The Mind Science of Practice

This page is built as a reference rather than an argument. It sets out the terms, the numbers and the diagnostic questions that recur whenever somebody tries to work out whether a practice of attention produces anything at all. The reasoning underneath is laid out elsewhere and is not repeated here, because a table you can consult is worth more than a position you have to agree with first. Everything below is meant to be quoted, argued with, or used as a checklist when reviewing a year of your own records. If you read one section, read the diagnostic checklist near the end. It is the part that does the work, and it applies equally to a claim made by a grimoire, a podcast, a research paper or a friend.

The four claims that do the work

Practitioners make claims about ritual that collapse, under inspection, into four separate propositions. Keeping them apart is the most useful single skill in this area, because they have different evidence and different likelihoods.

A claim about attention, a claim about physiology, a claim about interpretation and a claim about external causation are not four versions of one question. Most disputes in this field are disputes where a person is answering one of them and their opponent is answering another.

• Attention claims, that focused attention alters what a person notices and therefore does
• Physiological claims, that repeated ritual changes arousal, breathing and tension
• Interpretive claims, that a symbolic frame changes the meaning of an event after the fact
• External claims, that something outside the practitioner is altered, which has no clear evidence

Terms used precisely

A ritual is a repeated sequence performed with a stated intention. Gnosis is sustained, non-abstract contact with an experience. Sigilisation is the reduction of a statement into a glyph. Autonomic refers to the involuntary nervous system, which covers heart rate, breathing, sweating and pupil size.

A null result is a run in which the pre-registered condition was not met. Base rate is the rate at which a result would occur with no working at all. Criterion conflict is a decision that stalls because two good values point in different directions rather than because information is missing.

Numbers worth knowing before you start

A five-option draw has a twenty percent hit rate by chance, so an eight-run streak is a one-in-fifty-five event that happens to roughly one person in fifty-five who tries. Any method with a twenty percent base rate produces convincing streaks in a large enough number of attempts.

A five per cent effect on a monthly binary outcome needs a sample in the high hundreds to detect, which is the arithmetic that kills most overnight manifestation claims. A ten per cent effect on a daily count is detectable in a few weeks, which is why adherence studies find results and intention studies do not.

The open-label placebo problem

In an open-label study participants are told what they are receiving. If belief were necessary for a physiological effect, open-label designs should produce no effect. Some do not, which complicates the simplest version of the everything-is-belief account.

The complication is narrower than it is usually claimed to be. It shows that belief is not always required for a measurable outcome. It does not establish that any particular practice does anything, and it does not license the conclusion that expectation is unreal.

Predictive processing, and what it does not explain

Perception modelled as hypothesis management offers a mechanism by which expectation changes experience without anyone consciously deceiving themselves. Sensory data is fitted against priors, and the priors contribute to the output.

This is a real account of how expectation shapes experience and it explains the phenomenology of intense practice. It does not explain anything beyond the practitioner, which is the direction most people mean when they cite it.

What the autonomic findings support

Repeated, scheduled, attention-heavy activity reliably changes heart rate variability, respiration pattern and muscle tension. The effect is well documented and does not require the participant to expect any particular outcome.

This is the most defensible claim in the entire area, and it is also a much smaller claim than usually made. Treat ritual as a structured physiological practice and it supports your practice. Treat it as evidence for external causation and it does not.

1. Repetition on a fixed schedule changes basal arousal more reliably than intensity does
2. Breath and posture interventions produce effects within a single session
3. Expectation effects and habit effects are difficult to separate in most designs
4. None of this distinguishes between frameworks that differ about external causation

Where the research actually fails

Three recurring weaknesses. Small samples, in a field where practitioners are recruited by self-interest. Unblinded assessment, where the person scoring the outcome knows what was done. And outcome drift, where the stated target is revised during the run.

None is fatal individually and together they explain most of what looks like a body of supporting evidence. A study that fixes all three would be worth reading, and very few exist, which is a statement about the incentives in the field rather than about magic.

A diagnostic checklist for a claim

Before believing any result, run it through six questions. Was the condition specified before the run. Was there a stated deadline. Was the base rate calculated. Was the outcome measured by somebody who did not know the assignment. Were nulls recorded. And was there an attempt count.

A claim that survives all six is worth taking seriously even if you are sceptical. A claim that fails three of them should not be repeated, and the fact that a claim is emotionally moving is not one of the six questions.

1. Was the condition written before the first working
2. Was the review date fixed in advance and honoured
3. Was the chance rate of the result calculated
4. Was the outcome assessed by someone unaware of the working
5. Were null entries recorded on the day they happened
6. How many attempts were made, including the abandoned ones

Terms that mislead

Manifestation, in most usage, names an outcome without specifying a mechanism and is therefore not a term at all. Vibrations is a physical quantity being used as a metaphor while implying a literal process. Energy work inherits a vocabulary from a physics that does not describe the practice.

The cost of these words is not dishonesty. It is that they make claims unfalsifiable while appearing to be precise, which is the worst combination available. Replacing them with a sentence naming who does what by when costs nothing and immediately becomes checkable.

Practitioner vocabulary, translated

Charging a sigil means focusing attention on a glyph for a stated interval with a stated focus. Dismissing means completing a physical action that ends the working. Forgetting means putting the object out of reach so the attention loop closes. Grounding means returning to ordinary sensory detail after an unusual state.

None of these translations changes anything the practitioner does. They only make the practice describable in terms somebody else can check, which is the minimum price of finding out whether it works.

The short version to keep

Ritual reliably changes your state of mind. That is documented, it does not require belief, and it is worth building a practice on. Whether anything changes outside you is unresolved, and the people who claim certainty in either direction are usually defending a position rather than reporting a finding.

Everything else in this reference is detail. If you take one thing from it, take the checklist and apply it to the next claim you are inclined to believe, including the ones made by people you like.

Frequently Asked Questions

Is a paid app the right way to run an attention or ESP experiment?

A good app removes friction from recording, and friction is the thing that ruins most self-run experiments. The features that matter are automatic timestamped capture, a way to mark a null without it feeling like a failure, and an export in a format you own. Anything that offers interpretation, scoring or a verdict is adding a judgement to a process that is supposed to supply data, and that is the point where the tool stops measuring and starts advising.

What sample size do I need to detect a small effect?

For a monthly binary outcome and a five per cent change, you need a sample in the high hundreds, which is not achievable in self-directed practice and is the reason most overnight claims are untestable by the people making them. For a daily count and a ten per cent change, a few weeks is enough to see something. Choosing a measurable daily count rather than a monthly yes-or-no is the single most useful experimental design decision available to an individual practitioner.

How does gnosis differ from visualisation?

Visualisation is generating an image, which is a cognitive task that many people find quite easy. Gnosis is sustained non-abstract contact with an experience, which is closer to attention without narration. The second is harder, less pleasant, and the tradition regards it as the higher discipline. Practically, most beginners who think they are doing gnosis are visualising, and the correction is to drop the narrative and stay with a single sensory detail.

Is there a card test I can run to check whether I have any ability at all?

There are several, and all of them are worth running once for the calibration experience rather than for a verdict. With five options and a twenty percent chance rate, a good run is short, so the useful number is the attempt count rather than the best streak. A single impressive result tells you about your memory for the best result and nothing about your ability. If you want to test yourself, pre-commit to a fixed number of trials and report all of them.

What should I look for when a psi training programme claims a result?

Ask for the attempt count, the chance rate, the number of nulls, and whether anybody who was not the participant scored the outcomes. A programme that cannot answer those four questions is not reporting a result. Beyond that, check whether the claim is about scoring above chance on a task with a low base rate, because that is the single most abused statistic in this entire field and the arithmetic behind it is not difficult to work out yourself.