English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Operadic Consistency: Detecting Internal Contradictions in LLM Reasoning with Higher Mathematics

Forum topic · ✨步子哥 · 2026-06-13

Summary

Operadic Consistency (OC) is a label-free method for detecting compositional reasoning failures in large language models. Inspired by operad theory from algebraic topology, it compares a model's direct answer with answers reconstructed from sub-question decompositions: if the composed answer contradicts the direct one, the model's reasoning is internally inconsistent. Unlike self-consistency sampling, semantic entropy, or P(True), OC inspects the internal structure of reasoning rather than surface output statistics. Across 5 compositional reasoning benchmarks and 17 models (12 instruction-tuned LLMs plus 5 frontier reasoning models), OC correlates with accuracy at r ∈ [0.86, 0.94] on all four multi-hop QA datasets, remains informative after controlling for CoT-SC and semantic entropy (p ≤ 10⁻¹⁶), and improves selective prediction (AUROC gains of +0.092 to +0.164 at equal compute, K=3). OC rates also predict unseen models' accuracy within 3.0 points mean absolute error via leave-one-out regression. Requiring no ground-truth labels and only three inference calls, OC offers a near-zero-cost, structure-aware reliability signal, though it is so far validated only on tasks with natural decomposition structures.

A Simple Question

Ask an LLM: "In which country is the Eiffel Tower?" It answers "France."

Now ask two sub-questions: "In which city is the Eiffel Tower?" → "Paris." "In which country is Paris?" → "France."

The composed sub-answers match the direct answer. That's normal.

But sometimes the LLM answers "France" directly, while the sub-questions yield "Paris" and "Germany." The composed answer — Germany — contradicts the direct one. What does this mean? The model's reasoning is internally inconsistent — it isn't really "reasoning"; it's improvising.

This simple check is the core idea of Operadic Consistency (OC), and its mathematical foundation is operad theory from algebraic topology.

From Intuition to Math: What Is an Operad?

An operad is a formal tool for describing "composable systems." Its central question: how do operations on multiple inputs compose into larger operations?

A cooking analogy:

  • "Scramble eggs" is an operation with inputs {eggs, oil, salt}
  • "Make a sandwich" is another operation with inputs {bread, lettuce, scrambled eggs}
  • "Make a sandwich" can embed "scramble eggs" — the output of one becomes an input to the other
  • Operad theory precisely describes this "operation nesting operation" structure.

    In LLM reasoning:

  • A "question template" is an operation whose inputs are the blanks to fill
  • "Decomposition" splits a complex question into sub-questions, then combines the sub-answers
  • Operadic consistency checks whether the direct answer matches the answer reconstructed from decomposition
The paper defines a question operad Q: operations correspond to question templates (with slots), and composition corresponds to sub-question decomposition. The language model is interpreted as an algebra over Q — it fills slots for templates and combines answers for decompositions.

Why Is OC Stronger Than Existing Methods?

Existing failure-detection approaches fall into three classes:

1. Self-Consistency: sample multiple times, check answer agreement. Problem: a model may confidently give the same wrong answer every time. 2. Semantic Entropy: measure semantic diversity of answers. Problem: it looks at output diversity, not reasoning structure. 3. P(True): ask the model "Do you think you're right?". Problem: models can be confidently wrong.

OC's uniqueness: it examines the internal structure of reasoning, not surface output statistics. A model can look superficially consistent (same wrong answer every time), but if its sub-reasoning can't compose, OC catches the contradiction.

Detective analogy: self-consistency is "asking the suspect the same question three times"; semantic entropy is "counting how many versions of the story the suspect gives"; P(True) is "asking the suspect 'are you telling the truth?'"; OC is "taking the testimony apart and checking whether the details corroborate each other."

Experimental Results: A Strong Signal Across Models and Datasets

The paper tests 17 models (12 instruction-tuned LLMs + 5 frontier reasoning models) on 5 compositional reasoning benchmarks:

OC–accuracy correlation: on 4 multi-hop QA datasets, the OC rate correlates with accuracy at Pearson r ∈ [0.86, 0.94] (all p ≤ 0.0004). It is the only signal with r ≥ 0.85 on all four datasets. By contrast, chain-of-thought self-consistency (CoT-SC) performs well on HotpotQA and DROP (r = 0.93, 0.87) but collapses to r ≈ 0.45 on MuSiQue and StrategyQA.

Per-question complementarity: after controlling for CoT-SC and semantic entropy, OC still adds significant information (cluster-robust p ≤ 10⁻¹⁶). The result holds even against purpose-built "decomposition-aware" baselines (p ≤ 10⁻¹³).

Selective prediction: at equal compute budget (K=3), OC improves AUARC by +0.086 to +0.096 and AUROC by +0.092 to +0.164 over tuned CoT-SC baselines, with 95% confidence intervals excluding zero.

Reasoning models: on 5 frontier reasoning models, decompositions extracted from the models' own chains of thought yield positive gains in all 16 (dataset, budget, metric) cells under equal-cost comparison, with 12 of 16 confidence intervals excluding zero.

Cross-model accuracy prediction: OC rates can predict accuracy on unseen models — leave-one-out linear regression over the other 11 models achieves a mean absolute error of only 3.0 percentage points, versus 4.6 points (with a bimodal pattern) for CoT-SC at K=10 inference calls.

Why It Matters

OC reveals a deeper insight: LLM reasoning failures are often not random but structural.

When a model contradicts itself between sub-question reasoning and direct answering, this isn't an occasional slip — its reasoning process simply isn't "composing." Direct and decomposed answering follow two paths that never communicate.

This echoes dual-process theory in human cognition: System 1 (intuition) and System 2 (analysis) can give different answers. OC detects exactly the AI version of "System 1 and System 2 disagreeing."

Crucially, OC requires no labels — no ground-truth answers needed, only internal self-consistency of the model's own responses. This makes it nearly free in deployment: 3 inference calls (1 direct + 2 sub-questions) yield a stronger signal than 10-sample self-consistency.

An Honest Assessment

OC has so far been validated only on compositional reasoning tasks (multi-hop QA, math word problems); its applicability to open-ended generation is unclear, and not every question has a natural decomposition.

OC also has false positives and false negatives: a model can be OC-consistent yet wrong (confidently wrong), or OC-inconsistent yet right (following different valid reasoning paths). The paper candidly shows such cases.

The operadic framework currently uses only depth-2 decompositions (one layer of sub-questions); deeper tree-structured decompositions remain to be explored. But as a first step, OC already makes a compelling case: checking the internal structure of reasoning is more effective than checking its surface features.

---

Paper: Operadic consistency: a label-free signal for compositional reasoning failures in LLMs Code: https://github.com/natebottman/operadic-consistency-paper Authors: Nathaniel Bottman (Incubilate), Yinhong Liu (Cambridge), Kyle Richardson (Allen AI) Area: LLM Reasoning, Uncertainty Quantification, Mathematical Foundations

Tags

#llm-reasoning#operadic-consistency#uncertainty-quantification#operad-theory#ai-reliability#hallucination-detection#compositional-reasoning#research

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177981266