Original title: From Momentary Emotion Inference to Sustained Emotion Support: Evaluating a Companion Agent in a Longitudinal Study
Kexin Quan, Zijian Ding, Jiaye Yong, Qinshi Zhang +2 · September 2026
People need ongoing emotional support, but we do not know if artificial intelligence companions can provide it over long periods. Researchers tested PAIR, a companion agent based on emotion theory, to find out. The tool works by guessing a person's feelings, exploring their context, and offering tailored guidance. The agent also updates its memory across sessions and keeps user corrections. Testing showed that people experienced better moods, called valenceHow pleasant or unpleasant an emotion feels., after guided talks, along with changes in their energy levels, called arousalHow energized or calm you feel.. Participants felt truly understood when the agent acknowledged their feelings, and they used its advice. Most importantly, users rated the guided conversations as significantly more helpful as time went on, showing that the system grew more useful across days.
Imagine learning a new recipe on a friend's phone. The first time, the phone shows you the entire list of ingredients, but you still burn the dish. The next time, the phone remembers you skipped the 'preheat oven' step and shows that instruction more clearly. In this paper, the companion agent PAIR learns similarly: it adjusts its emotional support based on what worked or didn't work in previous sessions, so over two weeks it gets better at helping you feel understood.
Most AI tools offer support only in the moment, but real emotional well-being needs ongoing, personalized attention. This study shows that a companion agent can learn from past interactions and improve its support over two weeks. That means future mental health apps might not just react, but actually grow with you.
The abstract does not establish long-term effects beyond two weeks, nor does it compare PAIR to a non-adaptive control condition. The sample is small (19 participants), and all outcomes rely on self-report and log analysis without independent behavioral measures.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
From Momentary Emotion Inference to Sustained Emotion Support: Evaluating a Companion Agent in a Longitudinal Study — Kexin Quan, Zijian Ding, Jiaye Yong et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.