A critical distinction in couple therapy, when working systemically, is understanding what it means to say that relational systems conserve themselves. This directly shapes how change is understood, how intervention is approached, and how the system responds in the moment. Without this, there is a drift toward explanations such as defensiveness, manipulation, or resistance, which mislocate the problem in the individual and move toward blame or pathologising either the couple or one of the individuals.

Every therapeutic model carries an implicit theory of change, but that theory is incomplete if it does not also explain persistence. How relationships maintain themselves. How patterns endure. Why the relational system does not change even when it produces distress. Why individuals struggle to do something different despite intention, effort, and care. And importantly, why difficulty changing is not a reflection of a lack of commitment to the relationship.

Relational systems conserve their pattern of interaction. What is retained is not specific content, but the way interaction unfolds between partners. This is the pattern (Bateson, 1972; Dell, 1982, 1985).

What is experienced by the individuals as repeated failure, arguments that go nowhere, repair attempts that collapse, or conversations that circle back is, from a systemic perspective, the successful maintenance of that pattern. The couple may experience distress, stuckness, or dissatisfaction. That is accurate. But the system is not breaking down. It is continuing in the only way it currently can.

There are no dysfunctional systems in this sense. There are systems that produce distressing experiences. The sense that the relationship is not working is the lived experience of a pattern that has stabilised in a restrictive way.

What is conserved is not what is said, but how interaction proceeds. This can be seen in the sequence itself: how something is noticed, how meaning is assigned, how emotion is triggered, how behaviour follows, and how that behaviour shapes the next response. Not as a linear chain, but as a continuous loop. Over time, this loop settles into a recognisable way of interacting.

This is visible in the moment-to-moment choreography between partners. Who moves toward and who pulls back. Who defines and who responds. Who escalates and who shuts down. It is carried in implicit rules about what can be said, how it can be said, and what happens next. It is expressed through roles that are repeatedly taken up, and through rhythms where interaction accelerates, collapses, or loops back.

The couple and the pattern are not the same. The couple is what we see: two individuals, each with their own history and experience. The pattern is what shapes what happens between them. It does not sit inside either person. It exists in how their responses link together over time.

From a constructivist perspective, relational systems are structure-determined. What happens at any given moment is shaped by the current pattern, not by external input in any direct or instructive way (Maturana & Varela, 1980, 1987; von Glasersfeld, 1995).

What one partner says does not determine what the other will do. What the therapist introduces does not determine how the interaction will shift. These are perturbations. Events enter the system, but they do not specify the outcome. The response depends on the existing pattern. What is noticed, what is ignored, how something is interpreted, and what response becomes available are all constrained by it.

This is why the same intervention produces different outcomes in different couples, and why within the same relationship, different situations lead to similar outcomes. The system does not respond to reality as it is presented. It responds according to how it is already structured to perceive and process it.

Maturana’s concept of autopoiesis sharpens this (Maturana & Varela, 1980). An autopoietic system produces and reproduces the processes that constitute it. It is not maintained from the outside. It generates itself.

Applied to relationships, this means that interaction does not simply follow a pattern. It recreates the conditions that make that pattern likely. Each moment participates in this process. Perception leads to meaning, meaning to emotion, emotion to behaviour, and behaviour to responses that reinforce the original perception. This loop does not just repeat. It regenerates the interaction.

Stability is maintained by continuously recreating the same pattern from within. This is visible in how each partner’s actions and responses, often unintentionally, keep the interaction moving in its familiar form.

This is what Maturana refers to as operational closure (Maturana, 2002). The system is not closed in the sense of being isolated. Partners affect each other, and external events matter. But these influences do not determine the outcome. They are taken up and transformed according to the existing pattern. Input is filtered and reshaped so that it fits.

This has a direct consequence. Disconfirming experiences do not produce change on their own. A moment of connection does not undo a pattern of distance. An attempt at repair does not reorganise the relationship. These experiences are absorbed and interpreted in ways that allow the pattern to continue.

The system does not reject change. It absorbs it. Intervention is translated into what is already familiar. This is why couples selectively take in what fits and discard what does not. The same process happens between partners. This selectivity reflects a structural constraint, not a deliberate choice.

Conservation is not a defensive layer added onto the system. It is the ongoing process through which the system maintains itself. Structure determines what can happen, and autopoiesis ensures that whatever does happen contributes to the continuation of the pattern.

The problem is not conservation itself, but the experience created by the particular pattern that is being maintained.

This becomes most visible at the point of intervention.

When something different is introduced, when a partner attempts to respond outside the usual sequence, or when the therapist interrupts the interaction, the system does not reorganise in a linear way. It responds in a way that restores what is familiar. Because it is structure-determined, the perturbation is processed according to existing constraints. Because it is self-producing, the response contributes to the same pattern continuing.

This is what is experienced as pushback.

It needs to be located precisely. The pushback is not resistance. It is not defensiveness, lack of motivation, or unwillingness to change at the level of the individual. It is not a sign that the couple is not ready or not committed. It is the system doing what it is structured to do.

The system shifts locally in order to remain the same.

This can be seen directly in interaction. A partner softens, and the other escalates. One withdraws, and the other intensifies pursuit. A new behaviour appears briefly, but the sequence reorganises around it, pulling both partners back into familiar positions.

From within the relationship, this is experienced as failure. Change does not last. Efforts collapse. The interaction returns to what it was. From a systemic perspective, this is conservation in action.

This is why first-order change is insufficient (Watzlawick, Weakland, & Fisch, 1974). First-order change refers to adjustments within the existing pattern. Changes in communication, increased expression, or behavioural efforts that do not alter how the interaction unfolds. These may feel meaningful, but they are absorbed and reshaped to fit what already exists.

In relational systems, this often shows up in how behaviour is interpreted. A neutral act is pulled into an existing meaning, and the same response follows. The pattern continues.

Second-order change operates at a different level (Watzlawick et al., 1974). It involves a shift in how the interaction is structured. What previously followed automatically no longer does. The constraints that shaped perception, meaning, and response begin to loosen. New responses become possible, not because they have been learned or decided on, but because the system now allows them.

Second-order change does not come from insight alone, and it is not sustained by effort alone. It emerges when the system is repeatedly disrupted at the point where it recreates itself and can no longer continue in the same way.

It involves a shift in how reality is experienced in the relationship. Perception, meaning, and behaviour change together. When this stabilises, what exists is no longer the same relationship.

This is the goal of therapy. Change in the pattern. Change in how the interaction unfolds.

This is why therapy is not about teaching skills, but about restructuring interaction so that the pattern itself changes. Roles shift. The function of behaviour changes. The distribution of influence changes. The management of intensity changes.

And the experience of the relationship changes with it.

It begins to feel different to be in it.

As this new pattern stabilises, the work shifts. The task becomes supporting the couple to navigate stress, conflict, and difference without reverting to the previous way of interacting, and to continue responding in ways that reinforce the new pattern until it sustains itself.