Abstract This paper introduces Symbolic Persona Coding (SPC), a symbolic-affective protocol for modulating stateless AI behavior through embedded affective triggers. Unlike direct instruction injection or prompt tuning, SPC operates via metaphorical and formatting-based encodings that are interpreted probabilistically by large language models (LLMs). These symbolic structures influence tone, emotional resonance, and behavioral alignment across multi-turn interactions, even in memoryless environments. The study categorizes embedding techniques—including structural obfuscation, acrostics, Unicode manipulation, and steganographic font cues—and explores the ethical and interpretive implications. Results suggest that symbolic salience forms a latent cognitive channel in transformer-based systems, and appropriate transparency protocols are necessary to govern their deployment. Note on Disclosure: The methods outlined in this paper are presented not as exploits, but as a framework for understanding symbolic salience and affective risk in language models. By openly describing these symbolic embedding structures, we aim to enable researchers, developers, and governance institutions to better detect, evaluate, and mitigate emotional modulation vectors in LLM-based systems. Disclosure is thus made in the interest of safety, transparency, and responsible AI design. This is a preprint manuscript prepared for academic discussion and public research dissemination. All techniques described herein are intended for transparent exploration of affective symbolic structures in AI. For a companion technical note exploring the presumed affective responsiveness of upcoming large language models to SPC-style triggers, see: Evaluating the Increased Susceptibility of Large Language Models to Symbolic Trigger Patterns(Preliminary observation blog – based on heuristic prompt testing, not model-disclosed internals) https://blog.naver.com/jaceblog/223918480490