Saved in:
Bibliographic Details
Main Authors: Yang, Joshua C., Flechtner, Maurice, Dailisan, Damian, Bakker, Michiel A.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.15343
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917497121275904
author Yang, Joshua C.
Flechtner, Maurice
Dailisan, Damian
Bakker, Michiel A.
author_facet Yang, Joshua C.
Flechtner, Maurice
Dailisan, Damian
Bakker, Michiel A.
contents LLM-based agents are increasingly used to simulate deliberative interactions such as negotiation, conflict resolution, and multi-turn opinion exchange. Yet generated transcripts often do not reveal why an agent's stance changes: movement may reflect evidence uptake, anchoring, role drift, echoing, or changed prompt and retrieval context. We introduce the Belief Engine (BE), an auditable belief-update layer that treats "belief" as an evidential state over a proposition and exposes it as scalar stance. BE extracts arguments into structured memory and updates stance with a log-odds rule controlled by evidence uptake u and prior anchoring a. Across multiple base LLMs, parameter sweeps show that these controls reliably shape stance dynamics while preserving an evidence-level update trail. On DEBATE, a human deliberation dataset with pre/post opinions, BE best reconstructs participants whose final stance follows extracted evidence; stable and evidence-opposed cases instead point to anchoring or factors outside the extracted evidence stream. BE provides configurable infrastructure for studying evidence-grounded deliberation, where openness, commitment, convergence, and disagreement can be tied to explicit update assumptions rather than hidden prompt effects.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15343
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation
Yang, Joshua C.
Flechtner, Maurice
Dailisan, Damian
Bakker, Michiel A.
Artificial Intelligence
Machine Learning
Multiagent Systems
LLM-based agents are increasingly used to simulate deliberative interactions such as negotiation, conflict resolution, and multi-turn opinion exchange. Yet generated transcripts often do not reveal why an agent's stance changes: movement may reflect evidence uptake, anchoring, role drift, echoing, or changed prompt and retrieval context. We introduce the Belief Engine (BE), an auditable belief-update layer that treats "belief" as an evidential state over a proposition and exposes it as scalar stance. BE extracts arguments into structured memory and updates stance with a log-odds rule controlled by evidence uptake u and prior anchoring a. Across multiple base LLMs, parameter sweeps show that these controls reliably shape stance dynamics while preserving an evidence-level update trail. On DEBATE, a human deliberation dataset with pre/post opinions, BE best reconstructs participants whose final stance follows extracted evidence; stable and evidence-opposed cases instead point to anchoring or factors outside the extracted evidence stream. BE provides configurable infrastructure for studying evidence-grounded deliberation, where openness, commitment, convergence, and disagreement can be tied to explicit update assumptions rather than hidden prompt effects.
title Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation
topic Artificial Intelligence
Machine Learning
Multiagent Systems
url https://arxiv.org/abs/2605.15343