# "My Ex Is Crazy": True or False? > Source: > Published: 2026-08-13 19:06:00+00:00 ###### [Cognition](/us/basics/cognition) # "My Ex Is Crazy": True or False? ## A label ends the question. That is exactly what makes it feel like an answer. Posted August 13, 2026 [ Reviewed by Margaret Foley ](/us/docs/editorial-process) ### Key points - Under uncertainty, a settled answer can feel better than a fuller one. - Some psychology-sounding words are diagnoses. Most are not. - An agreeable chatbot can make a one-sided account feel like a verdict. It was a Monday in March, or that is how I remember it. I was 18, in a small mountain town, finishing a two-page letter to my first girlfriend about why I was ending it. Handwritten, because that is what you did. The reasons were ordinary, and I believe they were true. We wanted different things. We reacted badly to each other. Neither of us was going to give way. It took two pages because I was trying to be fair to her. Then, folding it, another thought arrived. What if none of this was ours? What if she was ill? I did not put it in the letter. I remember the relief of it, though, and the thought has never entirely left. The letter took two pages. The shortcut took a sentence. And the shortcut did not replace the two pages. It offered to explain them all at once. That is the part worth looking at. Relational reasoning is slow and stays uncomfortable because it keeps both people in the frame. A diagnostic label is fast and feels complete. It moves the cause out of the space between two people and lodges it inside one of them, where it can be filed. Relationships are seldom true or false. Psychologists call this pull the need for [cognitive closure](https://doi.org/10.1037/0033-295X.103.2.263): the desire for a definite answer when ambiguity is hard to bear. Under uncertainty we may seize on early information, and once a judgment has formed we become less willing to reopen it. A breakup creates exactly those conditions. The ambiguity is high, the cost of sitting in it is real, and no one will ever adjudicate what happened. [Memory](https://www.psychologytoday.com/us/basics/memory) does not hold the line either. In one study, people rated their [dating](https://www.psychologytoday.com/us/basics/mating) partners, then two months later rated them again and tried to [recall](https://doi.org/10.1177/0146167287132008) their original ratings. Those whose impressions had improved remembered their earlier evaluations as warmer than they had ac18tually been. Those whose impressions had soured remembered them as colder. Once a verdict is in, the record starts agreeing with it. What has changed since that Monday is not the reflex. It is the supply. At 18, the thought had nowhere to go. There was no search box in that room. The people I could have said it to knew one or both of us and might have pushed back. Now it has somewhere to go within seconds. Couples therapists [report hearing](https://www.theatlantic.com/family/2025/12/therapy-speak-therapy/685218/) the same vocabulary: narcissist, borderline, bipolar, avoidant. Around it sit terms for conduct or relationship dynamics: [gaslighting](https://www.psychologytoday.com/us/basics/gaslighting), love-bombing, [trauma bond](https://www.psychologytoday.com/us/basics/trauma-bonding). These words do not have the same status. [Narcissistic](https://www.psychologytoday.com/us/basics/narcissism) and borderline [personality disorders](https://www.psychologytoday.com/us/basics/personality-disorders) and [bipolar disorder](https://www.psychologytoday.com/us/basics/bipolar-disorder) are diagnoses. Avoidant [attachment](https://www.psychologytoday.com/us/basics/attachment) is not; nor are the last three. All can name something real. The trouble begins when a word settles a history before it has explained it. Search any of them next to partner or ex and the checklists appear, built around signs to recognize rather than evidence that would rule the label out. An agreeable [chatbot](https://www.psychologytoday.com/us/basics/artificial-intelligence) can go further. It works inside the account you give it, which you have already shaped, already sorted, already stripped of the parts that cut the other way, and it can hand that account back cleaner and more settled than you supplied it. In April 2025, OpenAI [rolled back](https://openai.com/index/expanding-on-sycophancy/) a version of its model after it became markedly more agreeable; the company later described it as validating doubts and reinforcing negative emotions in ways it had not intended. Whatever comes back, the account still has one witness. I want to be careful here. Some partners do meet diagnostic criteria. Some people have been gaslit or otherwise abused, and finding the accurate word can be where clarity begins. My question is narrower: What happens when the word arrives as a verdict before the evidence has earned it? Is this an explanation—or the end of one? [Cognition](https://www.psychologytoday.com/us/basics/cognition)Essential Reads I use three checks to slow a verdict I already hold. - Date the word. When did it first occur to me, and how had I described the same events before it did? - Rewrite one episode in verbs. What was said and done, in order, with no character labels. Some episodes get more serious in translation. Others lose most of their force. Either result tells you something. - Then ask what the label made unnecessary: what was happening between us, what else was going on around us, and what my own part in it was. References Khazan, O. (2025, December 12). Why couples therapists are sick of "therapy-speak." *The Atlantic.* [https://www.theatlantic.com/family/2025/12/therapy-speak-therapy/685218/](https://www.theatlantic.com/family/2025/12/therapy-speak-therapy/685218/) Kruglanski, A. W., & Webster, D. M. (1996). Motivated closing of the mind: "Seizing" and "freezing." *Psychological Review, 103*(2), 263–283. [https://doi.org/10.1037/0033-295X.103.2.263](https://doi.org/10.1037/0033-295X.103.2.263) McFarland, C., & Ross, M. (1987). The relation between current impressions and memories of self and dating partners. *Personality and Social Psychology Bulletin, 13*(2), 228–238. [https://doi.org/10.1177/0146167287132008](https://doi.org/10.1177/0146167287132008) OpenAI. (2025, May 2). *Expanding on what we missed with sycophancy.* [https://openai.com/index/expanding-on-sycophancy/](https://openai.com/index/expanding-on-sycophancy/) OpenAI. (2025, April 29). *Sycophancy in GPT-4o: What happened and what we're doing about it.* [https://openai.com/index/sycophancy-in-gpt-4o/](https://openai.com/index/sycophancy-in-gpt-4o/)