2. Experimental Organisms in American Alcohol Research

From the time of the first coherent definitions of alcoholism as a disease, termed “alcoholic insanity” or “neurosis” in the nineteenth century, experiments on non-human animals have been critical to its analysis. Up to the end of the nineteenth century, and following the delineation of the symptomology of “chronic alcoholism” by the Swedish physician Magnus Huss (Reference Huss1849), studies focused on the effects of high and sustained doses of alcohol on the body and mind (e.g., Stewart Reference Stewart1898; Lewis Reference Lewis1899). These early experiments were carried out predominantly in France, Russia, and Germany on a variety of organisms, including dogs and fowl (e.g., Magnan Reference Magnan1876). At the turn of the century, the bulk of research activities in this area moved to the United States, where dogs, rabbits, pigs, guinea pigs, monkeys, cats, birds, and wild rats were used, as well as the “white rat” which was employed for the first time in the laboratory of Clifton Fremont Hodge of Clark University (Hodge Reference Hodge, Atwater, Billings, Bowditch, Chittenden and Welch1903; see also Stewart Reference Stewart1898; Miles Reference Miles1930; Logan Reference Logan1999). Hodge carried out a series of studies in which animals were force-fed alcohol to study its effects on digestion and secretion, growth, and development, resistance to infection, toxicity, nutrition, and general physical pathology. Animal studies were critical to the debate between the temperance movement and its critics, as, with regards the effects of alcohol on the body “the great complexity of the human organism. . . has made the definite interpretation of the human experiment impossible. The method of physiological science is to reduce the problem to simplest terms in every way possible” (Hodge Reference Hodge, Atwater, Billings, Bowditch, Chittenden and Welch1903, 359–60).Footnote ⁷ Importantly, non-human animals were not considered to be simpler solely by virtue of their physiological adaptations. Hodge also noted that the “conditions of life” of the animals could be controlled in the laboratory, though the degree of control he exerted was limited, especially by later laboratory standards. Hodge's animals lived in a semi-naturalistic setting to ensure their physical and psychological health, and the animals were given “complete freedom” to wander in a sunny quarter-acre yard, so much so that one kitten was killed by a dog, and a dog run over by a car (Hodge Reference Hodge, Atwater, Billings, Bowditch, Chittenden and Welch1903). The set-up of an experimental environment granting such a degree of freedom to the animals was significant in terms of the validity of data obtained through these studies: according to Hodge, freedom of movement would remove the possibility that distress would confound the research results.

With the advent of Prohibition in the 1920s, interest in alcohol research waned in the US and the American Association for the Study and Cure of Inebriety disbanded (Blocker Reference Blocker1989). With the failure of Prohibition and its repeal in 1933, interest in alcohol abuse returned, with a more specific focus on the causes of alcoholism. In 1939, the Research Council on Problems of Alcohol was established, evolving into the Center of Alcohol Studies at Yale, directed by E. Morton Jellinek. This was the focal point of alcohol research and published the Quarterly Journal of Studies on Alcohol. Active in the treatment of alcoholics through the Yale Plan Clinics and closely linked to Alcoholics Anonymous, Jellinek (Reference Jellinek1942 and Reference Jellinek1960) vigorously promoted what was seen as the “new approach to alcoholism.” He defined five “species” of alcoholism, two of which were diseases due to physical dependency. The disease could be diagnosed by symptoms of tolerance and withdrawal, but most critical was the complete loss of control, in spite of immense physical, psychological, financial, and social consequences. Choice, will, and volition were overridden by an overwhelming desire for drink (Jellinek Reference Jellinek1960, 45). This vision of alcoholism as a disease proved popular because it removed stigma: alcoholism was not due to weakness of will or temperament, but an illness to be understood by science and treated by medicine (Schneider Reference Schneider1978; Conrad and Schneider Reference Conrad and Schneider1980).Footnote ⁸

For Jellinek, the study of alcoholism was fundamentally interdisciplinary. While empirical verification of various clinically-anticipated effects of alcohol as well as range of physiological functions had been well established, when it came to the “motivation in the genesis of the alcohol habit,” psychiatric speculation was rife while experiment offered a complete “tabula rasa” (Jellinek and McFarland Reference Jellinek and McFarland1940, 276). Jellinek reviewed the physiological contributions to this topic, noting in particular the work of the Johns Hopkins psychobiologist Curt Richter, as it provided possibilities for further psychological study. Richter's preferred experimental organism was the Norway rat, an animal that the laboratory scientist “could not possibly improve on” (Richter Reference Richter1968). The rat was rapidly becoming the generic animal model in psychology (Logan Reference Logan1999). Richter initially used the albino rat to which he was introduced by John Broadus Watson, but he rejected his mentor's use of this animal, privileging innate and spontaneous behavior over learning. He also rejected the tendency among biologists to associate the use of standardized cages with behaviorism: “The use of our cages makes it possible to put very definite questions to the rats, and to get definite answers . . . From the beginning. . . my interest focused entirely on what animals do on their own, that is, their innate behavior, not on what they can be taught to do” (Richter Reference Richter and Dewsbury1985, 376–7). In the 1930s, Richter devised a series of “cafeteria” experiments in which the animals were free to choose a variety of foods, believing that the “wisdom of the body” ensured the animal selected the correct nutritional requirements (Richter and Campbell Reference Richter and Campbell1940). This experimental set-up was particularly innovative in comparison to the reliance of earlier physiologists on force-feeding tubes and gastric fistula.

In the course of these studies, Richter discovered that rats also drank alcohol when it was provided in a “free choice” situation. The animals would accept beverages with alcohol volumes between 1–6 per cent due to their caloric content (preferring a 5.1 per cent concoction), but when offered the choice between alcoholic beverages and water, would take a sip of the former before turning to the latter. Rats considered alcohol a food – they did not drink to get drunk. Richter also noted a great deal of variation among the animals, and a small minority of “alcoholics” (more commonly among the wild rats), would drink larger volumes than others. He suggested this was the result of differing taste thresholds or perhaps endocrine deficiencies that could be corrected through supplements (Richter Reference Richter1926 and Reference Richter and Abramson1957; Richter and Campbell Reference Richter and Campbell1940).Footnote ⁹ Richter's experimental set-up stimulated numerous studies of self-selection among individually caged rats which were used to identify various metabolic, endocrinal, nutritional, chemical, and later, behavioral genetic differences between “drinker” and “non-drinker” animals within and between different strains (Eriksson Reference Eriksson1972; Eriksson et al. Reference Eriksson, Sinclair and Kiianmaa1980).

Jellinek was, however, less than impressed with evidence of hereditary factors in the generation of addiction (Jellinek and Jolliffe Reference Jellinek and Jolliffe1940). Increasingly drawn towards the social and behavioral sciences, he was more intrigued by the potential of experimental psychology to identify the causes of drinking in the early stages of alcoholism before physical dependence took hold. Particularly important in this respect was the work of the psychiatrist Jules Masserman. Masserman was committed to psychoanalysis but felt that, focused as it was on clinical cases, it had not yet fulfilled its opportunity to broaden and extend its approach through closer associations with general and comparative psychology (Masserman Reference Masserman1943). For Masserman, psychoanalytical concepts such as “inner needs” and “motivational conflicts” could be legitimately applied to non-human animals; humans differed only with regards to the variety and versatility of their “technics of adaptation” (Masserman Reference Masserman1968: 253). Devising a variety of conflicting situations (retrieving food at the risk of an air-blast), Masserman generated “neuroses” among cats. After injecting or force-feeding them alcohol, found that their neurotic reactions to the laboratory environment were relieved. He lent his support to further studies by the clinical psychologist, John Conger (Reference Conger1951 and Reference Conger1956) who carried out similar experiments with rats. Masserman and Conger argued that the animal laboratory provided support for arguments that drinking served to overcome fear, inhibition, and phobia, thus highlighting the importance of the environment as a trigger for the disease. Thus, not only was the physical and psychical make-up of experimental organisms of critical importance, but also the space in which they were located when setting up experiments.

While Masserman considered animal studies to be a means for further establishing the credentials of psychoanalysis as a biologically-grounded medical science, with the growing influence of behaviorism in addiction research, attention shifted from the role of alcohol for relieving conflicting drives, to a focus on its positive reinforcing qualities. For instance, animals were kept in experimental chambers (typically referred to as “Skinner Boxes”) in which food would be delivered only when the animals drank a drug solution or pressed a bar that would deliver alcohol, morphine, and cocaine (Keehn Reference Keehn1969 and Reference Keehn1986; cf. Campbell Reference Campbell2007). Once this behavior was established and animals were conditioned to request drugs, psychologists sought to identify the symptoms of physical dependence that sustained the use of intoxicating substances. Schedule-induced polydipsia, a particular method developed by the behaviorist psychologist John Falk, proved both influential and controversial. Falk (Reference Falk1961) discovered that an intermittent feeding schedule (one food pellet every few minutes) induced the rats to consume vast quantities of water. He, and others, then applied this finding to alcohol research by using the intermittent feeding schedule as a method for inducing rats to consume large quantities of alcohol, thereby inducing intoxication (Falk et al. Reference Falk, Samson and Winger1972 and Reference Falk, Samson and Winger1976; Lester Reference Lester1961; Mello and Mendelson Reference Mello and Mendelson1971).

3. Validating Experimental Models in Alcohol Research

This brief historical outline shows some of the variety of ways in which experimental organisms were enrolled into research programs to investigate the biological bases of drinking behavior. For early researchers, using experimental organisms to study alcoholism offered several advantages: the animal's environment could be controlled in ways that would be impossible in humans, as Hodge noted, and Richter similarly argued that using animals gave researchers unique opportunities to pose precise experimental questions. But using animals as experimental models for alcohol research also raised questions about the validity of these experiments; that is, about what constitutes a representative experimental set-up for human behavior.

One of the main questions of validity in this field concerns whether animals can be used to understand the “uniquely human” phenomenon of alcohol addiction. There is considerable evidence supporting the idea that alcoholism is peculiar to humans: animals, even when interested in using some alcohol, do not show a tendency towards consuming large amounts of it (in ways that the WHO would define as “abusive and damaging” [WHO 1955]). Mice and rats, the favorite experimental organisms of late twentieth-century biomedical research, are particularly reluctant to drink alcohol when given a choice between alcohol and water, as Richter's early work on the innate feeding behavior of rats showed. Later experiments such as Gerald McClearn and David Rodgers’ two-bottle choice experiments (where they gave animals bottles of alcohol and water and measure how much they drank from each) showed that only a few strains of mice will drink alcohol in large quantities and some strains of mice will refuse to drink alcohol entirely (McClearn and Rodgers Reference McClearn and Rodgers1959). More recent experiments have shown that even strains of mice that do seem to drink alcohol in large quantities (such as the C57BL, the standard strain with the highest affinity for alcohol) do not appear to consume it in a way that results in consistently high blood alcohol levels (Dole and Gentry Reference Dole and Gentry1984). Alcohol addiction in experimental animals thus is largely understood as induced, inasmuch as the behavior and preferences of animals have to be transformed in order for them to serve as experimental models for human alcoholism. Even in the few cases where researchers find evidence which supports the portrayal of alcohol addiction as innate for some types of animals, interest in human alcoholism is what drives research interests, not scientific interest in understanding alcohol's effects on non-human animals for its own sake.Footnote ¹⁰

If alcoholism in experimental animals was understood to be largely the result of inducements rather than innate preferences, what kinds of inducements should be considered legitimate for modeling human alcoholism? This question was strenuously debated in the field from the mid-1960s through the 1980s, and various strands of these debates continue into the present day. The use of behaviorist techniques in the 1960s generated some of the most severe criticism. Scientists were concerned with the validity of experimental set-ups such as Falk's schedule-induced polydipsia technique, and about the standards for what counted as a valid model for alcoholism more generally (cf. Broadhurst Reference Broadhurst1963). Critics suggested that in many experimental set-ups the animals were in fact choosing alcohol for its caloric properties or due to the sugary taste of many of the solutions (Freed and Lester Reference Freed and Lester1970). Some researchers also found that even after inducing rats, mice, and monkeys to consuming alcohol over considerable periods of time, the animals did not choose alcohol when placed in free-choice situations, nor did they display the symptoms of physical dependence (Freund Reference Freund1969; Ogata et al. Reference Ogata, Ogata, Mendelson and Mello1972; Mello Reference Mello1973; Mello and Mendelson Reference Mello and Mendelson1971). More generally, critics questioned the degree to which animals were acting “voluntarily.” In the behaviorist's laboratory, animals were trained to drink, and thus did not choose alcohol because of its intoxicating quantities as was the case among humans.

A series of papers published by alcohol researchers throughout this period (Lester Reference Lester1966; Lester and Freed Reference Lester and Freed1972 and Reference Lester and Freed1973; Cicero Reference Cicero, Majchrowicz and Noble1979 and Reference Cicero, Eriksson, Sinclair and Kiianmaa1980) argued that it was perfectly acceptable to use non-volitional techniques to explore various aspects of alcoholism such as physical dependence (Cicero et al. Reference Cicero, Snider, Perez and Swanson1971). But terms such as “animal” or “addiction model” needed to “be parsimoniously reserved for animal behavior which stringently meets both psychological and physiological criteria” (Lester and Freed Reference Lester and Freed1973, 106). A 1979 paper by Theodore Cicero articulated several criteria that an animal model should meet in order to be considered a valid model of alcoholism, such as: animals had to self-administer alcohol by the oral route and consume it in quantities that would result in pharmacologically significant blood alcohol levels; alcohol should be consumed for its pharmacological properties and not for its taste or caloric properties; animals should be willing to work for alcohol; and tolerance and dependence must emerge as a result, measured by reduced effects of alcohol consumption and acute withdrawal symptoms. These criteria became a touchstone for later debates in the field about model validity, with researchers attempting to assert the utility of particular animal models by arguing that they fulfilled all of Cicero's criteria (Bell et al. Reference Bell, Rodd, Lumeng, Murphy and McBride2006; Rodd et al. Reference Rodd, Bell, Sable, Murphy and McBride2004) and suggesting new additions to the list, such as a new criterion specifying that animal models should also show evidence of relapse (McBride and Li Reference McBride and Kai Li1998).

4. Situating Animals in Valid Experimental Contexts

What is most notable for our purposes here is that, despite the heavy focus in this field on using animals to understand the biology of alcoholism (especially genetic factors that predispose individuals to heavy drinking or the neurobiological mechanisms that underlie drinking behaviors), the validity of the experimental models depends to a large extent on the behaviors of the animals and the settings in which those behaviors are elicited, rather than the biology of the organism alone. Cicero's requirement that an animal must “orally self-administer alcohol,” for example, suggests that it is not just the rodent itself that models alcoholism, but a rodent situated in certain environments and experimental situations that allows it to become a model for the human. This criterion suggests that other methods for administering alcohol to animals – letting animals breathe in alcohol vapor, forcing them to drink alcohol by mixing it with their food or water, or even injecting it directly into their bodies – can be used to understand some aspects of the biology of alcoholism, but fall short of becoming representative for behavioral features of alcoholism such as craving.

The different positions taken by researchers in the field with regards to the implications of various manipulations to get experimental animals to drink demonstrate how the experimental setting is simultaneously a resource for overcoming experimental problems (such as rodents’ aversion to alcohol) and a site for drawing parallels to the human and assessing the validity of the model. In another controversial experimental design developed in the 1980s by Herman Samson, animals were trained to press a lever in a conditioning box to get drops of a sweet solution, and then the sugar in the solution was gradually reduced while the alcohol was gradually increased until the animals are drinking unsweetened alcohol (Samson Reference Samson1986). When trained in this way, even animals that had been selectively bred for low alcohol preference would drink a moderate amount, a result that the researchers argued demonstrated the powerful effect of the environment on drinking behaviors (Samson et al. Reference Samson, Tolliver, Lumeng and Kai Li1989). The “sucrose fading” procedure was effective for getting rodents to drink significant amounts of alcohol, but activated discussions in the field once again about taste, learned behaviors, and the motivation to drink alcohol. Using a sugary solution to make alcohol initially more palatable to rodents seemed to contravene Cicero's requirement that true animal models of alcoholism should consume alcohol freely and only for its pharmacological effects, and some in the research community argued that adding sugar into the experimental situation was adding an unnecessary complication that “preclude[d] seeing animals learning to work for alcohol in the absence of confounding factors” (Hyytia and Sinclair Reference Hyytiä and Sinclair1989, 161). Other researchers took the position that taste confounds were not just limited to models where sugar or food deprivation were involved, but were a potential problem in all experiments where animals had to ingest alcohol orally. Grahame and Cunningham (Reference Grahame and Cunningham1997) argued that the high-drinking C57BL mouse strain might be drinking because the taste of unadulterated alcohol was less aversive to them than to other strains, and suggested an alternative experimental design whereby mice could be trained to press a lever to have alcohol delivered directly into their stomachs via a pump. Other researchers argued that while the motivation to drink in various experimental designs was complicated by taste, the same could be said of drinking in humans. One group reported choosing the sucrose fading model because it offered a more “anthropomorphic” representation of drinking behaviors that bore a resemblance to the environmental conditions in which humans learn to drink. “Humans do not. . .usually initiate their drinking with pure ethanol,” they argued, and “in contemporary times, ‘wine coolers’ or mixed drinks are used as a ‘gateway’ to drinking large amounts of ethanol” (Gauvin, Moore, and Holloway Reference Gauvin, Moore and Holloway1993, 38).

In these cases, establishing and using a particular set-up involves making decisions about the scope of representation (whether to model general features of alcoholism such as motivation to drink or specific features such as the moment relapse), and tradeoffs between representing some aspects of alcoholism over others. None of these experimental set-ups are considered by researchers to be “isomorphic to the human condition”; rather, researchers argue that they each reveal “some aspects of a complex process” of addiction and abuse, such as preference for alcohol, dependence, withdrawal, and relapse (McBride and Li Reference McBride and Kai Li1998, 339). The main question for researchers is, how much of the human condition should researchers aim to model, and which aspects are most crucial to represent? Some experimental set-ups aim for a biological resemblance to human drinking, targeting the development of their models towards particular physiological features such as a blood alcohol level of 0.08 per cent (the legal limit for drunk driving in humans) (Rhodes et al. Reference Rhodes, Best, Belknap, Finn and Crabbe2005). Other researchers aim for behavioral similarities between animals and humans, placing less emphasis on the amount that rodents drink and more emphasis on whether experimental organisms appear to find alcohol motivating or rewarding (Green and Grahame Reference Green and Grahame2008).

The ways in which researchers decide on these priorities is linked to their understandings of the ultimate purpose of experimenting on nonhuman organisms. The aim of developing animal models of alcoholism, after all, is to produce information that will facilitate better treatments or a better understanding of the etiology of human alcoholism. Some researchers argue that if the purpose of animal models is to develop new medications, then it is not necessary for experimental organisms to manifest all (or even any) of the symptoms of human alcoholism as long as the experimental set-up as a whole predicts which drugs will be effective in humans. For instance, NIAAA intramural researcher Mark Egli argues that researchers’ efforts might be better directed towards developing “rapid, inexpensive tests which yield the same information as the more elaborate models, although they may be less obvious models of alcoholism” (Egli Reference Egli2005, 315). Others argue that experimental designs where the physiology and behavior of the animals strongly resemble human drinking can be used to identify novel genes or signaling pathways involved in drinking, which may point the way towards new drug targets or susceptibility genes in humans (Tabakoff et al. Reference Tabakoff, Saba, Printz, Flodman, Hodgkinson, Goldman and Koob2009).

Finally, researchers’ considerations of the contexts in which drinking behaviors are elicited and their implications for model validity can extend also to the animals’ home cage environments. In some cases, the ways in which animals are housed aim at controlling away rather than explicitly utilizing environmental context. For example, since stress is believed to impact on drinking behavior, some researchers use only mice bred in local facilities to avoid the stress caused by shipping them from a commercial supplier, or compare mice locally-bred with shipped mice in their experimental design (for an example of this design, see Mulligan et al. Reference Mulligan, Ponomarev, Boehm II, Owen, Levin, Berman and Blednov2008); and animal researchers in behavioral fields routinely control for many of these aspects of the laboratory environment, such as noise and light levels, food, and even the person conducting the experiment (Chesler et al. Reference Chesler, Wilson, Lariviere, Rodriguez-Zas and Mogil2002). In other cases, the home environment and early experience of animals are explicitly used as variables to model the conditions under which humans come to drink excessively. Masserman and Conger's early work on how stressful events change patterns of drinking in experimental organisms remains a topic of study in behavioral research today, with researchers investigating how stressful events early or late in life change patterns of drinking in both rodents and non-human primates (Sillaber et al. Reference Sillaber, Rammes, Zimmermann, Mahal, Zieglgänsberger, Wurst and Holsboer2002; Chester et al. Reference Chester, Blose, Zweifel and Froehlich2004; Fahlke et al. Reference Fahlke, Lorenz, Long, Champoux, Suomi and Higley2006). Another line of research that extends back to the 1960s uses the home cage environment as a variable to explore how living in environments that have been “enriched” with shelters, bedding materials, exercise wheels, and other objects rather than standard cage environments impacts behavior and development (Ramsden 2011a). Researchers have used these enriched set-ups to study, for example, how the rearing environment modulates the behavioral consequences of exposing rodents to alcohol while in utero (Hannigan, Berman, and Zajac Reference Hannigan, Berman and Zajac1993). Some lines of research have even investigated the relationship of drinking and social behavior in animals, such as the influence of dominance hierarchies on drinking patterns in rodents and especially in primates (Blanchard et al 1987; Peretti and Lewis 1969). Interest in the effect of social structure on drinking is high because of its obvious parallels to human drinking, although there are presently relatively few experimental programs focusing on social structure and alcohol intake in rodents (Crabbe Reference Crabbe2012).

6. Standardizing Genes, Organisms, and Environments

The standardization of rodents as experimental organisms undoubtedly is a key component of the research strategy currently employed in the alcohol field. Enthusiasm for genetic and genomic approaches to studying behavior at the National Institutes of Health in the 1990s drove increased funding for research on addiction with animals, especially mice (Hyman Reference Hyman, Parens, Chapman and Press2006), and availability of standardized inbred mouse strains is considered crucial for research in the field (see e.g. Wahlsten et al. Reference Wahlsten, Bachmanov, Finn and Crabbe2006, who compare results from two-bottle choice experiments conducted fifty years ago with contemporary results). Nevertheless, there are many instances in which alcohol researchers approach standardization as more than just a problem of creating a standard genetic organism. The large variability in alcohol-related traits in genetically heterogeneous populations of mice and rats has allowed researchers to use selective breeding techniques to develop genetic animal models to address questions about particular facets of alcoholism; such as the selectively bred “preferring” and “non-preferring” rats that are used as models of alcohol preference, and selectively bred lines of mice that are prone or resistant to having seizures during alcohol withdrawal. Further, researchers have critiqued the tendency of the mouse research community to use only a few standard genetic mouse strains (such as the widely used C57 Black 6 strain, or the 129 strain which is particularly amenable to genetic manipulation), pointing out that although strains such as the high-drinking Black 6 have been extensively studied for addiction behaviors many strains of mice have never even been tested for alcohol preference (Yoneyama et al. Reference Yoneyama, Crabbe, Ford, Murillo and Finn2008).

Changing expectations in the mouse research community more broadly about the “complexity” of behavioral traits have given rise to different ways of thinking about the standardization of genes, organisms, and environments. The Complex Trait Consortium, an interdisciplinary group of researchers from the mouse community, has argued that “existing and proposed mouse resources. . . are optimized to study the actions of isolated genetic loci on a fixed background,” and that these standard organisms are “less effective for studying intact polygenic networks and interactions among genes, environments, pathogens and other factors” (Churchill et al. Reference Churchill, Airey, Allayee, Angel, Attie, Beatty and Beavis2004, 1133). They proposed the development of new lines of inbred mice, made by breeding eight different strains, which would provide “a far broader representation of genetic variation in natural populations than current mapping resources” (Jackson Laboratory Reference Laboratory2003). Behavioral researchers have similarly argued that the approach of comparing a mutant to a standard model organism is unlikely to succeed in behavioral studies because of the small effects of individual genes and the presence of gene-gene interactions. Crawley and her colleagues note that the simplest way to achieve genetic standardization in knockout experiments “is to derive and maintain mutations in an isogenic genetic background, a standard practice in other model organisms such as yeast, Drosophila, and C. elegans”;Footnote ¹¹ but they reject this strategy as an appropriate one for behavioral researchers to employ because “behavioral characteristics of certain isogenic strains could overshadow the effects of the targeted mutation” (Crawley et al. Reference Crawley, Belknap, Collins, Crabbe, Frankel, Henderson and Hitzemann1997, 108). Instead, they recommend choosing different strains for knockout studies depending on the trait studied; for example, selecting a high drinking mouse strain such as the C57 when studying a gene that is expected to reduce drinking in order to be able to see a behavioral change more clearly. Some researchers have even gone so far as to argue against the development of generic standards for housing rodents to be used in behavioral studies, contending that simply adopting a single standard across laboratories may be counterproductive because it could limit the range of environmental contexts in which experiments are performed and thus result in the black-boxing of housing arrangements when interpreting results (Würbel Reference Würbel2000 and Reference Würbel2002). As one researcher puts it, the very idea of using organisms living in standard environments to model human behavior “fails to acknowledge the fundamental biological significance of the interactive nature of gene-environment relationships underlying most behavioral phenotypes” (Würbel Reference Würbel2002, 5). Richter, Garner, and Würbel (Reference Richter, Garner and Würbel2009) argue that researchers should instead adopt the practice of systematically varying the environment in order to observe how behavior changes depending on the context.

Finally, researchers have made productive use of existing variations in both animal genomes and the laboratory environment. In the alcohol field, comparisons between different strains of mice have long been used as a way to provide insight into the genetics of alcoholism, and researchers have even used the small genetic differences between substrains of C57 mice to identify candidate genes that influence alcohol consumption (Mulligan et al. Reference Mulligan, Ponomarev, Boehm II, Owen, Levin, Berman and Blednov2008). In the Mouse Phenome Project, another interdisciplinary collaboration between mouse researchers from many fields, a database has been established to collect information on the results of common mouse experiments and the laboratory conditions under which they were performed, in order to look at the influence of the environment on genotype. By collecting information on existing variation in experimental protocols and laboratory environments, researchers hope to produce a dataset that can be used to investigate what environmental factors influence mouse behavior (Bogue and Grubb Reference Bogue and Grubb2004).