Ergogenic aids have long attracted scientific interest for their potential to enhance neuromuscular performance, with caffeine being among the most extensively studied. While traditionally attributed to peripheral actions on skeletal muscle, accumulating evidence indicates that, at physiological doses, caffeine’s ergogenic effects are predominantly mediated by antagonism of central adenosine receptors. This antagonism leads to increased arousal, reduced inhibitory neuromodulation, enhanced corticospinal excitability, and altered motor unit recruitment and firing behavior. Importantly, the concentrations required to elicit direct effects on excitation–contraction coupling via ryanodine receptors exceed those compatible with human safety, rendering such mechanisms unlikely in vivo. This narrative review synthesizes contemporary neurophysiological evidence to propose that caffeine acts primarily by “tuning” motor system gain through central neurotransmitter modulation, rather than by directly augmenting muscle contractile properties. Additionally, we highlight unresolved questions regarding persistent inward currents, sex-dependent neuromodulatory influences—including the potential role of estrogen in regulating adenosine receptor expression—and the implications of repeated caffeine use during training for neural adaptation and motor control. Finally, we outline key methodological and conceptual directions for future research aimed at refining our understanding of caffeine’s neuromuscular effects in both acute and chronic contexts.

Resistance training lasting a few weeks increases maximal force mainly through neural adaptations that enhance the drive from the nervous system to muscle. While these adaptations have been well documented at the motor unit (MU) level during submaximal force contractions, the mechanisms underlying force increases during maximal voluntary contractions are poorly understood. This is due to a classic technical limitation in tracking MUs longitudinally during maximal force tasks. Here, we solved this technical challenge, enabling the investigation of MU adaptations during MVCs in both the trained and untrained limbs following unilateral resistance training. High-density surface electromyography was recorded from the biceps brachii of both limbs before and after a 4-week unilateral resistance-training intervention, and the same MUs were longitudinally tracked across sessions during MVCs by concatenation of three MVC trials of ∼5-s each. Unilateral training increased maximal force in the trained limb (+16%) and induced strength transfer to the untrained limb (+8%). In both limbs, maximal contractions after training were characterized by greater EMG amplitude, faster muscle-fiber conduction velocity, and higher MU discharge rates, indicating enhanced neural drive to the motoneuron pool. These adaptations were strongly associated with improvements in maximal force ( R ² > 0.7 for all). Importantly, longitudinal MU tracking revealed a non-uniform adaptation across the MU pool: MUs with higher baseline conduction velocity , indicative of higher recruitment threshold , exhibited the largest pre-post increases in discharge rate, whereas lower-threshold units showed smaller changes. Collectively, these findings demonstrate that gains in maximal force and their transfer to the untrained limb are primarily mediated by enhanced rate coding of higher-threshold MUs during MVCs.

Enhanced untrained muscle strength and force steadiness following unilateral resistance training (i.e., cross-education ) is typically attributed to neural responses. However, the mechanisms of these adaptations for spinal motoneurons remain unexplored. Therefore, we examined maximal-voluntary-force (MVF), steady-force variability (CovF), and longitudinally tracked motor unit adaptations in 10 individuals completing a 4-week unilateral strength intervention compared to 9 controls. High-density surface electromyography was recorded from the biceps brachii during steady (10%MVF) and trapezoidal (35%MVF) contractions. The relative proportion of common synaptic input (CSI) to motoneurons and its variability (CSI-V) were estimated using coherence and spectral analysis. Indirect estimates of persistent inward currents using firing rate hysteresis (ΔF) and motor unit recruitment thresholds (MURT) were assessed during the ramp forces (35%MVF). MVF increased in both the trained (+14%, p <0.001) and untrained limbs (+6%, p =0.004), and CovF decreased in both limbs ( p <0.001). Greater CSI was observed on both sides (p<0.01), concomitant with reduced CSI-V (p<0.01). ΔF increased exclusively in the trained limbs (+1.61±0.71 pps; p <0.001), and both sides exhibited lower MURT (p0.70, p 0.65, p 0.70, p <0.01). Our findings suggest that enhanced untrained muscle force and steadiness are mediated by increased relative strength of shared synaptic input with respect to independent noise and decreased variability of this shared input, with gains in trained muscle MVF being associated with ΔF. Key points: Unilateral resistance training improves strength and force steadiness in the contralateral untrained limb, suggesting neural adaptations without directly overloading the muscle. Despite established force-related modifications, the specific neural mechanisms mediating these functional responses remain largely unknown. 4-week unilateral training intervention enhanced muscle strength and force steadiness in the untrained limbs of 10 individuals, alongside a greater proportion of shared synaptic input, reduced variance in common input, and lower motor unit recruitment thresholds. We demonstrate that the neural mechanisms underlying improved strength and force control in muscles without mechanical overloading are associated with a higher relative shared input among motoneurons and reduced variance in these common input components. Graphical abstract legend unilateral training intervention reduced the variance in common synaptic input (CSI-V), which was associated with decreased variability in force steadiness (CovF) in both the trained and contralateral untrained limbs. On the exercised side, the increase in maximal voluntary force (MVF) was accompanied by a higher proportion of common synaptic input (CSI), a lower motor unit recruitment threshold (RT), enhanced persistent inward current (PIC) amplitude, and increased neural drive. In contrast, the contralateral untrained limb exhibited higher shared synaptic input and a lower RT but with unaltered PIC amplitude and neural drive (⇋). Overall, these adaptations resulted in a 14% increase in MVF in the exercised limb and a 6% increase in the untrained limb.