, and subjects had 4 seconds to respond. If the participant did not press a button within 4 seconds after the stimulus presentation, the trial ended, and the message "Too late" appeared on the screen. Upon pressing the corresponding button, the subject's rating value appeared in the middle of the screen for 0.5 seconds. After a random 0.5-2 s delay, the cue image of either peer or expert appeared in the middle of the screen representing which group’s evaluation participant will see for the item for 1 s and then the group rating for 1.5 s was shown with an upward arrow for above the subject’s rating, downward arrow for below subject’s rating, and the percent symbol for the same rating as subject’s choice (Fig. 3). In reality, the group rating was randomized to either match the subject's rating (no-conflict condition) or differ by being higher or lower (positive or negative conflict condition). This manipulation was designed to create a sense of social conflict between the subject and the group. The conditions were balanced among the 300 trials (1/2 no-conflict and 1/2 positive and negative conflict). Stimulus presentation was controlled with the Presentation software (Neurobehavioral Systems, Inc., Albany, CA, USA).

Thirty minutes after the initial rating session, subjects were asked to rate the same items again, this time without group cues or feedback, and in a new randomized order (second rating session, duration 30 minutes). This second session aimed to determine whether subjects altered their initial evaluations after being exposed to differing peer and expert opinions, in line with social influence theory. Importantly, subjects were not informed in advance about this second session. At the end of the experiment, the subjects were interviewed about the experiment, and the true nature of the experiment was revealed. Importantly, none of them reported guessing that the aim of the study was about social influence or conformity.

After the second session, in a self-paced behavioral session, subjects were presented with 80 items to test how much they could recall from the experiment. 40 items associated with either peer or expert group were picked from the first session Additionally, 40 novel control items were presented that were not included in the MEG sessions. The memory task consisted of two questions accompanied by an image of the clothing item: first “Which group evaluated the clothing?” below each item, with three response options (peers, experts, and neither). If the third option (“neither”) was chosen, the second question was not presented. The second question was: How was the group rating compared to yours? This question had three response options (same, bigger, smaller).

Following the memory task, a manipulation test questionnaire was administered to determine whether participants found the peer and expert groups and their feedback different^32,48,49. In addition to investigating the basic underlying processes of disagreement with peers and experts, all subjects answered online questionnaires related to the experiment to test possible correlations between conformity behavior and participants’ personality traits. We administered the Finnish version of the behavioral activation system/behavioral inhibition system (BAS/BIS)⁵⁰, the theory of mind assessment scale (Bosco et al., 2009), and a short version of the five-factor model (FFM)⁵¹ before the MEG session. Aspects of Identity Questionnaire⁵² and Emergent adult Peer Pressure Inventory (EAPPI)⁵³ were administered after the MEG session since they could be easily associated with social influence and conformity behavior and reveal the real aim of our experiment earlier than intended.

MEG data acquisition

MEG data were collected with a whole-head 306-channel (102 magnetometer channels and 204 planar gradiometer channels) Elekta TRIUX MEG device (MEGIN Oy, Helsinki, Finland) located in a magnetically shielded room at the Center for Interdisciplinary Brain Research, Dept. Psychology, University of Jyväskylä, Finland, with a 1000 Hz sampling rate and 0.1–330 Hz band-pass acquisition filter. Prior to data acquisition, participants were checked for magnetic interference and instructed to keep their heads still as much as possible. Visual stimuli with the size of 14 cm × 20 cm were presented with a DLP projector on the center of the screen situated in front of the subject at a distance of 100 cm. Concurrently, an electrooculogram (EOG) was recorded using electrodes above the right and below the left eye for the blink and saccade artifacts, and an electrocardiogram (ECG) with electrodes on the left and right clavicle to help detect cardiac artifacts. One ground electrode was attached to the right clavicle bone. The head position was recorded with five head-position indicator (HPI) coils, three at the forehead and one behind each ear. HPI coil positions were determined with reference to three anatomical landmarks: nasion and left and right preauricular points, which define the head coordinate system. All locations were digitized using a Polhemus Isotrak 3D tracker system (Polhemus, Colchester, VT, United States).

Data analysis

Behavioral data

Experimental manipulation test

A manipulation test was used to evaluate the efficacy of the manipulation of normative and informational influence by the two groups. 5 pairs of questions with ratings 1–7 were designed to assess the normative influence and 4 pairs for the informational influence. The analysis was conducted using SPSS (version 26). To determine the internal consistency and reliability of the questions, Cronbach's Alpha was calculated for each section: peer normative, expert normative, peer informational, and expert informational. Individual questions were averaged into a single value when the alpha value was above the predetermined threshold of 60%. A paired sample t-test was applied to compare the mean differences between the peer and expert groups for normative and informational influence.

Conformity

To assess whether individuals shifted their opinions toward peer and expert group's opinions between the first and second rating sessions, we categorized behavioral responses based on the direction of the groups’ opinion compared to the subject’s opinion in positive, negative, and no-conflict, as presented in Fig. 4. We assessed the opinion change using linear two-level modeling with random slopes¹⁵. The model's parameters were estimated with Bayesian estimation, as implemented in Mplus v8.2 to demonstrate that our null results were unlikely to be attributed to the low sample size. We assessed whether an individual's opinion was changed to the direction of the peer and expert feedback between the first and the second rating in positive, negative, and no-conflict feedback conditions (Fig. 1). Behavioral categories were transformed into dummy variables, which were used to estimate random slopes in the two-level model. We contrasted positive conflict and negative conflict with the no-conflict condition (random slopes S1 and S2, respectively). The initial rating was added as a covariate (S3) to control for the effect of regression to mean (RTM) due to repeated measurements⁵⁴. We examined the main effect of the feedback group on rating change as well as the interaction effect between the feedback group and feedback. We removed these factors from the final model after finding that these factors did not significantly influence rating change.

Memory test

We measured the accuracy of responses in the memory test with three categories: no-conflict, positive, and negative. We computed the hit rate (HR) for each category and the overall data. The HR is the ratio of correct responses to shown trials. To test if the responses were better than random guessing, we generated 10,000 random permutations of the category labels and computed the HR for each permutation. We then compared the actual HR to the permutation HRs to see how often the actual HR was higher than the permutation HRs.

MEG data preprocessing

First, a temporally extended signal space separation method (tSSS) from Maxfilter software (MEGIN, Helsinki, Finland) was used to remove external magnetic interference from MEG data and to correct for head movements, recorded with continuous HPI ⁵⁵. Head position was estimated in 200 ms time windows with 10 ms step for movement compensation and transformed to the mean head position across the MEG session. Independent component analysis (Fast ICA)⁵⁶ was applied to raw MEG data (low-pass filtered at 40 Hz and high-pass filtered at 0.5 Hz) to manually identify and remove signal artifacts corresponding to horizontal saccades, blinks, and cardiac activity. In this analysis, the subsequent MEG data processing steps and forward modeling were performed in MNE-Python, v0.19⁵⁷ using custom scripts. After ICA, the signals were downsampled by 3 (to 333.33 Hz) to reduce data size for further analysis. Since the experimental conditions defined by peer and expert ratings were the main variable in our experiment, we segmented the continuous data to epochs time-locked to the presentation of the peer and expert group feedback (Fig. 3) to investigate event-related neuronal responses. We grouped the epochs into positive conflict, negative conflict, and no-conflict trials, depending on whether the group's ratings were higher, lower, or equal to the participant’s rating, respectively. An epoch was rejected if any magnetometer channel exceeded 4e-12 T or any gradiometer channel 400e12 T/m. The trigger-to-stimulus delay, measured using a photosensitive resistor, was subtracted from each epoch.

Event-Related Field (ERF)

For the ERF analysis, the continuous MEG data was processed using a zero-phase FIR filter, with a low-pass filter set at 40 Hz and a high-pass filter set at 0.5 Hz. Epochs were selected from 200 ms before to 1100 ms after peer and expert feedback presentation onset. Baseline correction was applied to each trial by subtracting the mean of the pre-stimulus interval from − 200 to 0 ms. Evoked fields were estimated by averaging the epochs within each condition: conflict (negative and positive conflicts) and no-conflict.

Statistical analysis

To assess differences in neural responses between conflict and no-conflict conditions for peer and expert feedback and to identify relevant time windows, we conducted a nonparametric permutation test with a clustering method to correct for multiple comparisons⁵⁸. This analysis was performed for the time window between 0 and 1100 ms. For each sample (time point for ERF), the difference between conflict and no-conflict conditions was calculated using dependent samples t-statistic. Samples where these t-statistics exceeded an uncorrected threshold of α = 0.05 were clustered based on their spatial and temporal adjacency. Cluster-level test statistics were calculated by summing the t-statistics of the samples belonging to the same cluster. The largest cluster-level t-statistic was used as a test statistic as recommended by (among others) Maris and Oostenveld⁵⁸. A permutation distribution of cluster-level statistics was then calculated by randomly exchanging condition labels between epochs and computing positive and negative cluster-level statistics for each permutation, for a total of 5000 permutations. The observed cluster-level statistic was then tested against this permutation distribution to determine the permutation-based p-value.

Author contributions

FI: Conceptualization, Data collection, Analysis, Investigation, Visualization, Writing—Original Draft, Review & Editing; PL: Conceptualization, Methodology, Review & Editing; JM: Methodology, Software, Analysis; SM: Conceptualization, Data collection; TP: Conceptualization, Supervision, Resources; KH: Conceptualization; SM: Conceptualization, Software, Methodology, Formal Analysis, Supervision, Project Administration, Funding Acquisition, Review & Editing, Resources.

Competing interests

The authors declare no competing interests.

Data availability

The datasets are available from the corresponding author upon reasonable request.

Funding

This study was funded by Academy of Finland (Project 298456), Finnish Cultural Foundation (Suomen Kulttuurirahasto), and Alfred Kordelin Foundation.

ORCID

Fatemeh Irani http://orcid.org/0000-0002-2519-8944

Pessi Lyyra https://orcid.org/0000-0001-6230-4024

Joona Muotka http://orcid.org/0000-0002-7113-903X

Tiina Parviainen http://orcid.org/0000-0001-6992-5157

Simo Monto http://orcid.org/0000-0002-8191-6146

Cialdini, R. B. & Goldstein, N. J. Social Influence: Compliance and Conformity. Annu. Rev. Psychol.55, 591–621 (2004).
Wang, J., Molina, M. D. & Sundar, S. S. When expert recommendation contradicts peer opinion: Relative social influence of valence, group identity and artificial intelligence. Comput. Hum. Behav.107, 106278 (2020).
Deutsch, M. & Gerard, H. B. A study of normative and informational social influences upon individual judgment. J. Abnorm. Soc. Psychol.51, 629–636 (1955).
Klucharev, V., Hytönen, K., Rijpkema, M., Smidts, A. & Fernández, G. Reinforcement Learning Signal Predicts Social Conformity. Neuron61, 140–151 (2009).
Shestakova, A. et al. Electrophysiological precursors of social conformity. Soc. Cogn. Affect. Neurosci.8, 756–763 (2013).
Zubarev, I., Klucharev, V., Ossadtchi, A., Moiseeva, V. & Shestakova, A. MEG Signatures of a Perceived Match or Mismatch between Individual and Group Opinions. Front. Neurosci.11, (2017).
Klucharev, V., Smidts, A. & Fernández, G. Brain mechanisms of persuasion: how ‘expert power’ modulates memory and attitudes. Soc. Cogn. Affect. Neurosci.3, 353–366 (2008).
Ntoumanis, I. et al. “Expert persuasion” can decrease willingness to pay for sugar-containing food. Front. Nutr.9, (2022).
Engelmann, J. B., Moore, S., Monica Capra, C. & Berns, G. S. Differential neurobiological effects of expert advice on risky choice in adolescents and adults. Soc. Cogn. Affect. Neurosci.7, 557–567 (2012).
Meshi, D., Biele, G., Korn, C. W. & Heekeren, H. R. How Expert Advice Influences Decision Making. PLOS ONE 7, e49748 (2012).
Spears, R. Social Influence and Group Identity. Annu. Rev. Psychol. 72, 367–390 (2021).
Berns, G. S., Capra, C. M., Moore, S. & Noussair, C. Neural mechanisms of the influence of popularity on adolescent ratings of music. NeuroImage 49, 2687–2696 (2010).
Wasylyshyn, N. et al. Global brain dynamics during social exclusion predict subsequent behavioral conformity. Soc. Cogn. Affect. Neurosci. 13, 182–191 (2018).
Molleman, L., Ciranka, S. & van den Bos, W. Social influence in adolescence as a double-edged sword. Proc. R. Soc. B Biol. Sci. 289, 20220045 (2022).
Irani, F. et al. Brain dynamics of recommendation-based social influence on preference change: A magnetoencephalography study. Soc. Neurosci. 17, 397–413 (2022).
Gardner, M. & Steinberg, L. Peer influence on risk taking, risk preference, and risky decision making in adolescence and adulthood: an experimental study. Dev. Psychol. 41, 625–635 (2005).
Breiner, K. et al. Combined effects of peer presence, social cues, and rewards on cognitive control in adolescents. Dev. Psychobiol. 60, 292–302 (2018).
Suen, V. Y. M., Brown, M. R. G., Morck, R. K. & Silverstone, P. H. Regional Brain Changes Occurring during Disobedience to “Experts” in Financial Decision-Making. PLOS ONE 9, e87321 (2014).
Lourenco, F. S. et al. Consider the Source: Adolescents and Adults Similarly Follow Older Adult Advice More than Peer Advice. PLOS ONE 10, e0128047 (2015).
Mahmoodi, A., Bahrami, B. & Mehring, C. Reciprocity of social influence. Nat. Commun. 9, 2474 (2018).
Zonca, J., Folsø, A. & Sciutti, A. The role of reciprocity in human-robot social influence. iScience24, (2021).
Irani, F., Muotka, J., Lyyra, P., Parviainen, T. & Monto, S. Social influence in adolescence: Behavioral and neural responses to peer and expert opinion. Soc. Neurosci. 1–12 (2024) doi:10.1080/17470919.2024.2323745.
Mason, M. F., Dyer, R. & Norton, M. I. Neural mechanisms of social influence. Organ. Behav. Hum. Decis. Process.110, 152–159 (2009).
Klucharev, V., Munneke, M. A. M., Smidts, A. & Fernández, G. Downregulation of the Posterior Medial Frontal Cortex Prevents Social Conformity. J. Neurosci.31, 11934–11940 (2011).
Campbell-Meiklejohn, D. K., Bach, D. R., Roepstorff, A., Dolan, R. J. & Frith, C. D. How the Opinion of Others Affects Our Valuation of Objects. Curr. Biol.20, 1165–1170 (2010).
Chen, J., Wu, Y., Tong, G., Guan, X. & Zhou, X. ERP correlates of social conformity in a line judgment task. BMC Neurosci.13, 43 (2012).
Kim, B.-R., Liss, A., Rao, M., Singer, Z. & Compton, R. J. Social deviance activates the brain’s error-monitoring system. Cogn. Affect. Behav. Neurosci.12, 65–73 (2012).
Yu, R. & Sun, S. To Conform or Not to Conform: Spontaneous Conformity Diminishes the Sensitivity to Monetary Outcomes. PLOS ONE8, e64530 (2013).
Huang, Y., Kendrick, K. M. & Yu, R. Social conflicts elicit an N400-like component. Neuropsychologia65, 211–220 (2014).
Schnuerch, R., Richter, J., Koppehele-Gossel, J. & Gibbons, H. Multiple neural signatures of social proof and deviance during the observation of other people’s preferences. Psychophysiology53, 823–836 (2016).
Makgosa, R. & Mohube, K. Peer influence on young adults’ products purchase decisions. (2007).
Izuma, K. & Adolphs, R. Social Manipulation of Preference in the Human Brain. Neuron78, 563–573 (2013).
Levorsen, M., Ito, A., Suzuki, S. & Izuma, K. Testing the reinforcement learning hypothesis of social conformity. Hum. Brain Mapp.42, 1328–1342 (2021).
Hackel, L. M., Zaki, J. & Van Bavel, J. J. Social identity shapes social valuation: evidence from prosocial behavior and vicarious reward. Soc. Cogn. Affect. Neurosci.12, 1219–1228 (2017).
Cikara, M. & Van Bavel, J. J. The Neuroscience of Intergroup Relations: An Integrative Review. Perspect. Psychol. Sci.9, 245–274 (2014).
Nohlen, H. U., van Harreveld, F. & Cunningham, W. A. Social evaluations under conflict: negative judgments of conflicting information are easier than positive judgments. Soc. Cogn. Affect. Neurosci.14, 709–718 (2019).
Cascio, C. N., O’Donnell, M. B., Bayer, J., Tinney, F. J. & Falk, E. B. Neural Correlates of Susceptibility to Group Opinions in Online Word-of-Mouth Recommendations. J. Mark. Res.52, 559–575 (2015).
Baek, E. C. et al. Activity in the brain’s valuation and mentalizing networks is associated with propagation of online recommendations. Sci. Rep.11, 11196 (2021).
Chase, H. W., Swainson, R., Durham, L., Benham, L. & Cools, R. Feedback-related Negativity Codes Prediction Error but Not Behavioral Adjustment during Probabilistic Reversal Learning. J. Cogn. Neurosci.23, 936–946 (2011).
Martín, R. S., Appelbaum, L. G., Pearson, J. M., Huettel, S. A. & Woldorff, M. G. Rapid Brain Responses Independently Predict Gain Maximization and Loss Minimization during Economic Decision Making. J. Neurosci.33, 7011–7019 (2013).
Bogdan, P. C. et al. Direct feedback and social conformity promote behavioral change via mechanisms indexed by centroparietal positivity: Electrophysiological evidence from a role-swapping ultimatum game. Psychophysiology59, e13985 (2022).
Sassenhagen, J. & Draschkow, D. Cluster-based permutation tests of MEG/EEG data do not establish significance of effect latency or location. Psychophysiology56, e13335 (2019).
Parviainen, T. & Kujala, J. Event-Related Potentials (ERPs) and Event-Related Fields (ERFs). in Language Electrified: Principles, Methods, and Future Perspectives of Investigation (eds. Grimaldi, M., Brattico, E. & Shtyrov, Y.) 195–239 (Springer US, New York, NY, 2023). doi:10.1007/978-1-0716-3263-5_7.
Hajcak, G., MacNamara, A. & Olvet, D. M. Event-Related Potentials, Emotion, and Emotion Regulation: An Integrative Review. Dev. Neuropsychol.35, 129–155 (2010).
Hajcak, G. & Foti, D. Significance?... Significance! Empirical, methodological, and theoretical connections between the late positive potential and P300 as neural responses to stimulus significance: An integrative review. Psychophysiology57, e13570 (2020).
Luo, Y. et al. Social Comparison Manifests in Event-related Potentials. Sci. Rep.5, 12127 (2015).
Funkhouser, C. J. et al. Social Feedback Valence Differentially Modulates the Reward Positivity, P300, and Late Positive Potential. J. Psychophysiol.34, 255–267 (2020).
Burnkrant, R. E. & Cousineau, A. Informational and normative social influence in buyer behavior. J. Consum. Res.2, 206–215 (1975).
Roberts, J. A., Manolis, C. & (Jeff) Tanner Jr., J. F. Interpersonal influence and adolescent materialism and compulsive buying. Soc. Influ.3, 114–131 (2008).
Carver, C. S. & White, T. L. Behavioral Inhibition, Behavioral Activation, and Affective Responses to Impending Reward and Punishment: The BIS/BAS Scales.
Konstabel, K., Lönnqvist, J., Walkowitz, G., Konstabel, K. & Verkasalo, M. The ‘Short Five’ (S5): Measuring Personality Traits Using Comprehensive Single Items. Eur. J. Personal.26, 13–29 (2012).
Cheek, J., Tropp, L., Underwood, M. & Cheek, N. The Distinction between Social and Collective Identity Orientations in the Aspects of Identity Questionnaire. (2013).
Bradley, G. & Wildman, K. Psychosocial Predictors of Emerging Adults’ Risk and Reckless Behaviors. J. Youth Adolesc.31, 253–265 (2002).
Yu, R. & Chen, L. The need to control for regression to the mean in social psychology studies. Front. Psychol.5, (2015).
Taulu, S. & Hari, R. Removal of magnetoencephalographic artifacts with temporal signal-space separation: Demonstration with single-trial auditory-evoked responses. Hum. Brain Mapp.30, 1524–1534 (2009).
Hyvärinen, A. & Oja, E. Independent component analysis: algorithms and applications. Neural Netw.13, 411–430 (2000).
Gramfort, A. et al. MEG and EEG data analysis with MNE-Python. Front. Neurosci.7, (2013).
Maris, E. & Oostenveld, R. Nonparametric statistical testing of EEG- and MEG-data. J. Neurosci. Methods164, 177–190 (2007).

No competing interests reported.

Download PDF

Editorial decision: Revision requested
12 Nov, 2024
Reviews received at journal
11 Nov, 2024
Reviews received at journal
06 Nov, 2024
Reviewers agreed at journal
30 Oct, 2024
Reviewers agreed at journal
30 Oct, 2024
Reviewers invited by journal
08 Oct, 2024
Editor assigned by journal
08 Oct, 2024
Editor invited by journal
04 Oct, 2024
Submission checks completed at journal
01 Oct, 2024
First submitted to journal
18 Sep, 2024

You are reading this latest preprint version

Neural dynamics of perceived agreement and disagreement with peer and expert opinions: An MEG study

Status:

Version 1

Abstract

Figures

Introduction

Results

Behavioral Results

Experimental manipulation test

Conformity

Memory task

MEG results

Discussion

Methods

Participants

Stimuli and task

MEG data acquisition

Data analysis

Behavioral data

Experimental manipulation test

Conformity

Memory test

MEG data preprocessing

Event-Related Field (ERF)

Statistical analysis

Declarations

References

Additional Declarations

Status:

Version 1