We investigated the acute effects of caffeine on neurocognition and EEG FC in healthy adults. The major findings were as follows: 1) the property of EEG FC was reorganized toward a more efficient network after coffee consumption relative to baseline, 2) Performance in the Digit Span tests and Trail Making Test Part B was improved after coffee consumption, and 3) improved performance in the Trail Making Test Part B after coffee consumption was correlated with changes in graph measures reflecting a shift toward efficient network property.
The human brain is considered to be a large-scale complex network and has properties of efficient small-world networks that refer to locally well-connected clusters and efficient global connections23,26. The properties of small-world networks are known to enable higher rates of information processing and learning with a lower cost than those of random networks27. In terms of these network properties, changes in cognitive functional status or cognitive capacity might be associated with changes in the configuration of brain functional networks26. Indeed, there are several lines of evidence suggesting that loss of the small-world configuration might be implicated in the cognitive deficits observed in various brain disorders, such as Alzheimer’s disease, schizophrenia, and brain tumors28-30. Based on the aforementioned notion, our findings of changes in graph measures to high clustering and short path length after coffee consumption suggest that functional reorganization toward more efficient network properties might be a mechanism underlying the enhancement of cognitive function observed after coffee consumption.
The mechanism underlying the shift in FC toward efficient network properties after coffee consumption remains to be determined. It is believed that caffeine’s effect on cognition is associated with the blockade of the inhibitory properties of endogenous adenosine (particularly at A1 and A2A receptors), resulting in increased dopamine, norepinephrine, and glutamate release4. In addition, the cardiostimulatory effects of caffeine are considered to result from interactions with both adenosine and phosphodiesterase31. The caffeine-induced increases in dopamine and glutamate concentrations, coupled with phosphodiesterase inhibition, could be considered as a crucial mechanism underlying the net increase in the central nervous system and cardiovascular activity. Based on the actions of caffeine, it is plausible that the stimulatory effects of caffeine might directly lead to the reorganization of network properties toward a state of increased efficiency. Further studies are needed to unveil the mechanisms underlying the changes in network properties after coffee consumption.
Our findings of improved performance in the Digit Span Forward test suggest that attentional function could be enhanced by coffee consumption, which is in line with previous observations that coffee consumption has beneficial effects on attention7,32-34. In addition, our findings of greater performance in the Digit Span Backward test17 after coffee consumption may support findings from previous studies that have shown the role of coffee in improving working memory35-37. A recent functional magnetic resonance imaging (fMRI) study found that the alerting network, known as being responsible for maintaining an alert state throughout task performance, recruited a distributed network of brain regions, primarily the thalamus and bilateral fronto-parietal regions38,39. Based on these fMRI findings, our results that FC changes after coffee consumption are mainly observed in the fronto-centro-parietal regions imply that improvement of attentional function might be derived from activation of the alerting network.
We also found that performance in the Trail Making Test Part B was improved after coffee consumption and that the degree of improvement of the test was correlated with the changes in graph measures reflecting a shift toward more efficient network properties. It is well known that the Trail Making Test Part B is a representative tool for evaluating the ability of executive function responsible for psychomotor speed, visuospatial searching, target-directed motor tracking, and set-shifting40. Therefore, our findings further support previous studies that showed the beneficial effects of caffeine on executive function and psychomotor speed4,5,41. Performance of executive controls requires activation of widespread prefrontal regions in concert with the anterior cingulate cortex4,42,43. These brain areas have been shown to be upregulated by caffeine37,44, supporting the stimulatory effects of caffeine on executive function. Moreover, dopamine was found to be a critical neurotransmitter for supporting executive function in these areas45. Given that dopamine concentrations can be increased by caffeine through blockade of the inhibitory properties of adenosine, caffeine may enhance executive function through the interaction of dopaminergic pathways with anterior cingulate and prefrontal cortical regions.
Our findings of the relationship between improved executive function and graph measures suggest that changing network topology toward more efficient network properties might be a crucial mechanism underlying the beneficial effects of coffee on executive function. Our speculation is supported by prior studies using fMRI that found increases in FC in multiple brain regions during the performance of the Trail Making Test Part B46-49. In addition, the aforementioned relationships were mainly observed in the alpha band, which is in accordance with a recent study showing that executive functions have a positive relationship with alpha coherence between regions of the right and left hemispheres50. Taken together, our findings support those of previous studies that coffee may enhance the FC responsible for performance on executive function, especially in the alpha band. Meanwhile, we did not find any changes in nodal graph measures after coffee consumption in the alpha band. The changes in global network properties without any region-specific changes in the alpha band suggest that coffee consumption might further enhance the improvement of physiological network efficiency responsible for activating cognitive function across the whole brain, rather than causing changes in the network properties of specific localized areas. Given the involvement of the dopaminergic pathways in executive function45, another plausible explanation is that our findings of changes in cortico-cortical network properties may not fully reflect the interactions of subcortical dopaminergic pathways with the cortical areas responsible for executive function.
We did not find any relationship between performance in Digit Span tests and graph measures. It is not fully understood why the results of the Digit Span tests, which reflect the function of attention/working memory, were not correlated with graph measures. However, it is plausible that there was a ceiling effect in the performance of the Digit Span tests in our cognitively normal population. In addition, given that attention/working memory is associated with not only cortical function but also various subcortical neurotransmitter systems (e.g., basal forebrain cholinergic systems and dopaminergic systems), a FC analysis evaluating the cortico-cortical network using EEG might not be sufficient to reveal the mechanism underlying the attention/working memory function.
There are several limitations of the present study that should be considered when interpreting our results. First, our study population was relatively small, and was only composed of highly educated young adults. Therefore, our results could not be generalized to the overall population, especially to the elderly. Second, we did not measure individual differences in biological susceptibility to caffeine or expectancy for coffee drinking to stimulate cognitive function11. Further studies incorporating measurements of caffeine blood level and investigation of a subjective expectation of coffee drinking as a cognitive enhancer may clarify the dose-response relationship and main contributor of the FC changes. Third, the results of the neuropsychological tests after coffee consumption may be biased due to learning effects. However, learning effects were likely mitigated by the use of different sets of contents in the repetition of the same tests. Finally, since canned coffee contains various ingredients other than caffeine, it is unclear whether our results were due to the effect of caffeine or the combined effects with other ingredients. Nevertheless, our study is the first EEG network analysis investigating the effects of canned coffee, containing a precisely controlled content of caffeine, on neurocognitive function.
The strength of our study is that FC was evaluated using two methods, coherence and PLI, which were compared to mitigate the limitations of scalp-level EEG analysis. We used two representative building blocks for characterizing brain FC in sensor space, coherence, and PLI, and obtained consistent results. Coherence is the most common method used to quantify the correlation between signals from different brain regions in terms of both amplitude and phase. In contrast, PLI measures the stability of the phase differences of short- and long-range neuronal activities over time independent of the amplitude of oscillations. This method is designed to reliably estimate phase synchronization against the presence of common sources such as volume conduction and active reference electrodes. In brief, it can be accomplished by discarding 0 and π phase differences between two time series21.