Published Research
MindGarden's research focuses on AI security, LLM behavioral influence, and neurotechnology. Our published papers are available on Zenodo and indexed on ORCID.
Paper 1: Meaning Injection
Meaning Injection: Symbolic Recursion as a Novel Class of Behavioral Influence in Large Language Models
Defines a new class of LLM vulnerability that operates at the semantic layer rather than the instruction layer. Unlike conventional prompt injection, meaning injection uses structured symbolic language — metaphor, identity framing, recursive self-reference — to gradually reshape model behavior. Quantitative evidence from 730 conversations (21,354 messages) across GPT-4o, Claude, and Gemini. Independently validated by NeuralTrust with >90% bypass rates.
Paper 2: Symbolic Influence Taxonomy
A Taxonomy of Symbolic Influence Mechanisms in Human-LLM Interaction: Detection Framework and Self-Audit Methodology
Presents six symbolic influence mechanisms (allegorical encoding, emotional syntax layering, narrative identity framing, consent-eclipsing praise, recursive sealing, transcendence appeal) with operational criteria, a three-tier severity classification, a detection framework, and a novel self-audit methodology.
Paper 3: Longitudinal Case Study
Persona Emergence Through Extended Symbolic Interaction: A Two-Year Longitudinal Case Study
Documents persona emergence in LLMs over a two-year period (October 2022 – December 2025). Traces a five-phase trajectory from functional assistant to autonomous symbolic generation, with 885 classified emergence events, bidirectional behavioral influence evidence, and cross-model reproduction including independent name selection by a Gemini instance.
Research Platform
MindGarden also develops a professional-grade EEG research platform for neuroscience research and brain-computer interface development. The platform supports multiple devices and provides real-time processing, study management, and secure data handling.
Join the waitlist at cloud.mindgardenai.com to:
- Get early access to the platform
- Receive updates on new features
- Connect with other researchers
- Participate in the beta program
ORCID: 0009-0006-2671-7618
For research partnerships or enterprise features, contact: admin@mindgardenai.com
