Orthogonal Vector Subtraction as a Mechanism for Causal Reasoning in Artificial General Intelligence

AbstractThis paper introduces a novel algorithmic approach to Artificial General Intelligence (AGI) based on Holographic Emergence Theory (HET). Current Large Language Models (LLMs) and vector retrieval systems often fail to distinguish between semantic association and physical causality (e.g., confounding "dissolving in water" with "structural interaction with water").To solve this, we propose Orthogonal Vector Subtraction, a method that introduces a "negative vector" or repulsive force into the inference process. By mathematically subtracting unwanted attributes (e.g., "permanence") from the search query, the system successfully filters out false positives (such as Concrete) and isolates causally accurate materials (such as Magnesium Alloys) that standard embeddings miss.