The page opens with a plain-language overview of what this body of research does and does not show, followed by every study cited across the site, listed alphabetically by first author. Tags indicate which section or practice each reference supports.
In Parkinson's, the cells that produce dopamine in a small region deep in the brain gradually die. Dopamine is what the brain's movement circuitry uses to start a movement and set its size. Without enough of it, movements become slow, small and effortful, tremor can appear, and walking can suddenly stop altogether — the experience known as freezing.
The part people notice first is often the strangest. Movements that used to happen by themselves — swinging an arm, taking a step, writing your signature — stop being automatic. They still work, but only if you consciously attend to them. This is why so much of what helps in Parkinson's is essentially an attention trick: counting steps, marching to a beat, stepping over a line on the floor.
The research on this page asks whether attention itself can be trained. If watching a movement, imagining one, or deliberately calming the body can reach the motor system by a route not dependent on the damaged dopamine circuit, these might be useful additions to conventional treatment.
Watching someone perform an action, and imagining performing it yourself, both activate motor areas of the brain without any movement occurring. The proposal in Parkinson's is specific: these routes run mainly through the outer layers of the brain — premotor and parietal cortex — which are relatively spared, and so may partly bypass the damaged circuitry deeper down (Caligiore et al., 2017; Abbruzzese et al., 2015).
That is a hypothesis about mechanism, and worth saying so plainly. It explains why the idea is attractive; it does not demonstrate that it works.
There is a wrinkle here that does not arise in other conditions. Parkinson's slows imagined movement as well as real movement: when people are asked to imagine a task and the time is measured, the imagined movement is slowed much as the physical one is (Heremans et al., 2012). The imagination is affected by the very thing it is being asked to treat.
The same study found something useful, though. When visual cues were provided during the imagery, the slowing reduced and the images became more vivid. Imagery ability also varies between individuals, appears to relate to symptom severity (Readman et al., 2023), and can be measured formally (Malouin et al., 2007). This suggests guided imagery for Parkinson's should be cued and structured rather than left to the person's own imagination.
A few things about that study are worth knowing before leaning on it too far. It tested a hand-reaching task and a block-transfer task, not walking — so it says nothing directly about gait or freezing. It measured a single session in a lab, not what happens after weeks of practice. And the researchers screened out one participant beforehand for poor imagery ability, which is a reminder that not everyone imagines equally well, and that this kind of screening is something a clinician would normally do before relying on imagery as a training tool (Heremans et al., 2012).
Timing may matter too, though here the evidence is theory, not proof. The body cycles between alertness and rest roughly every ninety minutes (Kleitman, 1963; Rossi & Nimmons, 1991), the drowsy state before sleep is linked to unusually vivid, spontaneous imagery (Mavromatis, 1987; Schacter, 1976; Ghibellini & Meier, 2023), and sleep helps consolidate a skill practised earlier in the day (Walker et al., 2002; Stickgold & Walker, 2007). None of this has been tested with imagery in Parkinson's — it is a reason to place a session at a particular time of day, not evidence that doing so helps.
The first trial combined motor imagery with physical practice in a group setting: twelve people received both, eleven physical practice alone, twice a week for twelve weeks. The combined group got significantly faster at several movement sequences over the course of treatment, where the physical-practice-only group did not. On the standard Parkinson's rating scale, the combined group also improved significantly on two subscales over the treatment period — but that is a before-and-after comparison within the treatment group alone; the group that got imagery was not shown to improve significantly more than the group that didn't, on that scale. The authors noted the approach extends practice time at low cost and negligible risk (Tamir et al., 2007). It was a small, unblinded pilot study.
Watching movement has shown short-term effects. A single session of watching video of repetitive finger movements increased spontaneous movement speed, still measurable two days later — and the comparison was against a rhythmic acoustic cue, an established strategy in Parkinson's, rather than against nothing (Pelosin et al., 2013). Combining observation with imagery influences the size of hand movements (Bek et al., 2019), and a home training programme built around this has been developed with patients themselves (Bek et al., 2021). A randomised trial pairing imagery with virtual reality reported gains in balance, motor function and daily activities (Kashif et al., 2022a).
A separate randomised trial found that autogenic training — a relaxation technique built on visual imagery and body awareness — added to physiotherapy improved UPDRS motor scores more than physiotherapy alone, at both eight and twelve weeks (Ajimsha et al., 2014). It was not among the trials pooled in the meta-analysis below, and its technique is closer to relaxation than to motor imagery, but it is a positive result on the standard clinical measure and belongs in the picture.
One further trial asked a different question: not whether imagery improves movement, but whether it holds on to gains that physiotherapy has already produced. Fourteen people completed fifteen sessions of physiotherapy and were then randomised either to ten sessions of mental practice alongside home exercises, or to home exercises alone. Over the following weeks the mental practice group held its walking speed and its dynamic gait score, while the group doing home exercises unsupervised slipped back toward where it had started (Monteiro et al., 2018). It is a small trial — seven people in each group — and it says nothing about freezing, but it asks a question that matters in a progressive condition, where holding ground is its own result.
Watching movement has also been studied directly for freezing of gait, with more encouraging results: reduced freezing after observation-based training, in one case alongside changes visible on brain scans (Pelosin et al., 2010; Agosta et al., 2017).
Against this sit three results that must be given equal weight. A randomised trial found action observation and motor imagery had no effect on balance or freezing of gait (Bezerra et al., 2022). An earlier multicentre randomised trial found mental practice no better than relaxation for mobility (Braun et al., 2011) — though since no group received neither, that result is equally consistent with both helping. And the most recent meta-analysis pooled the trials using the standard motor rating scale as their main outcome — six studies, 152 people — and found no difference between treatment and comparison groups, whether tested on or off medication (Pettenuzzo et al., 2025).
The picture is different, and in some ways stronger, for the calming side of this work. In twenty people with moderate to severe tremor, relaxation guided imagery reduced tremor dramatically on accelerometer measurement, and abolished it entirely for between one and thirteen minutes in fifteen of them. Tremor remained below baseline half an hour later, and participants reported benefit lasting several hours. Relaxing music helped less; relaxing without guidance did not help at all (Schlesinger et al., 2009). A later study from the same group looked at motor fluctuations, and its detail is worth having. Time spent in the “on” state rose from 47.7% to 62.8% after listening to a recorded guided imagery session, while relaxing music produced no significant change. Movement scores fell after both sessions but significantly more after the imagery. Overall quality of life did not change — but the emotional subscale improved, after the imagery only (Schlesinger et al., 2014).
Two cautions belong with that study, both raised by its own authors. There was no randomisation, the assessments were not blinded, and every participant did the music session before the imagery session, so an order effect cannot be ruled out. The music condition was passive listening while the imagery session was delivered in person by a therapist, which means the comparison is not cleanly imagery against music. The authors describe their own findings as preliminary, and the three-month follow-up did not reach statistical significance. It is a promising signal from a small pilot, not a demonstrated treatment effect.
One finding from that study sits outside the usual evidence questions and is worth reporting anyway. Three months on, fourteen of the twenty-one participants were still choosing to listen to the guided imagery recording; none had continued with the relaxing music. Whether people keep using a free recorded practice, unsupervised, is a fair question to ask of a site like this one — and it is the only published answer this bibliography contains (Schlesinger et al., 2014).
For mood, a randomised trial in JAMA Neurology found mindfulness yoga improved anxiety and depression more than stretching and resistance training (Kwok et al., 2019). A mindfulness programme has been associated with structural changes on brain scans (Pickut et al., 2013), and a small uncontrolled study reported improvements in anxiety, depression, cognition and some motor measures (Dissanayaka et al., 2016).
Voice work sits alongside this as an established, separately evidenced treatment, with benefits demonstrated at two years and measurable changes in brain activity (Ramig et al., 2001; Narayana et al., 2010; Fox et al., 2012).
The honest position has four parts.
First, on the standard measure of Parkinson's motor symptoms, the pooled evidence for imagery and observation is null (Pettenuzzo et al., 2025). Six trials and 152 people is a very small body of evidence, and the encouraging studies are mostly small and short.
There is also a limitation that belongs to the method itself rather than to the evidence for it. A therapist can watch someone perform a movement or confirm they are watching a video; nobody can verify from the outside that mental practice is being done correctly, or at all. Reviewers of this literature name this as a reason action observation is simpler to deliver reliably than motor imagery (Abbruzzese et al., 2015). Everything on this site is imagery, which is the harder of the two to get right unaided — and one reason the sessions are voice-guided rather than left to written instruction.
Second, the picture for freezing and balance is mixed rather than settled — and the mix does not favour this site. Training built around watching movement has reported reduced freezing (Pelosin et al., 2010; Agosta et al., 2017) — though in Pelosin's trial freezing improved in both groups immediately after training, and only the observation group still held that gain at four-week follow-up, while the one randomised trial that combined observation with imagery found no effect on either freezing or balance (Bezerra et al., 2022). The encouraging findings belong to a different intervention from the one offered here. Freezing is among the most distressing symptoms, and it would be wrong to imply this approach addresses it.
Third, the relaxation findings, striking as they are, come from small unblinded studies of an outcome — tremor — known to respond to attention and stress. That does not make the effect unreal; anyone with Parkinson's knows what stress does to their tremor. But its size should be treated cautiously.
Fourth, and most importantly, none of this substitutes for exercise or medication. The largest evidence base in Parkinson's rehabilitation covers physical exercise: a Cochrane review of 154 trials and 7,837 people, of which 60 trials (2,721 people) measured motor symptoms and 48 (3,029 people) measured quality of life, showing benefit across almost every exercise type studied (Ernst et al., 2024). Set against the six trials pooled in the imagery meta-analysis, the comparison is not close. The intervention with the strongest evidence here is moving, not imagining.
That review is also worth reading for what it says about the category nearest to this site. Mind-body training — tai chi, yoga and similar — is included as its own exercise type, and shows a small beneficial effect on both motor signs and quality of life, at low confidence: real, but ranking below dance and gait/balance training for movement, and below aqua-based training for quality of life. The review's broader conclusion is that doing some form of exercise matters more than which form is chosen (Ernst et al., 2024).
Two rather different claims are gathered on this site, and they deserve to be held to different standards.
Imagery and observation for movement are mechanistically plausible and supported by some encouraging small trials, but the pooled evidence on the standard motor scale shows no benefit yet. It is reasonable to try; it would not be reasonable to promise.
Imagery and mindfulness for tremor, fluctuations, anxiety and mood rest on firmer ground, including a randomised trial in a major neurology journal. Stress makes Parkinson's symptoms worse in a way people recognise immediately in their own bodies, and practices that reduce it address something real.
What both share is a favourable balance of risk. No harm has been reported in any of these studies. They cost nothing, need no equipment, and can be practised on days when a walk is not possible. Used alongside exercise, medication and your clinical team — not in place of them — they are reasonable to add.
This overview summarises published research and is not medical advice. Never change medication or treatment without speaking to your neurologist or Parkinson's nurse.