How do tACS and temporal interference stimulation work for neuropsychiatric disorders?

Two techniques, two different objectives

The paper frames tACS and TI as complementary rather than competing tools. Conventional tACS delivers weak sinusoidal currents through scalp electrodes and is designed to interact with endogenous neural oscillations in a frequency- and phase-dependent way. TI instead applies two or more kilohertz-frequency electric fields with slightly different frequencies; where those fields overlap in tissue, they produce an amplitude-modulated field at the difference frequency.

That difference matters for what each method is trying to achieve. The review states that conventional tACS primarily modulates cortical and relatively superficial networks, and that its ability to engage deep structures with sufficient spatial selectivity remains limited. TI was introduced specifically to improve spatial steering and access to deeper neural targets, and initial animal studies showed that the region of maximal modulation could be steered by adjusting current ratios.

The paper also warns against treating a modeled field maximum as proof of selective deep engagement. Superficial tissues may still be exposed to stronger unmodulated carrier fields, and high-frequency carriers may not be biologically inert. A deep target that looks well placed in a model therefore does not by itself establish that the relevant circuit has been functionally engaged.

tACS acts through subthreshold spike timing

The mechanistic entry point for tACS is the neuronal membrane. Weak alternating fields shift the transmembrane potential in a frequency-dependent way, and the influence of the field weakens as frequency rises while phase lag accumulates. These shifts are subthreshold, but they can still change when a neuron fires by interacting with suprathreshold synaptic inputs.

The review traces this from membrane to network. Computational and experimental work indicates that tACS can bias spike timing, entrain local populations and modulate oscillatory power, phase synchronization and interregional coupling. In rats, transcranial electric stimulation entrained neuronal populations across cortical and hippocampal regions, and the proportion of phase-locked neurons increased with stimulation intensity and depended on the animal's behavioral state.

Cell type and dendritic location change the response. Layer 5 pyramidal-cell models show somatodendritic polarization that declines with increasing frequency, while interneuron models, including somatostatin- and parvalbumin-expressing cells, show distinct frequency-dependent phase-locking profiles. Apical dendrites display a sensitivity resonance in the 10–20 Hz range governed by local hyperpolarization-activated current, whereas soma and basal dendrites show monotonic low-pass responses. These details explain why stimulation frequency alone does not determine the behavioral outcome.

Excitability depends on region, state and field orientation

The review's preclinical evidence shows that tACS effects are not uniform across brain regions. In the rat motor cortex, pulse-train stimulation delivered at different phases of a continuous subthreshold sinusoidal waveform produced both excitatory and inhibitory effects, and modeling indicated that pulse timing and stimulation parameters can shift the balance between excitation and inhibition.

In awake nonhuman primates, tACS competed with ongoing endogenous oscillations for control of spike timing: lower intensities perturbed existing phase relationships, while stronger fields increasingly entrained firing to the applied waveform. In the visual middle temporal area of awake macaques, tACS reduced spike-frequency adaptation and produced a broadband increase in local field potential power.

The cerebellum shows strong directional and phase dependence. In anesthetized rats, rostrocaudally directed fields produced the largest modulation depth, followed by dorsoventral and mediolateral orientations. Negative and positive phases enhanced and suppressed cerebellar cortical multi-unit firing rates, respectively. Frequency optima also shifted with brain state: in anesthetized rats the 100–150 Hz range was most effective for phase locking with minimal delays in cerebellar nuclei, whereas a study in awake rats found 50 Hz most effective.

The paper also separates direct transcranial effects from peripheral co-stimulation. Neuronal entrainment persisted after local somatosensory input was blocked, supporting a direct transcranial contribution. But transcutaneous peripheral-nerve stimulation can reproduce or substantially contribute to some motor-system effects attributed to tACS, so the relative contributions likely depend on electrode placement, intensity, target region and outcome measure.

Safety evidence covers acute exposure, not repeated sessions

The review's safety claims are deliberately narrow. In rats, a single 20-minute session of 10-Hz tACS at 1.0–2.0 mA produced no detectable acute structural abnormalities, histological evidence of neuronal injury or stimulation-related thermal damage under the tested conditions.

That is an acute safety finding, not a long-term clearance. The authors state that repeated exposure, delayed tissue responses, and frequency- or dose-dependent safety thresholds remain insufficiently characterized. Across both techniques, available preclinical evidence generally supports an acceptable acute safety profile under the conditions tested, but safety across repeated sessions, prolonged exposure and the wider parameter space is not established.

TI adds a steerable envelope but unresolved biology

TI's proposed advantage rests on the physics of interference. Two kilohertz-frequency fields with slightly different frequencies create an amplitude-modulated envelope at the difference frequency inside neural tissue, and adjusting current ratios can steer where that envelope is maximal. The review describes this as a strategy for preferentially engaging deeper targets with improved spatial steering.

The biology of that envelope is still under investigation. The authors list unresolved questions: how neurons respond to amplitude-modulated fields, how direct neuronal polarization or activation by unmodulated high-frequency carriers contributes beyond envelope modulation, the extent of superficial and off-target effects, and how much neuromodulation is achievable at clinically feasible current levels.

The preclinical disease-model results are therefore best read as evidence of disease-specific effects rather than as established clinical mechanisms. In animal models, the review reports TI effects including restoration of hippocampal excitation/inhibition balance, enhancement of gamma oscillations and prefrontal-hippocampal information transfer, and reduction of amyloid-beta deposition in Alzheimer's disease models; orientation-dependent hippocampal circuit recruitment and reduction of interictal epileptiform discharges and pathological fast ripples in epilepsy; restoration of abnormal gamma-band activity in depression models; and modulation of mesocorticolimbic reward circuitry in substance use disorder. In a morphine-induced conditioned place preference model, non-invasive tTIS reduced morphine-associated preference without affecting locomotor activity, but the review notes that the specific mechanisms and therapeutic potential in addiction remain to be determined.

tACS has more human history than tTIS

The review's central translational comparison is a stage difference. tACS has undergone extensive human and clinical investigation, and the paper's timeline places its human translation in 2008, followed by mechanistic, preclinical and clinical expansion in psychiatric and neurological disorders. Human work cited in the review includes frequency-dependent perceptual effects over visual cortex, gamma-band enhancement of functional connectivity within sensorimotor networks, and alpha- and beta-band effects on network segregation and integration.

tTIS is at an earlier stage. The paper dates TI's methodological introduction to 2017, followed by mechanistic and preclinical investigations, early human exploration and initial clinical application in Parkinson's disease. The review characterizes tTIS as remaining at an earlier stage of mechanistic and translational development.

This asymmetry shapes how the two bodies of evidence should be compared. The tACS literature includes human and clinical investigation but still shows substantial interindividual and interstudy variability, which the authors call a major barrier to reproducibility and clinical optimization. The TI literature offers a spatial-steering rationale and preclinical disease-model effects, but its physiological basis and spatial capacity remain under active investigation.

What would make translation credible

The review's forward-looking argument is that fixed, empirically selected protocols are not enough. Because effects depend on stimulation parameters, electric-field distribution, individual anatomy, endogenous brain state and network dynamics, the authors call for mechanism-informed, state-dependent and individualized stimulation.

Their proposed standard is a chain of evidence: subject-specific electric-field modeling integrated with electrophysiology, neuroimaging, cell-type-specific and laminar recordings, multiscale computational models, and behavioral or clinical outcomes. That chain is meant to establish target engagement and dose-response relationships, not merely to show that a field reaches a modeled location.

The paper also asks for standardized reporting of stimulation parameters, modeled electric fields, physiological responses and adverse events to improve reproducibility and cross-study comparability. Its final criterion is explicit: the translational value of tACS and TI will depend on whether electric-field delivery can be reliably linked to neurophysiological responses, modulation of disease-relevant circuits and clinically meaningful outcomes.

Read together, the review's priority is not choosing between tACS and TI on the basis of a single frequency or a modeled hotspot. It is whether a given protocol can demonstrate, in the same study, that the field engages the intended circuit and that the engagement produces a measurable physiological and clinical effect. On current evidence, tACS has the longer human record but unresolved variability, while tTIS has the clearer spatial rationale but the shorter and more uncertain path to clinical use.

Sources used in this answer