Most conversations aren't only exchanges of information. They're also interactions between people who arrive with expectations about what happens next.

Someone expresses uncertainty and the other person offers an answer. Someone becomes upset and the other person tries to calm things down. One person challenges and the other begins defending. None of this is unusual. Conversation requires us to interpret what another person is doing and decide what kind of response makes sense.

Sometimes, though, the same responses begin appearing often enough that they form a pattern.

One person becomes the caretaker and the other becomes the person who needs care. One becomes the skeptic and the other the defender. One explains while the other asks questions. One initiates difficult conversations while the other waits for them to happen. Eventually the roles can become so familiar that both people enter them with remarkably little prompting.

I've come to think of this as a role mirror.

I don't mean that one person secretly projects a role that the other is compelled to reflect, or that relationships contain some hidden mechanism trying to complete a particular shape. The simpler explanation is interesting enough: what one person does changes the responses that make sense to the other, and those responses feed back into what happens next.

If I routinely take responsibility when something becomes uncertain, people around me may learn to wait for me to do it. Their waiting gives me another reason to step forward, and my stepping forward gives them another reason to wait. After enough repetitions, we can easily forget which part came first.

Now it looks as though I am simply “the responsible one.”

Perhaps I am unusually responsible.

Perhaps I've also been participating in a relationship that keeps giving me opportunities to practice that role.

The distinction matters because roles can become mistaken for identities. Someone who is confident in one relationship may become hesitant in another. A person who constantly explains things at work may prefer to follow when they're among friends. Someone who seems demanding around one person may behave very differently around somebody who responds to them differently.

We don't need to conclude that none of these behaviors reveal anything about the person. Repeated behavior is evidence. It simply isn't the whole person.

Relationships participate in what becomes visible.

This can make certain conversations feel strangely repetitive. The subject changes, but after a few minutes everyone seems to be standing in familiar places. One person expresses uncertainty, another starts instructing, the first resists being instructed, and now the second explains more forcefully because the explanation didn't work.

Each response makes some sense as a reaction to the previous one.

Together they recreate a conversation neither person particularly wanted.

That is where the mirror metaphor becomes useful to me. It asks us to look not only at what each person is saying, but at what each response makes easier or harder for the other person to do next.

If I respond to every expression of uncertainty with advice, am I gradually creating a conversation in which the other person's main options are to accept my advice or defend themselves against it? If someone repeatedly brings me problems and I immediately solve them, are they assigning me the role of solver, or am I helping create that role by stepping into it before anything else can happen?

There may not be a clean answer.

Usually both people are participating somehow, although not necessarily equally and not necessarily with equal freedom to change the interaction.

That qualification matters. A manager and employee don't enter a conversation with the same authority. Neither do a parent and young child. Social expectations, economic dependence, history and personality can all make one person's apparent “choice” considerably more expensive than the other's.

So noticing a role doesn't automatically free us from it.

It gives us another hypothesis about what is happening.

Sometimes that is enough to try something small. Instead of answering the question immediately, I can ask what the other person thinks. Instead of defending myself against criticism, I can ask which part they actually want changed. Instead of automatically calming someone who is upset, I can find out whether they want comfort, help or simply room to be upset.

The point isn't to refuse the familiar role.

Sometimes the familiar role is exactly right.

If someone trusts me because I've reliably helped them think through difficult problems, there is nothing inherently unhealthy about becoming the person they call when they need that kind of help. Roles make cooperation possible. Organizations depend on them, families develop them, friendships contain them and even casual conversations rely on temporary expectations about who is doing what.

The problem begins when the role becomes difficult to inspect.

Perhaps I keep helping long after help has become dependency because “helper” is now part of how I understand myself. Perhaps someone keeps challenging me because disagreement has become the only way our relationship knows how to create engagement. Perhaps two people keep reenacting caretaker and crisis roles because neither has discovered what the relationship looks like when there is no crisis to organize around.

Those possibilities are worth examining.

They aren't conclusions we should impose merely because the pattern looks familiar.

This is especially important because we can bring old role expectations into new relationships. If I've repeatedly encountered authority as controlling, I may interpret ordinary direction from a new manager through that history. If I've learned that expressing uncertainty causes other people to take over, I may expect the same response from someone who hasn't actually done that.

The role mirror can exist partly in my interpretation.

That doesn't make the experience imaginary. It means the pattern I'm noticing may contain information from more than one relationship.

Reality gets another turn.

What does this person actually do when I respond differently? Does the interaction change, or does it strongly pull toward the old arrangement? If I stop automatically supplying the expected response, does the other person adapt, become confused, escalate or simply continue normally?

Now we're learning something.

This is one reason changing our own behavior can be informative even when we're not trying to change the other person. A small variation can reveal how much of the pattern depended on our participation.

Sometimes the entire interaction changes surprisingly easily.

Sometimes it doesn't change at all.

Sometimes we discover that what felt like an expectation from the other person was largely an expectation we were carrying ourselves. Other times we discover that the role really is being reinforced, and departing from it produces immediate pressure to return.

Those are different situations.

The mirror is useful only if it helps us distinguish them.

I also think this gives us a better explanation for why the same person can appear so different across relationships. We aren't fixed collections of roles waiting for somebody to activate them, but neither do we enter each interaction without history. We bring habits, expectations and practiced responses with us. Other people bring theirs. The relationship that develops creates opportunities for some behaviors to repeat more often than others.

Over time, repetition can make the resulting pattern feel inevitable.

It rarely is.

That doesn't mean every relationship can become anything we want. History matters. Power matters. Temperament matters. Some patterns are deeply established, and some relationships may not survive an attempt to change them.

Still, it can be useful to separate two questions that are easy to collapse.

Who is this person?

What does this relationship repeatedly invite this person to become?

Neither question can completely answer the other.

That is what I want the role mirror to preserve. Not a theory that every interaction secretly seeks completion, and not a technique for escaping roles, but a reminder that behavior happens between people as well as within them.

Sometimes I really am being asked to play a familiar role.

Sometimes I'm volunteering for it before anyone asks.

Usually the interesting part begins when I can no longer tell which one is happening.

Then I can change something small and see what the relationship does next.