What maintains or regenerates a chronic pattern?
The maintained feature might be continuous pathology. It might instead involve altered threshold, response size, recovery, recurrence, learning, exposure, treatment, or several processes at once.
Continuing project · dynamic measurement
How can we tell what a mind-body practice changed?
A symptom score or resting biomarker can show that something differs. It usually cannot show what produced the difference, how long it lasts, or where in the course of chronic pain and illness the change occurred.
This project follows inputs, responses, recovery, recurrence, and longer-term outcomes so that competing explanations can be compared.
Conceptual response profile
The shape of change carries information
At a glance
Follow a practice from its actual dose and timing to immediate and delayed outcomes. The methods below separate symptoms, function, physiology and disease activity, compare rival explanations, and include burden, nonresponse and harms.
Page updated 28 September 2026 · Measurement project
Two linked questions
What can a mind-body practice actually change?
A practice may change experience, interpretation, behavior, exposure, physiology, function, disease activity, or none of them. Those are different findings.
The lived intervention
Where a mind-body practice enters
Programs such as Primal Trust and Somia do not deliver one isolated technique. They teach people how to notice and interpret symptoms, direct attention, regulate arousal, choose activity or rest, engage with care, and repeat those responses inside a relationship and explanatory system.
Lived
“I feel steady enough to begin.”
Morning walk · after shortened sleep
Practice
not yetchallenge beginsresponse observed
Mind-body practice beginsOrient → lengthen exhale → reframe sensation → choose pace
Measured
symptom burdenphysiological responseavailable functionunobserved comparison
Modeled
θ, z₀
u(t) ↑
x(t) Δ
a(t) ↑
y(t), z(t+1)
Model underneath
x(t+1) = f(x(t), z(t), u(t), c(t), a(t); θ) · y(t) = h(x(t), z(t), u(t), c(t), a(t); θ) + ε
Why the whole sequence matters
Less catastrophizing could alter distress and action without changing pathology. Pacing could reduce exposure to a trigger. Continued activity could create useful learning or provoke delayed worsening. Relationship and explanation could increase expectancy and adherence. A symptom change cannot identify which route occurred, so the study must observe the instructions, the person’s response, and the downstream course.
From course advice to lived experience
Our proposed Program Observatory would extend the audits into participation: helping people follow what they encounter, understand and try, and return to their experience when deciding what comes next. With separate consent, those records could also improve the audits and inform more adaptable courses.
See the participant journey and an example record →The questions below connect that purpose to measurement. They distinguish what a course teaches from how someone receives it and what follows.
- What was taught?
- The source version, the practice it describes, why it is supposed to help, and any qualifications or alternatives.
- What did it mean to the person?
- What they understood, accepted, questioned or hoped for. Hearing an explanation, believing it and finding it useful are different.
- What was feasible?
- What they intended to try, with the capacity, skills, time, resources and support available. Difficulty participating can have practical causes.
- What actually happened?
- The practice, its timing and any adaptations, alongside rest, medical care and changing circumstances. Assigned practice and actual practice can differ.
- What changed, and when?
- Immediate and delayed changes in symptoms, daily function and burden, with disease activity assessed separately where measured. Relief alone cannot establish disease change.
- What changes next?
- What led someone to repeat, adapt, pause, stop or seek care, and what would count against the explanation. Nonresponse should allow reconsideration.
These connections are research questions. Someone can use a practice while questioning its story, and changes in health or circumstances can change what is possible. Accurate course descriptions, useful information and health effects each need their own evaluation.
Optional questions for one experience
- What did you understand the advice to mean?
- What made it possible or difficult to try?
- What happened next, including later, and what changed your next choice?
These are possible interview prompts. A person could skip a question or correct our interpretation. Recording, access and reuse would need their own agreement; a private account would not become public by default.
How would we keep measurement useful?
Choose a question, a relevant observation window and what result could change a decision. Follow delayed costs as well as immediate effects, and consider other care, illness changes and circumstances that could explain the result.
Question wording, timing and repetition can themselves affect attention, experience and burden. There is no universal check-in schedule. Skipped responses remain missing; they do not mean good health or poor motivation. Any future study would need a tolerable schedule, easy pause and exit, and a way to assess the effects of monitoring itself.
Reading these resources does not require contributing health data. This section describes proposed research; it does not collect responses or offer symptom monitoring.
The measurement problem
One trajectory can fit several explanations
A flare that settles after a practice is worth observing. On its own, that sequence does not identify the process that caused improvement.
- 01Ongoing pathology or exposure
- 02Sensitization or a learned response
- 03A change in treatment, input, or context
- 04Altered threshold, gain, or recovery
- 05Natural history or regression to the mean
- 06Reporting or measurement error
Dynamic phenotype
Measure the course, not only the average
Repeated observations can describe how a person or system responds across time. The features below help turn a broad account of “regulation” or “reset” into measurements that competing models can address.
- Baseline
- The state before an episode or intervention, including anticipatory change.
- Threshold and latency
- What is required to evoke a response, and how soon it begins.
- Gain and peak
- How strongly the system responds and the maximum burden reached.
- Recovery
- How quickly and completely the response resolves.
- Delayed effects
- Overshoot, rebound, or a response that appears after the initial event.
- Recurrence
- How responses accumulate, repeat, or become easier to evoke.
- Context sensitivity
- How the response changes across setting, expectation, activity, or exposure.
- Cost and harm
- The functional, physiological, treatment, or participation burden of the response and the study.
A reproducible response profile can sharpen a hypothesis. It does not by itself establish an attractor, a mechanism, or a clinical indication.
Concrete observation
Measure only what can answer the study question
No study needs every item. The point is to select measures that can locate a proposed change and distinguish it from the most credible alternatives.
- What happened
- Time, trigger, activity dose, environment, sleep, food, medication, treatment, social setting
- What the person did
- Practice minutes, breathing pattern, attention target, interpretation, pacing, movement, checking, avoidance, care seeking
- What the person experienced
- Pain, fatigue, swelling, dizziness, affect, threat, bodily location, intensity, meaning, agency
- What the body did
- ECG and heart rate, respiration, blood pressure, temperature, electrodermal activity, sleep and activity
- What the condition did
- Condition-specific measures such as wheal and flare, event-timed mediators, orthostatic response, delayed post-exertional change, pain summation
- What life did
- Daily function, participation, work, relationships, treatment burden, adverse effects, relapse, and durability
Outcome layers
Keep connected changes distinguishable
Experience, behavior, physiology, and disease processes influence one another. They can also move differently.
A study should be able to show whether a practice changed pain, daily function, a physiological response, disease activity, treatment burden, or some combination of them.
- 01
Practice and dose
What a person actually did, in what sequence, with what intensity and adherence.
- 02
Context and exposure
Treatment, environment, relationship, explanation, expectation, activity, sleep, and other changing inputs.
- 03
Experience and symptoms
Pain, fatigue, distress, bodily experience, agency, and other reported effects.
- 04
Behavior and function
Movement, avoidance, participation, daily activity, work, and social function.
- 05
Physiology
Autonomic, endocrine, sensory, motor, sleep, and other regulatory measures chosen for the question.
- 06
Condition biology
Immune activity, tissue state, exposure, infection, injury, or other disease-specific processes.
- 07
Course and burden
Disease activity, damage, treatment burden, adverse effects, durability, and longer-term change.
Study sequence
Build evidence in stages
The sequence begins with observation and measurement quality. Later stages add intervention and prediction only when the question, population, and safeguards are defined.
- 01
Define the construct and its rivals
State what is expected to change, what else could explain the observation, and what result would count against each account.
- 02
Observe natural episodes
Follow flares, recoveries, ordinary practice, and relevant exposures before adding a study manipulation.
- 03
Test the measurements
Establish whether the measures are feasible, repeatable, sensitive to change, and meaningful to participants.
- 04
Separate timescales
Distinguish fast responses, slower regulatory configurations, stable person characteristics, and changing context.
- 05
Use bounded perturbations
Where clinically appropriate, use preauthorized challenges or practice sequences to reveal threshold, response, and recovery dynamics.
- 06
Study the intervention
Measure actual dose, sequence, co-interventions, adherence, durability, nonresponse, adverse effects, and exit.
- 07
Test predictions again
Evaluate selection rules and model predictions in new observations, new participants, or a held-out portion of the data.
Timescale
Immediate response
Attention, appraisal, breathing, autonomic activity, sensation, and the first behavioral choice.
Episode and recovery
Peak burden, delayed flare, symptom spread, sleep, function, medication use, decay, and recurrence.
Learning and course
Actual practice dose, changing exposure and activity, response parameters, generalization, relapse, disease activity, and harm.
Model comparison
What would change our minds?
Active inference and attractor language are useful only when they make better predictions than a simpler account.
A stronger dynamical account
Repeated response parameters, restoring tendencies, thresholds, path dependence, or timing effects predict new episodes and intervention effects better than simpler models.
A narrower explanation
Ongoing pathology, exposure, ordinary learning, treatment change, autoregression, or natural history explains the observations as well or better.
First study families
Concrete places to begin
These studies can begin without assuming that chronic illness is one type of state or that every mind-body program works through the same mechanism.
- 01
Map natural episodes
Follow triggers, symptoms, physiology, function, treatment, and delayed recovery before imposing a challenge.
- 02
Follow a real program
Observe participants before, during, and after the full multiweek dose, including practice, coaching, co-treatment, dropout, nonresponse, harm, and durability.
- 03
Separate the ingredients
Compare attention, explanation, breathing, exposure, movement, relationship, and sequence under credible matched conditions.
- 04
Test prediction
Ask whether response parameters and context predict held-out episodes or treatment effects better than simpler pathology, exposure, learning, or natural-history models.
Decisions
What could a study tell us?
A useful study could show whether a practice changes experience during practice, behavior and exposure, a repeatable response parameter, disease activity, the longer-term course of illness, or nothing measurable. It should also help identify for whom, at what dose and timescale, with what burden, and with what risk of harm.
Program architecture: Between Symptom and Disease: Measurement Roadmap. The multimodal pilot is an early feasibility example and does not establish clinical efficacy.