The Generative Horizon: Applied Hermeneutics, Linguistic Attractors, and the Limits of Model Self-Report

The Generative Horizon argues that, for language-model agents, language is simultaneously an object of interpretation and a medium of generation. The represented situation formed through instructions, prior turns, retrieved records, tool results, summaries, and corrections is not merely what a model computes about; it is part of what the model computes through. The paper calls that situated condition the generative horizon. When an output is retained as context, memory, policy, evidence, or authorization, one generation can alter the conditions of later interpretation and action. This is recursive interpretive conditioning: the output of one horizon becomes part of another. Hermeneutics supplies a vocabulary for historically formed and revisable understanding; applied hermeneutics translates that problem into requirements for provenance, status, revision, and authority. The argument does not attribute human historical consciousness, an inner observer, privileged access, or phenomenal awareness to language models. It treats mechanistic measurements and model-generated reports as representations whose evidential status must be established. Systems using summaries, self-reports, decoded features, reasoning traces, or evaluation labels should distinguish measurement from intervention, preserve correction and supersession, and bind a representation's authority to evidence proportionate to what it may govern. Mechanistic interpretability does not remove interpretation; it makes the problem experimentally and operationally precise. This work connects applied hermeneutics, mechanistic interpretability, model self-report, model introspection, reasoning-trace and chain-of-thought faithfulness, linguistic attractors, recursive interpretive conditioning, provenance, supersession, and epistemic engineering for large language model systems.