personality factors that relate to drinking. The chapter continues with a discussion of social processes and emphasizes research on the role of the family and peers, and on media and advertising. Because none of these influences occurs in isolation, the chapter takes an integrative approach that recognizes the interrelationships among factors. In particular, developmental issues and implications are highlighted throughout the presentation; they are integrated into the discussion rather than appearing in a discrete section. Additional discussion of developmental issues appears in chapter 3. Psychological Processes in Central to alcohol research is the basic question, Central to alcohol research is the basic question, "Why do people drink?" Negative Reinforcement Models Negative reinforcement models represent explanations of drinking motivation that involve the use of alcohol for relief of unpleasant feelings. Several such explanations have been postulated, including the tension reduction hypothesis and a recent refinement of the hypothesis called stress response dampening. Tension reduction hypothesis Over the past 30 years, the most popular theory of the motivation to drink has been the tension reduction hypothesis (Conger 1956). This theory suggests that alcohol is rewarding because it relieves tension, which is unpleasant. Two research issues are derived from this hypothesis: one involves the effect of alcohol consumption in reducing stress, and the other involves the role of stress in increasing alcohol consumption (Cappell and Greeley 1987; Cappell and Herman 1972). These are important research questions because if stress provides motivation for alcohol consumption, then severe or chronic stress might be associated with abusive drinking and with the development of alcohol dependence. Although seemingly straightforward, these research issues actually involve a host of difficult definitional and research design questions. Cappell and Greeley (1987) conducted an historical and conceptual review of the tension reduction hypothesis. Rooted in theory aimed at understanding the reinforcement effects that apply equally to positive and negative motivation, research concerning alcohol has tended to focus on negative conditions. As first proposed, the tension reduction hypothesis was based on animal research in which approach-avoidance conflicts were employed. This research paradigm sets up stressful situations in which the animal must choose whether or not to pursue a reward (e.g., food) when the pursuit is associated with an adverse event (i.e., a stressor). In highly stressful situations, pursuit of the reward is suppressed; in the presence of a tension-reducing substance, the tendency to avoid the stressor is overcome. In these animal experiments, Cappell and Greeley report that alcohol reliably reduced responses to stress, thereby suggesting that alcohol relieved tension. Tests of active avoidance and escape behavior in animals (i.e., experiments in which the animal avoids the stressor by performing rather than by inhibiting an action) produced mixed findings for alcohol's tensionreducing properties, although the lack of consistency in results may in part be methodological (Cappell and Herman 1972). Experimentally induced approach-avoidance conflict in humans has also suggested tension-reducing properties of alcohol, although human studies of response to social anxiety in the laboratory have produced more mixed results (Cappell and Greeley 1987). Studies of humans outside the laboratory are less consistently supportive of the tension reduction hypothesis. Some retrospective studies, in which individuals are asked to report about past drinking rates and stressful experiences or tension, show correlations between alcohol use and economic, job, and marital stress (Timmer et al. 1985), while others demonstrate support for the hypothesis only among certain subgroups, such as those living in rural areas (Neff and Husaini 1985). Young people, in particular, report greater alcohol use coincident with greater difficulty in social relationships (Labouvie 1986), stressful life events (Chassin et al. 1988; Wills 1985), and family conflict (Baer et al. 1987), although these relationships are generally small. These findings suggest the influence of developmental and social context factors in the tension-reducing effects of alcohol. Pohorecky recently reviewed the literature pertaining to alcohol and stress for animals (1990) and humans (1991), concluding that prospective designs, in which changes in stressrelated alcohol use are studied over time, have yielded only mixed support for theories of tension reduction. In a 3-year study, use of alcohol by air traffic controllers was found to be related to stress (DeFrank et al. 1987). When high school students were assessed at both the beginning and the end of the school year, Chassin et al. (1988) found that rates of alcohol use were associated with negative life events, among other factors. Yet, evaluation of a community sample (Cronkite and Moos 1984) suggested that alcohol use was affected more by cultural differences than by stressful life events, and studies of undergraduate college populations have seldom revealed relationships between stress and alcohol use (Brennan et al. 1986). These differential findings may indicate, as Cappell and Greeley (1987) concluded, that the tension reduction hypothesis is limited in its consideration of drinking as motivated by a single factor; tension reduction is but one of alcohol's many effects. A second limitation noted by Cappell and Greeley is the pharmacological effect of alcohol relative to dose. At certain doses, tension reduction may occur, but at other doses, tension may be enhanced. Thus, it is likely that tension reduction will occur only some of the time and perhaps only for certain individuals. More recent research on the relationship between stress and alcohol use has moved toward assessment of multiple explanatory factors to identify more specific predictive relationships. Such an approach has been used by Cooper et al. (1992), who hypothesized that stress may be associated with alcohol use among certain vulnerable individuals. In this research, vulnerability factors were defined as limited coping skills (i.e., use of avoidant coping) and holding strong positive expectancies about alcohol's effects. Analyses confirmed the authors' predictions. Indeed, among male subjects having both vulnerability factors, stressors accounted for 35 percent of the variance in alcohol use; stress this theory, since alcohol serves to inhibit the stress response, it might be especially reinforcing when consumed in times of stress. Persons experiencing stress-response-dampening effects are therefore likely to drink with greater frequency and perhaps also to consume greater amounts when they are under duress. Sher noted that the stress-response-dampening formulation differs from the tension reduction hypothesis in that alcohol use is seen as reinforced because of the inhibition of a specific physiological response to an explicit stressor. This specificity suggests the enhanced importance of certain drinking contexts as well as individual differences in reward potential. The management of stress is only one possible motivation for the consumption of alcohol; individuals may also differ in the degree of pleasure experienced as a result of drinking. Existing research suggests that stress-responsedampening effects may be stronger for certain individuals than for others. For example, Sher and Levenson (1982) demonstrated that personality styles consistent with alcoholism risk (i.e., behavioral undercontrol) were associated with greater stress-response-dampening effects among young men. Subsequent research (Finn and Pihl 1987; Levenson et al. 1987) documented stressresponse-dampening effects in both daughters and sons of male alcoholics, persons who are at heightened risk for alcohol problems. Finn and Pihl (1987), using a physiological measure (cardiovascular function), found great reductions in heart rate, suggesting stress reduction by alcohol in high-risk men. In a recent review of the literature, Pihl and colleagues (1990) cited a number of studies suggesting increased sensitivity of sons of male alcoholics to the stress dampening effects of alcohol as well as the possibility of greater susceptibility to its reinforcing effects, but the reasons for these findings remain unclear. Other Models of Alcohol The management of stress is only one possible motivation for the consumption of alcohol; individuals may also differ in the degree of pleasure experienced as a result of drinking. Newlin and Thomson (1990) recently summarized data on studies of alcohol administration to sons of alcoholics. Their review suggests that risk for alcoholism is related to increased sensitivity to alcohol's effects when blood alcohol levels are rising and increased (acute) tolerance to alcohol effects when blood alcohol levels are falling. Increasing blood alcohol levels are typically experienced as stimulating, whereas declining levels are typically experienced as depressing. Newlin and Thomson concluded that sons of alcoholics may experience enhanced pleasure during periods of rising blood alcohol and diminution of the unpleasant effects during periods of falling blood alcohol levels. Although more data are needed to test this concept, the idea is a very promising direction for research on drinking motivation. Further, there is no reason to assume that these individual differences in response to alcohol are restricted to sons of alcoholics. Heightened pleasure, reduced discomfort (greater tolerance), and greater stress response dampening are likely to be distributed across the population and to have greater prevalence in high-risk groups (Baer 1991; Sher et al. 1991). Cognitive Models of Motivation The general inconsistency in findings related to the direct stress-reducing effects of alcohol has led not only to work focusing on individual differences in tension reduction, but also to work on psychological mechanisms that might mediate alcohol's effects and thereby influence drinking motivation. Such explanatory models attempt to account for alcohol's sometimes stressreducing properties, as well as for other perceived effects of alcohol, such as enhancement of positive mood. Alcohol myopia Steele and Josephs (1988; Josephs and Steele 1990) found evidence suggesting that alcohol's tension-reducing effect can be mediated by the combination of cognitive and environmental factors: alcohol's effects on the ability to process information and the presence or absence of environmental distractors. Their research showed that distracting tasks relieved anxiety more effectively after alcohol had been consumed in comparison with times when no alcohol had been consumed. This finding was interpreted as suggesting that alcohol impairs cognitive functioning in information processing and constricts atten tion to the most pressing internal and external events. Thus, in the presence of distracting events, alcohol relieves stress by allowing the focus to be on a distractor (i.e., the more pressing event) rather than on an anticipated stressor. Conversely, in the absence of salient (pressing) distractors, anxiety is increased because attention has been directed toward the stressful event. This phenomenon has been called “alcohol myopia," which is defined as “a state of shortsightedness in which we process fewer cues less well" (Steele and Josephs 1988, page 197). The manifestations of alcohol myopia in increasing or decreasing euphoria or tension depend on the drinking context. For example, results of research using distraction tasks of varying complexity (Josephs and Steele 1990) are consistent with the attention-mediation interpretation, in that more complex tasks (i.e., tasks that demand more attention) had greater anxietyreducing effects than tasks requiring less attention. The authors also tested the effect of expectancies about alcohol's impact on performance and found that expectancy did not play a role in these findings. Steele and Josephs (1990) reviewed studies showing that social behaviors such as gambling, aggression, and self-disclosure occurred more frequently after intoxication. They concluded that alcohol can narrow the focus of attention and thereby decrease internal conflict felt by individuals. It blocks inhibitions (which tend to be more distant in consciousness), thereby making social behaviors more extreme when immediate cues are compelling. Indeed, the blocking of inhibition seen in subjects in studies reviewed by Steele and Josephs (1990) is similar to responses described by Cappell and Greeley (1987) in experiments involving approach-avoidance conflicts using animal models. Blocking of inhibitions and reducing of internal conflict are hypothesized by Steele and Josephs (1990) to inflate selfevaluations and lead to feelings of power and control, feelings that may be sought by some individuals through drinking. Future research employing this promising theory is likely to address the relationship of alcohol myopia to other known risk factors for alcohol problems. Expectancies An active area of research over the past 10 years has involved expectancies about alcohol's effects. Marlatt (1987) has suggested that alcohol is commonly perceived as a "magic elixir," capable of promoting social skills, sexual pleasure, confidence, power, and aggression. Advertising campaigns are often consistent with such a conceptualization. These beliefs about the effects of alcohol have been termed "expectancies" (Critchlow 1986; Goldman et al. 1987; Marlatt 1987). Expectancies relate not only to beliefs about physical and affective outcomes of drinking, but also to broader issues of sociability and social behavior (Leigh 1989). Expectancies relate not only to beliefs about physical and affective outcomes of drinking, but also to broader issues of sociability and social behavior. Studies examining nondrinking populations are particularly important for understanding the development of expectancies about alcohol's effects. Those who do not drink clearly have expectations, but these expectations are often based on factors other than personal experience with alcohol consumption. For example, children appear to have expectations about alcohol's effects long before drinking actually occurs: Miller et al. (1990) investigated alcohol expectancies among young children in first through fifth grades, identifying a developmental pattern in which positive beliefs about alcohol's effects increased with age. The most marked increase occurred among 8- to 10-year-olds in grades 3 and 4, a time consistent with developmental changes related to both receptive and expressive communication abilities. Children of this age thus might benefit from prevention efforts aimed at reducing risk for later alcohol problems. Further evidence of alcohol awareness among very young children was provided by Noll et al. (1990), who asked preschoolers to identify alcoholic beverages during a smelling game. A surprising proportion of preschoolers could identify alcoholic beverages and knew that only adults should use them. Success at the smelling game was related to heavier parental drinking rates. Developmental implications of these findings have been noted by Zucker and Fitzgerald (1991), in that the results linking parental drinking with early childhood knowledge about alcohol suggest the possibility of enhanced risk due to premature formation of understandings about and interest in alcohol. Expectancy studies have revealed that those who drink heavily tend to expect positive effects from alcohol, a finding noted in adolescent (Christiansen et al. 1982) as well as adult samples (Brown et al. 1985). Individuals also tend to make distinctions about alcohol's effects in small versus moderate doses (Southwick et al. 1981) and about alcohol's effect on oneself versus others (Rohsenow 1983). Expectancies among adolescents have been related to parental drinking rates (Brown et al. 1987) and family history of alcoholism (Mann et al. 1987). A recent study by Sher et al. (1991) examined the role of expectancies in college freshmen and found that children of alcoholics (COAS) reported stronger positive expectancies than non-COAs in three areas: tension reduction, social lubrication, and performance enhancement. The relationship between expectancies and family history proved to be quite complex in this study, however, when other characteristics thought to be associated with risk were considered. In fact, there was no direct relationship between family history and expectancies; family history influenced behavioral undercontrol, which in turn influenced expectancies. This result will be discussed in greater detail later in this chapter. A central challenge to research in alcohol expectancies has been to demonstrate that expectancies actually predict future behavior and do not simply reflect personal experience with alcohol. A central challenge to research in alcohol expectancies has been to demonstrate that expectancies actually predict future behavior and do not simply reflect personal experience with alcohol (Leigh 1989). Despite past research that did not support the predictive power of expectancies (e.g., Fromme et al. 1986), several recent studies are noteworthy. Christiansen et al. (1989) measured beliefs about alcohol in seventh and eighth grade students. Assessment of drinking status and rates 1 year later revealed that five of seven expectancy scores denoting specific beliefs about alcohol were predictive of initiation and rate of use. Those children deemed to be at high risk to initiate alcohol use believed alcohol would enhance social pleasure and improve cognitive and motor functioning. The importance of beliefs of performance enhancement is consistent with the findings of Sher et al. (1991). Stacy et al. (1990) similarly predicted drinking rates and the intention to drink from measures of expectancies over 4 weeks in samples of college students. This study examined differences in beliefs about negative and positive effects of alcohol and noted significant prediction only from perceptions of positive effects. Stacy et al. (1990) and Leigh (1989) suggested that beliefs about alcohol's effects that are most accessible in memory should be better predictors of behavior. Positive expectancies might be quite important in predicting alcohol use: Positive experiences, being more immediate, may have a greater effect on learning about drinking and its relevant effects than the more delayed negative experiences. The negative expectancies may be significant predictors if drinkers see the probability of negative outcomes as low (Leigh 1989). Stacy et al. (1991) investigated theoretical underpinnings of the expectancy construct and its predictive ability regarding alcohol and marijuana use from adolescence to adulthood. In a carefully controlled study using a large sample of young people who first completed questionnaires at approximately age 18 and again 9 years later, measures of anticipated effects of alcohol and marijuana were found to be predictive of later alcohol and other drug involvement. With respect to alcohol-specific findings, expected effects of alcohol during adolescence (termed "alcohol motivation") predicted alcohol/drug problems as an adult. It is noteworthy that adolescent alcohol/drug use did not predict later beliefs about alcohol's effects in this sample. The authors concluded that expectancies constitute strong motivational factors for behavior, but that support for reciprocal effects of adolescent behavior on adult expectancies was limited. In addition to studies based on the alcohol myopia construct, cognitive processing of information has been investigated in relation to expectancies. Because expectancies, like other beliefs, represent stored memories of things that are experienced, observed, and leamed, Goldman et al. (1991) have proposed studying the memory networks in which expectancies are thought to reside as a way of understanding how these beliefs about alcohol's effects may guide decisionmaking about alcohol use. The conceptualization of Goldman et al. focuses on the “semantic" portion of memory, where abstract knowledge and rules are thought to be stored. Using multidimensional scaling techniques, Rather et al. (1992) developed a preliminary alcohol "semantic network" and then compared net |