Clubgoers and Their Trendy Cocktails: Implications of Mixing
Caffeine Into Alcohol on Information Processing and Subjective
Reports of Intoxication
Cecile A. Marczinski and Mark T. Fillmore
University of Kentucky
Alcoholic drink preferences in college students have made an interesting shift recently, with
trends in consumption leaning toward caffeinated alcohol in various forms (e.g., Red Bull and
vodka or caffeinated beers such as Anheuser-Busch’s B-to-the-E). Despite the dramatic rise
in popularity of these beverages, little research has examined the combined effects of alcohol
and caffeine, which is problematic for adequately informing the public about the risk or lack
thereof of these drinks. The purpose of this study was to directly investigate the acute effects
of alcohol and caffeine, alone and in combination, on well-validated measures of cognitive
performance and subjective intoxication in social drinkers. Participants (N ? 12) performed
a psychological refractory period task that measured dual-task interference as the prolonged
reaction time to complete the 2nd of 2 tasks performed in close temporal sequence. Perfor-
mance was tested under 2 active doses and 1 placebo dose of caffeine (0.0 mg/kg, 2.0 mg/kg,
and 4.0 mg/kg) in combination with 1 active dose and 1 placebo dose of alcohol (0.0 g/kg
and 0.65 g/kg). As expected, alcohol impaired task performance by increasing dual-task
interference and increasing errors. The coadministration of caffeine counteracted the effects
of alcohol on interference but had no effect on the degree to which alcohol increased errors.
Subjective measures of intoxication showed that coadministration of caffeine with alcohol
reduced participants’ perceptions of alcohol intoxication compared with administration of
alcohol alone. The results highlight the complexity of drug interactions between alcohol and
Keywords: alcohol, caffeine, information processing, intoxication, college students
Alcoholic drink preferences in college students have
made an interesting shift in the last decade, with clubgoers
increasingly drawn to caffeinated alcoholic drinks in vari-
ous forms. With the national introduction of Red Bull to the
United States in 2001, young partygoers became enamored
with using the beverage as a mixer for their alcoholic drinks,
presumably for the purpose of reducing the depressant ef-
fects of alcohol and thus allowing them to party longer. This
increased popularity of caffeinated alcoholic beverages
among college students was quickly identified by the bev-
erage industry in North America. In 2005, several “energy”
beers and malt beverages were introduced, including An-
heuser-Busch’s B-to-the-E (54 mg caffeine in 6.6% alcohol
by volume), New Century Brewing’s Moon Shot (45 mg
caffeine in 4.8% alcohol by volume), Labatt’s Shok (60 mg
caffeine in 6.9% alcohol by volume), and Molson’s Kick
(55 mg caffeine in 5% alcohol by volume).
Some physicians have warned of the potential health
implications of mixing caffeine and alcohol, such as in-
creased risk of dehydration (American Medical Association,
2003). Although the various energy drinks and beers also
contain other stimulant compounds, such as taurine, theoph-
ylline, theobromine, and ginseng, the health concerns of
these drinks have been in relation to the high caffeine
content. Caffeine is the common stimulatory compound in
all of these beverages. Of interest, the U.S. Food and Drug
Administration (FDA) does not regulate the caffeine content
of energy drinks, and recent analyses have determined that
the caffeine content of these drinks can contain 150%–
300% of the amount of caffeine that the FDA allows for
cola beverages (McCusker, Goldberger, & Cone, 2006).
However, there have been surprisingly few investigations
into the presumption that these caffeinated alcohol drinks
allow drinkers to consume greater amounts of alcohol be-
cause the sedation and behaviorally impairing effects nor-
mally associated with the drug are offset by the coconsump-
tion of caffeine (Ferreira, de Mello, Pompeia, & de Souza-
Formigoni, 2006). Although to date there has been no
published study of North American drinkers’ motivations
Cecile A. Marczinski and Mark T. Fillmore, Department of
Psychology, University of Kentucky.
This research was supported by a grant from the Alcoholic
Beverage Medical Research Foundation, awarded to Cecile A.
Marczinski; National Institutes of Health National Research Ser-
vice Award DA07304 from the National Institute on Drug Abuse,
awarded to Cecile A. Marczinski; and National Institute on Alco-
hol Abuse and Alcoholism Grant R01 AA12895, awarded to Mark
Correspondence concerning this article should be addressed to
Mark T. Fillmore, Department of Psychology, University of Ken-
tucky, Lexington, KY 40506-0044. E-mail: email@example.com
Experimental and Clinical Psychopharmacology
2006, Vol. 14, No. 4, 450–458
Copyright 2006 by the American Psychological Association
for using caffeinated alcoholic drinks, a small survey of
college students in Brazil (N ? 136) examined their moti-
vations for using these drinks (Ferreira, de Mello, & For-
migoni, 2004). The authors reported that 76% of the sample
indicated regular use of energy drinks in combination with
alcohol (mainly whiskey, vodka, or beer). Of those who
reported use of caffeinated alcohol, 38% reported that the
combination drinks increased happiness, 30% reported eu-
phoria from these drinks, 27% reported uninhibited behav-
ior from these drinks, and 24% reported increased physical
vigor. Other reports from Ireland and Germany have impli-
cated these drinks in assaults and automobile accidents,
respectively, suggesting that the combination of caffeine
and alcohol impairs the ability to correctly assess level of
intoxication and the ability to drive more greatly than alco-
hol intoxication alone (Riesselmann, Rosenbaum, &
Schneider, 1996; Tormey & Bruzzi, 2001). Correspond-
ingly, Denmark, France, and Norway have placed bans on
the sale of Red Bull, citing health concerns.
Caffeine acts as an adenosine receptor antagonist, with
the central nervous system (CNS) stimulatory effects of
caffeine largely due to blockage of adenosine A2A recep-
tors that stimulate GABAergic neurons of inhibitory path-
ways to the dopaminergic reward system of the striatum
(Cauli & Morelli, 2005; Mandel, 2002; Nehlig, Daval, &
Debry, 1992). In addition, caffeine also acts as a CNS
stimulant as it is an indirect agonist of noradrenaline (Lader
& Bruce, 1989). Laboratory studies of the behavioral effects
of caffeine show that the drug can generally enhance skilled
performance by allaying fatigue, increasing vigilance,
speeding reaction time (RT), and prolonging effort (for
reviews, see Nehlig et al., 1992; Weiss & Laties, 1962). The
extent to which the coadministration of caffeine can coun-
teract or functionally antagonize alcohol-induced behavioral
impairment also has been studied in the laboratory; how-
ever, the findings have been mixed. Some studies have
shown that caffeine coadministration can reduce the impair-
ing effects of alcohol on some global performance tasks
(Burns & Moskowitz, 1990; Fillmore & Vogel-Sprott,
1999; Franks, Hagedorn, Hensley, Hensley, & Starmer,
1975; Kerr, Sherwood, & Hindmarch, 1991; Martin & Gar-
field, 2006; Rush, Higgins, Hughes, Bickel, & Wiegner,
1993). However, other studies have failed to demonstrate
consistent counteracting effects of caffeine (Fillmore &
Vogel-Sprott, 1995; Liguori & Robinson, 2001). Research
reviews also have noted these discrepancies with regard to
alcohol–caffeine interactions, leading to conclusions that
evidence for caffeine antagonism is equivocal (e.g., Fudin &
Reasons for these inconsistencies are not clear. However,
the tasks used in these studies varied widely in their com-
plexity and in the specific behavioral and cognitive mech-
anisms involved in their performance (e.g., memory, motor
coordination, RT). We recently argued that the equivocal
evidence for caffeine antagonism of alcohol-induced im-
pairment might reflect the fact that not all cognitive and
behavioral impairments from alcohol can be offset by the
coadministration of caffeine. For example, in a study of the
separate and combined effects of moderate doses of alcohol
(0.65 g/kg) and caffeine (2.0 and 4.0 mg/kg), we showed
caffeine could counteract alcohol-induced slowing of re-
sponse time but not the disinhibiting effects of the drug
(Marczinski & Fillmore, 2003). Thus, it appears that the
ability of caffeine to counteract alcohol-induced impairment
could depend greatly on the specific nature of the cognitive
and behavioral processes involved.
Another important consideration concerns the degree of
behavioral demands imposed by the particular activity being
performed. It is well known from studies of divided atten-
tion that alcohol impairment can be intensified in situations
of high behavioral demand (Holloway, 1995; Linnoila,
1974). A divided attention task essentially requires an indi-
vidual to perform two tasks simultaneously—for example,
manually tracking a moving object (e.g., a pursuit task)
while performing an auditory discrimination task (e.g., de-
tecting differences among tones). Related are dual-task sit-
uations. Typically, dual-task performance is measured by
requiring an individual to respond to each of two stimuli
(Tasks 1 and 2) presented in very close temporal proximity
(for a review, see Pashler, 1994). These situations often
illustrate the limits of human information processing. Char-
acteristically, an interference effect is observed as a slowing
of response time to the second stimulus (Task 2). The
delayed response time to the second stimulus is attributed to
the psychological refractory period (PRP) and is assumed to
reflect a limitation of information processing in which the
response to Task 2 must be delayed until processing of
Task 1 is complete (Johnston & Heinz, 1978). Performance
in dual-task situations is highly sensitive to the disruptive
effects of alcohol. Studies of alcohol effects in dual-task
situations show that moderate doses of alcohol dramatically
increase task impairment, even in simple tasks that show no
impairment from alcohol when performed in isolation (Fill-
more & Van Selst, 2002; Schweizer, Jolicoeur, Vogel-
Sprott, & Dixon, 2004).
Outside the laboratory, the disruptive effects of alcohol
often occur in complex, behaviorally demanding environ-
ments that require the simultaneous performance of multiple
activities (i.e., the operation of a motor vehicle). Laboratory
assessment of dual-task performance may hold greater eco-
logical validity as models of day-to-day performance of
activities outside the laboratory. As yet, no research has
applied the dual-task PRP model to the investigation of
alcohol–caffeine interactions. In the present study, we ex-
amined healthy adults and tested the separate and combined
effects of alcohol and caffeine on their ability to process
information in a dual-task situation. Performance was tested
under two active doses and one placebo dose of caffeine
(0.0 mg/kg, 2.0 mg/kg, and 4.0 mg/kg) in combination with
one active dose and one placebo dose of alcohol (0.0 g/kg
and 0.65 g/kg). The active alcohol dose (0.65 g/kg) used in
the study has been shown to impair information processing
in the dual-task situation (Fillmore & Van Selst, 2002). The
active caffeine doses (2.0 mg/kg and 4.0 mg/kg) were
selected to approximate the caffeine content found in the
various caffeinated beers and mixed alcoholic energy drinks
ALCOHOL AND CAFFEINE
currently on the market and typically consumed by college
students. In addition to examining effects on information
processing, the study also examined how this drug combi-
nation affects drinkers’ subjective reports of intoxication.
Twelve adults (6 women and 6 men) between the ages of 21
and 28 years (M ? 23.5, SD ? 2.7) participated in this study.
Determination of the appropriate sample size was based on power
analyses of data from a previous study that examined alcohol
effects on dual-task performance (Fillmore & Van Selst, 2002).
The racial makeup of the sample included 1 Asian and 11 Cauca-
sian participants. Participants had a mean weight of 69.6 kg
(SD ? 15.9). Volunteers completed questionnaires that provided
demographic information, drinking habits, and physical and men-
tal health status. Individuals with a self-reported psychiatric dis-
order, substance abuse disorder, head trauma, or other CNS injury
were excluded from the study. Volunteers with a score on the Short
Michigan Alcoholism Screening Test (Selzer, Vinokur, & Van
Rooijen, 1975) of 5 or higher were also excluded from the study.
Recent use of amphetamine, barbiturates, benzodiazepines, co-
caine, opiates, and tetrahydrocannabinol was assessed by means of
urine analysis. Any volunteer who tested positive for the presence
of any of these drugs was excluded from participation. No female
volunteers who were pregnant or breast-feeding participated in the
research, as determined by self-report and urine human chorionic
gonadotropin levels. Participants were recruited via notices posted
on community bulletin boards and by local newspaper advertise-
ments. All volunteers provided informed consent prior to partici-
pating. The University of Kentucky Medical Institutional Review
Board approved the study, and volunteers received $100 for their
participation in the entire seven-session study. Participants com-
pleted all of the sessions in 18.8 days on average (SD ? 4.0).
Apparatus and Materials
them to respond to two different stimuli (for Tasks 1 and 2)
presented in close succession. Task 1 was a go/no-go task. The go
stimulus was a 1, and the no-go stimulus was an X. The stimuli
were presented in black against a white background. Participants
were required to press the 1 key on the keyboard with their right
hand when the go stimulus was presented. No response was
required when the no-go stimulus was presented. The go/no-go
stimuli remained visible for 2,000 ms or were terminated once the
response to Task 2 occurred.
Task 2 was an auditory discrimination task. On each trial, the
auditory stimulus was either a high tone (1000 Hz) or a low tone
(125 Hz). The tone was presented for 500 ms. Using their left
hand, participants were required to press the a key when the high
tone was presented and to press the z key when the low tone was
presented. Participants had 2,000 ms from the onset of the tone to
respond; otherwise, no response was recorded as an error.
Each trial consisted of the following sequence of events: (a) the
presentation of a fixation point (an asterisk) for 250 ms; (b) a
randomly varying foreperiod of 120, 180, or 240 ms; (c) a Task 1
stimulus (1 or X); and (d) a Task 2 stimulus presented after a
stimulus onset asynchrony (SOA) of 50, 200, 600, or 800 ms
following the onset of the Task 1 stimulus. Each trial was sepa-
rated by an intertrial interval of 2,200 ms. To encourage accurate
Participants performed a dual task that required
responding, we displayed a feedback message (the word INCOR-
RECT) during the intertrial interval following any incorrect
Each test consisted of 192 trials. A test presented an equal
number of go and no-go stimuli for Task 1 (i.e., 96) and an equal
number of high- and low-tone stimuli for Task 2 (i.e., 96). The four
SOAs were presented equally often (48 times). There were 16
possible combinations of these variables for a trial (one possible
combination: Task 1 ? go stimulus, SOA ? 50 ms, Task 2 ? low
tone). Each combination was presented 12 times during a test in a
random order. A test required approximately 10 min to complete.
The task was operated using E-Prime experiment generation soft-
ware (Schneider, Eschman, & Zuccolotto, 2002) and was run on a
Simple auditory discrimination.
version of the Task 2 component of the PRP task. A test consisted
of 40 trials (20 high tone and 20 low tone) in random order. Tone
presentation and response requirements were identical to those
described in Task 2. A test required approximately 2 min. Unlike
the Task 2 component of the PRP task, the simple auditory
discrimination task was performed separately as a single task. This
task was a control task condition used to evaluate the effect of
alcohol and caffeine on simple tone discrimination when no inter-
ference from Task 1 was present. The task was also operated using
Personal Drinking Habits Questionnaire (Vogel-Sprott, 1992).
This questionnaire yielded three measures of a drinker’s current,
typical drinking habits: (a) frequency (number of drinking occa-
sions per week), (b) dose (milliliters of absolute alcohol per
kilogram of body weight typically consumed during a single
drinking occasion), and (c) duration (time span, in hours, of a
typical drinking occasion).
Caffeine use questionnaire.
This questionnaire yielded a mea-
sure of a participant’s typical daily caffeine consumption in mil-
ligrams per kilogram of body weight. The questionnaire required
participants to report their typical daily consumption of beverages
(e.g., coffee, tea, soft drinks) and foods (e.g., chocolate) containing
caffeine. Estimates of the caffeine content of foods and beverages
were taken from Barone and Roberts (1996).
Beverage rating scale.
Participants also completed a beverage
rating scale to report their perceived intoxication in each dose
condition. Participants estimated the alcohol content of the dose
they had received in terms of bottles of beer containing 5%
alcohol. The scale ranged from 0 to 10 bottles of beer, in 0.5-bottle
increments. Previous studies have noted that this beverage rating
scale provides a sensitive real-world measure of participants’ sub-
jective level of intoxication (Fillmore & Vogel-Sprott, 2000). In
previous studies, individuals gave estimates of the amount of
alcohol they thought they had consumed that closely corresponded
with the actual dose administered (Marczinski & Fillmore, 2005).
Blood alcohol concentrations (BACs).
from breath samples measured by an Intoxilyzer, Model 400 (CMI,
This task was an abbreviated
BACs were determined
Individuals who responded to the advertisements called the
laboratory and participated in a telephone-screen interview con-
ducted by a research assistant. Volunteers were told that the
purpose of the experiment was to study the effects of alcohol and
caffeine on performance. All sessions were conducted in the Hu-
man Behavioral Pharmacology Laboratory of the Department of
Psychology at the University of Kentucky and began between
MARCZINSKI AND FILLMORE
10 a.m. and 6 p.m. Before each test session, participants were
instructed to fast for 4 hr, abstain from alcohol for 24 hr, and
abstain from caffeine for 8 hr. Participants abstained from caffeine
for 9.8 hr on average (SD ? 3.2). Before each session, urine
samples were tested for the presence of drug metabolites (On Trak
TesTstiks; Roche Diagnostics Corporation, Indianapolis, IN) and
pregnancy in the female participants (Mainline Confirms HGL;
Mainline Technology, Ann Arbor, MI). In addition, a zero BAC
was verified for each participant at the start of each session.
Participants were tested individually by a research assistant who
was unaware of the research hypotheses. All testing was conducted
in a small room that consisted of a chair and a desk with a
computer that operated the tasks.
During familiarization, participants
provided informed consent, were weighed, and completed ques-
tionnaires. They also practiced the PRP and simple auditory dis-
PRP and simple auditory discrimination task
performance was tested under a 2 (alcohol dose) ? 3 (caffeine
dose) factorial within-subject design that crossed two doses of
alcohol (0.0 g/kg and 0.65 g/kg) with three doses of caffeine (0.0
mg/kg, 2.0 mg/kg, and 4.0 mg/kg). Thus, the design examined the
extent to which the behaviorally impairing effects of a single
active alcohol dose (0.65 g/kg) could be counteracted by two
active doses of caffeine (2.0 mg/kg and 4.0 mg/kg). Performance
under the resulting six dose conditions was tested on individual
sessions that were separated by a minimum period of 24 hr and
maximum period of 7 days. Dose administrations were double
blind, and dose order across the six sessions was randomized
Alcohol and caffeine administration.
that they might receive alcohol, caffeine, or both drugs during
some or all of the sessions. However, the exact contents of the
beverages were never disclosed to participants during the study.
Carbonated, lemon-flavored soda was used as the vehicle for
alcohol and caffeine administration. Participants consumed the
drink within 6 min in all dose conditions. Alcohol and caffeine
doses were calculated on the basis of body weight. The 0.65 g/kg
alcohol doses were administered as one part absolute alcohol and
three parts vehicle. The mean total volume of the drink that the
participants consumed was 230 ml. This 0.65 g/kg dose produces
an average peak BAC of 80 mg/100 ml approximately 60 min after
drinking. The alcohol dose was chosen on the basis of prior
research that showed that 0.65 g/kg of alcohol reliably impairs
PRP task performance (Fillmore & Van Selst, 2002). The placebo
alcohol dose (0.0 g/kg) consisted of the vehicle matched to the
total volume of the 0.65 g/kg alcohol dose beverage. A small
amount (3 ml) of alcohol was floated on the surface of the vehicle,
and the glass was sprayed with an alcohol mist that resembled
condensation and provided a strong alcoholic scent as the beverage
was consumed. Previous research has shown that individuals re-
port that this beverage contains alcohol (e.g., Fillmore, Carscad-
den, & Vogel-Sprott, 1998).
The caffeine doses were administered by adding 0.0 mg/kg, 2.0
mg/kg, and 4.0 mg/kg of tasteless, anhydrous caffeine powder to
the vehicle, either alone or in combination with the alcohol dose.
When mixed in an aqueous solution, this form of caffeine is
absorbed rapidly, within 60 min (Bonati et al., 1982). The caffeine
doses were chosen because they have been shown to counteract
alcohol impairment in other research and also were typical of the
caffeine content of the caffeinated, alcoholic beverages currently
popular among college students (e.g., Burns & Moskowitz, 1990;
Participants were told
Fillmore, 2003; Fillmore & Vogel-Sprott, 1995; Marczinski &
Fillmore, 2003). After dose administration, participants relaxed
and read magazines.
PRP and simple auditory discrimination testing.
PRP task performance followed by simple auditory discrimination
task performance was tested starting 35 min after drinking began,
a time corresponding with the ascending period of the blood
alcohol curve in the active alcohol dose conditions. Participants
completed the PRP task from 35 to 45 min after drinking. From 50
to 55 min, participants completed the simple auditory discrimina-
tion task, which was immediately followed by the completion of
the beverage rating scale. BACs were measured at 30, 45, 60,
and 90 min after drinking in each session, regardless of whether
doses contained alcohol. At 90 min postdrinking, the testing por-
tion of the session concluded, and participants relaxed in a waiting
room within the laboratory. They received a meal and remained at
leisure to read magazines or watch television until their BAC fell
below 20 mg/100 ml. Upon completing the final session, partici-
pants were paid and debriefed.
Criterion Measures and Data Analyses
The primary measures in this study concerned the degree to
which performance of Task 1 interfered with performance of
Task 2. Thus, the primary measure of interest was the PRP inter-
ference score, which quantified the level of interference from
Task 1 on Task 2 RT. In addition, Task 2 RT (RT2) and errors
were analyzed. All of these measures and the associated analyses
are described in detail below. In addition, manipulation checks
included analyses of Task 1 RT (RT1) and simple auditory dis-
crimination RTs. Dose effects on these basic performance mea-
sures were not predicted.
Task 2 Effects
In a dual-task context, RT2 increases as a function of
decreasing SOA. The least interference on RT2 should be evident
when Task 2 is presented at the longest period of time (SOA) after
Task 1. Therefore, RT2 should be shortest at the longest (i.e.,
800-ms) SOA. By contrast, interference on RT2 should be greatest
when Task 2 occurs at the shortest period of time after Task 1.
Thus, RT2 should be longest at the shortest (i.e., 50-ms) SOA.
Dose and gender effects on interference were examined by RT2
scores in each dose condition using a 2 (gender: male or fe-
male) ? 2 (alcohol dose: 0.0 or 0.65 g/kg) ? 3 (caffeine dose:
0.0, 2.0, or 4.0 mg/kg) ? 4 (SOA: 50, 200, 600, or 800 ms)
mixed-design analysis of variance (ANOVA), where gender was a
between-subjects factor and alcohol dose, caffeine dose, and SOA
were within-subject factors.
PRP interference score.
Typically, dual-task interference is
quantified by a PRP interference score whereby the magnitude of
the interference is calculated as the difference between RT2 at the
shortest SOA (maximal interference) and RT2 at the longest SOA
(minimal interference). Thus, a PRP interference score can be
expressed as a single value: RT2shortest SOA– RT2longest SOA(e.g.,
Fillmore & Van Selst, 2002; Van Selst, Ruthruff, & Johnston,
1999). Larger PRP scores indicate greater interference. These PRP
interference scores were submitted to a 2 (gender) ? 2 (alcohol
dose) ? 3 (caffeine dose) mixed-design ANOVA, where gender
was a between-subjects factor and alcohol dose and caffeine dose
were within-subject factors. Subsequent post hoc analyses used
simple effects t tests.
Prior to all analyses, the RT2 data were filtered to eliminate
trials with an incorrect response to either Task 1 or Task 2 or an
ALCOHOL AND CAFFEINE
RT of less than 100 ms or greater than 2,000 ms. The outlier
elimination procedures resulted in removal of less than 1% of
Task 2 errors.
Response errors for Task 2 were submitted to
a 2 (gender) ? 2 (alcohol dose) ? 3 (caffeine dose) ? 4 (SOA)
Task 1 Effects
Dose and gender effects on participants’ RT1 to the go
targets were analyzed by a 2 (gender) ? 2 (alcohol dose) ? 3
(caffeine dose) mixed-design ANOVA.
Task 1 errors.
The number of errors for Task 1 in the present
study was less than one per test, and this low level of errors
precluded any meaningful analysis.
Simple auditory discrimination.
participants’ simple auditory discrimination RT scores were ana-
lyzed by a 2 (gender) ? 2 (alcohol dose) ? 3 (caffeine dose)
mixed-design ANOVA. Overall errors were less than one per test
on all simple auditory discrimination tests, regardless of dose and
gender, thus precluding statistical analyses.
Dose and gender effects on
Drinking Habits and Caffeine Use
No gender differences were revealed by t tests on any
drinking habit measure or caffeine use measure (ps ? .09).
From the Personal Drinking Habits Questionnaire data, the
sample reported a mean drinking frequency of 1.6
(SD ? 0.9) times per week, with a mean dose per occasion
of 1.3 (SD ? 0.5) g/kg. For a person weighing 70 kg, this
alcohol dose would approximate five bottles of beer con-
taining 5% alcohol by volume. The mean duration of drink-
ing was 4.3 (SD ? 1.1) hr. The sample reported a mean
daily caffeine use of 6.8 (SD ? 7.7) mg/kg, which approx-
imates a mean level of daily caffeine exposure of 476 mg.
For a person weighing 70 kg, this caffeine dose would
approximate two 16-ounce (473 ml “grande” size) cups of
Starbucks breakfast blend coffee or about nine 355-ml cans
of soft drink, such as Pepsi (McCusker, Goldberger, &
Cone, 2003; McCusker et al., 2006).
BACs obtained in the three active alcohol dose conditions
were examined by a 2 (gender) ? 3 (caffeine dose) ? 4
(time) mixed-design ANOVA. There was no significant
main effect involving gender (p ? .34), caffeine dose (p ?
.40), or any significant interactions (ps ? .12). Thus, BAC
was not affected by gender or by the coadministration of
caffeine. A main effect of time, F(3, 30) ? 15.03, p ? .001,
was obtained, attributable to the rise and decline of BACs
during the course of a session. The 0.65-g/kg dose produced
a mean BAC of 70.8 mg/100 ml (SD ? 19.9) at 30 min, just
before the test, and rose to 84.1 mg/100 ml (SD ? 18.2)
at 45 min, in the middle of testing. The mean BAC declined
to 80.4 mg/100 ml (SD ? 14.7) at 60 min, at the conclusion
of testing, and continued to decline to 64.8 mg/100 ml
(SD ? 12.4) at 90 min after drinking, when the session
PRP Task Performance
PRP interference scores.
calculated for each dose condition (PRP interference ?
RT2shortest SOA– RT2longest SOA). A 2 (gender) ? 2 (alcohol
dose) ? 3 (caffeine dose) mixed-design ANOVA of the
PRP interference scores revealed a significant Alcohol
Dose ? Caffeine Dose interaction, F(2, 20) ? 4.19, p ?
.03. Figure 1 illustrates this interaction. This figure reveals
that PRP interference increased under alcohol alone. This
observation was confirmed by a simple effects t test that
found significantly greater PRP interference under alcohol
alone as compared with vehicle, t(11) ? 3.91, p ? .01. In
addition, the figure shows that the coadministration of both
active caffeine doses counteracted the alcohol-induced in-
crease in interference. It was revealed through t tests that the
coadministration of both active caffeine doses with alcohol
significantly reduced the PRP interference as compared with
alcohol alone (ps ? .02). Finally, the figure illustrates that
the administration of caffeine alone did not alter PRP inter-
ference as compared with vehicle. This observation was
confirmed, as PRP interference scores did not differ from
the vehicle in either caffeine dose condition (ps ? .23).
There was no significant main effect or interaction involv-
ing gender (ps ? .39).
A 2 (gender) ? 2 (alcohol dose) ? 3 (caffeine
dose) ? 4 (SOA) mixed-design ANOVA revealed a signif-
icant Alcohol Dose ? SOA interaction, F(3, 30) ? 4.72,
p ? .01. Figure 2 plots the alcohol-alone and vehicle con-
dition to illustrate the interaction. The figure shows the
typical dual-task interference (PRP) effect as an increase in
RT2 as a function of decreasing SOA. The figure also shows
how alcohol affects PRP differently as a function of the
SOA. Alcohol had the most pronounced slowing effect at
the shortest SOAs, when interference from Task 1 was the
greatest. There was no significant main effect or interaction
involving gender (ps ? .36).
Task 2 errors.
A 2 (gender) ? 2 (alcohol dose) ? 3
(caffeine dose) ? 4 (SOA) repeated measures ANOVA
revealed a main effect of alcohol, F(1, 10) ? 19.22, p ? .01.
Overall, the mean accuracy was very high but was reduced
by alcohol. Participants made an average of 9.1 errors per
test when alcohol was administered compared with an av-
erage of 5.5 errors per test when alcohol was not adminis-
tered. Caffeine administration, alone or in combination with
alcohol, did not alter error rates (ps ? .15). In addition, no
significant main effect or interaction involving gender was
found for Task 2 errors (ps ? .10).
PRP interference scores were
Effects on Task 1 and Simple Auditory
A 2 (gender) ? 2 (alcohol dose) ? 3 (caffeine dose)
repeated measures ANOVA of participants’ RT to the go
targets in Task 1 revealed no significant effects (ps ? .06).
Thus, alcohol and caffeine, alone or in combination, had no
effect on RT1 performance. The overall mean RT to go
targets on Task 1 was 493.3 (SD ? 135.5) ms.
MARCZINSKI AND FILLMORE
A 2 (gender) ? 2 (alcohol dose) ? 3 (caffeine dose)
repeated measures ANOVA of participants’ RT for simple
auditory discrimination revealed no significant effects
(ps ? .09). Thus, alcohol and caffeine, alone or in combi-
nation, had no effect on simple auditory discrimination
performance. The overall mean RT on the simple discrim-
ination task was 349.2 (SD ? 62.4) ms.
Dose and gender effects on the beverage ratings were
analyzed by a 2 (gender) ? 2 (alcohol) ? 3 (caffeine)
mixed-design ANOVA, which revealed a significant Alco-
hol ? Caffeine interaction, F(2, 20) ? 3.62, p ? .05. Table
1 illustrates this interaction. Participants reported greater
perceived alcohol effects under all active alcohol dose con-
ditions, and simple effects t tests revealed that beverage
ratings were greater under alcohol alone as compared with
vehicle, t(11) ? 8.64, p ? .001. Furthermore, coadminis-
tration of both caffeine doses reduced the alcohol-induced
increase in beverage ratings. Simple effects revealed that
coadministration of 2.0 mg/kg of caffeine with alcohol
significantly lowered beverage ratings as compared with
alcohol alone, t(11) ? 1.77, p ? .05, but coadministration
conditions. The graph on the left illustrates alcohol effects in combination with caffeine. The graph
on the right illustrates the effects of caffeine administration alone. The vehicle condition, 0.0 g/kg
alcohol (alc) ? 0.0 mg/kg caffeine (caf), is presented in each graph to facilitate active dose
comparison to vehicle. Error bars show standard errors of the mean.*p ? .05 for difference from
vehicle.?p ? .05 for difference from the 0.65 g/kg alc ? 0.0 mg/kg caf dose.
Psychological refractory period (PRP) interference effect scores under the six dose
stimulus onset asynchronies (SOAs) when 0.65 g/kg alcohol was
administered alone compared with vehicle.
Mean reaction time (RT) to Task 2 following the four
ALCOHOL AND CAFFEINE
of 4.0 mg/kg of caffeine did not (p ? .10). Finally, caffeine
alone did not significantly alter beverage ratings as com-
pared with vehicle (ps ? .08). There was no significant
main effect or interaction involving gender (ps ? .18).
This research investigated the separate and combined
effects of alcohol and caffeine on constraints on information
processing in a dual-task context. The results showed that
alcohol impaired information processing in a dual-task con-
text, as measured by the increased PRP to complete a
second task performed in close proximity to a first task.
Response accuracy to the second task was also impaired by
alcohol in the dual-task context. Moreover, alcohol impair-
ment was specific to the dual-task situation, because no
alcohol impairment was evident when Task 2 was per-
formed as a single, individual task. The results also showed
that the coadministration of caffeine had a nonuniform
counteracting effect on the various aspects of performance
that were impaired by alcohol. Caffeine doses completely
antagonized the alcohol-induced impairment of the PRP
interference effect as measured by RT. However, caffeine
had no antagonizing effect on the alcohol-induced impair-
ment of accuracy. Thus, the speed of reactions was restored
by caffeine, but not the accuracy of these actions. The
coadministration of caffeine also attenuated subjective re-
ports of intoxication, as measured by the beverage ratings,
as the participants reported feeling less intoxicated when
caffeine was coadministered with alcohol as compared with
the same dose of alcohol alone. Thus, participants reported
reduced intoxication in response to caffeine coadministra-
tion despite the fact that aspects of their performance re-
mained impaired (i.e., accuracy).
In this study, we focused on the unique situation of
dual-task information processing, as dual-task demands af-
ford a unique and valuable model for understanding the
pharmacological effects of alcohol and caffeine, alone and
in combination. Although laboratory studies often use sin-
gle tasks to study the pharmacological effects of various
drugs on behavior, single tasks often place only modest
demands on participants’ cognitive functioning and thus
might limit the ability to observe drug-induced impair-
ments. By contrast, dual-task performance affords a more
complex and possibly more naturalistic model of typical
human information processing. People’s attention is rou-
tinely divided among multiple task demands, and the dual-
task model captures the complexity of this type of informa-
tion processing. For example, common tasks performed
outside the laboratory, such as driving, are inherently mul-
titask natured and can be further complicated by voluntarily
adding other tasks, such as talking on a cellular phone.
Moreover, such dual-task activities may be the norm rather
than the exception outside the laboratory. The recent appli-
cation of dual-task models in studies of alcohol effects on
behavior shows that performance in dual-task situations is
highly sensitive to the disruptive effects of alcohol. Perfor-
mance in dual-task contexts can be impaired by moderate
doses of alcohol that do not impair the individual compo-
nent tasks (Fillmore & Van Selst, 2002; Schweizer et al.,
2004). Accordingly, such single-task analyses may under-
estimate the magnitude of alcohol impairment associated
with moderate doses that might be observed outside the
To our knowledge, no prior studies have used dual-task
models to examine any drug interactions, such as caffeine-
induced antagonism of alcohol impairment. Previous re-
views of single-task studies and numerous individual single-
task studies have noted mixed findings with regard to caf-
cognitive and behavioral functions (Gratton-Miscio & Vo-
gel-Sprott, 2005; Hasenfratz, Bunge, Dal Pra, & Battig,
1993; Martin & Garfield, 2006; Nehlig et al., 1992). How-
ever, we have recently argued that the equivocal evidence
for caffeine antagonism of alcohol-induced impairment
might reflect the fact that not all cognitive impairments from
alcohol can be offset by the coadministration of caffeine
(Marczinski & Fillmore, 2003). In the current study, we also
observed a dissociation in caffeine-induced antagonism of
alcohol-induced impairment. The moderate doses of caf-
feine resulted in complete counteraction of alcohol-induced
increase in PRP interference, as measured by RT2s. How-
ever, these doses of caffeine did not counteract the alcohol-
induced impairments on Task 2 response accuracy. Consid-
ering that individuals felt less intoxicated when caffeine was
coadministered with alcohol, a potentially worrisome out-
come may exist when one cognitive function recovers yet
another does not. Subjective perceptions of intoxication
level may function as feedback for an individual to termi-
nate the drinking episode or to avoid potentially hazardous
activities (e.g., driving). However, an individual who per-
ceives less behavioral impairment or feels less intoxicated
because of caffeine coadministration with alcohol may de-
cide to continue drinking and/or engage in risky behaviors
such as driving.
There are several other factors that might mediate alco-
hol–caffeine interactions that were not examined in this
study. First, the study does not address the role of expect-
ancies, which may be a potentially critical variable in the
motivation to consume caffeinated alcoholic drinks. In the
current study, dose administration was blind so as to allow
determination of the pharmacological effects of these drug
Mean Beverage Rating Scale Scores
0.0 g/kg alcohol ? 2.0 mg/kg caffeine
0.0 g/kg alcohol ? 4.0 mg/kg caffeine
0.65 g/kg alcohol ? 0.0 mg/kg caffeine
0.65 g/kg alcohol ? 2.0 mg/kg caffeine
0.65 g/kg alcohol ? 4.0 mg/kg caffeine
Scores represent participants’ estimated amount of alcohol
consumed during a session, expressed in terms of standard drinks.
MARCZINSKI AND FILLMORE
combinations. However, individuals who drink caffeinated
alcohol in various forms are fully cognizant of the drug
combinations they are consuming. Expectation may play a
critical role in the level of impairment observed and con-
tribute to subjective perceptions of intoxication. Indeed,
previous studies have noted that individual expectations
regarding the counteracting effects of caffeine on alcohol-
induced impairment of performance play a large role in the
actual performance impairments observed (for a review, see
Fillmore, 1999). In illustrating the ironic effects of expect-
ancy, one study led individuals to expect that caffeine would
counteract the impairing effects of alcohol. Those individ-
uals displayed much greater impairment on a psychomotor
task as compared with individuals who held no such expec-
tation (Fillmore, Roach, & Rice, 2002). One explanation for
this finding is that individuals who expect caffeine to coun-
teract some of their behavioral impairment might be less
motivated to compensate and resist the impairing effects of
alcohol. Given the evidence for the role of drug expectan-
cies as motivators for drug use (Goldman, Del Boca, &
Darkes, 1999) and as mediators of the drugged response
(Vogel-Sprott & Fillmore, 1999), more information on the
role of expectancies in alcohol–caffeine interactions is
The findings of the current study may also be limited to
college-age moderate alcohol drinkers who typically con-
sume moderate doses of caffeine on a regular basis. Thus, it
is unclear how individuals who do not regularly use caffeine
might respond to these drug combinations. With regard to
alcohol use, heavier drinkers and binge drinkers might also
respond differently to these drug combinations. Time of
testing under a dose might also be an important factor. The
current study measured behavioral effects during the as-
cending limb of the blood alcohol curve. Previous work has
revealed that various cognitive processes differentially re-
cover over the course of the blood alcohol curve within one
drinking episode. For example, subjects who performed a
cued go/no-go task demonstrated some recovery in RT, but
not in response accuracy, on the descending limb as com-
pared with performance on the ascending limb (Fillmore,
Marczinski, & Bowman, 2005). As such, both timing and
caffeine coadministration may exacerbate the differential
recovery of cognitive processes on the descending limb as
compared with the ascending limb of the blood alcohol
curve. This scenario has practical implications for potential
risky behavior, as individuals are more likely to make
decisions about whether to drive on the descending limb,
once drinking has concluded.
In summary, the dual-task model appears to be a highly
sensitive cognitive measure of the pharmacological effects
of moderate doses of alcohol and caffeine, alone and in
combination. The results of the current study suggest that
the new alcoholic drink preferences for caffeinated alcohol
in various forms warrant further investigation. It appears
that caffeine coadministration does counteract some aspects
of performance that are impaired by alcohol (i.e., response
speed) but not others (response accuracy). This finding
raises important new questions about the coadministration
of caffeine with alcohol. It is important to determine the
extent to which such counteraction of alcohol impairment
by caffeine might contribute to alcohol abuse by increasing
risk of binge use or other harmful patterns of alcohol con-
sumption. Also, it is unknown what brain mechanisms
might be responsible for the lack of uniform counteracting
effects of caffeine on impairments of speed and accuracy
under alcohol. Functional magnetic resonance imaging
analyses of regional brain activity during dual-task perfor-
mance under these drug combinations could provide some
explanation for why caffeine might fail to restore response
accuracy. A brain region of particular interest might be the
anterior cingulate, which is involved in error monitoring
during performance of choice–response tasks similar to our
dual-task model (Hester, Fassbender, & Garavan, 2004).
American Medical Association. (2003). Proceedings from the Ed-
ucational Forum on Adolescent Health: Youth drinking patterns
and alcohol advertising. Chicago: Author.
Barone, J. J., & Roberts, H. R. (1996). Caffeine consumption.
Food and Chemical Toxicology, 34, 119–129.
Bonati, M., Latini, R., Galletti, F., Young, J. F., Tognoni, G., &
Garattini, S. (1982). Caffeine disposition after oral doses. Clin-
ical Pharmacology and Therapeutics, 32, 98–106.
Burns, M., & Moskowitz, H. (1990). Two experiments on alcohol–
caffeine interaction. Alcohol, Drugs, and Driving, 5–6, 303–
Cauli, O., & Morelli, M. (2005). Caffeine and the dopaminergic
system. Behavioural Pharmacology, 16, 63–77.
Ferreira, S. E., de Mello, M. T., & Formigoni, M. L. (2004). Can
energy drinks affect the effects of alcoholic beverages? A study
with users. Revista da Associacao Medica Brasileira, 50, 48–
Ferreira, S. E., de Mello, M. T., Pompeia, S., & de Souza-
Formigoni, M. L. O. (2006). Effects of energy drink ingestion
on alcohol intoxication. Alcoholism: Clinical and Experimental
Research, 30, 598–605.
Fillmore, M. T. (1999). Behavioral effects of caffeine: The role of
drug-related expectancies. In B. S. Gupta & U. Gupta (Eds.),
Caffeine and behavior: Current views and research trends (pp.
207–219). Boca Raton, FL: CRC Press.
Fillmore, M. T. (2003). Drug abuse as a problem of impaired
control: Current approaches and findings. Behavioral and Cog-
nitive Neuroscience Reviews, 2, 179–197.
Fillmore, M. T., Carscadden, J. L., & Vogel-Sprott, M. (1998).
Alcohol, cognitive impairment and expectancies. Journal of
Studies on Alcohol, 59, 174–179.
Fillmore, M. T., Marczinski, C. A., & Bowman, A. M. (2005).
Acute tolerance to alcohol effects on inhibitory and activational
mechanisms of behavioral control. Journal of Studies on Alco-
hol, 66, 663–672.
Fillmore, M. T., Roach, E. L., & Rice, J. T. (2002). Does caffeine
counteract alcohol-induced impairment? The ironic effects of
expectancy. Journal of Studies on Alcohol, 63, 745–754.
Fillmore, M. T., & Van Selst, M. (2002). Constraints on informa-
tion-processing under alcohol in the context of response execu-
tion and response suppression. Experimental and Clinical Psy-
chopharmacology, 10, 417–424.
ALCOHOL AND CAFFEINE
Fillmore, M. T., & Vogel-Sprott, M. (1995). Behavioral effects of
combining alcohol and caffeine: Contribution of drug-related
expectancies. Experimental and Clinical Psychopharmacol-
ogy, 3, 33–38.
Fillmore, M. T., & Vogel-Sprott, M. (1999). An alcohol model of
impaired inhibitory control and its treatment in humans. Exper-
imental and Clinical Psychopharmacology, 7, 49–55.
Fillmore, M. T., & Vogel-Sprott, M. (2000). Response inhibition
under alcohol: Effects of cognitive and motivational conflict.
Journal of Studies on Alcohol, 61, 239–246.
Franks, H. M., Hagedorn, H., Hensley, V. R., Hensley, W. J., &
Starmer, G. A. (1975). The effect of caffeine on human perfor-
mance, alone and in combination with ethanol. Psychopharma-
cology, 45, 177–181.
Fudin, R., & Nicastro, R. (1988). Can caffeine antagonize alcohol-
induced performance decrements in humans? Perceptual and
Motor Skills, 67, 375–391.
Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol
expectancy theory: The application of cognitive neuroscience.
In K. E. Leonard & H. T. Blane (Eds.), Psychological theories
of drinking and alcoholism (2nd ed., pp. 203–246). New York:
Gratton-Miscio, K. E., & Vogel-Sprott, M. (2005). Alcohol, inten-
tional control, and inappropriate behavior: Regulation by caf-
feine or an incentive. Experimental and Clinical Psychophar-
macology, 13, 48–55.
Hasenfratz, M., Bunge, A., Dal Pra, G., & Battig, K. (1993).
Antagonistic effects of caffeine and alcohol on mental perfor-
mance parameters. Pharmacology, Biochemistry, and Behav-
ior, 46, 463–465.
Hester, R., Fassbender, C., & Garavan, H. (2004). Individual
differences in error processing: A review and meta-analysis of
three event-related fMRI studies using the GO/NOGO task.
Cerebral Cortex, 14, 986–994.
Holloway, F. A. (1995). Low-dose alcohol effects on human
behavior and performance. Alcohol, Drugs, and Driving, 11,
Johnston, W. A., & Heinz, S. P. (1978). Flexibility and capacity
demands of attention. Journal of Experimental Psychology:
General, 107, 420–435.
Kerr, J. S., Sherwood, N., & Hindmarch, I. (1991). Separate and
combined effects of the social drugs on psychomotor perfor-
mance. Psychopharmacology, 104, 113–119.
Lader, M. H., & Bruce, M. S. (1989). The human pharmacology of
the methylxanthines. In I. Hindmarch & P. D. Sonier (Eds.),
Human pharmacology: Measures and methods (Vol. 2, p. 179).
Chichester, England: Wiley.
Liguori, A., & Robinson, J. H. (2001). Caffeine antagonism of
alcohol-induced driving impairment. Drug and Alcohol Depen-
dence, 63, 123–129.
Linnoila, M. (1974). Effect of drugs and alcohol on psychomotor
skills related to driving. Annals of Clinical Research, 6, 7–18.
Mandel, H. G. (2002). Update on caffeine consumption, disposi-
tion and action. Food and Chemical Toxicology, 40, 1231–1234.
Marczinski, C. A., & Fillmore, M. T. (2003). Dissociative antag-
onistic effects of caffeine on alcohol-induced impairment of
behavioral control. Experimental and Clinical Psychopharma-
cology, 11, 228–236.
Marczinski, C. A., & Fillmore, M. T. (2005). Alcohol increases
reliance on cues that signal acts of control. Experimental and
Clinical Psychopharmacology, 13, 15–24.
Martin, F. H., & Garfield, J. (2006). Combined effects of alcohol
and caffeine on the late components of the event-related poten-
tial and on reaction time. Biological Psychology, 71, 63–73.
McCusker, R. R., Goldberger, B. A., & Cone, E. J. (2003). Caf-
feine content of specialty coffees. Journal of Analytical Toxi-
cology, 27, 520–522.
McCusker, R. R., Goldberger, B. A., & Cone, E. J. (2006). Caf-
feine content of energy drinks, carbonated sodas, and other
beverages. Journal of Analytical Toxicology, 30, 112–114.
Nehlig, A., Daval, J., & Debry, G. (1992). Caffeine and the central
nervous system: Mechanisms of action, biochemical, metabolic
and psychostimulant effects. Brain Research Reviews, 17, 139.
Pashler, H. (1994). Dual-task interference in simple tasks: Data
and theory. Psychological Bulletin, 116, 220–244.
Riesselmann, B., Rosenbaum, F., & Schneider, V. (1996). Alcohol
and energy drink: Can combined consumption of both beverages
modify automobile driving fitness? Blutalkohol, 33, 201–208.
Rush, C. R., Higgins, S. T., Hughes, J. R., Bickel, W. K., &
Wiegner, M. S. (1993). Acute behavioral and cardiac effects of
alcohol and caffeine, alone and in combination, in humans.
Behavioural Pharmacology, 4, 562–572.
Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime
user’s guide. Pittsburgh, PA: Psychology Software Tools.
Schweizer, T. A., Jolicoeur, P., Vogel-Sprott, M., & Dixon, M. J.
(2004). Fast, but error-prone, responses during acute alcohol
intoxication: Effects of stimulus–response mapping complexity.
Alcoholism: Clinical and Experimental Research, 28, 643–649.
Selzer, M. L., Vinokur, A., & Van Rooijen, L. (1975). A self-
administered Short Michigan Alcoholism Screening Test
(SMAST). Journal of Studies on Alcohol, 36, 117–126.
Tormey, W. P., & Bruzzi, A. (2001). Acute psychosis due to the
interaction of legal compounds: Ephedra alkaloids in “Vigueur
Fit” tablets, caffeine in “Red Bull,” and alcohol. Medical Sci-
ence Law, 41, 331–336.
Van Selst, M., Ruthruff, E., & Johnston, J. C. (1999). Can practice
eliminate the PRP effect? Journal of Experimental Psychology:
Human Perception and Performance, 25, 1268–1283.
Vogel-Sprott, M. (1992). Alcohol tolerance and social drinking:
Learning the consequences. New York: Guilford Press.
Vogel-Sprott, M., & Fillmore, M. T. (1999). Learning theory and
research. In K. E. Leonard & H. T. Blane (Eds.), Psychological
theories of drinking and alcoholism (2nd ed., pp. 292–327).
New York: Guilford Press.
Weiss, B., & Laties, V. G. (1962). Enhancement of human per-
formance by caffeine and the amphetamines. Pharmacological
Reviews, 14, 1.
Received April 19, 2006
Revision received August 7, 2006
Accepted August 7, 2006 ?
MARCZINSKI AND FILLMORE