Caffeine fatigue and cognition

[ Pobierz całość w formacie PDF ]
Brain and Cognition 53 (2003) 82–94
www.elsevier.com/locate/b&c
Caffeine, fatigue, and cognition
Monicque M. Lorist
a,b,
*
and Mattie Tops
a,c
a
Experimental and Work Psychology, University of Groningen, Groningen, The Netherlands
b
Department of Medical Physiology, University of Groningen, Groningen, The Netherlands
c
Department of Psychiatry, University of Groningen, Groningen, The Netherlands
Accepted 21 July 2003
Abstract
Effects of caffeine and fatigue are discussed with special attention to adenosine–dopamine interactions. Effects of caffeine on
human cognition are diverse. Behavioural measurements indicate a general improvement in the eciency of information processing
after caffeine, while the EEG data support the general belief that caffeine acts as a stimulant. Studies using ERP measures indicate
that caffeine has an effect on attention, which is independent of specific stimulus characteristics. Behavioural effects on response
related processes turned out to be mainly related to more peripheral motor processes. Recent insights in adenosine and dopamine
physiology and functionality and their relationships with fatigue point to a possible modulation by caffeine of mechanisms involved
in the regulation of behavioural energy expenditure.
2003 Elsevier Inc. All rights reserved.
Keywords: Caffeine; Fatigue; Reaction time; EEG; ERP; Adenosine; Dopamine; Arousal
1. Introduction
Almost all caffeine comes from dietary sources (e.g.,
coffee, tea, and cocoa beverages). An important source
of caffeine for children includes chocolate bars and soft
drinks. Most of the coffee is consumed at home, while
the second preferred place of consumption is at work.
Especially at these work places, coffee is considered a
pleasant occasion to break working hours (DAmicis &
Viani, 1993).
Caffeine use is self-limiting; subjects do not gradually
increase the amount of caffeine normally used. In ad-
dition, the intake of a high dose of caffeine is not rein-
forced by positive and pleasant behavioural effects. The
addictive potential of caffeine has been questioned fre-
quently in the past. In a recent study Nehlig and Boyet
(2000) found that in rats the functional activation of the
shell of the nucleus accumbens, an area involved in
addiction and reward, was only induced by the highest
dose of caffeine (10mg/kg). These findings showed that
the usual human consumption level of caffeine fails to
activate reward circuits in the brain, and therefore
provide evidence that caffeine has only very low addic-
tive potential.
In the present paper evidence is discussed regarding
the effects of caffeine on human behaviour. Since caffeine
Coffee is a beverage known all over the world, and
millions of humans drink it everyday. A significant
proportion of the effects of coffee is related to the actions
of caffeine, the best-known pharmacologically active
constituent of coffee. The reasons for humans to con-
sume caffeine are manifold. The common belief is that it
affects the energetic state of subjects. There is indeed a
considerable amount of research illustrating that the use
of caffeine does result in increases of subjective energy
and alertness (Bruce, Scott, Lader, & Marks, 1986;
Gevins, Smith, & McEvoy, 2002; Lieberman, 2001; Yu,
Maskray, Jackson, Swift, & Tiplady, 1991; Zwyghuizen-
Doorenbos, Roehrs, Lipschutz, Timms, & Roth, 1990).
In addition to these stimulant effects of coffee, it is a
pleasurable experience to consume a cup of coffee for
most people, and caffeine intake, either acute or chronic,
appears to have only minor negative consequences on
health.
Corresponding author. Fax: +31-50-363-6304.
E-mail address:
m.m.lorist@ppsw.rug.nl(M.M. Lorist).
0278-2626/$ - see front matter 2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0278-2626(03)00206-9
*
M.M. Lorist, M. Tops / Brain and Cognition 53 (2003) 82–94
83
is associated with enhanced cognition and some aspects
of cognition are closely linked to specific neurotrans-
mitter systems, we will review the effects of caffeine and
try to correlate these data with known effects on neu-
romodulator systems. Behavioural, EEG, and ERP
indices of performance will be examined.
in almost all brain areas. The highest levels are found in
the hippocampus, cerebral and cerebellar cortex, and
certain thalamic nuclei (Fastbom, Pazos, & Palacios,
1987; Goodman & Snyder, 1982), while only moderate
levels are found in caudate-putamen and nucleus ac-
cumbens. The presence of presynaptic adenosine A
1
receptors mediating inhibition of transmitter release has
been demonstrated on virtually all types of neurons.
There is considerable evidence for a link between
adenosine A
1
receptors and dopamine D
1
receptors (see
Ferre, Fredholm, Morelli, Popoli, & Fuxe, 1997).
Adenosine A
2A
receptors are found to be concentrated
in the dopamine-rich regions of the brain. There is little
evidence for A
2A
receptors located outside striatum,
nucleus accumbens, and tuberculum olfactorium, al-
though functional data clearly suggests the presence of
A
2A
receptors in hippocampus and cortex. In the dorsal
striatum, core and shell regions of the nucleus accum-
bens and the tuberculum olfactorium A
2A
and dopamine
D
2
receptors were found to be co-localized.
Svenningsson, Nomikos, and Fredholm (1999) have
argued that blockade of A
2A
receptors in striatopallidal
neurons is crucial for the stimulatory action of caffeine.
In addition, there is ample evidence that an intact
dopaminergic neurotransmission is necessary for caf-
feine to be stimulatory (Ferre, Fuxe, Von Euler,
Johansson, & Fredholm, 1992). Moreover, it has been
shown that the effects of a low dose of caffeine can be
mimicked by a selective adenosine A
2A
receptor antag-
onist, but not by a selective adenosine A1 receptor
antagonist (Svenningsson, Nomikos, Ongini, & Fred-
holm, 1997). Therefore, it seems justified to conclude
that the interaction between caffeine in relevant doses
and the dopaminergic transmission is based principally
on enhancement of postsynaptic dopamine D
2
receptor
transmission.
Dopamine is vital for the regulation of motor be-
haviour (e.g., co-ordinated motion) and for association
learning linked to behavioural reinforcement. Moreover,
a loss in striatal dopamine has been associated with a
reduction in internally initiated control of behaviour;
external cues seem to control behaviour instead of in-
ternal cues (Robbins, 1997). The antagonistic actions of
caffeine at the A
2A
adenosine receptors in the striatum
2. Pharmacology of caffeine
After oral ingestion, caffeine is rapidly and almost
completely (99%) absorbed from the gastrointestinal
tract into the bloodstream (Arnaud, 1993; Fredholm,
Battig, Holmen, Nehlig, & Zvartau, 1999). Peak plasma
concentrations are reached in about 30–60min after
consumption. Caffeine is widely distributed throughout
the body, and it passes through all biological mem-
branes, including the blood–brain barrier and the pla-
cental barrier. The elimination of caffeine occurs
primarily by metabolism in the liver. Less than 5% is
recovered unchanged in urine. The half-life of caffeine is
approximately 3–5 h, although individual clearance rates
vary considerably. For example, the clearance rate is
speeded up with 30–50% by nicotine, while it is doubled
in woman taking oral contraceptives.
3. Mechanisms underlying the central effects of caffeine
Caffeine, at doses comparable to those of typical
human exposure, are primarily related to its actions to
block adenosine receptors (Daly, 1993; Fredholm et al.,
1999; Phillis, 1991). The ability of caffeine to block
adenosine effects on these receptors can be observed
already at low concentrations achieved after a single cup
of coffee. Other mechanisms of action (e.g., inhibition of
phosphodiesterase, mobilisation of intracellular cal-
cium) demand higher concentrations of caffeine, un-
likely to be reached by normal use of caffeine containing
dietary sources.
Pharmacological studies indicate that the CNS effects
of caffeine are mediated particularly by its antagonistic
actions at the A
1
and A
2A
subtypes of the adenosine
receptors (Table 1). Adenosine A
1
receptors are present
Table 1
Central adenosine receptors affected by typical human caffeine exposure
Receptor Localization
Types of neurons
Effect of caffeine
Caffeine action
A
1
Almost all brain areas, especially
hippocampus, cerebral and cerebellar
cortex, certain thalamic nuclei
All types of neurons (aspecific)
Especially linked to dopamine
D1 receptors
Antagonistic Disinhibition of
transmitter release
A
2A
Dopamine rich regions: striatum,
nucleus accumbens, tuberculum
olfactorium, hippocampus? cortex?
Co-localized with dopamine
D
2
receptors
Antagonistic
Increase transmission via
dopamine D
2
receptors
84
M.M. Lorist, M. Tops / Brain and Cognition 53 (2003) 82–94
are in accordance with the established reduction in risk
of developing Parkinsons disease with increasing levels
of caffeine consumption (Chen et al., 2001).
while the number of false alarms did not change. They
interpreted this improvement as evidence for an increase
in the rate at which relevant information about the
stimulus builds up in the processing system. These re-
sults indeed indicate that the information processing
system seems more sensitive to relevant stimulus char-
acteristics after caffeine.
On the other hand, Flaten and Elden (1999) exam-
ined the effects of caffeine on pre-pulse inhibition.
Pre-pulse inhibition is supposed to index attentional
pre-processing of a stimulus presented prior to a startle
eye-blink reflex-eliciting stimulus. Their results showed
that caffeine did not facilitate automatic attentional
processes. Kenemans and Verbaten (1998) also illus-
trated the absence of an effect of caffeine on attention.
They examined the effect of caffeine on various aspects
of selective attention. A cueing task was used in which
cues were presented either at the location of a sub-
sequent target or at an alternative location, and a task
was used in which relevant information was sur-
rounded by irrelevant information. Their study showed
that RTs were shorter after subjects had caffeine (1.5
and 3mg/kg), however, these effects were not depen-
dent upon attentional demands of specific task condi-
tions. Therefore, they concluded that the effects of
caffeine on behaviour were the result of improvements
in preparation and/or execution of motor responses,
rather than the result of an effect on the attention
system.
Rees, Allen, and Lader (1999) found improvements in
psychomotor performance in human subjects after a
moderate dose of caffeine. These effects of caffeine on
motor performance seem in accordance with the con-
clusion of Kenemans and Verbaten (1998). Moreover,
the relationship between caffeine and motor behaviour
has been supported in several investigations illustrating
that caffeine reduced the time required to execute a re-
sponse (e.g., Jacobson & Edgley, 1987; Smith, Tong, &
Leigh, 1977). However, the effects of caffeine on motor
performance are not always beneficial; negative or no
effects are reported, as well (see Battig, 1985; Fredholm
et al., 1999; Van der Stelt & Snel, 1998).
The observed behavioural effects of caffeine are very
diverse and, although not mentioned above, there are
complicated interactions between stimulant actions of
caffeine and the arousal level of subjects and the nature
of task requirements. Even though sophisticated ex-
perimental paradigms can be used, and specific actions
on cognitive functions can be defined with some confi-
dence, behavioural measures do not seem to be su-
cient to delineate precisely the specific actions of
caffeine on the human information processing system
(see also Gevins et al., 2002). An alternative approach
to delineate the effects of caffeine on human information
processing is to make use of more direct measures of
brain activity.
4. Behavioural effects of caffeine
The effects of caffeine on performance have been, and
still are examined in many studies. More than 90 years
ago, Hollingworth (1912) published the first placebo-
controlled and double blind study, in which the effects of
caffeine on human performance and sleep were exam-
ined. However, despite the large number of studies, it
seems dicult to arrive at a coherent account of effects
of caffeine on human performance.
In general, observations point to an inverted
U-shaped dose–response curve for caffeine; lower doses
have positive effects on performance, while doses above
500mg cause a decrease in performance (e.g., Anderson
& Revelle, 1983; Patat et al., 2000). Similarly, lower
doses of caffeine are reliably associated with ‘‘positive’’
subjective effects, while higher doses of caffeine lead to a
clear increase in measures of anxiety and tension (e.g.,
Loke, 1988; Thayer, 1989).
Human information processing consists of many
cognitive operations ranging from the perception of
information to the selection and subsequent execution
of an action (e.g., button press). In addition, adequate
and ecient performance relies on higher-level cogni-
tive control processes, such as planning and prepara-
tion of activities. Although there is no strong
agreement on the effects of caffeine on specific cognitive
operations, there are indications that caffeine affects the
attention system. Central to the idea of attention is
that we can actively manipulate the impact that per-
ceptual stimuli have on our information processing
system (Kanwisher & Wojciulik, 2000). Attention can
act as a multiplier of the neural response to relevant
information, or can diminish the impact of irrelevant
information. Thus, attention can be used to actively
prepare or bias the human information processing
system for the processing of specific stimulus features
(Kastner, Pinsk, De Weerd, Desimone, & Ungerleider,
1999).
Using a paper and pencil version of a visual search
task, Marsden and Leach (2000) showed an increase in
performance eciency with caffeine. After 250mg black
coffee without sugar, subjects detected more targets
compared to a placebo condition. Ruijter, Lorist, Snel,
and De Ruiter (2000c) used a computer version of a
sustained attention task. They found, after a similar
dose of caffeine, that subjects showed higher levels of
perceptual sensitivity for relevant stimulus characteris-
tics, as indicated by the signal detection parameter
A
0
.In
line with these findings, Kenemans and Lorist (1995)
showed an increase in hit rate after caffeine treatment,
M.M. Lorist, M. Tops / Brain and Cognition 53 (2003) 82–94
85
5. EEG effects of caffeine
Caffeine is regarded as a mild stimulant acting on the
central nervous system, producing diverse and complex
effects, even when consumed in small quantities (Dews,
1984; Garattini, 1993). Behavioural indices of perfor-
mance may not provide an accurate picture of these
subtle and complex effects. Instead, measures of cortical
brain activity, regarded as an index of cortical arousal
(Rainnie, Grunze, McCarley, & Greene, 1994), might
serve as a more sensitive indication of the stimulating
effects of caffeine on brain functioning.
The electroencephalogram (EEG) shows more acti-
vation and changes towards faster frequency and lower-
amplitude activity with increasing arousal. Already
Gibbs and Maltby (1943) observed these effects after
subjects were treated with caffeine. A robust finding
observed in a number of studies concerns the reduction
after caffeine treatment of power in the lower a or h
band (6–9Hz; Bruce et al., 1986; Etevenon et al., 1989;
Newman, Stein, Trettau, Coppola, & Uhde, 1992;
Saletu, Anderer, Kinsperger, & Grunberger, 1987).
Kenemans and Lorist (1995) found similar changes in
brain-state indexed by the background EEG power
spectrum. The most pronounced effect was found in the
lower a range, while in the higher a and d range the effect
was smaller. In a study of Gevins et al. (2002), 200mg
caffeine did not elicit changes in resting EEG, however a
reduction in a band power was observed during task
performance. Contrary to these EEG effects, Gevins and
colleagues failed to find effects of caffeine on behavioural
measures. Patat et al. (2000) reported that caffeine
(600mg, slow release formulation) was able to coun-
teract the effects of sleep deprivation (36 h) on the EEG,
that is, caffeine increased the relative power in the a and
b frequencies, while it decreased h and d power. Jones,
Herning, Cadet, and Griths (2000) measured EEG for
3min while subjects had their eyes closed in order to
examine caffeine withdrawal effects. The effects illus-
trated that conform to the expectations caffeine with-
drawal decreases alertness as reflected in increased EEG
h power. In sum, the EEG data indeed supports the
stimulating effects of caffeine, although effects on specific
cognitive activities cannot be distinguished, using this
measure.
Fig. 1. A schematic representation of the actions of caffeine on the
human information processing system. Physiological and behavioural
indices thought to be related to different processes are depicted next to
the concerning processes (LRP, lateralized readiness potential; RT,
reaction time).
potentials (ERPs) are more convenient. ERPs are se-
quences of voltage deflections in the spontaneous elec-
trical activity of the brain, which are time-locked to
particular events such as the onset of a stimulus. They
are revealed, by averaging brain activity recorded during
many trials. ERPs can be recorded on trials in which
stimuli are presented to which a response is or should be
given, and stimuli that should be ignored, all within the
same experimental task.
In Table 2 those studies, which sought to establish the
effects of caffeine on the central nervous system, using
ERP measures, are presented.
6.1. Attention
The behavioural effects of caffeine indicated that
caffeine affects the attention system. Attention can
modify neural activity in specific cortical areas, which
are involved in the perceptual analysis of relevant
stimulus information (e.g., Kanwisher & Wojciulik,
2000), that is, attention may enhance the responsivity of
cells to specific stimulus features. Lorist et al. (1994a)
studied feature-based attention by examining the effect
of irrelevant information on the processing of relevant
information. A task was used in which stimulus quality
was manipulated, which is supposed to affect feature
extraction processes (Sanders, 1983). The non-degraded
stimuli consisted of a dot pattern surrounded by a
rectangular frame of dots. In the degraded condition,
dots were replaced from the frame into random position
within the frame. The spatial arrangements of the dot
patterns impaired the identification of the stimulus, as
reflected in increased RTs and decreased accuracy.
Caffeine had an effect on both the latency and amplitude
6. ERP effect of caffeine
Behavioural measures do not provide direct infor-
mation about the effects of caffeine on brain function.
These measures (e.g., RTs, errors) form the end product
of many different cognitive operations (see Fig. 1). To
delineate the specific effects of caffeine on the timing and
organisation of cognitive processes occurring in the
brain during task performance, event-related brain
 86
M.M. Lorist, M. Tops / Brain and Cognition 53 (2003) 82–94
Table 2
Effects of caffeine treatment on the amplitude of ERP com
ponents
N1
P2
N2b
P3
LRP RT
Accuracy Caffeine dose
Spilker and Callaway
(1969)
—
300/500mg
(dependent
on daily use)
Ashton, Millman,
Telford, and
Thompson (1974)
m (N1–P2) CNV: m
.
300mg
Elkins et al. (1981)
—
.
3/10mg/kgBW
Wolpaw and Penry
(1978)
m (Absence
of decrease
observed in
placebo)
—
300mg
Lorist, Snel, and
Kok (1994a)
m
.
m
.
m
200 + 50mg
Latency.
Lorist, Snel, Kok,
and Mulder (1994b)
m
m
m
m
.
—
200 + 50mg
Kenemans and Lorist
(1995)
Early positivity
(Cz/Pz)
Latency.
—
.
m
3mg/kg BW
Lorist, Snel, Mulder,
and Kok (1995)
m
.
. (No effects for
high display load)
—
3mg/kg BW
Latency.
Lorist, Snel, Kok,
and Mulder (1996)
m
.
m
m
.
—
200 +50mg
Lorist and Snel (1997)
m
Onset.
—
—
3mg/kg BW
Ruijter, Lorist, and
Snel (1999)
—
Fz: m
m
.
—
1, 3, and
7.5mg/kg BW
Ruijter, De Ruiter,
and Snel (2000a)
FPz: m
Att:mm(Targets)
.
—
250mg
Unatt .
Ruijter, De Ruiter,
Snel, and Lorist
(2000b)
—
Fz: m
N2: m
—
m (Hits) 250mg
m (A
0
)
Ruijter et al. (2000c)
Fz: m
m
—
—
250mg
., decrease; m, increase;
)
, no effect; and BW, body weight.
of the early exogenous N1 component. It was concluded
that caffeine indeed increased the receptivity of subjects
to external stimuli and moreover accelerates perceptual
processing.
In a selective search task in which subjects had to
search for a target letter on relevant spatial positions, a
similar enhancement of the N1 was found (Lorist et al.,
1994b; Lorist et al., 1995). However, this effect was
consistent across stimulus conditions; it was not limited
to relevant stimuli. Caffeine effects on the N1 component
were not always present (Elkins et al., 1981; Kenemans
& Lorist, 1995; Spilker & Callaway, 1969). Ruijter et al.
(2000a, 2000b) neither observed an effect on the N1
component in a task in which subjects had to attend
selectively to colour features nor in a task in which
subjects had to attend selectively to spatially arranged
bars of a specific size. They did report an enhancement
of the exogenous frontal P2 component in the caffeine
condition, which was interpreted as evidence supporting
a more general increase in responsivity of caffeine to
information, irrespective of stimulus relevance.
The effects of caffeine on selective visual attention
were also examined in a study by Kenemans and Lorist
(1995). Stimulus selection criteria in this study were
spatial frequency and orientation. Kenemans and Lorist
reported an increased positivity, specifically elicited by
targets and frequency relevant stimuli, in the ERP in the
60–150ms time interval after stimulus presentation. In
this study subjects performed virtually perfect concern-
ing the rejection of stimuli containing irrelevant spatial
frequencies. Improvements in the caffeine condition
therefore might be related to improvements in the
analysis of orientation. The observed positivity might be
a reflection of the orientation of stimuli, which have
relevant frequency characteristics.
The ERP results seem to be in agreement with theo-
ries of visual attention. Effects on the N1 appear to be
linked exclusively to spatial attention and are absent
during attention to non-spatial stimulus features such as
colour, size or spatial frequency (Hillyard, Mangun,
Woldorff, & Luck, 1995). Attention to these non-spatial
features is indexed by endogenous longer latency com-
ponents (e.g., N2b, P3). In addition to the P2 effects
mentioned earlier, Ruijter et al. (2000a) indeed reported
effects of 250mg caffeine on the N2b component,
reflecting active orienting towards relevant stimulus
features. The enlargement of the N2b component
in response to relevant stimuli and the smaller N2b
[ Pobierz całość w formacie PDF ]