Background

The rise in nutrition-related illness highlights the need for targeted health promotion and interventions across the lifespan and for future generations⁽ Reference Lopez, Mathers and Ezzati ^{1 )}. Traditionally the focus of such interventions was on prevention of chronic disease and premature death. However, there is now a large body of evidence demonstrating that cognitive decline accompanies certain metabolic health conditions such as type 2 diabetes, metabolic syndrome and obesity and that modifiable lifestyle factors including diet may contribute significantly to the risk of cognitive decline, including dementia⁽ Reference Kanoski and Davidson ^{2 )}. Consequently, there has been an increasing interest in the effects of nutrition on cognitive performance and more specifically how cognitive performance can be optimised using nutritional interventions. When looking across the lifespan, broadly speaking nutritional interventions offer opportunity to (i) optimise cognitive development during infancy and childhood, (ii) ensure the highest levels of cognitive function during adulthood and (iii) prevent cognitive decline in older age (see Fig. 1).

Fig. 1. Nutrition and cognition: potential for optimising cognitive performance across the lifespan.

The macronutrient glucose is perhaps most thoroughly researched in terms of its effects on cognition. Here the impact of diet-based glycaemic response and glucose regulation on cognitive processes across the lifespan will be reviewed. Before considering the relationship of glucose, glycaemic response and cognitive processes, some features of glucose metabolism important for the understanding of its role in cognition will be discussed.

Glucose: the major source of energy for the brain

All processes of cells (including nerve cells) require energy. In human subjects and most animals, ATP works as the main carrier of chemical energy. The human body uses three types of molecules to yield the necessary energy to drive ATP synthesis: fats, proteins and carbohydrates. Aerobic carbohydrate metabolism is the main source of energy available for brain tissue⁽ Reference McIlwain and Bachelard ^{3 )}. Compared with other organs, the brain possesses paradoxically limited stores of glycogen, which without replenishment are exhausted in up to 10 min. In nervous tissue, glycogen is stored in astrocytes. Astrocytes participate significantly in brain glucose uptake and metabolism and due to their location and metabolic versatility⁽ Reference Lehninger, Nelson and Cox ^{4 )}. The entry of glucose into the brain is mediated by the family of GLUT, which are adapted to the metabolic needs of the tissue in which it is found. The primary GLUT isoforms in the brain are GLUT1 and GLUT3 but others have been detected in different brain regions, at a lower level of expression⁽ Reference Maher, Vannucci and Simpson ^{5 )}.

The immense expenditure of energy by the brain relative to its weight and volume is thought to be due to the need to maintain ionic gradients across the neuronal membrane⁽ Reference Ames ^{6 )}. In addition, there is no break from the brain's energy demand as the rate of brain metabolism is relatively steady day and night, and may even increase slightly during the dreaming phases of sleep⁽ Reference Maquet ^{7 )}. Thus the energy requirements of brain tissue are exceptionally constant⁽ Reference Clarke and Sokoloff ^{8 )} and glucose deprivation can severely disrupt neuronal activity, producing electroencephalogram patterns characteristic of lowered cognitive functioning⁽ Reference Holmes, Hayford and Gonzalez ^{9 )}. Indeed, when blood glucose drops below 4 mm/l (72 mg/dl; hypoglycaemic condition), it can cause discomfort, confusion, coma, convulsions, or even death in extreme conditions⁽ Reference Nelson, Cox and Lehninger ^{10 )}. Conversely, persistent blood glucose concentrations above the normal range (hyperglycaemic condition) can also have damaging physiological effects. Because glucose exerts osmotic pressure in the extracellular fluid, extremely high blood glucose concentrations can cause cellular dehydration, loss of glucose in the urine, which can affect kidney function and deplete the body's supply of fluids and electrolytes⁽ Reference Guyton and Hall ^{11 )}.

Glucose brain metabolism: changes across the lifespan

The rate of glucose brain metabolism changes across the lifespan. Initially, there is a rise in the rate of glucose utilisation from birth until about age 4 years, at which time the child's cerebral cortex uses more than double the amount of glucose compared with adults. This high rate of glucose utilisation is maintained from age 4 to 10⁽ Reference Moeller, Ishikawa and Dhawan ^{12 )}. Childhood is a time of intense learning and therefore coincides with the most metabolically expensive period⁽ Reference Haymond ^{13 )}. The high energy demand of a child's brain requires the use of the majority of hepatically generated plasma glucose⁽ Reference Haymond and Sunehag ^{14 )}. In addition, glucose supply needs to be particularly stable as impairments are thought to occur at higher plasma glucose level (4·2 mm/l)⁽ Reference Jones, Borg and Boulware ^{15 )}. After this period, there is a gradual decline in glucose metabolic rate, reaching adult values by age 16–18 years⁽ Reference Chugani ^{16 )}. This is followed by a plateau phase until middle age after when a significant age-related decline in cerebral glucose metabolism can be observed⁽ Reference Moeller, Ishikawa and Dhawan ^{12 )}. This age-specific metabolic pattern of glucose consumption has not been observed in other species and it has been argued that this could be a driver or indeed a consequence of human cognition⁽ Reference Caravas and Wildman ^{17 )}.

Most children and young adults maintain circulating glucose within the normal range throughout cycles of feeding and fasting and balanced alterations in secretions of regulatory hormones⁽ Reference Nordlie, Foster and Lange ^{18 ,} Reference Pilkis, El-Maghrabi and Claus ^{19 )}. In contrast, older adults have a broader range over which circulating glucose is maintained and in addition have attenuated counter regulatory responses⁽ Reference Kent ^{20 ,} Reference Kalyani and Egan ^{21 )}. Circulating insulin levels tend to be elevated with age (approximately 8% higher than in young adults) and are indicative of reduced insulin sensitivity⁽ Reference Melanson, Greenberg and Ludwig ^{22 )}. Reduced insulin sensitivity or insulin resistance is a condition where individuals develop resistance to the cellular actions of insulin, characterised by an impaired ability of insulin to inhibit glucose output from the liver and to promote glucose uptake in fat and muscle⁽ Reference Reaven ^{23 )}. Both effects of insulin insensitivity on liver and muscle tissue cause elevations in peripheral blood glucose levels⁽ Reference Kahn, Hull and Utzschneider ^{24 )}. Changes in insulin action have been observed at different stages of development. Basal insulin secretion increases during puberty, falling back to pre-pubertal levels in adulthood⁽ Reference Caprio, Plewe and Diamond ^{25 )}. Yet, fasting glucose levels remain constant, implying an increase in tissue resistance to insulin coinciding with puberty⁽ Reference Savage, Smith and Dunger ^{26 )}. The reason for the puberty-induced reduction of insulin sensitivity appears to be growth-hormone related⁽ Reference Jørgensen, Møller and Wolthers ^{27 –} Reference Kyho and O’Sullivan ^{29 )}. Growth hormone secretion reaches a peak at around puberty and will begin to decrease by age 21 years⁽ Reference Savine and Sönksen ^{30 )}. It is commonly in middle age where insulin resistance and poor glucose tolerance become a health issue⁽ Reference Facchini, Hua and Abbasi ^{31 )}. Given that the brain uses glucose as a primary substrate for brain function, it is perhaps not surprising that conditions that affect peripheral and central glucose regulation and utilisation may also affect cognitive functioning. Moreover, based on the evidence earlier there might be critical periods in which alterations in cerebral glucose supply might have more pronounced effects on cognitive performance.

Acute administration of a glucose load: prototypical experimental paradigm

Over the past 30 years, a large body of literature has demonstrated beneficial effects of acute glucose administration on cognition in various populations⁽ Reference Messier ^{32 ,} Reference Smith, Riby and van Eekelen ^{33 )}. The general methodology used in these studies involves administration of an oral glucose load (usual range between 25 and 50 g glucose) after a period of fasting (ranging from 2 h to overnight fast) followed by assessment of cognitive performance and measurement of capillary blood glucose levels⁽ Reference Messier ^{32 ,} Reference Manning, Hall and Gold ^{34 )}.

Using this experimental paradigm, beneficial effects have been observed across different populations. For example, glucose administration has been shown to enhance cognitive performance in adolescents⁽ Reference Smith and Foster ^{35 )}, young adults⁽ Reference Sünram-Lea, Foster and Durlach ^{36 –} Reference Foster, Lidder and Sünram ^{45 )}, older adults⁽ Reference Meikle, Riby and Stollery ^{46 ,} Reference Riby, Meikle and Glover ^{47 )} and improvements have been observed in subjects with mild or severe cognitive pathologies, including individuals with Alzheimer's disease and Down's syndrome⁽ Reference Messier ^{32 ,} Reference Smith, Riby and van Eekelen ^{33 )}. In addition, facilitation of cognitive performance induced by elevations in plasma glucose levels has also been reported in patients with schizophrenia⁽ Reference Newcomer, Craft and Fucetola ^{48 ,} Reference Fucetola, Newcomer and Craft ^{49 )}. It is important at this point to note that these results do not reflect a negative effect of fasting on cognition and memory, as the degree of fasting in which participants engaged was not exceptional and participants do not reach blood glucose levels associated with hypoglycaemia.

In terms of cognitive tasks affected, benefits have been found to occur in a range of cognitive domains, including information processing and attention⁽ Reference Meikle, Riby and Stollery ^{46 ,} Reference Owens and Benton ^{50 –} Reference Stollery and Christian ^{53 )}, working memory⁽ Reference Sünram-Lea, Foster and Durlach ^{36 ,} Reference Sünram-Lea, Foster and Durlach ^{37 ,} Reference Owen, Scholey and Finnegan ^{42 ,} Reference Kennedy and Scholey ^{43 ,} Reference Hall, Gonder-Frederick and Chewning ^{54 )}, executive function⁽ Reference Brandt, Gibson and Rackie ^{55 ,} Reference Benton, Owens and Parker ^{56 )} , problem solving⁽ Reference Miller, Bourrasseau and Blampain ^{57 )} and long-term memory⁽ Reference Sünram-Lea, Foster and Durlach ^{36 –} Reference Sünram-Lea, Owen and Finnegan ^{38 ,} Reference Owen, Finnegan and Hu ^{40 –} Reference Owen, Scholey and Finnegan ^{42 ,} Reference Sünram-Lea, Dewhurst and Foster ^{58 –} Reference Sünram-Lea, Foster and Durlach ^{60 )}. The clearest enhancement effects of increased glucose supply have been observed for long-term memory over a variety of conditions and paradigms⁽ Reference Hoyland, Lawton and Dye ^{61 )}. As different aspects of cognition pertain to different neural structures and networks, this allows speculation about the areas of the brain that might be particularly susceptible to glycaemic fluctuations. The robust effects on long-term memory, suggest that glucose facilitation may be particularly pronounced in tasks that pertain to the hippocampal formation⁽ Reference Sünram-Lea, Foster and Durlach ^{36 )}. The level of task demand is a further moderating factor for cognitive enhancement by increased glucose availability. Indeed, in younger participants, glucose-related improvement of cognition appears to be related to the difficulty of the cognitive tasks. Tasks which are more cognitively demanding appear to be more sensitive to the effect of glucose loading⁽ Reference Sünram-Lea, Foster and Durlach ^{37 ,} Reference Kennedy and Scholey ^{43 ,} Reference Scholey, Harper and Kennedy ^{62 )}. In addition, depletion of memory capacity and/or glucose resources in the brain due to performing a concomitant cognitive task might be crucial to the demonstration of a glucose facilitation effect⁽ Reference Sünram-Lea, Foster and Durlach ^{37 )}.

While both young and older adults show cognitive improvement after the oral administration of glucose, the effects appear to be more profound in older individuals⁽ Reference Hall, Gonder-Frederick and Chewning ^{54 )}. Cognitive decline over the aging process has been well documented⁽ Reference Gold, McGaugh and Hankins ^{63 –} Reference Gold and Stone ^{65 )}. Traditionally, cognitive impairments are assumed to reflect deficits caused by damage of brain areas or systems in which cognitive processing in normal subjects occurs. However, more recently there has been a focus shift on specific physiologic and metabolic impairments that appear to contribute to the cognitive decline observed in ageing. Older adults have a broader range over which circulating glucose is maintained and in addition have attenuated counter regulatory responses⁽ Reference Andres and Tobin ^{66 ,} Reference Frank, Roland and Sturis ^{67 )}. These suboptimal metabolic and cognitive conditions are likely to make older individuals more susceptible to glucose facilitation of cognitive performance⁽ Reference Messier ^{68 ,} Reference Greenwood and Winocur ^{69 )}.

The energy cost for effortful, controlled or executive processes appears to be significantly higher than that for automatic or reflexive processes⁽ Reference Gailliot and Baumeister ^{70 )}. Effortful, controlled or executive processes are processes that are reliant on the central executive, in which thoughts, behaviours and actions are coordinated to allow goal directed and purposeful behaviour, while automatic and reflexive behaviours are evolutionarily predisposed or learned behaviours elicited by environmental stimuli⁽ Reference Miller and Wallis ^{71 )}. Indeed, lowered peripheral glucose levels following performance of a cognitively demanding task have been reported⁽ Reference Scholey, Harper and Kennedy ^{62 ,} Reference Fairclough and Houston ^{72 )}. This fall in plasma glucose could reflect a more efficient transfer of glucose to the brain which in turn results in increased provision centrally⁽ Reference Scholey, Harper and Kennedy ^{62 )}. One should be cautious when making assumptions about peripheral blood glucose levels and their putative effects on the brain, as other studies have failed to demonstrate such findings⁽ Reference Molden, Hui and Scholer ^{73 ,} Reference Boyle, Lawton and Allen ^{74 )}. Nevertheless, the evidence suggests that cognitively demanding tasks and in particular those relying on executive functions are sensitive to changes in glucose availability⁽ Reference Gailliot and Baumeister ^{70 ,} Reference Kelly, Sünram-Lea and Crawford ^{75 )}. Administration of a glucose drink would consequently provide the brain with sufficient metabolic resources for extensive cognitive processing and support the brain areas under greatest cognitive load, and thus lead to improved performance.

A further moderating factor of the impact of glucose on cognitive function is dose. As with many substances affecting cognitive performance, glucose displays an inverted U-shaped dose–response curve and its effect is time dependent⁽ Reference Gold ^{76 )}. For older adults 25 g glucose appear to be the optimal dose, with performance deterioration observed after administration of 75 g glucose⁽ Reference Parsons and Gold ^{77 )}. For young adults 25 g also seems to most reliably facilitate cognitive performance; however, there is evidence suggesting that the optimal dose or shape of the dose–response curve may be dependent on inter-individual difference in glucose metabolism, and the cognitive domain being assessed⁽ Reference Owen, Finnegan and Hu ^{40 )}. Of note, the cognitive enhancing effects of pharmaceutical substances such as stimulants (methylphenidate, modafinil) and acetylcholinesterase inhibitor (dementia drugs) in healthy individuals are generally moderate or small (as estimated by Cohen's d effect size) according to systematic reviews⁽ Reference Repantis, Schlattmann and Laisney ^{78 ,} Reference Repantis, Laisney and Heuser ^{79 )}. The effects of glucose administration are comparable with those from pharmaceutical interventions, with effect sizes for glucose effects range from 0·34 to 4·26, with typical values of 1·02, 0·81 and 1·07 for heavily loaded working memory and verbal episodic recognition and recall, respectively⁽ Reference Riby ^{80 )}.

Glucose facilitation of cognitive performance: putative underlying mechanisms

The precise mechanisms by which increased peripheral and/or central glucose availability affects cognitive processes are still unclear. There are two broad theoretical approaches: energetic demand models and domain specific models. Energetic demand models have their basis in the observation that the amount of mental effort involved in cognitive processing is an important determinant of a task's susceptibility to glucose enhancement⁽ Reference Scholey, Harper and Kennedy ^{62 ,} Reference Gailliot and Baumeister ^{70 ,} Reference Fairclough and Houston ^{72 )}. Domain specific theories, alternatively, stipulate that certain areas of the brain are more susceptible to changes in glucose availability⁽ Reference Foster, Lidder and Sünram ^{45 ,} Reference Riby ^{80 ,} Reference Messier, Durkin and Mrabet ^{81 )}. However, these different approaches are by no means mutually exclusive, their relative explanatory value depending on cognitive task and brain structure.

Glucose metabolism varies throughout tissue/cell types of the brain, with a clearly established correlation between increased energy metabolism and increased neuronal activity and energy metabolism⁽ Reference Sokoloff ^{82 )}. Both the rate of blood to brain glucose transport⁽ Reference Lund-Andersen ^{83 )} and glucose metabolism⁽ Reference Reivich, Gur and Alavi ^{84 )} are stimulated in different areas in the brain during cognitive tasks relevant to that area. There is evidence that performing cognitively demanding tasks increases total brain consumption by as much as 12%⁽ Reference Madsen, Hasselbalch and Hagemann ^{85 )}.

As described, glucose exerts quite robust effects on long-term memory tasks. The hippocampus is the brain region most strongly implicated in long-term memory performance⁽ Reference Aggleton and Brown ^{86 )}. Microdialysis measurements of brain glucose have shown a large decrease in hippocampal extra cellular fluid (32%) in rats tested for spontaneous alternation on a four-arm maze (a difficult memory task), while a smaller decrease (11%) was seen in rats tested on a simpler three arm-maze, suggesting that the changes observed in extra cellular fluid glucose are related to task difficulty. The fall in extra cellular fluid can be prevented by administration of glucose, which in turn leads to enhanced memory performance⁽ Reference McNay, Fries and Gold ^{87 )}. There is some evidence that the concentration of extracellular glucose in the brain after its transfer across the blood–brain barrier from plasma glucose varies with brain region from 1·3 mm/l in the hippocampus to 0·3–0·5 mm/l in the striatum⁽ Reference McNay, McCarty and Gold ^{88 )}. These findings suggest that the hippocampal area is particularly sensitive to energy fluctuations. However, the hippocampus has relatively greater glycogen stores compared with other areas suggesting that it has evolved some protection against temporary deficits (13 mm/l compared with 5–6 mm/l in the cerebral cortex⁽ Reference Dalsgaard, Madsen and Secher ^{89 )}.

There is evidence suggesting that the cognitive facilitation observed after glucose loading is due to an increase in enhancement of acetylcholine synthesis and/or release⁽ Reference Messier ^{32 )}. In addition, elevated insulin in response to hyperglycaemia rather than glucose levels per se may moderate memory performance⁽ Reference Watson and Craft ^{90 )}. Originally, insulin was considered only as a peripheral hormone, unable to cross the blood–brain barrier and to affect the central nervous system. However, there is now increasing evidence that neuronal glucose metabolism is antagonistically controlled by insulin and cortisol⁽ Reference Gray, Meijer and Barrett ^{91 ,} Reference Duarte, Moreira and Oliveira ^{92 )}.

The hippocampus, the brain region key to memory and learning, has particularly high levels of insulin receptors⁽ Reference Marks, Porte and Stahl ^{93 ,} Reference Dore, Kar and Rowe ^{94 )} which are known to promote cellular glucose uptake⁽ Reference Messier ^{32 ,} Reference Craft and Watson ^{95 )}. Insulin-sensitive GLUT such as GLUT4 are also enriched in the hippocampus⁽ Reference McEwen and Reagan ^{96 )}. Given the established role of the hippocampus in memory, elevated insulin in response to hyperglycaemia may boost glucose utilisation in the hippocampus and result in improved performance⁽ Reference Craft, Dagogo-Jack and Wiethop ^{97 )}.

Glucose might also act via peripheral physiological mechanisms, which in turn facilitate central mechanisms involved in cognition. Messier and White⁽ Reference Messier and White ^{98 ,} Reference White, Frederickson, McGaugh and Fenton ^{99 )} suggested that changes in cell membrane transport in the liver following administration of high doses of glucose and fructose (>1000 mg/kg) are detected by the coeliac ganglion, then transformed into neural signals and finally carried via the vagus nerve to the brain. In accordance with this suggestion, coeliac ganglion lesions (which block most of the efferents of the liver) have been shown to abolish the mnemonic effect of glucose⁽ Reference White, Frederickson, McGaugh and Fenton ^{99 )}. To date there is no concrete information available concerning how this proposed neural signal from the liver might influence cognitive performance when it reaches the brain. However, the nucleus of the solitary tract in the brain stem is the main relay station for afferent vagal nerve fibres and has widespread projections to numerous areas in the cerebral cortex, including the hippocampus and the prefrontal cortex⁽ Reference Clark, Naritoku and Smith ^{100 )}.

Research also shows that difficult tasks are more likely to be susceptible to glycaemic interventions. Difficult tasks include those involving executive functions pertaining to frontal brain regions: inhibition/self-control, working memory and mental flexibility⁽ Reference Manard, Carabin and Jaspar ^{101 ,} Reference Diamond ^{102 )}. Evidence suggested that tasks that demand such cognitive control and attentional resources appear to be more energy demanding⁽ Reference Gailliot and Baumeister ^{70 )}. Consequently, another area of the brain which appears to be particularly sensitive to energy fluctuations is the frontal cortex. The cerebral cortex, and in particular the prefrontal cortex, represents the neural basis of higher cognitive functions⁽ Reference Frith and Dolan ^{103 ,} Reference Fuster ^{104 )}. Aspects of higher-level cognition were probably one of the last cognitive abilities to develop ontogenetically. Based on the ‘last-in, first-out rule’, cognitive abilities that developed last ontogenetically are likely the first to become impaired when cognitive and/or physiological resources are compromised. Consequently, optimal performance on tasks pertaining to function of the pre-frontal cortex might require more energetic fuel than others. The research is not yet conclusive, but suggests that the underlying mechanism is multifarious. The most likely scenario is that glucose provides additional metabolic fuel under high demand conditions and that certain areas of the brain are more susceptible to limitations in fuel supply.

Glycaemic regulation and cognition

Over longer time periods, elevated blood glucose levels act as an allostatic load to biological systems and can accelerate disease processes. Chronic hyperglycaemic conditions negatively affect glycaemic regulation, i.e. the ability of the body to effectively regulate blood glucose levels and to remove glucose from the blood⁽ Reference Eizirik, Korbutt and Hellerström ^{105 ,} Reference Marshak, Leibowitz and Bertuzzi ^{106 )}. In addition to high carbohydrate loading, fat ingestion is also associated with development of insulin resistance through inflammation mediated mechanisms⁽ Reference Xu, Barnes and Yang ^{107 )}. Evidence suggests that the risk of impaired glucose regulation and type 2 diabetes is associated with a high trans fatty acid intake and a low poly-unsaturated to saturated fat intake ratio⁽ Reference Manco, Calvani and Mingrone ^{108 )}.

Consequently, glycaemic control is another important factor when considering cognition across the life-span⁽ Reference Lamport, Lawton and Mansfield ^{109 )}. Conditions in which glycaemic regulation is severely compromised are diabetes type 1 and type 2, impaired glucose tolerance and impaired fasting glucose. Cognitive impairments were indeed one of the earliest recognised neurological complications associated with diabetes⁽ Reference Miles and Root ^{110 )}. To date, numerous studies have compared cognitive functioning in diabetic patients with non-diabetic controls⁽ Reference Brands, Biessels and De Haan ^{111 )}. Although these studies differed widely with respect to patient characteristics (age, duration and type of diabetes) and cognitive tests used, the majority of these studies demonstrated cognitive impairments in this population, which included decreased performance on various attention and memory tasks⁽ Reference Lamport, Lawton and Mansfield ^{109 ,} Reference Tun, Nathan and Perlmuter ^{112 –} Reference Brands, Kessels and de Haan ^{115 )}. Risk factors associated with cognitive complications in diabetes appear to be (i) degree of metabolic control⁽ Reference Meuter, Thomas and Grüneklee ^{116 )} and (ii) repeated episodes of hypoglycaemia⁽ Reference Auer ^{117 )}. It is therefore not surprising that in children diagnosed with type 1 diabetes before age 10 years, cognitive complications are generally only observed if they have a history of hypoglycaemic seizures⁽ Reference Kaufman, Epport and Engilman ^{118 )}. It is evident from the literature that type 2 diabetes is the preventable metabolic condition associated with an increased risk of cognitive dysfunction⁽ Reference Strachan, Deary and Ewing ^{119 –} Reference Gallacher, Pickering and Elwood ^{121 )}.

However, there is now increasing evidence of a relationship between glycaemic control and cognitive functions in non-diabetic populations⁽ Reference Lamport, Lawton and Mansfield ^{109 ,} Reference Awad, Gagnon and Messier ^{113 )}. Cognitive decline over the ageing process has been well documented and it has been suggested that normal ageing may represent a condition in which there is greater vulnerability to disrupted glucose regulation⁽ Reference Gold and Stone ^{65 )}. Indeed, evidence to support this hypothesis is provided by the finding that memory performance in elderly participants with poor glucose regulation is impaired relative to elderly participants with good glucose regulation⁽ Reference Messier, Tsiakas and Gagnon ^{122 –} Reference Craft, Zallen and Baker ^{124 )}. Moreover, age-related changes in glucose metabolism have been identified as a risk factor for Alzheimer's disease⁽ Reference Messier ^{32 ,} Reference Watson and Craft ^{90 ,} Reference Hoyer ^{125 )}. Consistent with this notion is the finding that hyperglycaemia (induced through oral and intravenous glucose administration) can facilitate memory performance in Alzheimer's patients, at least in the early stages of the disease⁽ Reference Craft, Asthana and Newcomer ^{126 )}. Interestingly, alterations in blood glucose regulation seem to depend on the severity of the disease process. More specifically, high insulin levels are observable at the very early (very mild) stages and decline as dementia progresses. Moreover, memory facilitation can be achieved through glucose administration in the early stages and the degree of facilitation decreases at more advanced stages of the disease⁽ Reference Craft, Dagogo-Jack and Wiethop ^{97 )}. Indeed, as abnormalities in brain insulin resistance and deficiency have been observed in Alzheimer's disease, and the fact that molecular and biochemical hallmarks of Alzheimer's disease, such as neuronal loss, synaptic disconnection, tau hyperphosphorylation and amyloid-beta accumulation overlap with type 1 and type 2 diabetes, the term ‘type 3 diabetes’ has been suggested to account for the underlying abnormalities associated with Alzheimer's disease-type neurodegeneration⁽ Reference Suzanne ^{127 )}. A combination of diet and exercise has been demonstrated to have cognitive and metabolic benefits (improved glucose and insulin metabolism) in adults with impaired glucose tolerance⁽ Reference Watson, Reger and Baker ^{128 ,} Reference Goodyear, Laurie and Kahn ^{129 )}. Dietary lifestyle changes can have a positive impact throughout the lifespan and appear to not only reduce the risk of acquiring cognitive impairments, but can also attenuate existing impairments. For example, a recent study showed that a 4-week low-saturated fat/low-glycaemic index (GI) diet resulted in improved memory performance and insulin metabolism in adults with amnestic mild cognitive impairment⁽ Reference Bayer-Carter, Green and Montine ^{130 )}.

Perhaps more worryingly, performance decrements due to poor glucose regulation have been reported in younger individuals⁽ Reference Lamport, Lawton and Mansfield ^{109 ,} Reference Awad, Gagnon and Messier ^{113 )}. For example, recent studies have shown that even in a healthy young student population those with better glucose regulation (those who had the smallest blood glucose rise following glucose ingestion) perform better on tests of memory⁽ Reference Owen, Scholey and Finnegan ^{42 ,} Reference Benton, Owens and Parker ^{56 ,} Reference Awad, Gagnon and Messier ^{113 ,} Reference Donohoe and Benton ^{131 –} Reference Awad, Gagnon and Desrochers ^{133 )}, vigilance⁽ Reference Benton, Owens and Parker ^{56 ,} Reference Donohoe and Benton ^{131 )}, planning⁽ Reference Donohoe and Benton ^{131 )} and dichotic listening⁽ Reference Parker and Benton ^{134 )} compared with those with poorer glucose regulation. In addition, glucose administration preferentially improved performance in those with poorer glucose regulation and the effects are less likely to be observed in good glucose regulators in both old and young populations⁽ Reference Messier ^{32 )}. This would suggest that glucose control or tolerance is associated with cognition throughout the lifespan. Overall there appears to be some evidence that glucoregulation may exert direct effects on cognitive function in that those with poor glucoregulation may demonstrate mild cognitive deficit compared with good glucoregulation. However, research in young adults is limited, furthermore the methodologies for determining glucoregulatory control have been varied. Only a few studies have used a standardised oral glucose tolerance test for the evaluation of glucose tolerance in healthy young adults⁽ Reference Owen, Scholey and Finnegan ^{42 ,} Reference Donohoe and Benton ^{135 )}. The oral glucose tolerance test involves administration of a 75 g glucose load after a minimum 8 hour fast and is the gold standard test for the diagnosis of diabetes mellitus^{( 136 )}. Moreover, the majority of studies have only assessed one specific measurement of glucose tolerance. Several glucoregulatory indices have been previously evaluated for their relationship with cognitive performance in younger and older participants. These include: fasting levels, peak glucose levels, recovery and evoked glucose to baseline levels and incremental area under the curve⁽ Reference Owen, Scholey and Finnegan ^{42 )}. At a younger age, the deficits associated with poor glucoregulation may be minimal and hard to detect therefore it is important to identify the most sensitive marker. A study in our laboratory found area under the curve, which takes baseline blood glucose levels into account (area under the curve with respect to ground⁽ Reference Pruessner, Kirschbaum and Meinlschmid ^{137 )}), to be the best predictor of cognitive performance, whereas the most commonly used incremental area under the curve did not show a strong association⁽ Reference Owen, Scholey and Finnegan ^{42 )}. This suggests that overall circulating glucose levels may be an important factor in the assessment of glucoregulation in sub-clinical; populations with normal glucose tolerance as defined by the WHO^{( 136 )}. Indeed, a recent study identified fasting blood glucose levels as a predictor for cognitive performance⁽ Reference Hawkins, Gunstad and Calvo ^{138 )}. Young adults who were obese but otherwise healthy had higher fasting glucose levels compared with normal weight participants. In addition, higher glucose levels were associated with poorer cognitive performance on tests of inhibitory control, especially among individuals with pre-diabetic levels. Consequently, subclinical elevations in blood glucose may contribute to cognitive impairments before the development of clinically defined disease states.

The postprandial glycaemic response and cognition

When considering the nature of glucose availability, the rate at which food increases and maintains blood glucose, i.e. ‘the GI’ appears to be an important modulating factor. Shortly after intake of a high GI food there is a relatively rapid rise in blood glucose levels followed by a corresponding rapid decrease, whereas after the intake of a low GI food there is a relatively smaller rise in blood glucose followed by more stable blood glucose concentration. GI solely provides a measure of carbohydrate quality⁽ Reference Wolever, Vorster and Björck ^{139 )}, whereas glycaemic load (GL) takes into account the amount of carbohydrates consumed and is calculated by multiplying the amount of available carbohydrate in a food item by the GI of the food and dividing this by 100⁽ Reference Gilsenan, de Bruin and Dye ^{140 )}.

Although the effect of glucose administration has been extensively studied in an acute, short-term context, much remains to be done in order to establish the cognitive effects associated with foods of low or high GI and GL. Most studies examining the effects of GI on cognition have focused on the effect of breakfast on children's cognitive performance. Children may be particularly sensitive to breakfast interventions due to their greater energetic needs during this period compared with adults⁽ Reference Chugani ^{16 )}. Moreover, it has been suggested that in younger children, the overnight fast induces greater metabolic stress, as the higher the ratio of brain to liver weight and the greater metabolic rate per unit of brain weight, the greater the demand on glycogen stores⁽ Reference Pollitt, Leibel and Greenfield ^{141 )}. It has been shown that children at risk for malnourishment have improved cognition and learning at school if provided with breakfast⁽ Reference Hoyland, Dye and Lawton ^{142 )}. Moreover, in developed countries it has been found that skipping breakfast can result in impaired cognitive performance⁽ Reference Hoyland, Dye and Lawton ^{142 ,} Reference Benton and Parker ^{143 )}. This suggests that increased plasma glucose availability due to breakfast consumption leads to better cognitive performance. Mahoney et al. ⁽ Reference Mahoney, Taylor and Kanarek ^{144 )}, investigated the optimal rate of glucose supply following breakfast consumption, comparing a low GI breakfast with a high GI breakfast and found that when children consumed the low GI food they remembered significantly more than when they ate the high GI breakfast. Ingwersen et al. ⁽ Reference Ingwersen, Defeyter and Kennedy ^{145 )} compared the cognitive effects of a low GI breakfast and a high GI breakfast across the morning and found that performance on attention tasks was poorer 130 min after the high GI breakfast compared with the low GI breakfast. Furthermore, the low GI breakfast prevented a decline in memory performance. Overall, the results of studies assessing GI in children suggest that a lower postprandial glycaemic response may be protective against a decline in memory and attention throughout the morning⁽ Reference Mahoney, Taylor and Kanarek ^{144 –} Reference Young and Benton ^{151 )}. However, the evidence is far from conclusive⁽ Reference Brindal, Baird and Slater ^{152 ,} Reference Iovino, Stuff and Liu ^{153 )} and few studies have actually profiled the glycaemic response in children⁽ Reference Brindal, Baird and Danthiir ^{154 )}.

From a metabolic perspective, adolescence might also be a time where greater susceptibility to glycaemic variations is observed due to the specific metabolic conditions observed during that time of development⁽ Reference Caprio, Plewe and Diamond ^{25 ,} Reference Savage, Smith and Dunger ^{26 )}. However, few studies have looked at the effects of GI in adolescent populations and the results are somewhat contradictory. Wesnes et al. ⁽ Reference Wesnes, Pincock and Richardson ^{146 )} found that a low GI breakfast resulted in better memory performance and attention, but the age range used in this study was quite large (6–16 years). Other studies found performance benefits following a high GI intervention when assessing memory performance⁽ Reference Micha, Rogers and Nelson ^{149 ,} Reference Smith and Foster ^{155 )} whereas a low GI intervention proved to be beneficial for measures of attention/information processing⁽ Reference Micha, Rogers and Nelson ^{149 )}. Cooper et al. ⁽ Reference Cooper, Bandelow and Nute ^{150 )} found no difference between high GI and low GI on reaction times, but better performance on an executive function task following low GI.

In adult populations, the outcome of investigating the effects of GI has also been somewhat inconsistent. Some show beneficial effects on cognitive performance of low-GI foods⁽ Reference Benton, Ruffin and Lassel ^{147 ,} Reference Nilsson, Radeborg and Björck ^{156 ,} Reference Nilsson, Radeborg and Björck ^{157 )} whereas others show no such effects⁽ Reference Kaplan, Greenwood and Winocur ^{158 ,} Reference Dye, Gilsenan and Quadt ^{159 )}. Benton et al. ⁽ Reference Benton and Jarvis ^{148 )} compared three breakfasts varying in GL from 2·5 to 17·86 and found that the higher GL foods led to poorer memory performance. Lamport et al. ⁽ Reference Lamport, Dye and Mansfield ^{160 )} investigated the effects of low GI and high GI evening meals followed by a high GI standard breakfast on subsequent cognitive performance. Although no significant differences between evening meals on cognitive performance were observed, the high GI evening meal was associated with better memory performance the following morning after breakfast had been consumed.

To date only a few studies have been carried out into the effect of low GI and GL foods on glycaemic control and cognition in older adults, or populations with pre-existing metabolic and/or cognitive impairments. Kaplan et al. ⁽ Reference Kaplan, Greenwood and Winocur ^{158 )} found no differences between meals of different GI in performance in elderly adults. Nilsson, Radeborg and Björk⁽ Reference Nilsson, Radeborg and Björck ^{156 )} showed that in a sample ranging from 49 to 70 years, performance was better in the late postprandial period after consumption of a low-GI compared with a high-GI breakfast. In adults with type 2 diabetes consuming a low-GI carbohydrate meal, relative to a high-GI carbohydrate meal, has been shown to result in better cognitive performance in the postprandial period⁽ Reference Papanikolaou, Palmer and Binns ^{161 )}. However, two other studies by Lamport et al. ⁽ Reference Lamport, Dye and Mansfield ^{160 ,} Reference Lamport, Lawton and Mansfield ^{162 )} did not find any benefits following consumption of a low GL breakfast. All of these studies investigated the acute effects of postprandial glycaemic manipulation and it may be the case that for these populations cognitive effects will only be evident with chronic improvements in glycaemic control. Indeed, dietary interventions (combined with exercise interventions) have been shown to result in improved cognitive performance in adults with impaired glucose control when they were implemented for 12 months⁽ Reference Watson, Reger and Baker ^{128 )}.

Overall, it appears that a quick rise in blood glucose levels has some short-term benefits, most notably on memory performance; whereas over longer periods of time (i.e. throughout the morning) a more stable blood glucose profile seems to be more beneficial. In normoglycaemic samples, effects of low GI and/or low GL foods were usually observed in the late postprandial period (75–222 min) where they seem to prevent a decline in attention and memory⁽ Reference Mahoney, Taylor and Kanarek ^{144 ,} Reference Ingwersen, Defeyter and Kennedy ^{145 ,} Reference Benton, Ruffin and Lassel ^{147 )}. In populations with abnormalities in glucose regulation, benefits of low GI foods have been reported in particular following longer-term intervention.