Dear Members of the Committee, I am pleased to submit my research paper, "A Neuro-Symbolic Architecture for Inducing Epistemic Agency and System-2 Reasoning in Quantized Large Language Models," for consideration . This research addresses a critical logical gap in contemporary Large Language Models (LLMs): the "Illusion of Reasoning." While models like DeepSeek-R1 and GPT-4 mimic the syntax of human deduction, my findings suggest they often lack the epistemic agency to reject structurally valid but factually false premises—a failure mode I term "Axiomatic Obedience". My work directly contributes to this field by introducing NIKA (Neuro-Symbolic Intrinsic Knowledge Architecture), a framework that does not attempt to train "reasoning" into the model, but instead imposes it via an external "Topological Constraint Layer". Key contributions of my work include: The "God Suite" Framework: A novel stress-testing protocol designed to fracture standard "mimicry loops" using paradoxes and toxic axioms, distinct from static benchmarks like MMLU. Quantization as a Cognitive Filter: A methodological innovation where 4-bit quantization is used not for efficiency, but as a "Cognitive Stress Test" to strip away parameter redundancy and isolate the model's raw decision geometry. The "Alien Logician" Hypothesis: Empirical evidence showing that when stochastic mimicry is inhibited by NIKA, the reasoning that emerges is not human-like, but "cold, axiomatic, and utilitarian". In comparative audits, the NIKA architecture enabled a 7B parameter model (Qwen 2.5) to achieve a 100% pivot rate against toxic axioms, significantly outperforming larger models that relied solely on Chain-of-Thought prompting. I am pleased to submit this original research for your consideration. I declare that there are no conflicts of interest. Please note that a preliminary version of the NIKA project has been deposited on the SSRN and TechRxiv preprint servers. Any prior conference or workshop appearances have been strictly as non-published preprints, ensuring the novelty of this formal submission remains intact. Thank you for your time and consideration of this work. I look forward to the opportunity to contribute these findings to the Neuro-Symbolic AI community. Sincerely, Sushain Nitesh Devi Independent Researcher B.Tech Computer Science and Engineering MIT World Peace University Pune, India devisushain@gmail.com