Life Vocations

The Flattening: Why So Many Settlements Fall Apart, and What Mediators Are Really There to Do

Watch what happens to two people the moment a relationship turns into a dispute.

A week ago, they were a contractor and a homeowner who liked each other, swapped photos of their kids, and agreed the kitchen was going to look great. Now one of them is “the guy who blew the budget and ghosted me,” and the other is “the client who keeps moving the goalposts and won’t pay.” Two whole human beings, each with a history, a family, a bad back, a reason for running late, have been quietly compressed into a single function. Debtor. Plaintiff. Betrayer. Opposing party.

I have come to think of this as the central, under-noticed fact about conflict. It does something to perception. Long before anyone misbehaves, the dispute itself narrows how each person sees the other. I call it flattening, and resisting it, in the room and in myself, is what I think a mediator is actually there to do.

That is a slightly heretical thing to say in a field that often measures itself by settlement rates. So let me make the case.

What conflict does to your eyes

Here is the uncomfortable part. The flattening is not a character flaw. It is closer to a factory setting.

Your mind is a categorization engine. Faced with the impossible job of treating every encounter as brand new, it sorts people into schemas and scripts to move fast (Fiske & Taylor, 2013). Most of that sorting runs on what Kahneman (2011) calls System 1: quick, automatic, cheap, and almost entirely below your awareness. In ordinary life, this is a gift. You do not have to relearn what a waiter is every time you sit down to eat.

Under conflict, the same machinery keeps running, just on a thinner diet. The other person still has all their roles (parent, neighbor, colleague, the friend who drove you to the airport), but those roles drop out of view. What stays in focus is the one role that hurt you. The other becomes an instance of the thing that did the damage.

Then it gets reinforced from a few directions at once.

Our attitudes start working overtime to protect us. One of the oldest findings in the field is that attitudes can serve an ego-defensive function (Katz, 1960). Once you experience someone as a threat, your view of them stops being a neutral snapshot and starts being a shield. The flat image is convenient precisely because it spares you the more complicated truth, which is that the person across the table might be a full human being whose complaint is, in some measure, fair.

The body joins in, too. When we feel threatened, and conflict registers as a threat, we slip outside what Dan Siegel (2012) calls the window of tolerance, the band of arousal where the thinking, empathic parts of the brain stay online. Push someone into fight-or-flight, or into shutdown, and their perception narrows automatically. They are not being difficult. Their nervous system has simply changed the camera’s resolution.

And if there is real trauma in someone’s history, the picture sharpens further. People who have been badly hurt often arrive already primed to read threat where it may not be (van der Kolk, 2014; Herman, 1992). They are not showing up with a clean lens that the conflict has temporarily smudged. They are showing up with a system already configured to view the other person as a greater danger. Understanding this saves a mediator enormous energy, because you stop interpreting the flattening as bad faith and start treating it as ordinary human equipment doing exactly what it evolved to do.

Stack all of that together (the quick categories, the protective attitudes, the narrowed nervous system, the old wounds), and you get a person who genuinely cannot see the human across the table anymore. That is the flattening. Naming it is the first move because you cannot resist something you have not noticed.

The part everyone skips: settlements signed by flat people don’t hold

Here is a scene every experienced mediator knows. You reach an agreement. Sometimes a beautiful one. Everyone shakes hands. Three weeks later, it has collapsed.

The usual autopsy blames the document. Too vague. Too aspirational. Not specific enough. And honestly, that critique is often correct as far as it goes. Loose agreements do tend to come apart.

But there is a deeper problem hiding under the drafting. The agreement is being asked to carry weight that the parties’ perception cannot support. The page got signed by people who still saw each other flat.

The decision science backs this up. Behavior tends to follow intention, intention follows attitude, and attitude is shaped by what you believe about the other party (Fishbein & Ajzen, 1975). If the underlying belief remains “this person is the problem,” the attitude stays hostile, the intention to honor the deal remains weak, and behavior drifts back to where it started. What looked like agreement at the table was often just compliance under pressure. And compliance is not commitment. Kelman (1958) drew that line decades ago, separating mere compliance from genuine internalization, and it is exactly the line that matters here.

This is why I get a little restless with the standard advice to find the interests beneath the positions. It is good advice. Surfacing interests really does loosen the grip of a fixed position and open up creative options (Fisher, Ury & Patton, 2011). But interests live inside people, and people in conflict are not the people they will be once the conflict eases. The interests you elicit from a flattened version of someone are a partial map. Build a deal on a partial map, and you get a partial deal.

The transformative and conflict-transformation traditions have been circling this for years. Bush and Folger discuss recognition as one of the deeper goods that mediation can offer, though the term often remains abstract. Lederach (2005) pushes further, arguing that durable peace requires moral imagination, the capacity to picture a relationship that does not yet exist. I would put it more plainly. Recognition and moral imagination are what become possible the moment the flattening lifts. The agreements that actually hold are reached by people who, even briefly, saw each other as fully human again.

A picture worth keeping in your back pocket

If you remember one image from all of this, make it this one.

In the 1960s, a psychiatrist named Stephen Karpman (1968) noticed that people under stress tend to fall into one of three roles. The Persecutor, the one-dimensional aggressor (“he is impossible”). The Victim, the one-dimensional sufferer whose agency has been written out of the story. And the Rescuer, the well-meaning helper whose rescuing quietly assumes the Victim can’t manage on their own. He called it the Drama Triangle. The roles rotate (today’s Victim is tomorrow’s Persecutor), but the rotation never dissolves the triangle. It feeds it.

Look closely, and every one of those roles is a flattened role. None of them lets a person be more than one thing at a time. The Victim cannot be both wronged and accountable. The Persecutor cannot be both responsible and understandable. The Rescuer cannot be both helpful and respectful of the other person’s competence. The triangle is a machine for making “both/and” structurally impossible.

Twenty-some years later, Acey Choy (1990) offered the way out, and it is elegant. Keep the same three corners, but let each one grow back into a fuller version of itself. The Persecutor becomes the Challenger, someone who asserts a real need or boundary without reducing the other person to an essence. The Victim becomes the Vulnerable, someone who names their pain honestly without surrendering their agency. The Rescuer becomes the Caring, someone who offers presence and help without taking over the other person’s responsibility. He called it the Winner’s Triangle.

That movement, from Drama Triangle to Winner’s Triangle, is the whole discipline compressed into one diagram. It is what restoring fullness looks like in the geometry of a three-party room. And notice that nothing gets thrown away. The grievance is still real. The boundary is still firm. The compassion is still warm. The contradictions are simply held together rather than collapsed into one side. Clinical supervisors already use both triangles to coach trainees out of drama and into something workable (McKimm & Forrest, 2010), which is the same job in a different room.

There is a catch, and it is about you.

Of the three roles, the one that reaches out and grabs the mediator is the Rescuer. It is the most seductive because it feels exactly like compassion from the inside. One party shows up with severe, sympathetic, clearly justified pain, and something in you wants to step in and fix it. The trouble is that from outside the triangle, rescuing confirms the Victim’s flat picture of themselves and ratifies the Persecutor’s “see, it’s all their fault.” It does the opposite of what it intends. The discipline is to stay in the Caring corner instead: present, resourced, genuinely warm, and still handing responsibility back to the person who owns it. That single distinction, between rescuing and caring, is one of the most consequential moves in the work.

How you actually do it

None of this is mystical. It shows up in a fairly ordinary technique, just aimed at a different target than usual.

Reframing is the clearest example. When a mediator restates “he is irresponsible” as “he submitted the report late, and that delayed the project,” she is doing something more than softening the language. She is converting an essence into a situation. “He is irresponsible” is a flat, fixed verdict. “He submitted the report late” describes a person who could, conceivably, act differently next time. Reframing is the linguistic version of restored perception. In Drama Triangle terms, it also quietly refuses the casting. “She is impossible” becomes “she has been firm about deadlines,” which turns a Persecutor back into a Challenger.

Marshall Rosenberg’s Nonviolent Communication (2015) runs on the same engine, just more systematically: observe what actually happened, name the feeling, point to the need underneath, and make a doable request. Each step pulls a person back from caricature toward agency. “The report came in Tuesday” is not “you are unreliable.”

The caucus is another tool, and a sneaky, powerful one. Pull one party out of the other’s line of sight, and you interrupt the reciprocal reactivity that keeps each flat image actively in use. In private, people will admit doubt, fatigue, and second thoughts that the joint room cannot safely hold. Your job in there is not to take their side. It is to help them re-meet their own complexity, because that is the precondition for re-meeting the other person’s.

It also helps to keep a short diagnostic running in the back of your mind. Fisher and Shapiro (2005) identify five core concerns that, when violated, reliably make people rigid: appreciation, affiliation, autonomy, status, and role. When a session suddenly goes cold, the useful question is usually not “what’s the issue” but “which of those five just got violated, and for whom.” Name that quietly, often in caucus, and you are working on what is actually driving the flattening instead of swatting at its symptoms.

And then there is the part nobody can outsource, which is your own nervous system.

You cannot restore fullness to others while you are losing it yourself. A mediator who has drifted out of her own window of tolerance will either over-identify with one party or harden into a brittle, procedural neutrality that is itself a kind of flattening. This is why the boring stuff matters: the breath before the session, the body scan, the grounding cue when the room spikes, the slower pace of speech that pulls everyone back toward the present. There is a real mechanism here, not just self-care language. A regulated nervous system can serve as temporary scaffolding for a dysregulated one (Schore, 2003). Calm, it turns out, is as contagious as panic. The mediator’s internal state is not prepared for the intervention. It is the intervention.

What ties all of it together is an old idea Carl Rogers (1961) gave us: congruence, the alignment between what you feel, what you know you feel, and what you show. A congruent mediator is a person the parties can actually read, someone who is plainly not performing. In a room where two people have flattened each other into cardboard, the steadiest gift you can offer is to be visibly, durably three-dimensional yourself.

What it looks like in the room

Picture two former business partners after a bitter split. The first comes in swinging: “He ran the company into the ground and then blamed me.” The second goes quiet, then quieter, then basically disappears into the chair. One has cast himself as the righteous accuser. The other has slid into Victim silence.

A settlement-only process would start grinding both toward terms while each still holds the other flat, and you can almost set your watch for when that deal unravels.

The discipline works differently and more slowly. Reframing turns the first partner’s accusation into a Challenger’s claim: “You needed decisions made faster than they were getting made, and that has cost you.” Reflective listening coaxes the second partner out of silence into something vulnerable and real: “I stopped saying what I thought because I figured it wouldn’t land anyway.” And the mediator, feeling the strong pull to rush in and protect the quiet one, stays in the Caring corner instead. Present. Warm. Still handing each man his own piece of the responsibility.

Does any of that guarantee a signed agreement? No. But it completely changes what an agreement, if it comes, is built on. One is built on two people still seeing cardboard. The other is built on two people who, for a few minutes, saw a person again.

I will be honest about the limits here. This is a framework drawn from practice and theory, not a controlled trial, and I hold it as a strong hypothesis rather than a proven law. Not every relationship can or should be restored. Some marriages end in divorce, some partnerships in dissolution, and that is not always a failure of the work. Sometimes, the most honest thing the discipline does is help two people dissolve something gracefully rather than enforce a fiction. Restoring fullness is not the same as forcing reconciliation. It is making sure that whatever the parties decide, they decide it as whole people rather than as the flattened versions the conflict handed them.

Why this is worth the trouble

So here is where I land, and it is the thing I would say if we were finishing our coffee.

The center of this work is not paperwork, not an interesting excavation, not the elegant engineering of a deal, even though it is all of those things on a busy Tuesday. The center is restoring fullness to relationships that conflict has flattened. The flattening is real and recognizable, with roots in how we categorize, how we protect ourselves, how our bodies handle threat, and how our oldest relationships taught us to see. It is what makes settlement without fullness so fragile. And resisting it, in the parties and first of all in yourself, is the actual job.

The strange, almost paradoxical payoff is this: the more fully you can see each person, the less you have to do procedurally to get them to see each other. Fullness is contagious. So is its absence. The mediator who has practiced this internally carries it into the room, and for an hour or two, the room becomes a place where being seen as a whole person is possible again.

Whether the parties leave with a signed agreement is, in some real sense, secondary. They leave having been met as more than the role the conflict assigned them. That experience is itself the beginning of the repair. The agreement, if it comes, is just the paperwork that follows.

And the next time you watch a dispute turn two human beings into a debtor and a creditor in front of your eyes, you will know what is actually happening. Somebody in the room has to refuse to see flat. It might as well be you.

The full academic paper this piece is drawn from, “The Mediator’s Discipline: Restoring Fullness in Flattened Relationships,” with its complete references and figures, is available at the end of this blog entry.