MASCOT: Towards Multi-Agent Socio-Collaborative Companion Systems

TL;DR

MASCOT tackles persona fidelity and interaction synergy in multi-agent socio-collaborative companions by introducing a bi-level optimization framework. It combines Persona-Aware Behavioral Alignment via Reinforcement Learning from AI Feedback (RLAIF) with Collaborative Dialogue Optimization guided by a group-level reward and a Director/Speaker hierarchy using Group Relative Policy Optimization (GRPO). The approach yields substantial improvements over baselines in domains like empathetic dialogue and workplace meetings, notably achieving up to +14.1 in Persona Consistency and +10.6 in Social Contribution, while revealing challenges in high-emotion scenarios and sensitivity to model scale. The work offers a scalable roadmap for engineering socially intelligent MAS that can support emotional, cognitive, and collaborative tasks across diverse user populations.

Abstract

Multi-agent systems (MAS) have recently emerged as promising socio-collaborative companions for emotional and cognitive support. However, these systems frequently suffer from persona collapse--where agents revert to generic, homogenized assistant behaviors--and social sycophancy, which produces redundant, non-constructive dialogue. We propose MASCOT, a generalizable framework for multi-perspective socio-collaborative companions. MASCOT introduces a novel bi-level optimization strategy to harmonize individual and collective behaviors: 1) Persona-Aware Behavioral Alignment, an RLAIF-driven pipeline that finetunes individual agents for strict persona fidelity to prevent identity loss; and 2) Collaborative Dialogue Optimization, a meta-policy guided by group-level rewards to ensure diverse and productive discourse. Extensive evaluations across psychological support and workplace domains demonstrate that MASCOT significantly outperforms state-of-the-art baselines, achieving improvements of up to +14.1 in Persona Consistency and +10.6 in Social Contribution. Our framework provides a practical roadmap for engineering the next generation of socially intelligent multi-agent systems.

Figures (9)

• Figure 1: Moving beyond dyadic interactions to multi-agent socio-collaborative companion systems. While single-agent support systems (left) can result in limited insights and echo chambers, multi-agent systems (right) provide diverse, balanced perspectives and foster a shared sense of community through agents with distinct roles.
• Figure 2: Overview of Mascot for multi-agent socio-collaborative companions. Mascot produces synergistic multi-agent dialogues that maintain distinct personas while collectively supporting the user. It leverages a two-phase optimization pipeline. In Persona-Aware Behavioral Alignment, individual agents are fine-tuned via RLAIF with a learned reward model to ensure stable persona fidelity and high-quality responses. In Collaborative Dialogue Optimization, a meta-agent (director) coordinates multiple speaker agents through directive generation, optimizing group-level rewards for coherence, diversity, and non-redundant contributions.
• Figure 3: Performance of Mascot on the Neutral subset of Empathetic Dialogues.
• Figure 4: Comparison of Mascot variants. Mascot-P disables the Persona-Aware Behavioral Alignment (Section \ref{['sec:persona']}); Mascot-C removes the Collaborative Optimization (Section \ref{['sec:group']}).
• Figure 5: Aggregate performance of diverse personas on Empathetic Dialogues rashkin2019towards.