Source
Aldé (2026), LIT-1348, Synthese 207:204, read in full in NOTE-1055: the colour composition claims (§3, pp. 7–10), the unique hues objection (§4.1, pp. 11–13), the opposing dimensions objection (§4.2, pp. 13–15) and the reply to introspective doubt (§5, pp. 15–18).
What was actually shown
- The target. “The phenomenal character of a mental quality is fully determined by its similarity relations with the other phenomenal characters in the similarity space” (p. 4). If so, everything about colour character must be a geometric feature of the colour solid: once colours and axes are removed, the solid “encodes only its own shape” (p. 11).
- The facts. Exactly four hues are experienced as unmixed and the rest as weighted combinations of two adjacent ones, with red–green and blue–yellow combinations “prohibited”; with white and black these give six pure colours (p. 7). The evidence is reported from colour science: Hering, hue scaling, colour matching, and cross-cultural and infant studies (pp. 7–9). The critics’ studies are answered, not re-run (pp. 9–10).
- The unique hues objection. No geometric property of CIELAB distinguishes the unique hues from composite ones; better metrics are unlikely to; and colours missing from current models (chimerical colours, “olo”) would add mainly saturated teals and light blues (pp. 11–13, Figure 3).
- The opposing dimensions objection. The black–white axis is one gradient; axes through grey are two half-gradients meeting at grey. Current models are asymmetric, so every axis is geometrically unique, “yet only the black-white axis is so peculiar in its phenomenology, and it is not equally peculiar in its geometrical uniqueness” (p. 15). This argument does not need the unique hues.
- Introspection. Similarity structuralism is built on introspected similarity, so it cannot dismiss introspected composition without reason, and judgements that a colour contains others are component analyses of an ordinary, reliable kind (pp. 15–18).
What could have come out otherwise: the unique-hue loci might have been geometrically special, as extremal points or singularities. The paper argues they are not, by inspection.
What this does not say
- It does not show premise 3 by computation. No invariant search over a uniform colour space is reported; the paper says one “could” be done (p. 12). The promotion test asks for it.
- It does not refute structuralisms with other relations. Rosenthal’s discrimination spaces are set aside (pp. 4–5), and relations of composition, causation or neural organization are not considered. The alternative it points to, the opponent-process mechanism, is itself a structure, though not a similarity structure.
- It is about colour only. Other modalities are left open, with the conjecture that they would need the same non-structuralist treatment.
- The unique hues are not settled science. The paper grants that their stimulus loci vary widely across subjects and are not opposite in colour space, and that their neural basis is “elusive” (p. 9).
Connections
- THEORY-880 is the target, quoted from Lyre (2022). Lyre’s own reply to inversion relies on colour space being “highly asymmetric”; Aldé grants the asymmetry and argues it is the wrong asymmetry.
- THEORY-868 (Negro) objects to the same view from intuition. Aldé presents his arguments as stronger because they do not.
- CLAIM-412. The claim’s relations are interpretive, not similarities, so this THEORY does not reach it directly. It sets a test any account of character must meet: the composition of colour experience must come out of whatever organization is said to fix it.
- CLAIM-136. By analogy: Aldé states that a similarity structuralist must treat symmetric points as one point (p. 5), the strict version of CLAIM-136’s thesis applied to phenomenal characters.
Note of 2026-10-25: Isaac and the unique hues
Isaac (LIT-1361, read in NOTE-1163) is not the similarity structuralist this THEORY targets. He holds that geometry “by itself is not enough to generate meaning” and adds layers of constancy and of valence fields, so the conclusion here, that similarity relations alone cannot account for colour composition, does not reach his view as stated.
His DKL case bears on the test this THEORY sets, not on its premise. DKL, built from LGN opponency, “fails to accurately predict key phenomenal features, such as the appearance of unique hues” (Wuerger et al. 2005): pure yellow should be maximal activation at the yellow end of the blue–yellow circuit with the red–green circuit neutral, and it is not. That is a neural space failing to yield the unique hues, not the similarity geometry failing to mark them. Isaac reads it as a residual that motivates richer physiological correlates beyond LGN, not as a refutation. He also names hue superimportance (Judd 1970), hues being more discriminable than saturations, as one reason CIELAB is not an isotropic three-dimensional space, which bears on the source’s reliance on inspecting CIELAB. Neither is the invariant search the promotion test asks for. The comparison is the record’s.