Someone is sitting across the table talking about a colleague, and behind the listening a different conversation is already in progress. The one where it ends. The sentence they will use. Which friend gets told first. Where the books go.
Nothing has happened. By every available measure the relationship is fine, which is part of what makes the scene so hard to argue with: there is nothing to refute. And a good evening has gone somewhere else anyway, to a loss that has not occurred, and come back with nothing.
45%
of variance in a couple's current relationship quality that the best self-report models could account for
18%
the same models, same couples, predicting quality at the end of the study
11,196
couples across 43 longitudinal datasets
37,761
people in a meta-analysis of why relationships end
Mourning someone who has not gone anywhere
The clearest research on grieving a loss before it arrives does not come from relationships. It comes from end-of-life caregiving, where a 2016 systematic review by Mette Nielsen and colleagues pulled the literature apart to ask whether anticipating a death helps afterwards.
The answer was split in a way that matters here. Preparedness was associated with better functioning after the loss and less complicated grief. High pre-loss grief was associated with worse outcomes, including prolonged grief. Two things that sound identical, pointing in opposite directions.
That distinction carries a lot of this piece, so the limit on it should be said plainly: the transfer is an argument rather than a finding. Those caregivers were facing a death that was going to happen, and nobody has run the equivalent study on people who grieved a relationship that then lasted another nine years. What carries across is the shape. Being ready and being in mourning are not the same activity, and only one of them is paid for in advance.
The rehearsal cannot be cancelled by being right. If the ending comes, the months already spent inside it are not refunded, and if it does not come, they are not refunded either.
The price is specific and not the dramatic kind. It is the dinner that happened while attention was elsewhere, the holiday photographed from a slight distance by someone already filing it as a thing that will later hurt to look at. Presence is the only material a relationship is made of, and bracing spends it on a version of the future that no amount of spending improves.
Why the mind rehearses an ending it cannot prevent
Worth getting this part right, because the usual framing treats the rehearsal as a malfunction and it is not one.
Grupe and Nitschke's 2013 review in Nature Reviews Neuroscience makes the case that anxiety is organised around uncertainty about future threat, and specifically around the way uncertainty disrupts any ability to avoid a threat or soften it. Prediction, in that frame, is the next best thing to control. If the outcome cannot be changed, the one remaining move is to not be taken by surprise by it.
There is a more precise account of what the rehearsal buys, and it has experimental support. Michelle Newman and Sandra Llera's contrast avoidance model proposes that chronic worriers are not primarily afraid of bad outcomes. They are averse to the jolt from feeling fine to feeling terrible, and worry works by holding negative feeling at a steady level so nothing has far to fall. Llera and Newman found that pattern directly in 2014: worry reduced the likelihood of a negative emotional shift, and reduced the likelihood of a positive one too. People with generalised anxiety found a negative emotion induction harder to cope with after relaxing than after worrying.
So the rehearsal is not stupid. It is a trade, and the terms are legible. A continuous low ache for months, against a smaller drop on a day that may never arrive.
Then the awkward detail about the drop. Eastwick, Finkel, Krishnamurti and Loewenstein tracked people in relationships over months and compared what they predicted a breakup would do to them against what it did to those it happened to. The prediction overshot. People in love systematically overestimated how distressed they would be afterwards, which means the premium is paid against an inflated estimate of the thing being insured.
Rehearsing is not the same as preparing
The defence of bracing is usually that it is just being realistic, and one body of research takes that defence seriously rather than dismissing it. Julie Norem and Nancy Cantor described defensive pessimism in 1986: people who set low expectations, mentally run through everything that could go wrong, and then perform perfectly well, better in some cases than if they had been talked into optimism. Norem and Illingworth later showed that forcing defensive pessimists to think positively made their performance worse. Imagining the disaster is a legitimate strategy and should not be moralised at.
But look at what it is a strategy for. A presentation, an exam, a difficult conversation. Things with a date and a performance, where the run-through produces a changed slide, a rehearsed answer, an earlier train. It converts into something.
Rehearsing an ending converts into nothing, and the test is easy to apply. After an hour of it, is there anything a person could do differently tomorrow.
The preparing column is not a metaphor. What the rent would cost alone. Whether the savings cover three months. Which two friends get called, by name, with at least one of them spoken to about something real this year. One thing in the week that does not involve the relationship. All of it is genuine insurance, all of it takes a few hours, and none of it requires a rehearsed conversation. The clue that bracing is not preparation is that it has been running for a year and the list is still empty.
How bracing builds the distance it is braced for
Saying that this becomes a self-fulfilling prophecy is true and useless. The mechanism is traceable and has four steps.
The first is withdrawal, and it is small. Sandra Murray, John Holmes and Nancy Collins described a risk regulation system in which confidence about a partner's regard governs how much dependence a person is willing to risk. When confidence drops, the system protects by reducing exposure: less invested, less asked for, less said about the future. Nobody experiences this as a decision.
Second, the withdrawal is visible and its reason is not. What reaches the other side of the table is not fear. It is someone slightly harder to reach, who has stopped suggesting things, who answers a question about next summer with a shrug.
Third, the other person adjusts. Perceived partner commitment is one of the strongest predictors of relationship quality in the data, and it runs both ways. Almost nobody keeps investing at full rate into someone who appears to have stopped, and the adjustment is usually unconscious.
Fourth, the adjustment gets read as confirmation. The cooling that arrives is real, and it is also partly supplied.
There is a harsher branch, and the evidence for it is a diary study rather than a model. Geraldine Downey and colleagues followed dating couples in 1998 and found that people who expected rejection behaved, after ordinary conflicts, in ways that produced it: their partners reported more dissatisfaction and more thoughts of ending things in the days afterwards, and those relationships broke up more often. The effect appeared for women in that sample and not for men, which is a real limit on the finding rather than a detail to skip. Striking first does the most damage fastest, because it does not wait to be read.
| The move | What it is doing from inside | What the other person sees | What it produces |
|---|---|---|---|
| Pulling back first | Owning less of the height, to shorten the drop | Someone gone slightly quiet, for no stated reason | A partner who matches the new temperature, because nobody wants to be the only one still trying |
| Striking first | Getting it over with on terms that are at least yours | A disagreement that escalated out of proportion to its subject | A conversation about leaving that neither person planned to have |
| Grieving early | Spreading the loss out so less of it lands at once | Someone present and unreachable | A relationship that is already partly over, which was the forecast |
| Auditioning the ending | Checking whether the worst is survivable | Nothing at all. This one is entirely private | A vivid and detailed memory of something that did not happen |
Sometimes the foreboding is accurate
Any page on this subject owes a reader this and most of them do not pay it. Not all bracing is anxiety. Some people are in a relationship that is genuinely ending, and the body works it out before the language does, and being told the feeling is a distortion is both wrong and expensive.
The honest position starts with how badly anyone forecasts this. In 2020 Samantha Joel and a large collaboration ran machine learning over 43 longitudinal datasets covering 11,196 couples, looking for the self-report measures that best predict relationship quality. Relationship-specific variables accounted for up to 45 percent of the variance at the start of a study and up to 18 percent at the end of it. The present is reasonably legible. The future is much less so, with the best instruments the field has.
A 2010 meta-analysis by Benjamin Le and colleagues, covering 137 studies and 37,761 people, found the strongest predictors of a relationship ending to be commitment, love, dependence and how much the other person had been taken into the self. Personality measures predicted very little. What predicts an ending is the state of the bond, not the disposition of whoever is worrying about it.
Which leaves two cues worth more than the scene itself.
The first is whether the forecast survives translation. Written down as things a third person who knows both of you would have noticed independently, bracing usually evaporates, because its evidence is tonal, interior and legible from one seat only. Accurate foresight often survives the translation: plans stopped being made past a certain horizon, a direct question got a non-answer twice, the shared future stopped being mentioned by the other person at all.
The second is whether the forecast has ever been scored. A prediction made every month for two years, never revised, not yet right, is not a forecast. It is a habit with a plot. Someone who cannot remember one occasion when the rehearsed ending failed to arrive is not receiving information from the exercise.
When the unpredictability belongs to the other person
There is a version of this where the rehearsal is not about a nervous system at all.
If warmth arrives and withdraws with no explanation offered for the switch, if leaving gets raised as a lever whenever a disagreement is going the wrong way, if a conversation remembered clearly comes back reported as something that did not happen, then bracing for an ending is not a distortion. It is what accurate perception looks like when the thing being perceived is a person whose behaviour cannot be predicted. Calling that anxiety is one of the more reliable ways people stay put for years, and relationship advice is unusually good at doing it.
A separate thing, which is not a relationship question. If what is being managed is fear of the person rather than fear of losing them, if contact with other people has narrowed, if there is monitoring, or being put down, or a temper that has to be navigated to stay safe, no questionnaire describes that and neither does this article. Services exist for exactly it. They take messages and calls from people who have not decided what to call it yet and are only trying to establish whether this is normal, which is the stage they are built for rather than a later and more obvious one. They can be reached without any decision having been made, and without using a device somebody else has access to.
Reading your own bracing
Two things this piece is deliberately not about. Doubt that behaves like a compulsion, where the question is asked and re-asked and no answer can be stored, is a different pattern with its own name, and the standard relationship advice makes it worse. And the scanning upstream of all of this, the reading of four faces before the coat is off, is hypervigilance, which has its own entry.
What is left is narrower and more workable. The rehearsal is an attempt to control an outcome by predicting it, and the premium is paid in the only currency a relationship runs on, which is the attention of the person who is in the room.
The sentence that moves things is not a reassurance request and not a confession. It names the behaviour rather than the fear, because the behaviour is the part the other person has been reading. I have been pulling back this week and it was not about you. I have been getting ready for something that has not happened.
The first sign of change is not confidence. It is one whole evening attended by the person who was in the room.
Sources
- Nielsen, M. K., Neergaard, M. A., Jensen, A. B., Bro, F., Guldin, M.-B. (2016). Do we need to change our understanding of anticipatory grief in caregivers? A systematic review of caregiver studies during end-of-life caregiving and bereavement. Clinical Psychology Review, 44, 75-93.
- Grupe, D. W., Nitschke, J. B. (2013). Uncertainty and anticipation in anxiety: an integrated neurobiological and psychological perspective. Nature Reviews Neuroscience, 14, 488-501.
- Newman, M. G., Llera, S. J. (2011). A novel theory of experiential avoidance in generalized anxiety disorder: a review and synthesis of research supporting a contrast avoidance model of worry. Clinical Psychology Review, 31(3), 371-382.
- Llera, S. J., Newman, M. G. (2014). Rethinking the role of worry in generalized anxiety disorder: evidence supporting a model of emotional contrast avoidance. Behavior Therapy, 45(3), 283-299.
- Eastwick, P. W., Finkel, E. J., Krishnamurti, T., Loewenstein, G. (2008). Mispredicting distress following romantic breakup: revealing the time course of the affective forecasting error. Journal of Experimental Social Psychology, 44(3), 800-807.
- Norem, J. K., Cantor, N. (1986). Defensive pessimism: harnessing anxiety as motivation. Journal of Personality and Social Psychology, 51(6), 1208-1217.
- Norem, J. K., Illingworth, K. S. S. (1993). Strategy-dependent effects of reflecting on self and tasks: some implications of optimism and defensive pessimism. Journal of Personality and Social Psychology, 65(4), 822-835.
- Murray, S. L., Holmes, J. G., Collins, N. L. (2006). Optimizing assurance: the risk regulation system in relationships. Psychological Bulletin, 132(5), 641-666.
- Downey, G., Freitas, A. L., Michaelis, B., Khouri, H. (1998). The self-fulfilling prophecy in close relationships: rejection sensitivity and rejection by romantic partners. Journal of Personality and Social Psychology, 75(2), 545-560.
- Joel, S., Eastwick, P. W., et al. (2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences, 117(32), 19061-19071.
- Le, B., Dove, N. L., Agnew, C. R., Korn, M. S., Mutso, A. A. (2010). Predicting nonmarital romantic relationship dissolution: a meta-analytic synthesis. Personal Relationships, 17(3), 377-390.