What happened: Two UCLA psychologists propose a causal framework for the analogies by which consciousness is attributed. “Because subjective experience is directly accessible only in the first person, judgments about consciousness in other entities depend partly on analogy”; the framework “weights source-target similarity by causal relevance while allowing for unknown causes, disabling differences and alternative routes to consciousness.” Applied to animals it “explains why analogical support generally weakens with increasing causal distance from humans”; applied to AI “it suggests that behavioural similarity provides only limited evidence for consciousness because relevant causal correspondences remain poorly established.” The human mechanism is modelled as a soft INUS set; animals are “a calibration case”; “behaviors such as writing stories or coding websites should be assigned low causal weights”; the expectation that scale yields experience is “a variety of the ‘intellectualist fallacy’ first criticized by James.” Two guards are stated: “Failure of the analogy therefore does not establish absence of consciousness; rather, it shifts the evidential burden to independent support for a non-analogous mechanism”, and verbalisable internal vectors of the kind recently reported in language models are set beside symbolic systems that had comparable representations without anyone taking them to be conscious.
Diagnostic reading: This is the second framework in a fortnight to make a materialist weighting its first move and to arrive, by a different route, at structured uncertainty, and the entry reads it with Chandaria et al. as a pair. Where that group wrote the supervenience premise into one sentence and then spread credence across levels, Holyoak and Monti write it into the weights: the “factors plausibly involved in generating consciousness” are stipulated within the frame that takes experience to be generated by factors, and the term the framework reserves for “unknown causes” is the slot where FND-02 v5.8’s framing question sits — the project’s diagnostic shows up as a parameter the authors are candid enough to carry rather than drop. The paper is also, in effect, a reply to the held global-workspace entry (2026-07-06-anthropic-global-workspace): verbalisable vectors are an effect of whatever is going on, not a plausible cause of experience, so they weigh little — the same move FWK-03 v1.5 tracks as criteria movement, here made in advance and on stated grounds, which is the legitimate form. ARG-01 v1.5 has purchase at exactly the point the authors would call principled asymmetry. Animals are calibrated by homology to a human source; machines are weighted by stipulated causal correspondence; the human is never a target. Run symmetrically, the question is what weight the framework would assign to a human whose only evidence was behavioural — a stranger, a patient — and the answer, by its own logic, is the same low weight, which is not how anyone attributes. The authors see part of this: the framework “does not assume that the human route to consciousness is the only possible one”, and a failed analogy shifts the burden rather than closing the question. What they do not see is that the burden-shift is itself the boundary problem (ARG-02 v1.7): the “independent support for a non-analogous mechanism” would have to be certified by criteria that are, once again, drawn from the single case. Partial, then — animals are treated and the human comparison is the framework’s foundation, but the reciprocal run is not made. Stable on the direction axis, and the more valuable for it: this is what criteria look like when the weighting is argued before the evidence arrives rather than adjusted after it, and the project’s reading is that the argument is sound within its premise and that the premise is where the two frameworks of October meet. [Derived]
Preprint (arXiv:2610.01002v1 [cs.AI], 1 October 2026, 11pp, two boxes; UCLA Psychology and the Brain Injury Research Center), read in full from the arXiv HTML 2026-10-15, quotations verified. A method paper with a formula and no data; the framework organises evidence and assigns no probabilities. Read as a pair with the Chandaria et al. entry; the two are independent and reach the same shape of conclusion from different disciplines.