Introduction
Caffeine is found naturally in over sixty species of plants, a number of which are consumed for their psychoactive effects, including coffee (Coffea robusta/arabica), tea (Camellia sinensis), cocoa (Theobroma cacao), guaraná (Paullinia cupana), yerba maté (Ilex paraguariensis) and kola nut (Cola nitida/acuminata). These plants, or extracts of them, are regularly consumed as part of a normal diet, although caffeine can also be found in a number of other caffeine-containing products, such as soft drinks and medications. As a result, caffeine is consumed by the majority of the population in one form or another, with a study by Heatherley et al. (Reference Heatherley, Mullings and Tidbury1) demonstrating that 91 % of a UK sample consumed daily caffeine amounts equivalent to at least one cup of tea. Although caffeinated beverage preferences differ across countries(Reference Fredholm, Battig and Holmen2), tea and coffee represent the main sources of caffeine and are second only to water in terms of worldwide beverage consumption. It is therefore of great importance that we fully understand the health implications of these dietary components. Research into the effects of caffeine, as the main active compound within these beverages, is abundant, with the first investigation into its effects on behaviour being conducted in 1912(Reference Hollingworth3). However, the other compounds within these, prolific beverages have received surprisingly little research attention in terms of either their own independent psychoactive effects, or their ability to modify the behavioural effects of caffeine with which they are co-consumed. The aim of the present article is to provide a qualitative review of the literature to date examining behavioural effects of the main compounds consumed with caffeine. This will draw upon mechanistic and epidemiological data suggesting effects relevant to health before moving on to the main focus, intervention trials. Key issues to be considered are the quantities administered, the time-frame over which effects are monitored and the measures employed to assess any effects.
Caffeine
Caffeine is rapidly absorbed and distributed, with peak plasma levels being reached at about 30 min post-ingestion(Reference Magkos and Kavouras4). The action of caffeine is largely the result of antagonism of adenosine receptors, specifically A1 and A2A sub-receptors(Reference Fredholm, Battig and Holmen2). This blockade of the inhibitory action of adenosine has been linked to increased levels of dopamine, acetylcholine, noradrenaline, serotonin, glutamate and γ-aminobutyric acid(Reference Daly and Garattini5). In terms of vascular effects, despite exerting endothelium-dependent vasodilation via modulation of nitric oxide (NO) synthesis, caffeine has a net effect of reduced cerebral blood flow (CBF) as a consequence of vasoconstriction mediated by adenosine receptor antagonism(Reference Umemura, Higashi and Soga6).
The most widely accepted behavioural effect of caffeine is an increase in alertness, an effect that has been demonstrated in numerous studies (for example, Quinlan et al. (Reference Quinlan, Lane and Moore7), Smit & Rogers(Reference Smit and Rogers8) and Smith et al. (Reference Smith, Sturgess and Gallagher9)). Improvements to cognition, particularly reaction time (for example, Smit & Rogers(Reference Smit and Rogers8) and Durlach et al. (Reference Durlach, Edmunds and Howard10)) and vigilance (for example, Brice & Smith(Reference Brice and Smith11) and Childs & de Wit(Reference Childs and de Wit12)) are also well documented. It has been suggested that these effects merely represent alleviation of withdrawal in habitual caffeine consumers(Reference James13). However, comparisons of low/non-caffeine consumers' and regular consumers' behavioural responses to caffeine have demonstrated similar effects irrespective of habitual caffeine intake(Reference Richardson, Rogers and Elliman14, Reference Rogers, Martin and Smith15), even when there is no evidence of withdrawal in the habitual consumers(Reference Addicott and Laurienti16–Reference Hewlett and Smith18). There are, however, issues regarding the methodologies employed to test this suggestion, such as the possibility of non-consumers developing tolerance to these effects following repeated consumption of caffeine (for a review, see James & Rogers(Reference James and Rogers19)), and this is an area that requires further research before a definitive conclusion can be reached.
The majority of effects of caffeinated plants and beverages may be explained by the presence of caffeine. However, caffeinated plants contain a number of other components that either have their own direct action on the central nervous system or work indirectly to modulate the effects of caffeine.
Concomitant active components and their mechanisms
There are many phytochemicals present in natural sources of caffeine that have the potential to exert myriad independent, synergistic, additive or antagonistic effects to those of caffeine. These potentially bioactive phytochemicals are predominantly phenolic compounds, and in particular polyphenols, which contain multiple phenolic substructures. The one major exception to this is the amino acid l-theanine, which is present in tea and which modifies the activity of caffeine (see below). The primary phenolic compounds contained within tea are a range of flavan-3-ols, or catechins, of which epigallocatechin gallate (EGCG) predominates. However, tea also contains appreciable levels of the flavonol quercetin, phenolic acids, and, in the case of black tea, a range of theaflavins formed by the oxidation of the flavan-3-ols(Reference Lee and Ong20, Reference Ferruzzi21). Similarly, cocoa primarily contains flavan-3-ols, including catechin and epicatechin, along with their oligomer chain products proanthocyanidins(Reference Cooper, Campos-Gimenez and Alvarez22), and guaraná includes substantial amounts of, as yet ill-defined, tannins and flavan-3-ols(Reference Magna, Salamao and Vila23). The principal non-caffeine components in coffee comprise simpler phenolics, in particular chlorogenic acids, which are formed by esterification of several phenolic acids, most notably caffeic acid, and also quinic acid(Reference Ferruzzi21). Chlorogenic acid lactones, which may have independent effects on brain function, are also formed during roasting(Reference Farah, De Paulis and Trugo24).
Phenolics possess a number of properties that might underlie their purported health-promoting properties. Most pertinently, the notion that phenolics function as exogenous antioxidants has given way to an appreciation that phenolics such as polyphenols, and flavonoids in particular, interact with a number of cellular immune/stress signalling pathways(Reference Spencer25). In vitro and in/ex vivo research suggests that one common over-arching mechanism in this respect is modulation of multifarious elements of the protein kinase signalling pathways that connect extracellular stimuli, detected by membrane receptors, via signalling cascades (such as mitogen-activated protein kinase), to cellular changes and gene transcription. In terms of brain function, research suggests that the downstream effects of these flavonoid/signalling interactions include regulation of the expression of inflammatory factors such as cytokines, prostaglandins, and inducible NO synthase; neuroprotective effects, increased neurogenesis, and modulated vascular function via interactions with NO synthases(Reference Spencer, Vafeiadou and Williams26, Reference Williams and Spencer27).
To give one relevant example, the most abundant flavonoid in tea, EGCG, has been associated with a plethora of potentially beneficial effects in vitro/vivo, including in the face of neurodegenerative disorders(Reference Weisburger, Cadenas and Packer28, Reference Arab, Liu and Elashoff29) and dementia(Reference Rezai-Zadeh, Arendash and Hou30, Reference Lee, Yuk and Lee31). It is likely that the ability of EGCG to modulate signal transduction pathways, including protein kinases, and therefore cellular stress responses and survival(Reference Levites, Amit and Youdim32), underlies benefits in terms of inflammatory parameters(Reference Kim, Jeong and Lee33), mitochondrial function (for a review, see Mandel et al. (Reference Mandel, Avramovich-Tirosh and Reznichenko34)), protection against dopaminergic neurodegeneration(Reference Levites, Weinreb and Maor35) and up-regulation of a number of endogenous mammalian antioxidant enzymes(Reference Levites, Weinreb and Maor35–Reference Chou, Chu and Hsu37), with this latter effect abolished by inhibition of protein kinase C(Reference Levites, Weinreb and Maor35). Amelioration of oxidative damage following EGCG administration has also been associated with preserved cognitive function in rodents following a variety of neuronal insults(Reference Xu, Li and Chen38–Reference Unno, Takabayashi and Yoshida41). A further key mechanism of action of this flavonoid may be its potential to modulate blood flow, both in the periphery and the brain, via complex modulation of the expression of the three isoforms of NO synthase(Reference Lin and Lin42–Reference Lorenz, Wessler and Follmann45), which results in changes in vasodilation in the peripheral vasculature of animals(Reference Lorenz, Wessler and Follmann45–Reference Alvarez, Campos and Justiniano48), and in humans with coronary artery disease(Reference Widlansky, Hamburg and Anter49). Similar effects have also been observed in healthy non-smokers(Reference Shenouda and Vita50, Reference Alexopoulos, Vlachopoulos and Aznaouridis51) and smokers(Reference Nagaya, Yamamoto and Uematsu52) following the consumption of tea.
Studies exploring vascular effects of cocoa have produced similar results to those of EGCG. A number of observational studies have demonstrated a negative relationship between habitual cocoa intake and blood pressure(Reference Buijsse, Feskens and Kok53–Reference Djousse, Hopkins and North55). This is supported by cocoa intervention studies showing a reduction in blood pressure in healthy participants(Reference Fraga, Actis-Goretta and Ottaviani56, Reference Grassi, Lippi and Necozione57), as well as in participants who are overweight(Reference Berry, Davison and Coates58, Reference Faridi, Njike and Dutta59) and in those with evidence of glucose intolerance or cardiovascular risk factors(Reference Grassi, Desideri and Necozione60–Reference Grassi, Necozione and Lippi63). Others have failed to demonstrate these effects whether studying healthy participants or otherwise(Reference Monagas, Khan and Andres-Lacueva64–Reference Engler, Engler and Chen70) but meta-analyses have supported a case for anti-hypertensive effects of cocoa(Reference Taubert, Roesen and Lehmann71–Reference Desch, Kobler and Schmidt73). A number of studies have also explored effects of cocoa on flow-mediated dilation, an endothelium-dependent measure of vasodilation and a marker for NO availability. Significant increases in flow-mediated dilation have been demonstrated across a similar range of populations as studied in relation to blood pressure (many of whom suffer from endothelial dysfunction), whether studied chronically over periods of 8 d to 12 weeks(Reference Grassi, Desideri and Necozione60, Reference Heiss, Jahn and Taylor62, Reference Grassi, Necozione and Lippi63, Reference Davison, Coates and Buckley65, Reference Engler, Engler and Chen70), or acutely at 2 h post-administration(Reference Berry, Davison and Coates58, Reference Faridi, Njike and Dutta59, Reference Davison, Coates and Buckley65, Reference Heiss, Dejam and Kleinbongard66, Reference Njike, Faridi and Shuval68, Reference Heiss, Kleinbongard and Dejam74, Reference Vlachopoulos, Aznaouridis and Alexopoulos75). This acute effect of cocoa on flow-mediated dilation is correlated with increases in plasma NO species(Reference Heiss, Dejam and Kleinbongard66, Reference Heiss, Kleinbongard and Dejam74). Inverse relationships have also been observed between chocolate consumption and CHD(Reference Djousse, Hopkins and North55), cardiac mortality following acute myocardial infarction(Reference Janszky, Mukamal and Ljung76) and heart failure(Reference Mostofsky, Levitan and Wolk77, Reference Lewis, Prince and Zhu78). This latter study also showed that frequent chocolate consumers had a significantly lower prevalence of carotid atherosclerotic plaques. A significant inverse association between chocolate consumption and total stroke, cerebral infarction and haemorrhagic stroke has also been observed, with a stronger association for haemorrhagic stroke than cerebral infarction(Reference Larsson, Virtamo and Wolk79). Support for these findings comes from a recent systematic review of observational studies showing that cocoa consumption was associated with a 29 % reduction in risk of stroke, a relationship that survived adjustment for potential confounders(Reference Buitrago-Lopez, Sanderson and Johnson80). Studies in mice have also observed that epicatechin in isolation may aid the prevention of stroke damage through the Nrf2/HO1 (nuclear factor (erythroid-derived 2)-like 2/haeme oxygenase-1) pathway(Reference Shah, Li and Ahmad81).
Whilst the phenolic coffee chlorogenic acids have not benefitted from as much research as EGCG or cocoa flavan-3-ols, in vitro evidence does suggest that both chlorogenic acid and chlorogenic acid lactones interact with mitogen-activated protein kinase signalling pathways, bolstering antioxidant and neuroprotective properties(Reference Chu, Brown and Lyle82). In vivo these effects translate, for instance, into improved cognitive function in rodents in the face of a scopolamine challenge, with concomitant up-regulation of endogenous antioxidant systems, and additional cholinesterase inhibition in healthy rats(Reference Kwon, Lee and Kim83). Any interaction with signalling pathways could feasibly underlie the few direct demonstrations of altered brain function in humans by polyphenols, including those instances of modulated brain function described below with respect to EGCG, chlorogenic acid and cocoa flavan-3-ols.
The majority of research into the neuropharmacology of l-theanine has involved animal studies and therefore its mechanism of action in humans has yet to be wholly understood. l-Theanine is a derivative of, and has a chemical structure similar to that of, the neurotransmitter glutamic acid(Reference Nathan, Lu and Gray84). It crosses the blood–brain barrier via the leucine-preferring transport system(Reference Yokogoshi, Kobayashi and Mochizuki85) and has been shown to influence neurotransmitter concentrations. Levels of dopamine in the rat striatum were increased in a dose-dependent manner following l-theanine(Reference Yokogoshi, Kobayashi and Mochizuki85), as were those of γ-aminobutyric acid when l-theanine was administered intraperitoneally to mice(Reference Kimura and Murata86). Modulation of serotonin levels has also been observed. Yokogoshi et al. (Reference Yokogoshi, Kato, Sagesaka and Takiharamatsuura87) reported a dose-dependent reduction in levels of serotonin in the brain of spontaneously hypertensive and Wistar–Kyoto rats. A later study also demonstrated decreased global brain serotonin following l-theanine administration; however, region-specific increases were observed in the striatum, hippocampus and hypothalamus(Reference Yokogoshi, Kobayashi and Mochizuki85). A putative glutamate antagonist, l-theanine is capable of binding to the glutamate receptor subtypes AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid), kainite and NMDA (N-methyl-d-aspartate) glycine(Reference Kakuda, Nozawa and Sugimoto88). This antagonism of glutamate receptors by l-theanine has been linked to neuroprotection against cerebral ischaemia when administered intraperitoneally to mice(Reference Egashira, Hayakawa and Mishima89) and via intracerebroventricular injection to gerbils(Reference Kakuda, Yanase and Utsunomiya90).
Epidemiological evidence
Support for mechanistic studies described above, that imply a role for components other than caffeine in the effects of caffeinated products on behaviour, comes from epidemiological evidence showing differential effects on cognition as a function of caffeine source. In one of the first epidemiological studies of caffeine's effects on behaviour, Jarvis(Reference Jarvis91) showed a significant dose–response relationship between habitual caffeine consumption and performance of both simple and choice reaction time tasks after controlling for sociodemographic, lifestyle and health factors. These effects were more apparent in older, rather than younger, participants, which may indicate a cumulative effect of habitual caffeine consumption on cognitive performance, or may simply indicate a greater scope for improvement in older volunteers. Nevertheless, these data suggest that tolerance to the effects of caffeine is incomplete. The study also showed that the relationship across all ages was stronger for coffee than for tea. Given the dose–response relationship between caffeine and performance, this favourable correlation for coffee may simply be indicative of the higher caffeine levels present. It is also important to note that recent caffeine intake was not monitored in this study, so the possibility that the relationship reflects acute effects of caffeine/caffeinated beverage consumption cannot be ruled out. A number of other studies have provided evidence in support of a relationship between caffeine and cognitive performance, but no attempt was made to differentiate these effects as a function of caffeine source(Reference Hameleers, Van Boxtel and Hogervorst92–Reference Ritchie, Carriere and de Mendonca94). The latter of these studies found that caffeine consumption did not relate to dementia risk over a 4-year period. This is contrary to the findings of Maia & de Mendonca(Reference Maia and de Mendonca95) who carried out a retrospective study of caffeine consumption in fifty-four patients with probable Alzheimer's disease (AD) and fifty-four matched control participants, finding that caffeine intake in the 20 years preceding diagnosis of AD was inversely related with AD. A similar finding was reported by Lindsay et al. (Reference Lindsay, Laurin and Verreault96), but only coffee consumption was negatively related with AD; there were no effects for tea consumption, a finding supported by Eskelinen et al. (Reference Eskelinen, Ngandu and Tuomilehto97).
When considering the effects of coffee consumption on cognitive decline in older adults, Johnson-Kozlow et al. (Reference Johnson-Kozlow, Kritz-Silverstein and Barrett-Connor98) demonstrated a significant positive association between coffee consumption and cognitive performance. This effect was only observed in women. Interestingly, this study also explored the effects of decaffeinated coffee and found no such associations, in either sex, indicating the importance of caffeine to these effects. van Gelder et al. (Reference van Gelder, Buijsse and Tijhuis99) explored the effects of coffee consumption in men only, finding an inverse relationship between consumption and cognitive decline. Similar results were presented, in men and women, by Wu et al. (Reference Wu, Lan and Chen100). Conversely, in a twin study, Laitala et al. (Reference Laitala, Kaprio and Koskenvuo101) failed to show any relationship between coffee consumption in middle age and cognitive performance in old age (approximately 28 years later), after controlling for other, related, sociodemographic and health factors. One explanation for inconsistencies in results is provided by data from Corley et al. (Reference Corley, Jia and Kyle102) who observed a positive relationship between current caffeine consumption and cognitive function but only when caffeine was consumed as ground coffee. A number of studies do not control for the type of coffee consumed, which, when considering effects of non-caffeinated compounds present in coffee, is likely to have effects that are not controlled for even if an attempt is made to correct for caffeine level.
A similar issue exists within epidemiological studies of tea, whereby the brewing method and size of cup are rarely accounted for. Another factor that is likely to be of importance is the type of tea consumed. For instance, Shimbo et al. (Reference Shimbo, Nakamura and Shi103) showed that consumption of caffeine-containing drinks as a whole was actually related to a higher risk of mental ill health among females in Japan, but when looking specifically at green tea, no such effect was evident. More promisingly, Kuriyama et al. (Reference Kuriyama, Hozawa and Ohmori104) demonstrated a significant inverse relationship between green tea consumption and cognitive impairment, a relationship not observed for black or oolong tea, or coffee. However, in a Chinese community where black and oolong tea were the predominant tea types consumed, Ng et al. (Reference Ng, Feng and Niti105) observed a significant inverse relationship between total tea intake and cognitive impairment that was most apparent in relation to black and oolong tea. No association with coffee was observed. In a naturalistic study of tea, Bryan et al. (Reference Bryan, Tuckey and Einother106) explored the relationship between reported beverage consumption (using the diary method completed four times per d) and subjective work performance, cognitive–affective associates of work performance, mood, work-related strain and recovery. Tea consumption was shown to be negatively related to ratings of tiredness, but this effect was only apparent when sugar was not added to tea. A similar relationship between tea and work performance was observed, whereby higher tea consumption resulted in higher levels of performance but only if consumed without additives (milk, sugar, both). Non-caffeinated beverage consumption positively correlated with relaxation ratings; and evening consumption of non-caffeinated beverages showed a positive relationship with recovery (psychological detachment from work, relaxation and mastery of experiences that provide opportunities for learning).
The impact of flavonoid consumption from a range of caffeinated and non-caffeinated sources on cognitive function and decline has also been explored. The Paquid study explored the role of flavonoid consumption in the development of dementia in 1367 French participants over the age of 65 years with a 5-year follow-up(Reference Commenges, Scotet and Renaud107). This study demonstrated a significant inverse relationship between flavonoid intake and incidence of dementia. This dataset was further explored by Letenneur et al. (Reference Letenneur, Proust-Lima and Le Gouge108) who assessed mental state (Mini-Mental State Examination) and cognitive performance at baseline and at re-assessment four times over a 10-year period. A positive relationship between flavonoid intake and cognitive performance at baseline as well as an inverse relationship between intake and cognitive decline over the 10-year follow-up were demonstrated; however, the specific source of flavonoid was not explored in either analysis. Nurk et al. (Reference Nurk, Refsum and Drevon109) did, however, study the specific relationship of chocolate with cognitive performance, as well as the association of performance with flavonoid-rich tea and wine in approximately 2000 elderly men and women living in Norway. In a multiple adjusted model, they found that, in five out of six cognitive tests, the performance of consumers of chocolate (based on previous year's consumption) was significantly better than that of non-consumers and tea drinkers performed better in four out of six tests as compared with non-tea drinkers. Effects of tea were approximately linear but an increasing intake of chocolate was associated with a plateau in increased performance at 10 g/d.
Taken together, data from epidemiological studies would seem to indicate that effects on cognitive function and decline are more robust for coffee than tea, with data specifically examining effects of cocoa being too sparse to draw any conclusions. Of the nine studies reviewed above that explored effects of coffee, six showed a positive relationship with cognitive function(Reference Jarvis91, Reference Lindsay, Laurin and Verreault96–Reference Wu, Lan and Chen100), whereas only three of the six studies that explored effects of tea showed this relationship(Reference Jarvis91, Reference Kuriyama, Hozawa and Ohmori104, Reference Ng, Feng and Niti105), with one of these only showing effects of green tea(Reference Kuriyama, Hozawa and Ohmori104). Of those studies that included both tea and coffee, two observed positive findings of coffee but not tea(Reference Lindsay, Laurin and Verreault96, Reference Eskelinen, Ngandu and Tuomilehto97), one showed a relationship for both coffee and tea but the effect was stronger for coffee(Reference Jarvis91), and two showed effects of tea but not coffee(Reference Kuriyama, Hozawa and Ohmori104, Reference Ng, Feng and Niti105). One possible explanation for these inconsistencies is that studies that failed to show benefits of coffee were conducted in countries where consumption of coffee is very low and a similar interpretation can be applied to studies showing a lack of benefits of tea. One potential reason for greater neuroprotective effects of coffee than tea comes from studies showing a positive relationship between coffee and insulin sensitivity (for a review, see Muley et al. (Reference Muley, Muley and Shah110)), which may result from reductions in C-peptide levels seen following both caffeinated and decaffeinated coffee consumption(Reference Wu, Hankinson and Willett111, Reference Loopstra-Masters, Liese and Haffner112). However, the lack of a relationship between decaffeinated coffee and cognitive decline observed previously suggests that there are other important factors(Reference Johnson-Kozlow, Kritz-Silverstein and Barrett-Connor98).
Intervention studies
Coffee (Coffea robusta/arabica)
Despite being one of the most widely consumed beverages in the world, particularly in Western society, research into the effects of coffee per se is limited. A number of studies exist that have explored the effects of caffeinated coffee v. decaffeinated coffee(Reference Warburton113–Reference Smith, Brice and Nash116). However, although informative in terms of the effects of caffeine when consumed in combination with other concomitant compounds present in coffee, they do not allow us to assess the impact of these other components. Studies comparing tea and coffee have highlighted some effects of coffee not observed following tea, such as decreased sedation and faster choice reaction time(Reference Quinlan, Lane and Moore7, Reference Hindmarch, Rigney and Stanley117) (further details are provided below). A number of studies have also explored expectation in relation to the effects of caffeine in different vehicles and, in some cases, this has allowed for an interpretation of the contribution of caffeine, v. other components in coffee, to any effects observed. For instance, Andrews et al. (Reference Andrews, Blumenthal and Flaten118) demonstrated that decaffeinated coffee significantly lengthened startled blink onset latency as compared with caffeinated coffee, caffeinated juice and non-caffeinated juice. Although this is interpreted in terms of expectation effects and compensatory mechanisms, it is equally possible that this effect is indicative of components within the decaffeinated coffee exerting an action that is modulated (and therefore not observed) when consumed in combination with caffeine. This suggestion was explored by Cropley et al. (Reference Cropley, Croft and Silber119) (see Table 1 for further details) who examined the effects of chlorogenic acid (CGA) on mood and cognition in healthy, elderly participants. Acute effects of CGA were explored by manipulating their levels in a decaffeinated coffee vehicle, producing a regular-CGA decaffeinated condition (224 mg CGA, 5 mg caffeine) and a high-CGA decaffeinated condition (521 mg CGA, 11 mg caffeine) that were compared with each other as well as with caffeinated coffee (244 mg CGA, 167 mg caffeine) and placebo. Although the expected effects of caffeine were observed, very few effects of CGA were detected, with high CGA increasing alertness, and decreasing accuracy of negative bias indices (the difference between responses to sad and happy faces). However, as these effects were only observed in comparison with regular ‘CGA decaffeinated’ they allow us to isolate CGA as the compound responsible for these effects. The effects of caffeinated coffee in this study may be indicative of a synergistic effect between caffeine and other components within coffee, including CGA. However, without the inclusion of a caffeine-only arm it is not possible to ascertain this. Although findings from this pilot study combined with epidemiological data provide a suggestion that non-caffeine components within coffee may have some behavioural effects in human subjects, the evidence as it stands is far from convincing, particularly given the lack of effects of CGA on cognition. The observation that effects of high CGA are only apparent when compared with low CGA, as opposed to placebo, would indicate that at lower levels CGA is actually detrimental to those parameters affected. However, evidence that caffeinated and decaffeinated coffee, as well as CGA, are able to prevent apoptosis induced by H2O2 and increase NADPH:quinine oxidoreductase 1, which is regulated by the Nrf2 pathway, suggests that neuroprotection via antioxidant mechanisms is a strong possibility(Reference Kim, Lee and Shim120). CGA isolated from other sources has also demonstrated similar protective effects against oxidative stress(Reference Kim, Park and Jeon121, Reference Nakajima, Shimazawa and Mishima122). Further support for antioxidant properties of CGA comes from ex vivo work showing its ability to reduce malondialdehyde in the hippocampus and frontal cortex. Short-term behavioural benefits have also been demonstrated in a rodent scopolamine challenge model with associated acetylcholinesterase inhibition observed in vitro and ex vivo (Reference Kwon, Lee and Kim83). Further work directed at the effects of this compound in humans as well as research into other coffee components, such as the diterpenes kahweol and cafestol, which also have demonstrated antioxidant(Reference Hwang and Jeong123, Reference Lee and Jeong124) as well as anti-inflammatory properties(Reference Kim, Jung and Jeong125, Reference Kim, Jung and Lee126), is needed.
Randomised controlled trials assessing the effects of coffee on brain function
CGA, chlorogenic acid; decaf, decaffeinated; CA, caffeine; CRVAS, Caffeine Research Visual Analogue Scales; VVLT, Visual Verbal Learning Test; RVIP, rapid visual information processing; MMN, mismatch negativity; EFRT, Emotional Face Recognition Task; IT, inspection time; SCWT, Stroop Colour-Word Test; EEG, electroencephalogram; ERP, event-related potential; PD, post-dose; NBI, negative bias indices.
Tea (Camellia sinensis)
Perhaps surprisingly, the effects of tea on cognition and mood have received far greater attention than coffee, with a number of studies comparing the effects of tea with coffee and thus also providing some information on the effects of coffee. In one of several open-label studies Quinlan et al. (Reference Quinlan, Lane and Aspinall127) investigated the effects of 100 mg caffeine in differing vehicles (tea, coffee, water) on mood and physiological measures. Tea resulted in greater skin temperature increases as compared with coffee or water, suggesting greater vasodilation as a consequence of flavonoid content. This effect was not replicated when explored further by comparing tea containing 37·5 and 75 mg caffeine with coffee containing 75 and 150 mg, and hot water or no drink. The only significant effect of vehicle was greater lowering of sedation ratings following coffee than tea(Reference Quinlan, Lane and Moore7). Significant increases in critical flicker fusion following 100 mg caffeine in tea are not demonstrated following the same dose in coffee(Reference Hindmarch, Quinlan and Moore128), an effect replicated following 75 mg caffeine, along with a differential effect on choice reaction time in favour of coffee(Reference Hindmarch, Rigney and Stanley117). However, sedation ratings were decreased by caffeine irrespective of vehicle, representing a failure to replicate earlier findings(Reference Hindmarch, Quinlan and Moore128).
In a naturalistic study at high altitude, Scott et al. (Reference Scott, Rycroft and Aspen129) demonstrated that tea drinking was associated with lower fatigue ratings in regular tea drinkers. More recently De Bruin et al. (Reference De Bruin, Rowson and Van Buren130) described two randomised, controlled trials of black tea as compared with placebo tea. In the first of these, 1040 mg tea solids (containing 100 mg caffeine and 46 mg l-theanine) were shown to significantly improve intersensory attention and performance of a switch task. Participants also felt more alert and less calm following tea. In a second study, these effects on the switch task and alertness ratings were replicated following 760 mg tea solids (60 mg caffeine, 24 mg l-theanine). The lack of effects on ‘calm’ and intersensory attention in the second study may be the result of the slightly lower caffeine and l-theanine levels but differences in the times of testing and the tea employed make it difficult to assess this. The finding of reduced ‘calm’ in study 1 is in keeping with the stimulant effects of caffeine but is difficult to reconcile with the generally promoted relaxed alertness induced by tea.
Although the preceding studies are useful in demonstrating ecologically valid effects of tea drinking, they do not allow for ascertaining the contribution of non-caffeine components to these effects. The earlier studies also face the obvious problems associated with open-label trials, particularly when conducted by, or in collaboration with, tea manufacturers. In an attempt to overcome these issues, Steptoe et al. (Reference Steptoe, Gibson and Vounonvirta131) explored the effects of chronic tea administration (containing 72 mg caffeine) as compared with a caffeine-matched placebo in terms of effects on stress responsivity and recovery. The findings showed that tea was able to attenuate the effects of stress on platelet activation and cortisol levels, as well as increasing ‘relaxation’ ratings, relative to placebo. The use of a caffeinated placebo in this study allows for the conclusion that components within tea are able to modulate the effects of caffeine and/or produce direct ‘anti-stress’ effects; an assertion not possible when assessing caffeinated tea against a non-caffeinated placebo (as above). Support for a direct effect of non-caffeine components within tea comes from Henry & Stephens-Larson(Reference Henry and Stephens-Larson132) who found that hypertension in mice exposed to 3–5 months of psychosocial stress was significantly reduced by chronic administration of decaffeinated green tea as compared with water. With the exception of the latter study(Reference Steptoe, Gibson and Vounonvirta131), human research on tea to date has contributed little to our knowledge regarding its behavioural effects beyond that attributed to caffeine; studies demonstrating differences between tea and coffee are inconsistent and do not allow for assessment of which components are involved in any differential effects of these two caffeine sources. Two of the main components studied regarding potential bioactivity in tea are covered in the following sub-sections.
l-Theanine
l-Theanine (γ-glutamylethylamide) is found almost uniquely in tea and is one of the predominant amino acids present, making up between 1 and 2 % of the dry weight, resulting in about 25–60 mg l-theanine per 200 ml cup of tea. l-Theanine has been used in Asia for centuries in order to produce feelings of relaxation(Reference Heese, Jenkinson and Love133), but it is only recently that human studies exploring this effect have been conducted. A number of studies have demonstrated electroencephalogram (EEG) modulation following l-theanine administration. Kobayashi et al. (Reference Kobayashi, Nagato and Aoi134) found that 200 mg l-theanine led to increased alpha waves in the occipital and parietal regions of the brain within 40 min of ingestion when administered to resting participants, which the authors suggest is indicative of ‘relaxation’ without ‘drowsiness’. Nobre et al. (Reference Nobre, Rao and Owen135) assessed the effect of 50 mg l-theanine on resting EEG activity in healthy young volunteers and found significantly increased alpha activity as compared with placebo. However, the ability of a compound to modulate alpha activity does not unequivocally confirm anxiolytic properties, as similar findings have been demonstrated with non-anxiolytics(Reference Nathan, Lu and Gray84). l-Theanine has also been shown to reduce background alpha activity in the parieto-occipital region during performance of a demanding task whilst increasing attention-related (anticipatory) alpha activity(Reference Gomez-Ramirez, Higgins and Rycroft136). This reduction in background alpha activity was replicated in a similar study exploring effects during a visuospatial task but the effect on anticipatory alpha activity was not(Reference Gomez-Ramirez, Kelly and Montesi137). However, when considering the effects of 100 mg l-theanine and 50 mg caffeine both alone and in combination, Foxe et al. (Reference Foxe, Morie and Laud138) failed to find any effect of l-theanine on alpha band activity during performance of a sustained attention task when administered in isolation. One possible explanation for the differential results produced by Foxe et al. (Reference Foxe, Morie and Laud138) is the use of largely non-caffeine consumers and non-tea drinkers, with only two participants drinking tea and at a level of only one cup per d.
Physiological and subjective responses to l-theanine also suggest an anxiolytic effect. Kimura et al. (Reference Kimura, Ozeki and Juneja139) observed that heart rate, salivary IgA and the subjective perception of stress were lower following l-theanine as compared with placebo in response to a 20 min mental arithmetic task. A 250 mg dose of l-theanine has also been shown to slow reaction time on a visual probe task, indicating reduced anxiety in healthy young volunteers(Reference Rogers, Smith and Heatherley140). In contrast, Lu et al. (Reference Lu, Gray and Oliver141) were only able to demonstrate the relaxant effects of 200 mg l-theanine, as measured by the ‘tranquil-troubled’ item of the Bond–Lader visual analogue scales, during resting conditions. This effect was not apparent when anxiety was induced through the anticipation of electric shock. Despite the assertion that l-theanine produces ‘relaxation without drowsiness’, studies assessing the behavioural effects of l-theanine have demonstrated decrements in performance. Impairment has taken the form of a reduction in the number of serial sevens subtractions completed (as well as increasing subjective headache ratings)(Reference Haskell, Kennedy and Milne142) and slowed reaction time in response to an auditory task(Reference Gomez-Ramirez, Higgins and Rycroft136). However, most studies have failed to show any behavioural effects of l-theanine in isolation(Reference Gomez-Ramirez, Kelly and Montesi137, Reference Haskell, Kennedy and Milne142–Reference Owen, Parnell and De Bruin144), with only one study to date demonstrating a positive effect; Foxe et al. (Reference Foxe, Morie and Laud138) demonstrated a significant reduction in omission and commission errors on a sustained attention to response inhibition task following 100 mg l-theanine as compared with placebo.
It has been suggested that the most beneficial effects of l-theanine are observed when administered in combination with caffeine, due to a modulation of caffeine's stimulatory effects. For instance, Haskell et al. (Reference Haskell, Kennedy and Milne142) reported a number of effects following a combination of 250 mg l-theanine and 150 mg caffeine that were not apparent when each compound was administered alone. Improvements included increased speed on several tasks (simple reaction time, delayed word recognition, numeric working memory) and improved sentence verification accuracy. Participants also rated themselves as being more ‘alert’ and less ‘tired’. Combining 250 mg caffeine with 200 mg l-theanine resulted in the abolition of systolic and diastolic blood pressure increases seen following administration of caffeine alone(Reference Rogers, Smith and Heatherley140). A combination of 100 mg l-theanine and 50 mg caffeine was also capable of improving performance accuracy (as measured by increased hit rate and target discriminability) and reducing overall tonic (background) alpha power(Reference Kelly, Gomez-Ramirez and Montesi143), as well as simultaneously improving speed and accuracy on an attention-switching task at 60 min post-dose and performance on a memory task at 60 and 90 min post-dose(Reference Owen, Parnell and De Bruin144). Similar dose levels of l-theanine (97 mg) in combination with caffeine (40 mg) improved accuracy on a switch task with a trend towards faster responses on an intersensory task of attention(Reference Einother, Martens and Rycroft145). This combination also improved accuracy on a switch task as well as increasing subjective alertness and reducing task-induced fatigue(Reference Giesbrecht, Rycroft and Rowson146). Foxe et al. (Reference Foxe, Morie and Laud138) also found a significant reduction in omission and commission errors on the sustained attention to response inhibition task following 100 mg l-theanine combined with 50 mg caffeine, as well as attenuation in tonic alpha activity. However, no synergistic effect was observed with the effect of the combination being equivocal to that seen with caffeine alone.
With the exception of an anxiolytic effect, these findings demonstrate very few positive effects of l-theanine in isolation; in fact, the over-riding conclusion is that l-theanine impairs attention. Although positive outcomes have been indicated by those studies that have explored its effects in combination, the majority do not allow delineation of effects of this combination over and above those of caffeine alone. Of those studies that do allow this distinction(Reference Foxe, Morie and Laud138, Reference Rogers, Smith and Heatherley140, Reference Haskell, Kennedy and Milne142), there is a lack of interaction effects or findings are shown to be equivocal between those of the combination and those of caffeine in isolation(Reference Foxe, Morie and Laud138). Studies of l-theanine, either in isolation or in combination with caffeine, have also failed to explore the longer-term effects. Given the habitual nature of tea drinking, it is necessary for further work to be directed at ascertaining the effects of chronic consumption of l-theanine. The same observation can also be applied to EGCG research, covered in the following section.
Epigallocatechin gallate
Only two intervention studies to date have assessed the effects of EGCG on neural activity and cognitive performance in humans. In a randomised, double-blind, placebo-controlled cross-over study Wightman et al. (Reference Wightman, Haskell and Forster147) assessed CBF in the frontal cortex utilising near-IR spectroscopy. A group of twenty-seven healthy, young participants performed a range of cognitive tasks at baseline and then again 45 min after consumption of 135 or 270 mg pure EGCG. The results showed that, during the post-dose task performance period, the administration of 135 mg EGCG resulted in reduced CBF in the frontal cortex as indexed by significantly lower concentrations of both oxygenated and total Hb, in comparison with placebo. No effects were observed on cognitive performance. This reduction in CBF suggests that EGCG was exerting a vasoconstrictive effect, but in the absence of any modulation of behavioural parameters, the identification of a positive or negative influence of this effect on neurocognitive function is difficult. On the most simplistic level, optimal performance should require increased delivery of blood, and therefore metabolic substrates, during neuronal demand. However, it is also possible that this reduction in CBF with no negative impact on performance is indicative of increased neural efficiency. Scholey et al. (Reference Scholey, Downey and Ciorciari148) examined the effects of 300 mg Teavigo on electrical brain activity, as assessed through resting-state EEG, and mood in healthy human participants. EGCG increased cerebral activity in the form of alpha, beta and theta waves measured 180 min post-consumption. Ratings of ‘calmness’ were also increased at the same time and ‘stress’ ratings were significantly decreased. Cognitive and cardiovascular functioning effects from this study are, as yet, unpublished.
The paucity of data available relating to the behavioural effects of EGCG makes it impossible to assess its efficacy at the present time. Based upon human data to date, there is very little evidence of behavioural effects, with the exception of modulation of subjective ratings pertaining to ‘relaxation’ in one study(Reference Scholey, Downey and Ciorciari148). This effect aligns with the common perception of tea as relaxing but is contrary to previous data from tea as a whole showing decreased calmness(Reference De Bruin, Rowson and Van Buren130); although it should be noted that this reduction in calmness has not been replicated. The finding of increased cerebral activity(Reference Scholey, Downey and Ciorciari148) would seem to be contradictory to the observation of decreased CBF(Reference Wightman, Haskell and Forster147), but it should be noted that increased cerebral activity was observed following more than double the dose shown to decrease CBF and this may well represent opposing effects at different doses.
Randomised controlled trials of tea are summarised in Table 2.
Randomised controlled trials assessing the effects of tea on brain function
CA, caffeine; L-T, l-theanine; VAS, visual analogue scales; PD, post-dose; BP, blood pressure; HR, heart rate; EEG, electroencephalography; SART, Sustained Attention to Response Task; CRT, choice reaction time; VST, Visual Search Task; EMRT, Ergocentric Mental Rotation Task; SBP, systolic blood pressure; DBP, diastolic blood pressure; RT, reaction time; IWR, immediate word recall; SRT, simple reaction time; DV, digit vigilance; RVIP, rapid visual information processing; SWM, spatial working memory; LR, logical reasoning; NWM, numeric working memory; DWR, delayed word recall; WRec, word recognition; PRec, picture recognition; SV, sentence verification; S3, serial 3 subtractions; S7, serial 7 subtractions; CRVAS, Caffeine Research Visual Analogue Scales; STAI-Y1, State Trait Anxiety Inventory state component; s-IgA, salivary IgA; ECG, electrocardiography; HRV, heart rate variability; MAS, Manifest Anxiety Scale; ALP, Alprazolam; BDI-II, Beck Depression Inventory-II; AA, anticipatory anxiety; STAI-Y2, State Trait Anxiety Inventory trait component; BAI, Beck Anxiety Inventory; CFF, critical flicker fusion; DASS, Depression, Anxiety and Stress Scales; MAPS, Mood Alertness and Physical Symptoms questionnaire; EGCG, epigallocatechin 3-gallate; CV, cardiovascular; LORETA, low-resolution brain electromagnetic tomography; NIRS, near-IR spectroscopy.
Cocoa (Theobroma cacao L)
Research into effects of cocoa was first triggered by observations of a lack of blood pressure increases with age, and the virtual non-existence of hypertension in Kuna Indians living off the coast of Panama, who drank large quantities of cocoa on a daily basis. These effects were attributed to cocoa flavan-3-ols and a number of studies have demonstrated increased peripheral blood flow following flavan-3-ol-rich cocoa (for example, Engler et al. (Reference Engler, Engler and Chen70), Heiss et al. (Reference Heiss, Kleinbongard and Dejam74), Fisher et al. (Reference Fisher, Hughes and Gerhard-Herman149) and Flammer et al. (Reference Flammer, Hermann and Sudano150)). However, it is also possible that theobromine, a methylxanthine present in cocoa, may have a role to play in these effects. Crews et al. (Reference Crews, Harrison and Wright151) compared the effects of cocoa, in the form of 37 g dark chocolate (containing about 11 g cocoa (397 mg/g total proanthocyanidins)) plus 237 ml cocoa beverage (containing 11 g cocoa (357 mg/g total proanthocyanidins)), with placebo products over a 6-week intervention period in ninety cognitively intact elderly participants. Despite the older age of participants in this study and the long intervention period, effects on cognition were not apparent. One possible explanation for this lack of effects is the use of neuropsychological tasks that may not be sensitive enough to detect small changes in cognition. Alternatively, the lack of significant benefits may indicate that cognitive effects are only apparent acutely. This suggestion is supported by findings from Field et al. (Reference Field, Williams and Butler152) who demonstrated improvements to visual contrast sensitivity (as assessed by reading numbers that became progressively more similar in luminance to their background) and the time to detect motion direction, following dark chocolate (720 mg cocoa flavan-3-ols; 38 mg caffeine; 222 mg theobromine) as compared with white chocolate. This study also demonstrated improvements to a visual spatial memory task following dark chocolate. Although both studies demonstrate the ability of cocoa to improve cognition, it is difficult to draw firm conclusions regarding the mechanism of action of these effects.
Studies covering the effects of theobromine and cocoa flavan-3-ols that allow some delineation of effects of cocoa make up the next two sub-sections of the present review.
Theobromine
In the first of only four studies to date that have explored the effects of theobromine on behaviour, Mumford et al. (Reference Mumford, Evans and Kaminski153) failed to find group subjective effects of 560 mg of theobromine. However, four of the seven participants were able to discriminate this dose of theobromine from placebo. Smit et al. (Reference Smit, Gaffan and Rogers154) administered methylxanthine levels equivalent to that found in a 50 g bar of dark chocolate (250 mg theobromine, 19 mg caffeine) in two different encapsulated forms – cocoa powder, or theobromine + caffeine only. These active treatments were compared with placebo in terms of effects on mood and performance at 1 and 2 h post-intake. Both active treatments were observed to reduce simple reaction time (SRT) and energetic arousal. The only other significant effect on mood or performance was a significant time × treatment interaction, indicating a delayed effect of cocoa powder on hedonic tone, which was evident during the 2 h post-dose assessment as compared with the peak observed at 1 h following caffeine + theobromine only. This would suggest a modulation of the effects of the methylxanthines by other components when administered as cocoa powder, at least in terms of the time-course of effects. This may stem from behavioural effects of other components, or alteration in absorption time as a consequence of the food matrix. In a second study the effects of 60 g ‘chocolate’ with varying levels of methylxanthines (placebo = zero; low = 8 mg caffeine +100 mg theobromine; high = 20 mg caffeine +250 mg theobromine) were explored. Only the high-methylxanthine treatment was shown to reduce SRT, whereas both the high- and low-methylxanthine treatments improved accuracy of rapid visual information processing. No effects on mood were observed. It is important to note that the ‘chocolate’ in this study did not contain cocoa powder and therefore does not inform of the role, if any, of flavan-3-ols in these effects.
The differential contributions of caffeine and theobromine to the effects observed following their combination were explored by Mitchell et al. (Reference Mitchell, Slettenaar and Van der Meer155) who examined the effects of encapsulated theobromine (700 mg), caffeine (240 mg) and their combination on mood, performance and autonomic measures as compared with placebo at 1, 2 and 3 h post-ingestion. Caffeine increased alertness at each time point, as well as increasing contentment at 2 and 3 h post-ingestion and increasing interest in tasks and eagerness to do tasks. Theobromine decreased calmness at 3 h and increased interest in tasks. These effects on mood were not evident on performance measures and the combination of the two components appeared to eradicate any individual effects. As expected, caffeine increased blood pressure whilst decreasing heart rate, whereas theobromine decreased both systolic and diastolic blood pressure, and increased heart rate.
Cocoa flavan-3-ols
The majority of studies of cocoa have involved examining the effects of cocoa, containing methylxanthines, with enriched levels of flavan-3-ols. In one of the first studies to extend the findings of improved peripheral vasodilation seen following cocoa to CBF, Sorond et al. (Reference Sorond, Lipsitz and Hollenberg156) assessed the impact of short-term flavan-3-ol-rich cocoa consumption on elderly adults' CBF as assessed by transcranial Doppler. Middle cerebral artery mean blood flow velocity was assessed at baseline and following 7 d consumption of a dairy product-based flavan-3-ol-rich cocoa beverage containing 450 mg flavan-3-ols (with 18·3 mg caffeine and 336·5 mg theobromine). No effects were observed in this study, which may relate to effects of the known vasoconstrictor caffeine. Given that dietary control was simply limited to assessment taking place ≥ 2 h post-meal, it is possible that dietary intake of caffeine before participation resulted in a reduction of CBF as previously demonstrated (for example, Cameron et al. (Reference Cameron, Modell and Hariharan157), Lunt et al. (Reference Lunt, Ragab and Birch158), Sigmon et al. (Reference Sigmon, Herning and Better159) and Kennedy & Haskell(Reference Kennedy and Haskell160)). Alternatively, caffeine withdrawal may have led to increases in CBF as previously shown (for example, Addicott & Laurienti(Reference Addicott and Laurienti16) and Field et al. (Reference Field, Laurienti and Yen161)), making it difficult to demonstrate further superimposed increases. Another explanation for the lack of effects of flavan-3-ols on CBF is that these are only evinced when measured acutely.
Support for the assertion that CBF effects of cocoa flavan-3-ols may only be detectable acutely comes from Francis et al. (Reference Francis, Head and Morris162) who conducted a randomised, double-blind, balanced cross-over controlled trial where participants were assessed using functional MRI during a letter pair-switching task following 5 d consumption of flavan-3-ol-rich cocoa (172 mg) or a matched control (13 mg flavan-3-ol). Blood oxygenation level-dependent (BOLD) signal intensity was shown to be significantly increased by flavan-3-ol-rich cocoa as compared with the control. However, despite increased activation in brain areas relevant to the task, no significant effects on cognitive performance were observed. CBF responses to either a flavan-3-ol-rich cocoa drink (516 mg) or control (39 mg) at 2, 4 and 6 h post-consumption using arterial spin labelling also showed significant increases in grey matter CBF at 2 h following flavan-3-ol-rich cocoa, with a return to baseline levels achieved by 6 h post-consumption(Reference Francis, Head and Morris162). In the only study to date to detect an effect of cocoa flavan-3-ols on cognition (as compared with a methylxanthine-matched placebo), Scholey et al. (Reference Scholey, French and Morris163) reported significant improvements 90 min post-administration of flavan-3-ol-enriched cocoa at levels of 520 and 994 mg. Of the two enriched drinks, the 520 mg drink was more effective, leading to consistent improvements to serial threes performance and a robust attenuation of mental fatigue increases induced by performance of the demanding tasks. More recently, Camfield et al. (Reference Camfield, Scholey and Pipingas164) explored the effects of chronic cocoa flavan-3-ol consumption in the absence of an acute dose following low (0 mg), medium (250 mg) and high (500 mg) levels in a sample of 40- to 65-year-olds. Steady-state visually evoked potentials (SSVEP) were recorded using steady-state probe topography, a form of electrophysiological brain imaging that measures steady-state evoked potentials in response to task-irrelevant visual flickers, similar to event-related potentials but with greater temporal resolution. Assessments at baseline and following 30 d supplementation, whilst participants completed a spatial working memory task, showed a decrease in posterior-parietal SSVEP amplitude following the medium flavan-3-ol dose as compared with placebo. In light of a lack of effects on cognition, it is suggested by the authors that this decrease in amplitude may reflect an increase in neural efficiency, whereby participants are able to perform at the same level with reduced activation.
Despite mixed findings with regards cognition, the effects of cocoa on CBF are extremely promising and data from Field et al. (Reference Field, Williams and Butler152) and Scholey et al. (Reference Scholey, French and Morris163) show that cognitive benefits are possible, with the latter demonstrating these effects independently of methylxanthine content. Irrespective of behavioural benefits, the vascular effects of cocoa are certainly of interest. The potential of cocoa present in chocolate, a food that is viewed as palatable by the majority of the population, to provide benefits in preventing conditions such as stroke and vascular dementia is an exciting proposition worthy of further exploration.
Randomised controlled trials of cocoa are summarised in Table 3.
Randomised controlled trials assessing the effects of cocoa on brain function
CI, cognitively intact; MMSE, Mini Mental State Examination; CF, cocoa flavanols; TB, theobromine; CA, caffeine; BSRT, Buschke Selective Reminding Test; WMS-III, Wechsler Memory Scale-III; TMT, Trail Making Test; WAIS-III, Wechsler Adult Intelligence Scale-III; DSST, Digit Symbol Substitution Task; A-DACL, Activation-Deactivation Adjective Check List; SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; PD, post-dose; MX, methylxanthine; CHO, carbohydrate; CS, contrast sensitivity; WM, working memory; CRT, choice reaction time; SRT, simple reaction time; ERTT, Emotive Reaction Time Task; Q, questionnaire; CP, cocoa powder; VAS, visual analogue scales; RVIP, rapid visual information processing; SST, steady-state topography; SSVEP, steady-state visually evoked potential; SWM, spatial working memory; CBF, cerebral blood flow; ASL, arterial spin labelling; BOLD, blood oxygenation level-dependent response; S3, serial 3 subtractions; S7, serial 7 subtractions; STAI-Y1, State Trait Anxiety Inventory state component; TCD, transcranial Doppler; MCA, middle cerebral artery; VR, vasoreactivity; MFV, mean blood flow velocity; CVR, cerebral vasoreactivity; ABP, ambulatory blood pressure.
* Levels taken from Langer et al. (Reference Langer, Marshall and Day176).
Guaraná (Paullinia cupana)
The plant species guaraná originates from the central Amazonian Basin, and has a long history of local usage, initially as a stimulant by indigenous people(Reference Henman165), and more recently as a ubiquitous ingredient in Brazilian soft drinks. The putative stimulant properties are generally taken to reflect the presence of caffeine, which comprises 2·5–5 % of the extract's dry weight, although other purine alkaloids (theophylline and theobromine) are present in smaller quantities(Reference Weckerle, Stutz and Baumann166). Guaraná also has a high content of saponins, tannins(Reference Espinola, Dias and Mattei167) and catechins (see Carlson & Thompson(Reference Carlson and Thompson168)), and these may well underlie the demonstrated antioxidant properties of the plant(Reference Mattei, Dias and Espinola169).
Few studies to date have examined the cognitive effects of guaraná in human subjects (and some of these include guaraná co-administered with other potentially psychoactive components – see Table 4 for a summary). The first controlled study in healthy young adults administered guaraná for 3 d at a dose of 1000 mg then assessed effects acutely at 1 h post-dose. No effects were found on tests of psychomotor speed and accuracy, working memory, episodic memory, or the mosaic test(Reference Galduróz and Carlini170). In addition, no effects on anxiety or quality of sleep were observed. In a follow-up study the same doses and tasks were used to assess chronic (5 months) effects in an elderly population. Only one improvement was observed, a significant effect of guaraná on mosaic performance at 5 months(Reference Galduróz and Carlini171).
Randomised controlled trials assessing the effects of guaraná on brain function
GA, guaraná; CA, caffeine; DSST, Digit Symbol Substitution Task; STAI, State-Trait Anxiety Inventory; PD, post-dose; CDR, cognitive drug research; RVIP, rapid visual information processing; VAS, visual analogue scales; PG, Panax ginseng; SV, sentence verification; S3, serial 3 subtractions; S7, serial 7 subtractions; CDB, Cognitive Demand Battery.
More recently a randomised, double-blind, placebo-controlled, counterbalanced study examined the effects of a range of doses (37·5, 75, 150 and 300 mg) of guaraná (standardised to 11 % caffeine), over the course of 6 h(Reference Haskell, Kennedy and Wesnes172). Improvements to a ‘secondary memory’ factor following both 37·5 and 75 mg guaraná were observed, as well as increased self-rated ‘alertness’ following the highest dose and increases in self-rated ‘contentment’ following all doses of guaraná. The low level of caffeine in the doses of guaraná with the memory-enhancing effects (4·5 and 9 mg caffeine) and a declining pattern of effects with ascending dose in terms of ‘contentment’ suggest that these effects were not attributable to the caffeine content of the guaraná. These results also shared some similarities with another study examining the effects of the same 75 mg extract of guaraná, 200 mg of ginseng and their combination over the course of 6 h. Improvements to a ‘speed of attention’ factor, a serial sevens subtraction task and a sentence verification task were observed following 75 mg of guaraná. Although less pronounced, performance on the ‘secondary memory’ factor was again improved at 2·5 h post-dose(Reference Kennedy, Haskell and Wesnes173). A vitamin/mineral preparation supplemented with 222 mg guaraná also showed positive effects on performance of a rapid visual information processing task, coupled with an attenuation of mental fatigue associated with task performance(Reference Kennedy, Haskell and Robertson174).
The limited literature on behavioural effects of guaraná suggests that caffeine content alone does not account for any behavioural effects observed, with multi-dose exploration of guaraná suggesting that 75 mg (containing only 9 mg caffeine) was the optimum of those doses studied. Obviously, the lower doses of guaraná also contained lower doses of any other potential active ingredients, but since so little is known about the other constituents of guaraná – its effects so often being attributed to caffeine – it is difficult to assess what level of these constituents is the optimum, or indeed what level is present. It is also possible that at higher doses, the effects of the higher levels of caffeine are in some way masking the effects of the other active ingredients. This highlights the complexity of studying naturally concomitant psychoactives. As saponins and tannins have effects that may, directly or indirectly, affect behaviour, it is extremely likely that any effects of caffeine within guaraná are concomitant with effects of other components. It is also possible that guaraná simply has no behavioural effects as suggested by null findings(Reference Galduróz and Carlini170, Reference Galduróz and Carlini171); however, the lack of observed effects in these studies must be considered in light of a lack of effects of 25 mg caffeine, suggesting that the tasks employed may simply not be sensitive enough.
Summary and conclusions
A wealth of research spanning over 100 years has examined the effects of caffeine on cognition and mood in human subjects; however, research exploring the impact of other concomitant compounds is limited. This research is indicative of the prolific intake of caffeine, often being described as ‘from womb to tomb’. However, aside from consumption of caffeine pills, caffeine is not consumed in isolation, and is most often consumed alongside other plant components that have demonstrated some evidence of behavioural or behaviourally relevant activity. Data from epidemiological studies also illustrate differences in the relationship with cognition or cognitive decline as a function of source of caffeine, again suggesting that co-occurring phytochemicals may underlie any effects seen. In terms of intervention studies, less research has been directed towards the effects of coffee than tea or cocoa, which is unexpected given the popularity of this drink. This is also surprising, as a number of epidemiological studies demonstrate habitual coffee consumption, but not tea, to have a positive association with cognitive function(Reference Jarvis91, Reference Lindsay, Laurin and Verreault96, Reference Eskelinen, Ngandu and Tuomilehto97). Although tea(Reference Kuriyama, Hozawa and Ohmori104, Reference Ng, Feng and Niti105) has been shown to have a negative association with cognitive impairment, only coffee, not tea, has demonstrated an inverse relationship with AD(Reference Lindsay, Laurin and Verreault96, Reference Eskelinen, Ngandu and Tuomilehto97). A number of other health benefits have been demonstrated in relation to tea(Reference Widlansky, Hamburg and Anter49–Reference Nagaya, Yamamoto and Uematsu52), with similar inverse relationships demonstrated for cocoa consumption(Reference Djousse, Hopkins and North55, Reference Janszky, Mukamal and Ljung76–Reference Lewis, Prince and Zhu78), indicative of vascular improvements. These vasodilatory effects also extend to CBF, with negative relationships with stroke being seen for both cocoa(Reference Larsson, Virtamo and Wolk79, Reference Buitrago-Lopez, Sanderson and Johnson80) and tea(Reference Shen, Song and Ma175). Similarities in effects between tea and cocoa are unsurprising given the number of compounds shared between the two. Given these effects of tea on cerebral vasculature, and effects of EGCG in isolation on peripheral vasculature, it would seem a fair assumption that administration of EGCG would lead to increases in CBF. However, the only study to date to explore direct effects of EGCG on CBF demonstrated a decrease in total and oxygenated Hb. This is potentially expected given the myriad of effects on NO, but may also indicate a complex interaction of effects within components of tea, many of which have not been investigated in relation to cognitive effects in humans. It is also possible that the effects seen acutely following EGCG are not representative of the cumulative effect following long-term consumption.
In addition to potential problems surrounding task selection and task sensitivity, there are also a number of other issues within this field of research that the limited literature on guaraná, showing the greatest psychoactive effects at lower doses (containing very little caffeine), illustrates very well. Given the lack of previous human data, it is often difficult to select an appropriate dose and/or time-frame to test in and this can lead to an assumption that a nutritional intervention is behaviourally inactive. Referring to Tables 1–4, it is apparent that the over-riding focus of intervention studies in this field has been on acute effects of these interventions. The one exception to this, where some consideration has been given to chronic consumption, is in research on the effects of cocoa. However, even in this instance the effects of chronic consumption have often been evaluated following an acute dose that may have opposing effects to those seen chronically. Given that the majority of these compounds are consumed daily, there is a need for further exploration of their effects long term. Another factor that becomes apparent when referring to intervention studies in this field is that the majority of studies have focused on healthy young adults, who are presumably functioning near optimal performance and are unlikely to show improvements as a consequence of a nutritional intervention. In cases where older adults are considered, participants tend to be elderly, where complications caused by concomitant medication and underlying health issues are more common, and where the potential damage caused by ageing may be too great to be overcome by these compounds. Sample sizes in these studies also tend to be very low, which could result in null findings or, conversely, spurious significant findings. Another problem inherent here is that caffeine is likely to have a greater effect than its concomitant compounds and may therefore mask any effects exerted by them. A number of studies in this area fail to describe restrictions placed on caffeine consumption before participation and more importantly do not take into consideration the habitual caffeine consumption of participants, which given the potential impact of withdrawal and/or tolerance in relation to caffeine should be considered in all studies that involve caffeine administration. An additional concern when studying plants is the variability in levels of compounds, with the majority of extracts being standardised to caffeine content only.
Of the non-methylxanthine compounds covered in the present review, only the tea components EGCG and l-theanine have been considered in isolation. Data from these studies published to date have generally failed to show any effects on cognition, despite showing modulation of brain activity as indexed through EEG or changes in CBF. This may suggest that the physiological effects of these compounds are not evident on behaviour or that in isolation they are too subtle to be detected by the majority of cognitive tasks. Findings from Foxe et al. (Reference Foxe, Morie and Laud138) showing previously elusive improvements following l-theanine may provide support for the assertion that these effects are detectable with appropriate sensitive tasks. Similar findings from the cocoa flavan-3-ol literature demonstrate that this problem may not be restricted to studies that have considered the effects of compounds in isolation, with a number of studies failing to show any effects on cognition despite modulation of activity in brain areas relevant to the task employed(Reference Crews, Harrison and Wright151, Reference Francis, Head and Morris162, Reference Camfield, Scholey and Pipingas164). This may indicate that the tasks employed simply are not sensitive enough to detect these changes, that the participants are performing at ceiling and that other cohorts may show an effect, or that the intervention period is not long enough for beneficial effects to be seen. Positive effects of cocoa on cognition presented(Reference Field, Williams and Butler152, Reference Scholey, French and Morris163) would seem to dispute the impact of ceiling effects or length of intervention, having shown acute effects in healthy young adults. However, given the paucity of research in this area and the exploratory nature of the trials conducted it remains possible that these compounds simply do not affect behaviour.
Further research in this area is warranted, not least because of the extremely high level of intake of caffeinated beverages, which contain a number of other components that have been shown to have activity relevant to brain function, either in vitro or in vivo. Although it is important to assess the impact of these compounds in isolation, it is also important to be aware that this may not reflect the effect elicited when consumed with caffeine or as a whole. Each of these compounds is likely to have its own peak level in terms of effects on behaviour but this is also liable to be modulated by the combination with other substances. In terms of exploring ecologically valid effects, the most informative approach to studying this would seem to be that employed in the cocoa literature, whereby cocoa enriched with flavan-3-ols is compared with a control with low levels of flava-3-ols but matched for caffeine level.
Acknowledgements
This review received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Each author contributed to the writing of the manuscript. The authors declare no conflicts of interest.
