Deep learning-based control of electrically evoked activity in human visual cortex.

Monkeys learn to report their own sensory cortical population activity

Sensory-guided human-machine joint learning accelerates the acquisition of motor imagery brain computer interface control

Brain–computer interfaces (BCIs) offer the potential to restore function and augment human capabilities. However, non-invasive electroencephalography (EEG)-based BCIs still face challenges in learning efficiency and control precision, particularly for naïve users performing complex tasks. Here, we present a sensory-guided joint learning framework that integrates human motor learning with adaptive machine learning to improve BCI training and performance. In 31 BCI-naïve participants, the framework enabled rapid skill acquisition, achieving average online discrete accuracies of 86.0% for one-dimensional (1D) and 77.5% for two-dimensional (2D) motor imagery tasks, along with continuous control accuracies of 77.5% (1D) and 66.9% (2D). Mechanistically, tactile guidance reduced user exploration and accelerated neural adaptation, while sample reweighting aligned decoder updates with human learning trajectories. By coupling reinforcement-driven neural plasticity with adaptive algorithmic optimization, this framework advances BCI training from passive calibration to active human–machine joint learning, enabling practical and scalable neural interfaces for communication and rehabilitation. Motor imagery brain-computer interfaces are promising neurotechnologies but limited by slow user learning and unstable decoder adaptation. Here, the authors develop a novel sensory-guided joint learning framework that coordinates subject learning and decoder adaptation to improve BCI acquisition.

Development of a Primary Visual Cortex Model to Investigate Cortical Visual Prosthesis Stimulation

Selective gating of neural modulation through frequency- and behavior-dependent modes during cortical electrical stimulation

Electrical stimulation is widely used to modulate neuronal activity, yet its effects on neuronal circuits in vivo remain poorly understood. This, in turn, has hindered the principled design of stimulation protocols and raised questions about reproducibility that constrain the field’s translational impact. Here we combine cortical sinusoidal electrical stimulation (sES) with Neuropixels recordings to characterize stimulation-driven responses in more than 2,700 well-isolated neurons across 53 brain areas in 14 behaving, head-fixed mice. We uncover two distinct, concurrent modes of neural modulation. First is a sustained, brain-wide spike-phase entrainment effect that depends on stimulation frequency: entrainment to slow stimulation is supported by non-synaptic electric field propagation while anatomical connectivity dominates entrainment to higher stimulation frequencies. Second, we find a transient, spatially localized spike-rate modulation mainly mediated through anatomical connectivity that only emerges at high stimulation frequencies by selectively recruiting inhibitory neurons. We show that the two distinct modes are differentially shaped by behavior. By identifying how stimulation frequency governs the mechanism of neural engagement and how behavioral state selectively gates brain-wide entrainment but not local inhibitory recruitment, our results provide a mechanistic foundation for designing targeted, reproducible neuromodulation strategies.

Temporal interference stimulation: A new paradigm for non-invasive deep brain stimulation-mechanistic insights, clinical advances, and future directions.

Neurological and psychiatric disorders frequently arise from dysfunctional deep-brain circuits, yet targeting these subcortical structures with conventional non-invasive neuromodulation remains a significant challenge due to the lack of focal precision at depth. Temporal Interference (TI) stimulation has emerged as a transformative paradigm, leveraging the intersection of multiple high-frequency electric fields to generate a low-frequency amplitude-modulated envelope within deep-seated targets. This biophysical strategy enables the modulation of subcortical dynamics while minimizing the activation of overlying cortical tissues. Emerging preclinical evidence demonstrates that TI can robustly orchestrate neurotransmitter release, facilitate synaptic plasticity, and ameliorate deficits in both motor and cognitive domains. Preliminary clinical translations further underscore its potential in enhancing memory precision, accelerating motor skill acquisition, and suppressing epileptic biomarkers. The mechanistic understanding of TI has evolved from passive low-pass filtering to include nonlinear ion-channel rectification and, more recently, network-mediated inhibition-particularly the recruitment of parvalbumin-positive interneurons in superficial layers-as a critical determinant of spatial selectivity. Human intracranial studies have further refined this framework, revealing that TI operates in a subthreshold regime, produces carrier-independent deep modulation, and elicits a sustained carry-over effect absent from unmodulated kilohertz stimulation. The efficacy of TI is fundamentally governed by a complex interplay of controllable parameters-including carrier frequency offset (Δf), current intensity, electrode geometry, and timing-alongside uncontrollable factors such as individual anatomical heterogeneity and endogenous brain states. Furthermore, advanced computational modeling, particularly finite element simulations incorporating personalized head models, has become indispensable for characterizing electric field distributions and achieving individualized, high-precision targeting. This review provides a comprehensive synthesis of TI's mechanistic foundations, safety profiles, and therapeutic trajectory, while critically discussing the integration of closed-loop systems, multi-target paradigms, and patient-specific optimization as the next frontiers in non-invasive deep brain stimulation.

Inhibitory Evoked Potentials as a Spatially Dependent Intraoperative Marker of Clinical Tremor Reduction

Deep brain stimulation (DBS) of the ventral intermediate nucleus (Vim) of the thalamus may be used to treat medication refractory essential tremor. Using recordings from in vivo human Vim neurons, our previous work has suggested that evoked potentials (that we termed quasi-evoked inhibition) ∼2 ms following high frequency microstimulation pulses may be related to inhibitory synapses onto the Vim. Here, we investigate whether (i) quasi-evoked inhibition is related to clinical tremor reduction, and (ii) if quasi-evoked inhibition is dependent on the stimulation location within the Vim. By developing an objective determination of the presence or absence of quasi-evoked inhibition and utilizing accelerometer recordings, we showed that recordings with quasi-evoked inhibition at 100 Hz microstimulation exhibit greater tremor reduction than those without (P < 0.05, BF > 30). The number of stimulation pulses with quasi-evoked inhibition is also correlated with tremor reduction (rho = 0.18, P < 0.05) at all stimulation frequencies >=100 Hz. Furthermore, by analyzing microelectrode trajectories reconstructed from structural MRIs, we found that proximity to the ventral caudal border (P < 0.005) and to a previously established sweet spot (P < 0.05) are anti-correlated with the number of stimulation pulses with quasi-evoked inhibition. Our findings suggest that quasi-evoked inhibition is a potential biomarker of tremor reduction by means of network inhibition, and the more posterior regions of the Vim may allow for better recruitment of inhibition. This may be useful for closed-loop stimulation design.