An Evolutionary Approach to Binge Drinking Impression Formation: A Cross-Cultural Comparison Between France and Peru

Social Theories of Binge Drinking

Over the past two decades, various sociopsychological theories have intended to propose a framework for generating testable and predictable hypotheses. Cognitive theories, for instance, have proposed to analyze alcohol expectancies and drinking refusal self-efficacy in order to explain four distinct types of drinking styles (normal/social drinkers, binge drinkers, regular heavy drinkers, and problem drinkers or alcoholics; see Oei & Morawska, 2004). Attitudinal theories usually refer to Ajzen’s model (attitudes, social normal, and perception of control of binge drinking, see Ajzen, 1985) of planned behavior (explaining up to 58–69% in binge drinking intentions; see Norman et al. 2007; Johnston & White, 2003). Social bond theory (Durkin et al., 1999) and self-control theory (Gibson et al., 2004) proposed to integrate beliefs, peer behavior, and environmental factors to explain and predict binge drinking frequency.

These theories present clear interests as they try to conceptualize binge drinking through attitudinal, motivational, and situational characteristics. However, they somehow failed to clearly distinguish between solitary and collegial binge drinking. While the former is more relevant through the lens of alcoholism, the latter appears to have an important social component, which may be to express competition, masculinity, or entertainment. To tackle the deep-rooted behaviors that constitute binge drinking, a possible strategy is to go beyond proximal explications and propose an existing evolutionary framework that spans the different physical and ecological factors involved in that particular type of risk (Hill & Chow, 2002).

Costly Signal Theory

According to costly signal theory (Zahavi, 1975; Zahavi & Zahavi, 1999), only men who could afford those type of signals would engage in such behaviors. Originally rarely used in evolutionary biology, this theory gained popularity after Grafen (1990) found a pair of evolutionary stable strategies that are a function of a male’s true genetic quality, his level of advertising, and the observed value of the male by the female (which depends largely on the perceived value of the male). Any male peacock (Loyau et al., 2005) capable of maintaining and displaying such characters indicate a high genetic quality, yet they also expose the owner to increased possible aggressions (i.e., it may be more difficult to fight or to escape because of the physiological cost of the feature, see Mateos & Carranza, 1997).

Among the myriad of risks at play in human interactions, binge drinking is a feature that fits very well with Grafen’s criteria in the Zahavian framework. Indeed, social event generally allows for a public test of an individual’s tolerance, which may be explicitly written in some drinking games’ rules (level of advertising). On the other hand, the binge drinker can hardly lie about his or her digestive capacities and would generally give up once his or her limits are reached (observed value). Finally, how the individual tolerates the alcohol is a direct signal of true genetic quality and his or her capacity to tolerate the toxic effects of alcohol abuse on the body (true genetic quality). Indeed, tolerance to alcohol is understood from an evolutionary point of view as a context-dependent trait that may have evolved through intrasexual competition, is often exaggerated, and explains human cognitive sex differences notably in gastric alcohol dehydrogenase (Geary, 2017). Thus, tolerating the toxic effects of alcohol could be understood through the lens of costly signal theory as it informs about the ancestral true genetic quality of the individual.

To tackle the binge drinking phenomenon, we attempt to evaluate how the different risks displayed in equivalent binge drinking videos would impact the impression formations of the binge drinker. We predict that the video describing sexual impotence should be the most detrimental in the evaluation of the binge drinker in young adults in the middle of the sexual competition (18–35 years old; see Kanny et al., 2018). Despite Zahavian hypothesis being primarily applied to intersexual selection (and so should work better with a female sample), we induce that male participants would also be affected when evaluating the competitiveness of the masculine target.

Short- Versus Long-Term Risks

The time line of the risks displayed in our videos made our theory compete with a more proximal behavioral economical one. Adolescents have been shown to be more willing to gamble and take risks when exposed to ambiguous economic risks, while actually taking fewer risks than adults when the odds at play were clearly exposed (Tymula et al., 2012). Since Tymula and colleagues (2012) also found a positive correlation with age (as a continuous predictor), we think that our relatively young sample (24–25 years old) could still be impacted by short-term messages when compared to long-term ones. As a consequence, a parallel hypothesis may postulate that the sexual and cognitive deficits spots will be more impactful globally than the long-term video.

We ran two studies to test our predictions in a between-subject design. First, we ran the study with French participants between 18 and 35 years old through social networking sites (i.e., mainly Facebook) in a mostly occidental individualist society. Then, we tried to replicate those findings in an online sample of Peruvians to test whether those results hold in a very different context (nonoccidental collectivist). Taken together, those findings should be used as a first step toward understanding formation impression during binge drinking episodes.

Situation Embodiments (All) and Intrasexual Competition (Men Only)

We systematically applied contrasts to compare the effect of the sexual video to the three other conditions (cognitive vs. long term vs. control) while controlling for the sexual versus cognitive and the sexual versus control comparisons (Brauer & McClelland, 2005) and excluded any outliers. ² Neither men nor women participants were less willing to be in Nicolas’s state after the sexual videos (M = 4.27, SD = 1.82 vs. M = 4.32, SD = 1.86) when compared to the other videos as revealed by a one-way analyses of variances (ANOVAs; p = .79 and .89, respectively). Similarly, the intrasexual competition measurement (combining the will to be in Nicolas’s team as well as how much participants would be willing to play against him in a sport competition) did not reveal any difference after being exposed to the sexual message (M = 3.96, SD = 0.80) when compared to the other conditions (M = 3.89, SD = 0.49, p = .91, see Table 1).

OPEN IN VIEWER

Table 1.

Means (M), and Standard Deviations (SD), of the Different Dependent Variables Across the Different Conditions in Study 1 for the First Hypothesis (Sexual Condition vs. Others).

Measures/Conditions	Sexual	Cognitive	Long-Term Diseases	Control
	M (SD)	M (SD)	M (SD)	M (SD)
Willingness to be in the target’s situation (men)	4.32 (1.86)	4.25 (1.77)	4.40 (1.76)	4.21 (1.96)
Intrasexual competition (men)	3.89 (0.49)	3.66 (0.70)	3.93 (0.56)	4.19 (0.94)
Willingness to be in the target’s situation (women)	2.82 (1.68)	2.21 (1.12)	2.62 (1.56)	2.67 (1.41)
Target’s reproductive success (women)	4.50 (1.26)	3.57 (1.70)	4.43 (1.63)	3.70 (1.90)
Target’s attractiveness (women)	2.32 (1.49)	3.21 (2.08)	3.29 (1.35)	2.30 (1.35)

Reproductive Success and Attractiveness (Women Only)

Whereas women did not evaluate differently the reproductive success of the target (p = .24), we observed a statistical trend with Nicolas being evaluated as marginally less attractive after viewing the sexual spot (M = 2.32, SD = 1.49) when compared to participants exposed to the cognitive, long-term, and control conditions (M = 2.73, SD = 1.56), t(96) = 1.79, p = .08, 95% CI [−.01, .19], η a 2 = .06 (see Table 1). There was no interaction between the experimental condition and the actual alcohol consumption of the participant (i.e., 0 = none to less than one drink, 1 = one drink, 2 = two drinks, 3 = three drinks, 4 = four drinks, 5 = five drinks or more, p = .78).

When using contrasts to compare the short- versus long-term conditions (sexual and cognitive spots vs. long-term video and control), women also tended to find Nicolas particularly less attractive after viewing the short-term clips (sexual and cognitive, M = 2.67, SD = 1.77) when compared to the long-term disease video (M = 3.29, SD = 1.35), t(96) = 2.46, p = .02, 95% CI [−.01, .19], η a 2 = .06 (see Table 2). Again, no interaction emerged when plotting our experimental condition against participants’ self-reported alcohol consumption in a two-way ANOVA interaction (p = .62).

OPEN IN VIEWER

Table 2.

Means (M) and Standards Deviations (SD), of the Different Dependent Variables Across the Different Conditions in Study 1 for the Second Hypothesis (Short- vs. Long-Term Risks).

Measures/Conditions	Short-Term Risks	Long-Term Risks	Control
	M (SD)	M (SD)	M (SD)
Willingness to be in the target’s situation (men)	4.29 (1.80)	4.40 (1.76)	4.21 (1.96)
Intrasexual competition (men)	3.79 (0.59)	3.93 (0.56)	4.19 (0.94)
Willingness to be in the target’s situation (women)	2.58 (1.50)	2.62 (1.56)	2.67 (1.41)
Target’s attractiveness (women)	2.67 (1.77)	3.29 (1.35)	2.30 (1.35)

Limitations

As in any research program, our studies had limitations that must contribute to making further studies even more precise to attest transparency and replicability. First, since some of the students of that particular high school participated as actors in our binge drinking spots, it was ethically impossible to ask participants to evaluate them. This goes against both research ethical standards and appropriate academic education cooperation. As a result, we had to create an impression formation task following the diffusion of our spots, making the transfer of sexual and competitive evaluation much more difficult to evaluate that if we would have asked to judge directly the individual acting in the different videos. On the other hand, it can be a way to sidestep eventual experimental biases by transferring the spot influences over another task as it is common in social psychology.

Additionally, we did not test for the impact of the spot on attitude or consumption toward alcohol. We tested an indirect impression formation task after the exposure to a spot. The main interest here was to see how binge drinking signals could impact the way they judge an intoxicated peer during a social gathering. Showing that specific signals impact more their impressions than others could be of great help for the health and educational community and contribute more deeply to understand binge drinking as a multifaceted phenomenon.

The main replication of this project may target young high school or university students outside of scholar context to see how much of the cognitive versus sexual signaling really impacts their impression formation. One way to replicate this program would be a longitudinal online investigation to see right after, +1 week, and +1 month after this exposure how much the different spots impacted their attitude toward alcohol and how robust and strong the effect is. Another advantage of the online protocol is that, contrary to a classroom where data can get really noisy (student–student interactions, professor presence, scholar environment, etc.), the online version can be made isolated at home, similarly to the way they would be talking to their friend online or planning a social event.

On the other hand, doing a laboratory setting of our experiment, adding implicit or physiological measures, might be a good way to correlate self-report postimpression to direct in vivo implicit reactions and physiological changes. First, an eye-tracker study would permit to see which part of the different spots attract the most student’s attention, and pupil dilation would tell which moment is most impactful (either describing fear or excitement). Moreover, adding electromyography measures to evaluate electrical activity produced by skeletal muscles (Weiss et al., 2015) and distinguish, combined with eye-tracking measures, which part of the videos are the most stressful could help develop future health programs that are both impactful and measurable. Evidence-based policy must be evaluated scientifically on the long term and, like any other bottom-up approach, health programs must always be turned to measurable hypothesis-generating studies.

The literature is also inconsistent about whether impression formation tasks can in fact reduce drinking intentions. Public health researchers used framing effects (from the prospect theory, Kahneman & Tversy, 2013) to show that a loss-framed message (e.g., engaging in binge drinking ruins your health) would tend to increase risky drinking when compared to a gain-framed one (e.g., refraining from binge drinking is good for you). Gerend and Cullen (2008) were the first to experimentally test (in laboratory conditions) the effect of framed messages on quantity of alcohol use. Participants exposed to short-term consequences messages reported fewer drinks per drinking occasion when compared to participants exposed to long-term consequences messages. They also found that participants in the short-term conditions decreased their drinking consumption after being exposed to gain-framed messages compared to participants in the loss-framed condition. However, Bernstein and colleagues (2015) found that in the loss-framed condition, participants actually reported fewer heavy episode drinking (HED) compared to the gain-framed condition (though they found a main effect of short-term consequences messages on reducing HED compared to long-term consequences messages). On the other hand, Hutter and colleagues (2015) found no main effects in a large sample of UK students (N = 666). They reported that gain-framed messages with incongruous pairing (i.e., loss/refrain or gain/engage) decreased alcohol consumption 2 weeks later compared to other combinations. Our study was the first to propose an impression formation of binge drinker in short- versus long-term consequences messages. Yet, our messages are only loss-framed, and thus, a gain versus loss-framed manipulation should be compared to test how the framing interacts with the impression formation.

Our data suggest a gender difference with women being more influenced by the different conditions (the message videos) when compared to men participants. This can be the result of different operationalizations across genders (intersexual measures for women and intrasexual competition ones for men). However, a meta-analysis conducted by Tannenbaum and colleagues (2015) suggests that, for instance, fear appeals are more efficient when targeting a larger female audience. According to the regulatory fit theory (Higgins et al., 2008), people can be promotion or prevention focused. Our videos end with a prevention message focusing on specific health risks of binge drinking and so are definitely prevention-based messages. Past research indicated that women are more prevention focused than men (Kurman & Hui, 2011; Lockwood et al., 2005), possibly explaining the gender gap we found in the different studies.

Finally, further research should include impression formation tasks including men tolerating alcohol’s effect to see how the influence of the different signaling may vary the response to the following task. Indeed, one possible weakness in our spots is that the men not tolerant to alcohol are sending, de facto, signals of low genetic quality to the participants. To that extent, possible floor effects could have emerged making the target (Nicolas) unfit whatever the situation following the spot is. Getting short- or long-term signals, sexual, or nonsexual would in that case make no difference whatsoever about the target’s capability. On the contrary, measuring the transfer of those signals upon a tolerant man would probably show higher variations between spots leading to more subtle measurements.