Everyone already believes exercise helps mental health. What almost nobody can tell you is how — which is usually where these claims fall apart under questioning. A new mediation analysis of the German ImPuls trial, Changes in repetitive negative thinking and stress perception mediate treatment effects of a transdiagnostic exercise intervention, published August 12, 2026 in Psychological Medicine, actually answers the how. Six months of structured running reduced symptoms in 400 people diagnosed with depression, anxiety disorders, PTSD, or insomnia — and it did it almost entirely by cutting two things: how much they ruminated, and how stressed they felt. Not by fixing their sleep, which is the mechanism everyone would have guessed first. Musonius Rufus, the Roman Stoic who taught Epictetus, spent a lecture called “On Training” arguing that this shouldn’t surprise anyone, because the soul was never separable from the body it happened to be running around in.

The Quick Verdict

Question	What we found
Who ran the study?	Anna Katharina Frei and a team led with Gorden Sudeck and Sebastian Wolf, University of Tübingen and collaborating institutions
Where and when?	Psychological Medicine, published August 12, 2026
How big was it?	400 physically inactive adults across 10 sites in Baden-Württemberg, Germany
Who was studied?	People diagnosed with major depressive disorder, agoraphobia, panic disorder, PTSD, or primary insomnia
What was the intervention?	30 minutes of moderate-to-vigorous outdoor running, 2-3 times weekly for 6 months, plus behavioral coaching, versus standard treatment as usual
What actually explained the benefit?	Reduced repetitive negative thinking and reduced perceived stress fully mediated the drop in symptom severity. Sleep quality did not improve.
What did Musonius Rufus already argue?	In “On Training” (Peri Askeseos), that the soul and body are one composite thing, so physical hardship — cold, hunger, exertion — is how courage and self-control actually get built

The Quick Version

A 2026 mediation analysis of 400 people with depression, anxiety disorders, PTSD, or insomnia found that six months of running relieved symptoms almost entirely by lowering rumination and perceived stress — not by improving sleep, which researchers expected to matter more. Musonius Rufus, Epictetus’s teacher, taught nineteen centuries earlier that a human being is a composite of soul and body, not a mind that happens to be riding one around, and that physical hardship is how the soul’s own capacities — courage, self-control — get trained. The 2026 data is oddly specific about exactly which mental capacities exercise touches, and it lines up with the ones he named.

What the 2026 Study Actually Found

This isn’t the ImPuls trial’s headline result — that came earlier, when the main trial showed the six-month running intervention outperformed treatment as usual on overall symptom severity across a transdiagnostic sample. What Frei and her colleagues did in this new analysis was go back into the data and ask the harder question: through what pathway. Feeling better after exercise is easy to believe and hard to explain. Sleep is the obvious guess — you run, you’re tired, you sleep better, everything downstream improves. That’s the hypothesis the researchers were testing against.

It didn’t hold. Across 10 outpatient sites in Baden-Württemberg, 400 physically inactive adults with major depressive disorder, agoraphobia, panic disorder, PTSD, or insomnia were assigned either to the ImPuls program — 30 minutes of moderate-to-vigorous outdoor running two to three times a week for six months, four weeks supervised and the rest self-directed with therapist check-ins, plus coaching on goal-setting and barrier management — or to standard treatment as usual, which in the German outpatient system already includes therapy and medication where indicated. At six and twelve months, the running group had significantly lower repetitive negative thinking and significantly lower perceived stress than controls. Those two variables, statistically, did almost all the work of explaining the reduction in overall symptom severity. Sleep quality, measured the same way at the same timepoints, did not budge more than it did in the control group.

That’s a specific, checkable claim, and it cuts against the tidy story most people tell about why exercise helps: burn energy, sleep deeper, feel calmer. The mechanism these researchers found is cognitive and perceptual, not restorative. Running didn’t make people more rested. It seems to have made them think in shorter, less repetitive loops, and made the same objective stressors feel smaller. The authors’ own framing, as reported in PsyPost’s coverage of the findings, is that changes in perceived stress and repetitive negative thinking function as key transdiagnostic mechanisms — meaning this pathway held across five different diagnoses, not just one. A depressed person, a person with panic disorder, and a person with insomnia were all getting relief through the same two channels, which is a stranger and more interesting finding than “exercise helps depression.”

Musonius Rufus Already Made This Argument, Without a Single Data Point

Musonius Rufus doesn’t get quoted as often as his student. That student was Epictetus — this site has covered how Epictetus turned enslavement into the sharpest account of freedom Stoicism ever produced, and it’s worth knowing that Epictetus built that account on a framework he inherited directly from Musonius, whom he mentions repeatedly in his own Discourses. Musonius himself was exiled twice — once under Nero, once again after the Pisonian conspiracy, to a bare island called Gyaros — which is a detail worth sitting with before dismissing his ideas about hardship as theoretical.

In a lecture collected under the title “On Training” (Peri Askeseos), preserved in Cora Lutz’s translation of his surviving Lectures and Sayings, Musonius makes an argument that reads almost like a rebuttal written in advance to anyone who thinks philosophy is a purely mental exercise. “Since it so happens that the human being is not soul alone, nor body alone, but a kind of synthesis of the two,” he says, “the person in training must take care of both, the better part, the soul, more zealously.” Note what he doesn’t say. He doesn’t say the body is irrelevant, or a lesser vessel to be managed and ignored while the real work happens upstairs. He says there are two kinds of training — one proper to the soul alone, and one common to both soul and body — and the second kind isn’t optional scaffolding. It’s where actual virtues get built.

Musonius is specific about what that common training looks like: “we discipline ourselves to cold, heat, thirst, hunger, meager rations, hard beds, avoidance of pleasures, and patience under suffering.” And he’s specific about what it does. The body gets stronger and more capable of enduring hardship — fine, expected. But he claims the soul gets trained too, through the exact same practices: courage, by patience under hardship; self-control, by abstaining from pleasure. Not courage trained by thinking bravely. Courage trained by being cold and staying there. This is the same logic behind why fear is supposed to be part of courage rather than its opposite — the virtue isn’t the absence of discomfort, it’s what gets built by repeatedly not flinching from it.

What Two Mechanisms Actually Explain Exercise’s Mental Health Benefit?

1. Reduced repetitive negative thinking. The tendency to loop on the same worries, replaying a conflict or rehearsing a worst case, dropped significantly in the running group relative to controls — and this reduction statistically accounted for a large share of the overall symptom improvement.
2. Reduced perceived stress. Not a change in how many stressful things happened to participants, but a change in how threatening those events felt once they happened — this also fully mediated the effect, independent of rumination.

Sleep quality was tested as a third candidate mechanism and didn’t clear the bar. If you’ve been telling yourself that exercise “works” mostly by tiring you out into better sleep, this particular trial doesn’t back that up.

Where This Doesn’t Hold Up

None of this makes Musonius Rufus a behavioral scientist, and the honest version of this piece has to say so directly.

The ImPuls mediation analysis is a well-powered, multi-site trial with a genuine active comparison — treatment as usual, not a waitlist — which is a real strength. But it’s still a specific population (physically inactive German outpatients, diagnosed and already engaged with the mental health system), a specific intervention (supervised group running with behavioral coaching, not solo jogging with no support structure), and a specific measurement window (six and twelve months). It says something real about mechanism within that trial. It doesn’t establish that any form of exercise, done any way, produces the same mediation pattern for everyone, or that rumination and stress perception are the only channels through which movement affects mood in general.

Musonius’s claim is a different kind of claim entirely, and stacking it next to a 2026 mediation model risks flattening what’s actually interesting about the overlap. He wasn’t describing a measurable psychological process. He was making a metaphysical argument about what a human being is — a synthesis, not a mind with an attached machine — and deriving an ethical program from that premise. He’d have had no way to distinguish “reduced perceived stress” from “reduced repetitive negative thinking” as separate variables, because he didn’t have variables. What he had was a conviction, argued from first principles and tested against his own exiles, that treating the body as irrelevant to virtue was a category error. The 2026 data doesn’t prove that conviction. It’s compatible with it in a way that’s genuinely interesting, which is a smaller and more honest claim.

It’s also worth saying plainly: this is not a case for running instead of treatment. The ImPuls intervention was studied alongside standard care in a system where participants already had access to therapy and medication, not as a replacement for either. Research on stoic-adjacent practices and the brain keeps turning up the same caveat — voluntary discomfort and structured discipline show real effects, but they’re additions to care, not substitutes for it, especially for PTSD and major depressive disorder, where untreated symptoms carry real risk.

A Practice Worth Trying

Neither Frei’s team nor Musonius wrote this as a five-step plan, but the shared shape of what they found points somewhere specific.

1. Treat the run as training for a mental skill, not a mood fix. If you’re lacing up expecting an immediate calm-down, you’re measuring the wrong outcome. The trial’s effect showed up over six months, through a change in thinking patterns, not a single post-run glow.
2. Notice what stops looping, not just how tired you feel. After a run, check whether the specific worry you started with is still replaying at the same intensity. That’s closer to the mechanism this study actually found than checking whether you feel sleepy.
3. Don’t substitute intensity for consistency. The ImPuls protocol was moderate, not punishing — 30 minutes, two to three times a week. Musonius’s list of hardships (cold, hunger, hard beds) reads extreme by comparison, but his own framing was about patience under sustained, ordinary difficulty, not occasional heroics. The same logic shows up in the case for daily mental micro-practice over occasional intense sessions — the training principle doesn’t change much between mind and body.
4. If sleep is the thing that’s actually broken, treat it as its own problem. This trial is a useful corrective here: running did not fix sleep in this sample. If insomnia is the primary complaint, this particular intervention isn’t the evidenced fix for it, whatever else it might help.

The Composite Was the Point All Along

Frei and her colleagues weren’t testing Musonius Rufus. They were testing why a running program helps people with five different diagnoses, and they found an answer more specific than “exercise is good for you” — it works through cutting rumination and lowering perceived stress, not through better sleep, and it works that way across diagnostic categories that don’t usually get grouped together. Musonius wasn’t running trials. He was arguing, from a bare island he’d been exiled to for saying inconvenient things to power, that a soul trained only in the abstract was a soul left half-built. Cold, hunger, hard beds, exertion — none of it was punishment in his account. It was where courage and self-control actually got manufactured, because there was no soul sitting apart from the body to train some other way. Nineteen centuries later, a mediation model built on 400 people’s psychiatric symptoms keeps landing on the same two capacities he named. He didn’t know why. He was right anyway.

---

This post discusses a 2026 clinical trial on exercise and psychiatric symptoms alongside a classical philosophical text. It isn’t a substitute for professional mental health treatment. If you’re managing depression, PTSD, panic disorder, or another diagnosed condition, talk to a clinician before changing your treatment plan — exercise in this trial was studied as an addition to care, not a replacement for it.