Introduction
The idea that individual decisions are largely shaped by the behavior of those around us is a cornerstone of social psychology (Ajzen & Fishbein, 1980; Asch, 1955; Cialdini & Goldstein, 2004; Milgram, 1974). Until recently, most neurobiological research examined decision-making mechanisms in isolation from social context. However, advances in neuroscience have not only allowed us to explore the biological foundations of decision-making but have also led to the emergence of a new interdisciplinary field—neuroeconomics (Glimcher & Rustichini, 2004)—which combines neuroscience, economics, social psychology, and other disciplines to form new models of decision-making mechanisms (Klucharev, Smids, Shestakova, 2011). This review analyzes key neurobiological studies from recent years, which suggest a link between social influence and fundamental neurobiological (dopaminergic) mechanisms of learning processes.
Conformity Behavior
Early neurobiological studies of social influence focused on the mechanisms of conformity. Conformity is a specific form of social influence in which an individual changes their beliefs, evaluations, and behavior to align with the norms of a particular social (reference) group, even without direct instruction or demand. According to Robert Cialdini and Jeff Goldstein (2004), conformity arises under the influence of two types of social norms:
- Prescriptive norms specify expected or required behavior and usually carry a moral judgment (e.g., “Don’t drive under the influence”).
- Descriptive norms describe the typical behavior of most people in a group or situation, regardless of moral value (e.g., fashion trends or smoking in youth subcultures).
While prescriptive norms regulate behavior and violations may result in sanctions, descriptive norms simply inform about the dominant behavior in a group. Descriptive norms and the associated social approval (validation) are highly effective and can influence behaviors such as tax compliance or adherence to environmental standards (Kenrick, 2003). Interestingly, most people are unaware of the impact of descriptive norms in daily life (Bryan & Test, 1967).
Conformity can be motivated by:
- The desire for accurate interpretation of reality and effective behavioral strategies
- The desire for social approval
- The desire to maintain positive self-esteem and avoid cognitive dissonance
Social psychology emphasizes the role of public approval in reinforcing conformity (Cialdini & Goldstein, 2004), while behavioral economics focuses on sanctions for norm violations (Fehr & Fischbacher, 2004). Both approaches can be described using reinforcement learning theory, where social validation reinforces norm-congruent behaviors and punishes deviations. The nervous system may evaluate actions not only by individual goals but also by their alignment with social norms, treating discrepancies as errors requiring behavioral correction.
Neuroeconomic studies (Klucharev et al., 2009, 2011; Shestakova et al., 2012) have shown that social influence mechanisms are part of the fundamental neural system for performance monitoring, as described in reinforcement learning models (Sutton & Barto, 1998). Klucharev et al. (2009) demonstrated that deviation from group norms generates a neural response similar to the “reward prediction error” (RPE)—a key parameter in reinforcement learning. The RPE signal acts as a learning signal, indicating the need to modify behavior to avoid future errors. Thus, when an individual’s opinion diverges from the group, an RPE signal is generated, prompting a change in opinion to align with the social norm.
The Dopaminergic System and Reinforcement Learning
Adaptive behavior relies on the ability to learn from the outcomes of one’s actions (Thorndike, 1911). Reinforcement learning theory formalizes this process (Sutton & Barto, 1998), where the difference between expected and actual outcomes is the RPE, which triggers learning. The subject updates their prediction of the value (V) of a situation or stimulus as follows:
Vnew = Vold + h (R - Vold)
Here, R is the actual reward, h is the individual learning rate, and (R – Vold) is the RPE. The RPE is generated at the moment the outcome is revealed.
The temporal difference model also postulates an RPE signal, modeling it as a continuum of prediction updates until the outcome occurs (Niv & Schoenbaum, 2008). Research links RPE generation to the brain’s dopaminergic system—networks of neurons using dopamine as a neurotransmitter (Schultz, 2007). Dopaminergic neuron activity in the midbrain matches theoretical RPE models: increased firing for positive RPEs (better-than-expected outcomes) and decreased firing for negative RPEs (worse-than-expected outcomes). Neuroimaging in humans confirms RPE signals in the substantia nigra (SNc), ventral tegmental area (VTA), nucleus accumbens (NAc), and medial prefrontal cortex (pMPFC) (Gehring et al., 1993; McClure et al., 2004; O’Doherty et al., 2004).
The rostral cingulate zone (RCZ) of the frontal cortex is especially important for RPE generation. RCZ is activated when behavior must change due to errors or high error probability (Cohen & Ranganath, 2007; Ridderinkhof et al., 2004). Its activation predicts subsequent behavioral changes (Kerns et al., 2004). RCZ activity is modulated by dopaminergic RPE signals, regardless of whether the outcome is better or worse than expected (Holroyd & Coles, 2002; Matsumoto et al., 2007). The NAc is also involved in reward prediction and is activated by cues preceding reinforcement (Knutson & Wimmer, 2007). Disruption of RCZ-NAc connections impairs decision-making and learning (Hauber & Sommer, 2009; Parkinson et al., 2000). This suggests a specialized brain system for performance monitoring and RPE generation, involving RCZ and NAc.
The Social Influence Mechanism Hypothesis
If every action outcome is evaluated for both its subjective “physiological” value and its “social” value (conformity to group behavior), then deviation from group norms should be integrated into the overall RPE signal. Disagreement with the group should generate an RPE signal similar to that seen in classic errors (e.g., monetary loss or incorrect answers), involving RCZ and NAc activation. The strength of RCZ and NAc activation may indicate an individual’s susceptibility to social influence. Importantly, conflict with social norms is not a behavioral error in the strict sense, but any deviation from majority behavior (Montague & Lohrenz, 2007).
Social Influence Research Using Non-Invasive Neuroimaging
The first neuroimaging study of social influence was by Gregory Berns et al. (2005), using a paradigm similar to Solomon Asch’s to study the neural basis of conformity during mental rotation tasks. Later research developed new paradigms using functional brain mapping. Klucharev et al. (2009) created an experiment comparing individual and group opinions on facial attractiveness, modeling opinion conflict. In such conflicts, RCZ activation (reflecting the RPE signal) and NAc deactivation were observed, indicating that opinion disagreement automatically generates neural activity resembling an RPE signal. The amplitude of NAc deactivation correlated with subsequent opinion change due to social influence. Other studies confirmed that differences between individual and group opinions modulate RCZ (Behrens et al., 2008; Berns et al., 2010; Campbell-Meiklejohn et al., 2010; Falk et al., 2010) and NAc (Campbell-Meiklejohn et al., 2010; Klucharev et al., 2009) activity—regions involved in performance monitoring and RPE generation.
The role of RCZ in conformity was tested using transcranial magnetic stimulation (TMS) to temporarily suppress RCZ activity (Klucharev et al., 2011). TMS suppression of RCZ reduced the degree and likelihood of conformity, showing that susceptibility to social influence can be explained by modulation of the performance monitoring system, leading to behavioral changes.
Other studies support RCZ involvement in opinion change under social influence. For example, RCZ and NAc activation was observed when participants’ opinions conflicted with those of experts in music evaluation tasks (Campbell-Meiklejohn et al., 2010), with higher signal amplitude in more conforming individuals. Berns et al. (2010) found that peer group opinions influenced teens’ musical preferences, with RCZ activation linked to opinion change. Behrens et al. (2009) showed that social information is processed by structures associated with reinforcement learning, with both social and personal information integrated in the ventromedial prefrontal cortex.
Izuma and Adolphs (2013) found that the direction of opinion change and RCZ activation under social influence depend on positive or negative self-identification with the reference group. RCZ activation correlated with the degree of opinion change, even four months later. These and other fMRI studies indicate that the dopaminergic system, including NAc and RCZ, is involved in changing individual opinions under social influence. Notably, increasing brain dopamine with methylphenidate increases conformity (Campbell-Meiklejohn et al., 2012).
But do experimentally observed opinion changes reflect true internal reevaluation (private acceptance) or just outward compliance (public compliance)? NAc activation encodes preference levels even when direct evaluation is not required (Elliott et al., 2004; O’Doherty et al., 2004; Tricomi et al., 2004). Studies by Mason et al. (2009) and Zaki et al. (2011) showed that changes in ratings of abstract symbols and facial attractiveness due to group conflict were accompanied by long-term changes in NAc activation, suggesting internalization of group opinions.
Feedback-Related Negativity (FRN) as an EEG Marker of Social Influence
The hypothesis that social influence is mediated by the performance monitoring system involved in reinforcement learning is supported by EEG studies (Chen et al., 2012; Kim et al., 2012; Shestakova et al., 2012). The feedback-related negativity (FRN) component of event-related potentials (ERPs) is best explained by reinforcement learning theory (Holroyd & Coles, 2002; Nieuwenhuis et al., 2004; Walsh & Anderson, 2012). The cingulate gyrus (including RCZ) is considered the main FRN generator (Cohen & Ranganath, 2007; Gehring & Willoughby, 2002; Hewig et al., 2007; Miltner et al., 1997; Nieuwenhuis et al., 2005; Tucker et al., 2003). FRN reflects the RPE signal—phasic decreases in dopaminergic firing that disinhibit RCZ, resulting in more negative ERP deflections (Holroyd & Coles, 2002).
Using ERPs, Shestakova et al. (2013) showed that conflict with group opinion triggers neural activity in the frontocentral cortex (negative ERP deflection at ~200 ms), similar to FRN. Similar results were found by Chen et al. (2012) in a task modeled after Asch’s line judgment paradigm.
Pharmacological studies show that FRN amplitude increases with dopamine agonists (amphetamine) and decreases with antagonists (haloperidol, pramipexole) (de Bruijn et al., 2006; Zirnheld et al., 2004), supporting the role of dopaminergic structures in FRN generation, though other neurotransmitter systems may also be involved (Jocham & Ullsperger, 2009).
Overall, EEG and fMRI research supports the model of social influence as a function of the fundamental performance monitoring system described by reinforcement learning models.
Neurobiological Correlates of Persuasion
Unlike conformity, where opinion change occurs without explicit targeted influence, persuasion often results from an active, directed process. Persuasion is not homogeneous, as individuals may change opinions for various reasons. Richard Petty and John Cacioppo (1986; see also Petty & Wegener, 1999) distinguish two types of persuasion: (a) direct, based on logical argumentation, and (b) indirect, based on triggering an “automatic compliance mechanism.” Factors influencing persuasion effectiveness include: 1) communicator authority; 2) message content; 3) message delivery; 4) audience’s ability to process arguments.
Although few studies have examined the neurobiology of persuasion, some objective markers have been identified. For example, fMRI studies show that activation level and intensity in the anterior medial prefrontal cortex (including RCZ) while watching anti-smoking videos better predict whether someone will quit smoking than their self-reported persuasiveness ratings (Chua et al., 2009; Falk et al., 2010, 2011). Thus, activation in this brain area may be a more reliable marker of message internalization than survey data.
Cognitive Dissonance and Social Influence
The concept of cognitive dissonance (Festinger, 1959) and structural balance theory (Heider, 1958) are foundational to social influence theory. The idea is that individuals prefer situations where new information aligns with existing beliefs and judgments; when new information conflicts, people seek to minimize the discrepancy.
Cognitive dissonance is a state of psychological discomfort caused by conflicting ideas, beliefs, values, or emotions. This discomfort motivates people to change their attitudes. Experiments show that after being forced to perform unpleasant actions (e.g., eating unusual food—Zimbardo et al., 1965—or writing an essay supporting higher tuition—Steele et al., 1981), most participants become more positive toward the previously uncomfortable topic, reducing dissonance. Dissonance can also arise if an individual’s attitude toward an object differs from that of a liked group, or if their opinion matches that of a disliked group. In both cases, people try to resolve the conflict by changing their attitude toward the object or the group.
Izuma and Adolphs (2013) suggest that the neurobiological mechanism underlying opinion change under social influence may be the drive to reduce cognitive dissonance. RCZ activation is observed not only in social influence situations but also during psychological discomfort from conflicting beliefs or emotions (Izuma et al., 2010; van Veen et al., 2009). Izuma concluded that RCZ activity reflects an “index of internal consistency” of opinion or behavior, i.e., internal dissonance during social influence (Izuma & Adolphs, 2013; Izuma, 2013). However, cognitive dissonance may be a special case of reinforcement learning: if group opinion is treated as a parameter signaling reward probability, the temporal difference model predicts RPE generation both when the participant’s opinion diverges from a “preferred” group and when it matches a “non-preferred” group. In both cases, there is a mismatch between outcome and expectation, as shown by Izuma et al. (2013). Thus, cognitive dissonance may reflect a conflict between current outcomes and long-term goals, functionally similar to the RPE mechanism.
Conclusion
Research using non-invasive neuroimaging confirms that the mechanism of social influence is closely linked to the brain’s performance monitoring system. The nervous system constantly tracks not only basic behavioral outcomes (rewards or punishments) but also the alignment of individual behavior with group norms. Both the absence of expected rewards and deviation from group behavior are treated as behavioral errors, reflected in RPE signals generated by the brain’s dopaminergic system. This signal triggers behavioral change to align with social norms.
Why do humans exhibit such automatic conformity? There are likely strong evolutionary reasons. The concept of evolutionarily stable strategies (Dawkins, 1976) suggests that in a stable environment, most individuals will follow the same strategy only if it offers advantages over alternatives, and deviations are punished by natural selection. Thus, from both evolutionary theory and game theory perspectives, it is rational to follow the majority.
Understanding the neurobiological mechanisms underlying social behavior regulation is valuable not only for basic science—expanding research methodologies and enabling more precise models and experimental testing—but also for practical applications such as social advertising and public health campaigns. Neurobiological data reveal a dynamic brain system that constantly monitors our behavior, compares it to our expectations and group behavior, and automatically adjusts our actions for optimal outcomes. Although neurobiological research in this area is relatively new, it has already demonstrated the fruitful synthesis of neuroscience and social science methods in formulating a fundamental neurobiological mechanism of social influence—a process that shapes our opinions, daily behavior, and even scientific activity.