You are on page 1of 8

Physiology & Behavior 73 (2001) 585 – 592

Cognitive demand and blood glucose


Andrew B. Scholey*, Susan Harper, David O. Kennedy
Human Cognitive Neuroscience Unit, Division of Psychology, University of Northumbria, Newcastle upon Tyne, NE1 8ST, UK
Received 14 September 2000; received in revised form 22 January 2001; accepted 19 March 2001

Abstract

Previous research has identified that glucose administration can enhance cognitive performance, especially during more intense cognitive
processing. There appears to be a reciprocal relationship between falling glucose levels and cognitive performance, particularly under
conditions of cognitive demand. The present placebo-controlled, double-blind, balanced, crossover study examined the possibility that a high
cognitive load may produce changes in blood glucose levels. A secondary aim was to examine the effects of glucose on tasks of varying
cognitive demand. The effects of a glucose drink on participants’ performance of a serial subtraction task (computerised Serial Sevens), a
somatically matched control task (key-pressing), a short interval Word Memory task and a Word Retrieval (Verbal Fluency) task were
assessed. The change in blood glucose during the demanding computerised Serial Sevens was compared to the change occurring during the
key-pressing control. Glucose consumption significantly improved performance on Serial Sevens, with a trend for improved performance on
Word Retrieval and no effect on the Word Memory task. Compared with the control task, Serial Sevens resulted in a significant reduction in
blood glucose in both drink conditions. This accelerated decay was significantly greater following glucose than placebo. It is suggested that
the amount of cognitive load associated with task performance is an index of its sensitivity to enhancement by glucose. Furthermore, a period
of intense cognitive processing leads to a measurable decrease in levels of peripherally measured blood glucose, which may be linked to
increased neural energy expenditure. However, the relative contribution of central and peripheral (e.g. cardiac) activity to this effect has yet to
be determined. D 2001 Elsevier Science Inc. All rights reserved.

Keywords: Glucose; Cognition; Cognitive demand; Cognitive load; Mental effort

1. Introduction [28]. It seems plausible that the degree of cognitive demand


is a key factor in a task’s susceptibility to enhancement by
There is a wealth of evidence documenting the beneficial glucose and other metabolic substrates [7,8,14,21,22,29,30].
effects of a glucose drink on cognitive performance in To assess this, Kennedy and Scholey [14] directly compared
healthy young adults. This effect has been reported as a the effects of a glucose drink on three tasks of differing
direct consequence of a glucose load or in relation to the cognitive demand (as measured by subjective ratings and by
extent of change in blood glucose following the drink. the level of physiological arousal engendered). Glucose
While a great deal of research in this field has focused on significantly improved performance only on the task which
declarative verbal memory, a number of reports have was rated as being the most (subjectively) cognitively
emerged documenting glucose-related enhancement in other demanding and which elicited the highest level of physio-
cognitive domains. These include beneficial effects on logical arousal.
reaction times [23], rapid visual information processing Following a glucose load, blood glucose levels rise,
[5,8], the Stroop paradigm [5], Porteus maze and Block peak, then fall back to baseline levels over the subsequent
Design tasks [7], the Brown –Peterson working memory hour or so. A number of studies have identified a clear
task [17], driving simulator performance [15], face recog- association between the rate at which glucose levels fall and
nition [20], serial subtractions [14] and kinaesthetic memory better cognitive performance. Arousal-linked mechanisms
certainly contribute to the post-apex fall in blood glucose
during cognitive processing. However, it is possible that the
* Corresponding author. Tel.: +44-191-227-4468; fax: +44-191-227-
phenomenon also involves an increased uptake of blood
3190. glucose by active brain substrates. Most previous research in
E-mail address: a.scholey@unn.ac.uk (A.B. Scholey). the field has relied on inference to examine the relationship

0031-9384/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved.
PII: S 0 0 3 1 - 9 3 8 4 ( 0 1 ) 0 0 4 7 6 - 0
586 A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592

between performance and glucose, either by correlating the The aims of this study were therefore twofold. The
extent of the change in blood glucose with cognitive primary aim was to assess the effects of cognitive demand
measures [14] or by comparing performance of participants on blood glucose levels both in the presence and absence of
allocated to groups on the basis whether their change in a glucose load. This was achieved by direct comparison of
blood glucose levels lay at the extremes of the sample the change in blood glucose levels during Serial Sevens
distribution [5,7,8,24]. with the change occurring during the somatically matched
Donohoe and Benton’s [8] results are directly relevant to control task. A secondary aim was to further examine the
the present experiment. In their study, following a glucose effects of a glucose drink on three tasks of differing
or placebo drink, participants sat quietly or performed the cognitive demand. One of these was the cognitively
demanding rapid visual information processing task for 10 demanding computerised Serial Sevens task described
min. Subsequently, those in the demand condition had above. The other two tasks (Word Memory and Verbal
significantly lower blood glucose levels than those who Fluency) have been reported to be susceptible to glucose
had sat quietly. From the data presented, it is not possible administration, but here task parameters were deliberately
to ascertain the extent to which this effect would have changed to reduce task demands.
been evident if the two drink conditions had been con-
sidered separately. Nevertheless, the findings are consis-
tent with the notion that there is increased glucose 2. Methods
utilisation during more intense cognitive processing. It is
possible that increased neural activity contributes to these 2.1. Participants
effects. Such an interpretation is not without problems
however. Firstly, cognitive demand is itself associated with Eleven male and nine female volunteers (mean age 22.7
arousal, including increased heart rate [3,6,14,32], and years, range 20 – 30) took part in the study. Prior to partic-
secondly, task performance involves somatic energy ipation, each gave informed consent and completed a
expenditure at the very least by the musculature involved medical health form as approved by the University of
in response generation. Both of these processes will Northumbria Division of Psychology Ethics Board.
inevitably involve cellular uptake of glucose, and while
the former is more difficult to control for, the latter can be 2.2. Cognitive measures
partially addressed by the inclusion of an appropriately
matched control task. 2.2.1. Computerised Serial Sevens
Serial Sevens involves the repeated subtraction of seven Serial Sevens involves presenting a participant with a
from a given starting number. Previous work in our labo- starting number from which they must subtract 7, then
ratory has shown that glucose significantly enhances per- subtract 7 from the resultant number and so on. The original
formance on this task [14] and that the degree of fall in (verbal) Serial Sevens test [12] has appeared in a number of
blood glucose correlates with individuals’ performance. forms and is sensitive to both lowered [31] and raised [14]
Furthermore, Serial Sevens was rated as highly demanding blood glucose levels.
by participants and performance of the task was associated A computerised version of the Serial Sevens test was
with an increase in heart rate — although a control task in implemented. A randomly generated starting number
which participants counted upwards at the same rate did not between 800 and 999, which disappeared at the first
elicit this response [14]. In the present study, a computerised key-press, was displayed on the monitor of a desktop
form of the task was implemented in which responses were computer. Responses were made in the form of three-digit
made via the numeric keypad of a desktop computer. This numbers, which were recorded via the numeric keypad.
allowed the use of a somatically matched control task with a Each key-press resulted in an asterisk being displayed in
negligible cognitive load — a similar number of repeated the first, second or third position, as appropriate, in a box
key-presses over the same task period — thus going some on the screen. Pressing the ‘enter’ key signaled the end of
way to overcoming one of the drawbacks of earlier experi- each response and cleared the three asterisks in the box
ments described above. ready for the next three-digit number. We have preliminary
A second complication of previous investigations relates (unpublished) data suggesting that performance remains
to the timing of task performance with respect to a glucose reasonably stable over several minutes; thus, the task may
load. Such studies have tended to use an interval of 20 min draw increasingly on cognitive resources over time. It was
between drink consumption and any cognitive processing. therefore felt that any effect on blood glucose may be
At this time point, typically, peoples’ blood glucose levels more pronounced after a period of cognitive processing.
are still rising. It was felt that any impact of cognitive load For this reason, the duration of the task was increased from
on glucose levels may be more pronounced if superimposed the 2 min used in previous studies to 5 min following the
onto falling blood glucose. For this reason, in the present first key press.
study, a longer glucose-task interval of 45 min was used for The task was scored for both total number of subtractions
this aspect of the experiment. and number of errors. In the case of a mistake, subsequent
A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592 587

responses were scored as positive if they were correct in Previous research has found that 25 g of glucose is a
relation to the new number. suitable dose to produce enhanced performance on both
memory and non-memory cognitive tasks (e.g. [Ref. 14]).
2.2.2. Serial Sevens control
A control task was devised to involve a similar amount of 2.3.3. Procedure
somatic/peripheral energy expenditure as the Serial Sevens Participants attended two sessions, one for each of the
but with a minimal cognitive load. The computer generated two drink conditions, which took place 1 week apart,
a metronome tone at a rate of 20 beeps per minute. and for each participant at the same time of day.
Participants were required to make four key presses at each Sessions lasted approximately 1 hour and were con-
tone — the number ‘5’ in the centre of the numeric keypad ducted on an individual basis in the same environment
three times followed by the ‘enter’ key. Again, the task and by the same experimenter (who was blind to
duration was 5 min following the first key-press. This participants’ drink condition) throughout the study. All
frequency was based on the results of Kennedy and Scholey testing took place between 09:00 and 13:00 h and
[14] where 15 – 20 responses per minute were typically participants were requested to fast from midnight on
generated in a verbal version of the task. The present form the nights before testing.
therefore involved similar, or slightly greater, somatic On arrival at the first session, participants were randomly
energy expenditure. allocated to one of two drink conditions (glucose or pla-
cebo), and the first blood glucose reading was taken. The
2.2.3. Word retrieval participant then consumed the glucose or saccharine drink,
A 2-min Word Retrieval (Verbal Fluency) Task was studied the word list for 5 min and sat quietly until 40 min
utilised. A relatively easy form of the task was used where following the drink. The second blood glucose sample was
participants were asked to generate (out loud) as many taken and, starting at 45 min following the first blood
words as they could beginning with either the letter ‘‘S’’ glucose reading, the participant completed either the com-
or the letter ‘‘A’’. This task has been described as assessing puterised Serial Sevens task or the key-pressing control task
‘‘access to and retrieval of information stored in long-term for 5 min. Immediately following this, a third blood glucose
memory’’ [11]. reading was taken and the participant completed the other
task (Control/Serial Sevens) for 5 min. This was directly
followed by a fourth and final blood glucose measurement,
2.2.4. Word memory then the test phase of the Word Memory task (60 s) and the
Participants were presented with a sheet containing one Word Retrieval (Verbal Fluency) task (2 min). The second
of two lists of 15 words, which were matched for frequency visit was similar except that participants were in the oppo-
and concreteness. Again, task parameters were designed to site drink condition. At the end of their second visit, each
minimise the cognitive load. Participants were allowed 5 participant was thanked and debriefed.
min to study this list. Recall consisted of writing down as The order of word lists, letters used for Verbal
many of the words from the list as possible within 1 min. Fluency and whether computerised Serial Sevens pre-
The task was scored as number of correctly recalled words. ceded or followed the key-pressing control were balanced
across participants.
2.3. Physiological measures
2.3.4. Statistics
2.3.1. Blood glucose The blood glucose level data from the placebo and
Blood samples were taken with ‘Boehringer Soft Touch’ glucose conditions were analysed by two-way analysis of
single-use blood sampling lancets (Boehringer, Germany). variance (ANOVA), the factors being measurement [first
Blood glucose levels were measured using a Lifescan One (baseline), second, third and fourth]  condition [Placebo/
Touch system (Johnson and Johnson, UK), which uses an Glucose] with repeated measures on both factors. Appro-
enzymatic method similar to that reported to be both priate preplanned comparisons were made using the Bon-
accurate and consistent [18]. ferroni t statistic.
The effects of drink on performance scores for Serial
2.3.2. Drinks Sevens (both number of responses and errors), Word
On each visit, participants received either a glucose or a Retrieval (responses) and Word Memory (words) were
placebo drink (depending on the condition to which they compared by two-way ANOVA, drink order (glucose first/
were allocated). The glucose drink was made up of 25 g placebo first)  drink (glucose/placebo) with the latter as a
glucose powder dissolved in 250 ml of water and 25 ml within-subjects factor.
‘Summer Magic,’ no added sugar, apple and blackcurrant To assess the effects of computerised Serial Sevens and
squash. The placebo drink was similar but with 30 mg of the key-pressing control on blood glucose levels, the pre-task
saccharine substituting for the glucose. The two drinks were blood glucose level was subtracted from the corresponding
similar in taste and ‘mouth feel.’ post-task level for each drink condition. The resulting
588 A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592

measures were subjected to a two-way ANOVA, drink 3.2. Effects of glucose on cognitive measures
(glucose/placebo)  task (Serial Sevens/control). Order
effects on this measure were considered separately using There was a significant effect of glucose on performance
two-way ANOVAs, task order  drink order [1]. of the Serial Sevens task [ F(1,18) = 7.95, P=.011]. Individ-
uals generated significantly more responses when in the
3. Results glucose condition than when in the placebo condition (see
Fig. 2a). The number of errors made during Serial Sevens
3.1. Blood glucose levels did not differ significantly between the two conditions
[ F(1,18) = 2.05, P = .17]. However, the glucose condition
There were significant main effects of condition (glu- was associated with numerically fewer errors (mean = 4.85
cose/placebo) [ F(1,19) = 19.52, P < .001] and of measure- compared with 6.15 in the placebo condition), ruling out the
ment [first (baseline), second, third and fourth reading] possibility of a condition-specific ‘‘speed-accuracy trade-
[ F(1,19) = 24.41, P < .001]. There was also a significant off’’ on this measure.
condition  measurement interaction [F(3,57) = 23.25, For the Verbal Fluency task, there was a strong trend for
P < .001; Fig. 1]. an increased number of responses in the glucose condition
Planned Bonferroni t tests comparing blood glucose compared with the placebo condition [ F(1,18) = 3.98,
readings between the placebo and glucose conditions at P = .061; see Fig. 2b]. There was also a significant inter-
each of the four sampling times confirmed glucose absorp- action between performance and drink order on this measure
tion. While there was no significant difference between [ F(1,18) = 4.95, P = .039]. There were no other similar
baseline glucose levels [t(57) = 0.36, n.s.], glucose readings significant interactions with drink order. There was no
following the drink were significantly elevated in the significant difference between the placebo condition and
glucose condition compared to the placebo at the second the glucose condition on Word Memory performance
[t(57) = 10.85, P < .01], third [t(57) = 6.06, P < .01] and [ F(1,18) = 0.28, P = .603; see Fig. 2c].
fourth [t(57) = 7.07, P < .01] readings.
3.3. Effects of cognitive demand on blood glucose levels

There were no significant task order or drink order


interactions with any of the task-related blood glucose
measures. There was a significant main effect of task on
the change in blood glucose [ F(1,18) = 11.38, P = .003]. The
magnitude of fall in blood glucose during 5 min of Serial
Sevens was significantly greater than the fall during the 5
min of the key-pressing control task regardless of the type of
drink consumed. There was also a significant main effect of
drink condition [ F(1,18) = 8.59, P = .009]. The change in
blood glucose was greater in the glucose condition than in
the placebo condition irrespective of task. These data are
presented graphically in Fig. 3.

4. Discussion

These findings confirm that glucose administration is


capable of enhancing cognitive performance in healthy young
adults. Furthermore, they lend support to the notion that more
cognitively demanding tasks are most affected by glucose.
The results also demonstrate that a high cognitive load can
itself result in a significant uptake of blood glucose — an
effect which can be detected by the extent of change in
peripheral blood.
Fig. 1. Blood glucose levels immediately before, 40 min and two later time- The significant improvement of computerised Serial
points following consumption of a placebo (6) or glucose (.) drink. Each Sevens performance (Fig. 2a) is consistent with previous
point represents mean ± standard error ( ** P < .01 compared to the placebo
findings from our laboratory [14], where the number of
condition at the corresponding time point). Timing of the placebo/glucose
drink and the 5-min tasks used to assess the impact of cognitive load are verbally generated responses was similarly increased by
indicated. The order in which Serial Sevens or the control task were glucose administration. The results reported here suggest
performed as Tasks A and B was balanced across participants. that the computerised version of the task may be a useful
A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592 589

with drink order. However, improved performance on this


measure is consistent with reports of a similar trend on the
same task [14], suggesting that the ability to access
information from long-term memory may, in some circum-
stances, be susceptible to enhancement by glucose — an
effect which has been identified previously in the elderly
[16]. It seems possible that if the task parameters were
changed slightly, it would be more sensitive to glucose.
Indeed, Donohoe and Benton [7] reported a highly signifi-
cant ( P < .001) effect of glucose on a Verbal Fluency task
with higher demands (utilising three letters of relatively
lower frequency each for 1 min).
The lack of a significant improvement in Word Memory
scores (Fig. 2c) appears to be somewhat at odds with certain
reports. However, closer inspection of the literature sug-
gests a number of reasons for this discrepancy. Not all
studies have reported a direct effect of glucose on verbal
memory. For example, the effect has not always been
observed in young adults [2] and, when present, it appears
to be less pronounced in younger than elderly participants
[10]. Those reports of positive effects of glucose on verbal
recall have tended to rely on relatively large groups of
participants to achieve significance at a levels of .05. On
the other hand, previous studies have used different
amounts of glucose. For example, Benton and Owens [4]
used a 50 g drink and found an association between blood
glucose and word recall scores.
In the Word Memory task used in the present study,
overlearning of target material was deliberately allowed
during the 5-min presentation phase, possibly resulting in
ceiling performance. The results do raise the possibility that
it is the level of demand rather than the type of cognitive
domain measured by a task which is the primary factor in
determining its susceptibility to enhancement by glucose.

Fig. 2. Cognitive performance for the placebo (white bars) and glucose
(black bars) conditions. (a) Number of subtractions performed during
computerised Serial Sevens; (b) number of words generated during the Word
Retrieval (Verbal Fluency) task; (c) number of words recalled in the Word
Memory task. Each bar depicts mean ± standard error. There was a
significant effect of condition on Serial Sevens performance ( * P < .05
compared to the corresponding placebo score) and a trend for enhanced
performance in the Verbal Fluency task ( + P = .060). There was no difference
between conditions for scores on the Word Memory task ( P = .603).
Fig. 3. Effects of a task with a high cognitive load on blood glucose levels
for the placebo (white bars) and glucose (black bars) condition. The graph
tool for the automated assessment of the effects of cogni-
represents the change between pre- and posttask blood glucose levels
tive demand. The trend for improved performance on the during a control (key-pressing) task and computerised Serial Sevens. Each
Word Retrieval (Verbal Fluency) task presented in Fig. 2b bar depicts mean ± standard error ( ** P < .01 compared to the correspond-
is difficult to interpret due to the significant interaction ing placebo condition; *** P < .005 between tasks).
590 A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592

Foster et al. [9] reported a significant main effect of conditions (placebo and glucose). The relatively modest fall
glucose, using a relatively low N (10 per group), on free in the control-placebo condition may simply reflect chance
recall using a modified California Verbal Learning Task. fluctuations in glucose levels possibly coupled with uptake
Interestingly, during learning of target material, they by musculature involved in generating the 400 key-presses
required participants to simultaneously perform self-gener- during the 5 min of the control task. The significantly
ated alternation of motor sequences (where subjects may greater fall in blood glucose levels in the control-glucose
have believed that performance was being assessed), and condition is almost certainly a consequence of these mech-
the mnemonic load of the task was deliberately increased to anisms coupled with the cellular uptake of glucose pro-
prevent ceiling effects. Additionally, Messier and Gagnon moted by the endocrine factors involved in regulating blood
[19] identify complex, rather than simple, verbal declara- glucose (including insulin secreted in response to the
tive memory as the impairment which is most amenable to glucose load). This effect will be augmented to some degree
reversal by glucose. This is consistent with studies of by the uptake of glucose by cardiac muscle since part of the
rodent spatial working memory; glucose did not increase physiological response to glucose ingestion itself includes
the level of rats’ spontaneous alternation in a three-arm an increase in heart rate [14].
radial maze [25], whereas performance on a more difficult Moving to the Serial Sevens-placebo condition, the fall
four-arm plus version was enhanced by both systemic and in blood glucose level is of a significantly greater magni-
central injection of glucose [26,27]. While not wishing to tude than in the control-placebo condition. This may be a
overinterpret the present data, comparing the trend for a manifestation of several processes which constitute a
glucose-associated improvement in the Verbal Fluency task compound response to cognitively loaded processing.
(involving retrieval of past-learned material) with the non- These include increased heart rate [14], other physiological
significant effect on Word Memory (involving the retrieval indices of arousal and possibly the delivery of glucose to
of recently overlearned material) supports the contention active brain mechanisms. The fact that this change in
that it is the cognitive demand rather than the cognitive blood glucose is greatest in the Serial Sevens-glucose
domain of a task which dictates its susceptibility to condition must be a reflection of these processes coupled
enhancement by glucose. with the additional glucoregulatory endocrine effects.
How well is this notion supported by findings from Given this pattern of results, it is tempting to tentatively
investigations into the effects of a glucose drink on speculate that cognitive demand may also result in an
performance in non-mnemonic arenas? Glucose has been elevated rate of neural glucose metabolism, which in turn
shown to enhance performance on later (but not earlier) facilitates task performance.
phases of driving simulator performance [15], choice (but Interestingly, this magnitude of change in blood glucose
not simple) reaction time [23], later rather than earlier due to a high cognitive load is similar to that presented by
periods of rapid visual information processing [5,8], incon- Donohoe and Benton [8] where the difference between the
gruent (i.e. more difficult) trials of the Stroop paradigm [5], quiet control and the cognitive demand condition was in the
more difficult examples of the Porteus maze and Block order of 0.5 –0.7 mmol/l glucose (Donohoe and Benton [7],
Design tasks [7] and more demanding (Serial Sevens) but p. 417, Fig. 2). In the present study, the change due to
not easier (Serial Threes) serial subtractions [14]. Such cognitive demand was 0.4 mmol/l (combining the placebo
findings are consistent with the idea that the degree and/or and glucose data) — a comparable figure despite differences
duration of cognitive demand is an important factor con- in the task used, the amount of glucose consumed, the
tributing to the extent to which task performance may be timing of the task relative to the temporal haemodynamics
enhanced by glucose. of glucose, and the fact that no somatic matching of a
The matter is not straightforward. Clearly, for facilitation control task was attempted in their study. There are numer-
to occur, there must be ‘room for improvement’ due either ous methodological factors, which may contribute to the
to a cognitively impaired sample population or to the task similarity in these results, and clearly we are some way from
parameters themselves. Studies need to be performed into quantifying such effects.
the effects of glucose on relatively pure cognitive domains, The idea of increased utilisation of glucose during
where task demands can be systematically titrated, as well intense cognitive processing is consistent with previous
as comparing cognitive domains where the demand associ- work, which has identified an association between the extent
ated with individual tasks can be matched. to which an individual’s glucose levels fall and their
The main focus of the present study was an examination cognitive performance [7,8,14]. The notion that there may
of the effects of cognitive demand on blood glucose levels. be a direct neural component to this effect is supported to
Glucose levels fell more rapidly during Serial Sevens than some extent by findings from brain imaging, especially
during the key-pressing control and fell to a greater extent in positron emission tomography (PET). For example, it was
the glucose than the placebo condition (Fig. 3). There are a shown that increasing the load of a working memory task
number of processes which contribute to these effects. (N-back) resulted in an overall increase in the uptake of
Firstly, from Fig. 3, it is clear that glucose levels are falling labelled glucose as measured by PET [13]. The relative
during the 5 min of the task (demanding or not) in both contribution of peripheral and central mechanisms to glu-
A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592 591

cose uptake associated with cognitive demand needs further [7] Donohoe RT, Benton D. Cognitive functioning is susceptible to the
level of blood glucose. Psychopharmacology 1999;145:378 – 85.
investigation before firm conclusions can be drawn.
[8] Donohoe RT, Benton D. Declining blood glucose levels after a cog-
It should be reiterated that Serial Sevens is associated nitively demanding task predict subsequent memory. Nutr Neurosci
with an increase in heart rate over and above that seen in a 1999;2:413 – 24.
somatically matched control task [14]. This increased [9] Foster JK, Lidder PG, Sunram S. Glucose and memory: fractionation
cardiac muscle activity will inevitably contribute to the of enhancement effects. Psychopharmacology 1998;137:259 – 70.
accelerated decay of glucose observed in the present [10] Hall JL, Gonder-Frederick LA, Chewning WW, Silviera J, Gold PE.
Glucose enhancement of performance on memory tests in young and
experiment. Additionally, there are a number of physiolog- aged humans. Neuropsychologia 1989;27:1129 – 38.
ical responses during arousal which may modulate cogni- [11] Halpern DF. Sex differences in cognitive abilities. New Jersey: Law-
tive performance. As well as heart rate changes, these rence Erlbaum, 1992.
include the release of ACTH, vasopressin, opioid peptides, [12] Hayman M. Two minute clinical test for measurement of intellectual
and adrenal catecholamines. Of these, the action of adrena- impairment in psychiatric disorders. Arch Neurol Psychiatry
1942;47:454 – 64.
line may further complicate interpretation of these data. [13] Jonides J, Schumacher EH, Smith EE, Lauber EK, Awh E, Satoshi M,
Adrenaline release results in a breakdown of hepatic Koeppe RA. Verbal working memory load affects regional brain acti-
glycogen stores and a subsequent increase in blood glucose vation as measured by PET. J Cognit Neurosci 1997;9:462 – 75.
[33]. There is clearly a dynamic balance between release [14] Kennedy DO, Scholey AB. Glucose administration, heart rate and
and uptake of blood glucose. It would appear that the cognitive performance: effects of increasing mental effort. Psycho-
pharmacology 2000;149:63 – 71.
arousal processes associated with intense cognitive process- [15] Keul J, Huber G, Lehman M, Berg A, Jakob EF. Einfluss von Dex-
ing might shift this balance in favour of mechanisms which trose auf Fahrleistung, Konzentrationsfaehigkeit, Kreislauf und Stoff-
deplete blood glucose. Future studies examining the inter- wechsel in Kraftfahrzeug-simulator (Doppelblind-studie im Cross-
action of glucose levels with tasks, which similarly raise over Design). Akt Ernahr Mad 1982;7:7 – 14.
cardiac activity and/or arousal in the absence of a cognitive [16] Manning CA, Hall JL, Gold PE. Glucose effects on memory and
other neuropsychological tests in elderly humans. Psychol Sci
load, may prove informative. 1990;1:307 – 11.
The mechanisms underlying the cognition-enhancing [17] Martin PY, Benton D. The influence of a glucose drink on a demand-
effects of glucose are unknown. Several authors have ing working memory task. Physiol Behav 1999;67:69 – 74.
suggested that the cholinergic system may be involved in [18] Mathews DR, Holman RR, Bown E, Steenson J, Watson A, Hughes S,
Scott D. Pen-sized digital 30-sec blood glucose meter. Lancet
many of the observed cognitive effects of glucose. We have
1987;1:778 – 9.
previously speculated that cholinergic modulation is [19] Messier C, Gagnon M. Glucose regulation and cognitive functions:
unlikely to account for all of the observed cognitive effects relation to Alzheimer’s disease and diabetes. Behav Brain Res
of glucose [14]. In conclusion, glucose administration 1996;75:1 – 11.
preferentially facilitated performance of the task with the [20] Metzer M. Glucose enhancement of a facial recognition task in young
highest cognitive demand. This demand was itself associ- adults. Physiol Behav 2000;68:549 – 53.
[21] Moss MC, Scholey AB. Oxygen administration enhances memory
ated with an accelerated decay in blood glucose levels, formation in healthy young adults. Psychopharmacology 1996;124:
particularly in the glucose condition. Whatever the mech- 255 – 60.
anisms underlying such effects, these results do lend [22] Moss MC, Scholey AB, Wesnes K. Oxygen administration selectively
support to the increasing body of evidence that, under enhances cognitive performance in healthy young adults: a placebo-
conditions of cognitive demand, there may be a reciprocal controlled double blind crossover study. Psychopharmacology
1998;138:27 – 33.
relationship between task performance and falling blood [23] Owens DS, Benton D. The impact of raising blood glucose on reaction
glucose levels. times. Neuropsychobiology 1994;30:106 – 13.
[24] Parker PY, Benton D. Blood glucose levels selectively influences
memory for words dichotically presented to the right ear. Neuropsy-
chologia 1995;33:843 – 54.
[25] Raggozzino ME, Hellems K, Lennartz RC, Gold PE. Pyruvate infu-
References sions into the septal area attenuate spontaneous alternation impair-
ments induced by intraseptal morphine injections. Behav Neurosci
[1] Armitage P, Hills M. The two-period crossover trial. Statistician 1995;109:1074 – 80.
1982;31:119 – 31. [26] Ragozzino ME, Unick KE, Gold PE. Hippocampal acetylcholine re-
[2] Azari NP. Effects of glucose on memory processes in young adults. lease during memory testing in rats: augmentation by glucose. Proc
Psychopharmacology 1991;105:521 – 4. Natl Acad Sci USA 1996;93:4693 – 8.
[3] Backs WR, Seljos KA. Metabolic and cardiorespiratory measures of [27] Ragozzino ME, Pal SN, Unick K, Stefani MR, Gold PE. Modulation
mental effort: the effects of level of difficulty in a working memory of hippocampal acetylcholine release and of memory by intrahippo-
task. Int J Psychophysiol 1994;16:57 – 68. campal glucose injections. J Neurosci 1998;18:1595 – 601.
[4] Benton D, Owens DS. Blood glucose and human memory. Psycho- [28] Scholey AB, Fowles K. Retrograde enhancement of kinaesthetic
pharmacology 1993;113:83 – 8. memory by low dose alcohol and by glucose: evidence for independ-
[5] Benton D, Owens DS, Parker PY. Blood glucose memory and atten- ent mechanisms unrelated to mood change. J Psychopharmacol
tion. Neuropsychologia 1994;32:595 – 607. 2000;14:A46.
[6] Carroll D, Turner JR, Prasad R. The effects of level of difficulty of [29] Scholey AB, Moss MC, Neave N, Wesnes K. Cognitive performance,
mental arithmetic challenge on heart rate and oxygen consumption. Int hyperoxia and heart rate following oxygen administration in healthy
J Psychophysiol 1986;4:167 – 73. young adults. Physiol Behav 1999;140:783 – 9.
592 A.B. Scholey et al. / Physiology & Behavior 73 (2001) 585–592

[30] Scholey AB, Moss MC, Wesnes KA. Oxygen and cognitive perform- [32] Turner JR, Carroll D. Heart rate and oxygen consumption during
ance: the temporal relationship between hyperoxia and enhanced mental arithmetic, a video game, and graded exercise: further evi-
memory. Psychopharmacology 1998;140:123 – 6. dence for metabolically exaggerated cardiac adjustment. Psychophysi-
[31] Taylor LA, Rachman SJ. The effects of blood sugar level changes on ology 1985;20:544 – 9.
cognitive function, affective state and somatic symptoms. J Behav [33] Wenk GL. An hypothesis of the role of glucose in the mechanism of
Med 1987;11:279 – 91. cognitive enhancers. Psychopharmacology 1989;99:431 – 8.

You might also like