Effects of caffeine and glucose, alone and combined, on cognitive performance

Abstract

Objective

To study the effects of consuming caffeine and glucose, alone and combined, on cognitive performance.

Methods

Seventy-two healthy subjects (36 women; age range 18–25) were tested early in the morning, having fasted overnight. Using a double-blind, randomised design, subjects received one of the following beverages: water (150 ml); water plus 75 mg of caffeine; water plus 75 g of glucose; water plus and 75 mg of caffeine and 75 g of glucose. Attention, manual dexterity, visuo-spatial and frontal functions, memory (immediate, consolidation and working) and subjective state were all assessed.

Results

The combination of caffeine and glucose had beneficial effects on attention (sequential reaction time tasks) and on learning and consolidation of verbal memory, effects not being observed when either substance was administered alone. Caffeine only showed improvement in simple reaction time and glucose in simple and one sequential reaction time tasks and in the manual dexterity assembly task.

Conclusions

The results indicate that the synergistic effects of caffeine and glucose can benefit sustained attention and verbal memory, even with adequate levels of activation of the subjects. However, further studies are required, controlling for different levels of cognitive effort and also considering measurements of neural activity. Copyright © 2010 John Wiley & Sons, Ltd.

INTRODUCTION

Caffeine and glucose are part of our everyday lives, the former being taken for its stimulant effects, the latter for its restorative effects, and both for their pleasurable effects. Caffeine is present in coffee as well as in a wide variety of other drinks (i.e. tea, soft drinks). The quantity of caffeine in one cup of coffee is highly variable, the estimated range being 50–180 mg. Caffeine is rapidly absorbed (30–40 min) and has a half-life of between 3 and 6 h (Rogers, 2007). Glucose is the major source of energy for the brain and is essential for the normal functioning of the central nervous system (Sieber and Traystman, 1992). Numerous studies have observed that consumption of caffeine and/or glucose can benefit cognitive performance.

A great number of studies have been carried out into the effects of caffeine on human behaviour and cognition. Caffeine has beneficial effects on measures of reaction time and sustained attention tasks (Haskell et al., 2005; Kelemen and Creeley, 2001; Oei and Hartley, 2005; Smith, 2002, 2009; Smith et al., 2005). It has also been found to increase subjective alertness and mood and reduce fatigue (Adan et al., 2008; Hewlett and Smith, 2007; Smith, 2009; Smith et al., 2005). It improves motor-skill performance in tasks such as a simulated driving task (Brice and Smith, 2001) and handwriting (Tucha et al., 2006).

At low doses (less than 100 mg), caffeine does not always produce beneficial effects, these being more apparent at moderate and high doses (Childs and de Wit, 2006) and in situations with a deficit in activation, such as in the case of fatigued subjects (Brice and Smith, 2001; Hogervorst et al., 2008; Smith et al., 2005), and working at night or during sleep deprivation (Killgore et al., 2006; Wesensten et al., 2005). Moreover, sensitivity to the mood and performance-enhancing effects of caffeine is related to habitual intake levels; high-caffeine consumers are more likely to perceive broadly positive effects (Attwood et al., 2007; Ghisolfi et al., 2006; Hewlett and Smith, 2006; Rogers et al., 2003, 2005). Although the administration of caffeine does not improve either speed or precision in memory tasks (Boxtel et al., 2003; Childs and de Wit, 2006; Haskell et al., 2005; Hogervorst et al., 1998, 2008; Kelemen and Creeley, 2001; Oei and Hartley, 2005), in the first study to use functional magnetic resonance imaging (Koppelstaetter et al., 2008), it was recently found that it modulates neuronal activity during performance of a working memory task.

Evaluation of the cognitive effects of glucose has mainly focused on its learning and memory-improving action, with beneficial effects observed on a number of different parameters. It has been shown that glucose increases immediate and delayed recall in episodic memory tasks both in young and older healthy adults (Foster et al., 1998; Messier, 2004; Riby et al., 2004, 2006), although these facilitation effects were more pronounced in older subjects and mediated by each individual's gluco-regulatory efficiency. Cognitive facilitatory effects have also been found in other studies which evaluated declarative learning (Sünram-Lea et al., 2002a, b) and memory performance (Messier, 2004; Meikle et al., 2004; Riby et al., 2004, 2006; Stone et al., 2005), with effects lasting up to 24 hours after administration. However, there is little evidence that glucose can boost semantic memory retrieval (Messier, 2004; Riby et al., 2006). Some studies also obtained beneficial effects of glucose intake on non-mnemonic cognitive measures, such as rapid information processing and executive functions, that would appear to be associated with a relatively high cognitive load (Kennedy and Scholey, 2000; Meikle et al., 2004; Scholey et al., 2001, 2006). In general, lower doses of glucose (25 g) appear to be more effective in young adults while higher doses (50–75 g) are more often found to improve performance in older adults (Messier, 2004).

Although there is much evidence that glucose and caffeine improve various aspects of cognitive performance, very few studies have investigated the effects of the two substances in combination. Beneficial effects of combined caffeine and glucose have been found in sustained attention and working memory (Smit et al., 2006), and in situations of extended cognitive demand (Kennedy and Scholey, 2004). Moreover, better performance was observed in a selective attention task coupled with direct effects on visual cortical processing and decision-making assessed by event-related brain potentials (Rao et al., 2005). The studies evaluating the effects of energy drinks containing caffeine, glucose and other purportedly active ingredients have noted improvements in attention and declarative memory tasks without significant changes in mood (Scholey and Kennedy, 2004; Smit and Rogers, 2002). This has led to the suggestion that there may be a synergistic effect between caffeine and glucose on performance, as these results could not be explained by the effects of glucose or caffeine individually (Scholey and Kennedy, 2004).

This study aims to analyse the effect of consuming caffeine (75 mg) and glucose (75 g), alone and combined, on a battery of performance tasks and subjective state. Participants were healthy undergraduate students, non-consumers or low consumers of caffeine (<100 mg caffeine/day), who were evaluated first thing in the morning, having fasted overnight, under optimum activation conditions after a period of normal, restorative sleep.

METHODS

Participants

Seventy-two right-handed healthy undergraduate students (36 women; age range 18–25; 21.07 ± 1.70 year) were recruited from the University of Barcelona. Subjects with chronic disorders, nervous system disorders or history of mental illness were excluded, as well as habitual drinkers or smokers and those on medication. All participants had intermediate circadian typology, were good sleepers and reported an undisturbed sleep period of at least 6 h during the night prior to the experimental session. Daily caffeine consumption was controlled, with only non-consumers (N = 46; 63.9%) or low consumers (<100 mg/day; N = 26; 36.1%) included. Low consumers caffeine consumption included a maximum of one coffee per day and none of them were daily consumers of tea or cola beverages. The participants abstained from caffeine for a minimum of 18 h and fasted for at least 8 h prior to the experiment. Women had regular menstrual cycles and were studied in their follicular phase, when gastric emptying of glucose intake is slower and glycaemia lower (Brennan et al., 2009). The study was approved by the ethics committee of Hospital Clínic de Barcelona. Written consent was obtained from all participants, who were financially compensated for taking part in the study.

Caffeine and glucose measurements

Blood glucose levels (mg/dl) were determined using a Blood Glucose Sensor and disposable Blood Glucose Test Strips (Glucocard Gmeter Kit Menarini). Saliva samples for caffeine levels were obtained by asking participants to expectorate into a tube. Levels of caffeine (ng/ml) were determined by high performance liquid chromatography (HPLC) in the Scientific and Technical Department of the University of Barcelona, following the procedure described in Childs and de Wit (2006). Samples of blood and saliva were taken immediately on arrival at the laboratory in order to confirm compliance with 8 h fasting for food and 18 h of caffeine abstinence. A further saliva sample was taken 30 min after intake of the beverage and blood glucose levels were repeated at 30 and 60 min post-intake.

Performance tasks

The set of performance tasks used required 30–45 min to be completed and was applied to all the participants in the following order.

Rey Auditory Verbal Learning Memory Test (RAVLT; Schmidt, 1996)—Verbal declarative memory task (immediate and delayed recall): This consists of the presentation of a list of 15 words which subjects are asked to recall in five immediate trials (learning function). Another 15-word list is then read out which acts as interference before going on to request recall of the first list in two further trials. The first trial evaluates the effect of the interference and the second delayed recall (20 min after the final trial of learning). Delayed recall is regarded as an indicator of memory consolidation (Boxtel et al., 2003). The difference between number of words remembered in the last learning trial vs. the delayed recall was obtained as an estimation of forgetting.

Purdue-Pegboard (Lafayette Instruments Company, 1999)—Premotor test which measures fine motor coordination, dexterity and manipulation speed: It consists of taking a peg from a cup and inserting it into a hole as quickly as possible, firstly with the dominant hand (30 s), secondly with the non-dominant-hand (30 s) and thirdly, with both hands (30 s). Lastly, there is a task of assembling pegs, collars and washers, alternating the dominant hand with the non-dominant (60 s).

Benton Judgement of Line Orientation Test (Lezak et al., 2004)—Evaluates visuo-spatial function: The subject has to match lines to a model they are presented with. The computerised version was used.

Wisconsin Card Sorting Test (WCST, 2003): The computerised version was administered with the paradigm of 128 trials with the sequence colour, shape and number. This test measures prefrontal functions, assessing different dimensions: cognitive processing speed, concept formation, inhibition capacity and cognitive flexibility.

California Computerized Assessment Package (CalCAP; Miller, 1990)—Computerised reaction time program which measures sustained attention, reaction time and visual scanning speed: The abbreviated version with four tasks was used; simple reaction time (press a key as soon as anything appeared in the screen), choice reaction time (press a key as soon as a specific number appeared), sequential reaction time 1 (press a key when two of the same number appeared in sequence) and sequential reaction time 2 (press a key only when two numbers in increasing order appeared).

Digit Span of WAIS (Wechsler, 2001)— Test in which participants are read a progressively longer series of numbers ranging from two to nine digits and then asked to repeat the series forward (general attention) and backward (verbal working memory): The total score proposed as criteria in the test (forward + backward series remembered) was considered.

Lastly, a set of eight unipolar Visual Analogue Scales (VAS; Adan et al., 2008) was used as a subjective measure of activation and mood. Each scale is 100 mm long and asks the subjects to mark a cross at some point between ‘very little’ (left) and ‘very much’ (right), based on how they feel at the time of response (range of scoring 0–100), choosing from among alert, sad, tense, energetic, satisfied, tired, calm and sleepy. The VAS has two subscales: (a) concentrated, energetic, tired and sleepy for subjective activation level; (b) happy, calm, sad and tense for subjective effect. To obtain the score for the two subscales, the following formulas are used: subjective activation = [concentrated + energetic + 200 − (tired + sleepy)]/4; mood = [happy + calm + 200 − (sad + tense)]/4.

Experimental design

A first individual session was held to verify that subjects met inclusion criteria, explain what the experiment consisted of and the requirements of their participation and to obtain their written consent. The experiment started between 9 AM and 9:30 AM and after baseline measurements of glucose, saliva and subjective state, subjects received the treatment beverage. In a double-blind randomised design, participants received one of the four beverages (18 subjects, nine women per group). The individual beverages comprised: (1) placebo containing water (150 ml); (2) water plus 75 mg of caffeine; (3) water plus 75 g of glucose; (4) water plus 75 mg of caffeine and 75 g of glucose. There was no difference between groups in habitual caffeine consumption (χ²₃ = 3.792; p = 0.285). On the basis of the pharmacokinetics of caffeine and glucose, the performance recordings started 30 min after oral intake of the beverage. Subjective state was recorded again at 30 and 60 min post-intake and after completion of the battery of tasks.

Statistical analysis

Performance in each task was analysed by multivariate analysis of variance (MANOVA), when more than one measure was recorded, or by analysis of variance (ANOVA) when only one measure was recorded, with the beverage as between-subjects factor (placebo, caffeine, glucose and caffeine + glucose). A repeated measures analysis was carried out to study the development of the learning in the RALVT task and the subjective evaluations, as well as the change in the measurements of blood glucose and caffeine levels. Post-hoc comparisons were made with contrast options implemented in the analysis of variance. The partial eta squared (η) was used to measure the effect sizes considering that a partial eta squared of 0.01 was small, 0.04 medium and 0.1 large. All analyses were carried out with the SPSS package (version 15.0) and statistical tests were two-tailed with type I error set at 5%.

RESULTS

There was no difference between beverage groups in terms of mean age or number of hours of sleep, either in general or the night before the recordings (Table 1). Nor were there significant differences in baseline levels of blood glucose, caffeine and subjective state (activation and mood). Table 1 shows the descriptive statistics and results of the contrasts in different variables. The two post-intake glucose measurements were highly significant between groups, with values similar to baseline in the placebo and caffeine groups and significantly higher in the glucose and caffeine + glucose groups (p = 0.0001). The post-intake measurement of caffeine was similar to baseline in the placebo and glucose groups, while there was a significant increase in the caffeine and caffeine + glucose groups (p = 0.0001), the increase being greater in the caffeine group than in the caffeine + glucose group (p = 0.001).

Table 1. Descriptive statistics (mean and standard error) for age, habitual sleep and sleep pre-test (night before recordings) in hours and minutes for each beverage group. Recordings of glucose (mg/dl), caffeine (ng/ml) and subjective activation and mood (visual analogue scales from 0–100) performed during the experimental sessions (pre10 = 10 min before beverage; post30 and post60 = 30 and 60 min after beverage; final = measure on completion of the performance tasks)

	Placebo	Caffeine	Glucose	Caffeine + glucose	F(3,68); p
Age	21.00 ± 0.44	20.72 ± 0.45	21.55 ± 0.40	21.00 ± 0.28	0.775; 0.523
Habitual sleep	7:34 ± 0:07	7:55 ± 0:09	7:46 ± 0:07	7:53 ± 0:08	1.284; 0.287
Sleep pre-test	7:07 ± 0:08	7:22 ± 0:09	7:11 ± 0:07	7:19 ± 0:06	0.706; 0.552
Glucose pre10	76.50 ± 2.12	75.39 ± 1.28	80.67 ± 3,24	79.56 ± 1.18	1.373; 0.258
Glucose post30	75.93 ± 2.30	74.69 ± 2.39	157.67 ± 12.36	141.20 ± 8.01	32.958; 0.0001
Glucose post60	71.06 ± 2.05	70.78 ± 1.48	128.39 ± 13.14	124.22 ± 7.81	17.090; 0.0001
Caffeine pre10	4.41 ± 0.84	5.76 ± 1.36	5.37 ± 1.01	6.58 ± 1.08	0.680; 0.567
Caffeine post30	4.10 ± 0.82	267.85 ± 29.23	5.38 ± 0.92	174.28 ± 24.94	46.153; 0.0001
Activation pre10	57.59 ± 3.72	53.78 ± 3.83	56.12 ± 2.49	59.20 ± 2.98	0.487; 0.693
Activation post30	62.21 ± 3.52	62.61 ± 4.18	63.04 ± 3.68	61.04 ± 2.39	0.059; 0.981
Activation post60	58.46 ± 3.85	64.57 ± 4.55	63.02 ± 1.99	66.18 ± 3.14	0.896; 0.448
Activation final	66.89 ± 3.56	64.45 ± 4.53	70.64 ± 2.73	67.17 ± 3.01	0.520; 0.670
Mood pre10	76.81 ± 2.11	76.14 ± 3.05	74.82 ± 1.73	76.93 ± 2.91	0.149; 0.930
Mood post30	77.47 ± 2.57	75.96 ± 3.19	71.68 ± 2.31	75.19 ± 3.26	0.730; 0.536
Mood post60	74.21 ± 2.60	74.32 ± 3.04	75.07 ± 1.82	75.12 ± 2.75	0.035; 0.991
Mood final	77.68 ± 2.32	78.31 ± 3.16	75.87 ± 1.88	77.79 ± 2.64	0.174; 0.914

Table 2 shows the descriptions for the performance of the four CalCAP tasks for the beverage groups. A main effect of beverage was obtained for the simple reaction time task, considering mean response time (F_(3,68) = 3.280; p = 0.026; η = 0.128) z-score (F_(3,68) = 3.494; p = 0.020; η = 0.135) and percentile (F_(3,68) = 2.759; p = 0.049; η = 0.110). In all cases, the post-hoc comparisons showed poorer performance in the water group with respect to the other three (0.006 > p < 0.044), with no significant differences between the caffeine, glucose and caffeine + glucose groups (Table 2). No main effect of beverage was obtained in the choice reaction time task (F_(3,68) < 0.804; p > 0.496), the sequential reaction time 1 task (F_(3,68) < 2.099; p > 0.110) or the sequential reaction time 2 task (F_(3,68) < 2.291; p < 0.085). However, the post-hoc contrasts in the sequential tasks revealed differences between the beverage groups in execution. In the sequential reaction time 1 task, the placebo group had a greater mean response time than the glucose group (p = 0.042) and the caffeine + glucose group (0.028), in addition to a lower z-score (placebo-glucose: p = 0.052; placebo-caffeine + glucose: p = 0.026) and lower percentile (placebo-glucose: p = 0.044; placebo-caffeine + glucose: p = 0.048) (Table 2). In the sequential reaction time 2 task, execution was poorer in the placebo group than in the caffeine + glucose group, with a greater mean response time (p = 0.016), a lower z-score (p = 0.024) and a lower percentile (p = 0.019) (Table 2).

Table 2. Performance data (mean and standard error) for four tasks of abbreviated California Computerized Assessment Package (CalCAP). Reaction time in milliseconds

Tasks	Placebo	Caffeine	Glucose	Caffeine + glucose
Simple
Reaction time	389.35 ± 18.50	320.16 ± 17.98	318.16 ± 17.04	336.44 ± 14.02
z score	−0.699 ± 0.204	0.112 ± 0.110	0.080 ± 0.108	−0.080 ± 0.142
Percentile	36.70 ± 4.95	54.56 ± 3.88	53.33 ± 3.45	47.78 ± 4.81
Choice
Reaction time	401.28 ± 10.58	391.67 ± 10.77	381.39 ± 9.95	390.11 ± 8.74
z score	0.187 ± 0.254	0.574 ± 0.229	0.674 ± 0.150	0.448 ± 0.216
Percentile	59.77 ± 6.51	67.17 ± 6.96	71.33 ± 4.51	63.11 ± 6.51
Sequential 1
Reaction time	487.72 ± 17.75	468.83 ± 13.99	441.77 ± 14.08	431.28 ± 18.83
z score	0.636 ± 0.189	0.850 ± 0.144	1.132 ± 0.154	1.245 ± 0.195
Percentile	68.61 ± 5.07	77.56 ± 4.52	83.33 ± 3.42	83.06 ± 4.24
Sequential 2
Reaction time	581.41 ± 26.52	525.61 ± 22.42	549.83 ± 24.47	498.22 ± 20.71
z score	0.262 ± 0.244	0.782 ± 0.200	0.537 ± 0.214	0.948 ± 0.175
Percentile	58.00 ± 6.50	74.06 ± 5.74	65.78 ± 6.11	77.67 ± 4.54

Table 3 shows the descriptions for performance in the Purdue Pegboard, Benton Judgement of Line Orientation Test and RAVLT (total words remembered in learning trials, interference and forgetting). Significant differences between beverage groups were only found in the Purdue Pegboard assembly task (F_(3,68) = 2.938; p = 0.039; η = 0.116). Post-hoc comparisons between groups uncovered differences between the glucose and the placebo (p < 0.023) and the caffeine (p < 0.008), performance being better in the glucose group. The MANOVA performed for the Benton test measures (adjusted total score, errors and reaction time) only revealed significant differences for the beverage in reaction time (F_(3,68) = 2.912; p = 0.041; η = 0.115). Post-hoc comparisons between groups show a shorter execution time in the glucose and caffeine + glucose groups, the differences being significant with respect to the caffeine group (p = 0.009 and p = 0.017, respectively).

Table 3. Performance data (mean and standard error) for each beverage group in Purdue Pegboard, Benton Judgement of Line Orientation Test (total score with correction for gender) and Rey Auditory Verbal Learning Memory Test (RAVLT)

Tasks	Placebo	Caffeine	Glucose	Caffeine + glucose
Purdue Pegboard
Dominant hand	14.66 ± 0.44	14.39 ± 0.42	14.62 ± 0.44	15.34 ± 0.45
Non-dominant hand	13.48 ± 0.39	14.12 ± 0.36	14.12 ± 0.39	14.51 ± 0.39
Both hands	22.44 ± 0.68	23.59 ± 0.79	23.44 ± 0.77	23.33 ± 0.78
Assembly	35.75 ± 1.51	34.86 ± 1.44	40.74 ± 1.32	37.85 ± 1.50
Benton
Total score	26.50 ± 0.79	27.19 ± 0.73	26.78 ± 0.77	26.65 ± 0.85
Errors	4.51 ± 0.90	3.72 ± 0.73	4.19 ± 0.85	5.30 ± 0.90
Reaction time	5417.22 ± 474.36	6344.16 ± 481.55	4618.73 ± 333.71	4769.95 ± 424.71
RAVLT
Total words learning	56.50 ± 1.35	58.78 ± 1.32	56.89 ± 1.28	60.56 ± 1.14
Interference	1.28 ± 0.32	1.22 ± 0.42	1.61 ± 0.39	1.17 ± 0.34
Forgetting	1.55 ± 0.35	0.84 ± 0.33	1.66 ± 0.42	1.16 ± 0.30

There were no significant differences between beverage groups in the WAIS Digit Span subtest (F_(3,68) = 0.772; p = 0.514; η = 0.033). The total score in this working memory task was similar regardless of the beverage administered (placebo: 18.50 ± 0.81; caffeine: 18.11 ± 0.82; glucose: 17.22 ± 0.61 and caffeine + glucose: 17.00 ± 0.79). Nor was a significant main effect of beverage found in any estimation of execution in the WCST (F_(3,68) < 0.736; p > 0.536), either when considering the quality (number of trials, correct trials, perseverative errors, learning) or the speed, in this frontal-function task.

There were no significant differences between beverage groups in the RAVLT learning when considering the function of the five memory trials (F_(3,68) = 1.181; p = 0.324; η = 0.050). In all cases, the number of words remembered followed an ascending pattern from the first to the last trial. Differences were found, however, when memory was analysed for the groups in each of the five trials individually, with differences found in the second to last (RALVT 4: F_(3,68) = 4.218; p = 0.009; η = 0.157) and the last (RALVT 5: F_(3,68) = 6.125; p = 0.001; η = 0.213). In both cases, the mean number of words remembered is higher in the caffeine + glucose group than in the other three (0.05 < p > 0.001), with no difference being observed between the placebo, caffeine and glucose groups (Figure 1). Lastly, there were no differences between beverage groups in the RAVLT when analysing the total words remembered in learning trials (F_(3,68) = 2.155; p = 0.101), the interference effect (F_(3,68) = 0.295; p = 0.829) and forgetting (F_(3,68) = 1.14; p = 0.338). However, the post-hoc contrasts in the total words remembered revealed differences between the beverage groups, the caffeine + glucose group had greater recall than the placebo (p = 0.028) and the glucose group (0.046) (Table 3). Finally, differences were found in the memory consolidation (F_(3,68) = 3.321; p = 0.025; η = 0.128) that was also greater in the caffeine + glucose group than the placebo group (p = 0.023) or the glucose group (0.004) (Figure 1).

Figure 1. Number of words remembered in Rey Auditory Verbal Learning Memory Test (RAVLT) for each trial of immediate recall (A1-A5) and for delayed recall or memory consolidation recorded 20 min after (20')

Although a significant linear variation was obtained for the four temporal recordings taken of subjective activation (F_(3,68) = 31.93; p = 0.0001; η = 0.320), there were no differences between beverage groups (F_(3,68) = 1.834; p = 0.149). Regardless of the beverage, an increase in subjective activation was observed from baseline to the final recording at the end of the experiment. Moreover, none of the inter-recording analyses found differences between any of the groups (p > 0.125). The evaluations of mood showed significant quadratic evolution for the four temporal recordings taken (F_(3,68) = 5.505; p = 0.022; η = 0.075), although the changes were similar in all the beverage groups (F_(3,68) = 1.272; p = 0.291). Once again, none of the inter-recording analyses found differences between beverage groups (p > 0.455). The evaluations of mood fell between the first and third recordings, only to go up again in the fourth and last recording, regardless of the beverage consumed.

DISCUSSION

We found a differential pattern between the three beverages with active substance them of beneficial effects on cognitive performance, although the effects were modest compared to placebo. Only very few studies have specifically set out to evaluate the synergy between caffeine and glucose, and none has considered such an extensive battery of tasks as has been used in our study. Moreover, the recordings were taken under conditions with no arousal deficit, since different variables which could have affected performance and biased the results, such as sleep quality and duration and the circadian typology of the participants, were controlled. The doses selected in our study (low for caffeine and high in the case of glucose) correspond with the approximate contents of two soft drinks containing caffeine and had not been used in previous studies.

The administration of caffeine only provided a beneficial effect in the execution of the simple reaction time task compared to placebo. Moreover, in the Benton visuo-spatial task, the speed of execution was similar to placebo and lower than the groups which consumed glucose and caffeine + glucose. This is consistent with the earlier literature; improvements in performance are particularly observed in activation deficit situations (Brice and Smith, 2001; Smith et al., 2005; Wesensten et al., 2005; Killgore et al., 2006) and in moderate or high caffeine consumers (Attwood et al., 2007; Ghisolfi et al., 2006; Hewlett and Smith, 2006). Thus, we observed that 75 mg of caffeine had a minimal impact upon behavioural performance in habitual non-consumers or low consumers (Rogers et al., 2003, 2005) recorded first thing in the morning under adequate activation conditions. Future studies, however, should investigate whether or not caffeine increases neuronal activity during the performance of several cognitive tasks in the absence of effects on execution estimations, as was observed for the working memory (Koppelstaetter et al., 2008).

Compared to placebo, the consumption of glucose had beneficial effects on execution in the simple reaction time tasks and sequential reaction time 1 task (press a key when two of the same number appeared in sequence) and the Purdue Pegboard test assembly task. Studies which have investigated the effects of glucose consumption on execution have not tended to incorporate manual dexterity tasks and attention tasks of reaction time. Our results signal the need for future studies which determine the dose-response ratio in the execution of such tasks, since it is possible that the beneficial effects only occur at high doses such as that selected in this study. In contrast, consumption of glucose was not found to have any beneficial effects on the learning and memory tasks (Digit Span and RAVLT) or on executive function (WCST). Many previous studies have obtained beneficial effects from the consumption of glucose on verbal memory tasks such as RAVLT or more complex tasks (Meikle et al., 2004; Messier, 2004; Riby et al., 2004, 2006; Stone et al., 2005; Sünram-Lea et al., 2002a, b). This discrepancy may be due to the dose of glucose we used, since all the significant data with young subjects were obtained at low doses (25 g).

The combined administration of glucose and caffeine, compared to placebo, has been shown to improve execution in all reaction time tasks except choice, and learning (immediate memory) and memory consolidation of RAVLT. In accordance with previous studies, results indicate that the synergistic effects of caffeine and glucose can benefit high-demand sustained attention processes (Kennedy and Scholey, 2004; Rao et al., 2005; Smit et al., 2006) such as the sequential 2 task, the most complex and last to be executed of the CalCAP, and verbal memory (Scholey and Kennedy, 2004; Smit and Rogers, 2002). This does not occur with the consumption of caffeine or glucose alone, suggesting that combined caffeine and glucose may be a more effective cognition enhancer for this type of task (Scholey and Kennedy, 2004), even under circumstances of adequate levels of activation in young subjects. Further research is needed to evaluate the potential pharmacokinetic interactions of the co-administration of caffeine and glucose and their impact on behaviour. While it is known that the co-administration increases intestinal glucose absorption (Van Nienwenhoven et al., 2000), our data suggest that caffeine absorption could also be different when taken alone or combined with glucose, although there is not any publication to this regard.

Subjective state (activation and mood) was independent of the type of beverage administered, both when considering the change over the course of the experiment and for each recording taken. This result is consistent with the only previous study to have evaluated the effects of glucose in self-assessments of activation and stress (Riby et al., 2004) and with two studies in which caffeine and glucose were administered at lower doses (Kennedy and Scholey, 2004; Smit and Rogers, 2002). However, it contrasts many studies which have observed beneficial effects on activation with the administration of caffeine at low doses (Adan et al., 2008; Brice and Smith, 2001; Hewlett and Smith, 2007; Smith, 2002, 2009; Smith et al., 2005). The wide dispersion within groups in self-assessments and the non-naturalistic administration may be the factors responsible for our data. It should be emphasised that the administration of 75 mg of caffeine and 75 g of glucose had no negative effects on mood (tension, nervousness, anxiety, etc.), as might be expected after consuming two soft-drinks containing caffeine very quickly. Nevertheless, further study should investigate the effects of caffeine, glucose and their combination in extreme-time periods or during the post-lunch, and also in subjects with high fatigue levels or sleep deprivation and in different pathological conditions. Under such conditions, greater benefits might be expected, both in performance and subjective state, as is already well established with the administration of caffeine (Brice and Smith, 2001; Killgore et al., 2006; Rogers, 2007; Smith et al., 2005; Wesensten et al., 2005) and, to a lesser extent, glucose (Messier, 2004; Riby et al., 2004, 2006; Stone and Seidman, 2008), although to date, the possibility of such benefits with caffeine and glucose combined has yet to be explored.

Finally, we should mention several limitations in our study. Firstly, we used a single dose in each condition, and it is therefore difficult to determine the real degree of the synergistic effect between caffeine and glucose. Secondly, the results of this study can only be generalised to young healthy subjects fasted for food and recorded early in the morning, and under non-naturalistic conditions. In addition, we cannot rule out the fact that other factors, such as those intrinsic to the beverage (e.g. sensory attributes, smell and taste) or of a psychological nature (e.g. expectancy), may play a significant role in mediating the responses. Moreover, further studies are required to investigate the effects of caffeine and glucose, alone and in combination, with repeated doses, controlling for different levels of cognitive effort, and also considering measures of neural activity such as event-related brain potentials or functional neuroimaging.