Does Mindfulness Ease Anxiety Beyond Placebo Expectation?

Mindfulness may ease anxiety because it changes your relationship with thoughts, not merely because you expect a soothing result. A 2024 randomized balanced-placebo experiment by Bos et al. found that people who received brief decentering instructions reported lower anxiety immediately afterward than people given sham instructions. The study did not find a statistically significant benefit from expectancy framing alone.

That result targets a question that has bothered mindfulness research for years: does practice work because people learn to see thoughts as mental events, or because a calm voice tells them that they ought to feel calmer? The answer from this study is encouraging, but narrow. It came from 128 nonclinical online volunteers during the COVID-19 pandemic, after one 20-minute audio session, with anxiety measured only at the end of that session. It supports an immediate effect under those conditions. It does not establish a durable treatment for an anxiety disorder.

What decentering means

Decentering means noticing a thought without treating it as a fact, command, or definition of who you are. "I am going to fail" becomes "I am having the thought that I will fail." The difference may sound grammatical, but it changes the mental task. Instead of arguing with the thought or obeying it, you observe it.

That differs from relaxation, distraction, or simple breath focus. Those practices can feel helpful, and many mindfulness programs include them. But decentering is a metacognitive skill: it creates distance between the person and the thought content. Whether that skill is the active ingredient for anxiety is the question Bos and colleagues tried to isolate.

A placebo design built for the mechanism question

Bos et al. used a 2×2 balanced-placebo design with 128 participants, or roughly 32 people in each cell. Participants received either mindful decentering instructions or sham audio instructions. They also received either high-expectancy framing, intended to raise belief that the exercise would help, or sham expectancy framing. The design therefore separated the instruction itself from the expectation that it would work.

After the intervention, participants completed the short form of the State-Trait Anxiety Inventory, a measure of current state anxiety. The mindful-decentering condition produced significantly lower post-session state anxiety than the sham condition. The publication details available here do not report the effect estimate or confidence interval, so the size and precision of that difference cannot be judged from the reported summary. Expectancy framing showed no statistically significant main effect, and the decentering-by-expectancy interaction was also non-significant.

The authors also asked participants to write after the task. People in the mindful condition used more decentering-related language than people in the sham condition. That check makes it more plausible that the two audio conditions prompted different mental approaches. It does not prove that the participants achieved decentering as psychologists define it. A person can repeat the language they just heard without changing how they relate to a distressing thought.

The study shows that a brief decentering instruction reduced immediate self-reported anxiety more than a sham instruction in this sample. It does not show that expectation never matters.

Why the null interaction needs restraint

The absence of a significant expectancy effect or interaction does not settle the issue. Factorial studies need more participants to detect interactions than to detect main effects, and this experiment had about 32 participants per condition. A non-significant interaction can reflect genuine independence between decentering and expectation, but it can also reflect limited statistical power. The study gives direct evidence against the idea that expectancy alone explains the observed immediate effect. It provides weaker evidence that expectancy plays no role at all.

Demand characteristics remain another concern. Participants may infer what a mindfulness exercise is supposed to do, even if researchers manipulate expectancy separately. The word-count check cannot distinguish genuine cognitive distancing from relaxation, distraction, or verbal mimicry. Established decentering measures would have made the mechanism claim more secure.

One mindfulness mechanism does not fit every condition

Even if decentering drove the effect in this experiment, mindfulness does not appear to run through one universal pathway. In a randomized clinical trial of mindfulness-based stress reduction for people with PTSD, Kang et al. identified interoceptive awareness, rather than decentering, as the mediator of therapeutic effects on PTSD symptoms. Interoception involves sensing and interpreting internal bodily signals.

PTSD is not interchangeable with everyday state anxiety. It often includes hyperarousal, avoidance, and strong bodily threat responses. Still, the trial provides an important correction to any simple claim that decentering is the active ingredient in mindfulness. Different symptoms, populations, and programs may recruit different processes. A thought-heavy pattern of worry may respond differently from trauma-related hyperarousal.

Broader delivery research also supports a more modest conclusion: self-guided formats can engage people and improve outcomes, but format alone does not reveal mechanism. In teachers, Montero-Marin et al. found improvements in stress and well-being with both self-taught and instructor-led mindfulness programs. Anxiety was not the primary endpoint, and the study did not isolate decentering. Its useful contribution is practical: a live instructor is not the only route to meaningful participation in mindfulness practice.

Similarly, a randomized trial of a 21-session self-guided digital program for adults with anxiety and depression symptoms found reductions in anxiety and depression. The intervention combined cognitive-behavioral and mindfulness-based strategies, so Staiano et al. cannot tell us whether decentering caused the improvement. It does show that a digital format can produce symptom change without therapist delivery when the program is longer and more comprehensive than a single audio exercise.

The COVID context may have amplified the signal

Bos et al. recruited during the pandemic, when many people faced an acute and shared stressor. That setting may have made a brief intervention easier to detect: someone already keyed up may have more room for state anxiety to fall than someone dealing with ordinary, diffuse stress. This is a plausible explanation, not a demonstrated one.

A 2025 pilot randomized trial by Sparacio et al. adds a useful counterweight. In adult Singaporeans, a brief smartphone-delivered mindfulness intervention for stress reduction produced modest or null efficacy signals. The studies are not identical: they differed in setting, platform, population, and outcome context. Sparacio's pilot also had limited power, so it cannot prove that brief smartphone mindfulness fails. But the contrast makes it hard to assume that one short decentering exercise will reliably lower anxiety outside an acute-crisis setting.

The wider literature supports mindfulness-based training for psychological distress in some groups, although effects vary by intervention and population. For example, a systematic review and meta-analysis in medical students reported reduced psychological distress and improved well-being after mindfulness-based training. That finding concerns multi-study training programs, not a 20-minute decentering audio, placebo expectancy, or a specific anxiety mechanism.

What to do with this evidence

For a person experimenting with mindfulness, explicit cognitive distancing has a stronger mechanistic rationale than treating mindfulness as a generic relaxation ritual. Practice language that asks you to label a thought, notice it arise, and let it pass can target the process tested by Bos et al. A useful prompt is: "What thought am I having right now, and what evidence says I must act on it?"

Do not mistake that rationale for a clinical protocol. The Bos study has no follow-up, no clinical anxiety sample, and no evidence that one-session effects persist, accumulate, or generalize to panic disorder, generalized anxiety disorder, social anxiety disorder, or PTSD. For persistent or impairing anxiety, evidence-based clinical care remains the appropriate standard.

The cleanest verdict is also the least glamorous: decentering looks like a real candidate mechanism for immediate anxiety relief, independent of expectancy in one well-aimed experiment. It needs replication with validated process measures, larger samples, clinical populations, and follow-up long enough to tell whether a momentary shift becomes a skill.