Source
Isaac (2025), Philosophy and the Mind Sciences 6, LIT-1361, read in full in NOTE-1163 from the publisher’s open-access PDF. §5 (“Making progress on consciousness”) with footnote 15, footnote 2 of §2.1, and the closing puzzle of §6.
What was actually shown
Nothing was tested in the paper. It answers the question “how can we be sure it studies phenomenal experience at all, rather than merely patterns of behavior and their neural substrate?” in three parts, the third its “own considered view”.
- Relations are behaviourally manifest. “phenomenal relations can be empirically investigated without independent access to phenomenal relata”. Whether your rose looks like mine cannot be settled, but whether it looks closer to the poppy or to the carnation can, through discrimination and matching, and the patterns can be compared.
- Introspection keeps the target. Phenomena found by observation (afterimages, the Purkinje shift) guide models at both levels, and phenomenology is “front-loaded” into design. But intent “is also only part of the answer”, because models such as CIELAB were built with no phenomenological motive.
- Convergence. CIELAB is derived from colour matching, “purely behaviorally realized judgments of indifference”; OSA-UCS from “explicitly reflective and consciously assessed judgments of sameness of distance”. Their metrics agree coarsely, with “some degree of distortion” (Oleari 2001, cited). “To the extent that they do exhibit the same structure, however, the fact that they are derived from different kinds of behavior, detouring in different ways through conscious phenomenal experience, constitutes evidence that they target experience itself.” The same is proposed for odour (descriptor atlases against perfumers’ practice) and timbre (similarity judgements against descriptors): experience “is the least common denominator”.
The paper states the counter-evidence itself. “observed convergence is never perfect”, and “in many cases, the discrepancies between different methods … are greater than the levels of uncertainty each method asserts individually” (also footnote 2). Some take this as evidence against a well-defined metric (MacLeod and Willen 1995 for visual space; Decock 2006 for colour; Barwich 2019 for smell). Isaac’s reply is that convergence is “a regulative ideal”: the structuralist is committed to explaining and correcting the discrepancies. And in §6 the models “rely constitutively on averaging over subjects”, so “the target is a kind of ideal, a ‘standard observer’”, which he leaves as a puzzle.
What could have failed: convergence could have been absent, and for visual space and odour the critics say it is.
What this does not say
- Not that convergence excludes the rival explanations. Two procedures can agree because they share the stimulus’s physical geometry, or a common decision and response process, without sharing an experience. The “least common denominator” argument names experience as the common factor and does not rule these out. That is the gap THEORY-1204’s promote_when asks to close.
- Not that convergence has been shown. It is coarse for colour, the best case, and disputed elsewhere.
- Not a measure. The paper gives no degree of agreement that would count as enough, and no account of which discrepancies are corrigible.
- Not about individuals. Averaged models target a standard observer; whether any individual’s experience is reached is left open.
- Not about whether a state is conscious, only about whether a model of character targets experience. Blindsight and masking are raised as the worry, and the reply is the three parts above, not a test for consciousness.
Connections
- THEORY-868. Negro holds that structuralist methods support only that experiences stand in relations. This account is a claim of that weaker, methodological kind, and is compatible with it.
- THEORY-880. Lyre’s programme assumes that a behaviourally built Q-structure is the structure of experience. This account is Isaac’s argument for that step, and its weakness is the same: shared stimulus geometry would also produce agreement.
- CLAIM-138. The relations here are fixed by named behavioural procedures, independently of the structure claimed, which is the remedy for the Newman objection that the claim describes.
- QUESTION-255. The question’s line is evidence for experience. This is one structuralist answer to what makes behavioural evidence evidence of experience, at the level of character only.
Note of 2026-10-30: narrowed
The old title stated the coarseness of convergence, and discrepancies beyond each method’s uncertainty, as the record’s own fact. The record has read only Isaac’s report of the comparisons (Oleari; MacLeod and Willen; Decock; Barwich are cited, not read), so the clause is now his concession.