Are placebo effects and expectancy bias overstating transcranial magnetic stimulation?

How big is the placebo effect in TMS studies?

The placebo effect in TMS is not small — it's often large enough to rival the active treatment. In a meta-analysis of repetitive TMS (rTMS) for primary insomnia, sham stimulation reduced insomnia symptoms by a very large margin: the effect size (a measure of improvement) was -0.98 after 10 days, -1.16 after 20 days, and -2.14 after 30 days [2]. To put that in perspective, the active rTMS group improved more, but 73.5% of that improvement was already achieved by the sham group alone [2]. That means most of the benefit people reported came from the placebo response, not the magnetic stimulation itself.

For schizophrenia, a meta-analysis of 44 randomized sham-controlled trials found that placebo effects on negative symptoms were moderate (effect size 0.44) and statistically significant [1]. Placebo also produced small-to-moderate improvements in memory, executive function, working memory, and processing speed [1]. These are not trivial effects — they are large enough to make a real TMS treatment look more effective than it actually is if the sham group isn't properly accounted for.

In a smaller trial on auditory hallucinations, both active and sham TMS led to significant improvement in hallucination scores, with no difference between the two groups [4]. This means the entire benefit seen in the active group could be attributed to placebo. The authors specifically noted a 'marked placebo effect' [4].

Does expectancy bias actually change how the brain responds to TMS?

Yes, and this is a key finding that goes beyond just 'believing' something works. A 2021 study showed that when people could predict a TMS pulse — either because they triggered it themselves or because a reliable cue warned them — the motor-evoked potential (the electrical response in their muscles) was significantly smaller than when the pulse came unexpectedly [5]. In plain terms, if you know a TMS pulse is coming, your brain dampens its own response to it. This is a form of sensory attenuation, similar to why you can't tickle yourself. The implication is that in a typical TMS session, where patients know exactly when and where the stimulation will occur, their brain may be actively reducing the stimulation's impact. This could make real TMS seem less effective than it could be, while also making sham stimulation seem more effective because the patient's expectations (and the brain's predictive mechanisms) are engaged in both conditions.

This finding also has a practical consequence: it suggests that the way TMS is delivered (predictable vs. unpredictable) can change the outcome, and that many studies may not have controlled for this factor. So expectancy bias isn't just a psychological confound — it has a measurable neurophysiological signature.

When does the placebo effect matter most — and when does real TMS still win?

The placebo effect is most pronounced in conditions that rely on subjective self-report, such as pain, mood, insomnia, and hallucinations. In the insomnia meta-analysis, the placebo effect was huge [2]. In the schizophrenia hallucination trial, placebo matched active treatment [4]. In the larger meta-analysis on negative symptoms and cognition, placebo effects were significant but smaller than the active treatment effects in most cases [1]. So for objective motor outcomes, the picture is different. In a stroke rehabilitation study, low-frequency rTMS produced measurable changes in brain inhibition and motor function that were significantly greater than sham [3]. The sham group did not show the same rebalancing of cortical excitability or improvement in motor outcomes [3]. This suggests that when the outcome is an objective physiological measure (like brain inhibition or muscle strength), real TMS can outperform placebo more clearly.

In summary, placebo and expectancy bias can overstate TMS benefits substantially — especially for subjective symptoms. But for objective motor or neurophysiological outcomes, real TMS often shows a genuine advantage. The key takeaway: always check whether a TMS study used a sham control and whether the outcome was subjective or objective.

About These Sources

This answer is built on 5 peer-reviewed studies — published from 2017 to 2026, 2 from 2024 or later, 1 in Q1–Q2 journals — selected as the most relevant from 5 studies that passed quality screening, drawn from 73 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions