J. biosoc. Sci.

(1996) 28, 463^79

THE SUPPLY OF GLUCOSE TO THE BRAIN AND

COGNITIVE FUNCTIONING
DAVID BENTON, PEARL Y. PARKER AND RACHAEL T. DONOHOE
Department of Psychology, University of Wales Swansea, Swansea SA2 8PP
Introduction
Glucose: the fuel of the brain

Unlike other organs the energy requirement of the brain is met almost exclusively by
aerobic glucose degradation (Siesjo, 1978). The energy requirement of the brain is 20-30%
of the whole organism at rest, although its weight is only 2%. The energy stores in the
brain are extremely small when compared with the high rate of glucose utilisation: thus
the brain is reliant on a continuous glucose supply. Only about 30% of glucose is required
for direct energy production; much of the remainder is used for the synthesis of amino
acids, peptides, lipids and nucleic acids (Siebert, Gessner & Klasser, 1986). Thus a source
of glucose is essential for the synthesis of physiologically active amines such as serotonin,
noradrenaline and acetylcholine. Although it is well accepted that hypoglycaemia can
result in the disruption of cognitive functioning, this is a rare phenomenon and it has
usually been assumed that levels of blood glucose, within the normal range, do not
influence intellectual functioning. This assumption is discussed in this paper.
Blood glucose is normally maintained in the range 4-5-5 mmol/L. If glucose levels
are too high tissue dehydration results. When blood glucose falls below about
2-2 mmol/L the symptoms of hypoglycaemia occur, including trembling, sweating,
confusion and problems of perception. Nearly all the carbohydrate absorbed from the
gut circulates in the blood as glucose, the main substrate of tissue respiration. When
the levels of blood glucose fall, insulin secretion stops and the release of other
substances including the catecholamines, glucagon, growth hormone and
glucocorticoids help optimal values to be again achieved.
Neural metabolic rate and cognitive functioning

Young adults
The development of positron emission tomography (PET) scans has allowed the
amount of glucose metabolised in discrete regions of the brain to be visualised. There
are two basic strategies. In the 'resting state' procedure the use of glucose by the resting
brain is related to cognitive performance at another time. In the 'activation' paradigm
the brain's use of glucose while performing cognitive tasks is measured.
Weiss (1986) commented that it 'would defy the most fundamental laws of
thermodynamics, if differences in brain power would not find their counterpart in
463

464

D. Benton, P. Y. Parker and R T. Donohoe

individual differences in brain energy metabolism'. There is some evidence to support

such a position, although the relationship depends on the age of the subject and the
nature of the test. There are many reports that mental inactivity, rather than the
performing of a cognitive task, is associated with a lower rate of brain metabolism. For
example, Reivich & Alavi (1983) found that light stimulation increased cerebral glucose
metabolism in the primary visual cortex; similar changes can be measured in other
areas of the brain after various cognitive tasks such as intense counting or reading.
Haier et al. (1992a) found that the learning of a complex visuospatial motor task was
associated with an increased use of glucose by the brain.
There is clear evidence from PET scans that increased neural activity is associated
with increased glucose metabolism, and the question arises as to the means by which
additional glucose is directed to the relevant areas of the brain. Regional glucose
utilisation and regional blood flow vary greatly throughout the brain and are closely
related, yet regional differences in brain glucose concentration are limited (Hawkins et
al, 1983): these data could only arise if the supply of glucose matched its utilisation.
Metabolically more active nuclei contain more capillaries than those that are less active
(Craigie, 1920), and an increase in the availability of glucose results, in part at least,
from the recruitment of capillaries, a mechanism similar to that in muscle.
Studies have repeatedly failed to find a relationship between the metabolic rate of the
resting brain and measures of cognitive functioning (Duara et al, 1984; Haxby et al,
1986). In contrast, the metabolic rate when performing a cognitive task has been found to
be associated with performance on that task. Haier et al. (1988) found a negative
correlation between scores on Raven's Advanced Progressive Matrices and the rate at
which glucose was metabolised in the cortex. Parks et al. (1988) and Boivin et al. (1992)
reported negative correlations between verbalfluencyand resting glucose metabolic rate
in the frontal cortex of normal subjects. Berent et al. (1988) reported a negative correlation
between WAIS memory scores and the use of brain glucose by the resting brain.
These data have been interpreted as evidence for a brain efficiency model of
intelligence; the intelligent brain works efficiently, rather than hard. This efficiency may
derive from only using brain areas necessary for the particular task. It seems that
subjects performing a complex task will use a limited set of brain circuits while poorer
performers use more circuits. Haier et al. (1992b) found that the learning of a complex
task was associated with a decrease in the use of glucose in various areas of the brain.
They interpreted the findings as evidence consistent with the use of fewer extraneous
areas of the brain after learning a task.
More generally PET studies have shown that cognitive demand is associated with
increased glucose metabolism in localised regions of the brain. In PET studies
radioactively labelled glucose appears in the areas of the brain that are metabolically active
within a few minutes of being injected into the blood stream. This observation indicates
that the brain relies on glucose taken from the blood when neural activity is stimulated.
The elderly
With PET studies it is necessary to distinguish findings obtained from younger and
older patients, particularly those with dementia-related disorders. In contrast to
younger subjects, in the elderly there are several reports of a greater use of glucose by

Brain glucose and cognitive functioning

465

the resting brain in those with more preserved cognitive ability (Berent et al, 1988;
Chase et al, 1984; Foster et al, 1983, 1984; Haxby et al, 1985; Karbe et al, 1989).
Kuhl et al (1982) reported a negative correlation between the rate of cerebral
metabolism and age, although regional changes in the rate of brain glucose metabolism
have not been detected (de Leon et al, 1984). There is a report of positive correlations
between the rate of regional cerebral glucose metabolism and intelligence scores (Chase
et al, 1984). Riege et al. (1985) related regional brain glucose metabolism to memory
and found performance on the memory tests were correlated with glucose metabolism
in some, but not all, areas of the brain. Two factors distinguished those below 42 and
above 48 years of age. One reflected memory for material verbally processed and the
metabolic ratio in Broca's area; the other was denned by tests requiring sequential or
organisational coding of information and metabolic measures of the thalamic regions.
Those with high superior frontal and low caudate-thalamic metabolic measures were
those who performed well in tests of memory for sentences, stories, designs and
complex patterns. Berardi et al. (1991) found that in the healthy elderly, those who had
better visual than verbal memory, had a higher right rather than left hemisphere
metabolic rate. The data on this topic are not, however, totally consistent, Duara et al
(1984) found no association between brain metabolic rates, age and cognitive scores.
However, the majority of evidence suggests a positive relationship between preserved
cognitive ability and the ability of the brain to metabolise glucose.
Alzheimer's disease
A number of workers have begun to suggest that an abnormality of glucose metabolism
may be the primary cause of early onset Alzheimer's dementia (Hoyer, Oesterreich &
Wagner, 1988; Hoyer, 1993). Landin et al. (1993) confirmed the finding of Adolfsson
et al (1980) when they found low blood glucose levels in Alzheimer patients and
suggested that it was a hypometabolic disorder. In fact, metabolic abnormalities in
senile dementia of Alzheimer's type (SDAT) are not restricted to neural tissue; they can
be obtained in cell cultures of peripheral tissue, leading to the suggestion that it is a
systemic disease (Blass & Zemcov, 1984). In spite of low blood glucose levels Alzheimer
patients have been found to have a greater insulin response during a glucose tolerance
test (Bucht et al, 1983) possibly a reflection of insulin resistance in peripheral tissues.
The rate of glucose transport across the blood-brain barrier is decreased in Alzheimer's
disease (Jagust et al, 1991) and Kalaria, Mitchell & Harik (1988) found a marked
reduction in a glucose transport molecule in the brains of Alzheimer patients. Given
the importance of a continuous supply of glucose for the brain, a problem associated
with the transport of glucose across the blood-brain barrier could be crucial. In those
suffering from SDAT there is a marked decline in glucose metabolism in those areas
of the brain associated with functional loss, although the decline in glucose utilisation
is general, occurring in other parts of the brain (Foster et al, 1983).
It is known that SDAT is associated with a particularly large degeneration of
neurones that use acetylcholine as a neurotransmitter. Why, if SDAT is a general
metabolic disease, is there a relatively selective effect on cholinergic mechanisms? The
answer is unknown but there is good evidence that anything that impairs glucose
utilisation, including lack of oxygen, hypoglycaemia and thiamin deficiency, influences

466

D. Benton, P. Y. Parker and R T. Donohoe

cholinergic functioning particularly (Gibson, Barclay & Blass, 1982). Hoyer (1993)
suggested that these disorders of glucose metabolism occur first at acetylcholine and
glutamine synapses, resulting in neuronal damage.
It must be considered whether the metabolic changes in SDAT contribute to brain
dysfunction or just reflect it, although it is possible that both mechanisms are
important. Given the essential nature of glucose it is tempting to see problems with its
delivery to the brain and its metabolism as primary defects in SDAT. The alternative
that these changes are simply a reflection of the decreased use of cerebral glucose must,
however, be kept in mind. The finding of metabolic problems in other tissues and the
improvement of cognitive functioning following a glucose drink (see below) suggest
that the metabolic changes, in part at least, play a causal role. Hoyer et al. (1988) asked
whether a problem of glucose metabolism is the primary abnormality that leads to
SDAT. They looked at patients who were just beginning to display symptoms of
dementia and found there was a 44% decrease in the metabolic rate of glucose, but
cerebral blood flow and the metabolic rate of oxygen were unchanged. They suggested
that abnormalities in the breakdown of glucose played a pivotal role in SDAT.
Glucose and memory

Interest in the influence of blood glucose levels on cognitive functioning was stimulated
by attempts to develop drugs that facilitate memory and attention. Wenk (1989)
reviewed the evidence supporting the hypothesis that glucose had a role in the action
of drugs that enhance memory, an effect mediated by the release of adrenaline from
the adrenal medulla. The general hypothesis that cognitive enhancing drugs produce
their effect by increasing the availability and uptake of glucose is suggested by four
related types of evidence. Some cognitive enhancing drugs do not cross the blood-brain
barrier; others are effective when injected peripherally but not when injected directly
into the brain; many cognitive enhancing drugs are not effective following
adrenalectomy; finally, cognitive functioning is correlated with glucose regulation in
aged animals and humans. Given the frequently reported facilitation of memory that
results from adrenaline release (Gold & Stone, 1988), a possible mechanism is that
memory is facilitated by the release of adrenaline from the adrenal glands, that
stimulates the liver to release glucose, that in turn is taken up by the brain. There is
evidence that the injection of glucose into the ventricles facilitates memory (Lee,
Graham & Gold, 1988).
Hall & Gold (1990), in a study of mice with adrenalectomy-induced memory
deficits, noted that an injection of glucose restored memory to that of the
sham-operated animals. Gold (1986) and Stone, Rudd & Gold (1990) found that
moderate doses of glucose were associated with enhanced memory, although further
increases were disruptive: Parson & Gold (1992) reported a similar inverted-U shaped
relationship in elderly humans. Messier & White (1987) found that whereas 2g/kg of
glucose enhanced memory in rats, 1 g/kg or 3 g/kg did not. As all three doses of glucose
resulted in broadly the same increase in blood glucose it appeared that the level of
blood glucose was unimportant. The finding of similar rises in blood glucose following
different doses of glucose suggested the importance of differential rates of absorption.
Gold, Vogt & Hall (1986) noted that glucose was an exception to the generalisation

Brain glucose and cognitive functioning

467

that adrenergic antagonists attenuate the effects of most treatments on memory: it

appears that glucose has its impact beyond the action of a peripheral adrenergic receptor.
To date only a few authors have looked for a similar phenomenon in healthy human
adults. Lapp (1981) found that lists of words were more easily learned by subjects with
high (> 7 mmol/L) rather than low (< 4-4 mmol/L) blood glucose. Benton & Sargent
(1992) found that the eating of breakfast was associated with better memory performance
later in the morning, and that memory performance correlated with blood glucose levels.
Other human studies have focused on the elderly. Gonder-Frederick et al. (1987)
found that a story was more easily remembered following moderate increases in blood
glucose; a similar but smaller effect was also observed in young subjects (Hall et al,
1989). In their study of seventeen healthy elderly subjects Manning, Hall & Gold (1990)
found significant improvements in two sub-tests of the Wechsler Memory Scale;
however, the glucose beverage did not enhance digit span or other measures of
cognitive functioning. The giving of a glucose drink improved the memory of a group
of healthy elderly if administered either shortly before or immediately after hearing a
passage. Memory outlasted the transient increase in blood glucose and was better 24
hours later (Manning, Parson & Gold, 1992). Craft et al. (1993) reported, in those with
mild SDAT, that memory for a story was facilitated when glucose was maintained at
12-5 mmol/L. Craft, Zallen & Baker (1992) found that glucose administration improved
the memory of Alzheimer patients in whom blood glucose levels tended to stay high,
that is those unable to deal adequately with blood glucose. A generalisation supported
by the reports of studies with rats and both elderly and young adult humans is that
blood glucose levels in the range 8-3 to 9-7 mmol/L optimally enhance memory.
Most studies of the impact of glucose on memory have employed very crude
measures and concepts of memory; many subtleties remain to be explored as it is
improbable that all aspects of memory will be similarly influenced by any
manipulation. There are preliminary suggestions that glucose influences some but not
all aspects of memory. In the elderly, Manning et al. (1990) found that a glucose drink
improved the memory for a story and a list of words, but not digit span, memory for
a complex figure or tests of attention, intelligence or motor functioning.
Gonder-Frederick et al. (1987) found that the taking of glucose increased the ability to
recall a story, but other sub-tests of the Wechsler memory scale were not significantly
improved. Lapp (1981) found that the memory of 16-year-olds for lists of nouns was
greater when blood glucose levels were high. Benton & Owens (1993) reported an
association between blood glucose and memory for word lists but not spatial memory.
Craft, Murphy & Wemstrom (1994) have most systematically examined the impact
of glucose on various aspects of memory. Glucose administration enhanced the ability
to recall a paragraph in older males and younger males whose blood glucose levels fell
during the task. They concluded that the effect of glucose was restricted to declarative
memory as other measures of working and procedural memory were uneffected.
Glucose tolerance

There have been many suggestions that the tendency to develop low blood glucose
levels may be associated with characteristic behaviour, in particular that some
individuals react to a meal containing sugar by the release of large amounts of insulin,

468

D. Benton, P. Y. Parker and R. T. Donohoe

such that after several hours blood glucose falls to levels that are described as
hypoglycaemic. Harris (1924) introduced the term spontaneous hypoglycaemia when
he described, in non-diabetic patients, symptoms similar to those that result from
insulin-induced hypoglycaemia. There are two groups of symptoms associated with
hypoglycaemia. Adrenergic symptoms reflect the release of adrenaline and include
sweating, palpitations, weakness and subjective feelings of anxiety; a blood glucose
level of approximately 2-2 mmol/L is about the point at which this syndrome can be
expected, although there are individual differences. If glucose values fall further then
neuroglycopenic symptoms occur, central nervous system impairment manifests itself
as confusion, amnesia, bizarre behaviour and blurred vision.
The term neuroglycopenia refers to the symptoms resulting from an inadequate
supply of carbohydrate to the neurone. If low glucose levels persist then coma and even
death can result. There have been suggestions that a coma induced by low blood
glucose levels may be associated with neurological damage (Deary, 1992).
A way of examining the ability of an individual to control their levels of blood
sugar is offered by the glucose tolerance test that is used to diagnose diabetes, some
forms of liver disorder, and, less commonly, food-stimulated hypoglycaemia. The test
examines the dynamic way in which the body handles sugar. Initially a baseline blood
glucose value is measured in an individual who has fasted overnight and has not eaten
breakfast. A drink containing glucose is administered and the levels in the blood are
monitored, for 3 hours in the case of diabetes, and longer in the study of
hypoglycaemia. After eating, glucose levels in the blood rise until they stimulate the
pancreas to produce insulin, causing the levelling out and eventual fall of glucose
values. The diabetic in this test is characterised by blood sugar values that increase and
then stay high, as there is a failure to release the insulin that stimulates the cells to
remove glucose from the blood.
Depending on the size and duration of the insulin response, some individuals end
up with blood glucose values in the glucose tolerance test less than the fasting levels,
so called rebound or reactive hypoglycaemia. Although there can be marked individual
differences within the normal range, typically plasma glucose will rise by about 50%
during the first hour after the drink, it will approach the fasting value around the
second hour and may then, in some cases, fall below fasting values.
How many people suffer from rebound hypoglycaemia?
There is a marked contrast between the frequency with which popular writers and
many clinicians suggest that what is eaten can induce hypoglycaemia. Fredericks &
Goodman (1969) begin their book with the claim that 'for one person in ten sugar is
a deadly food, paving the way towards . . . all the tortures of neurotic and even
psychotic symptoms', a view conflicting with Marks & Rose (1981) who, when writing
a clinical text book on the topic, noted that 'most investigators with extensive
experience of all types of spontaneous hypoglycaemia due to organic causes consider
reactive hypoglycaemia an uncommon or rare variant'.
One of the most extensive studies of glucose metabolism using the glucose tolerance
test was carried out by Lev-Ran & Anderson (1981) who examined 650 normal
subjects. The median nadir was 3-6 mmol/L, the 10th percentile was 2-6 mmol/L, the

Brain glucose and cognitive functioning

469

5th was 2-4 mmol/L and the 2-5th was 215 mmol/L: even in the glucose tolerance test,
very low glucose values are not common. In contrast to the extensive data on blood
glucose changes under experimental conditions there is relatively little evidence dealing
with changes in blood glucose concentration during everyday conditions. In a study of
five normally fed males, blood glucose never fell below 3-9 mmol/L (Hansen &
Johansen, 1970) and in a similar study of eight individuals the levels did not fall below
4-4 mmol/L (Genuth, 1973). Alberti, Dornhorst & Rowe (1975), after monitoring
nineteen normally fed individuals, reported 'very little variation in glucose
concentrations during the day'; they commented that the changes are quite different to
those obtained during a glucose tolerance test. A clear picture emerges: when normally
fed healthy subjects are studied their glucose levels tend to be stable and, although
glucose values rise after a meal they typically return to levels well above those required
to diagnose hypoglycaemia. In normal individuals, fed in a usual manner, a
hypoglycaemic response is uncommon. In fact a diagnosis of reactive hypoglycaemia
is best achieved by the use of a meal tolerance test in which a load of sugar is combined
with fat and protein as the combination of macro-nutrients greatly influences the
nature of the change in blood glucose.
There is, however, reason to suggest that there are individual differences in the
response to low blood glucose levels. Widom & Simonson (1990) found the threshold
for the impairment on a series of cognitive tasks was in the range 2-2 to 2-8 mmol/L
blood glucose. There were large individual differences, with the performance of some
individuals being disrupted at 4 mmol/L, whereas it was normal in others at levels even
below 2-2 mmol/L. Although the origin of these individual differences is unknown
Snorgaard et al. (1991) found that non-diabetics, in whom food relieved the symptoms
of hypoglycaemia, tended to have cognition disrupted at higher glucose levels.

Is it low levels or the ability to use glucose that is important?

To date much of the work on glucose tolerance has assumed that blood glucose
only influences behaviour when levels are low enough to induce hypoglycaemic
reactions. There is increasing evidence that a second possible mechanism should be
considered. Even when blood glucose levels are within the normal range there are
individual differences in the ability to control blood glucose levels, that are associated
with differences in psychological functioning.
Several types of study suggest a relationship between the control of blood glucose
levels and aggression (Benton, 1988). Bolton (1973, 1976), an anthropologist, gave a
glucose tolerance test to Quolla Indians in Peru, a people known for their violence. A
history of violence, often associated with having committed murder, was associated
with a tendency to develop lower levels of blood glucose in the test. In a long series of
studies, Virkkunen has reported a tendency to develop low blood glucose in a glucose
tolerance test in violent offenders (Virkkunen, 1982, 1983, 1984, 1986; Virkunnen &
Huttunen, 1982). Benton, Kumari & Brain (1982) gave a glucose tolerance test to
apparently normal members of British society and found a relationship between the
tendency for glucose levels to fall and self-reports of irritability. Benton, Brett & Brain
(1987) found that, after a glucose drink, children displayed fewer signs of frustration
and irritability when faced with an impossible although plausible computer game.

470

D. Benton, P. Y. Parker and R. T. Donohoe

In the majority of these cases discussed here, the subjects do not satisfy the clinical
criteria that would allow a diagnosis of reactive hypoglycaemia. However, to dismiss
these data for this reason may be premature; such a consistency of findings, in such
disparate settings, deserves to be taken seriously. The possibility that it is not
hypoglycaemia that is important, but rather individual differences in the ability to
tolerate blood glucose within the normal range, is suggested by recent research on
memory and other aspects of cognition.
Glucose tolerance and cognition

Although a decline in the use of blood glucose by the elderly brain can be seen in
part as a reflection of a decline in the number of neurones, there is evidence of an
association between a more general inability to control blood glucose levels and poor
memory. In elderly subjects, glucose tolerance has also been shown to be predictive of
memory performance (Hall et al, 1989; Manning et al, 1990; Craft et al, 1993). In
elderly humans the magnitude of the change in blood glucose levels following ingestion
of a glucose containing drink was significantly correlated with performance on the
Wechsler memory scale (Hall et al, 1989); the elderly whose blood glucose levels
remained high, that is they poorly regulated blood glucose, had poorer scores on
memory tests. Craft et al. (1992) similarly found in the normal elderly that memory was
better following a glucose drink if blood glucose levels fell after the initial rise, rather
than staying raised: that is, an inability to move the glucose from the blood to the cells
was indicative of poor memory. Poor blood glucose regulation is also associated with
cognitive dysfunction in diabetics (Holmes et al, 1983; Holmes, Koepke & Thompson,
1986; Ryan & Longstreet, 1984; Ryan et al, 1984). Stone et al. (1990) reported a similar
phenomenon in aged rats, the extent of the increase in blood glucose after a glucose
injection being inversely correlated with memory. It appeared that a decreased ability
to take glucose from the blood into cells is associated with a poor memory.
Control of blood glucose levels and cognitive functioning in young adults

An obvious way of examining the possible impact of increasing blood glucose levels is
to compare the effect of a glucose containing drink with a placebo. Figure 1 illustrates
the results of one such study. Verbalfluencyis measured by asking a subject to mention
in a minute as many words as they can that begin with certain letters. Verbal fluency
is used by neuropsychologists as a test of frontal-lobe functioning: high scores tend to
result from a strategic approach where the subjects keep asking themselves a series of
questions such as the names of animals, flowers or vegetables that begin with that
letter. Figure 1 shows that the performance of a group of young healthy adults did not
differ initially, although the taking of a glucose drink, as opposed to a placebo,
increased performance. There are similar reports that a glucose drink improved the
ability to sustain concentration (Benton et al, 1987; Keul et al, 1982; Moser, Plum &
Buckmann, 1983). However, studies of cognitive functioning following a glucose drink
do not universally report an improvement in performance; rather individual differences
in glucose tolerance often predominate.
Studies of the relationship between changing blood glucose levels and cognitive

Brain glucose and cognitive functioning

471

45

Placebo
40

| Glucose Drink

|
W

CD30

rr
LU
25
20

BEFORE DRINK

AFTER DRINK

Fig. 1. The impact of a glucose drink on verbal fluency. Verbal fluency was measured
before and after taking a glucose drink or a placebo. The verbal fluency of a group of
80 females, aged 20 years, was significantly higher after a glucose drink (/><0-001)
(Benton & Donohoe, 1995).

performance in young healthy adults can be accounted for by the suggestion that the
level and provision of brain glucose influences cognitive functioning. As it is not
technically possible to monitor local changes in glucose concentrations in the human
brain, changes in blood glucose levels, and known aspects of physiology are relied on
to suggest likely reasons for alterations in the level of brain glucose. The results of a
series of neuropsychological studies can be explained by the assumption that two
physiological mechanisms are important.
First, Lund-Anderson (1979) marshalled the evidence that an equilibrium develops
between the level of plasma and blood glucose. In a resting animal an increase in blood
glucose results in an increase in brain glucose that reaches an equilibrium in about 15
minutes. If such a mechanism were important in humans then, in those sitting quietly
without demands being placed upon them, it would be expected that higher blood
glucose levels would be associated with higher levels of glucose in the brain. If the
provision of glucose to the brain limits cognition then a higher level of blood glucose
would be predicted to be associated with better performance. Benton & Owens (1993)
found that memory for a word list correlated with blood glucose levels when entering
the laboratory, an association that existed throughout the range of blood glucose levels
and not only at levels that could be expected to induce the symptoms of
hypoglycaemia. Figure 2 similarly shows that in a sample of 186 young healthy adults
the ability to recall a story from the Wechsler memory scale was associated with blood
glucose levels on entering the laboratory. Benton, Owens & Parker (1994) reported, in
young adults, a positive correlation between blood glucose levels and forgetting; those
with a higher initial blood glucose remember more.

D. Benton, P. Y. Parker and R. T. Donohoe

472

14

12

R
CO

u_
O10
<

O
LU
DC

<3.6 3.7-4.7 4.8-6.3 >6.3

BLOOD GLUCOSE (mmol/L)
Fig. 2. Blood glucose levels and memory of a Wechsler story. The blood glucose levels
of 186 females, aged 20 years, who had been sitting quietly were related to their ability
to recall a story. The memory of those with high blood glucose levels was greater, and
those with low levels poorer, than those with intermediate values (/><0-03).

The ability to tolerate blood glucose is the second physiological mechanism that
accounts for the findings. After a meal, or a glucose containing drink, blood glucose
levels increase markedly for about half an hour after which, in most people, they fall,
reaching baseline level about 2 hours afterwards. If the levels of blood glucose increase
to particularly high levels, and stay in this range for several hours, then this is indicative
of diabetes. In non-diabetics the speed at which blood glucose levels fall differs. Some
tolerate glucose well and the levels fall rapidly. Some tolerate glucose less well and
blood glucose levels fall less quickly. A repeated finding is that those whose blood
glucose levels fall rapidly (i.e. they show better glucose tolerance) perform better on
various cognitive tests.
Figure 3 illustrates one example. Following a glucose drink, changes in blood
glucose were monitored while taking a test of abstract reasoning, the Graduate and
Managerial Assessment (Blinkhorn, 1985), an intelligence test, performance on which
correlates very highly with performance on Ravens matrices. The scores of those with

Brain glucose and cognitive functioning

10

473

BLOOD GLUCOSE
Falling
^ ^ > 3.6 mmol/L
Rising

55
rr

I
CO

Fig. 3. Changes in blood glucose while completing an intelligence test. Those whose
blood glucose levels rose or fell markedly while performing an intelligence test were
distinguished. Falling blood glucose levels were associated with better performance

poor glucose tolerance, that is their blood glucose levels were rising, were poorer than
those whose blood glucose levels were falling markedly, that is they displayed good
glucose tolerance. Those whose blood glucose levels did not fall while performing the
test attempted as many questions but were less likely to get them right.
Similar findings have been reported in young adults with the Rapid Information
Processing Task (Benton et al, 1994): high blood glucose while performing the task was
associated with poorer performance. Individuals unable to move glucose rapidly from
the blood stream, when faced with a prolonged cognitive task, perform more poorly.
Benton & Donohoe (1996) found that those whose levels of blood glucose remained
higher performed more poorly on the block design test of the WAIS and were more
susceptible to set in the water jars test. Both these tests measure frontal lobe functioning.
Similarly, in both the healthy elderly and those with Alzheimer's disease, poor
glucose tolerance predicts poor memory (Hall et al, 1989; Manning et al, 1990; Craft
et al, 1992). In fact, the association between the ability to control blood glucose and
memory is stronger in the elderly but it is not simply a sign of pathology as it is found
in young adults.

474

D. Benton, P. Y. Parker and R. T. Donohoe

Placebo ^ ^

Glucose drink |

No breakfast

Breakfast

Fig. 4. Breakfast blood glucose and memory. In 137 females and 47 males eating
breakfast was associated with a greater recall of a list of words. In those who had fasted
the taking of glucose drink resulted in performance similar to that in those who had
eaten breakfast (/><0-015) (Benton & Parker, 1996).

If the provision of blood glucose limits cognition, an obvious prediction is that

fasting will be cognitively disruptive. Benton & Parker (1996) compared those who had
and had not eaten breakfast. Failure to eat breakfast was associated with poorer recall
of a word list, a story from the Wechsler memory scale and recall of trigram
consonants. Figure 4 shows the impact of giving a glucose containing drink or a
placebo to those who had and had not eaten breakfast. Those who had not eaten
breakfast recalled significantly fewer words, something that was reversed by the taking
of a glucose drink. Similarly the adverse effect of not eating breakfast on the recall of
trigrams was reversed by taking a glucose drink. Thus, on two tasks, the adverse effects
of not eating breakfast were reversed by increasing the level of blood glucose. The
conclusion that a low level of glucose can explain all the negative effects of missing
breakfast is too simple; a poor recall of the Wechsler story, that resulted from a failure
to eat breakfast, was not reversed by a glucose drink. It seems that not eating breakfast

Brain glucose and cognitive functioning

475

influences cognition by more than one mechanism although the provision of glucose
appears to be one of these mechanisms.
Discussion
It has been commonly assumed that blood glucose levels in the range that most of us
experience most of the time do not influence cognitive functioning. At the same time
it is widely accepted that very low blood glucose levels are associated with a disruption
of cognitive functioning. Hypoglycaemic levels (< 2 mmol/L) cause obvious physical
and psychological symptoms. Levels between 2 and 3 mmol/L produce less obvious
problems: for example, reaction times and the ability to sustain attention may be
compromised. In contrast, Fig. 1 reports psychological consequences of blood glucose
levels that are much greater than these levels. In fact, the data in all the text-figures
are from subjects whose blood glucose levels are typically much greater than the levels
that have been traditionally assumed to be psychologically disruptive.
Technically there are no means of monitoring the changing levels of glucose in local
areas of the human brain. Without such data one can only speculate about the likely
mechanisms. The assumption that cognitively demanding situations can locally deplete
the brain of glucose, its basic fuel, does however account for thefindings.If a lack of
brain glucose can become a problem then it would be predicted that a high baseline
level of blood glucose would be beneficial. If there are no demands of a physcial or
psychological nature, the higher the level of blood glucose then the higher the levels of
glucose that are found in the brain (Lund-Anderson, 1979). It would be predicted that
those with a higher level of blood glucose would perform cognitive tasks more
efficiently. Figure 1 and other reports (Benton et al, 1994) confirm that prediction.
If cognitive demand was associated with the depletion of blood glucose then those who
were better able to transport glucose from the blood stream to the brain would be better
able to deal with that demand. The performance of a cognitive task, as opposed to sitting
quietly, has been found to be associated with a more rapid fall in blood glucose levels. Based
on the evidence from PET scans it is reasonable to suggest that additional blood glucose is
leaving the blood stream when a cognitive task is being performed and is entering the brain.
It is an obvious hypothesis that when those whose blood glucose levels do not fall are
contrasted with those whose levels fall markedly one is comparing individuals who differ in
the ability to rapidly transfer glucose to the brain. If so it could be predicted that a greater
fall in blood glucose would be associated with better cognitive performance. Figure 3 and
other data (Benton et al., 1994; Benton & Donohoe, 1995) confirm this prediction.
There is increasing evidence of an association between the ability to control blood
glucose and cognitive functioning. The phenomenon is more easily demonstrated in the
elderly, but can be also demonstrated in young healthy adults.
References
ADOLFSSON, R., BUCHT, G., LITHNER, F. & WINBLAD, B. (1980) Hypoglycemia in Alzheimer's

disease. Metabolism, 38, 572 576.

ALBERTI, K. G. M., DORNHORST, A. & ROWE, A. S. (1975) Metabolic rhythms in normal and

diabetic man. Israel J. med. Sci. 11, 571-580.

476

D. Benton, P. Y. Parker and R. T. Donohoe

BENTON, D. (1988) Hypoglycaemia and aggression: a review. Int. J. Neurosci. 41, 163-168.
BENTON, D., BRETT, V. & BRAIN, P. F. (1987) Glucose improves attention and reaction to
frustration in children. Biol. Psychol. 24, 95-100.
BENTON, D. & DONOHOE, R. (1996) Cognitive functioning is susceptible to the level of blood
glucose, (submitted for publication).
BENTON, D., KUMARI, N. & BRAIN, P. F. (1982) Mild hypoglycaemia and questionnaire measures
of aggression. Biol. Psychol. 14, 129-135.
BENTON, D. & OWENS, D. (1993) Blood glucose and human memory. Psychopharmacol. 113,
83-88.
BENTON, D., OWENS, D. & PARKER, P. (1994) Blood glucose memory and attention.
Neuropsychologia, 32, 595-607.
BENTON, D. & PARKER, P. Y. (1996) Breakfast blood glucose and cognition. Am. J. din. Nutr.
(in press).
BENTON, D. & SARGENT, J. (1992) Breakfast blood glucose and memory. Biol. Psychol. 33,
207-210.
BERARDI, A, HAXBY, J. V., GRADY, C. L. & RAPOPORT, S. I. (1991) Asymmetries of brain glucose

metabolism and memory in the healthy elderly. Devi Neuropsychol. 7, 87-97.

BERENT, S., GIORDANI, B., LEHTINEN, S., MARKEL, D., PENNEY, J. B., BUCHTEL, H. A.,

STAROSTA-RUBENSTEIN, S., HICHWA, R. & YOUNG, A. B. (1988) Positron emission tomographic

scan investigation of Huntington's disease: Cerebral correlates of cognitive function. Ann.

Neurol. 23, 541-546.
BLASS, J. P. & ZEMCOV, A. (1984) Alzheimer's disease: a metabolic systems degeneration?
Neurochem. Pathol. 2, 103-114.
BLINKHORN, S. F. (1985) Graduate and Managerial Assessment. NFER-Nelson, Winsor.
BOIVIN, M. J., GIORDANI, B., BERENT, S., AMATO, D. A., LEHTINEN, S., KOEPPE, R. A., BUCHTEL,

H. A., FOSTER, N. L. & KUHL, D. E. (1992) Verbal fluency and positron emission tomographic
mapping of regional cerebral glucose metabolism. Cortex, 28, 231-239.
BOLTON, R. (1973) Aggression and hypoglycaemia among the Quolla: a study in
psycho-biological anthropology. Ethnology, 12, 221-251.
BOLTON, R. (1976) Hostility in fantasy: a further test of the hypoglycaemia-aggression
hypothesis. Aggres. Behav. 2, 257-274.
BUCHT, G., ADOLFSSON, R., LITHNER, F. & WINBLAD, B. (1983) Changes in blood glucose and

insulin secretion in patients with senile dementia of Alzheimer type. Ada med. scand. 213,
387-392.
CHASE, T. N., FEDIO, P., FOSTER, N. L., BROOKS, R., D I CHIRO, G. & MANSI, L. (1984) Wechsler

adult intelligence scale performance: cortical localization by flurodeoxyglucose F 18-positron

emission tomography. Archs neurol. 41, 1244-1247.
CRAFT, S., DAGOGO-JACK, S. E., WIETHOP, B. V., MURPHY, C , NEVINS, R. T., FLEISCHMAN, S.,

RICE, V., NEWCOMER, J. W. & CRYER, P. E. (1993) The effects of hyperglycemia on memory

and hormone levels in dementia of the Alzheimer type: a longitudinal study. Behav. Neurosci.
107, 926-240.
CRAFT, S., MURPHY, C. & WEMSTROM, J. (1994) Glucose effects on complex memory and
non-memory tasks: the influence of age, sex and glucoregulatory response. Psychobiology, 22,
95-105.
CRAFT, S., ZALLEN, G. & BAKER, L. D. (1992) Glucose and memory in mild senile dementia of
the Alzheimer type. /. din. exp. Neuropsychol. 14, 253-267.
CRAIGIE, E. H. (1920) On the relative vascularity of various parts of the central nervous system
of the albino rat. / . comp. Neurol. 31, 429-464.
DEARY, I. J. (1992) Diabetes hypooglycaemia and cognitive performance. In: Handbook of
Human Performance. Edited by A. P. Smith & D. M. Jones. Academic Press, London.

Brain glucose and cognitive functioning

All

DE LEON, M. J., GEORGE, A. E., FERRIS, S. H., CHRISTMAN, D. R., FOWLER, J. S., GENTES, C. I.,

BRODIE, J., REISBERG, B. & WOLF, A. P. (1984) Positron emission tomography and computed
tomography assessments of the aging human brain. J. Computer Assisted Tomography, 8, 88-94.
DUARA, R., GRADY, C , HAXBY, J., INGVAR, D., SOKOLOFF, L., MARGOLIN, A., MANNING, R. G.,

CUTLER, N. R. & RAPOPORT, S. I. (1984) Human brain glucose utilization and cognitive
function in relation to age. Ann. Neurol. 16, 702-713.
FOSTER, N. L., CHASE, T. N., FEDIO, P., PATRONAS, N. J., BROOKS, R. A. & Di CHIRO, G. (1983)

Alzheimer's disease: focal cortical changes shown by positron emission tomography.

Neurology, 33, 961-965.
FOSTER, N. L., CHASE, T. N., MANSI, L., BROOKS, R., PATRONAS, N. J. & Di CHIRO, G. (1984)

Cortical abnormalities in Alzheimer's disease. Ann. Neurol. 16, 649-654.

FREDERICKS, C. & GOODMAN, H. (1969) Low Blood Sugar and You. Constellation International,
New York.
GENUTH, S. M. (1973) Plasma insulin and glucose profiles in normal, obese and diabetic persons.
Ann. intern. Med. 79, 812-822.
GIBSON, G. E., BARCLAY, L. & BLASS, J. (1982) The role of the cholinergic system in thiamin
deficiency. Ann. N. Y. Acad. Sci. 378, 382-403.
GOLD, P. E. (1986) Glucose modulation of memory storage processing. Behav. neur. Biol. 45,
342-349.
GOLD, P. E. & STONE, W. S. (1988) Neuroendocrine effects on memory in aged rodents and
humans. Neurobiol. Aging, 9, 709-717.
GOLD, P. E., VOGT, J. & HALL, J. L. (1986) Glucose effects on memory: behavioral and
pharmacological characteristics. Behav. neur. Biol. 46, 145-155.
GONDER-FREDERICK, L., HALL, J. L., VOGT, J., Cox, D. J., GREEN, J. & GOLD, P. E. (1987)

Memory enhancement in elderly humans: effects of glucose ingestion. Physiol. Behav. 41

503-504.
GRADY, C. L. (1984) Neuropsychology and cerebral metabolism in normal aging. Ann. intern.
Med. 101, 358-360.
HAIER, R. J., SIEGEL, B. V., MACLACHLAN, A., SODERLING, E., LOTTENBERG, S. & BUCHSBAUM,

M. S. (1992a) Regional glucose metabolic changes after learning a complex visuospatial motor
taska positron emission tomographic study. Brain Res. 570, 134-143.
HAIER, R. J., SIEGEL, B. V., NUECHTERLEIN, K. H., HAZLETT, E., W U , J. C , PAEK, J., BROWNING,

H. L. & BUCHSBAUM, M. S. (1988) Cortical glucose metabolic rate correlated of abstract

reasoning and attention studied using positron emission tomography. Intelligence, 12,199-217.
HAIER, R. J., SIEGEL, B., TANG, C , ABEL, L. & BUCHSBAUM, M. S. (1992b) Intelligence and

changes in regional cerebral metabolic rate following learning. Intelligence, 16, 415-426.
HALL, J. L. & GOLD, P. E. (1990) Adrenalectomy-induced memory deficits: role of plasma
glucose levels. Physiol. Behav. 47, 27-33.
HALL, J. L., GONDER-FREDERICK, L. A., CHEWNING, W. W., SILVEIRA, J. & GOLD, P. E. (1989)

Glucose enhancement of performance on memory tests in young and aged humans.

Neuropsychologia, 27, 1129-1138.
HANSEN, A. P. & JOHANSEN, K. (1970) Diurnal patterns of blood glucose, serum free acids, insulin,
glucagon and growth hormone in normals and juvenile diabetics. Diabetologia, 6, 27-33.
HARRIS, S. (1924) Hyperinsulinism and dysinsulinism. J. Am. med. Ass. 83, 729.
HAWKINS, R. A., MANS, A. M., DAVIS, D. W., HIBBARD, L. S. & Lu, D. M. (1983) Glucose

availability to individual cerebral structures is correlated to glucose metabolism. J. Neurochem.

40, 1013-1018.
HAXBY, J. V., DUARA, R., GRADY, C. L., CULTER, N. R. & RAPOPORT, S. I. (1985) Relations

between neuropsychological and cerebral metabolic asymmetries in early Alzheimer's disease.

J. Cerebral Blood Flow Metab. 5, 193-200.

478

D. Benton, P. Y. Parker and R T. Donohoe

HAXBY, J. V., GRADY, C. L., DUARA, R., ROBERTSON-TCHABO, E., KOZIARZ, B., CUTLER, N. R.

& RAPAPORT, S. I. (1986) Relations among age visual memory and resting cerebral metabolism
in 40 healthy men. Brain & Cognition, 5, 412^27.
HOLMES, C. S., HAYFORD, J. T., GONZALEZ, J. L. & WEYDERT, J. A. (1983) A survey of cognitive

functioning at different glucose levels in diabetic persons. Diabetes Care, 6, 180-185.

HOLMES, C. S., KOEPKE, K. M. & THOMPSON, R. G. (1986) Simple versus complex impairment
at three blood glucose levels. Psychoneuroendocrinol. 11, 353-357.
HOYER, S. (1993) Intermediary metabolism disturbance in AD/SDAT and its relation to
molecular events. Progr. Neuropsychopharmacol. 17, 199-228.
HOYER, S., OESTERREICH, K. & WAGNER, O. (1988) Glucose metabolism as the site of the primary
abnormality in early-onset dementia of Alzheimer type. / Neurol. 235, 143-148.
JAGUST, W. J., CEAB, J. P., HUESMAN, R. H., VALK, P. E., MATHIS, C. A., REED, B. R., COXSON,

P. & BUDINGER, T. F. (1991) Diminished glucose transport in Alzheimer's disease: dynamic

PET studies. J. Cerebral Blood Flow Metab. 11, 323-330.
KALARIA, R. N., MITCHELL, M. J. & HARIK, S. I. (1988) Chemical pathology of cerebral vessels
in Alzheimer's disease. Soc. Neurosci. Abstr. 14, 154.
KARBE, H., HERHOLZ, K., SZELIES, B., PAWLICK, G., WIENHARD, K. & HEISS, W. D. (1989)

Regional metabolic correlated of Token test results in cortical and subcortical left hemisphere
infarction. Neurology, 39, 1083-1088.
KEUL, J., HUBER, G., LEHMANN, M., BERG, A. & JAKOB, E. F. (1982) EinfluB von dextrose auf

fahrleistung, konzentrationsfahigkeit, kreislauf und stoffwechsel im kraftfahrzeug-simulator

(Doppelblindstudie im cross-over-design). Aktuelle Ernaehrungsmedizin, 7, 7-14.
KUHL, D. E., METTER, E. J., RIEGE, W. H. & PHELPS, M. E. (1982) Effects of human aging on

patterns of local cerebral glucose utilization determined by the [18F]fluorodeoxyglucose

method. J. Cerebral Blood Flow Metab. 2, 163-171.
LAPP, J. E. (1981) Effects of glycemic alterations and noun imagery on the learning of paired
associates. J. Learning Disabil. 14, 35-38.
LANDIN, K., BLENNOW, K., WALLIN, A. & GOTTFRIES, C. G. (1993) Low blood pressure and

blood glucose levels in Alzheimers disease: evidence for a hypometabolic disorder? /. intern.
Med. 233, 357-363.
LEE, M. K., GRAHAM, S. N. & GOLD, P. E. (1988) Memory enhancement with postraining
intraventricular glucose injections in rats. Behav. Neurosci. 102, 591-595.
LEV-RAN, A. & ANDERSON, R. W. (1981) The diagnosis of postprandial hypoglycemia. Diabetes,
30, 996-999.
LUND-ANDERSON, H. (1979) Transport of glucose from blood to brain. Physiol. Rev. 59,305-352.
MANNING, C. A., HALL, J. L. & GOLD, P. E. (1990) Glucose effects on memory and other
neuropsychological tests in elderly humans. Psychol. Sci. 1, 307-311.
MANNING, C. A., PARSON, M. W. & GOLD, P. E. (1992) Antergrade and retrograde enhancement

of 24-h memory by glucose in elderly humans. Behav. new. Biol. 58, 125-130.
MARKS, V. & ROSE, F. G. (1981) Hypoglycaemia, 2nd edn. Blackwell Scientific Publications,
Oxford.
MESSIER, C. & WHITE, N. M. (1987) Memory improvement by glucose, fructose, and two glucose
analogs: a possible effect on peripheral glucose transport. Behav. neur. Biol. 48, 104-127.
MOSER, L., PLUM, H. & BUCKMANN, M. (1983) Der einfluB von dextrose auf diet psychophysische
leistungsfahigkeir des autofahrers. Aktuelle Ernaehrungsmedizin, 8, 247-249.
PARKS, R. W., LOEWENSTEIN, D. A., DODRILL, K. L., BARKER, W. W., YosHn, F., CHANG, J. Y.,

EMRAN, A., APICELLA, A., SHERAMATA, W. & DUARA, R. (1988) Cerebral metabolic effects of

a verbal fluency test: a PET scan study. J. din. exp. Neuropsychol. 10, 565^575.
PARSON, M. & GOLD, P. E. (1992) Glucose enhancement of memory in elderly humans: an
inverted-U dose-response curve. Neurobiol. Aging, 13, 401-404.

Brain glucose and cognitive functioning

479

REIVICH, M. & ALAVI, A. (1983) Positron emission tomographic studies of local cerebral glucose
metabolism in humans in physiological and pathological conditions. Adv. metabol. Disorders,
10, 135-176.
RIEGE, W. H., METTER, E. J., KUHL, D. E. & PHELPS, M. E. (1985) Brain glucose metabolism

and memory functions: age decrease in factor scores. J. Gerontol. 4, 459-467.

RYAN, C. & LONGSTREET, C. (1984) Lower verbal IQ scores in adolescents with IDDM. Diabetes,
33, 525-532.
RYAN, C , VEGA, A., DRASH, A. & LONGSTREET, C. (1984) Neuropsychological changes in

adolescents with insulin-dependent diabetes. J. consult, din. Psychol. 52, 335-342.

SIEBERT, G., GESSNER, B. & KLASSER, M. (1986) Energy supply of the central nervous system.
Bibliotheca Nutritio Dieta, 38, 1-26.
SIESJO, B. K. (1978) Brain Energy Metabolism. Wiley, Chichester.
SNORGAARD, O., LASSEN, L. H., ROSENFALCK, A. M. & BINDER, C. (1991) Glycaemic thresholds

for hypoglycaemic symptoms impairment of cognitive function and release of counter

regulatory hormones in subjects with functional hypoglycaemia. J. intern. Med. 229, 343-350.
STONE, W. S., RUDD, R. J. & GOLD, P. E. (1990) Glucose and physostigmine effects on

morphine-induced and amphetamine-induced locomotor activity in mice. Behav. neur. Biol. 54,
146-155.
STONE, W. S., WENK, G. L., OLTON, D. S. & GOLD, P. E. (1990) Poor blood glucose regulation

predicts sleep and memory deficits in normal aged rats. J. Gerontol. 45, B169-173.
VIRKKUNEN, M. (1982) Reactive hypoglycemic tendency among habitually violent offenders.
Neuropsychobiology, 8, 35^10.
VIRKKUNEN, M. (1983) Insulin secretion during the glucose tolerance test in antisocial
personality. Br. J. Psychiat. 142, 598-604.
VIRKKUNEN, M. (1984) Reactive hypoglycemic tendency among arsonists. Ada psychiat. scand.
69, 445^52.
VIRKKUNEN, M. (1986) Insulin secretion during the glucose tolerance test among habitually
violent and impulsive offenders. Aggress. Behav. 12, 303-310.
VIRKKUNEN, M. & HUTTUNEN, M. O. (1982) Evidence for abnormal glucose tolerance among
violent offenders. Neuropsychobiol. 8, 30-34.
WEISS, V. (1986) From memory span and mental speed toward the quantum mechanics of
intelligence. Personal, individ. Diffs, 7, 737-749.
WENK, G. L. (1989) An hypothesis on the role of glucose in the mechanism of action of cognitive
enhancers. Psychopharmocol. 99, 431-438.
WIDOM, B. & SIMONSON, D. C. (1990) Glycaemic control and neuropsychologic function during
hypoglycaemia in patients with insulin-dependent diabetes mellitus. Ann. intern. Med. 112,
904-912.