The current study shows that young adult HDs not only experience reduced subjective effects of alcohol, but also demonstrate a blunted response in the brain's reward system. Alcohol significantly activated the NAcc in the SDs, but not the HDs. In both groups, NAcc activation significantly correlated with self-rated intoxication. Fearful faces significantly activated the amygdala and other temporal regions only in SDs; this activation was modulated by alcohol. HDs demonstrated no differences in their response to fearful vs neutral faces during the alcohol or placebo sessions.
HDs Report Reduced Subjective Effects of Alcohol
On the BAES, HDs reported significantly lower scores on stimulation and slightly lower scores on sedation. They also reported reduced DEQ ratings on all questions, except the question ‘Would you like more of what you received?' This is consistent with previous reports (King et al, 2002
) that found HDs reported only slightly higher scores on ‘want more' while reporting lower scores of ‘feel effects' and ‘feel high.' This suggests that HDs may be at particularly high risk for the development for alcohol use disorders, as the desire to continue drinking despite demonstrating reduced positive effects of alcohol can result in consuming greater amounts to reach desired effects.
These differences in self-reported effects of alcohol are indicative of tolerance, defined as the need for increasing amounts of alcohol in order to achieve intoxication or a desired effect, or a markedly diminished effect with continuous use of the same amount of alcohol (APA, 1994). In the HDs, binge drinking, or consumption of multiple drinks in rapid succession, may have led to increased tolerance. Tolerance has been demonstrated among HDs in a variety of self-report studies, which have demonstrated that heavier drinking behavior is correlated with lower subjective responses to alcohol effects (Portans et al, 1989
; Heath and Martin, 1991
; Ramchandani et al, 2002
Though acquired tolerance is one explanation for reduced subjective effects in HDs, another possibility is that the decreased effects of alcohol in the HDs may have been pre-existing. Low response to alcohol is thought to be a risk factor for alcoholism, as the need to ingest more drinks to achieve a desired affect leads to heavier drinking patterns among these individuals (Schuckit et al, 2009
). This hypothesis suggests that the HDs in this study may drink more heavily because of an inherently lower sensitivity of the brain's reward circuitry to alcohol.
HDs Demonstrate Reduced Activation to Alcohol in the Nucleus Accumbens
Consistent with our previous report (Gilman et al, 2008
), the SDs in this study showed greater activation to alcohol than to placebo in several brain regions, including the left nucleus accumbens. In contrast, accumbens activation to alcohol was not seen in the HDs at a BAC of 0.08
g%. There are several explanations for the reduced activation of the NAcc to alcohol in HDs. In vivo
microdialysis experiments in rats have shown that oral self-administration of alcohol causes significant increases in extracellular dopamine (DA) levels in the NAcc (Weiss et al, 1993
; Gonzales and Weiss, 1998
; Melendez et al, 2002
), and that this increase in DA is reduced by repeated alcohol exposure (Smith and Weiss, 1999
). Franklin et al (2009
) showed that at moderate doses (eg, 1.0
g/kg), there was a significant desensitization of DA D2 receptors following alcohol exposure. In human PET scans, research has shown that alcoholics exhibit a decreased number of DA D2 receptors in the striatum compared with non-alcoholics (Volkow et al, 2009
). The reduced NAcc activation that we observe in the HDs could be related to either desensitization or to a reduced number of D2 receptors, as human DA function had been linked to magnitude of the BOLD signal in NAcc (Schott et al, 2008
The reduced BOLD response in the NAcc may be a cognitive effect as well as a pharmacological effect. Research has suggested that striatal activity may be modulated by cognitive processes. Yoder et al (2009
) conducted an experiment with IV alcohol in which they informed participants that they either would or would not receive alcohol. They found alcohol-induced DA release when participants were not expecting alcohol, but not among subjects who expected and then received alcohol (Yoder et al, 2005
). The authors attribute this effect to reward prediction error (Schultz, 2002
), which states that DA neurons in the ventral striatum fire when an individual does not expect a reward and one is delivered. Yoder et al
suggest that with repeated drinking, the perceived value of alcohol may diminish and not match the expected rewarding effects from earlier drinking experiences. In the real world, HDs may respond by consuming larger quantities of alcohol, but in our controlled experiment, blood alcohol levels were clamped, which may have created a negative prediction error (ie, they expected to feel intoxicated but their BAC was not high enough). The alterations in expectation combined with delivery of alcohol's euphoric effects at increasingly high BACs may drive further heavy drinking behavior.
Although we were able to detect clear differences between SDs and HDs in the NAcc activation to alcohol, the number of drinks per week did not directly correlate with activation in the NAcc. This indicates that the neural response to alcohol is complex, and may be dependent on several factors, such as differences in expectation, personality characteristics, or genetic predisposition, that may not be reflected in a time-line follow back measure of alcohol consumption.
SDs, but not HDs, Show Decreased Amygdala Activity to Fearful Faces
Few neuroimaging studies have examined the effect of acute alcohol on emotional processing. A recent alcohol challenge study also reported reduced amygdala activity to fearful faces under intoxication (Sripada et al, 2011
). Most studies of alcohol and emotion, however, have investigated subjective, and not neural, responses to alcohol, and results have been inconclusive. In an experiment in which participants received either alcohol or placebo before viewing negative, positive, and neutral pictures, alcohol diminished the magnitude of the startle response and the skin conductance response regardless of the valence of the stimuli (Stritzke et al, 1995
), suggesting that alcohol diminished overall emotional reactivity. In a similar study by Gabel et al (1980)
, healthy participants viewed negative, positive, and neutral pictures and were then given the choice to drink alcohol. Both negative and positive stimuli resulted in increased arousal, but alcohol consumption was highest after the participants had viewed the positive slides (Gabel et al, 1980
). Curtin et al (1998)
used the startle reflex as a measure of the effect of alcohol in a stressful situation. Sober and intoxicated individuals were presented with cues that signaled either safety or a threat of electric shock. Autonomic activity, including SCR, heart rate, EMG, and magnitude of startle responses, were all decreased in the intoxicated condition, but no interaction was found between intoxication and threat condition (Curtin et al, 1998
). These studies, therefore, have not demonstrated a specific anxiolytic interaction between alcohol and emotional cues, but suggest that alcohol may modulate emotion through ‘dampening' of emotional reactivity to any emotion. This interpretation could indicate that the modulation of amygdala reactivity to fearful faces in the SDs is not specific to negative emotion. Future studies can present stimuli of different emotional categories (ie, happy, sad, disgust, surprise) to better explain alcohol's influence on different types of emotional processing.
HDs do not Demonstrate Significant Limbic Activity in Response to Fearful Faces
We replicated our earlier results (Gilman et al, 2008
) that showed that alcohol modulated brain activity to fearful faces in the amygdala and other frontal and temporal brain regions in SDs. During the placebo condition, fearful compared with neutral faces activated several higher-order visual regions related to emotion in the SDs, in addition to the bilateral amygdala. Other studies have reported similar results (Devinsky et al, 1995
; Phillips et al, 2003
; Vuilleumier, 2005
). In SDs, the increased response to the fearful faces that we observed in the placebo condition was not observed during the alcohol condition, suggesting that alcohol modulated the neural response to a threatening stimulus.
Fearful faces did not activate the amygdala or other temporal regions during either the placebo or alcohol session in HDs. Research has shown that young adults at risk for alcoholism show reduced amygdala activity to fearful faces compared with low-risk subjects (Glahn et al, 2007
). These high-risk individuals also show blunted cortisol responses to stress (Sorocco et al, 2006
). Hypoactivation of the brain's threat-detection circuitry may confer a greater risk for developing alcoholism, suggesting that deficient amygdala activity may either be a risk factor for, or may be a result of, heavy drinking.
Limitations and Conclusions
This study has several limitations. Our sample size is not large enough to allow us to look for higher-order interactions such as how family history of alcoholism, gender, or personality variables affect the neural response to alcohol. Second, we used only one level of alcohol exposure, which affected both groups differently. Future studies could expose HDs to higher BACs, equating subjective alcohol effects, and investigate whether differences in brain responses remain. Third, we cannot determine if HDs exhibit a reduced response to alcohol as a result of acquired tolerance or pre-existing lower sensitivity. This limitation is present in most studies comparing HDs to controls and is a challenge for future research. In addition, because this was a passive viewing task, there are no measures of task performance, and therefore, we do not know if the SDs were intoxicated to the point of altered perception. Finally, though the groups were not significantly different in smoking status, there were more smokers in the HD group, which could have affected activation of brain reward systems.
A caveat of this analysis is that significant three-way interactions between group, alcohol, and emotion were not observed. There were also no significant interactions between group and emotion, or emotion and alcohol. This negative finding may be a limitation of using factorial ANOVAs with voxel-wise data. Voxel-wise 3-way interactions are often difficult to interpret, and there is no standard approach to correcting for multiple comparisons within a whole-brain ANOVA. Furthermore, data have to meet certain assumptions, such as compound symmetry (equal correlation across levels of a factor) and homoscedasticity (equal variance across the levels of a between-subject factor, such as group) (Brown and Forsythe, 1974
), and patient groups often exhibit more variance than controls. In addition, voxel-wise three-way ANOVAs often produce ‘edge' artifacts with inflated t
values. Linear contrasts may offer a solution to directly investigate effects of interest without limitations of compound symmetry, homoscedasticity, and unbalanced designs.
This is the first study to detect a difference in the neural response to alcohol between heavy and SDs, and has clinical implications. Many recovering alcoholics cannot return to moderate levels of drinking, and one explanation is that the brain's reward system may be permanently altered by years of heavy drinking. Another possibility is that these individuals have pre-existing deficiencies in the brain's reward systems, and it therefore will take a greater amount of alcohol for them to achieve the desired effect of the alcohol. This study also has implications for the field of drug abuse, because it suggests that heavy drug users may have lower sensitivity and/or increased tolerance to the drug of abuse, which would make characterization of drug effects in heavy users challenging.