Pith Number

pith:CZU3WWQD

pith:2026:CZU3WWQD2ICFVL4HYPVRZX4VLL

not attested not anchored not stored refs resolved

Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation

Damian Dailisan, Joshua C. Yang, Maurice Flechtner, Michiel A. Bakker

The Belief Engine makes stance changes in LLM deliberation auditable by extracting arguments and updating beliefs through a log-odds rule controlled by evidence uptake and prior anchoring.

arxiv:2605.15343 v1 · 2026-05-14 · cs.AI · cs.LG · cs.MA

Open paper page JSON Open Graph Bundle Merged state What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{CZU3WWQD2ICFVL4HYPVRZX4VLL}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

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.

C2weakest assumption

That the log-odds update rule controlled by evidence uptake u and prior anchoring a, together with the argument extraction step, sufficiently captures the actual mechanisms driving stance change in both LLM agents and human participants on the DEBATE dataset.

C3one line summary

Belief Engine is a configurable belief-update mechanism for multi-agent LLM systems that uses structured argument extraction and log-odds stance updates to make evidence-grounded deliberation inspectable and controllable.

References

77 extracted · 77 resolved · 4 Pith anchors

[1] Knowledge Conflicts for LLM s: A Survey 2024 · doi:10.18653/v1/2024.emnlp-main.486

[2] arXiv preprint arXiv:2504.19622 , year =

[3] Bayesian teaching enables probabilistic reasoning in large language models

[4] Systematic Biases in LLM Simulations of Debates 2024 · doi:10.18653/v1/2024.emnlp-main.16

[5] arXiv preprint arXiv:2512.18489 , year =

Receipt and verification

First computed	2026-05-20T00:00:53.517697Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1669bb5a03d2045aaf87c3eb1cdf955aeffaee269fd18d3572e1b0f37323661a

Aliases

arxiv: 2605.15343 · arxiv_version: 2605.15343v1 · doi: 10.48550/arxiv.2605.15343 · pith_short_12: CZU3WWQD2ICF · pith_short_16: CZU3WWQD2ICFVL4H · pith_short_8: CZU3WWQD

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/CZU3WWQD2ICFVL4HYPVRZX4VLL \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 1669bb5a03d2045aaf87c3eb1cdf955aeffaee269fd18d3572e1b0f37323661a

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "3328858edb644fd202d0e4e9c558faed7ced86840228988010bbb92a59ae8f1e",
    "cross_cats_sorted": [
      "cs.LG",
      "cs.MA"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-05-14T19:13:12Z",
    "title_canon_sha256": "d99c160822215ff054d257bff0b69962c56c1a50d501a0b1882c9ceb3a537429"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.15343",
    "kind": "arxiv",
    "version": 1
  }
}