A Social Contextual Review of the Effects of Alcohol on Emotion

Abstract

Drinking and drinking problems are complex phenomena. Understanding the etiology of alcohol use disorder requires consideration of biological, psychological, and social processes. It is our view that the last of these dimensions is just beginning to receive adequate scrutiny. In this selective review, we discuss the concept of a biopsychosocial analysis of the effects of alcohol. After briefly addressing biological and psychological research on alcohol’s emotional effects, we bid to make a case for the vital role that social processes play in understanding why people drink. The bulk of the paper describes research illustrating the contributions that a social psychological perspective can make to advance understanding of the rewarding effects of alcohol. Overall, studies incorporating social contexts have revealed reliable evidence that alcohol enhances emotional experience in many social environments and have identified socio-contextual variables that moderate responses to alcohol. Further, these studies have broadened the scope of constructs thought to be socially rewarding, including social bonding, relationship functioning, and humor enjoyment. Our analysis concludes by identifying research areas we believe would profit from additional research.

1. Introduction

The consumption of alcohol has long been recognized for both its rewarding and hazardous effects (Steele & Josephs, 1990). On the one hand, drinking is linked to violence, homicide, and a host of damaging health outcomes (World Health Organization, 2018). On the other hand, alcohol has been enjoyed throughout the world for millennia, and is often linked to important celebrations and rituals (MacAndrew & Edgerton, 1969; Slingerland, 2021). To better understand why people drink and why some go on to develop problems with alcohol, investigators have examined the effects of alcohol that reinforce drinking behavior. Indeed, since the mid-1900’s, hundreds of experiments have been conducted to evaluate the extent to which alcohol consumption enhances positive affect and relieves negative affect or stress (Sayette, 1993).

Perhaps surprisingly, many laboratory studies evaluating the effects of alcohol have failed to confirm what researchers, policymakers, and the general public have often taken for granted: that drinking alcohol makes us feel good and relieves anxiety (Sayette, 2017). Here, we aim to selectively review this experimental literature through a biopsychosocial lens. After briefly discussing biological and psychological research on alcohol’s effects, we present our case for the vital role that social processes play in understanding why we drink. The bulk of the paper illustrates how contributions stemming from social psychological theory and research can advance the understanding of both the rewarding and the negatively reinforcing effects of alcohol. This review will conclude with consideration of areas we believe would profit from additional research.

2. Early Work

Many early examinations of alcohol’s reinforcing properties centered around the impact of alcohol on stress or conflict (see Sayette, 1993). Masserman and Yum (1946) reported that alcohol reduced what was then referred to as experimental conflict neurosis in cats. Subsequently, Conger (1951, 1956) formulated the tension reduction hypothesis (TRH). Within the context of drive reduction theory (Dollard & Miller, 1950), he argued that ethanol consumption would attenuate the intensity of fear or anxiety under particular conditions. Importantly, he also asserted that these negatively reinforcing effects of alcohol consumption would in turn motivate an organism to drink in stressful situations.

The TRH has generated more research than any other psychological theory of alcohol use disorder (AUD) (Leonard & Blane, 1999). The general premise underlying much of the early research testing alcohol’s effects on stress and emotion emphasized the pharmacological properties of ethanol. Some research suggested that alcohol’s anxiolytic effects were peripherally-mediated, perhaps restricted to those cardiovascular functions driven primarily by beta-adrenergic action (Levenson et al., 1987). The bulk of the research investigating the pharmacological effects of ethanol on emotion, however, implicated CNS mediation (Koob & Bloom, 1988; Pohorecky, 1990; Sher, 1987). For instance, acute ethanol administration is thought to dampen responses to fearful stimuli in limbic structures which would normally be engaged by threatening or uncertain social outcomes (Bjork & Gilman, 2014).

By the 1970s, investigators asserting that ethanol directly affects anxiety through the central nervous system had to confront a litany of animal and human studies that failed to provide supporting evidence (see Sayette, 2017). Wilson (1978) noted that “available evidence now clearly shows that there is no simple direct relation between the pharmacologic effect of alcohol and its behavioral consequences.” (p. 317). Even proponents of a biological account were somewhat uncomfortably arriving at this conclusion, noting that these mixed findings are “… perplexing since the rationale underlying the TRT [tension reduction theory] seems so plausible and so consistent with commonplace experience” (Cappell, 1975, p. 202). As with stress relief, lab findings regarding the impact of alcohol on positive emotions have also been inconsistent (Sayette et al., 2016). Some studies reveal that alcohol can induce elation (e.g., Lukas & Mendelson, 1988; Martin et al., 1993), while others have failed to observe emotion-enhancement effects (see Sher, 1987). Accordingly, and as detailed elsewhere (Sayette, 2017), research began to move away from a direct biological analysis of ethanol on emotion to consider a broader biopsychosocial approach to understanding alcohol’s effects. The approach continues to influence research to this day.

Perhaps influenced by Engel’s (1977) biopsychosocial model of disease, the cognitive revolution in psychology (Dember, 1974), and placebo-based effects on emotion (Wilson, 1982), this biopsychosocial perspective began to appear in reviews of the alcohol-emotion literature in the 1980s. Wilson (1982) concluded that alcohol’s effects on emotion were influenced by a range of biological (e.g., dose, physiological responsiveness to alcohol), psychological (e.g., social learning histories, learned expectations about the effects of alcohol), and social factors (e.g., the immediate setting in which drinking occurs): “In view of the influence of all these psychosocial and pharmacological factors on the relationship between alcohol and anxiety, it is simply no longer useful to ask the question: Does alcohol reduce tension? … the more meaningful question to pose is: Under what conditions, at which doses, in whom, on what measures, and how does alcohol affect specific forms of tension?” (p. 124).

A wave of cognitive-based research followed from this emerging biopsychosocial perspective. Instead of targeting direct, CNS-mediated effects on emotion, this work shifted to address the impact of ethanol on cognitive processes, which in turn might, or might not, enhance emotion. Cognitive models began with the recognition that alcohol’s effects were variable and likely depended, as Wilson had suggested, on the circumstances in which drinking occurred. Although these models were not exclusively focused on anxiety relief (see Hull & Van Treuren, 1986; Steele & Josephs, 1990), initial tests concentrated on alcohol’s anxiolytic effects. A comprehensive discussion of these cognitive models is beyond the scope of this paper (see Sayette, 2017). Instead, we briefly consider three models developed in the mid-1980s to early 1990s that continue to be examined.

First, the self-awareness model posits that alcohol intoxication reduces one’s momentary level of self-awareness (Hull, 1981, 1987). Specifically, alcohol is thought to inhibit encoding processes, and thereby serves to attenuate performance-based self-evaluation. In situations when such evaluation is unpleasant, alcohol is proposed to reduce stress responses, thus heightening the probability of drinking. Hull’s (1987) position that alcohol inhibits the processing of information as self-relevant has been tested in studies showing that alcohol interferes with the use of self-focused statements during a speech and that persons who are high in self-consciousness (Fenigstein et al., 1975) are more sensitive to alcohol’s anxiolytic effects. Generally, though, studies testing the self-awareness model have offered mixed support (see Sayette, 2017). As noted elsewhere (Hull & Reilly, 1983; Sayette, 1993), however, it is challenging to assess self-awareness levels without inadvertently probing for self-awareness and influencing the very cognitive processes that one wishes to measure (Yankofsky et al., 1986).

Arguably, the most influential theory among this set of cognitive models has been Steele and Joseph’s (1988) attention-allocation model. As with the self-awareness model, the attention-allocation model posits that alcohol impacts stress indirectly through its impairment of cognitive processing (Josephs & Steele, 1990; Steele et al., 1986; Steele & Josephs, 1988). Alcohol’s purported “narrowing of perception” to immediate stimuli and its attenuation of cognitive abstracting capacity limits attention to the most immediate, salient aspects of experience (see also Taylor & Leonard, 1983). Accordingly, the activity in which an intoxicated drinker engages determines the effects of alcohol. Intoxication during concurrent distraction is thought to weaken stress responding, whereas, without a neutral or pleasantly distracting activity, intoxication is not predicted to relieve stress, and may even increase anxiety by focusing attention on the then-salient stressor.

Tests of the attention-allocation model reveal that alcohol combined with a pleasant distraction diminishes anxiety. Without distraction, however, alcohol did not reduce (see Josephs & Steele, 1990) and even increased anxiety (Steele & Josephs, 1988). Further studies addressed the mechanisms underlying this model (Curtin et al., 2001; Fairbairn & Sayette, 2013; Fleming et al., 2013). Because most drinking that occurs in the real world includes distractions, a major feature of the attention-allocation model is that it offers an explanation for alcohol’s stress-relieving capacities (Josephs & Steele, 1990). Although it is difficult to reconcile Steele and colleagues’ data with those of studies that have found anxiolytic effects in the absence of distraction (see Sayette, 1993), the attention-allocation model nevertheless offers a plausible explanation for both the anxiolytic and anxiogenic effects of alcohol.

Like the self-awareness and attention-allocation models, the appraisal-disruption model emphasizes the direct effects of alcohol on cognition (Sayette, 1993). According to this model, intoxication impairs the cognitive processes associated with the appraisal of new information. Specifically, alcohol is thought to impede the initial perception of stressful information by preventing the activation of associated memories and concepts. This view is consistent with a number of studies that have found alcohol to impair organizational processes (see Saults et al., 2007; Sayette, 1993). The appraisal-disruption model emphasizes the match between the complexity of a stressor and the appraisal processing resources available during encoding. Complex stressors that are difficult to appraise provide optimal conditions for stress relief (Curtin et al., 1998). In addition, when intoxication precedes exposure to a stressor, weakened appraisal may buffer stress by protecting the drinker from fully experiencing the stressor. If the stressor has already been sufficiently appraised, however, subsequent drinking may not attenuate stress—unless, consistent with the attention-allocation model, one were to become distracted by friends or other stimuli and “forget” his or her own discomfort. Subsequent research using a variety of cognitive tasks (e.g., mediated semantic priming task; color-naming Stroop task) probed the cognitive processes thought to underlie the appraisal-disruption model (Sayette, Martin, Perrott, & Wertz, 2001; Sayette, Martin, Perrott, Wertz, et al., 2001). Consistent with the model, alcohol was found to constrain the spread of activation of associated information in memory. The appraisal-disruption model accommodates many of the apparently contradictory findings reported in past investigations (Sayette, 1993). Nevertheless, methodologically it is difficult to conduct research on this model, as studies are hard-pressed to avoid informing participants of a stressor at the study outset. Further, the appraisal-disruption model does not entirely settle the question regarding which types of information are most sensitive to alcohol’s effects (Lang et al., 1999; Sayette, 1993).

While most of the foregoing work has focused on stress relief, some research focusing on cognition and information processing has examined positive reinforcing effects of alcohol. Alcohol expectancy theory posits that alcohol ingestion produces a generalized pharmacological effect that interacts with culturally-derived beliefs, or expectancies, to affect responding (Goldman, Brown, & Christiansen, 1987). From this perspective, individuals who believe that alcohol can transform experiences in a positive manner, enhance pleasure, or bolster sexual experiences, for example, are more likely to feel these effects while consuming alcohol. According to alcohol myopia theory, which subsumes the attention-allocation model described above, by focusing attention on one’s most immediate experiences, alcohol intoxication reduces the salience of information that might otherwise inhibit responding (Steele & Josephs, 1990). Thus, drinking may enable one to act on impulses, whether they be altruistic or antisocial, that prove rewarding, at least in the short term. While the self-awareness model (Hull, 1987) traditionally has focused on reducing negative self-evaluation, such effects often enable an individual to interact more freely and thus derive greater reward from their environment. Building on self-awareness theory and research by Curtin and colleagues investigating the circumstances under which alcohol is most likely to be reinforcing (e.g., Bradford, Shapiro, & Curtin, 2013; Moberg & Curtin, 2009), a social-attribution analysis suggests that alcohol can enhance mood when negative outcomes are perceived to be unstable and/or self-relevant (Fairbairn & Sayette, 2014). This approach proposes that alcohol’s effects within a social context are largely explained by its tendency to free individuals from preoccupation with social rejection, allowing them to access social rewards.

In summary, the self-awareness, attention-allocation, and appraisal-disruption models sought to explain a contradictory set of findings by identifying the circumstances under which the direct pharmacologic effects of alcohol would enhance, or in some cases, worsen emotional responding. Subsequent research has continued to explore moderating factors (e.g., certainty and predictability of a stressor, Bradford et al., 2013; Hefner et al., 2013; Hefner & Curtin, 2012; Moberg & Curtin, 2009). Yet, despite the advances generated by these studies, one is left with a trio of nagging observations: 1) alcohol is the most social of drugs; 2) nearly every study testing the self-awareness, attention-allocation, and appraisal-disruption models recruited social drinking participants who rarely reported consuming alcohol alone (Fairbairn & Sayette, 2014); and yet 3) virtually all examinations of these models relied on paradigms in which participants were alone (though see Fairbairn & Sayette, 2013).

One possible explanation for this focus on isolated drinking paradigms is that because solitary drinking is associated with alcohol problems, it is assumed social drinking is not problematic or etiologically relevant (Skrzynski & Creswell, 2020, 2021). There is, however, evidence that studying alcohol’s effects in a social environment is critical to a holistic, biopsychosocial understanding of the development and maintenance of AUD. Social factors are widely believed to play a key role in early drinking behaviors (Chassin et al., 2013). Adolescents’ social networks, for example, exert an important impact on the initiation of alcohol use (Mundt, 2011), and repeated drinking in early adolescence, a risk factor for later problematic drinking, occurs most often in social settings (Degenhardt et al., 2015). Moreover, as Chassin et al. (2013) notes, there is neuroscientific evidence suggesting that peers make alcohol particularly rewarding for adolescents. As has been detailed elsewhere (Creswell 2021; Fairbairn & Velia, 2019), there is also compelling evidence that social drinking is an important consideration in later AUD etiology. Both solitary and problem drinkers do most of their drinking in social settings (Bourgault & Demers, 1997; Creswell et al., 2015; Single & Wortley, 1993; Skrzynski et al., 2018). Even severe problem drinkers who go on to do much of their drinking in solitary settings, largely start out drinking in social contexts (Warner & White, 2003). Further, social drinking contexts are associated with more positive reinforcement related to drinking (Creswell, 2021) and increased alcohol consumption (Caudill & Marlatt, 1975; Monk et al., 2020). Not to mention, social drinking is associated with numerous risky drinking behaviors that serve as potential markers of AUD, such as binge drinking, drunk driving, violence, aggression, sexual assault, and risky sexual behavior (Bersamin et al., 2012; Bushman & Cooper, 1990; Cleveland et al., 2019; Mair et al., 2015; Park et al., 2008; Smith et al., 1999).

To the degree that laboratory analog studies seek to capture the essential elements of real-world behavior, it is fair to wonder just what has been lost by failing to develop and test models of drinking and AUD in a social context. That is, research that considers the social aspect of a biopsychosocial analysis of alcohol is critical to developing a comprehensive understanding of why alcohol can be so alluring. We now turn to work primarily published in the past decade that has sought to draw on research and theory derived from experimental social psychology, and in particular small groups research, to begin to examine the impact of alcohol on emotion in a social context.

3.0. Alcohol in a Social Context

Only the rare experiment has examined the social context which surrounds drinking (see Fairbairn & Sayette, 2014). As noted above, this omission is striking when one considers that alcohol is perhaps the most social of drugs, typically consumed in social situations, among both light (Cahalan et al., 1969; Demers et al., 2002; Single & Wortley, 1993) and problem (Bourgault & Demers, 1997; Cahalan et al., 1969) drinkers, frequently for its presumed social enhancement effects (Brown et al., 1980; Cooper et al., 1992). It, therefore, is hard to imagine that examining the motivations underlying alcohol consumption in almost exclusively isolated laboratory contexts offers the full scope of alcohol’s reinforcing properties. As Engel (1977) observed, humans exist in social contexts, and consideration of such contexts is critical to describing the full scope of disease. The field’s focus on individual responses to alcohol has limited its ability to define the complex, dynamic, and reciprocally-influencing processes that motivate drinking in natural settings. In the following discussion, we contemplate how this focus on individual processes (including past studies conducted in our lab) has offered an incomplete understanding of alcohol’s effects on emotion. We then review experimental research that has considered social context in alcohol response and discuss how this work has begun to fill in some of these gaps. Finally, we comment on what remains under-examined and suggest ways for the field to move forward to address limitations in our knowledge of alcohol reward.

The alcohol field’s focus on individual states has limited its ability to generalize findings to the settings in which drinking typically occurs. In a natural drinking environment, however, many factors may influence emotional responses, which an isolated laboratory is ill-suited to simulate: noise and smell, feelings of bonding or group identity, conversation content, and the emotions of others, to name a few. Indeed, the countless contextual factors that surround drinking likely interact to affect a complex, multiply-determined response to alcohol. Without considering these effects, it is hard to know how well laboratory findings generalize to the experience of drinking in natural settings. Extending the theories that emerged from tightly controlled laboratory studies to the social environment is, therefore, a critical next step for understanding alcohol reward. As an illustration, Curtin and colleagues have published an impressive program of work that demonstrates selective alcohol effects on responses to uncertain threats (Bradford et al., 2013,2022; Hefner et al., 2013; Hefner & Curtin, 2012; Moberg & Curtin, 2009). While these researchers have considered the boundary conditions of this effect (e.g., Bradford et al., 2022; Hefner et al., 2013; Hefner & Curtin, 2012), it would be informative to understand how these findings apply to the social stressors that drinkers typically experience. The social environment is inherently unpredictable, and so it is plausible that alcohol may globally dampen stress in social drinking settings. On the other hand, some social stressors might be perceived as more imminent and unambiguous (e.g., some perceptions of ostracism). Would alcohol fail to produce its prized anxiolytic effects in the face of such stressors? Investigating factors such as predictability and certainty to better understand alcohol’s rewarding effects in social settings would benefit the field.

In addition to limiting generalizability, prior laboratory studies relying on individual-based paradigms necessarily ignore critical aspects of human motivation. Acute individual changes in affect are surely not the only experience that reinforces human behavior. In fact, collective emotions, described by Goldenberg et al. (2020) as “macrolevel phenomena that emerge from emotional dynamics among individuals who are responding to the same situation” (p. 2), tend to be stronger in magnitude and longer-lasting than individual emotions (Goldenberg et al., 2020). It therefore seems plausible that such group experiences are even more motivationally salient than individual affective responses. Further, sharing an emotion with another person may be rewarding irrespective of the valance of that emotion. Research on shared reality, the experience of having common inner states with others, reveals that such collective awareness is of great value in satisfying fundamental human motivations (Echterhoff et al., 2009; Hardin & Higgins, 1996). Consider, for example, a group of friends who become angry at the same unjust situation, or family members who share in a moment of grief, or perhaps most profoundly veterans who shared the horrors of battle. These experiences, though on an individual level emotionally painful, may also generate a powerful and lasting sense of connection and validation.

The foregoing discussion is influenced by research demonstrating that forming social bonds is a fundamental human motivation (Baumeister & Leary, 1995). Further, people consistently report that alcohol consumption facilitates and enhances social experience (Brown et al., 1980; Cooper et al., 1992). This combination of research on social motivation and alcohol expectancies makes clear the need for alcohol administration research to integrate social context into its experimental paradigms. Human motivation is surely more than individual reports of being happy or sad; it is emboldened by the laughter at your joke, excited by the flirtatious glance from your crush, confident because of the ease at which you carry a conversation, self-assured because of the compliment you received, amused by the bartender’s mint-less attempt at a mojito, hopeful because you found someone that shares your distress over Tuesday’s presentation, nostalgic because the smell, the sound, the warmth, the buzz of it all reminds you of the time you went out on your 21^st birthday with friends you haven’t seen in ages. To understand why humans drink and to have at our disposal the full range of processes subject to intervention, we ought to target these complex and powerful social motivators.

A number of studies have examined social processes using trained confederates to structure the interaction (Abrams & Wilson, 1979; de Boer et al., 1994; Keane & Lisman, 1980; Wilson & Abrams, 1977). These studies have steered away from testing unscripted social experience, as their goals were better served with the control gained from precise experimental manipulations. It also is challenging to conduct unstructured group research (Fairbairn & Sayette, 2014). Such studies can be unwieldy, as the spontaneous nature of the social interaction in the absence of a trained confederate creates an unpredictable experimental setting. In addition, larger samples are typically needed to address power requirements linked to the interdependence of participants’ emotional experience within a group. Further, the dynamic nature of the social experience can make reliance on traditional self-report assessments of mood during or following an exchange intrusive and less sensitive to subtle, fleeting changes in emotion that might be captured, for example, using a multimodal approach (e.g., micro-analytic coding of facial expressions, Sayette et al., 2012).

Unfortunately, despite the advantage of tight experimental control, confederated-based studies have generated inconsistent findings (see Fairbairn & Sayette, 2014). Some of these inconsistencies may be due to methodological concerns, including arbitrarily varying the size of groups within a single study, little attention to the level of acquaintanceship amongst group members, and lack of adequate beverage control conditions to fully parse pharmacological and dosage-set effects of alcohol (Sayette et al., 2012). Of note, alcohol’s reinforcing properties have been most consistently detected during unscripted social interactions (Fairbairn & Sayette, 2014).

3.1. Early Unscripted Research

Fifty years ago, a simple study was conducted that should have exerted a greater impact on the field. Pliner and Cappell (1974) examined the effects of alcohol on self-reported emotion and observer-coded amusement in both social and solitary drinking situations. Participants who drank in the social condition experienced primarily rewarding emotional responses to alcohol (e.g., euphoria), while those who drank alone reported primarily physiological changes (e.g., dizziness). Subsequent studies have similarly underscored the need to consider social context, either by demonstrating selective reinforcement in social drinking settings (del Porto & Masur, 1984; Doty & de Wit, 1995; Kirkpatrick & De Wit, 2013), or by observing emotional enhancement in social situations using uniquely social measures, such as speech behavior and smile coordination (Kirchner et al., 2006; Samson & Fromme, 1984).

3.2. The Group Formation Project

In response to this early social research conducted in controlled, scripted settings, we developed an unscripted group formation paradigm to address some of the methodological limitations of previous work (Sayette et al., 2012). We tested the impact of alcohol on initial group formation among 240 groups of three verified strangers. Beverage condition was explicitly manipulated, with groups assigned to each of three beverage conditions: alcohol, placebo, and no-alcohol control. Groups were seated around a table while they drank and interacted freely. These interactions were recorded, allowing for unobtrusive measures of affect throughout the drinking period. Participants’ affect-related facial expressions were then coded across millions of frames of video using Ekman et al’s (2002) Facial Action Coding System. In addition to a self-report measure of social bonding, we assessed the coordination of group members’ expressive behavior. Other real-time measures of group bonding included content-free speech behavior (Sayette et al., 2012) and group volume (Fairbairn, Sayette, Amole, et al., 2015). Analyses of this dataset provided broad, multimodal evidence that a moderate dose of alcohol enhances affect (both decreasing negative affect and increasing positive affect). Importantly, alcohol also elevated group measures of social bonding, including increased shared smiles (while controlling for individual smile behavior), reduced silences, and increased sequential speech behavior (Sayette et al., 2012).

While it is reassuring to find experimental corroboration of the widely presumed notion that alcohol enhances emotion, results from this group formation project are not startling. Any bartender worth their salt would agree that alcohol has the power to make drinkers happier and facilitate social interactions. Yet in the context of a large alcohol response literature, which has not reliably observed such effects, it is notable that we found consistent, multimodal evidence of alcohol-induced emotional enhancement. Many of these rewarding effects, which may offer clues for understanding the etiology of AUD (e.g., shared smiling, group bonding, coordinated speaking), are impossible to observe in solitary settings. Perhaps most importantly, the initial findings summarized in (Sayette et al., 2012) offered critical validation for this paradigm and supported its utility in answering questions that might be less obvious to the local bartender.

Lab studies that illuminate why, when, and for whom alcohol is rewarding are crucial to understanding problematic drinking (King et al., 2014; Schuckit et al., 2011), and may inform intervention work. However, previous laboratory work examining the mediators and moderators of alcohol reward has largely tested these effects in isolated contexts (e.g., Sayette et al., 1994) with affect explicitly manipulated, raising questions about the generalizability of findings to the typical drinking experience, and further precluding the investigation of mediators and moderators that exist only within the social context. The group formation project permitted examination of social variables that both mediate and moderate alcohol reward. These findings have been discussed elsewhere, and here we mention just a few of these observations.

First, regarding mediators, emotional inertia may underlie the capacity of alcohol to improve affective states (Fairbairn & Sayette, 2013). Second, alcohol’s effects on emotional contagion (Fairbairn, Sayette, Aalen, et al., 2015) may help to explain the differential impact of alcohol on emotion in male and female participants. For instance, alcohol selectively enhanced the probability that a male participant would “catch a smile” in a social-drinking context and the data suggest that the social rewards associated with alcohol consumption are especially pronounced for male drinkers (Fairbairn, Sayette, Aalen, et al., 2015). There appears to be a particular reward associated with enjoying a moment of shared joy. In addition, several analyses focused on moderators of alcohol’s emotional effects, including personality traits thought to be linked to AUD. Specifically, those high in extraversion gained particular reward from alcohol in a social context (Fairbairn, Sayette, Wright, et al., 2015). Fourth, the group formation paradigm permitted examination of the effects of alcohol on social discomfort in groups containing both white and black participants versus those containing only white participants (Fairbairn et al., 2013). This work allows for a better understanding of alcohol’s impact on intergroup relations and, importantly, highlights the role of self-presentational concerns. In summary, this unstructured group drinking environment offered a platform to evaluate the complex relationship between alcohol and emotion in a social context. These studies suggest that certain individuals are particularly sensitive to alcohol’s socially reinforcing effects, and that such effects are possibly mediated through a variety of social processes.

3.3. Recent Social Psychological Research

In the past decade, researchers have continued to elaborate on the effects of alcohol in social settings and thus have offered a more comprehensive, biopsychosocial understanding of alcohol reward. One set of studies has relied on traditional measures of alcohol response in explicitly social, and oftentimes unscripted, settings. This work has increased the generalizability of results related to individual measures of alcohol response and identified contextual moderator variables, some of which are only observable in a social setting. Another set of studies has offered creative new experimental paradigms for studying alcohol’s effects and assessed novel responses. In the following paragraphs, we review in more detail the contributions of this emerging literature.

3.3.1. Studies in Social Contexts Implementing Traditional Measures of Alcohol Response

A number of recent studies have addressed the limitations of early work in several notable ways. Corbin et al. (2015) contrasted alcohol responses in both a group lab and a more ecologically-valid group bar setting. Low-arousal positive effects of alcohol (e.g., feeling relaxed), relative to placebo, were more pronounced in the less stimulating lab drinking environment (a similar pattern also emerged for the high arousal positive effects of alcohol, assessed during the ascending limb of the BAC curve). Building on this finding, Corbin et al. (2021) had participants consume either an alcohol or placebo beverage in one of four possible drinking contexts: solitary lab, group lab, solitary bar, and group bar. Results again indicated that the positive effects of alcohol were more influenced by context than were the negative effects of alcohol. For example, the high arousal positive effects of alcohol (e.g., lively) were stronger in nonbar contexts and alcohol’s low arousal positive effects were only observable in the group nonbar context. Conversely, the negative effects of alcohol (e.g., aggressive, woozy) were minimally impacted by context. While in part these findings may reflect the power of a placebo comparison condition to exert rewarding effects in the bar setting, they nevertheless illustrate that when considering alcohol’s effects, context matters.

Other recent investigations have considered a broader variety of contextual features. Kirkpatrick and De Wit (2013), for example, examined the effects of alcohol vs. placebo in both unacquainted dyads and isolated contexts. Within the dyadic interactions, the number of people consuming alcohol varied systematically, such that participants were exposed to each of four possible beverage combinations on separate days: (1) both members drank alcohol, (2) both drank placebo, (3) and (4) one participant drank alcohol while the other drank placebo. Though the pattern of results was complex, drinking in a social context did appear to increase certain positive subjective effects of alcohol (i.e., stimulation), especially when the partner also received alcohol. Further, having an intoxicated partner increased other subjective pleasant effects (e.g., high and social), regardless of the participant’s beverage condition. Research testing dyads composed of platonic friends has similarly found alcohol to enhance positive mood (Bowdring & Sayette, 2021).

Fairbairn et al. (2018) used mixed laboratory and ambulatory assessment methods to survey the effects of alcohol on emotion in social settings, and further to test the moderating impact of social familiarity on alcohol response in participants’ natural drinking environments. Results from the laboratory arm of the study add to the growing literature that demonstrates consistent alcohol mood enhancement in social contexts; alcohol, administered in a group setting, decreased negative mood and increased positive mood compared to the control beverage (Fairbairn et al., 2018). Notably, data from the ambulatory arm of the study revealed similar alcohol emotional enhancement in social contexts, but not in isolated contexts. Further, data collected over the ambulatory assessment period indicated that social familiarity moderated the impact of alcohol on mood, with alcohol increasing positive affect and decreasing negative affect more in situations with more unfamiliar company (Fairbairn et al., 2018). In the lab setting, however, we have observed significant mood-enhancing effects of alcohol among familiar dyads (Bowdring & Sayette, 2021), suggesting the need for additional investigations into the relationship between social familiarity and alcohol reward.

Kang et al. (2018) examined the impact of alcohol, administered in groups, on subsequent stress responding while participants were either alone (separated from the group they drank with) or with group members. Alcohol, administered in this group setting, reduced stress responding overall. The attenuating effect of alcohol on startle response, however, was greater in the alone condition compared to the group condition. This finding diverges from past research that demonstrated more rewarding effects of alcohol on emotion in social contexts (del Porto & Masur, 1984; Doty & de Wit, 1995; Pliner & Cappell, 1974). Unlike most previous research, however, this study examined the effect of alcohol on physiological stress reduction, as opposed to subjective emotional experience (e.g., enhancement of positive emotion), suggesting that the influence of the social environment may vary between divergent emotional responses. [Although not necessarily in a social setting, we refer readers interested in the impact of alcohol on both subjective and physiological stress reduction to Sayette (1993), Bradford et al. (2013), and Bradford et al. (2022)].

While much of this article has emphasized the reinforcing effects of alcohol in social contexts, it is important to note that social factors also contribute to understanding the impact of alcohol on antisocial behaviors, including risky sexual or driving behavior and aggression (though such behavior also can be reinforcing, see Fairbairn & Sayette, 2014). Indeed, several models described above also have addressed alcohol-related aggression and/or risky behavior, including the appraisal disruption model (Ito, Miller, & Pollock, 1996), alcohol myopia (Giancola et al., 2010), self-awareness (Hull & Van Treuren, 1986), and alcohol expectancy (Goldman et al., 1987). Research illustrates the importance of social contextual factors (e.g., provocation by another person) in understanding the impact of alcohol on aggression (Giancola et al., 2002; Taylor & Leonard, 1983). Other research has focused on social mechanisms (e.g., social information processing) to examine the impact of alcohol on aggressive and coercive behaviors (Farris, Treat, & Viken, 2010; Sayette, Wilson, & Elias, 1994).

In sum, a number of recent investigations have leveraged social alcohol administration paradigms to examine the effects of alcohol on a variety of individual emotional responses. These studies have revealed that alcohol, consumed in social settings, has reinforcing effects on emotions, a result which contrasts with the mixed findings from studies testing participants in isolation. A novel contribution of this work is the identification of specific socio-contextual features that may further impact alcohol response, including physical setting (Corbin et al., 2015, 2021), partner intoxication status (Kirkpatrick & De Wit, 2013), and social familiarity (Fairbairn et al., 2018). Some of this new research also suggests that contextual features may have divergent effects across different alcohol responses (Kang et al., 2018). This emerging literature reinforces the importance of considering a host of ecologically valid settings in future work to better understand when, where, and for whom alcohol will prove reinforcing.

3.3.2. Studies Examining New Domains of Alcohol Response

Another set of contemporary investigations has developed novel experimental paradigms to assess potentially rewarding experiences apart from individual self-reports of affect, including group bonding processes, interactions in couples, humor enjoyment, responses to moments of ostracism, and eye gaze in virtual interactions.

Group Bonding.

Since our group formation bonding data were published (Sayette et al., 2012), a few recent investigations have used creative approaches to examine the impact of alcohol, or alcohol cues, on group bonding and cohesion processes. Moltisanti et al. (2013) tested the impact of alcohol priming (using alcohol words or neutral beverage words in a lexical decision task) on group bonding in a collaborative group task. Amongst participants who were primed with alcohol words, those with higher social alcohol expectancies reported more task-related group cohesion than did those with lower expectancies, whereas group bonding did not vary by expectancy in the neutral prime condition. Apparently, among individuals with high social alcohol expectancies, merely exposure to alcohol cues, without even expecting or drinking alcohol, can lead to more rewarding group experiences. Examining bonding from a more literal perspective, a recent study evaluated the impact of alcohol on physical distancing (Gurrieri et al., 2021). Alcohol led unacquainted dyads to draw physically closer to each other during a social exchange. Alcohol did not, however, affect physical distancing among familiar partners. These findings suggest that alcohol may serve “to overcome a natural caution people feel towards strangers” (Gurrieri et al., 2021, p. 1). Similarly, research has also found alcohol to increase the amount of time dyads choose to engage in an interaction (Kirkpatrick & De Wit, 2013). These findings align with those of our group administration study, indicating that alcohol, across a wide range of measures, is associated with increased group bonding and cohesion, especially among strangers (Sayette et al., 2012).

Interaction Behaviors and Quality in Couples.

Alcohol’s social-enhancement effects also can extend to relationship functioning in romantic couples (Homish & Leonard, 2005). In the laboratory, Testa et al. (2014) examined the impact of alcohol on interactional behaviors in heterosexual couples. Specifically, couples (in one of four beverage conditions: both alcohol, both no alcohol, alcohol in the male, alcohol in the female) engaged in conflict resolution paradigms that were coded by trained observers for negative (e.g., psychological abuse, withdrawal) and positive (e.g., acceptance, humor) behaviors in the interaction. When a participant consumed alcohol, they increased their positivity and decreased their negativity, particularly when their partner was sober. Further analyses revealed that alcohol’s rewarding effects during these interactions were pronounced in couples with low relationship quality, which in turn could underly their increased risk for problematic drinking (Fairbairn & Testa, 2017). A recent EMA study also reinforces the idea that alcohol has positive effects on couple interactions: in a community sample of couples, simultaneous drinking by both partners increased the likelihood of couple intimacy in the subsequent hours (Testa et al., 2019). Notably, however, the results reported by Testa et al. (2014), that alcohol’s enhancing effects were most pronounced when one partner was sober, contrast with this real-world finding that congruent, but not solo, drinking in couples leads to alcohol reward. Further, Testa et al’s (2014) findings, diverge somewhat from those reported by Kirkpatrick and De Wit (2013), in which some of alcohol’s subjective stimulating effects were enhanced in unacquainted dyads particularly when both members were consuming alcohol. As discussed below, these inconsistencies underline the need for more research aimed at understanding alcohol’s effects in groups with varied intoxication statuses and across degrees of social integration.

Humor Enjoyment.

Humor enjoyment is one of the leading perceived benefits of drinking (Orford et al., 2004). While experimental research examining the effects of alcohol on humor enjoyment is not new (e.g., Ruch, 1994), in recent years it has become a topic of study within a social context. We recently evaluated the impact of alcohol on humor enjoyment while trios of strangers listened to a comedy routine (Sayette et al., 2019). Alcohol elevated positive affect across both self-report and behavior-expressive (facial expression analysis) measures. A subsequent analysis revealed that alcohol selectively increased the likelihood of smiling specifically during the non-punchline moments (see Fairbairn et al., 2020). More research is needed to continue to address the impact of alcohol on humor perception.

Responses to Ostracism.

An additional social domain that has been a focus of recent work concerns the impact of alcohol on response to being socially rejected. Recently, across three studies, Fairbairn et al. (2021) had groups consume an alcoholic or non-alcoholic beverage (in two of the studies this was a no-alcohol control beverage, in one it was a placebo beverage) and then complete a Cyberball ostracism task (Williams & Jarvis, 2006) in which participants were either included or excluded from most ball tosses. Interestingly, alcohol did not enhance subjective experience following social exclusion. Outside the lab, however, Hales et al. (2015) exposed participants to a Cyberball ostracism manipulation and found that subjective intoxication levels were associated with less distress following ostracism. Furthermore, it appears that being ostracized also may motivate people to drink alcohol (Bacon & Engerman, 2018).

Virtual Interactions.

Finally, though this work is in its infancy, the COVID-19 pandemic has increased the urgency of understanding the impact of alcohol during virtual interactions. One recent study used eye-tracking methods to evaluate the time that participants spent during an interaction gazing at themselves or a partner (Ariss et al., in press). Results indicated that alcohol increased self-directed attention, failing to demonstrate its typically potent social-affective enhancement in this virtual context. Given its increasing popularity, more work is needed to fully describe alcohol’s effects in virtual domains.

In sum, alcohol’s rewarding effects seem most evident in social settings. In addition to group studies revealing effects of alcohol on emotion, an emerging literature is testing the effects of alcohol on other socially motivating experiences. Largely, this literature illustrates that alcohol is associated with enhanced social experience, including increased group bonding, humor enjoyment, and couple positivity. It is plausible that these responses, and many more yet-to-be-examined social experiences, motivate alcohol use.

4.0. Conclusions and Future Directions

This review highlighted alcohol administration studies that have incorporated social contexts into their investigations. A modest but growing literature has demonstrated that alcohol enhances emotion in these social contexts and has begun to identify specific contextual variables that moderate these social effects. Another small group of studies has examined alcohol’s impact on socially motivating experiences using a broad set of responses that extend beyond an individual’s self-reported mood. While these innovative studies have advanced understanding of alcohol response, there are many stones yet left unturned. We conclude this review with nine proposed research directions aimed at more fully integrating social processes into our biopsychosocial analyses of alcohol.

1. Most social alcohol administration studies have focused on neutral, non-threatening interactions among strangers and in romantic couples. While this aligns with many natural drinking contexts, there are other types of interactions worth considering. More research is indicated, for example, to examine alcohol’s effects among groups varying in familiarity (Bowdring & Sayette, 2021; Fairbairn, 2017), as well as in more conflicted situations (Sayette et al., 2004).

2. There is also continued need to examine alcohol response in more diverse social groups. For instance, the social alcohol studies reviewed here (including our own work) have either focused on heterosexual couples or did not assess sexual orientation, used dichotomous categorizations of gender, and centered on cis-gender individuals. Future research would do well to incorporate more inclusive representations of partnership and gender. In addition, the studies reviewed here have recruited primarily white participants and have not focused on the relationship between alcohol, the social environment, and social identity. Continued work in this realm is necessary for a complete biopsychosocial understanding of alcohol’s effects.

3. Varying levels of intoxication among group members are common outside the laboratory (e.g., a group of drinkers that includes a sober designated driver), and research with dyads has demonstrated that the intoxication status of a drinking partner can influence alcohol-related effects (Kirkpatrick & De Wit, 2013). To our knowledge, however, mismatched beverage conditions have not been tested in studies with groups of more than two people. Systematically varying alcohol concentrations within groups may address important questions that thus far have been neglected in the experimental literature.

4. Most alcohol administration studies meticulously control alcohol dosing. While this approach allows for precise manipulation of alcohol levels, it differs from typical drinking behavior. Studies that allow alcohol consumption to vary tend to designate BAC or the amount of alcohol consumed as the dependent variable (e.g., taste test research, see Marlatt et al., 1973). Studies should also consider creative, hybrid designs, in which socio-emotional responses continue to serve as dependent variables, while also permitting participants to modulate their drinking pace and volume, in order to approximate real-world settings where the amount of beverage consumed is determined by the drinker. While processes of reciprocal influence between alcohol consumption and emotion would make such research challenging, ultimately it may offer a more ecologically valid approach to understanding the effects of alcohol.

5. With few exceptions (Kirkpatrick & De Wit, 2013), most social lab-based alcohol studies have tightly restricted interaction time frames. For instance, in our research, participants must remain seated around a table throughout the entire 36-minute drinking period (Sayette et al., 2012). Such constraints diverge from most real-world drinking experiences, in which people are free to come and go as they wish. The worry that one’s partner (or a member of one’s group) may abruptly exit the interaction for greener pastures is off the table. So, too, “captive” participants in existing paradigms may be motivated to engage in friendly conversation in part because they know they are stuck with each other. Kirkpatrick and De Wit (2013) offered a clever alternative to this approach by placing dyads in a room together and allowing them to freely engage in a variety of activities, including but not limited to interacting with each other. More studies that permit participants to engage in interactions at their will would necessarily be unwieldy, but if done properly, have potential to yield valuable data.

6. As noted above, there has been a recent push to expand the range of emotional and motivational responses captured in alcohol studies. Outside the addiction field, measures of nonverbal behavior such as proximity, posture, and thin-slice judgments have yielded insights regarding interaction dynamics (e.g., Ambady & Rosenthal, 1992; Bernieri et al., 1996; LaFrance & Broadbent, 1976). Further, while verbal content has been assessed during individual speeches (e.g., Hull et al., 1983; Sayette, 1994), to our knowledge, no study has yet tested alcohol’s impact on natural speech content in a group interaction. We recently have begun to undertake such content analysis in a group study to investigate how alcohol affects the words we utter, the topics we discuss, and more generally, the manner in which we express ourselves. Understanding the dynamic interplay between these instantaneous responses may additionally illuminate the social mechanisms that underline alcohol reward.

7. Research that evaluates the effects of alcohol on socially motivating experiences remains in its infancy. We are optimistic, however, that soon the alcohol literature will include novel approaches to studying the impact of alcohol on a variety of socio-motivational constructs, such as affiliation, status, and mate seeking and retention (Neel et al., 2016). As one example, social psychological investigations reveal that favoring members of one’s own group can be rewarding, due to increased self-esteem and in-group identity (Lemyre & Smith, 1985; Oakes & Turner, 1980). Testing the influence of alcohol on such intergroup relations may elucidate group-level processes that motivate drinking, while also identifying social mechanisms underlying some of alcohol’s hazardous effects (e.g., violence).

8. Research has yet to identify emergent group experiences, such as initial group rapport at the outset of an interaction, which may serve to moderate alcohol’s subsequent effects on emotion. Such research would add to a nascent literature examining social contextual moderators (in contrast to the more typically assessed, individual-level moderators) of alcohol response (Sayette et al., 2016).

9. Finally, a complete biopsychosocial understanding of alcohol response necessitates integration of research across these three broad domains. More work is needed to incorporate social stimuli into biologically-based paradigms and methods. Alcohol researchers would do well to adapt recent social neuroscientific paradigms (e.g., exposing participants to social stimuli; for example Collier & Meyer, 2020) to their investigations. Further integration might expand to include virtual reality and augmented reality methods as well (Sayette & Goodwin, 2020).

In this article, we have advocated for continued investigation into alcohol’s effects within a variety of social settings and suggested several directions for future work. We recognize, however, that these recommendations raise methodological challenges. Unscripted social interactions are inherently unpredictable and introduce substantial variability into experimental paradigms, making causal inference difficult. Group members also influence each other, leading to covariance that needs to be statistically addressed. Additionally, examination of alcohol’s social effects requires consideration of both individual responses (e.g., whether a participant smiles) and group-level phenomena (e.g., coordinated smiling among group members). These variables highlight the value of including dynamic, real-time measures of emotional responding, which can be extremely labor intensive to code and analyze. Moreover, the most ecologically valid paradigms may also be challenging to implement both practically and ethically (e.g., most IRBs would be skeptical of a study that allows participants to drink with no restrictions). All of these considerations may discourage laboratory researchers from pursuing social alcohol administration work. Yet, we argue that these complexities also make a compelling case for conducting controlled group-level experimental work that is well suited to causal inference. We believe such work offers an important complement to the relatively large body of studies (including some from our own lab) that have examined the effects of alcohol on individuals. Toward this aim, we offer several recommendations for addressing methodological challenges in order to harness some of the spontaneous and perhaps unwieldy data that stem from unscripted group-level research.

First, multilevel modeling is useful to examine individual and group effects, while also statistically accounting for the interdependence of group members (see Sayette et al., 2012). Second, large sample sizes are required to adequately power such analyses and can be important in detecting effects in inherently noisy data. Third, while causal inference can be challenging in less tightly controlled social paradigms, tools such as survival analysis (see Fairbairn et al., 2015) can probe mediational pathways in a fashion that accounts for the passing of time. Fourth, while coding of real-time behavior is labor intensive, thin-slice techniques that require assessment of only a small sliver of the behavioral stream can save time at little cost to accuracy (see Ambady et al., 2000 for review). Fifth, innovative data collection methods such as ecological momentary assessment (e.g., Fairbairn et al., 2018) and field experiments with traveling research labs (e.g., Hales et al., 2015) make the collection of real-world data possible while also avoiding many of the practical constraints placed on laboratory work.

In sum, after nearly a century of research aimed at understanding why we drink, alcohol’s prized effects remain elusive. Most alcohol administration studies have focused on its impact on individual emotional experience assessed in isolated experimental settings. While important information has been gleaned from this approach, the field would benefit from further investigating social processes and integrating such research more fully into a comprehensive, biopsychosocial analysis of alcohol.