Neural Underpinnings of Expecting Alcohol: Placebo Alcohol Administration Alters Nucleus Accumbens Resting State Functional Connectivity

Abstract

Background:

Using balanced placebo designs, seminal alcohol administration research has shown individuals’ beliefs about whether they have consumed alcohol, irrespective of the actual presence of alcohol, can determine level of alcohol consumption and impact social behavior. Despite the known effect of expecting alcohol on drinking behavior, few studies have used the placebo manipulation to directly investigate the neural underpinnings of the expectancy-related effects that occur following perceived alcohol consumption in humans. The present paper examined placebo responses in the laboratory to better understand the neural basis for the psychological phenomenon of expectancies.

Methods:

As part of a larger within-subjects study design, healthy young adults (N=22, age_mean+SD=23±1) completed resting state fMRI scans and measures of subjective response before and after consuming placebo beverages. Effect of placebo beverage consumption (pre- versus post-beverage consumption) on functional connectivity within prefrontal cortical networks was examined using the CONN Toolbox. Relations between perceived subjective response to alcohol with functional connectivity response following placebo beverage consumption were examined.

Results:

Compared to pre-beverage scan, placebo beverage consumption was associated with increased positive functional connectivity between right nucleus accumbens – ventromedial prefrontal cortex and subcallosal cingulate cortex (p_FDR<0.05). Subjective ratings of intoxication (i.e., feeling ‘drunk’) positively correlated with placebo beverage-related increases in nucleus accumbens – subcallosal cingulate cortex functional connectivity.

Conclusion:

Results suggest placebo response to alcohol is associated with increased functional connectivity within a key reward network (nucleus accumbens – ventromedial prefrontal cortex and subcallosal cingulate cortex) and puts forth a mechanism by which alcohol expectancies may contribute to the subjective experience of intoxication.

1. INTRODUCTION

In a seminal study, Marlatt and colleagues (1973) found that an individual’s beliefs, rather than a biological trigger from the presence of alcohol, lead to increased alcohol consumption (1). Applying the balanced placebo design to alcohol, researchers fully crossed what participants expected to receive (Alcohol/No Alcohol) with what they actually received (Alcohol/No Alcohol) in an effort to separate the biological effects of alcohol from the psychology effects of what a drinker expected to happen when they drank. Using a blinded taste test to measure amount of beverage consumed, participants’ beliefs about whether they received alcohol, rather than the actual presence of alcohol, determined their higher level of consumption. The potential influence of non-pharmacological factors on alcohol-seeking behavior presented an exciting opportunity for the development of novel intervention strategies.

Placebo response is thought to be driven by alcohol expectancies – that is, a person’s beliefs about what happens when drinking alcohol. Self-reported measures of alcohol outcome expectancies consistently show a persons’ beliefs about the effects of alcohol correlate with the amount of alcohol people consume (3–5). This literature indicates the critical role of alcohol expectancies on drinking behavior. Uncovering the neural processes that underlie alcohol expectancies marks a critical next step that can inform novel interventions.

Functional MRI (fMRI) studies have examined the neural correlates of alcohol expectancies using self-reported measures. This work suggests a neural basis for alcohol expectancies, i.e., alcohol expectancies positively relate to limbic and striatal functional connectivity during resting (6–8). Placebo alcohol administration combined with fMRI marks a rigorous method that can extend this work by directly capturing alcohol expectancies in the lab while systematically controlling alcohol cue exposure and drinking setting. Additionally, placebo-related change in neural systems can be distinguished from baseline neural phenotypes that may indirectly relate to alcohol expectancies. Alcohol administration methods have been used in combination with fMRI to examine the effects of alcohol on brain function (9–14). This work, however, has typically examined the contrast of alcohol versus placebo conditions with the goal of isolating the acute pharmacological effects of alcohol. Few studies have used the placebo manipulation to directly investigate the neural underpinnings of the expectancy-related effects that occur following perceived alcohol consumption. Using a balanced placebo design, Gundersen and colleagues (2008) reported the pharmacological and psychological effects of alcohol had distinct effects on anterior cingulate cortex (ACC) activation during a working memory task (15). This study used a between-subjects design, did not collect baseline fMRI (to investigate how placebo manipulation alters brain function within an individual), and focused on males.

The present paper revisits the placebo response in the laboratory to better understand the neural basis for the psychological phenomenon of expectancies. We examined brain function following placebo beverage consumption, compared to pre-beverage consumption, in healthy young adults using resting state fMRI. Resting state functional connectivity is sensitive to the acute effects of alcohol (13, 16–18) and can be effectively used to examine the neural correlates of mood/subjective feelings during acute alcohol intoxication in the absence of task-induced changes in mood/stimulation (i.e., negative mood induction during emotional task, attentional demands of working memory task). The data reported here were obtained from a larger study design using within-subjects placebo-controlled alcohol administration methods to investigate subjective and neural response to alcohol (NCT04063384). We hypothesized placebo beverage consumption would be associated with an increase in resting state functional connectivity within prefrontal cortical (PFC) networks—including PFC, ACC, amygdala, insula, and striatum—that govern reward and emotion processes. We predicted we would see changes within these PFC networks because 1) alcohol expectancies (measured with self-report surveys) relate to variation in functional connectivity within these PFC networks (6–8), 2) alcohol alters PFC network activity and connectivity (9–14), and 3) variation in how these networks respond to alcohol is associated with differences in the subjective experience of intoxication (16, 19–23). We further hypothesized that greater increase in functional connectivity within PFC networks would relate to greater subjective ratings of intoxication to placebo beverage (referred to as ‘subjective response’).

2. METHODS AND MATERIALS

2.1. Participants and clinical data collection

Twenty-six healthy young adults (ages 21 – 26 years) enrolled in the study and completed a fMRI session prior to placebo beverage consumption (pre-beverage) and another fMRI session following placebo beverage consumption (post-beverage). The first participant was excluded due to a subsequent change in fMRI scanning parameters to optimize image acquisition. Participants were recruited through advertising on the University of Texas at Austin campus and in the surrounding areas. Telephone screening was used to obtain participant background information. Exclusion criteria included possible alcohol dependence, defined as Alcohol Use Disorder Identification Test (AUDIT) score >15, ever being in an abstinence-oriented treatment program for alcohol use, reporting wanting to quit drinking but not being able to, any medical, religious, or other reasons for not drinking alcohol, adverse reaction to alcoholic beverages, reporting never consuming four (men) or three (women) or more drinks on a drinking occasion in the past 12 months, unwillingness to have a friend/family member drive them home after alcohol sessions, self-reported pregnancy or nursing, or positive pregnancy test. Additional exclusion criteria included history of major medical illness with possible neurological or central nervous system outcomes, a medical condition or previous surgery preventing participation in magnetic resonance imaging (MRI) scanning, or a history of heart attack, heart trouble, high blood pressure, diabetes, or liver disease.

All eligible participants provided written informed consent and completed clinical assessment. The Structured Clinical Interview for DSM-5 Research Version (24) was used to confirm presence/absence of diagnosis of major depression, bipolar disorder, anxiety disorders, alcohol/substance use disorders, and history of suicide attempt at enrollment. Additional exclusion criteria included history of mood, psychosis, or anxiety disorders, lifetime suicide attempt, and use of psychotropic medications for >1 month. The Wechsler Abbreviated Scale of Intelligence-Second Edition (WASI-II) was used as a measure of full-scale intelligence quotient (FSIQ-2) (25). IQ<85 was an exclusion criterion. Family history of problematic drinking was assessed using the Family History - Research Diagnostic Criteria - Epidemiological Version (26). Urinalysis was conducted on the days of the scans and beverage administration to assess for pregnancy and substance use. Participants were asked to abstain from any alcohol or drug use for 24-hours preceding MRI scans and beverage administration. Participants were screened for a 0.000 g% breath alcohol concentration (BrAC) upon arrival for MRI and beverage session study visits. All study procedures were approved by the University of Texas at Austin Institutional Review Board.

2.2. Beverage administration procedures and placebo manipulation instructions

Following enrollment and pre-beverage MRI scan, participants completed a placebo beverage administration session. Figure 1 outlines experimental design and timeline of experimental procedures, including pre- and post-beverage consumption resting state fMRI scans, measures of subjective response, and BrAC testing. All beverage sessions occurred after noon. Prior to study enrollment (during phone screening), participants were told that as part of the study they would come to our lab and consume beverages containing alcohol on two separate occasions. Participants were informed would not be dosed to exceed a BrAC higher than .08g%. Participants were also told they could not drive to the session since it is unsafe to drive a vehicle after consuming alcohol and were asked to fast from food for four hours prior to beverage consumption. Before beginning consumption of beverages, participants ate weight-adjusted, 1 calorie per pound snack of pretzels. An algorithm was used to calculate amount of alcohol that would be used if dosing each participant to a target peak breath alcohol concentration of .08g% based on the participants’ age, biological sex, height, and weight. Participants were given 10 minutes to consume each of the two beverages containing a 1:3 mixture of decarbonated tonic water (presented as vodka) to mixer (cranberry juice, diet cherry 7up, and lime juice). The decarbonated tonic water was served from a vodka bottle and drinks were mixed and poured in view of the participant (visual cue) as part of standard placebo manipulation procedures (27). Beverages were topped with an alcohol floater. Specifically, 95% alcohol (from a lime juice concentrate squeeze bottle) was added to beverages immediately before participants began drinking (gustatory cue). The amount of alcohol added to the beverage was minimal and not sufficient to generate a breath alcohol concentration >0.000. The table was also wiped with tequila (olfactory cue) before the participant entered the room for beverage consumption. A minimum of two research staff members were present during beverage procedures.

Figure 1.

Study design and timeline of experimental procedures

Following enrollment and clinical evaluation, participants completed resting state fMRI scans, measures of subjective response and breath alcohol concentration (BrAC) before and after placebo beverage consumption. Placebo beverage consumption occurred over a 20-minute period. Post-beverage BrAC testing, subjective response measures, and placebo manipulation check occurred 10 minutes after beverage consumption was complete (to match 10 minutes allowed for absorption if alcohol had been administered).

**In the overarching study design, alcohol and placebo conditions were counter-balanced; half of participants completed placebo beverage consumption on their first beverage session day, while half of participants completed placebo beverage consumption on their second beverage session day (there were, on average, two days between alcohol and placebo sessions) (sensitivity analyses were conducted to examine effects of order of placebo—i.e., before or after alcohol consumption—on findings).

2.3. Measures

2.3.1. Recent substance use

The Timeline Followback (TLFB), a researcher-administered measure, was used to obtain daily reports of alcohol use for 30 days prior to the date of baseline MRI assessment (28). Using a calendar marked with holidays and specific days, participants reported which days they consumed alcohol and the number of standard drinks they consumed on these days. Standard drinks were defined as 12 oz beer, 5 oz wine, or 1.5 oz liquor (straight or in a mixed drink). Total number of drinks consumed during the past 30 days, total number of drinking days during the past 30 days, and average number of drinks per drinking day during the past 30 days were calculated.

2.3.3. Breath alcohol concentration (BrAC)

Alco-Sensor IVs (Intoximeters, Inc., St. Kouis, MO) were used for BrAC testing prior to beverage administration (pre-beverage) to confirm a 0.000 BrAC at start of session. Although no participant had a BrAC >0.000 following placebo beverage consumption, we repeated BrAC testing after completion of beverage consumption (post-beverage) to enhance believability of the placebo manipulation. We timed post-beverage BrAC measurement collection during the placebo to be identical to timing of BrAC measurement collection during an alcohol session (which is collected 10 minutes following completion of beverages to allow for a 10-minute absorption period). Participants were shielded from BrAC readouts.

2.3.2. Subjective response and placebo manipulation check

A modified version of the Drug Effects Questionnaire (DEQ) (29) was used to assess subjective response. In this modified DEQ, participants used visual analog scale to indicate how much they were feeling (1) drunk, (2) the effects of alcohol, and (3) high on a scale from no change to extremely. These measures were collected prior to beverage administration (pre-beverage) and again following beverage procedures (30 minutes) (post-beverage). After the post-beverage subjective response assessment, participants were told, “We do not use standard glasses or mixers for research that are used in other bars. Please estimate the number of standard alcohol drinks you were served during this experiment.” Only participants who reported having consumed at least one standard alcoholic beverage were included in the final sample. Three individuals did not report they thought they had consumed alcohol and were excluded from further analyses. Participants entered the scanner immediately after completing the post-beverage subjective response measures. Table 1 details demographic, clinical, recent substance use and family history characteristics for the final sample (N=22). As sex differences in alcohol expectancies and their neural correlates have been observed (30, 31), supplementary table 1 presents demographic, clinical, and recent substance use and family history characteristics for the final sample stratified by biological sex.

Table 1.

Demographic, clinical, and family history characteristics

	Placebo Manipulation fMRI Sample (N=22)
Demographics	Mean Age (SD)	22.7 (1.4)
	Number of Females (%)a	11 (50)
	Number of Males (%)a	11 (50)
	Mean WASI-II FSIQ-2b (SD)	123 (13)
	Asian (%)c	7 (32)
	Non-Hispanic White (%)c	8 (36)
	Hispanic (%)c	5 (23)
	Mixed Race (%)c	2 (9)
	Body Mass Index (BMI)	22.4 (4.4)
Alcohol and Substance Use Characteristics d	Current Cannabis Use Disorder, mild (%)	1 (5)
	AUDITe (Past Year)	4.5 (3.6)
Past 30-Day Alcohol Use f	Total Drinks (SD)	19.6 (22.7)
	Total Drinking Days (SD)	5.7 (4.6)
	Drinks/Drinking Days (SD)	3.1 (2.1)
Positive Urinalysis Toxicology Screen	Tetrahydrocannabinol (%)	3 (14)
Family History	Problematic Alcohol Use (%)	1 (5)

Demographic, clinical, and family history characteristics for the final sample (participants who reported having consumed at least one standard alcoholic beverage following placebo beverage consumption).

Abbreviations: fMRI: functional magnetic resonance imaging; FSIQ, full-scale intelligence quotient; AUDIT, Alcohol Use Disorder Identification Test

^aCategories based on biological sex

^bFSIQ-2 represents the composite score for full-scale intelligence quotient comprising verbal comprehension and matrix reasoning subtests on the Wechsler Abbreviated Scale of Intelligence-Second Edition (WASI-II)

^cCategories based on participant self-identification

^dNo participants met criteria for current or past alcohol use disorder

^eAUDIT represents mean sum score

^fPast 30-day alcohol use was measured with the Timeline Followback (TLFB)

2.4. MRI acquisition and preprocessing

All imaging occurred at the University of Texas at Austin Biomedical Imaging Center with a 3-Tesla Siemens VIDA MR Scanner (Siemens, Erlangen, Germany) using a 64-channel head coil. Sagittal structural MRI images were acquired with a three-dimensional MPRAGE T1-weighted sequence with parameters: repetition time (TR)=2400ms, echo time (TE)=2.18ms, matrix=208×300×320, flip angle=8°; field of view=167mmx240mmx256mm, 0.8-mm slices and one average with isotropic voxel geometry (0.8×0.8×0.8 mm³).

Participants completed resting state fMRI prior to and following placebo beverage consumption. During a resting state fMRI scan, lasting six-minutes, participants viewed the Headspace Studios’ Inscapes video. Participants were asked to avoid sleeping or closing their eyes and to clear their mind of any intentional thought. The video depicts an inanimate object changing shape and has been used successfully in resting state fMRI studies (32). Differences in functional connectivity while viewing the Inscapes video, compared to more traditional resting state paradigms (when participant views a fixation point), have been reported (32); hence this study may not necessarily be capturing the brain “at rest”. We utilized the Inscapes video to help improve signal to noise, minimize motion, and decrease concerns that our placebo manipulation may change an individual’s ability to “rest” and not think about anything. FMRI data was acquired with a single-shot echo-planar imaging sequence aligned with the anterior-posterior commissure plane with multiband factor: 6, TR=778ms, TE=30ms, matrix=86×86, flip angle=52°, field of view=215×215mm², and 60 2.5-mm thick slices without gap (voxel size= 2.5×2.5×2.5mm³).

The CONN toolbox (www.nitrc.org/projects/conn) (33) for SPM12 was used to preprocess fMRI data and to calculate functional connectivity within PFC networks (34, 35) during resting state scans. Preprocessing steps included realignment and unwarping, slice-timing correction (to correct for within-scan acquisition time differences between slices), outlier detection, segmentation (into gray matter, white matter, and cerebrospinal fluid tissue classes) and normalization to MNI space, and smoothing using a 5mm FWHM Hamming filter (36). Volumes were examined for outliers using the Artifact Detection Tools toolbox in CONN. Thresholds were set for spikes for global signal >3 standard deviations from the mean and subject motion >0.5mm. There were no differences in motion between pre- and post-beverage scans. Denoising included aCompCor (anatomical component analysis correction) regression (37), followed by quadratic detrending and band-pass filtering (0.008 – 1 Hz). Nuisance regressors included cerebrospinal fluid and white matter, six motion parameters and their first derivative, scrubbing parameters, and session effects. A priori ROIs were defined using the FSL Harvard Oxford Atlas in the CONN toolbox and included ventromedial prefrontal cortex (vmPFC), subcallosal cingulate cortex (SCC), dorsal ACC, bilateral frontal orbital cortex, bilateral inferior frontal gyrus pars triangularis, bilateral inferior frontal gyrus pars opercularis, bilateral NAcc, bilateral amygdala, and bilateral insula. We focused on these ROIs because acute alcohol has been found to alter activity and connectivity within and between these regions (9–14).

2.5. Statistical analyses

Analyses were performed in the CONN Toolbox version 21.a and JMP Pro Statistical Software version 15.1.0.

2.5.1. Subjective response to placebo beverage

Within-subject repeated t-tests were used to examine effect of placebo beverage on BrAC and DEQ measures (pre-beverage versus post-beverage consumption).

2.5.2. Functional connectivity response to placebo beverage

Using the CONN Toolbox (33), we performed ROI-to-ROI bivariate correlation first-level analyses. At the second level, we used paired t-tests to compare ROI-to-ROI resting state functional connectivity between pre-beverage and post-beverage scans using ROI-based inferences. The ROI-based inferences method uses functional network connectivity multivariate parametric statistics to define a cluster of connections for each row of the ROI-to-ROI matrix, grouping all connections that arise from the same ROI as a new cluster of connections (see figure 2a for illustration of how connection clusters were defined). Then, a multivariate parametric general liner model analysis for all connections included in these clusters of connections was performed. Results were considered significant if they survived a false discovery rate (FDR) correction p_FDR<0.05 ROI-level threshold with a post-hoc uncorrected p<0.01 height (connection-level) threshold (38). Significant results are reported below.

Figure 2.

Region of Interest (ROI)-to-ROI group-level analysis in the CONN Toolbox

(A) Matrix display of group-level results (unthresholded). Each colored box represents the t-statistic for effect of placebo beverage consumption for each region of interest (ROI)-to-ROI connection (post-beverage > pre-beverage). Magenta color indicates increase in ROI-to-ROI functional connectivity from pre- to post-beverage consumption. Black brackets denote how connection clusters were defined by the ROI-based inferences method in CONN. (B) “Connectome Ring” illustrating ROI-to-ROI connections that show a significant main effect of placebo beverage consumption (post-beverage > pre-beverage). Results were considered significant if they survived a false discovery rate (FDR)-correction p_FDR<0.05 ROI-level threshold with a post-hoc uncorrected p<0.01 height (connection-level) threshold. Compared to pre-beverage, placebo beverage consumption was associated with increased functional connectivity in a cluster of connections between right nucleus accumbens (NAcc) with ventromedial prefrontal cortex (vmPFC) and subcallosal cingulate cortex (SCC) (F_(3,19)=10.1, p_FDR=0.005). (C) Inferior and lateral views of ROIs showing a significant effect of placebo beverage consumption in functional connectivity analysis. ROIs were defined from the Harvard-Oxford Atlas in CONN.

Abbreviations: vmPFC: ventromedial prefrontal cortex, SCC: subcallosal cingulate cortex, L NAcc: left nucleus accumbens, R NAcc: right nucleus accumbens, L Amyg: left amygdala, R Amyg: right amygdala, R IFG oper: right inferior frontal gyrus pars opercularis, R IFG tri: right inferior frontal gyrus pars triangularis, L IFG oper: left inferior frontal gyrus pars opercularis, L IFG tri: left inferior frontal gyrus pars triangularis, L FOrb: left frontal orbitalis, R FOrb: right frontal orbitalis, dACC: dorsal anterior cingulate cortex.

2.5.3. Associations between subjective response with functional connectivity response to placebo beverage

Fisher-transformed correlation coefficients between each ROI-to-ROI connection showing a significant effect of placebo beverage were extracted from CONN for further analyses in JMP. We used multiple linear regression to examine associations between post-beverage subjective response with post-beverage functional connectivity. Specifically, we modeled associations between adjusted post-beverage DEQ measures (each DEQ measure modeled separately as the predictor variable) and post-beverage resting state functional connectivity (dependent variable) while controlling for corresponding ROI-to-ROI pre-beverage resting state functional connectivity. Adjusted post-beverage DEQ measures were calculated by subtracting pre-beverage DEQ response from post-beverage DEQ response. Significance was defined as p<0.0167 (Bonferroni correction to account for three DEQ measures: “high,” “drunk,” and “feeling effect of alcohol”). Significant results are reported below.

2.5.4. Post hoc sensitivity analyses

We conducted post-hoc sensitivity analyses for the above models in JMP controlling for (1) biological sex, (2) recent alcohol use (average drinks per drinking day over past 30 days), and (3) excluding outliers. Outliers were identified as resulting residuals from the main model that lie above/below 1.5 times the interquartile range, and outliers on the DEQ measures predictor variables that were above/below 1.5 times the interquartile range. In the study design, alcohol and placebo conditions were counter balanced; half of participants completed placebo beverage consumption on their first beverage session day, while half of participants completed placebo beverage consumption on their second beverage session day (there were on average two days between beverage sessions). Therefore, we also conducted post-hoc sensitivity analysis for the above models controlling for session order (i.e., if placebo session came before or after the alcohol session in the overarching study design). Results that no longer remained significant (p>0.05) are reported below.

3. RESULTS

3.1. Subjective response to placebo beverage

Twenty-two young adults reported having consumed at least one standard alcoholic beverage following placebo beverage consumption and thus comprised the final sample. As expected, there was no significant increase in BrAC from pre-beverage to post-beverage consumption. As a group, there were increases in all DEQ subjective response measures following placebo beverage consumption. Table 2 details subjective response and manipulation check for the final sample. Supplementary table 2 details subjective response and manipulation check for the final sample stratified by biological sex.

Table 2.

Subjective response before and after placebo beverage consumption

	Placebo Beverage Administration
	Pre-Beverage	Post-Beverage	p-value
Feeling Alcohol Effectsa (SD; range)	1.0 (3.8; 0–18)	10.0 (8.6; 0.4–36)	<0.0001
Drunka (SD; range)	1.2 (3.8; 0–18)	6.5 (6.2; 0–20.9)	<0.0001
Higha (SD; range)	1.3 (3.9; 0–18)	7.0 (6.5; 0–18.7)	0.0002
Estimated Number of Standard Drinks (SD; range)	---	1.9 (0.6; 1–3)	---

Subjective response [Drug Effects Questionnaire (DEQ) measures] before and after placebo beverage consumption for the final sample (participants who reported having consumed at least one standard alcoholic beverage following placebo beverage consumption). Within-subject t-tests were used to assess effect of placebo beverage consumption on DEQ measures.

^aScales range from 0 to 36 where 0 indicates ‘normal (no change)’

3.2. Functional connectivity response to placebo beverage

Compared to pre-beverage functional connectivity, placebo beverage was associated with increased functional connectivity in a cluster of connections between right NAcc with vmPFC and SCC [F_(3,19)=10.1, p_FDR=0.005]. Post-hoc connection-level comparisons indicated the right NAcc showed increased functional connectivity with vmPFC [t₍₂₁₎=3.3, p_FDR=0.04, r²=0.11] and SCC [t₍₂₁₎=3.1, p_FDR=0.04, r²=0.13] following placebo beverage, compared to pre-beverage functional connectivity. Results remained significant in sensitivity analyses. Figure 2 shows ROI-to-ROI group-level analysis results in CONN and figure 3 shows within-subject changes in right NAcc functional connectivity from pre- to post-placebo beverage consumption. Results remained significant in sensitivity analyses.

Figure 3.

Within-subject changes in right nucleus accumbens functional connectivity from pre- to post-placebo beverage consumption

Individual participant pre-beverage and post-beverage consumption right nucleus accumbens (NAcc) –ventromedial prefrontal cortex (vmPFC) and subcallosal cingulate cortex (SCC) resting state functional connectivity fisher transformed correlation coefficients. Lines indicate within-subject changes from pre- to post-beverage consumption. Dashed lines indicate female data points; solid lines indicate male data points. Post-hoc connection-level comparisons indicated the right NAcc showed increased functional connectivity with vmPFC (t₍₂₁₎=3.3, p_FDR=0.04, r²=0.11) and SCC (t₍₂₁₎=3.1, p_FDR=0.04, r²=0.13).

3.3. Associations between subjective response with functional connectivity response to placebo beverage

Feeling more ‘drunk’ following placebo beverage was associated with greater right NAcc – SCC functional connectivity following placebo beverage (t₍₂₀₎=3.5, p=0.002, overall model: F_(2,19)=6.9, p=0.006, r²=0.42). Feeling greater ‘effects of alcohol’ following placebo beverage was also associated with greater right NAcc – SCC functional connectivity following placebo beverage (t₍₂₀₎=3.1, p=0.006, overall model: F_(2,19)=5.5, p=0.01, r²=0.37) (figure 4). Pre-beverage resting state functional connectivity did not relate to post-beverage resting state functional connectivity. Results remained significant in sensitivity analyses.

Figure 4.

Subjective ratings of intoxication relate to functional connectivity following placebo beverage consumption

Relations between perceived subjective response following placebo beverage consumption (‘drunk’ and ‘feeling alcohol effects’) with right nucleus accumbens (NAcc) – subcallosal cingulate cortex (SCC) functional connectivity following placebo beverage consumption. Feeling ‘drunk’ was positively associated with right NAcc - SCC functional connectivity following placebo beverage consumption (p=0.002). ‘Feeling alcohol effects’ was positively associated with right NAcc – SCC functional connectivity following placebo beverage consumption (p=0.006). Diamonds indicate female data points; circles indicate male data points. Models controlled for pre-beverage ROI-to-ROI functional connectivity.

*Indicates post-beverage drug effects questionnaire measures were adjusted for pre-beverage responses.

4. DISCUSSION

The current study examined placebo responses in the laboratory to better understand the neural basis for the psychological phenomenon of alcohol expectancies in healthy young adults. Results support our hypothesis, suggesting the placebo response to alcohol is associated with functional connectivity changes within PFC networks. Specifically, we observed an increase in positive functional connectivity between right NAcc – vmPFC and right NAcc – SCC following placebo beverage consumption, compared to the pre-beverage functional connectivity. Subjective ratings of intoxication following placebo beverage consumption positively related to increased functional connectivity following placebo beverage consumption, further supporting that functional connectivity changes in this network underlie the placebo response to alcohol. Overall, this study furthers our understanding of the neural underpinnings of alcohol expectancies in humans and proposes a possible mechanism by which non-pharmacological, expectancy-related processes contribute to the subjective experience of intoxication.

The NAcc is a central node in the reward network and thought to mediate the positive-reinforcing properties of alcohol (39). The NAcc shares rich connections with vmPFC and SCC (40)—regions that are also involved in the positive perception and pleasure of drinking, as well as in mood regulation (41). Together this network regulates functions associated with reward, motivation, and incentive salience (42). Prior studies have shown NAcc and vmPFC/SCC are involved in the expectation of both drug and non-drug reward (43–48). Using positron emission tomography, Volkow and colleagues found expectation of methamphetamine administration (expecting methamphetamine but receiving placebo) was associated with increased glucose metabolism in the NAcc and ventral ACC (Brodmann area 25) (48). While the current study did not collect a measure of reward per se, it is possible increased functional connectivity within this network following placebo beverage consumption reflects the expectation of, and perhaps even the experience of, reward and pleasure following perceived alcohol consumption. Positive alcohol expectancies (beliefs about the desirable effects of alcohol use) are robust predictors of alcohol use (3). While speculative, it is possible engagement of this network during perceived alcohol consumption marks a neural mechanism by which expectancies influence drinking behavior.

Human neuroimaging studies have shown acute alcohol activates the NAcc, with degree of activation thought to reflect the subjective experience of intoxication (19, 23). Similarly, the current study found placebo-related increases in NAcc – SCC network functional connectivity directly related to the self-reported experience of intoxication following placebo beverage consumption. These results add to the literature, suggesting NAcc networks may contribute to the subjective experience of intoxication through non-pharmacological, expectancy-related factors. Interestingly, prior studies examining NAcc – PFC resting state functional connectivity following acute alcohol consumption have been mixed, with either decreased or no change in functional connectivity reported (13, 14). Results from the current study indicate the non-pharmacological factors associated with alcohol use may relate to distinct changes in NAcc – ventral PFC networks during rest. It is important to note that participants in this study viewed the Headspace Studios’ Inscapes video during resting state data acquisition. Results therefore may differ from, and not be directly comparable to, these prior studies examining effects of acute alcohol on resting state functional connectivity, which employed a more traditional resting state design (i.e., viewing a fixation cross) (32). Prior work has suggested alcohol expectancies (measured with self-report surveys) relates to functional connectivity of striatal and limbic systems at rest (6–8). Thus, despite resting state methodological differences, results converge to suggest the placebo response to alcohol may relate to functional connectivity changes within this key reward network. Preclinical and clinical studies have shown drug expectancies influence response to drug such that response is enhanced when drug is expected (48–54), and future within-subject studies are needed to directly test the interactive effects of pharmacological and non-pharmacological factors on neural response to alcohol.

FMRI studies have observed changes in frontal cortex connectivity and vmPFC activity in response to alcohol cues (55, 56). It is possible that alcohol cues associated with placebo beverage manipulation elicited the observed neural response in the current study. As functional connectivity response to placebo beverage related to perceived feelings of being drunk, we hypothesize findings extend beyond cue exposure. Studies including an alcohol cue plus expectancy and an alcohol cue minus expectancy condition is needed to isolate the expectancy effect and test this hypothesis. Zhornitsky and colleagues (2018, 2019) found self-reported alcohol expectancies mediated the relationship between frontal cortical connectivity (during alcohol cue exposure and at rest) with problematic alcohol use in nondependent adult drinkers (6, 56). This work suggests neural response to alcohol cues may interact with alcohol expectancies to influence alcohol use outcomes. Recent work has also suggested the neural placebo response plays an important role in future alcohol use outcomes. A placebo-controlled, within-subjects alcohol administration study using fMRI found functional connectivity during the placebo (but not alcohol) condition predicted increased alcohol use problems over a five-year period (22). Overall, findings suggest an important role of the neural placebo response in alcohol use outcomes and highlight the need for longitudinal research to examine how variation in the placebo response, underlying neural systems, and craving/cue exposure contributes to alcohol use outcomes over time.

Limitations

This within-subjects fMRI study has many strengths; however, several limitations should be noted. Findings should be interpreted with caution owing to small sample size. Three participants (12%) did not report believing they had consumed beverages containing alcohol. As the goal of the study was to examine the neural correlates of the placebo response, these participants were excluded from analyses. It is difficult to interpret our three placebo manipulation failures as few studies that use placebo report their success rates. Nonetheless, it is important to consider limitations associated with the placebo manipulation. For example, drinking contexts (i.e., physical and social) affect subjective response to alcohol. Corbin and colleagues found social context (i.e., group versus solitary) is an important factor in studies focused on alcohol expectancies (57). The current study was conducted in a laboratory setting. While participants were the only individual consuming beverages, a minimum of two research staff members were present during beverage procedures. Future studies examining the placebo effect would benefit from study designs employing group drinking, and possibly bar laboratory settings. Additionally, the current study did not include a ‘Told No Alcohol/Get No Alcohol’ condition, so it is possible placebo-related effects could be associated with any type of beverage consumption. As discussed, it is also possible neural responses observed following placebo beverage consumption could be related to alcohol cue exposure. As a next step, studies should include and counterbalance ‘Told No Alcohol/Get No Alcohol’ (no alcohol expectancy) and ‘Told Alcohol/Get No Alcohol’ (alcohol expectancy) conditions. Despite this limitation, results from the current study highlight the importance of considering the placebo response in research examining effects of alcohol intoxication and risk for problematic alcohol use.

This study focused on young adults (ages 21 – 26) and excluded young adults showing possible signs of alcohol dependence, defined as an AUDIT score greater than 15. While the sample included individuals exhibiting a range of recent alcohol use, no participants met criteria for an alcohol use disorder, according to DSM-5. Alcohol expectancies are related to drinking experiences (58–61), and it is unclear if results from this study would generalize to older individuals and those with heavier patterns of alcohol use. However, the study’s focus on young adults within a narrow age range (ages 21 – 26) and without prior or current alcohol use disorder minimizes experience-related differences in alcohol expectancies. There is also evidence indicating individuals with family history of alcohol use disorder exhibit enhanced dopaminergic striatal response to alcohol expectation (62). The current sample had only one individual with a family history of problematic alcohol use. Studies have revealed important sex-differences in relations between alcohol expectancies and problem drinking (30), and neuroimaging work has uncovered a possible neural basis for these differences (6, 30, 31, 56). Due to small sample size, we were underpowered to investigate sex differences in the placebo response to alcohol. Future studies, with larger sample sizes, should examine the neural underpinnings contributing to variation in placebo response in young adults with family history of alcohol use disorder and other high-risk populations, and explore interactions with biological sex, to further our understanding of risk factors for alcohol use disorder. Future work should also include longitudinal designs capable of investigating the predictive value of the placebo response.

Conclusion

The current study used placebo administration to further our understanding of the neural correlates of alcohol expectancies in healthy young adults. Results suggest the placebo response to alcohol is associated with increased functional connectivity between NAcc – vmPFC and SCC, and that these brain changes relate to perceived subjective feelings of intoxication. Together, results put forth a possible mechanism by which alcohol expectancies contribute to the subjective experience of intoxication. Results also highlight the need to consider the placebo response in studies examining acute effects of alcohol intoxication. Alcohol expectancies play a critical and distinct role in alcohol use behavior; a greater understanding of the placebo response may provide a unique opportunity to inform intervention aimed at reducing risk for heavy alcohol use and alcohol use disorder.

Highlights.

• Examined neural correlates of the placebo response to alcohol

• Placebo alcohol alters nucleus accumbens (NAcc) functional connectivity

• NAcc functional connectivity relates to perceived feelings of intoxication