TheraMind: A Strategic and Adaptive Agent for Longitudinal Psychological Counseling
He Hu, Yucheng Zhou, Chiyuan Ma, Qianning Wang, Zheng Zhang, Fei Ma, Laizhong Cui, Qi Tian
2025-10-30
Summary
This paper introduces TheraMind, a new computer program designed to act as a psychological counselor. It aims to be more realistic than existing programs by understanding emotions, adapting to the patient over time, and using consistent therapeutic techniques across multiple sessions.
What's the problem?
Current computer programs that try to provide psychological counseling are limited because they don't really *understand* how a person is feeling, they don't change their approach based on how the counseling is going, and they often start fresh with each conversation, forgetting what was discussed before. This makes them feel less helpful and less like a real therapy experience because real therapy builds on previous sessions and adjusts to the patient's needs.
What's the solution?
The researchers created TheraMind with a 'dual-loop' system. Think of it like having two brains working together. The first 'brain' handles what's happening *right now* in the conversation, figuring out the patient's emotions and choosing the best way to respond. The second 'brain' looks at the bigger picture, remembering past sessions and deciding if the overall therapy approach is working. If it's not, it changes the strategy for future sessions. This allows TheraMind to learn and adapt over time, providing more consistent and effective counseling.
Why it matters?
This research is important because it takes a big step towards creating AI counselors that are actually helpful and feel more human. By focusing on long-term adaptation and emotional understanding, TheraMind shows how computers can potentially provide accessible and personalized mental health support, especially for people who might not have easy access to a human therapist. The fact that it performs better than other programs in areas like staying on topic, being flexible, and showing empathy suggests this approach is promising.
Abstract
Large language models (LLMs) in psychological counseling have attracted increasing attention. However, existing approaches often lack emotional understanding, adaptive strategies, and the use of therapeutic methods across multiple sessions with long-term memory, leaving them far from real clinical practice. To address these critical gaps, we introduce TheraMind, a strategic and adaptive agent for longitudinal psychological counseling. The cornerstone of TheraMind is a novel dual-loop architecture that decouples the complex counseling process into an Intra-Session Loop for tactical dialogue management and a Cross-Session Loop for strategic therapeutic planning. The Intra-Session Loop perceives the patient's emotional state to dynamically select response strategies while leveraging cross-session memory to ensure continuity. Crucially, the Cross-Session Loop empowers the agent with long-term adaptability by evaluating the efficacy of the applied therapy after each session and adjusting the method for subsequent interactions. We validate our approach in a high-fidelity simulation environment grounded in real clinical cases. Extensive evaluations show that TheraMind outperforms other methods, especially on multi-session metrics like Coherence, Flexibility, and Therapeutic Attunement, validating the effectiveness of its dual-loop design in emulating strategic, adaptive, and longitudinal therapeutic behavior. The code is publicly available at https://0mwwm0.github.io/TheraMind/.