Modeling Multi-Party Interaction in Couples Therapy: A Multi-Agent Simulation Approach DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing E

Canwen Wang, Angela Chen, Catherine Bao, Siwei Jin, Yee Kit Chan, Jessica R Mindel, Sijia Xie, Holly Swartz, Tongshuang Wu, Robert E Kraut, Haiyi Zhu · 2026

Read Modeling Multi-Party Interaction in Couples Therapy: A Multi-Agent Simulation Approach DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing E by Canwen Wang, Angela Chen, Catherine Bao, Siwei Jin, Yee Kit Chan, Jessica R Mindel, Sijia Xie, Holly Swartz, Tongshuang Wu, Robert E Kraut, Haiyi Zhu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-10970.txt.

Introduction

Canwen Wang and co-authors model couples therapy with a multi-agent simulation. Couples therapy helps partners resolve conflict and grow closer. Training new therapists leans on textbooks and roleplay. Those tools often miss the complexity and emotional nuance of real couple dynamics. A simulator could let trainees practice against lifelike partners and a therapist agent.

The system is multimodal and multi-agent, modeling multi-party interaction in the therapy room. Each party is played by an agent with its own stance. Trainees can face hard moments safely and repeat them. The work speaks to clinical training and to the wider study of how AI can model delicate human interaction without putting real couples at risk.

Catalog record

CREATORCanwen Wang, Angela Chen, Catherine Bao, Siwei Jin, Yee Kit Chan, Jessica R Mindel, Sijia Xie, Holly Swartz, Tongshuang Wu, Robert E Kraut, Haiyi Zhu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE73 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-10970

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Modeling Multi-Party Interaction in Couples Therapy: A Multi-Agent Simulation Approach DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Human-centered computing Empirical studies in HCI Canwen Wang, Angela Chen, Catherine Bao, Siwei Jin, Yee Kit Chan, Jessica R Mindel, Sijia Xie, Holly Swartz, Tongshuang Wu, Robert E Kraut, Haiyi Zhu Abstract . Couples therapy, or relationship counseling, helps partners resolve conflicts, improve satisfaction, and foster psychological growth. Traditional approaches to training couples therapists, such as textbooks and roleplay, often fail to capture the complexity and emotional nuance of real couple dynamics. We present a novel multimodal, multi-agent simulation system that models multi-party interactions in couples therapy. Informed by our systematic research, this system creates a low-stakes environment for trainee therapists to gain valuable practical experience dealing with the critical demand-withdraw communication cycle across six couple-interaction stages. In an evaluation study involving 21 US-based licensed therapists, participants blind to conditions identified the engineered agent behaviors (i.e., the stages and the demand-withdraw cycle) and rated overall realism and agent responses higher for the experimental system than the baseline. As the first known multi-agent framework for training couples therapists, our work builds the foundation for future research that fuses HCI technologies with couples therapy. 1. Introduction Standardized patients have long been used in medical training as a way to give trainees realistic yet safe practice opportunities, with systematic reviews showing positive effects on learner skill development and even patient outcomes ( 42 ; 34 ; 30 ) . Simulation-based training with virtual patients extends these benefits while also addressing issues of cost and scalability, providing repeatable practice environments unconstrained by human availability. Recent systems have leveraged large language models (LLMs) to simulate patient interactions for counselor training, such as Roleplay-doh ( 24 ) . Yet these systems almost exclusively model one-on-one interactions, leaving open the question of how to simulate conversations involving more than one patient at a time, which is central to couples therapy or relationship counseling. Couples therapy is a core modality in mental health care with evidence for improving relationship distress and supporting individual well-being ( 36 ) . Multiple structured interventions, such as Emotionally Focused Couple Therapy (EFT) and Behavioral Couple Therapy (BCT), offer procedures for targeting patterns that maintain conflict and disconnection ( 11 ; 16 ; 20 ; 14 ) . Yet training clinicians to navigate the multi-party realities of couples work remains difficult. Trainees must track two perspectives simultaneously, manage rapid emotional shifts, and intervene in live interaction cycles—skills that are hard to cultivate using textbooks and videos alone and imperfectly approximated by role-play due to issues with ecological validity, bias, cost, and scalability ( 27 ; 9 ; 31 ; 35 ; 28 ) . From a human-computer interaction (HCI) perspective, these constraints motivate interactive, simulation-based training tools that afford experiential practice with controllable complexity and timely feedback in multi-party, emotion-laden dialogue. Across approaches and settings, communication problems are among the most frequently reported presenting concerns for couples experiencing relationship difficulties, conflict, or emotional strain ( 12 ; 26 ; 40 ) . A particularly robust pattern in distressed couples is the demand-withdraw cycle (also termed pursue-withdraw ): one partner pursues change with criticism or pressure while the other avoids or defends, amplifying negative affect during and after conflict discussions ( 13 ; 8 ; 25 ) . EFT and related approaches explicitly organize intervention around identifying and de - escalating this cycle in-session ( 18 ) . For training, this points to a need for practice environments that reliably surface the cycle, let trainees test timing- and wording - sensitive interventions, and do so without relying on peers or supervisors to play clients. Our paper introduces a multimodal, multi-agent simulation system for couples therapy that models the dynamics of two virtual patients and enables experiential practice with the complexities of multi-party dialogue. Our system models multi-party interaction with two virtual patients and a module that controls the stage of dialogue (the stage controller ), centering on the demand-withdraw pattern and six recurrent stages: greeting, problem raising, escalation, de-escalation, enactment, and wrap - up . The system enables trainees to engage with agents that (a) display demand–withdrawal patterns, (b) exhibit stage-appropriate behaviors, and (c) shift from therapist-facing to partner-to-partner exchanges during conflict, thereby creating opportunities to practice calming, reframing, and restructuring strategies. This design was informed through a systematic, three-stage research process. First, we synthesized theories and literature on couples therapy. Second, we conducted semi-structured interviews with two experienced couples therapists, who converged on a six-stage session flow and emphasized nonlinearity and emotional shifts as triggers for intervention. Third, we performed a secondary analysis of seven publicly available couples therapy transcripts (including 1,621 conversation turns), which confirmed that these patterns recur in authentic sessions. We evaluated our couple agent design in a within - subjects study with 21 licensed U.S. therapists. Participants were asked to interact with an experimental system embodying the six stages and demand–withdraw dynamics and a baseline version without these behaviors; participants were blind to condition and order. We found that therapists were more likely to identify the engineered stages and the demand–withdraw cycle in the experimental system and rated both overall realism and agent responses higher than the baseline, with gains concentrated in problem raising , escalation , and de - escalation . To our knowledge, this is the first multi - agent framework tailored to simulating realistic couples - therapy interaction. Our contributions are threefold: (1) a stage - grounded, multi - agent simulation of couples therapy that operationalizes six recurrent stages and the demand-withdraw pattern with multimodal feedback for trainees; (2) a formative foundation combining theory, expert interviews, and corpus analysis to derive stage definitions, agent behaviors, and transition logic; and (3) an empirical evaluation with licensed therapists showing higher perceived realism and clearer recognition of stage - and cycle - specific behaviors relative to a baseline, with stage - specific effects concentrated in problem raising, escalation, and de - escalation. The remainder of the paper reviews related work on couples therapy and simulation (Section 2), details our system design and implementation (Section 3), and reports quantitative and qualitative results from the evaluation (Section 4), followed by limitations and future directions (Section 5). 2. Related Work 2.1. Couples Therapy In the field of mental health, most therapy takes the form of individual sessions where one person seeks support. Yet, beyond the one-on-one encounter lies the powerful domain of multi-party therapy, such as couples therapy, where the room holds more than one story. Couples therapy is a form of psychotherapy that involves both members of a committed partnership, whether married or unmarried, heterosexual or homosexual, and focuses on the patterns of interaction between partners ( 37 ) . Also known as marital therapy or relationship counseling, couples therapy is a psychotherapeutic process in which a trained clinician works with two individuals involved in a romantic relationship to resolve conflicts, improve relationship satisfaction, and promote emotional and psychological growth within the partnership ( 5 ; 16 ; 17 ) . Couples therapy has gained growing recognition in mental health care because of its three unique features,

Read the complete text