Updating your model of your partner

You carry a compressed file of your partner in your head. It is fast, cheap, and mostly accurate enough for daily life. It is also old. The compression was done largely in the first few years of the relationship and has been patched lightly since then. Most disagreements between long-term partners are between two slightly-wrong cached models, each one operating on the other's outdated cache. The first move of an update is to acknowledge: I am running a cache. It is not the partner. The partner is somewhere behind it.

You hear yourself saying "he always," "she never," "they would never," "of course they will." Each of these is the cache speaking. The actual partner may always, may never, may would, may will — or may not, anymore, because they have changed. When you catch yourself making a confident prediction about your partner that requires no information, notice. That is the cache. Sometimes it is right. Sometimes it has missed the last three updates. The frequency with which you make these predictions, and the frequency with which you turn out to be wrong, is a diagnostic.

Once a year, do a deliberate survey. Pick a quiet evening. Ask, with no agenda: what do you want more of in your life. What do you want less of. What are you newly interested in. What have you stopped caring about. What do you wish I knew that I don't. The first time you do this, it will feel artificial and might be awkward. By the third year, it will be one of the most useful conversations of the year. Karen Fingerman's research suggests that couples who maintain explicit periodic check-ins outperform couples who rely on inference; inference at scale becomes projection.

Updating the model does not mean erasing the history. Your partner's past, the patterns you have observed over decades, the things you have learned through long experience — these remain valuable data. The update is a recency-weighted refinement, not a wipe. Treat the old model as a strong prior; treat new evidence as a real signal. When new evidence consistently contradicts the prior, update the prior. When new evidence is sparse, hold the prior. This is just Bayesian decency, applied to a person you love.

A particular failure: holding a partner inside a frame built from one period of their life that they have outgrown. They were anxious in their thirties; you still treat them as anxious in their fifties. They were a poor communicator early on; you still walk in expecting them to communicate poorly. They had a difficult patch with their family in 2011; you still react as if they are still in it. The frame becomes a prison. The fix is hard: notice the frame, name it to yourself, then deliberately look for current evidence rather than confirmatory evidence. This is the practice of seeing the actual partner, not the wax figure.

The opposite failure: comparing the current partner to an idealized earlier version. Sometimes the comparison is real — they have genuinely declined in some respect — but more often the earlier version is half-fiction, polished by memory. The current partner cannot beat a polished memory. Notice when you are doing this. Ask whether the earlier version really was what you remember, and whether the current version really is less than that, or just different. Difference is not deficit. Most apparent decline is reorganization.

Make a list. What did you used to ask your partner that you have stopped asking. Their opinion on movies, on politics, on how their day went, on what they are thinking about, on what they want for their birthday, on what they would change about the marriage if they could. Many couples reduce their question diet to logistics — what's for dinner, who's picking up what. The question diet starves the model. Reintroduce a few of the lost questions. Watch the model update.

Adam Phillips's writing reminds us that we always relate to a partly-imagined version of the other person; the imagination is part of intimacy, not a flaw in it. The work is not to eliminate imagination but to keep it in conversation with reality. Let the imagination be checked against the actual person, frequently, with curiosity. The marriages that go bad on this axis are the ones in which the imagination has run on autopilot for a decade and no longer touches the actual partner at any point.

Track the last five times your partner surprised you in good faith. If you cannot think of any, the model is over-fitted. Surprise is data — it means there is information in the partner that your model does not contain. Welcome it. Couples who treat surprise as annoying ("you should know me by now") punish the very thing they need. Couples who treat surprise as interesting ("oh, tell me more about that") get a steady stream of useful updates. Esther Perel's claim about erotic mystery scales here: the partner who can still surprise you is the partner you can still desire.

Sometimes the most useful question is: how do you see yourself right now. How would you describe yourself to a stranger. What do you think is true about you that I might not know. Some partners will refuse the question; some will rise to it. Those who rise to it will say things you did not know. Listen. Do not argue. Do not correct from your old model. Take it in as fresh data. Update.

Once you have a model, you tend to notice evidence that confirms it and miss evidence that contradicts it. This is universal. The fix is to deliberately search for the disconfirming evidence. If your model says your partner is conflict-avoidant, look for the times this week they brought up something hard. If your model says they are not affectionate, count the gestures of affection you may have stopped noticing. The point is not to gaslight yourself into a rosier model; it is to expose the model to data it might be filtering out.

Sometimes the most powerful move is to say it out loud: I realize my model of you is out of date. I want to know who you actually are now. Help me update. Most partners are moved by this. It is, in effect, a request to be taken seriously as a developing person rather than as a known quantity. Marriages in which this request is made occasionally tend to retain a sense of liveness for decades. Marriages in which neither partner has ever made it tend to feel, by year twenty, like a museum of an earlier marriage.

The update is not a project with an end date. It is the discipline of staying interested in the actual person across from you, indefinitely. The discipline is small in any given week. It is enormous over decades. The couples who do it have, at eighty, a marriage that is alive in a way that defies most predictions. The couples who do not do it have a museum, and museums, however beautifully maintained, are not where love lives.

1. Phillips, Adam. Monogamy. New York: Pantheon, 1996. 2. Phillips, Adam. Missing Out: In Praise of the Unlived Life. New York: Farrar, Straus and Giroux, 2012. 3. Perel, Esther. Mating in Captivity: Unlocking Erotic Intelligence. New York: HarperCollins, 2006. 4. Perel, Esther. The State of Affairs: Rethinking Infidelity. New York: HarperCollins, 2017. 5. Fingerman, Karen L. Aging Mothers and Their Adult Daughters: A Study in Mixed Emotions. New York: Springer, 2001. 6. Fingerman, Karen L., and Frank F. Furstenberg. "You Can Go Home Again." New York Times, May 30, 2012. 7. Pipher, Mary. Women Rowing North: Navigating Life's Currents and Flourishing as We Age. New York: Bloomsbury, 2019. 8. Freedman, Marc. The Big Shift: Navigating the New Stage Beyond Midlife. New York: PublicAffairs, 2011. 9. Bonanno, George A. The Other Side of Sadness: What the New Science of Bereavement Tells Us About Life After Loss. New York: Basic Books, 2009. 10. Gawande, Atul. Being Mortal: Medicine and What Matters in the End. New York: Metropolitan Books, 2014. 11. Levine, Carol. Always On Call: When Illness Turns Families into Caregivers. Nashville: Vanderbilt University Press, 2004. 12. Didion, Joan. The Year of Magical Thinking. New York: Knopf, 2005.