Transcranial Magnetic Stimulation: A Loud Warning

By Neuroskeptic | March 18, 2013 5:07 am

Transcranial Magnetic Stimulation (TMS) is popular tool in neuroscience.

A TMS kit is essentially a portable, powerful electromagnet, called a ‘coil’. Switching on the coil causes it to emit a magnetic pulse, and this magnetic field is strong enough to evoke electrical activity in the brain. So, by placing the TMS coil next to someone’s head, neuroscientists can activate the brain region closest to the coil.

The effects can be dramatic: TMS over the motor cortex, responsible for muscle control, causes muscle twitches. Stimulation of the visual cortex makes people ‘see’ flashes of light. It’s as easy as waving a magic wand.

One major application of TMS is in research, and it works like this: suppose you want to test the theory that a particular part of the brain is involved in performing a particular cognitive task. You zap people with a TMS pulse aimed at the brain area of interest, just when they’re in the middle of doing the task. If your theory’s correct, and the brain area is involved, this ought to impair their performance on the task, by disrupting or overstimulating the region just when it’s needed.

Pretty neat, but there’s a catch. TMS is noisy – every pulse makes a loud “click” sound. The problem is that this noise could also affect performance on the task, by distracting the subject for example. So in order to be sure that the effects of TMS on cognition are really about the brain stimulation, not just the noise, you need a control condition. Sometimes neuroscientists use TMS aimed over a different part of the brain as a control; others use ‘sham TMS’, a device that looks and sounds like a TMS coil, but without the magnets.

However, according to a paper just out by Maastricht neuroscientists Duecker and Sack, controlling for the effects of noise is a lot harder than it sounds: Pre-stimulus sham TMS facilitates target detection.

Duecker and Sack used sham TMS on volunteers performing a simple task in which volunteers had to push a button as soon as they saw a visual target appear on screen. They found that the “clicks” of the sham TMS could help people respond quicker, especially if the clicks happened within a narrow time-window before the target appeared.

The effect of the sham TMS also depended on where the “coil” was: noises coming from the left side of the head made people faster to respond to targets that appeared on the left side of the screen. This is worrying, because sometimes people put the real TMS on one side of the head and the sham on the other side.

Duecker and Sack write that they’re not criticizing the use of sham TMS, but that it must be used carefully because:

depending on the experimental design and research question, these specific effects of sham TMS can lead to false positive or false negative results. Our findings thus have direct implications for many TMS experiments, and could be used to guide the choice of an appropriate control condition in the context of a concrete research question.

ResearchBlogging.orgDuecker F, & Sack AT (2013). Pre-stimulus sham TMS facilitates target detection. PloS ONE, 8 (3) PMID: 23469232

  • John

    I hope that they take on the fact that TMS causes muscle twitches in the face and scalp next. That is controlled for less often than the noise.

  • DS

    Hi Neuroskeptic

    You wrote:

    “You zap people with a TMS pulse aimed at the brain area of interest,
    just when they’re in the middle of doing the task. If your theory’s
    correct, and the brain area is involved, …”

    But when one “zaps” a brain with a TMS pulse one does not know which area/areas of the brain are being stimulated (or disrupted). I know that researches like to think they know in some vague sense but vague is just not good enough to claim that a particular area and only that area is being affected by TMS.

    I think it is best to keep the descriptions of TMS experiments accurate. The best we can presently do this, IMO, is to describe the methodology used and refrain from making claims that specific regions have been “zapped”. For instance one could accurately state that the TMS was positioned and oriented relative to a head in some reproducible manner and a current waveform of some shape, amplitude, and polarity was applied.

    Any claims that a specific area or areas were alone “zapped” would require real measured evidence. There is plenty of fodder for scientific story telling with TMS. This skeptic will resist the impulse.

    • Steffen Nestler

      DS, there are now commercial options available which allow
      stimulation with reference to a specific brain area, given an
      MRI scan of the respective person.
      (cf. )
      Employing 3D markers as well as a camera and depending
      on the use of a stand or the steadiness of the person who is
      stimulating, TMS pulses can be delivered accurately to the
      0.1 mm level. Whether this is always entirely desirable is,
      I suppose, a separate question.

      • DS


        Commercial options or not the physics does not allow the targeting of arbitrary regions of the brain with a user specified electric field.

        • Steffen Nestler

          Hello DS, thank you for your response!
          I’m not sure I quite understand, having re-read both of
          your messages.
          Firstly, I forgot to mention that the area can be defined in
          the accompanying software from a number of sources,
          such as MNI/Talairach co-ordinates or previously obtained
          functional activations (meaning predominantly fMRI).
          Secondly, if we assume that these parameters provide a
          certain standardization across participants which is in fact
          measurable, would this not constitute more of a
          ‘philosophical’ rather than a factual disagreement?
          Certainly, were one to argue that stimulating one point may
          not affect the whole temporal lobe (to pick an exaggerated
          example), one could select a grid of target points (as, in
          my understanding, is quite common in motor cortex
          research). And there’s obviously also the problem of
          connectivity between brain regions, but this applies to
          a lot of neuroscience and hopefully techniques such as
          DTI will assist us in answering relevant queries.

          • DS

            NO! In general one simply can not target a specific region only with a user specified E field. There is no controversy on this. The physics is clear. Only in very specific cases would the physics permit otherwise and even in those cases one would still have to give independent evidence that this is indeed being accomplished or one will be subject to much skepticism from physicists.

            But I fully expect the claims of specific targeting to continue as they have been. The dynamics of good story telling will permit it … for a while.

          • basneggers

            For empirical evidence on the accuracy of Neuronavigated TMS:


            Of course the E field deployed and the actual neuronal stimulation are not always right underneat the pulse center, and depend on local conductive morphology, but its not that bad as DS says. Effects are local to the stimulated location, and distal along connections as concurrent TMS – FMRI shows. But indeed we need more research to fully grasp what is going on.

          • DS


            I thank you for your participation in this discussion and hope you will not take offence but what can this: “Effects are local to the stimulated location” possibly mean?

            Also, your first link points to an article that uses fMRI as a validator of TMS spatial localization (localization of what remains a big question) but fMRI is in no position to validate effects of induced E fields. The fields could easily have excitatory, inhibitory or combined effects that do not be represented in fMRI. And I am being kind to fMRI by assuming that what does show up can be trusted to correlate with the applied B or E field applied in TMS.

            Your second link points to something that amounts to little more than a promotional advertisement.

          • basneggers

            Sure, in a strict biophysical sense you are right of course. FMRI is able to catch macroscopic vascular consequences of electrical neuronal activation as induced by the effects of E fields on neuronal membranes, by virtue of the neurovascular coupling. The E field neuron interaction is not yet well understood indeed, but see recent advances in detailed modelling for progress:


            But what is the “user specified field” you keep referring to? The B field generated by a figure of 8 coil is well known and easily characterised by Biot-Savart equations in the absence of coupling. The induced primary and secondary E field is a much more complex story as you can read in the above paper.

            But psychologists as neuroskeptic (i guess) usually argue in terms of units like gyri, sulci, brodmann areas and thelike, they usually are not that interested in the science of what happens on a smaller scale. Just a different perspective (and sorry for this shameless generalization). And with fmri validation one verifies effective activation at that scale relevant to psychologists. Neuroscientists and biophysicists tend to disagree, and i count myself to the latter.
            But for TMS effects the level of description at that macroscopic scale often suffices for psychology.

            Needless to say, we need to understand TMS at the microscopic level a lot better. We embarked on a big new project to do just that. Will get back here in 3 years and post results :-)
            but this shortcoming does not need to keep psychologists from doing interesting neurocognitive research.

          • DS

            This is the problem with neuroscience. Story replaces method.

          • Steffen Nestler

            DS, once again, I’m not the expert in the field,
            but there are a few things that I feel urged to

            Firstly, we are not talking about electrical stimulation
            (that would be tDCS), but the magnetic kind.
            Secondly. I don’t think people would argue that
            ‘only’ a particular area is being stimulated (see
            my earlier comment involving connectivity et al.),
            but rather, that it is _predominantly_ over that
            area, since the magnetic field rapidly decreases
            away from the target.
            Thirdly, I was taught that high-frequency stimulation
            increases activity (potentiation?) whereas low-
            frequency stimulation decreases brain activity
            (depolarization)? Unfortunately, I don’t have a
            reference at this point and I will gladly admit that
            the exact neurophysiological processes are not
            well understood as yet.

            DS, I’m very happy that you have brought these
            points up (as well as Neuroskeptic), since TMS-
            centered discussions are necessary and im-
            portant. However, I can’t help but think that we
            are debating two somewhat disparate issues
            and I’m still trying to wrap my head around the
            angle that you’re coming from. :o)

          • DS

            Right, we are talking about TMS not tDCS.

            More accurately, the B field rapidly decreases away from the coil, not the target. If one wanted to target the maximum B field amplitude at 2 cm below the surface of the cortex (the target) that would simply be impossible with a current source external to the brain.

            I do not know all the TMS literature but I have read enough to know that there is a lot of variation with respect to the characteristics of applied coil current and the potentiation, stimulation or inhibitory effects on brain tissue. Just changing the polarity of the current in the coil can change the effect on the neuronal tissue.

          • DS

            I don’t think we are debating disparate issue. I think we are debating the ability of TMS to “focus” an E field on a particular region of the brain primarily. Secondarily, I think we are debating what the effect an such an E field would be on the brain – ie stimulate, inhibit, potentiation.

            One thing that I hope we could agree upon is that any researcher that claims to target deep tissue alone without effecting more superficial tissue would need a lot of independent proof for such a claim.

          • DS

            My primary concern is to make sure that readers of this blog understand that one can not simply choose an arbitrary region of the brain to target and expect that TMS will be be able to create an E field localized to that region.

  • Ethan Solomon

    I’m a little confused. So long as people are comparing their true-TMS conditions to sham controls, what’s the big deal? Sham in the same place that you zap, and compare control and TMS-condition performance. Control performance might be higher or lower than baseline because of the click, but so what? Minus the actual zap, it’s all on an even playing field.

    What am I missing here?

    • Felix Duecker

      Being first author of this article, I would like to clarify a few things.

      The central goal of this experiment was to find out whether the click of the TMS coil influences task performance in a time- and site-dependent way. In order to mimic real TMS without the (confounding) presence of direct neural effects, we simply used a sham TMS coil. So, this experiment only uses sham TMS to identify the non-neural effects of TMS. It is not about sham TMS per se!

      As you point out, these non-neural effects are not problematic when using appropriate control conditions. We make the same point in the original article… However, many experiments actually do not follow your advise to use real and sham TMS over the same area at the same time points. In our view, this is bad practice and we provide empirical evidence that this is indeed risky.

      There is one additional aspect to the discussion that I would like to mention. Many TMS experiments are not as simple as you seem to suggest. For example, it is often interesting to investigate the functional relevance of two brain areas in one experiment. In such situations, one might argue that they control for each other. Or, one might include TMS over the vertex as control site an use this for comparison with the two experimental conditions. Our results suggest that this is not always sufficient for controlling the non-neural effects of TMS. Again, the best control would probably be to use sham TMS over both brain areas. I am pretty sure that there are not a lot of studies out there that have done this…

      Finally, a very short remark about why not just use real and sham TMS over the same area and that’s it?. I agree that this controls for the non-neural effects of TMS. However, this approach is suboptimal for another reason: it fails to demonstrate neural specificity. Ideally, you would like to show that your TMS effects do not occur when stimulating any brain area but specifically when targeting the brain area in question. For this, you need additional stimulation sites which brings you back to my previous point.

      I hope this puts things into perspective…

      • Neuroskeptic

        Hi Felix, thanks very much for commenting. That’s very helpful!

  • Boon Tee Tan

    Was the e-m pulse adequately focused or aimed at the “region” of the brain “closest” to TMS? The e-m wave could penetrate much deeper.
    Would the control condition be appropriate? This article lacks clarity.

    • DS

      Any TMS article which claims to target a specified region of the brain will lack clarity until the targeting of the region is validated by some independent means.

  • hnarf

    This is probably a stupid question, but wouldn’t it be possible to avoid these problems simply by giving the subjects a pair of powerful ear plugs (assuming you’re running an experiment that doesn’t require them to respond to any auditory stimulus)? Or if auditory stimulus is needed, give them a pair of sound-insulated headphones that can play whatever sounds are needed but that otherwise block out external noise.

    • basneggers

      You will still feel muscles on the scalp contract, the second confound one has to deal with…

  • Pingback: News Mash: Magnetic Brain Stimulation To Help With Cognitive Tasks & Managing Pain? Yup. | Anguished Repose()

  • Pingback: Transcranial Magnetic Stimulation: A Loud Warning : Neuroskeptic | Neuro Physiology Blog()

  • Beth

    Love to see you comment on TMS used for depression.

  • glenn williams

    Here is my experience 10 treatments in. It may be working. I think it is. .. but they increase the amplitude every two sessions or so. They say You may experience some discomfort ” Translation : This is really going to hurt ! 5 second migraine and 20 second pause for 37 minutes or until it reaches 3,000 pulses. I am at 105″%” now and it rattles my teeth so hard I am wanting a football or boxing mouthpiece.
    Like I said, I think it is working, and will be worth the temporary physical pain, if it can remove some of the emotional pain.



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar

Login to your Account

E-mail address:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »