Search papers, labs, and topics across Lattice.
SAVOIR, a novel reinforcement learning framework, tackles the credit assignment problem in social intelligence by using Shapley values to distribute rewards based on an utterance's potential to enable favorable future trajectories. This prospective valuation, combined with Shapley value's axiomatic guarantees, ensures fair and efficient credit distribution. Experiments on the SOTOPIA benchmark show SAVOIR achieves state-of-the-art performance, even surpassing GPT-4o and Claude-3.5-Sonnet with a 7B model, highlighting the distinct capabilities required for social intelligence compared to analytical reasoning.
Social intelligence isn't just scaled reasoning: a 7B model trained with SAVOIR beats GPT-4o and Claude-3.5-Sonnet on social interaction tasks.
Social intelligence, the ability to navigate complex interpersonal interactions, presents a fundamental challenge for language agents. Training such agents via reinforcement learning requires solving the credit assignment problem: determining how individual utterances contribute to multi-turn dialogue outcomes. Existing approaches directly employ language models to distribute episode-level rewards, yielding attributions that are retrospective and lack theoretical grounding. We propose SAVOIR (ShApley Value fOr SocIal RL), a novel principled framework grounded in cooperative game theory. Our approach combines two complementary principles: expected utility shifts evaluation from retrospective attribution to prospective valuation, capturing an utterance's strategic potential for enabling favorable future trajectories; Shapley values ensure fair credit distribution with axiomatic guarantees of efficiency, symmetry, and marginality. Experiments on the SOTOPIA benchmark demonstrate that SAVOIR achieves new state-of-the-art performance across all evaluation settings, with our 7B model matching or exceeding proprietary models including GPT-4o and Claude-3.5-Sonnet. Notably, even large reasoning models consistently underperform, suggesting social intelligence requires qualitatively different capabilities than analytical reasoning.
