Same currents, different field: delta f = 0 carrier-only stimulation is not temporal interference without the beat
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.
Ines: I'm Ines.
Marcus: And I'm Marcus, here with Yuki.
Ines: We're looking at the paper: "Same currents, different field".
Marcus: Here's the gist of the paper. Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation, assumed to reproduce TI's kilohertz carrier field without its low-frequency beat Without the frequency difference, the carrier fields interfere coherently, changing the peak strength, power, and direction of the resulting field depending on electrode polarity and montage geometry Using computational field modeling, they demonstrate that these mismatches can be substantial and widespread They recommend that TI studies report electrode polarity, model both carrier-field exposures, and choose controls according to their study objectives The Failure of the Carrier-Only Control Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation, assumed to reproduce TI's kilohertz carrier field without its low-frequency beat.
Ines: However, matching the carrier currents does not match the kilohertz field underlying the AM.
Marcus: Without the frequency difference, the two carrier fields interfere coherently, and the field's maxima and minima can differ substantially from those under TI.
Ines: Formally, let E1 and E2 be the field amplitude vectors of each electrode pair at a given point, which TI and carrier-only stimulation share in the quasi-static regime.
Paper summary: Ines: So to wrap up this discussion on Same currents, different field: delta f equals zero carrier-only stimulation is not temporal interference without the beat, what's the big picture here? It basically challenges the idea that carrier-only stimulation is a perfect control for temporal interference.
Marcus: The authors of this paper are saying that you have to account for how electrode polarity and montage geometry interact when you remove that low-frequency beat. This means your choice of control condition has to be tailored specifically to what you are trying to study in your experiment, not just picking one standard way.
Yuki: From a population genetics view, this suggests that when we look at real biological samples, the spatial differences between these controls could be really important for understanding how different brain regions respond. It points toward needing more complex modeling to understand those responses across a species.
Ines: Right. The implication is that if you're doing temporal interference studies, you need to explicitly state your electrode polarity and model both carrier-field exposures so that the results are interpretable in context of what the authors found in this paper.
Marcus: They also suggest that for clinical studies, maybe prioritizing sensory matching with a sham device is fine, but for studying the actual mechanisms of temporal interference, you need controls planned with an explicit field model. The dependence on polarity can even be used as a tool to cancel out the carrier field at certain targets.
Yuki: It pushes us toward more detailed experimental design where we consider the full context of how fields interact in different brain structures. We can't just look at one measurement in isolation anymore; we need to account for all those spatial and geometric factors mentioned.
Conclusion: Ines: So, what does this title really mean for us? Why is it "Same currents, different field"?
Marcus: It means just matching the current flow isn't enough to get a fair comparison for temporal interference. The resulting electric field is fundamentally different because of that frequency offset.
Yuki: From a genetic standpoint, this suggests that what we measure in the lab might not translate directly to how these fields affect real brain tissue in people.
Ines: Exactly. It means if we just look at the carrier current, we’re missing a whole piece of the puzzle about how the field actually moves things around.
Marcus: And it points to a huge problem for setting up experiments; you have to be super specific about your setup because that zero frequency difference thing doesn't work as well as people think.
Yuki: It means we need much more careful modeling before we can trust any of these field measurements when looking at population data.
Ines: Right. And the authors are basically saying that for anyone doing temporal interference studies, you have to be really careful about what control you choose based on your actual goal.
Marcus: They suggest that clinical studies might use a simpler sham control, but for understanding the underlying mechanism of temporal interference, you need a control that models both the carrier current and the resulting field exposure.
Yuki: And they're pushing for these studies to be explicit about electrode polarity and to report on both the peak fields under temporal interference and those under carrier-only stimulation.
Ines: It’s a strong call for better experimental design across the board if we want reliable results in this area. This opens up a lot of new ways to test our theories, but it requires us to change how we plan these investigations from the start.
Marcus: And it makes me wonder how much more modeling is going to be needed before we can actually make sense of these complex field interactions across different brain regions.
Department of Biomedical Engineering, University of Wisconsin-Madison · Department of Psychiatry, University of Wisconsin-Madison
q-bio.NC
Submitted: 2026-11-10
Updated: 2026-11-10
Code: https://github.com/idossha/carrier-only-stim
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 89/100
The gist: Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation…
Key concepts
- Temporal Interference (TI)
- TI occurs when the relative phase between two carrier fields sweeps through every value once per beat. This causes the resulting magnetic field to reach a maximum strength determined by the worst-case carrier exposure, which is higher than in simple carrier-only stimulation.
- Carrier-Only Stimulation ($\Delta f = 0$)
- This control uses the same currents but no frequency difference between carriers. Because there is no beat, the relative phase of the two fields remains fixed. This results in a simpler field pattern that does not fully capture the complex, time-varying peaks seen in true TI.
- Peak Field Strength ($ ext{HFmax}$)
- This refers to the maximum magnetic field strength generated at a specific point. In TI, this peak can be much larger than in carrier-only stimulation because the relative phase changes. The paper shows that carrier-only stimulation only matches TI's peak when electrode fields are orthogonal.
- Spatial Mismatch
- This describes how the magnetic field distribution differs between the two conditions across different brain regions (montages). Modeling reveals that for many setups, carrier-only stimulation results in a field strength 1.5 to 2 times lower than TI's peak, and significant mismatches occur in areas like the insula.
Terminology
Summary
Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation, assumed to reproduce TI's kilohertz carrier field without its low-frequency beat Without the frequency difference, the carrier fields interfere coherently, changing the peak strength, power, and direction of the resulting field depending on electrode polarity and montage geometry Using computational field modeling, they demonstrate that these mismatches can be substantial and widespread They recommend that TI studies report electrode polarity, model both carrier-field exposures, and choose controls according to their study objectives
The Failure of the Carrier-Only Control
Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation, assumed to reproduce TI's kilohertz carrier field without its low-frequency beat. However, matching the carrier currents does not match the kilohertz field underlying the AM. p. 0 2026) have included
“carrier-only” stimulation as a control, delivering the same carrier currents without the frequency difference (ΔF = 0), and
this has been considered the most rigorous comparison (Peterchev 2025; Demchenko et al. However, matching the
carrier currents does not match the kilohertz field underlying the AM. Without the frequency difference, the two carrier
fields interfere coherently, and the field's maxima and minima can differ substantially from those under TI. Formally, let E₁ and E₂ be the field amplitude vectors of each electrode pair at a given point, which TI and carrier-only
stimulation share in the quasi-static regime (Wang et al. Formally, let E₁ and E₂ be the field amplitude vectors of each electrode pair at a given point, which TI and carrier-only
stimulation share in the quasi-static regime (Wang et al. In TI, the relative phase between carriers sweeps through
every value once per beat, so the field necessarily reaches HFmaxTI = max(|E₁ + E₂|, |E₁ − E₂|), labeled “worst-case carrier
exposure” in (Cassarà et al. In TI, the relative phase between carriers sweeps through
every value once per beat, so the field necessarily reaches HFmaxTI = max(|E₁ + E₂|, |E₁ − E₂|), labeled “worst-case carrier
exposure” in (Cassarà et al. However, without a carrier-frequency offset, the relative phase is fixed. However, without a carrier-frequency offset, the relative phase is fixed. With equal
carriers in phase, peak HFmax0 (0 for ΔF = 0) is |E₁ + E₂|, which matches HFmaxTI where the fields are at most 90° apart but
falls below it at larger angles, and to zero for equal and opposite fields (Fig.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Spatial Mismatches Across Montages
The time-averaged power, which drives heating, also differs. TI's carriers add incoherently, PTI ∝ E12 + E22, whereas those of carrier-only stimulation add coherently, P0 ∝ E1 ± E22. Where carrier-only stimulation matches TI's peak, it holds that peak continuously and deposits up to twice TI's power (for equal, parallel E1 and E2); where the fields interfere destructively, it deposits much less. The carrier-only field also oscillates along one fixed axis, whereas TI's peak direction swings over every beat. TI and carrier-only stimulation thus share peak and power only where E1 and E2 are orthogonal. p. 0 1A), which matters for orientation-dependent effects
such as conduction block. TI and carrier-only stimulation thus share peak and power only where E₁ and E₂ are orthogonal. This mismatch is not an edge case, and its extent depends on the montage. Using the leadfield of one head model (Ernie),
we computed HFmaxTI and HFmax0 for all 3,848,925 four-electrode montages on a 10–10 system, counting grey matter asWhere the paper says it
Where the paper says it
Quantitative Mismatch in Published Models
The extent of this mismatch depends on the montage. Using two published montages in 15 MRI-based head models, they found that for the hippocampal montage, carrier-only stimulation matched TI's peak in the left hippocampus at 1.95 times TI's power and left only 9% of grey matter mismatched. Conversely, for the Schaeffer et al. p. 0 (Grossman et al. 2017) to ours (Schaeffer et al.Where the paper says it
Recommendations for Future Research
The authors suggest that studies using carrier-only stimulation should (i) report the phase lock and the electrode polarity of every channel, without which its field is undefined. They also suggest that studies should (ii) model and report HFmaxTI, HFmax0 and both arms' carrier power, in the target and off target, alongside TImax. Alternatively, a high-frequency beat (e. p. 1 We suggest that studies using carrier-only stimulation
(i) report the phase lock and the electrode polarity of every channel, without which its field is undefined; and (ii) model
and report HFmaxTI, HFmax0 and both arms' carrier power, in the target and off target, alongside TImax. Alternatively, a highfrequency beat (e. We therefore
recommend that TI studies report electrode polarity, model both carrier-field exposures, and choose controls according to
their study objectives. Keywords: temporal interference stimulation; carrier-only stimulation; kilohertz electric field; electric field modeling
Temporal interference (TI) stimulation rests on the premise that the low-frequency amplitude modulation (AM) of two
kilohertz carrier fields acts on neurons beyond the carriers themselves, and there is accumulating evidence that it does
(Grossman et al. , carriers of 4. 5 kHz instead of 5. Ultimately, no control simply separates the AM from the carrier field. Because direct kilohertz effects plausibly depend on
field strength and orientation as well as frequency, a difference between TI and carrier-only stimulation may reflect the
missing envelope, a lower carrier peak, a higher carrier power or a fixed field direction, and a null difference does not
show that the envelope is inactive.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Conclusion on Control Selection
TI studies should therefore treat carrier-only stimulation as an independently defined condition and select controls according to their objectives. Clinical studies may prioritize sensory matching with a ramponly sham, whereas studies of TI mechanisms need controls that account for carrier-field exposure and should be planned with an explicit field model. p. 1 TI studies should therefore treat carrier-only stimulation as an independently defined
condition and select controls according to their objectives. Clinical studies may prioritize sensory matching with a ramponly sham, whereas studies of TI mechanisms need controls that account for carrier-field exposure and should be planned
with an explicit field model. Nevertheless, the dependence on polarity can
even be put to use. For example, reversing pair 2 in the hippocampal montage largely cancels the carrier field at the target
(Fig.Where the paper says it
Where the paper says it
Supplementary Data Details
The angle between the electrode pairs ranges from 1° to 99° across the four montages of Fig. 1. Across all 3,848,925 montages of Fig. p. 7 2
The angle between the electrode pairs
Which polarity is mismatched follows, roughly, the angle θ between the two pairs on the scalp, each drawn + → −, which
ranges from 1° to 99° across the four montages of Fig. Across all 3,848,925 montages of Fig. Across all 3,848,925 montages of Fig. 1C, the share
of grey matter mismatched with E₁ + E₂ rises with θ (Spearman ρ = 0. 1C, the share
of grey matter mismatched with E₁ + E₂ rises with θ (Spearman ρ = 0. 92), and its median is 7%, 40% and 84% at θ = 45°,
90° and 135° (Fig. 1D and 1E; diamonds, additional examples of
Fig. The two published montages
Table S1 gives the medians and ranges over the 15 head models behind Fig. However, without a carrier-frequency offset, the relative phase is fixed. With equal
carriers in phase, peak HFmax0 (0 for ΔF = 0) is |E₁ + E₂|, which matches HFmaxTI where the fields are at most 90° apart but
falls below it at larger angles, and to zero for equal and opposite fields (Fig. Carrier-only values are for phase-locked channels in phase, with pair 2 as each study lists it. a The nearest 10–10 sites to the
EGI electrodes reported by Schaeffer et al.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Data Availability and AI Use
Code that reproduces the analyses is available at https://github.com/idossha/carrier-only-stim. p. 4 Data Availability
Code that reproduces the analyses is available at https://github. com/idossha/carrier-only-stim. :
Conceptualization, Formal analysis, Funding acquisition, Investigation, Validation, Visualization, Writing – original draft,
Writing – review & editing. Declaration of generative AI and AI-assisted technologies in the manuscript preparation process
During the preparation of this work, the authors used Claude Opus 5. Declaration of generative AI and AI-assisted technologies in the manuscript preparation process
During the preparation of this work, the authors used Claude Opus 5. 5 (Anthropic) in order to assist with text editing,
restructuring, and clarity improvements. 5 (Anthropic) in order to assist with text editing,
restructuring, and clarity improvements. No scientific content was generated by AI.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Integration of explicit field modeling into temporal interference stimulation protocols allows researchers to choose controls according to their study objectives, as suggested by Haber et al. This enables studies to report both carrier-field exposures and model HFmaxTI and HFmax0 for target and off-target regions alongside TImax.
-
Development of a highfrequency beat (e.g., carriers of 4.5 and 5.5 kHz instead of 5.00 and 5.01 kHz) could preserve TI's peak field strength, total carrier power, and envelope distribution while minimizing neuronal modulation by using a modulation frequency that neurons likely follow far less than a slow beat.
-
AI systems can be trained to identify montage-dependent mismatches between carrier-only stimulation and true temporal interference by analyzing the relationship between the angle theta between electrode pairs and the fraction of grey matter mismatched with E1 + E2.
-
AI tools can be used for automated validation of carrier-only stimulation controls by comparing their resulting peak field strength to simulated TI fields across various montage geometries, flagging instances where mismatches fall more than 10% below HFmaxTI.
Abstract
Carrier-only stimulation (Δf = 0) is generally considered the most rigorous control for temporal interference (TI) stimulation, assumed to reproduce TI's kilohertz carrier field without its low-frequency beat. We show that this assumption is incorrect: without the frequency difference, the carrier fields interfere coherently, changing the peak strength, power, and direction of the resulting field depending on electrode polarity and montage geometry. Using computational field modeling, we demonstrate that these mismatches can be substantial and widespread. We therefore recommend that TI studies report electrode polarity, model both carrier-field exposures, and choose controls according to their study objectives.
Related papers
- Self-Supervised Learning of Local Predictive Directions as a Neural Computation
- Synchronized Hearts, Desynchronized Brains: A Nonlinear Dynamics Perspective for Function and Failure
- Message interaction dynamics covary with brain volume across mammalian connectomes
- Planning as Dynamics Relaxation: Hippocampal Recurrent Network Realizes Optimal Goal-Directed Navigation
- The Platonic brain bridge hypothesis: brain networks as architectural priors for multimodal large language models
- BrainPFN: Amortized Brain-Behaviour Prediction via Prior-Data Fitted Networks