The Caffeinated Brain Part 1: The Effects of Caffeine on Event-Related Potentials (ERPs)-A Systematic and Mechanistic Review.

抄録

Introduction: Caffeine is a widely consumed adenosine receptor antagonist with well-documented effects on arousal and performance, but its time-resolved neurophysiological signature across stages of information processing remains fragmented across event-related potential (ERP) paradigms.Objectives: This systematic and mechanistic review aimed to (i) identify and catalog human ERP studies testing caffeine effects, (ii) synthesize findings by task domain and ERP component family, and (iii) evaluate moderators including dose, timing, abstinence/withdrawal control, sleep status, and habitual use.Methods: Following PRISMA 2020 and PRISMA-S, we searched multiple databases (PubMed/MEDLINE, Embase, APA PsycINFO, Web of Science Core Collection, Scopus, IEEE Xplore, and Cochrane Central Register of Controlled Trials) from inception to 28 November 2025 and conducted a structured narrative synthesis using SWiM (Synthesis Without Meta-analysis, no prespecified quantitative pooling). Risk of bias was assessed using RoB-2 (Risk of Bias 2, including crossover extension) and ROBINS-I (Risk Of Bias In Nonrandomized Studies of Interventions). Of 761 records, 63 controlled human studies met the inclusion criteria. The evidence most consistently supported stage- and context-dependent modulation. Within the P3 family, target-related P3b/P300 latency was frequently shortened, or fatigue-related slowing was prevented, often without parallel increases in amplitude. P300 amplitude findings were mixed and context-dependent: amplitude was often unchanged in rested or low-demand paradigms, but increased or was restored when caffeine counteracted fatigue, sleep loss, sustained attention demands, or high workload. Preparatory activity (CNV/slow negativity) showed selective effects, while early sensory components were comparatively stable in many paradigms; higher doses (approximately 200-400 mg) were associated with weaker early auditory sensory gating in some studies.Conclusions: Across heterogeneous paradigms, caffeine was associated with context-dependent ERP changes rather than a uniform amplification of ERP amplitudes. The most consistent pattern was shorter or preserved latency of late positive ERP components, particularly in tasks requiring stimulus evaluation or target detection. In some fatigue, sleep deprivation, sustained attention, or high-demand paradigms, caffeine was also associated with larger or restored P300/P3b amplitudes. These findings are compatible with state-dependent changes in attentional engagement or stimulus evaluation, but mechanistic interpretation remains limited by heterogeneity in task paradigms, ERP definitions, dosing, abstinence procedures, and participant caffeine use profiles. Methodological heterogeneity, small samples, inconsistent control of habitual use and withdrawal, and the predominance of healthy young adult samples limit generalizability, particularly to children, older adults, clinical populations, and long-term high-dose caffeine users.