Toward an Integrated Model of Lasting Change

A person decides to begin writing. They choose a subject, open a document and produce a page.

An intention has become an action. But that does not yet tell us whether a writing practice has formed, what might allow it to continue or what will happen when the conditions around it change.

This problem appears across many forms of human change. One different response is not yet a stable pattern. Repetition can continue without becoming resilient to interruption. A temporary stop does not necessarily mean that everything previously learned has disappeared.

To explain lasting change, we must therefore look beyond the moment when behaviour first becomes different. We need to examine what made another response available, what happened when it was enacted, how later conditions changed and whether the developing pattern became supportable, adaptable or recoverable over time.

No single mechanism can answer all of those questions.

What an integrated model must explain

Theories of change emphasise different processes. Some focus on motivation and intention. Others examine reinforcement, habit, identity, self-regulation, relationships, environmental conditions or wider systems. A major review found dozens of distinct theories across the behavioural and social sciences, illustrating the absence of one accepted complete account. Davis and colleagues, 2015

Integration cannot mean placing every possible influence into one enormous list. That would replace "one mechanism explains everything" with the equally unhelpful conclusion that "everything matters."

A useful model must instead distinguish the functions that different processes can perform.

One process may help make a response available. Another may help convert intention into action. Others may alter what happens after the action, support its repetition or make it easier to resume after interruption. A mechanism that helps change begin need not be the one that allows it to continue.

The framework developed here organises five broad explanatory movements:

• the configuration from which change begins;
• the availability and enactment of another response;
• the consequences and learning that follow;
• the embedding and stabilisation of a developing pattern;
• adaptation and recovery when conditions change.

These movements are not universal stages or necessary components of every change. They identify functions that may need to be investigated. They can overlap, recur, operate in parallel or carry different importance in different cases.

An environmental change, for example, may make a behaviour available before motivation changes. Performing that behaviour may then alter skill and expectation at the same time that repetition begins to affect identity and routine. The process need not follow one clean line.

The model is therefore not a formula. It is a framework for understanding how a different response may become possible, repeatable, supported and recoverable.

Change begins from an existing configuration

No act of change begins from nowhere.

Before someone sits down to write, they already have a history with writing. They possess some degree of skill. They may expect the work to feel absorbing, difficult, exposing or pointless. Their body may be rested or exhausted. They may have a clear task, an uninterrupted hour and an accessible workspace—or none of them.

Their present possibilities are shaped by previous attempts, current resources, relationships, routines, responsibilities and the meaning the activity already carries.

The COM-B framework proposes that behaviour emerges through interacting conditions of capability, opportunity and motivation. A person needs relevant physical and psychological capacities, sufficient social and environmental opportunity, and processes that energise or direct behaviour. The framework also allows behaviour to change those conditions in return. Michie, van Stralen and West, 2011

These categories do not provide a complete causal explanation. They help expose why wanting an outcome cannot be considered separately from whether a person can act and whether the conditions permit action.

We can describe the relevant conditions at a particular moment as a starting configuration. But giving them this name is not enough. An explanation must still show what a specific condition does.

Does it change whether the response is accessible? Does it alter expected effort, cost or consequence? Does it remove an opportunity, strengthen a competing response or make one course of action more credible?

This protects configuration from becoming a container into which every conceivable influence is placed.

Relations among conditions may matter as much as their isolated presence. Skill may not produce action when the opportunity is absent. Available time may not help when the next step remains undefined. High motivation may overcome substantial effort once without doing so repeatedly.

Complex-systems approaches are useful because they direct attention towards such interactions, histories and changing relationships. They also caution that a process observed on average across a group may not operate identically within every person. Heino and colleagues, 2021

A starting configuration conditions what can happen next. It does not dictate it. A small number of decisive constraints may matter more than a long inventory of weak influences.

Before action, there must be availability

A response can be possible in principle without being readily available in practice.

Someone may be capable of writing and sincerely want to do it, yet writing remains difficult to access in the relevant moment. The task may be undefined. Time may be fragmented. Necessary tools may be missing. Competing responsibilities may dominate attention, or the expected emotional cost may be unusually high.

This is why intention and action must remain separate.

Research on the intention–behaviour gap shows that people do not always enact what they intend. Intention matters, but its translation into behaviour varies with conditions including opportunity, planning, perceived control, self-efficacy and the stability of the intention. Conner and Norman, 2022

Several states that are often collapsed are therefore different:

• a response can be possible;
• it can become more or less available;
• a person can desire or intend it;
• they can prepare for it;
• they can enact it;
• the action can produce—or fail to produce—the intended result.

Availability is graded. An action can remain technically possible while being improbable, effortful or difficult to sustain.

Preparation may alter that gradient. A defined task, an opened document or an available period of time can make writing more accessible. These are examples of changes in behavioural conditions, not instructions that guarantee action. Preparation remains distinct from enactment, just as enactment remains distinct from its outcome.

When sufficiently relevant conditions align, another response may become more available and sometimes possible. There is no universal threshold, and not every condition must be optimal. Different configurations may support the same action.

Once action occurs, it can begin changing the conditions surrounding the next opportunity.

Action changes the next moment

Suppose the person writes a paragraph.

The paragraph matters as an outcome, but the action also produces consequences. The writer may experience progress, relief, difficulty or uncertainty. They practise the skill. The subject may become clearer. Another person may respond. The expected effort of returning may rise or fall.

Action produces consequences that can be registered through experience, learning, bodily response, social reaction or environmental change. Intention alone cannot produce all of this information.

Research identifies several plausible pathways through which enacted behaviour may affect later behaviour. These include reinforcement, skill acquisition, changing beliefs about capability and consequences, self-regulation, social influence and environmental change. The pathways are supported to different degrees and should not be mistaken for one experimentally established feedback mechanism. Carey and colleagues, 2019

The process unfolds through time:

• a response occurs;
• it produces consequences;
• those consequences may alter skill, affect, expectation, meaning, relationships or environmental conditions;
• the changed conditions shape a later opportunity to act.

An earlier action can therefore affect a later action without causing itself.

The consequences are not necessarily supportive. Writing may make the next session more approachable, but it may also reveal that the project is poorly defined. Praise may encourage continuation, create pressure or make little difference. An action can strengthen the developing pattern, strengthen a competing response, produce mixed effects or leave the relevant conditions largely unchanged.

Feedback matters, but favourable reinforcement cannot be assumed.

From repetition to embedding

If writing continues, repetition may gradually alter how the activity is organised.

Practice can improve skill and fluency. Uncertainty may decrease. The writer may become better able to begin, tolerate an imperfect first version and recognise a workable next step. An activity that once required extensive preparation may become more accessible.

When a response recurs in sufficiently stable contexts, cues may also become associated with it. A particular place, time or preceding event can begin prompting the response more directly. Habit research supports this role for repeated behaviour in stable contexts, but repetition alone does not guarantee automaticity. Gardner, 2015

Nor does habit formation follow one fixed schedule. A systematic review and meta-analysis found substantial variation across people and behaviours. Claims that every behaviour becomes habitual after the same number of days are therefore misleading. Singh and colleagues, 2024

Repeated behaviour may also contribute to self-interpretation. Someone who keeps returning to the page may increasingly regard writing as something they do. That interpretation can affect what they notice, expect and consider credible in the future.

Identity is not a separate force directing behaviour. It remains revisable and develops through action, interpretation, relationships and context. Research suggests that behaviour can inform identity while identity may support later behaviour, but neither process is automatic or necessary in every case. Verplanken and Sui, 2019

The pattern can also become embedded in its surroundings. Writing may become linked to a routine, workspace, set of tools, deadline or shared expectation. This embedding may be intentional or incidental, extensive or partial, robust or fragile.

The crucial shift is not simply that the action has occurred many times. It is that repetition may have changed some of the conditions supporting what happens next. Skill, cues, expectations, self-interpretation, relationships or the environment may now make the response differently available.

Much direct research on habit and maintenance concerns health behaviour. Its distinctions can clarify wider forms of change, but identical mechanisms should not be assumed across writing, health, relationships and structurally constrained behaviour.

Stability comes from aligned support

A developing pattern can receive support from several sources.

A writing practice might be supported by skill, manageable effort, a clear project, access to tools, recurring opportunities or expectations involving another person. A compatible self-interpretation might make returning more credible. An alternative place to write might reduce dependence on one environment. A familiar route back might limit the consequences of an interruption.

Theories of behaviour-change maintenance reflect this plurality. A systematic review identified five recurring explanatory themes: maintenance motives, self-regulation, psychological and physical resources, habit, and environmental and social influences. It found no single settled theory that explains maintenance completely. Kwasnicka and colleagues, 2016

The relevant supports can interact. Greater skill may reduce effort. Reduced effort may make repetition more likely. Repetition may strengthen contextual cues. Social expectations may help preserve an opportunity to act.

This does not mean that every support matters or that adding more always improves durability. One decisive constraint can outweigh many weak supports.

Some supports can perform overlapping functions. If a practice depends entirely on high motivation, it may become vulnerable when motivation falls. Skill, accessible routines or supportive relationships may sometimes provide another route to the same relevant function.

Such redundancy reduces dependence on one condition only when the alternative support is genuinely functional and remains available under the disruption concerned. A second support that fails under the same conditions adds little protection.

Stability can therefore be understood as a property of how relevant supports are organised. It is not a substance added to the behaviour, and it does not make the pattern independent of the processes sustaining it.

Nor is stability the same as rigidity. A response that continues only while every condition remains unchanged may be consistent but fragile. Lasting change need not preserve one exact performance forever.

Changed conditions can provide further evidence about how the pattern is supported.

Disruption can reveal how change is supported

The writer travels, becomes ill, encounters a heavier workload, loses access to the usual workspace or receives discouraging feedback.

The practice may continue in roughly the same form. It may continue less frequently. The person may change when, where or how they write while preserving the relevant activity. They may stop temporarily and later return. Or the pattern may weaken until the earlier arrangement becomes re-established.

These outcomes are not equivalent.

A lapse is a bounded interruption. It is not automatically a full reversal. Relapse research—much of it drawn from health and addiction contexts—supports distinguishing a temporary lapse from a sustained return to an earlier pattern. That evidence cannot be transferred unchanged to every domain, but the distinction prevents one interruption from becoming proof that nothing changed. DiClemente, 2022

Within this model, recovery means meaningful re-entry followed by persistence appropriate to the outcome being evaluated. Wanting to return is not recovery. One isolated action may initiate re-entry, but it does not establish that the pattern has been restored.

Adaptation is different again. Someone may no longer be able to write for an uninterrupted hour but may preserve the practice in a changed form. The behaviour is not identical, yet a relevant function continues.

Disruption therefore provides evidence about some of the pattern's supports and dependencies. It is not a complete or neutral measure of durability. What can be inferred depends on the kind of disruption, the outcome, the observation period and the evidence available.

Disruption is also not a required test. Severe stress should not be imposed to prove that change is real. But when conditions change naturally, the response can reveal which supports remained available, which were replaceable and which had been carrying more of the pattern than was previously visible.

A framework for questions, not a universal formula

To judge whether change lasted, we must first specify what changed.

A different behaviour, a developed skill, an altered expectation, a revised identity, a relationship pattern and an environmental arrangement are not interchangeable outcomes. Each may change across a different timescale and require different evidence.

Context matters too. A writing practice may continue at home but disappear during travel. It may survive changes in motivation while remaining dependent on protected time. That does not make the change unreal, but it defines the conditions under which it has become durable.

There is no universal duration after which all change becomes lasting. The relevant period depends on how often the outcome has an opportunity to occur, the variation expected within it and the forms of disruption or reversal that matter.

Measurement shapes the conclusion. Self-reported identity, observed behaviour and environmental change answer different questions. A group-level association does not automatically describe what changed within one person. Repeated and intensive longitudinal measurement can reveal fluctuation and temporal order, but measurement alone does not establish causation. Dunton, 2019

The framework therefore returns us to four central questions:

• What became differently possible or available?
• What did action change about the conditions of later action?
• How did the developing pattern become supported or embedded?
• What happened when relevant conditions varied?

For the writer, the answer might involve increased skill, reduced uncertainty, accessible tools, a recurring opportunity and a workable route back after interruption. For change involving health, relationships or structural constraints, the decisive configuration may be very different. The writing example cannot determine those pathways.

Across domains, the explanatory arc remains provisional. Conditions may make another response available. Action may change later conditions. Recurring cycles may become embedded. Relevant supports may stabilise the pattern, while adaptation or recovery may preserve or restore it when circumstances change.

These are possibilities to investigate, not stages every person must complete.

That is why this is a model toward integration rather than a final theory. It does not prescribe one pathway, make every mechanism equally important or claim that change can be understood apart from the outcome, person, context and timescale involved.

Lasting change is not a moment preserved unchanged, but a reorganised pattern that can keep generating, supporting and restoring a different way of living.