Attention & Behavioural Self-Engineering

We propose revisiting the methodologies used in products for productivity and focus, sleep tracking, physical activity, nutrition, and other parts of everyday human behaviour.

Most of these methodologies treat the object of observation as an end metric: how long someone slept, how far they walked, what they ate, how long they worked, or how many tasks they completed. Measurement ends with a description of what happened.

We have not found products that treat events in a person’s life as changes to the probability architecture of future behavioural expression, and that build intervention at the level of that architecture.

A record of behaviour contains no mechanism for how that behaviour arose. It therefore does not determine what must change for that behaviour to change systematically.

The proposed object of measurement is not behaviour itself, but the probability architecture of its expression.

Tonicity

At every moment, a person exists in a concrete material configuration: a particular nervous-system and bodily state, a particular room, time of day, social context, information environment, set of available objects, commitments, skills, and already-formed behavioural patterns.

This configuration does not determine the next action, but it makes the potential actions available to a person unequally probable.

We call such a configuration a tonic state, and tonicity the degree to which the current probability architecture corresponds to the behaviour a person needs to express.

That changes the meaning of observation.

If someone intended to go to bed at 11:00 PM but went to bed at 2:14 AM, the relevant fact is not only a delay of three hours and fourteen minutes.

We need to reconstruct which sequence of events made staying awake until 2:14 AM more probable than going to sleep at 11:00 PM.

They drank coffee at 7:30 PM. At 9:00 PM they began working. At 10:40 PM they received a message. At 11:05 PM they were still at the computer. The light remained bright. The phone was beside the bed.

Each event matters not in isolation, but as a change to the subsequent probability architecture of behaviour.

A person’s life event can therefore be understood as:

a change in the state of a person or their environment that can alter the probability distribution of subsequent behavioural expression.

NASCW+I

Nutrition, Activity, Sleep, Connections, Work/Study, and Infrastructure are not six independent objects of optimisation or six “healthy habits”. They are a coordinate system for registering events that can alter the architecture of subsequent behaviour.

Nature Release is interested not in an event’s category by itself, but in the transition:

Event → State Change → Shift in the Probability of Subsequent Behaviour → Behavioural Expression

From this follows the fundamental distinction between execution and construction.

Execution and Construction

Execution is the expression of behaviour that is an agent’s immediate goal.

I want to run → I run.

Construction is the expression of behaviour whose goal is to change the probability architecture of another behaviour.

I want to run in the morning → I arrange to run with my brother → I prepare my clothes → I stop eating late → I stabilise my sleep time → I remove the need to make extra decisions in the morning → I run.

Rather than repeatedly trying to make a person execute a desired behaviour, Nature Release should increase the probability that they will construct the conditions for its later expression.

This is self-engineering in the literal sense. A person becomes both the object and the subject of engineering.

Construction is itself behaviour and is therefore not outside the model. It also has its own probability architecture. Nature Release becomes one element of that architecture: an environment that raises the probability of constructing one’s own tonic state.

The result is a recursive structure:

NR → Probability of construction ↑ → Tonic state changes → Probability of desired behaviour ↑ → Behavioural expression → New observations → Further construction.

Discipline, Willpower, Enthusiasm

Discipline causes nothing.

Discipline is not an explanation: “He trains because he is disciplined.”

This is a tautology: observed behaviour is explained by the name of that same observed behaviour.

Discipline is the retrospective degree to which actually expressed behaviour corresponds to previously specified behaviour over a chosen period.

Planned to train 12 times → trained 11 times → high discipline relative to that plan and period.

Why the result was 11 out of 12 is a question of tonicity. Discipline comes after behaviour, not before it.

At this point we can also set aside the need to introduce willpower as an independent causal variable.

The phrase “He did it thanks to willpower” contains no mechanism.

If an observed “act of will” is itself behaviour, we must explain the probability of expressing that behaviour. Moving the explanation one level back by using the word willpower does not solve the problem of causality.

Until an independent mechanism is found, willpower can remain a phenomenon that has not yet been operationalised, rather than a variable in the model.

Enthusiasm is different. For now, we can describe it phenomenologically:

Enthusiasm is a short-lived state of heightened subjective attractiveness of an activity or its expected outcome, accompanied by a greater probability of immediately beginning the relevant behaviour and/or constructing plans to continue it.

This definition deliberately does not explain what causes enthusiasm. That is appropriate: first, the existence of a reproducible phenomenon is established; then its predictors are studied.

The decay of enthusiasm is especially interesting: in the evening the probability of construction and execution is high; in the morning enthusiasm disappears, yet the architecture created in the evening may continue to affect the probability of behaviour.

That is why:

Enthusiasm can initiate construction. Tonicity can outlast enthusiasm.

Nature Release Concept

The central unit of analysis in Nature Release is not a “habit”. It is a probability transition:

There was a state S₀ in which B₀ was most probable.

The user wants B₁.

Which changes ΔS are needed so that, in state S₁ = S₀ + ΔS, behaviour B₁ becomes sufficiently probable?

Nature Release then observes the result and updates its own model.

The result is, in practice, HADI applied to the person:

Observe → Hypothesise → Construct → Express → Measure → Update.

Not:

Track your life.

Build better habits.

Become disciplined.

But:

Observe what produces your behaviour. Construct what produces the behaviour you want.