Does Total Knowledge Mute Emotion? A Thought Experiment using Large Language Models

Abstract This paper examines whether comprehensive knowledge acquisition attenuates emotional experience through a thought experiment that uses large language models (LLMs). We propose that LLMs, which possess extensive propositional knowledge yet lack phenomenal consciousness, serve as philosophical instruments for isolating the knowledge variable from other cognitive attributes. Our analysis develops two competing theoretical frameworks: the epistemic saturation hypothesis (ESH), which predicts emotional dampening via reduced prediction error, and the information hazard hypothesis (IHH), which anticipates emotional intensification via heightened awareness of negative states. We introduce a value uncertainty matrix that maps entities along the dimensions of knowledge completeness and embodied value schemas, thereby positioning hypothetical total knowledge agents (TKAs), LLMs, and humans within this conceptual space. Our synthesis suggests that total knowledge transforms rather than eliminates emotion and, therefore, shifts affective experience from uncertainty-driven volatility toward knowledge-dependent sublime states. This thought experiment reveals three insights: (1) vast propositional knowledge alone generates no intrinsic valence; (2) emotional experience requires embodied value computation; and (3) human cognitive enhancement toward total knowledge would likely preserve complex emotions while attenuating basic reactive effects. In parallel, the same thought experiment demonstrates how computational systems lacking consciousness can shed some light on fundamental questions about knowledge, emotion, and the nature of human experience.