Alcohol use disorder (AUD) is a prevalent public health problem in the United States and worldwide. It is estimated that alcohol drinking results in 3–6% of annual deaths worldwide and represents a leading cause of preventable death (Sacks et al., 2015; GBD, 2016; Di Castelnuovo et al., 2019; Alcohol-Related Disease Impact, 2023; NIAAA, 2024). Despite significant research efforts and the emergence of several FDA-approved therapeutics, there has been limited success in treating AUD, as hazardous alcohol drinking has continued to rise in recent years (Grant et al., 2017; Daly and Robinson, 2021; Irizar et al., 2021; NIAAA, 2024). Therefore, understanding mechanisms regulating the consumption of alcohol remains an important priority for biomedical research.

Many environmental factors influence alcohol drinking and excessive alcohol consumption. The factors included, and often overlooked, are the influences of social dynamics on alcohol drinking (Heilig et al., 2016; Sudhinaraset et al., 2016; Mac Killop et al., 2022). For instance, during the COVID-19 pandemic, hazardous alcohol drinking rose, likely due to increases in environmental stress and increased social isolation due to lockdowns (Irizar et al., 2021; Killgore et al., 2021; Weerakoon et al., 2021). On the other hand, epidemiological data indicate that social drinking with peers or partners is associated with the facilitation of hazardous alcohol consumption (Weitzman et al., 2003; Kuntsche et al., 2004; Nogueira-Arjona et al., 2019; Monk et al., 2020; Creswell, 2021). A substantial experimental human research literature indicates that social context can facilitate hazardous alcohol consumption through normative (drinking for ingratiation) and informational influences (drinking “because others do”) (Deutsch and Gerard, 1955; Quigley and Collins, 1999; Larsen et al., 2010; Dallas et al., 2014) and that the magnitude of this influence can depend on contextual cues (Tomaszewski et al., 1980; Doty and de Wit, 1995; Dallas et al., 2014; Monk et al., 2018). However, given the difficulty of controlling the alcohol consumption history and individual genetic and developmental differences, there are challenges to infer causality in such studies. Therefore, animal models of social alcohol drinking are critical to understanding the complicated interplay between social dynamics and alcohol consumption.

The majority of studies in rodent models utilize single-housed animals and relatively simple environments, which do not model human conditions. A number of previous studies in rodents have attempted to study the effects of more complex environments, including social housing and enrichment, on addiction and alcohol drinking (Pohorecky, 2010; Logue et al., 2014; Varlinskaya et al., 2015; Camarini et al., 2018; Koskela et al., 2018; Ryabinin and Walcott, 2018; Venniro et al., 2020; Stefaniuk et al., 2023). Various effects have been observed. For example, a comprehensive study comparing the effects of group housing in standard (impoverished) conditions and complex “intellicage” approaches suggested that both social enrichment and non-social environmental enrichment decrease alcohol drinking in male C57 mice (Holgate et al., 2017). However, this study did not include single housing in the complex condition and used lickometer approaches. Although the overall number of licks in lickometer cages correlates with alcohol drinking in many studies (Ford et al., 2005; Griffin et al., 2007; Anacker and Ryabinin, 2013), lick sizes may differ across various fluids, circadian phases, or experimental conditions. To circumvent this potential problem, we have recently adopted the tracking of consummatory behaviors using radio frequency identification (RFID) tags in Heardsman-2 (HM2) cages (Thomsen et al., 2017; Caruso et al., 2021; Fulenwider et al., 2021; Walcott and Ryabinin, 2021). The HM2 cages are typically large, allowing group housing of rodents. Each HM2 cage is equipped with two channels. Each channel leads to a bottle placed on a precision balance. Only one animal can enter the channel at a time. By a combination of an RFID reader and photoelements located in the channels, the consummatory behaviors of each individual rodent housed either singly or in groups can be tracked with high temporal resolution. HM2 cages can be considered more complex housing conditions because consummatory behavior in these cages requires animals to learn to climb into the channels to get access to fluids and make active choices to enter the channels for specific fluids. In contrast, in a standard cage setup, the drinking spouts protrude into the cage and are next to each other, not requiring memorization of the location of the fluid and making the choice of the fluid easy.

Using the HM2 caging system in male C57 mice, we have observed that while single-housed mice display very low alcohol intake in a two-bottle choice (2 BC) procedure, mice housed at four animals per cage had a higher intake of alcohol per mouse (Caruso et al., 2021; Fulenwider et al., 2021). This finding suggested that non-social and social differences in housing conditions can produce varied effects on alcohol intake: non-social complexity of HM2 cages contributes to lower alcohol intake, whereas the social complexity of group housing contributes to increased alcohol intake in these cages. However, this interpretation was hampered by three potential caveats. First, the previous studies did not perform parallel experiments across all four housing conditions: simple environment + single housing, simple environment + group housing, complex environment + single housing, and complex environment + group housing, leaving open the possibility that the mismatch in effects is due to laboratory conditions. Second, the previous studies in mice were performed only in males, and thus did not allow the assessment of the generality of effects. Third, the previous studies did not adjust the cage size between the single and group-housed animals in the large HM2 enclosures, leaving open the possibility that the higher drinking in the group-housed mice was not due to social effects but due to the availability of space per mouse. Therefore, the current study aimed to overcome these caveats.

Specifically, the goals of the study were to definitively determine the effects of social housing on alcohol drinking in male and female mice and to assess the dependence of these effects on the non-social differences between environments. To attain these goals, we first compared 2 BC alcohol drinking in C57 mice housed either at one or at four per cage in the standard shoebox cages. Having observed social inhibition, rather than facilitation, of alcohol drinking in these relatively “simple” housing conditions, we then compared 2 BC alcohol drinking in C57 mice housed either at one or at four per cage in HM2 cages. We adjusted the space available to each animal in HM2 cages between the single and group-housed conditions to evaluate whether the observed effects were due to social factors. Taking advantage of the full capabilities of the HM2 system, we also analyzed various behaviors that contributed to differences in alcohol drinking across housing conditions.

Experiments were performed in male and female mice, and the results demonstrated a lack of sex differences in facilitating the effects of social housing on alcohol drinking in HM2 cages.

The experiments herein provide evidence that the effects of social factors on voluntary ethanol consumption in mice depend on the non-social properties of the housing environment. Specifically, in standard cages, single-housed mice consume more alcohol than group-housed mice (Figure 2A). In contrast, in HM2 cages, group-housed mice show higher alcohol preference (Figures 3E,F) and make more approaches to consume from the ethanol-containing bottles than single-housed mice (Figure 4B). This facilitating effect of group housing on alcohol preference and approach to ethanol was independent of the sex of the animals.

The vast majority of studies investigating voluntary alcohol consumption in rodents use single-housed animals in standard shoebox cages. While these housing conditions allow easy assessment of voluntary alcohol intake, there are substantial caveats to the interpretation of these studies. Most importantly, single housing is well known to be stressful to rodents, resulting in increased reactivity of the hypothalamic–pituitary–adrenal axis and increased signs of anxiety and depression (Berry et al., 2012; Krügel et al., 2014; Ieraci et al., 2016). Therefore, despite this standard practice, alcohol consumption in single-housed mice should hardly be taken as a “baseline” condition. The lower alcohol consumption in group-housed mice in these cages observed here is in agreement with a substantial number of previous reports (Thiessen and Rodgers, 1965; Deatherage, 1972; Parker and Radow, 1974; Advani et al., 2007; Lopez et al., 2011; Lopez and Laber, 2015; Panksepp et al., 2017). This decrease in alcohol intake can be theorized to reflect lower stress levels in socially housed animals. However, the relationship between stress and alcohol intake in rodents is complex, and exposure to stress can also decrease, instead of increase, alcohol intake (van Erp et al., 2001; Norman et al., 2015). An alternative possibility is that the presence of cagemates within the tight space of standard cages creates competition for the drinking cylinders. In agreement with this idea, the inhibitory effects of group housing on intake of 3% ethanol in our study were more pronounced in male than in female mice (Figure 2A). This sex difference could reflect typically higher levels of aggression in male versus female rodents (Hashikawa et al., 2018; Karamihalev et al., 2020; Fulenwider et al., 2022), which could result in increased competition for resources. Also in agreement with this interpretation, social housing decreased consumption of not only 6% ethanol but also water in the second phase of the experiment in standard housing conditions (Figures 2B,D).

Another caveat of interpreting results in shoebox cages is the possibility that animals are spilling the fluid, in addition to consuming it. This possibility decreases the precision of measurements. We cannot exclude the possibility that spillage is greater in group-housed animals versus single-housed mice. Related to this caveat, measurements of alcohol intake by group-housed mice in standard shoebox cages are also not precise because the intake values were estimated based on the total fluid consumed by the cage and the body weight of each of the four mice in each cage (see Methods). However, since we detected a statistically significant decrease in ethanol intake in the social housing condition in standard shoebox cages compared to single-housed animals (Figure 2B), it is unlikely that the overall decrease in ethanol consumed per cage would have masked social facilitation of this intake in a proportion of mice. The limitations of studying voluntary social alcohol consumption in standard shoebox cages and the opposite effects of social housing on ethanol intake in standard versus HM2 cages demonstrate the utility of the HM2 cages for studying social facilitation of drinking.

The social facilitation of alcohol preference was visible compared to low alcohol preference (Figures 3E,F) and intake (Figures 3C, D) in mice single housed in HM2 cages. We have previously also observed low alcohol intake in single-housed male mice in these cages (Fulenwider et al., 2021). The intakes of alcohol in HM2 cages in the current study were similar to the ones observed in the previous study (4 g/kg/day on average at the 6% phase of the experiment Figure 3B) and substantially lower than the intake of approximately 7.5 g/kg/day observed in standard shoebox cages (Figure 2B). Therefore, prior to discussing why social housing enhances alcohol preference in these cages, we need to consider why alcohol intake is low in these cages compared to the standard shoebox conditions in single-housed mice. The lower intake of alcohol in HM2 cages versus standard cages cannot be attributed to the differences in the sizes of the cages because the intake of alcohol in the HM2 system was the same whether we used the full-size HM2 enclosures for single-housed mice [as in Fulenwider et al. (2021)] or decreased their size using a partition (current experiment). However, importantly, fluid consumption in shoebox cages and HM2 cages could be guided by very different behavioral mechanisms. In the shoeboxes, the spouts of the two drinking cylinders protrude into the center of the cage and are located close to each other (8.5 cm distance). Since the position of the ethanol-containing cylinder is alternated daily, there is no need for the mouse to remember the position of the preferred fluid. As a result, the choice of intake in this type of housing is most likely based on daily sampling. In contrast, the entrances to the channels in HM2 cages are located much further apart (27 cm), and the mouse must climb into and through the channel to reach the drinking spout. Since the fluid at the end of the channel is not alternated from day to day, the mouse is more likely to rely on its memory to access the fluid of choice. This behavior in HM2 cages, therefore, is more likely to demand a more deliberate action from the animal without being able to rely on simple sampling of fluids. As a result, mice could be less efficient at reaching their fluid of choice than in the shoebox cages. The HM2 cages could be, therefore, considered a more complex environment than shoebox cages not because of any special complexity of the cage, but because consummatory behaviors in these cages require more effort.

Interestingly, although intakes of ethanol were similar between the current and previous experiment using different sizes of HM2 enclosures, water intake, and as a result, preference for ethanol was different. In the previous experiment, using unpartitioned HM2 cages, single-housed mice consumed approximately 40 g/kg/day of water (Fulenwider et al., 2024), whereas here they consumed an average of over 100 g/kg/day of water (Figure 3D). Consequently, preference for 6% ethanol was consistently around 70% in large unpartitioned cages, and slightly below 50% in the smaller, partitioned cages in the present study (Figure 3F). This comparison between studies suggests that decreasing the available area of HM2 cages, as we have done here, made the mice less discriminating in their choice of fluid. The reason for this effect is not clear and would need additional parametric experiments. For example, it would be useful to test whether this observation would apply to 2 BC procedures with other fluids, or even just with water alone. In any case, equalizing the space available to each mouse in HM2 cages allowed us to investigate the social facilitation of alcohol preference in the current experiments.

Specifically, we observed that while the intake of ethanol was on average nominally higher in socially housed versus single-housed mice in HM2 cages, this difference did not reach statistical significance (Figures 3A,B). In contrast, alcohol preference was significantly higher in the group-housed versus single-housed mice (Figures 3E,F). This difference in preference was primarily driven by a difference in water consumption, which was lower in socially housed versus single-housed mice (Figures 3C,D). These differences in preference and water intake were independent of sex or concentration of ethanol in the 2 BC procedure. The differences were also consistent across the circadian cycle, where hourly analysis of consumption showed nominally higher ethanol intake and lower water intake in socially housed mice during the dark phase of the circadian cycle (Figure 8). The relationship of fluid consumption reversed between housing conditions only at 1–2 h during the dark cycle, which could not have contributed to the overall differences in intake between groups.

While differences in ethanol intake between single and socially house mice in HM2 cages did not reach statistical significance, the number of entries into the ethanol channels when animals consumed alcohol (CVs) was significantly higher in group-housed versus single-housed mice during the 6% ethanol period (Figure 4B). Times spent in the channels (CV time) showed similar group differences as in analyses of CVs, but were complicated by sex differences, where socially housed males spent more time in the ethanol channels than females (Figure 5A). Overall, the group differences in CV times were less robust than differences in CVs (Figures 4, 5). Since CV times is a composite measure comprised of number of visits to the channel and the duration of each visit, these data indicate that differences in fluid consumption in this procedure are primarily driven by the number of visits to specific channels. Importantly, the increase in CVs into the ethanol channel was not accompanied by an increase in the number of visits when animals were not consuming ethanol (NNVs). In fact, NNVs to both ethanol and water channels were lower in socially housed versus single-housed mice (Figures 6A,B). The latter observation serves to indicate that the increase in ethanol CV is not a consequence of any potential locomotor stimulating effects of alcohol consumption or social housing.

Taken together, analyses of behavior in HM2 cages showing increased alcohol preference, decreased water intake, increased number of CVs to the ethanol channel, and decreased NNVs to both channels, suggest that socially housed mice were more deliberate in their attempts to access the ethanol channel, which can reflect higher motivation to access ethanol, than single-housed mice. It is important to consider, therefore, why socially housed mice could have higher motivation to access ethanol than single-housed mice. One possibility is that individual mice in social cages could exhibit individual influences on their cagemates to encourage their motivation for alcohol. We have observed such influences previously in other species of rodents. Specifically, we reported that pair-housed prairie voles (Microtus ochrogaster) can coordinate their alcohol intake resulting in significant correlations of alcohol intake between each member of a pair (Anacker et al., 2011a,b; Anacker and Ryabinin, 2013). This coordination of intake likely contributes to a high incidence of “cage effects” in prairie voles housed in HM2 cages at four per cage. In other words, prairie voles housed at four per cage are likely to contain either only high or only low alcohol-consuming animals (Walcott and Ryabinin, 2021). However, social behaviors are known to vastly differ between prairie voles and mice. Thus, while mice do prefer a social environment to isolation (Panksepp and Lahvis, 2007), in contrast to voles (and humans), they do not form long-term attachments between specific adult individuals (Insel et al., 1998; Young et al., 1999; Charles et al., 2014). In accordance, the social environment could influence alcohol drinking in these two species via different mechanisms. In agreement with this idea, comparing the distribution of high-, low- and medium-consuming individuals between cages with socially housed mice did not identify a strong clustering of mice with similar drinking levels (Figure 9). While there were differences between cages, the majority of cages contained animals with different levels of alcohol intake and preferences. Therefore, it seems likely that the increased motivation to consume ethanol in mice socially housed in HM2 cages could not be attributed to animals within each cage mimicking each other’s drinking. Instead, a more likely explanation is that the overall context of the presence or absence of cagemates influenced alcohol preference. At this point, it is difficult to speculate about precise neural mechanisms that could contribute to the effects of social context on alcohol drinking in HM2 cages. While there has been much research describing differences in gene expression and neurocircuits between socially and single-housed animals (Siuda et al., 2014; Yorgason et al., 2016; Zilkha and Kimchi, 2018), previous studies have been primarily performed in the context of standard housing and might not apply to conditions found in HM2 cages.

Although the presence of animals with varied alcohol intake within the same cage of socially housed mice suggests a lack of matching of drinking behavior between animals, it does not mean that individual animals in the cage do not influence each other’s drinking. Thus, differences in alcohol intake could be not due to mimicking behavior, but due to their different rank within a social hierarchy. Social hierarchies have been shown to influence alcohol intake in humans (Dumas et al., 2018), non-human primates (McKenzie-Quirk and Miczek, 2008), rats (Duncan et al., 2006), prairie voles (Anacker et al., 2014), and mice (Hilakivi-Clarke and Lister, 1992). However, rodent studies on the relationship between alcohol drinking and social hierarchy have not been performed in the context of HM2 cages. Our previous studies using the tube test of social dominance indicated that it usually takes more than 10 days to establish stable social hierarchies in mice socially housed in HM2 cages (Fulenwider et al., 2024). Confoundingly, the establishment of hierarchies in a social environment is promoted by tube testing (Zhou et al., 2017). Therefore, it is not clear whether the total of 10 days of housing in HM2 cages (5 days of drinking water and 5 days of 2 BC with ethanol) without repeated testing for social dominance, as was done here, was sufficient to result in a social hierarchy. Moreover, the social dominance ranking established by the tube test in HM2-housed mice in our previous study was not easily translatable to differences in consumption of sucrose, possibly due to ease of finding sufficient time to access this palatable fluid. Although alcohol and sucrose consumption could be differently regulated by social rank, it is unclear whether social hierarchies have contributed to differences in alcohol drinking within each cage of socially housed mice in our study.

Furthermore, being able to socially housed mice when testing individual fluid consumption is progress over testing alcohol intake in mice singly housed in shoebox cages, the HM2 cage analysis of drinking also has caveats. First, as mentioned previously, intakes in this procedure are not very high and the alcohol exposure only lasted 5 days. Therefore, interpreting these studies as addressing mechanisms of development of AUD can be questioned. In pilot experiments performed in our laboratory, we did not observe increases in alcohol intake when mice in HM2 cages were consuming alcohol for approximately 1 month or if the 2 BC procedure was preceded by “no choice” alcohol exposure (data not shown). These pilot experiments suggest that alcohol intakes and associated behaviors observed here are likely not to differ from the ones that could be achieved if additional paradigms thought to induce dependence in mice were employed. Therefore, the social effects of alcohol consumption studied here are more likely to model what is considered normative social drinking. Population statistics indicate that while 84% of the adult US population consumed alcoholic beverages at some point in their lifetime, only 6.3% reported current heavy alcohol use (SAMHSA, 2022a,b). We calculate that the top 6% of mice in our HM2 experiments consume an average of 9.5 g/kg/day. This is a substantial intake of ethanol, comparable to that observed in 2 BC procedures with higher concentrations of ethanol in mice individually housed in shoebox cages (McClearn and Rodgers, 1959; Belknap et al., 1993; Seemiller et al., 2022). Such doses consumed in 2 BC procedures are capable of producing hyperalgesia upon withdrawal following less than 2 weeks of exposure, which is a sign of at least mild alcohol dependence (Smith et al., 2016, 2017). Thus, future studies focusing on subpopulations of individual mice consuming high doses of alcohol in HM2 cages could be translationally important.

Another potential caveat of studies in HM2 cages is that at a time when mice are consuming fluid, they are in the channels by themselves, facing away from other mice in the cage. This situation is different from when human subjects are usually engaged in social drinking. It is not clear whether this difference plays a substantial role in regulating alcohol consumption. However, we note that prairie voles socially housed in HM2 cages consume very similar amounts of alcohol as when they are singly housed or pair-housed in shoebox cages (Anacker et al., 2011a; Walcott and Ryabinin, 2021). Therefore, at least in a species with translational validity for human social behaviors, this theoretical deficiency in the experimental setup of HM2 cages does not contribute to changes in alcohol drinking.

Despite the two caveats above, assessing ethanol intake in mice socially housed in HM2 cages provides substantial advantages over traditional preclinical methods used in studies on AUD. Efforts to allow testing alcohol drinking in socially housed animals via various other methods are underway in other research groups [reviewed in Ryabinin and Walcott (2018)]. While many of these methods have their own advantages, we note that currently they are based on either measuring proximity to the drinking spout (Logue et al., 2014; Varlinskaya et al., 2015) or registering licking behavior (Holgate et al., 2017; Stefaniuk et al., 2017; Frie and Khokhar, 2024). Although applying varied methods to assess behaviors has advantages over a single method, our studies in HM2 cages indicate that results from measures of approach to the fluid do not always coincide with measures of intake. Thus, although 6% ethanol CVs were significantly different between single and group-housed animals (Figure 4B), ethanol intakes were not significantly different between these groups (Figure 3B). In contrast, while water CVs at 3% ethanol were not significantly different between the groups (Figure 4C), water intake was significantly different between single and group-housed mice (Figure 3C). These findings reiterate the importance of distinguishing between the appetitive and consummatory phases of self-administration behaviors (Craig, 1917; Kringelbach and Berridge, 2016; Sherrington, 2023). Measuring just approach behavior or just intake could lead to different conclusions. Future studies should strive to incorporate analyses of both phases of alcohol consumption into their analysis.

Taken together, our experiments in standard housing conditions and automated HM2 cages, show that the effects of social housing on alcohol drinking depend on non-social parameters of the housing environment. We demonstrate that social housing increases alcohol preference and approach to the alcohol solution in environments requiring more deliberate effort to access consumed fluids. It will be important to start identifying molecular mechanisms underlying the potentially increased motivation to consume alcohol under these social circumstances. Identification of such mechanisms could help design therapeutic strategies to decrease problematic patterns of alcohol consumption in subjects with AUD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was approved by Institutional Animal Care and Use Committee at OHSU, Portland, OR, United States. The study was conducted in accordance with the local legislation and institutional requirements.

MJ: Data curation, Formal analysis, Investigation, Visualization, Writing – original draft. JZ: Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. YZ: Investigation, Methodology, Writing – review & editing. AR: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – original draft.