For more details, please check out our preprint and a 1-minute explanatory video:
📄 Preprint (bioRxiv): biorxiv.org/content/10.6...
🎥 1-minute video: www.youtube.com/watch?v=6MCV...
Together, these findings point toward a novel "structure-based taxonomy" of divergent experiences, grounded directly in phenomenology. Our results open up a new avenue to quantitatively assess the commonality and individuality of subjective experiences.
Crucially, however, this structural alignment is not universal. Indeed, we identified two distinct aligned clusters, corresponding to color-neurotypicals and atypicals. Moreover, we discovered "structural intermediates": individuals who do not robustly align with either cluster.
Importantly, this structural correspondence constrains the plausibility of the longstanding “inverted qualia” argument in philosophy: if my red were your green, the relational geometries would be incompatible, and our algorithm would not have identified such precise alignment.
We demonstrate that, among certain individuals, color qualia structures of four participants can be “correctly” aligned in an unsupervised manner. That is, "my red" corresponds to "your red" not by assumption, but because it represents the optimal mathematical mapping.
The key method is an unsupervised alignment of qualia structures. Our method finds the optimal, maximally structure-preserving correspondence between color experiences across individuals, without presupposing any fixed mapping, such as your “red" should correspond to my “red”.
By collecting a massive dataset of 4,371 similarity judgments of 93 colors from each of the participants, we empirically constructed individual color qualia structures. This unprecedentedly rich dataset enables us to directly address the original question.
Is your “red” my “red”? To tackle this age-old question, we utilized the “qualia structure” approach, which characterizes qualitative aspects of consciousness, or “qualia” through their internal relationships (i.e., “red” is similar to “orange" but dissimilar to “green”)
Excited to share our new preprint w/@yukoy.bsky.social , Chihiro Hiramatsu, @naotsuchiya.bsky.social & @oizumim.bsky.social ! We present the world’s first empirical evidence supporting that "my red" is "your red", through an unsupervised alignment of color qualia structures at the individual level.
