Is There Even an Elephant in the Room?

The science of consciousness has produced over 350 distinct theoretical accounts, yet the field lacks a systematic means by which these theories can be brought into structured dialogue with one another. Existing resources serve important functions — Kuhn's Landscape of Consciousness provides comprehensive cataloging, the Stanford Encyclopedia offers philosophical depth, and individual publications develop their respective positions — but none provides a framework for direct, standardized comparison, and some explicitly disavow any attempt to adjudicate among theories. Without such structured dialogue, the situation resembles the parable of the blind men and the elephant: each theory may be describing a genuine aspect of the phenomenon (the elephant), but without structured communication regarding assumptions, predictions, and experimental implications, these accounts cannot be reconciled or integrated. If, alternatively, the differences in starting assumptions reflect that there is no unified phenomenon to converge upon (no elephant), this too can only be recognized through rigorous cross-examination of what each theory claims.

We present ElephantRoom, an open-access platform for structured comparison of empirically testable theories of consciousness. Building on the consciousness research landscape mapping of Kriegleder and Noichl, which charted the field's structure, this project moves from overview to evaluation. Theories that meet explicit inclusion criteria — (1) self-nomination as a theory of consciousness, (2) at least two peer-reviewed publications, and (3) at least one empirically distinguishable prediction — are put through a standardized set of five questions centered on the hard problem. The questions are designed to be theory-neutral, forcing concrete claims about mechanisms, boundary conditions, and falsifiability rather than inviting meta-theoretical positioning. Responses are extracted from the source literature using large language models and subsequently refined through community review. Based on the quality and substance of these answers, the AI also assigns scores on five orthogonal evaluation dimensions — Hard Problem, Mechanism, Empirical Fit, Boundary Precision, and Falsifiability — and the results are presented as radar charts that preserve the multidimensional profile of each theory rather than collapsing it into a single composite score.

A central challenge in any comparative exercise is moderation. Academic discourse on consciousness is prone to the social dynamics inherent in evaluating competing research programs: defensiveness, scope disputes, and arguments over weighting criteria. We explore the use of AI not as an objective arbiter — no such thing exists — but as a uniformly applied analytical lens that holds no allegiance to any particular theory. The AI generates initial content from source papers using identical prompts and evaluation rubrics for every theory, providing a consistent baseline that the research community can then challenge, refine, and override. This division of labor — AI as scaffold, community as authority — aims to make comparison tractable at scale while keeping legitimacy where it belongs: with the researchers.

The platform currently includes fifteen theories spanning multiple levels of analysis (cellular to whole-organism) and theoretical orientations (physicalist, functionalist, eliminativist, emergentist, embodied). It is designed to grow through community contributions: researchers may submit new theories, propose questions, edit AI-generated content, and challenge dimensional scores. The goal is not consensus but productive confrontation — surfacing the specific empirical disagreements between theories and identifying the experiments that could resolve them.