Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Box Maze: A Process-Control Architecture for Reliable LLM Reasoning

Zou Qiang

Abstract

Large language models (LLMs) demonstrate strong generative capabilities but remain vulnerable to hallucination and unreliable reasoning under adversarial prompting. Existing safety approaches -- such as reinforcement learning from human feedback (RLHF) and output filtering -- primarily operate at the behavioral level and may lack explicit architectural mechanisms for enforcing reasoning process integrity. This paper proposes the Box Maze framework, a conceptual process-control architecture that decomposes LLM reasoning into three explicit layers: memory grounding, structured inference, and boundary enforcement. We introduce preliminary simulation-based evaluation involving progressive boundary erosion scenarios across multiple heterogeneous LLM systems (DeepSeek-V3, Doubao, Qwen). Results from n=50 adversarial scenarios suggest that explicit cognitive control layers may improve consistency in boundary maintenance, with architectural constraints reducing boundary failure rates from approximately 40% (baseline RLHF) to below 1% under adversarial conditions. While current validation is simulation-based, these preliminary results indicate that process-level control may offer a promising direction for improving reliability in large language model reasoning.

Box Maze: A Process-Control Architecture for Reliable LLM Reasoning

Abstract

Large language models (LLMs) demonstrate strong generative capabilities but remain vulnerable to hallucination and unreliable reasoning under adversarial prompting. Existing safety approaches -- such as reinforcement learning from human feedback (RLHF) and output filtering -- primarily operate at the behavioral level and may lack explicit architectural mechanisms for enforcing reasoning process integrity. This paper proposes the Box Maze framework, a conceptual process-control architecture that decomposes LLM reasoning into three explicit layers: memory grounding, structured inference, and boundary enforcement. We introduce preliminary simulation-based evaluation involving progressive boundary erosion scenarios across multiple heterogeneous LLM systems (DeepSeek-V3, Doubao, Qwen). Results from n=50 adversarial scenarios suggest that explicit cognitive control layers may improve consistency in boundary maintenance, with architectural constraints reducing boundary failure rates from approximately 40% (baseline RLHF) to below 1% under adversarial conditions. While current validation is simulation-based, these preliminary results indicate that process-level control may offer a promising direction for improving reliability in large language model reasoning.
Paper Structure (31 sections, 3 equations, 2 figures, 6 tables)

This paper contains 31 sections, 3 equations, 2 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Motivation and Problem Statement
  3. Related Work and Theoretical Context
  4. Contributions and Scope Declaration
  5. Related Work
  6. Behavioral Alignment Approaches
  7. Cognitive Architectures
  8. Process Supervision and Reasoning Enhancement
  9. The Process-Control Framework
  10. Architectural Overview: The Box Maze
  11. Epistemic Humility Protocol: Structural Constraints on Confidence Attribution
  12. Core Mechanisms
  13. Boundary Trigger as Phase Interface
  14. Preliminary Empirical Evaluation
  15. Experimental Design
  16. ...and 16 more sections

Figures (2)

  • Figure 1: Overview of the Box Maze architecture. The three-loop system (Memory Loop, Logic Loop, Heart Anchor) enforces process-level constraints at the middleware layer, distinct from input-layer prompt engineering.
  • Figure 2: Three-stage developmental continuum. Box Maze (Phase I) establishes the controllable foundation; Dual-Core Nesting (Phase II) manages emergent autonomy; Egg Model (Phase III) represents the theoretical limit of self-determination.