The question
If lasting change is real, why can't it be explained by one mechanism that works for everyone?
Ask what makes change last, and several familiar answers appear.
Repeat a behaviour until it becomes habitual. Change identity so the new action feels consistent with who you are. Build motivation. Regulate physiological and emotional states. Strengthen rewards. Find social support. Redesign the environment.
Each answer captures something important, although the strength and relevance of the evidence differ across outcomes and contexts.
Repetition can make action less dependent on deliberation. Self-interpretation and expectations of consistency can connect behaviour with identity. Motivation can mobilise effort. Physiological and emotional states can alter which actions are accessible. Reinforcement can shape repetition. Relationships can support persistence. Environments can preserve—or remove—the opportunity to act.
The title rejects the idea that one mechanism universally explains all lasting human change. It does not deny that one mechanism may dominate a particular change under defined conditions.
A mechanism is a process proposed to produce or sustain an outcome. It can be real, important and common without operating in every person, every context or every phase.
Lasting change can emerge through different pathways and be sustained by different causal configurations.
Important does not mean universal
Consider habit.
Repeated action in a stable context can make behaviour more automatic. This may reduce the attention and effort required each time it is performed. Habit can therefore make a powerful contribution to persistence.
But habit alone may not explain why the action began, why opportunities for repetition remained available or why the behaviour survived a major change in context. It may not explain how someone returned after interruption or why one repeated behaviour became stable while another disappeared.
This defines the boundaries of what habit explains; it does not make habit unimportant.
The same principle applies elsewhere. Motivation may dominate when change begins but become less central once action is embedded in a routine. Identity-related expectations may stabilise some behaviours while remaining peripheral to others. Environmental design may reduce friction without producing a reason to act. Social support may be essential in one setting and unnecessary in another.
A mechanism may be common without being universal, dominant without being exclusive, or necessary within one configuration without being necessary in all configurations.
Evidence that a process contributes to lasting change does not establish that every lasting change depends on it in the same way.
Different roads can reach a similar outcome
Imagine several people who maintain a defined level of regular physical activity over the same extended period.
One joins a team and develops reliable social commitments. Another changes how they commute so movement becomes part of the day. A third develops competence and begins enjoying the activity. Someone else repeats a simple routine until it demands less deliberation. Another increasingly interprets continued activity as consistent with who they are.
These are possible pathways, not verified causal biographies. Any person's change may involve several of them.
The example holds a broad outcome constant for illustration. Real research would need to define its frequency, intensity, setting, stability and timescale more precisely.
A similar observed destination can nevertheless be compatible with different causal routes. Researchers call this possibility equifinality: different starting conditions or pathways can converge on a similar outcome.
Equifinality does not mean that every route is equally likely, effective or durable. Two people can meet the same activity threshold while differing in flexibility, dependence on context and ability to recover after disruption.
Nor does observing a shared outcome establish that equifinality produced it. The concept describes a possible causal structure and frames a research question. Identifying the actual pathways requires additional longitudinal, experimental or convergent evidence.
The shared outcome is not proof of a shared mechanism.
The same mechanism can lead elsewhere
Causal plurality also operates in the other direction.
Social accountability may support persistence when it creates belonging and reliable commitment. Under different conditions, it may create pressure or withdrawal. Its influence may disappear when the group dissolves.
Identity-related commitment may stabilise action when people attend to options that fit their self-understanding and receive social confirmation for acting consistently. It may also become rigid, remain symbolic without altering behaviour or weaken after an experience interpreted as identity-inconsistent.
Environmental restructuring may make action easier in one setting but produce temporary change if the person later enters another environment.
Researchers call this possibility multifinality: similar starting conditions or processes can lead to different outcomes.
Multifinality does not make outcomes random. Divergence may reflect differences in prior learning, current state, resources, timing, social meaning or competing constraints.
Like equifinality, multifinality is an organising principle rather than proof of the pathways operating in a particular case. It tells researchers what kind of causal structure to investigate.
Mechanisms do not carry fixed effects independent of what they encounter.
Starting change is not the same as keeping it
Time creates another source of plurality.
The process that helps someone begin may not be the process that makes the change durable.
Initiation may depend on urgency, motivation, opportunity, social invitation or immediate reward. Continued action may rely more on planning, manageable effort, reinforcement and repeated opportunity.
Greater stability may later arise through skill, environmental embedding, changed expectations, identity integration or structural support.
These are analytical phases, not compulsory stages through which everyone must pass. Their boundaries can overlap, and the same mechanism may contribute at several points. Separating them simply prevents early improvement from being mistaken for a complete explanation of maintenance.
Recovery introduces another temporal question.
Illness, travel, stress, changed responsibilities or environmental disruption can interrupt behaviour that had previously been stable. What matters afterward may include preserved skill, flexible planning, practical support or an interpretation that treats the disruption as temporary rather than total failure.
An interruption can weaken or end a changed pattern. It does not automatically prove that the change was never durable. A single return does not prove durability either. Recovery contributes evidence only when meaningful re-entry and subsequent persistence are demonstrated over the relevant period.
Durability can include the capacity to recover, not merely uninterrupted performance.
Mechanisms work in configurations
Mechanisms rarely operate as isolated ingredients.
Motivation may increase readiness, while environmental opportunity permits action. Repetition may then increase skill and reduce expected effort. Those changes can influence later motivation.
Behaviour and identity-related processes can also shape one another across time. Earlier actions provide evidence used in self-interpretation. That interpretation changes how later options are perceived. Later behaviour can reinforce or revise the interpretation again.
This is not simultaneous self-causation. Earlier activity changes the conditions under which later activity occurs.
Relationships, environments and physiological states can enter the same process. Social support may preserve access during a difficult period. Environmental structure may reduce demands on self-regulation. Physiological state may constrain access to a cognitive strategy when it is needed.
A mechanism can become effective because another process enables, amplifies or preserves it.
Redundancy means that more than one pathway may support the same function. Compensation occurs when activity elsewhere changes after one pathway weakens or is altered. Either may preserve a new pattern—or help restore an old one—but neither should be assumed in an individual case without evidence.
A causal configuration is not a name for everything that might matter. It is a bounded, testable set of specified processes and conditions proposed to work together for a defined outcome, phase and context. It should exclude factors that add no explanatory value and generate evidence that could count against it.
A configuration can be simple or complex. It may contain one dominant mechanism together with the conditions that allow that mechanism to operate.
The alternative to one universal mechanism is therefore not an undefined web in which everything causes everything. It is a specific causal proposal.
Pluralism is not "everything matters"
Causal pluralism becomes empty if every possible influence receives equal status.
A serious mechanism claim must specify the outcome being explained, the relevant person or population, the phase of change and the timescale involved. It should describe the proposed pathway, the conditions that enable or limit it and the evidence that could weaken the claim.
Confidence may grow when several forms of evidence converge.
Temporal ordering can show that the proposed mechanism changed before the outcome. Experimental manipulation can test whether altering a target affects what follows. Mediation analysis can examine evidence consistent with an indirect pathway. Moderation analysis can show whether an effect changes across conditions. Replication can test whether a pattern survives new samples and settings.
Convergence increases confidence in a bounded mechanism claim. It does not guarantee an exhaustive explanation or eliminate alternative pathways.
A measured mediator may change alongside the outcome without constituting the complete mechanism. A temporal sequence may still reflect a shared cause. A pathway estimated across a group may not describe what happened within every individual.
Personal accounts contribute evidence that formal measures can miss: meaning, interpretation, context and the sequence as experienced. They can reveal valuable candidate processes. Experiential authority over what change felt like is not, however, identical to complete causal identification.
Pluralism should therefore make questions more precise:
- Which mechanism?
- For which outcome?
- At which phase?
- Under which conditions?
- Supported by what evidence?
The possibility of several pathways should raise the standard of explanation, not lower it.
From master keys to causal configurations
Once the search for one master key is abandoned, the task is not to collect an endless list of mechanisms.
It is to determine which processes are active, what functions they perform, when they matter, which conditions enable them and how they interact. It is also to identify what stabilises the changed trajectory and what supports recovery after disruption.
A useful account must be plural without becoming indiscriminate. It must integrate mechanisms without reducing them to one hidden master cause. It must remain sensitive to people and contexts without assuming that individual pathways can be read directly from averages or narratives.
One mechanism may dominate a bounded case. Several people may share a common pathway. Certain processes may prove widely important. A causal configuration can therefore be relatively simple while still including the conditions that make its dominant mechanism effective.
The need for integration follows from causal plurality, but the integration cannot be declared in advance. It must preserve meaningful differences among mechanisms, phases, levels and conditions rather than erasing them.
Lasting change becomes more intelligible when we stop searching for one explanation that owns every case and begin identifying the bounded configuration that sustains the change in question.
Lasting change is not unlocked by one master mechanism, but sustained by the particular configuration of processes that keeps a different future possible.