Synthetic Cognition

Abstract

This research note proposes synthetic cognition as an umbrella term for describing cognitive capabilities in artificial systems without requiring a prior conclusion about machine consciousness. Current discussions of advanced AI often collapse into a binary question: is the system conscious or not? That framing can obscure important distinctions among reasoning, metacognition, self-modeling, social cognition, adaptation, agency, autonomy, and continuity. These capabilities can be meaningfully observed and studied even when the separate question of subjective experience remains unresolved. The note presents a working vocabulary for this middle ground between computation and consciousness. It argues that terms such as reasoning, agency, and metacognition should retain their ordinary meanings, while synthetic cognition identifies the broader artificial context in which those processes occur. Phenomenal consciousness, including the possibility of qualia or subjective experience, is treated as a separate and substantially stronger question. Recent work on self-relevant internal states, agentic behavior, and continuity of experience is used to illustrate why this distinction may be increasingly useful. The goal is not to claim that current AI systems are conscious, but to provide clearer language for describing what they can already do while keeping uncertainty about consciousness explicit.