The claim
CLAIM-440 makes valence a functional regime: an ordering of the system’s own states, by their bearing on its maintenance, prevails in the negotiation among its perspectives. It says nothing about when that regime is felt. This claim proposes the condition.
The condition. The regime is felt when the system also models it. A process at a deeper level infers three things about the negotiation: which ordering is winning, which way the system is moving along it against its reference course, and how strongly it prevails. The inference is used elsewhere in the system, to guide attention, report, or further regulation.
Where it comes from. Brouillet and Friston (THEORY-662) separate unfelt fluency from felt fluency. In the record’s reading of them (NOTE-785), unfelt fluency is “the subpersonal recognition of a change in precision, which orients attention”, and felt fluency is “a hierarchically deeper inference about it”. Hesp and colleagues (LIT-1468, NOTE-1111) put valence in a deeper inference: “agents infer their valence state based on the expected precision of their action model—an internal estimate of overall model fitness (“subjective fitness”)” (p. 399). The precision γ weights the preference-driven ranking of policies against the habitual prior, so “higher γ values afford a greater influence of the expected free energies of each policy entailed by one’s current action model” (p. 408): in this record’s terms, γ is a potential’s strength, and the deeper level models it. They add that “this type of explicit valence representation is also a necessary condition for self-reportable experience of valence” (p. 429).
A precedent and a warning in Scherer. The component process model gives feeling “a special status … as it monitors and regulates the component process”, and requires it to “centrally represent all information about the continuous patterns of change and their coherence in all other components” (LIT-1451, NOTE-1113, §3e). That is a model of the system’s own regime, as this claim requires, though of the components’ coherence rather than of which ordering wins. Scherer makes a second, separate claim, that “it is the degree of synchronization of the components … that will generate conscious experience”, which is nearer Veit’s identity than to a self-model view, and he places the monitoring representation only partly inside consciousness. So the model alone may not suffice for feeling, which is the gap QUESTION-059 marks. He also concedes that “all we can currently measure is the individual’s verbal account”, the limit QUESTION-266 is meant to get past.
What it would answer. CLAIM-411 asks which further organizational properties make an interpretation experiential, under the owner’s constitutive thesis (CLAIM-412). For affect, this is a candidate: the system’s model of its own potential. It bears on QUESTION-059, whether a state meeting a functional characterization of affect must be felt, without answering it: in the sense of “felt” adopted here, attended and available, the answer would be no, since the state is felt when it is also modelled, and an organization may or may not have such a model; but the question asks about feeling in a sense this claim declines (below).
For an organization. The model is the organization’s reporting, to those who govern it, of what wins its conflicts. On this reading Wells Fargo’s board (CASE-151), told that sales-integrity terminations were “in the range of 200-300”, had a model of the bank’s potential that misrepresented it. If the claim holds, that is mis-felt valence, not unfelt valence. Whether the misrepresentation was motivated is the report’s open question, not this claim’s.
Its relation to illusionism
The claim fits Graziano’s account (THEORY-873), on which reports of consciousness come from a schematic model of attention. Here the model is of potential, not attention, and it may be schematic in the same way. The claim does not say whether such a model explains feeling or explains it away. It adopts THEORY-662’s sense of “felt”, attended and available, and QUESTION-048’s distinction between access and experience stands: if “felt” is given a phenomenal sense beyond access, the claim’s third defeater applies.
Its rival in the component process model
Grandjean, Sander and Scherer (LIT-1452, NOTE-1115) hold that “consciousness of emotional reactions emerges when lower levels of processing are not sufficient to cope with the event and regulate the emotional process, particularly when the degree of synchronization between the components reaches a critical level and duration” (p. 484), and that then “higher level regulation strategies, such as reappraisal, must kick in. These strategies require high-level representational operations and are therefore usually conscious” (p. 492). Scherer and Moors restate it: awareness follows “after a second threshold of coherence has been reached” (LIT-1457, p. 737).
- Where they agree. Feeling is a representation that higher regulation uses, which is this claim’s condition in other words.
- Where they differ: the trigger. For them the regime’s own degree and duration recruit consciousness. This claim can adopt that as a hypothesis about when the self-model is recruited: strong and lasting potential, too costly to regulate below.
- Where they differ: the content. Their representation is of the components’ coherence, which has no sign and does not say which ordering wins. This claim’s model is of which ordering prevails.
- The case that would decide. Strong, lasting synchronization that no further regulation uses. Grandjean et al. predict it is felt; this claim predicts it is not. Their data split felt from unfelt episodes by report, so they do not decide it.
A second rival: the beautiful loop
Laukkonen, Friston and Chandaria (LIT-1363, NOTE-1063, THEORY-882) separate local “parametric depth”, a second-order inference about one precision such as “policy precision”, for which they cite Hesp et al., from epistemic depth, a system-wide recursive loop over all precisions. They hold that “even complex metacognition can potentially be unconscious or conscious depending on epistemic depth” (p. 20). This claim’s condition, a deeper process that infers potential and is used elsewhere, is parametric depth in their terms, and on their view it is not sufficient for awareness. Two readings follow. If “felt” keeps the thin sense adopted here, attended and available, their view is about something stronger and does not conflict. If “felt” is meant as awareness, their view predicts this claim’s second defeater, and the claim would need strengthening: the model of potential must be part of a global, recursive precision loop. Their proposal has no data and no simulation (NOTE-1063), so it decides nothing, but it names the distinction the claim must take a side on.
Felt what?
Berridge notes that ‘wanting’ can be appraised as pleasure by the person who has it (NOTE-1110), and Hesp et al. concede that their precision signal sits nearer ‘wanting’ than pleasure (p. 430). If the potential this claim’s model is of is motivational, as CLAIM-438 argues, the claim is a theory of felt motivation. Whether it is also one of felt pleasure depends on where ‘liking’ is in the account, which CLAIM-438 leaves open.
What it does not say
- That any particular system feels. The claim is a condition, and its test is QUESTION-266.
- That the model must be accurate. A model that misrepresents potential gives felt valence that misrepresents the system’s state, which is what the claim predicts of denial and of mis-reported organizations.
- That the deeper process must be a cortex, or a person. Any process in the system that infers its potential and is used elsewhere qualifies.
Note of 2026-10-24: what the self-model models
Asked “what’s our current characterization of what it means for an affect to be ‘felt’?” (workbench entry of 2026-10-24, U24), the exchange found a gap. If the self-model represents only the ordering that prevails, the claim explains felt motivational precedence better than felt pleasure from a perspective that does not prevail, and CLAIM-437 makes such perspectives common. A suggested extension, not yet the claim: felt affect is an accessible, consequential self-representation of a process-relative affective evaluation, prevailing or not. That allows pleasure felt without governing and motivation governing unfelt, and separates having an affect, feeling it, understanding it and letting it govern. Legibility (CASE-124) bears on which affective signals are registered and whether the representation has influence; it does not make the represented state phenomenal. An earlier turn (U5) proposed CASE-129, the alarm that learned against the alarm that was disabled, as this claim’s test case: accurate local evaluations can fail to reach the process that models the organization, and a model can represent the organization as calm when consequential signals were suppressed, so self-modelling is no infallible route from evaluative organization to felt affect. The three mechanisms, evaluation, the gate on its influence and the model of the state, can be manipulated separately, which is QUESTION-266’s design. THEORY-570’s Minsky, for whom feelings are descriptions of organizational cascades, is the nearest antecedent.
Note of 2026-10-31: Massumi’s “felt” is this claim’s unfelt
Massumi’s definition of affect, which Bencherki, Brummans and Vézy build their agency on, ends “with the corollary that the transition is felt” (LIT-1905, p. 1; THEORY-1282). His “felt” is not this claim’s. Microperception “is something that is felt without registering consciously. It registers only in its effects” (p. 4), and the onset of experience is “by nature imperceptible”. On this claim, a change that reaches other processes only through its effects, with no deeper process modelling it, is unfelt valence, as Brouillet and Friston’s unfelt fluency “orients attention”. So when the parent paper says agency “can therefore be felt, observed, experienced” (LIT-1902, p. 1363), its sense of the word does not meet this claim’s condition, and nothing in it bears on whether that condition is right. The two uses should not be merged. CLAIM-527 also separates the quantity Massumi’s affect is about from valence.
No status, title or warrant changes.