A new study shows that even a moderate dose of caffeine alters brain activity during sleep, increasing complexity and nudging neural systems toward a high-efficiency processing state, especially in young adults during deep sleep.
Left column shows statistical differences (blue: reduced during caffeine, red: increased during caffeine), while SVM/LDA columns show classification accuracy between conditions (green). Dots indicate statistical significance (gray: p < 0.05, white: p < 0.01). Study: Caffeine induces age-dependent increases in brain complexity and criticality during sleep.
Caffeine ingestion leads to a reduction in long-range temporal correlations and also causes an increase in brain complexity, especially during non-rapid eye movement (NREM) sleep.
A recent study in Communications Biology assessed sleep electroencephalography (EEG) reports to understand the effects of caffeine on the brain, particularly during sleep.
Caffeine, brain health, and sleep architecture
Caffeine can have a disruptive effect on sleep quality by raising sleep latency and reducing sleep efficiency. On the contrary, research has demonstrated caffeine’s neuroprotective qualities, particularly against Parkinson’s disease (PD). Therefore, the impact of caffeine on health is complex. Given the widespread consumption of caffeine-containing products, it is crucial to understand the effects of caffeine on the brain during sleep across various age groups.
Caffeine delays the onset of rapid eye movement (REM) sleep and worsens the quality of awakening. Lack of sleep deteriorates the proper functioning of sleep-related brain processes. It also leads to other health issues such as weight gain, diabetes, hypertension, cardiovascular diseases, and depression.
Despite encouraging progress in research, existing knowledge on the mechanisms by which caffeine alters brain dynamics during sleep is in its early stages. There are not only direct effects through adenosine signaling, but also a host of downstream effects on other neurotransmitter systems, as well as complex interactions that drive the diverse effects of caffeine on brain function and sleep architecture. Additionally, these complex interactions may affect the balance between neural excitation and inhibition, which plays a key role in how the brain processes information during different sleep stages.
About the study
This study conducts an in-depth assessment of caffeine's effects on the brain’s electrophysiological signals while sleeping. The specific focus was on brain complexity and criticality. Periodic and aperiodic components in the EEG spectra were disentangled, thereby improving upon existing power spectral investigations.
A previous study has shown that caffeine increases brain entropy during wakefulness. However, whether similar effects extend to sleep EEG signals is still unknown. This study also assessed whether caffeine-induced changes in slow-wave sleep (SWS) and increased lighter sleep stages (e.g., N1 and N2) resulted in an increase in EEG complexity during non-rapid eye movement (NREM) sleep. The authors note that some of the observed increases in brain complexity may be partly due to caffeine shifting the proportion of lighter NREM stages, making it difficult to disentangle direct effects on brain dynamics from indirect effects via changes in sleep architecture. Additionally, it explored age heterogeneity in how caffeine impacts the brain, given that adenosine receptor density decreases with aging, which reduces the time spent in deep sleep.
Sleep EEG data were collected from 40 healthy participants under caffeine (200 mg) and placebo conditions during two non-consecutive nights. Entropy measures, complexity metrics, and power spectral density (PSD) were extracted from the EEG. Machine learning (ML) and inferential statistics were separately used for NREM and REM sleep.
Study findings
Concerning caffeine-driven changes in EEG oscillations, statistically significant reductions in power in the alpha, theta, and delta frequency bands were noted in NREM sleep. For delta and theta, this spanned parietal and central channels. Across parietal and frontal channels, a widespread increase in beta power was noted, which could be attributed to elevated GABA levels. During REM sleep, theta power was lowered in the occipital, parietal, and temporal channels.
Caffeine affected the aperiodic component of the spectrum. The change in the scaling behavior of the EEG spectra was attributed to altered EEG self-similarity, which could be associated with the shift in neural criticality. The discrepancies observed between the corrected and uncorrected spectral power indicate that the aperiodic component of the power spectrum may be a candidate EEG feature modified by caffeine. The study emphasizes that correcting for the aperiodic component reveals caffeine’s effects on rhythmic brain activity that might otherwise be missed, highlighting the importance of separating these components in EEG analysis.
All EEG complexity measures exhibited a consistent increase under caffeine. Both the Detrended Fluctuation Analysis (DFA) scaling exponent and the slope of the aperiodic activity showed a consistent reduction, implying a shift towards the critical regime. Although the caffeine-induced alterations in brain signals were widespread and prominent during NREM sleep, a weaker effect that was limited to the occipital regions was observed during REM sleep.
The most accurate association between caffeine and placebo sleep EEG was obtained via Spectral Sample Entropy as a feature in NREM sleep. The overall agreement between the single-epoch and subject-wise ML results indicates the sensitivity of the selected features of the caffeine effect and the robustness of the observations.
The random forest (RF) classifier, trained on all extracted features, achieved significantly higher decoding accuracy in NREM sleep than REM sleep. The complexity and criticality-related measures performed considerably better than the spectral features for classifying caffeine and placebo samples during NREM. The distribution of features ranked by importance during REM was found to be more heterogeneous and less structured.
Caffeine exerted significant effects during NREM sleep on several EEG features in both young (20–27 years) and middle-aged (41–58 years) adults. The younger age group exhibited more prominent effects. However, caffeine-induced changes in REM sleep were only significant in the younger age group, affecting specific features, including sample entropy (SampEn), spectral entropy (SpecEn), Lempel-Ziv complexity (LZc), and the DFA scaling exponent. The study found no significant age differences in the impact of caffeine during NREM sleep, but clear age-related differences were seen in REM, suggesting that age-related changes in adenosine receptor density may diminish caffeine’s effect on REM sleep in middle-aged adults.
The authors also discuss that the observed shift in brain activity patterns reflects a movement toward a theoretical “critical regime” or “edge of chaos,” where the brain is thought to be most sensitive, adaptable, and efficient at processing information.
Conclusions
This study demonstrated that caffeine intake resulted in a significant increase in EEG complexity, particularly during non-rapid eye movement (NREM) sleep. Caffeine also affects the brain, allowing it to process information most efficiently. Compared to the middle-aged group, younger adults exhibited a significantly increased brain entropy response to caffeine during REM sleep.
The effects were more widely distributed in NREM compared to REM sleep. However, the authors caution that these findings are based on healthy adults, and that caffeine’s effects may differ in individuals with sleep disorders or neurodegenerative conditions. Results should also be interpreted in the context of both direct and indirect changes to sleep architecture.