Neural Synchrony Between Socially Interacting Language Models

TL;DR

This work introduces SyncR^2, a representation-level neural synchrony measure based on affine mappings predicting one LLM's future representations from another's current ones during social interactions. Using the Sotopia environment, the authors show that SyncR^2 captures both social engagement and temporal proximity, and that higher synchrony aligns with stronger social performance across diverse model families. Controlling for confounds like instruction-following and long-context reasoning, the study finds a robust link between neural synchrony and social capabilities, while analyses hint at underlying Theory of Mind and social predictive coding mechanisms. These findings establish neural synchrony as a practical proxy for studying the internal dynamics of LLM social interactions and point to avenues for designing more socially adept AI agents.

Abstract

Neuroscience has uncovered a fundamental mechanism of our social nature: human brain activity becomes synchronized with others in many social contexts involving interaction. Traditionally, social minds have been regarded as an exclusive property of living beings. Although large language models (LLMs) are widely accepted as powerful approximations of human behavior, with multi-LLM system being extensively explored to enhance their capabilities, it remains controversial whether they can be meaningfully compared to human social minds. In this work, we explore neural synchrony between socially interacting LLMs as an empirical evidence for this debate. Specifically, we introduce neural synchrony during social simulations as a novel proxy for analyzing the sociality of LLMs at the representational level. Through carefully designed experiments, we demonstrate that it reliably reflects both social engagement and temporal alignment in their interactions. Our findings indicate that neural synchrony between LLMs is strongly correlated with their social performance, highlighting an important link between neural synchrony and the social behaviors of LLMs. Our work offers a new perspective to examine the "social minds" of LLMs, highlighting surprising parallels in the internal dynamics that underlie human and LLM social interaction.

Figures (11)

• Figure 1: An illustration of our analysis framework. (1) Two LLM agents engage in a social interaction, generating responses conditioned on their backgrounds, goals, and shared history. (2) Hidden representations are extracted at each turn from both LLMs. (3) Neural synchrony is measured by learning affine mappings to predict one agent’s future representations from the other’s current ones.
• Figure 2: Example synchrony heatmaps of $A$ predicting $B$. The horizontal and vertical axes denote source layers $l_A$ and target layers $l_B$, respectively, with each cell showing the test-set $R^2$ score for the corresponding layer pair. Gray cells indicate negative test-set $R^2$ values. For each $l_A$, the best-predicting $l_B$ is highlighted with a black box. Model names are abbreviated by omitting the suffix "Instruct" due to space constraints. Note that Llama-3.2-3B-Instruct has 28 layers, whereas the other models have 32 layers.
• Figure 3: Examining synchrony of Mistral-7B-Instruct-0.3 predicting Llama-3-8B- Instruct under different conditions. Left: Experimental group with genuine interaction. Middle: Control group 1 (w/o social engagement), where one agent passively consumes the dialogue without actually engaging in the interaction. Right: Control group 2 (w/o temporal proximity), where source representations are paired with lagged future representations.
• Figure 4: Neural synchrony declines under control settings. (a) Control group without social engagement: bar plots show the mean $\mathit{SyncR^2}$ across models within each family. (b) Control group without temporal proximity: line plots show $\mathit{SyncR^2}$ for all model pairs within each family as a function of lag $k$ ($k\!=\!0$ is the experimental group). Model names are abbreviated (see Appendix \ref{['appendix:abbreviations']} for the full name correspondence). Error bars denote the standard errors.
• Figure 5: Neural synchrony under different relationship closeness. $\mathit{SyncR^2}$ is aggregated over all model pairs.