Interpreted Temporal Fields - Anticipation and the Composition of Generative Assessment

Posted on 2026-07-20 In Philosophy , Social Science Views: Disqus: Word count in article: 23k Reading time ≈ 1:25

ENGLISH

Interpreted Temporal Fields

Anticipation and the Composition of Generative Assessment

Wanhong Huang · huangwanhong@serendip.ngo

Abstract

A previous paper proposed generative obstruction as a primitive explanatory object, defining it over the continuations available to a system in fact and restricting attributions of obstruction to parties within a relation. It deferred a class of cases its resources could not examine: conflict arising from foreclosure a party anticipates in place of foreclosure it undergoes. This paper takes up that case. It argues that the reference models the earlier paper established are temporally composite, assembled from an interpreted past, a present state, and an anticipated future, each entering through interpretation; and that a party’s assessment of its own generative prospects is a function of all three. The extension reaches anticipatory conflict and offers a reading of grievance without introducing a further primitive, since both operate through models the earlier framework already required. Two limits are stated. The composition of the estimated space is tense-neutral while its evolution remains temporally ordered, which preserves the path-dependence of the earlier account and forecloses a stronger reading sometimes drawn from arguments of this kind. And the extension sharpens an evidential difficulty already present: where obstruction is assessed over an interpretive field, it becomes inaccessible to observation that lacks access to that field, so the framework grows in causal reach as it contracts in testability.

Keywords: generative obstruction; estimated generative space; anticipation; interpreted past; grievance; generative relational being.

A note on the standing of this paper. This is the second paper in a series, and it depends throughout on the first. Its central move is to substitute, as the object over which obstruction is assessed, a party’s model of its available continuations for those continuations themselves. The substitution is defended as an extension of resources the first paper already required in place of a new commitment, and readers who reject the first paper’s account of reference models will find little here that stands without it. Several claims are entered as open questions where the resources to settle them were unavailable. Objection, correction, and counter-evidence are welcome at huangwanhong@serendip.ngo.

§1 Introduction

The preceding paper in this series [1] proposed generative obstruction as a primitive explanatory object and defined it by a partial order on the continuations available to a constrained generative system: an interaction obstructs a system when the continuations available to it under the interaction are properly contained in those available otherwise. Conflict was treated as the manifestation of such obstruction, and the mechanisms by which obstruction is produced were arrayed without privilege among them.

That paper also established, in the course of addressing a difficulty in its own definition, that attributions of obstruction require reference models. The definition compares what is available under an interaction with what is available otherwise, and the comparison requires a specification of the otherwise. No such specification is available as a fact about a relationally constituted system, since the counterfactual it requires removes the couplings within which the system was constituted. The earlier paper proposed in place of a fact a working model, held by a party for the purpose of coordination, carrying no ontological standing and answerable to nothing beyond its role. It argued further that a model treated as accurate degrades the channel by which it would be corrected, and that the rigidity with which a model is held accordingly bears on obstruction where the model’s content does not.

One class of cases was deferred. A party may act on a foreclosure it anticipates in place of one that has occurred. The colleague has not been promoted, the resource has not been allocated, the position remains open; and the party conducts itself as though its continuations were already closed. Conflict of this kind is common, and the earlier framework assigned it outside its scope, since the criterion there concerned foreclosures that had taken place. The assignment was made on a reading the earlier paper declined to examine, and it left a note to that effect.

This paper examines the reading. Its thesis may be stated compactly. The comparison Definition 1 requires has two terms, and the earlier paper read both as concerning what is available in fact. Read as concerning what a party’s working model represents as available, the definition reaches anticipated foreclosure directly, since a model of what is available is what a projection of the future supplies. Write $\hat{G}$ for the space of continuations a party’s model represents as available, distinguishing it from the space $G$ of continuations available in fact. The proposal is that obstruction, as parties attribute it and as it conditions their conduct, is assessed over $\hat{G}$.

The thesis carries three consequences, and the paper is organized around them. The first concerns what $\hat{G}$ contains. A model of available continuations is assembled from what a party takes to have happened, what it takes to obtain now, and what it takes to be coming. Each of these enters through interpretation. The past enters as reconstructed in place of as it occurred, and the same event admits reconstructions that support incompatible assessments. The future enters as projected, and a projection conditions conduct whether or not the projected event occurs. The present enters as construed, since a party’s reading of its current constraints is itself a reading. Assessment is accordingly a function of three temporally distinguished arguments, each interpretation-mediated, and the composition of that function is the subject of §3.

The second concerns what this leaves standing. An argument of this shape can be pressed toward a stronger conclusion, that the present enjoys no privileged status in generative assessment. The stronger conclusion is resisted here, and the resistance matters for what the earlier paper established. Assessment occurs at a time, over a field composed at that time, and issues in conduct at that time. The argument dissolves the privileged status of the present as content: a party’s current circumstances have no special claim to be the argument that matters, and interpreted pasts and projected futures enter alongside them. The present survives as locus: the evaluation happens now, and the field evolves from one such evaluation to the next. This distinction preserves the path-dependence the earlier paper derived, and abandoning it would remove that result.

The third concerns what the extension costs. The earlier paper found that reference models are licensed for parties within a relation and unlicensed for observers outside one, which restricted well-formed attributions of obstruction to the parties themselves. The present extension sharpens that restriction. Where obstruction is assessed over an interpretive field, an observer requires access to that field, and access to a party’s model of its own prospects is harder to obtain than access to the party’s circumstances. The framework accordingly gains in causal reach and loses in testability, and §6 states the trade without softening it.

One clarification belongs at the outset, since a claim in this vicinity is easily overstated. The earlier paper declined to specify a trajectory a system ought to have followed, on the ground that the specification would require a warrant the descriptive facts fail to supply. The move to $\hat{G}$ might appear to answer that difficulty, since a projected trajectory raises no question of what ought to be. It does answer it, and it answers it by abandoning the ground on which any trajectory could be criticized: a model may be arbitrarily mistaken, and the framework retains no resources with which to say so. The difficulty is avoided in place of solved, and §6 declines the stronger claim explicitly.

The remainder proceeds as follows. §2 distinguishes $\hat{G}$ from $G$ and argues for its causal efficacy. §3 sets out the temporal composition of $\hat{G}$ and the functional form of assessment, and establishes what is tense-neutral and what is not. §4 treats three instances: a reinterpreted past, an anticipated future, and grievance. §5 extends the earlier paper’s result to the temporal field, and supplies the conditions without which the extension would reach too far. §6 states what the extension costs, and §7 what it leaves open.

§2 The Estimated Generative Space

This section introduces the object over which the remainder of the paper operates, distinguishes it from the object the earlier paper employed, and argues that it bears on conduct in a manner the earlier object does not.

2.1 The Estimated Space Distinguished from the Available Space

The earlier paper wrote $G$ for the space of continuations available to a system: what it can go on generating from where it stands, given its history and its constraints. Obstruction was defined by proper containment on this space, $G’ \prec G$, and the definition concerned continuations available in fact.

Write $\hat{G}$ for the space of continuations a party’s working model represents as available. The circumflex carries the sense the earlier paper assigned it: this is a provisional construct held for coordination, subject to revision, carrying no ontological standing.

The two spaces are related loosely and come apart in both directions. A party may represent as available continuations that are foreclosed in fact. A researcher who takes a line of work to remain open when its premises have failed holds a $\hat{G}$ larger than $G$ in that respect. A party may equally represent as foreclosed continuations that remain available. A candidate who takes a position to be filled when it remains open holds a $\hat{G}$ smaller than $G$.

The divergence is the normal case in place of a defect. §3 argues that $\hat{G}$ is assembled from interpreted materials, and interpretation of a system’s own prospects proceeds under conditions that guarantee imperfection: the party’s information is partial, its constraints are imperfectly known to it, and its future is underdetermined by what it can observe. A $\hat{G}$ coinciding with $G$ would be an achievement requiring explanation.

The earlier paper’s treatment of reference models applies here without amendment. A model that diverges from what it models is not thereby faulty, since it was never held as a representation answerable to correspondence. It is held for coordination and answerable to whether it serves that. The earlier paper added a condition on how such models are held, and §5 extends that condition to the present case.

2.2 The Space to Which Conduct Answers

The claim on which this paper turns may now be stated.

Claim 1. A party’s conduct is conditioned by $\hat{G}$. Where $\hat{G}$ and $G$ diverge, conduct answers to the former.

The argument for it is short. A party acting on its prospects acts on prospects as it takes them to be. The continuations available to it in fact bear on what its actions achieve; they bear on which actions it selects only through its representation of them. A party unaware that a continuation is foreclosed pursues it; a party persuaded that an available continuation is closed abandons it. This is a general feature of action under a representation and requires no special apparatus.

Two qualifications keep the claim within bounds. It concerns selection of conduct in place of outcome. Conduct selected against a mistaken model meets the world as it is, and what results is determined by $G$ in the ordinary way. A party pursuing a foreclosed continuation fails to reach it. The claim asserts the occurrence of the pursuit, and attributes that occurrence to $\hat{G}$.

The claim leaves the relation between the spaces open. Nothing here holds that $\hat{G}$ is unconstrained by $G$, and a great deal ordinarily constrains it: a party encounters its constraints, receives correction, and revises. The claim asserts of the divergent case alone that action answers to the model.

2.3 The Causal Efficacy of Representations

Claim 1 has a consequence that carries the paper’s principal extension. If conduct answers to $\hat{G}$, and if conduct enters the conditions under which systems generate, then a representation bears on generative dynamics whether or not it is accurate.

The structure was described by Merton as the self-fulfilling prophecy [11], and one instance is worth stating because its mechanism is transparent. Where participants in a market take a decline to be coming, they act on the expectation by selling, and the selling produces the decline. No fact about the market’s prospects need have obtained beforehand. The expectation was mistaken at the moment it was formed and accurate by the time it was assessed, and what closed the gap was conduct the expectation itself conditioned.

The same structure operates in the cases this paper treats. A party takes a colleague’s advancement to foreclose its own prospects. The advancement has not occurred and the foreclosure is a projection. The party conducts itself accordingly: it withholds cooperation, it competes where it would have collaborated, it treats the colleague as an obstacle. The colleague, meeting this conduct, responds in kind. The prospects of both are now diminished in fact, and the diminution was produced by a representation of a diminution that had not occurred.

The argument of §5 depends on three features of this structure. The representation does its work before the represented event. Conduct is conditioned at the moment the representation is held, and the represented event may occur later, or otherwise, or never. This is what places anticipated foreclosure within the framework’s reach and is the substance of what the earlier paper deferred.

The representation may be self-confirming. Conduct conditioned by a representation may produce the state the representation asserted, and where it does, the party receives confirmation its own conduct arranged. This is the closure structure the earlier paper described in its treatment of symbolic mechanisms, arising here in the particular case of a model of one’s own prospects.

The mechanism operates without error on either side. Neither party in the case above need be mistaken about anything except the initial projection, and the initial projection need be no more than a reasonable reading of an uncertain situation. This matters for the same reason it mattered in the earlier paper: a mechanism requiring ill intent or gross error would explain a narrower class of cases than the phenomena present.

2.4 The Status of $\hat{G}$ Within the Series

A framework proposing a minimal primitive base owes an account of its inventory whenever a new object appears prominently, and the account is due here.

The earlier paper established four things about reference models: that attributions of obstruction require them, that parties hold them, that they are provisional, and that they condition conduct sufficiently for their rigidity to bear on obstruction. The present section names the space such a model represents and asserts that conduct answers to it. Both follow from the earlier arguments and add to them nothing the earlier paper lacked.

The change lies in prominence. The earlier paper introduced reference models to address a difficulty in its definition and treated them thereafter as apparatus, subordinate to the space $G$ over which its primitive was defined. This paper takes them as the object over which assessment is conducted, so that $\hat{G}$ occupies in the present treatment the position $G$ occupied in the earlier one. The shift is substantial and it is a shift of emphasis, leaving the inventory unaltered.

Whether an emphasis of this magnitude preserves the minimality the earlier paper claimed is a fair question. §6 takes it up, together with the other costs the extension carries.

§3 The Temporal Composition of the Estimated Space

§2 established that conduct answers to $\hat{G}$ and left the composition of that space unexamined. This section supplies the composition, states the functional form assessment takes, and fixes the two respects in which the account is temporally ordered and the one in which it is not.

3.1 Three Horizons

A party’s model of the continuations available to it draws on three sources, distinguished by their temporal position.

The first is the past as the party reconstructs it. A history of foreclosures, achievements, and constraints enters the model as evidence bearing on what remains available, and it enters as reconstructed in place of as it occurred. That recall is reconstructive in place of reproductive is a long-standing finding in the experimental literature on memory [2, 10], and the present claim is the weaker one that a party’s reading of its own history enters $\hat{G}$ as read. It follows also from the earlier paper’s account of generation: what a system has produced enters the conditions of its subsequent production, and a party’s reading of its own history is one such production, revised as later productions condition it.

The second is the present as the party construes it. Current constraints bear on available continuations directly, and they bear on them through a construal: a party’s reading of what presently constrains it is a reading, subject to the same partiality as its reading of anything else. The present enters the model as interpreted material alongside the other two, its immediacy notwithstanding.

The third is the future as the party projects it. Projections of coming states bear on the model with no less force than the other two, and §2 argued that they condition conduct before the projected states obtain or fail to. A projection is interpretive throughout, being assembled from the party’s reading of its situation and its expectations about what follows from it.

Each horizon enters through interpretation, and the uniformity is the section’s first claim. Work in the memory literature reports substantial overlap between the processes supporting recollection of the past and those supporting simulation of the future [14]. That finding concerns cognitive architecture in place of the compositional claim entered here, and it is noted as consistent with the uniformity in place of evidence for it.

Claim 2. The estimated generative space $\hat{G}$ is composed from three interpretively mediated sources: a reconstructed past, a construed present, and a projected future. No source enters the composition unmediated.

The claim’s content lies in the uniformity of the mediation. An account admitting the interpretive character of projections while treating the past as fixed material would sort the horizons by their epistemic standing, on the ground that the past has occurred and the future has not. Claim 2 declines the sorting. Occurrence fixes an event; it leaves open what the event contributes to a model of available continuations, and the contribution is what enters $\hat{G}$.

3.2 The Functional Form

Claim 2 licenses a compact statement of the assessment a party performs. Write $\Pi$ for the reconstructed past, $\Sigma$ for the construed present, $\Phi$ for the projected future, and $\hat{G}$ for the estimated space assembled from them:

$$\hat{G} ;=; F(\Pi, \Sigma, \Phi).$$

The expression is a schema in place of a model, and the distinction should be entered before it is used. It fixes what the arguments of an assessment are and in what manner they stand to the assessment; it specifies no relation among them. Nothing here supplies $F$, and whether the horizons compose additively, whether they interact, and whether one dominates under specifiable conditions are open questions entered in §7. No claim of this paper is derived from the expression. Every argument below runs in prose and would survive the notation’s removal, which compresses the statement of a dependence in place of establishing one.

The arguments are the party’s, and the assessment is the party’s. $\Pi$, $\Sigma$, and $\Phi$ are the party’s reconstruction, construal, and projection, and $\hat{G}$ is the space its model represents. The earlier paper’s restriction of attributions to parties within a relation applies to the whole expression, and §6 states what this leaves an observer able to say.

The form concerns assessment in place of capacity. The space $G$ of continuations available in fact is determined by the system’s history and constraints in the manner the earlier paper described, and $F$ says nothing about it. Where the two diverge, Claim 1 holds that conduct answers to $\hat{G}$, and $G$ determines what that conduct achieves.

3.3 Composition and Evolution

An argument of the shape given above can be pressed toward a conclusion this section resists, and the resistance requires statement because the conclusion is attractive and its cost is easily missed.

The attractive conclusion is that generative assessment is temporally unstructured: that since all three horizons enter the same evaluation on the same footing, temporal position is immaterial to the framework, and the present enjoys no privileged status. The first half of this is granted below. The second half would remove a result the earlier paper derived, and it is declined.

The distinction the section proposes is between composition and evolution.

Composition is tense-neutral. The three horizons enter $F$ on the same footing, and their temporal positions do not order their contributions. A foreclosure reconstructed from twenty years ago may bear on $\hat{G}$ more heavily than a constraint obtaining now, and a projection of a foreclosure that may never occur may bear more heavily than either. The horizons are distinguished by their sources and are not ranked by them. This much of the attractive conclusion is granted, and §4 exhibits it.

Evolution is temporally ordered. The assessment $F(\Pi, \Sigma, \Phi)$ occurs at a time. Its arguments are the party’s reconstruction, construal, and projection as they stand at that time; its output conditions conduct at that time; and that conduct enters the conditions under which the party reconstructs, construes, and projects at times following. Write $\hat{G}_t$ for the estimated space at $t$: the sequence ${\hat{G}_t}$ is a trajectory, ordered, and each term is conditioned by its predecessors.

Claim 3. The composition of $\hat{G}$ is tense-neutral: the three horizons enter assessment without ordering by their temporal positions. The evolution of $\hat{G}$ is temporally ordered: assessment occurs at a time, conditions conduct at that time, and conditions subsequent assessment through that conduct.

The pair carries a consequence the earlier paper’s central derivation requires.

That paper derived the path-dependence of obstruction from the historical conditioning of generative capacity: what a system can produce next depends on what it has already produced, so that whether an interaction obstructs depends on the history in which it falls. The derivation requires an ordered sequence of states. An account on which temporal position is immaterial throughout would supply none, and path-dependence would be lost.

Claim 3 preserves the derivation by locating the tense-neutrality where it belongs. The composition of an assessment draws on three horizons without ranking them; the sequence of assessments is ordered, and the earlier paper’s argument applies to that sequence unchanged. The estimated space evolves path-dependently over an interpretive field in place of a physical one, and the structure of the argument survives the substitution.

3.4 The Present as Locus

One further distinction follows from Claim 3 and is worth drawing explicitly, since the vocabulary of privilege has been used above in two senses.

The present loses its privilege as content. A party’s current circumstances have no special claim to determine its assessment of available continuations, and they compete with reconstructions and projections on terms Claim 2 makes uniform. This is a substantive result. Accounts locating a party’s prospects in its present situation are, on this reading, attending to one argument of $F$ and treating it as the whole.

The present retains its privilege as locus. Assessment happens now, over a field composed now, issuing in conduct now. The reconstructed past is a present reconstruction; the projected future is a present projection; and the party performing the assessment occupies the present in a sense the horizons it draws on do not. Removing this would leave the framework unable to locate an assessment in time, and with it the ordering Claim 3 requires.

The two senses are compatible and the distinction between them is what permits the section’s argument to run. A framework may hold that the present has no special claim to determine what a party takes to be available, while holding that the taking occurs in the present. The first is a claim about the arguments of $F$; the second is a claim about where $F$ is evaluated.

§4 Three Instances of Temporal Composition

The claims of §3 are exhibited here on three cases, selected to isolate the contribution of each horizon. The first varies the reconstructed past with the other two held fixed; the second varies the projected future likewise; the third exhibits a case in which a reconstruction long past dominates an assessment. The cases are constructed, and the standing of constructed cases within this series is as the earlier paper described: they exhibit an argument in a particular setting, forgoing evidential weight for the control they supply.

4.1 A Reconstructed Past

Two parties stand in a close coupling. One failed to mark an occasion the other treated as significant, and the failure lies some months back. The event is settled: it occurred, it occurred as it occurred, and neither party disputes the description.

The reconstructions available to the second party differ, and two are worth setting beside each other.

On the first reconstruction, the failure indicates a disposition: the party who failed attends to the other’s concerns weakly, has done so throughout, and will continue. The reconstruction bears on $\hat{G}$ directly. Continuations requiring sustained attention from that party are represented as unavailable, since the disposition the reconstruction attributes forecloses them. The estimated space contracts.

On the second, the failure indicates a circumstance: the party who failed was under load, managed the occasion poorly, and attended as well as the circumstance permitted. This reconstruction leaves the same continuations represented as available, since a circumstance passes where a disposition persists.

The event is identical across the two. Its contribution to $\hat{G}$ differs, and it differs because reconstruction differs. This is Claim 2 exhibited on the first horizon: the past enters the estimated space as reconstructed, and the reconstruction is what the space is composed from.

The reconstructions are not distinguished by accuracy, and the case is constructed so that neither party is positioned to settle which is correct. A disposition and a circumstance are compatible with the same behaviour on a single occasion, and the evidence that would separate them lies in conduct yet to occur. The party reconstructing is choosing under underdetermination, and the choice conditions conduct before the evidence that would inform it arrives.

The conduct so conditioned bears on the evidence. A party representing continuations as foreclosed conducts itself accordingly: it withdraws the attention it takes to be unreciprocated, and the withdrawal alters what the other party does. The reconstruction thereby participates in producing the evidence by which it would be assessed, which is the structure §2 described and §5 takes up.

4.2 A Projected Future

A party works within an organization in which a position will be filled. A colleague is a candidate. The position remains open, the decision lies ahead, and no allocation has occurred.

The party projects. On the projection, the colleague is appointed, and the appointment forecloses the party’s own advancement for a period it estimates as considerable. The projection enters $\hat{G}$, and the estimated space contracts: continuations requiring the advancement in question are represented as unavailable.

Conduct answers to the contracted space. The party withholds cooperation it would otherwise extend, competes where it would have collaborated, and construes the colleague’s conduct through the projection it holds. Note that no continuation available to the party in fact has been foreclosed. The space $G$ stands as it stood; the party acts on $\hat{G}$.

The colleague, meeting this conduct, responds to it. Cooperation withdrawn is cooperation unavailable, and the colleague adjusts its own model accordingly, which conditions its conduct in turn. The prospects of both parties are diminished in fact by the exchange, and the diminution was produced by a projection of a diminution that had not occurred.

This is the case the earlier paper deferred, and the present resources reach it. The earlier framework’s scope criterion asked whether satisfying a demand would foreclose continuations, and whether frustration degraded capacity describably. Anticipated foreclosure satisfies neither, since no foreclosure has taken place. Read over $\hat{G}$, the case is straightforward: the party’s estimated space has contracted, its conduct answers to the contraction, and the obstruction it attributes is one its model represents.

The case establishes less than it might appear to. The projection may be accurate. The colleague may be appointed and the advancement may be foreclosed, in which case the party’s conduct was answering to a space that came to coincide with the available one. The mechanism described here operates in both cases, and its operation does not turn on the projection being mistaken. The case exhibits the conditioning of conduct before the projected event resolves, whichever way it resolves.

4.3 A Reconstruction Long Past

The third case exhibits the tense-neutrality of composition claimed in §3.3, and it treats a phenomenon the earlier paper listed among the mechanisms it had omitted, one with a substantial literature of its own [7, 3].

A party sustained a foreclosure some years back: an opportunity was closed to it by another party’s action, and the closure was real. The conditions that produced it have since passed. The other party’s position has changed, the arrangement that permitted the closure has been dissolved, and nothing presently obtaining would produce a comparable foreclosure.

The reconstruction remains. On it, the other party is one that forecloses, the arrangement is one that permits foreclosure, and continuations requiring the other party’s cooperation are represented as unavailable. The estimated space is contracted by a reconstruction of conditions that no longer obtain, and conduct answers to the contraction.

The case turns on three features. The reconstruction outweighs the construed present. The party’s present circumstances would support an assessment on which the continuations in question are available, and the reconstruction dominates them. This is the tense-neutrality of composition: horizons enter $F$ without ranking by temporal position, and a reconstruction from years back may bear more heavily than a constraint obtaining now.

The reconstruction is accurate as to what occurred. The foreclosure happened, and the party’s account of it is correct. The case exhibits a reconstruction accurate about the past and inaccurate about available continuations, and the two come apart because the conditions producing the foreclosure have passed while the reconstruction has not been revised.

Grievance, so described, requires no new mechanism. The earlier paper listed grievance among the mechanisms it had omitted and offered no argument that it reduces to those it treated. The present account reaches it as a case of temporal composition: an unrevised reconstruction contracting an estimated space, with conduct answering to the contraction. This offers a reading of a mechanism the earlier framework lacked, through resources §2 argued were already present. Whether the reading captures what the grievance literature describes is a question for that literature to settle.

4.4 The Structure Common to the Three Cases

The cases differ in which horizon carries the assessment and agree in three respects. In each, the estimated space diverges from the available one, and conduct answers to the estimated space. In each, the divergence arises through interpretation of material the party possesses in place of through ignorance of material it lacks. And in each, the conduct so conditioned bears on the evidence by which the interpretation would be assessed, so that the interpretation participates in producing its own confirmation.

The third respect is where the paper’s extension of the earlier result operates, and §5 takes it up. Absent that treatment, the cases would establish only that interpretation matters, which is common ground and would extend the framework by nothing.

§5 Rigidity Over the Temporal Field

The cases of §4 exhibit interpretation entering assessment and conditioning conduct. Taken alone they establish that interpretation matters, which the framework requires no apparatus to say. This section supplies the apparatus, extending the earlier paper’s result to the estimated space and fixing the conditions without which the extension would reach further than its argument supports.

5.1 The Extension and Its Necessity

The earlier paper argued that a working model held of another party degrades the channel by which it would be corrected, where the model is treated as accurate in place of provisional. The argument proceeded in three steps. A correction indicating that the model errs is anomalous on a model treated as accurate, which makes available a construal attributing the anomaly to the corrected party in place of the model. Corrections so construed achieve less than the correction intended and cost the correcting party more than they return, so their rate falls. The model is thereafter held against a thinner correction channel, and the conditions permitting the construal are strengthened.

The estimated space is subject to the same argument, and the extension requires one observation. The terms composing $\hat{G}$ are interpretations: a reconstruction of the past, a construal of the present, a projection of the future. Each is corrigible in principle, and each is corrected in the same manner, by material the interpretation failed to anticipate. A reconstruction is corrected by conduct the disposition it attributes would fail to produce; a projection is corrected by the projected event resolving otherwise; a construal is corrected by constraints proving other than construed.

Where a term is held as accurate, the argument applies to it unchanged, and the correction channel for that term degrades.

Claim 4. The terms composing $\hat{G}$ are subject to the degradation the earlier paper described. Where a reconstruction, construal, or projection is held as accurate in place of provisional, the channel by which it would be corrected degrades, and the term tends toward unrevisability.

Claim 4 might be read as an addition to an argument already complete. Without it, the cases of §4 would license a conclusion the framework should decline: that interpretations obstruct. Every party interprets, every interpretation may err, and a framework holding that erroneous interpretation obstructs would find obstruction wherever it looked. This is the difficulty the earlier paper met in the case of unmet demand and declined by the same route: the obstruction-relevant property is the rigidity with which a term is held in place of the content the term asserts.

5.2 The Content Condition

The first condition follows from Claim 4. A reconstruction, construal, or projection carries no obstruction in virtue of its content. A party may reconstruct a past unfavourably, construe present constraints pessimistically, and project a future in which its prospects are foreclosed, holding each of these provisionally and revising as material arrives. Such a party is coordinating, and the account finds no obstruction in its holdings however unfavourable they are.

The corresponding case is a party holding a favourable interpretation rigidly. A reconstruction attributing good will, held as accurate and closed to correction, degrades its own channel by the argument of §5.1, and the degradation proceeds whatever the reconstruction asserts. Favourable content supplies no protection.

The condition applies across the three horizons uniformly, and the uniformity is the paper’s second extension of the earlier result. Tense supplies no more protection than content: a rigid projection and a rigid reconstruction degrade their channels alike, and §4.3 exhibited a reconstruction from years back operating in the manner §4.2’s projection operated.

Claim 5. Obstruction attaches to the rigidity with which a term of $\hat{G}$ is held, leaving aside the term’s content and its temporal position. Unfavourable interpretations held provisionally obstruct nothing; favourable interpretations held rigidly obstruct in the manner Claim 4 describes.

5.3 The Conduct Condition

The second condition concerns which holdings enter the account at all. A term of $\hat{G}$ bears on obstruction through the conduct it conditions. A projection a party holds and acts on enters the relational dynamics by way of the acting; a projection held privately, with no conduct answering to it, remains outside them. The account concerns the estimated space as it conditions conduct, and an interpretation with no conduct answering to it falls outside its reach.

A party’s private assessments are its own. A projection of foreclosure that alters nothing in a party’s dealings with another produces no obstruction of that other, whatever distress it occasions the party holding it. The account describes relational dynamics and takes an interpretation into account where the dynamics register it.

The condition admits degrees and resists sharp application. Conduct is conditioned in ways ranging from the explicit to the barely marked, and the threshold below which a conditioned conduct fails to register is unspecified here. This is a vagueness of the same kind the earlier paper conceded in its treatment of relational significance, and it is conceded on the same terms: the vagueness attaches to the account’s application in place of its central claim.

5.4 The Position of Symbolic Mechanisms

The earlier paper treated symbolic mechanisms as those by which obstruction becomes self-reproducing, on the ground that fixing an interpretation regulates the interactions that would revise it. The present account places that treatment in a wider setting.

The terms of $\hat{G}$ are interpretations, and mechanisms fixing interpretations therefore operate on them directly. Narratives fix reconstructions of the past, a point developed in the literature on collective memory [8]; institutional categories fix construals of present constraints; and expectations transmitted through a community fix projections of what follows. Each is a route by which a term of $\hat{G}$ comes to be held as accurate, and Claim 4 then applies.

This suggests a reading of the earlier paper’s treatment. Symbolic closure was described there as a mechanism producing persistence in obstruction; it may be described here as a mechanism producing rigidity in the terms of an estimated space. The two descriptions pick out the same structure, the second locating it with respect to the object this paper takes as central.

Whether the second description supersedes the first, or whether the two are complementary treatments of one phenomenon at different levels, is left open. The question bears on how the series is organized in place of on what either paper claims, and §7 enters it.

§6 The Costs of the Extension

The extension carries four costs. Two sharpen difficulties the earlier paper already recorded; one concerns the framework’s standing as a dynamical account; and one concerns a claim the argument might be taken to establish and does not.

6.1 Access to the Interpretive Field

The earlier paper found that attributions of obstruction are licensed for parties within a relation and unlicensed for observers outside one. The ground was that reference models are justified by their role in coordination, and an observer coordinating with neither party has no such role to appeal to. The restriction was a licensing restriction: it concerned what an observer is entitled to assert.

The present extension adds an epistemic restriction alongside it, and the two are distinct.

Where obstruction is assessed over $\hat{G}$, an attribution of obstruction reports on a party’s estimated space. That space is composed from the party’s reconstruction, construal, and projection, and an observer requires access to these to report on it. Such access is harder to obtain than access to a party’s circumstances, and it is harder in a manner the framework’s own commitments explain. A party’s circumstances are, within limits, publicly available. Its reconstruction of its history is available through report, and report is conduct, which §5.3 holds to be conditioned by the very space the report concerns.

The consequence is that an observer’s access to $\hat{G}$ runs through material the framework holds to be conditioned by $\hat{G}$. This is a circularity in the evidential situation in place of the argument, and it admits no resolution the present resources supply.

Two mitigations are available, neither developed here. Conduct other than report bears on $\hat{G}$ and is observable, so that an estimated space may be inferred from what a party does in place of what it says. And divergence between $\hat{G}$ and $G$ becomes observable in its consequences, since conduct answering to a contracted space produces outcomes that conduct answering to the available space would not. Whether either supports the inferences an empirical programme would require is entered in §7.

The cost is accordingly this. The framework’s causal reach has been extended to cases the earlier paper could not treat, and its evidential position has been weakened in the same movement. A reader who valued the earlier account for its comparative testability should regard the present extension as a trade in place of an improvement.

6.2 The Standing of Constructed Cases

The earlier paper employed a constructed case and stated its standing: such cases exhibit an argument in a setting, forgoing evidential weight for control. The present paper employs three, and their standing is the same. The point requires restatement because the extension increases the temptation to treat them as more.

The cases of §4 are recognizable. A reader will supply instances from experience, and the supplying is what makes the cases legible. Recognition is not evidence. That a described mechanism is familiar establishes that the description is intelligible and leaves open whether the mechanism operates as described, how frequently, and under what conditions. The cases exhibit Claims 2 through 5 in settings where their operation is visible, and they establish none of them.

6.3 The Framework as a Dynamical Account

An argument of this shape can be taken to remove the framework from the class of dynamical theories, on the ground that its objects are interpretations in place of physical configurations. The reading is declined.

The substitution effected here is of the state space. Where the earlier paper’s assessments ranged over $G$, determined by a system’s history and constraints, the present assessments range over $\hat{G}$, composed as §3 describes. A system whose states are estimated spaces, evolving as §3.3 holds them to evolve, has the shape a dynamical account requires: states, an ordering on them, and conditioning of later states by earlier.

Neither this paper nor the earlier one supplies such an account, and §3.2 is explicit that its notation establishes nothing. The substitution specifies more precisely what a formalization would formalize, leaving the formal programme of the earlier paper intact with its state space replaced. The reading declined above would have removed that programme; the reading taken here leaves it standing and unexecuted.

6.4 The Humean Difficulty

The earlier paper declined to specify a trajectory a system ought to have followed, on the ground that such a specification requires a warrant the descriptive facts fail to supply [9]. The present extension might be taken to answer that difficulty. A projected trajectory raises no question of what ought to be, since a projection claims to represent what is coming in place of what is owed, and a framework operating over projections might appear to have escaped the problem.

It has escaped the problem on terms less favourable than they appear. The difficulty was never that trajectories are unavailable. It was that a trajectory carrying normative weight requires a warrant, and the descriptive resources supply none. Substituting $\hat{\gamma}$ for $\gamma^*$ dissolves the difficulty by removing the normative claim: a projection makes none, and no warrant is owed. The dissolution is complete and it costs the framework the ground on which any projection could be criticized. A party’s reconstruction may be self-serving, its projection may be wildly mistaken, and the account retains no resources with which to fault either. It can say that a term is held rigidly, which §5 establishes to bear on obstruction. It cannot say that a term is wrong in a sense carrying any weight.

The Humean difficulty is accordingly avoided in place of solved, and the avoidance is purchased at the price the earlier paper’s declining of normative questions already entailed. A framework wishing to say that some interpretations are better than others would require the normative layer both papers defer, and would face the difficulty there in its original form.

§7 Questions Left Open

The items below are ordered by their bearing on the account’s standing, beginning with those on which it is most exposed.

7.1 The Structure of the Composition

§3 states that $\hat{G}$ depends on three interpretively mediated horizons and supplies no account of the dependence. Supplying one is the principal formal task the paper leaves undone, and three questions organize it. Whether the horizons compose additively or interact. An additive composition would permit the contribution of each horizon to be assessed independently; an interactive one would make the contribution of a reconstruction depend on the projection accompanying it, which the cases of §4 suggest without establishing.

Whether one horizon dominates under specifiable conditions. §4.3 exhibits a reconstruction dominating a construed present, and the conditions under which such domination occurs are unexamined here.

Which structure represents an estimated space. The earlier paper posed the corresponding question for $G$ and noted that a continuation space admits contraction, expansion, and deformation. An estimated space admits these and admits also the divergence from $G$ that this paper treats, and a representation adequate to both is wanted.

7.2 Testing an Account of Models

Claims 4 and 5 are falsifiable in the manner the earlier paper’s central claim was falsifiable, and the difficulty of testing them is greater.

The prediction is that a term of $\hat{G}$ held as accurate degrades its correction channel, and that obstruction attends the degradation. Testing it requires observing the terms of parties’ estimated spaces and the rigidity with which they are held. §6.1 records that access to these runs through conduct the terms themselves condition.

Two routes were mentioned there and neither is developed. Inference from conduct other than report may support claims about a party’s estimated space, and the inference would require an account of which conduct bears on which terms. Divergence between $\hat{G}$ and $G$ is observable in its consequences, and an account of which consequences distinguish a contracted estimated space from a contracted available one would be required. Whether either supports an empirical programme is open.

7.3 The Organization of the Series

§5.4 observed that symbolic closure admits two descriptions: as a mechanism producing persistence in obstruction, which the earlier paper gave, and as a mechanism producing rigidity in the terms of an estimated space, which the present account permits. Whether these are one phenomenon described at two levels, or two phenomena related in a manner requiring statement, bears on how the series is organized.

A related question concerns the primitive. The earlier paper defined obstruction over $G$; the present paper assesses it over $\hat{G}$; and the relation between the two definitions requires more than the present treatment supplies. §2.4 held that the inventory is unaltered and the emphasis has moved. Whether a third paper should define the primitive over $\hat{G}$ directly, treating the earlier definition as the special case in which model and available space coincide, is open.

7.4 Relation to Predictive Accounts in Cognitive Science

The apparatus assembled here bears a structural resemblance to apparatus developed in the cognitive sciences under the headings of predictive coding, predictive processing, and active inference [13, 5, 4, 6, 12], and the resemblance was noted in the earlier paper’s treatment of reference models without being pursued. No claim of derivation is entered, and the accounts differ in the respects set out below.

The resemblance is structural: a system acting on a generative model of its situation, revising the model against material the model failed to anticipate, and conducting itself so as to bring its situation into accord with what the model represents. Whether the resemblance supports transfer in either direction is open, and two obstacles are visible. The cognitive accounts concern a system modelling an environment that returns no model; the present account concerns parties modelling each other. And the cognitive accounts specify an objective, prediction error, which the present account declines to supply for reasons §6.4 gives.

7.5 The Normative Layer

§6.4 found that the account retains no resources with which to fault an interpretation. A framework wishing to supply them would face the Humean difficulty in its original form, and the candidate the earlier paper identified remains the candidate here: unrevisability, as a property of a relation, is what survives after trajectories are deprived of normative standing. Why the openness of a relation to revision by its parties should carry normative weight is unanswered by either paper.

§8 Closing

The earlier paper in this series defined generative obstruction over the continuations available to a system in fact, and deferred a class of cases its resources could not reach: conflict arising from foreclosure a party anticipates in place of foreclosure it undergoes.

This paper has argued that the reference models the earlier paper required are composed of three interpretively mediated horizons, that assessment of a party’s own prospects is a function of all three, and that conduct answers to the space so composed. The extension reaches anticipated foreclosure and offers a reading of grievance, both through resources the earlier account already contained. Its composition is tense-neutral and its evolution is temporally ordered, which preserves the path-dependence the earlier paper derived. And the property bearing on obstruction remains the one the earlier paper identified: the rigidity with which a term is held, leaving aside the term’s content and its temporal position.

The costs have been stated at length. The framework’s evidential position is weaker than before, since assessment over an interpretive field requires access to that field, and the access runs through conduct the field conditions. The Humean difficulty is avoided in place of solved, at the price of any ground on which an interpretation might be faulted. And whether a framework organized around estimated spaces retains the minimality the series began with is a question the series has yet to settle.

Statement on AI Assistance

The author used Claude (Anthropic) for discussion, drafting assistance, and editorial revision. All theoretical claims, arguments, and conclusions are the author’s own, and the author takes full responsibility for the content.

References

[1] Huang, Wanhong. Generative Obstruction: Toward a Minimal Theory of Conflict in Generative Relational Being. Working paper, Serendip Commons Society, 2026.

[2] Bartlett, Frederic C. Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge University Press, 1932.

[3] Cederman, Lars-Erik, Kristian Skrede Gleditsch, and Halvard Buhaug. Inequality, Grievances, and Civil War. Cambridge: Cambridge University Press, 2013.

[4] Clark, Andy. “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science.” Behavioral and Brain Sciences 36, no. 3 (2013): 181–204.

[5] Friston, Karl. “The Free-Energy Principle: A Unified Brain Theory?” Nature Reviews Neuroscience 11, no. 2 (2010): 127–138.

[6] Friston, Karl, Thomas FitzGerald, Francesco Rigoli, Philipp Schwartenbeck, and Giovanni Pezzulo. “Active Inference: A Process Theory.” Neural Computation 29, no. 1 (2017): 1–49.

[7] Gurr, Ted Robert. Why Men Rebel. Princeton: Princeton University Press, 1970.

[8] Halbwachs, Maurice. On Collective Memory. Edited and translated by Lewis A. Coser. Chicago: University of Chicago Press, 1992.

[9] Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge and P. H. Nidditch. 2nd ed. Oxford: Clarendon Press, 1978.

[10] Loftus, Elizabeth F., and John C. Palmer. “Reconstruction of Automobile Destruction: An Example of the Interaction between Language and Memory.” Journal of Verbal Learning and Verbal Behavior 13, no. 5 (1974): 585–589.

[11] Merton, Robert K. “The Self-Fulfilling Prophecy.” The Antioch Review 8, no. 2 (1948): 193–210.

[12] Parr, Thomas, Giovanni Pezzulo, and Karl J. Friston. Active Inference: The Free Energy Principle in Mind, Brain, and Behavior. Cambridge, MA: MIT Press, 2022.

[13] Rao, Rajesh P. N., and Dana H. Ballard. “Predictive Coding in the Visual Cortex: A Functional Interpretation of Some Extra-Classical Receptive-Field Effects.” Nature Neuroscience 2, no. 1 (1999): 79–87.

[14] Schacter, Daniel L., Donna Rose Addis, Demis Hassabis, Victoria C. Martin, R. Nathan Spreng, and Karl K. Szpunar. “The Future of Memory: Remembering, Imagining, and the Brain.” Neuron 76, no. 4 (2012): 677–694.

中文

被诠释的时间场

预期与生成性评估的构成

黄万宏 · huangwanhong@serendip.ngo

摘要

前一篇论文提出生成性阻碍为一个原初解释对象，在一个系统事实上可用的诸延续之上定义它，并把对阻碍的诸归属限制在关系之内的各方。它推延了一类它的资源无从考察的情形：起于一方所预期之封闭、而非它所承受之封闭的冲突。本文接手那一情形。它论证前一篇论文所确立的诸参照模型是时间上复合的，由一个被诠释的过去、一个当下状态与一个被预期的未来装配而成，每一个都经由诠释而进入；并且一方对其自身生成性前景的评估，是这三者的一个函数。这一扩展够到预期性冲突，并提供一种关于怨愤的读法，而不引入一个进一步的原初对象，因为两者都通过前一套框架业已要求的诸模型来运作。有两条界限被陈述。被估计之空间的构成是无时态偏向的，而其演化仍是时间上有序的，这保存了前一套说明的路径依赖，并封闭了有时从这类论证中引出的一个更强的读法。而这一扩展锐化了一个业已在场的证据困难：当阻碍在一个诠释场之上被评估时，它对缺乏进入那个场之通路的观察变得不可及，因而框架在因果触及上增长，同时在可检验性上收缩。

关键词： 生成性阻碍；被估计的生成性空间；预期；被诠释的过去；怨愤；生成性关系存在。

关于本文定位的一则说明。 这是一个系列中的第二篇论文，且它通篇依赖于第一篇。它的核心动作是把一方对其可用延续的模型，替换那些延续本身，作为阻碍在其之上被评估的对象。这一替换被辩护为对第一篇论文业已要求之诸资源的一个扩展，而非一个新的承诺，而拒绝第一篇论文关于参照模型之说明的读者，将在此处发现少有能脱离它而立的东西。若干主张被作为悬而未决的问题引入，在无从落实它们的资源不可用之处。反对、更正与反证，欢迎寄至 huangwanhong@serendip.ngo。

§1 引言

本系列前一篇论文 [1] 提出生成性阻碍为一个原初解释对象，并借一个受约束的生成性系统可用延续上的偏序来定义它：一个互动阻碍一个系统，当在该互动之下对它可用的诸延续，真包含于它以别样方式可用的诸延续之中时。冲突被当作此种阻碍的显现，而阻碍凭之被产生的诸机制被不加特权地排列开来。

那篇论文还在处理其自身定义中一份困难的过程中，确立了对阻碍的诸归属要求参照模型。定义把一个互动之下可用的东西同以别样方式可用的东西加以比较，而这一比较要求对那个别样方式的界定。作为关于一个由关系构成之系统的事实，没有任何这样的界定可用，因为它所要求的反事实移除了该系统在其内被构成的诸耦合。前一篇论文提出以一个工作模型代替一个事实，由一方为协调之目的而持有，不携带任何本体论地位，且不对超出其角色的任何东西负责。它进一步论证一个被当作准确的模型败坏本可用来矫正它的通道，而一个模型被持有的刚性因而与阻碍相关，而该模型的内容则否。

有一类情形被推延。一方可能依据一个它所预期的封闭、而非一个已经发生的封闭行动。那位同事尚未被晋升，那份资源尚未被分配，那个职位仍然空缺；而该方举止仿佛它的诸延续已然封闭。这种冲突是常见的，而前一套框架把它指派到其范围之外，因为那里的判据关乎已然发生的诸封闭。这一指派是基于一种前一篇论文拒绝去考察的读法作出的，而它就此留下一则说明。

本文考察这一读法。它的论题可被紧凑地陈述。定义 1 所要求的比较有两个项，而前一篇论文把两者都读作关乎事实上可用的东西。读作关乎一方的工作模型表示为可用的东西，定义便直接够到被预期的封闭，因为一个关于什么可用的模型，正是对未来的一个投射所供出的东西。以 $\hat{G}$ 记一方的模型表示为可用的延续之空间，把它同事实上可用的延续之空间 $G$ 区分开来。所提议的是，阻碍，如各方所归属它、以及如它条件化它们的行为那样，在 $\hat{G}$ 之上被评估。

论题携带三个后果，而本文围绕它们组织。第一个关乎 $\hat{G}$ 包含什么。一个关于可用延续的模型，由一方当作已经发生之物、它当作如今成立之物，以及它当作正在到来之物装配而成。这些中的每一个都经由诠释而进入。过去作为被重构者、而非作为它发生的样子而进入，而同一个事件容许支持互不相容之评估的诸重构。未来作为被投射者而进入，而一个投射条件化行为，无论被投射的事件是否发生。当下作为被构想者而进入，因为一方对其当前诸约束的读解本身是一个读解。因而评估是三个时间上有别的自变量的一个函数，每一个都经由诠释而中介，而那个函数的构成是§3的主题。

第二个关乎这留下什么立着。一个这种形状的论证可被推向一个更强的结论，即当下在生成性评估中不享有任何享有特权的地位。这个更强的结论在此被抵制，而这一抵制对前一篇论文所确立的东西要紧。评估发生在一个时刻，在一个于那个时刻被构成的场之上，并在那个时刻发为行为。该论证溶解当下作为内容的享有特权的地位：一方的当前诸境况没有特别的凭据成为那个要紧的自变量，而被诠释的诸过去与被投射的诸未来同它们并列进入。当下作为所在而存留：评价发生在此刻，而该场从一次这样的评价演化到下一次。这一区分保存了前一篇论文所推导的路径依赖，而放弃它会移除那个结果。

第三个关乎这一扩展耗费什么。前一篇论文发现参照模型对关系之内的各方获许可，而对关系之外的观察者未获许可，这把对阻碍的良构归属限制到各方自身。当下的扩展锐化那一限制。当阻碍在一个诠释场之上被评估时，一个观察者需要进入那个场的通路，而进入一方关于其自身前景的模型，比进入该方的诸境况更难获得。因而框架在因果触及上有所增益，而在可检验性上有所丧失，而§6陈述这一权衡而不加软化。

有一则澄清属于开篇，因为在这一临近处的一个主张容易被夸大。前一篇论文拒绝去界定一个系统本应遵循的一条轨迹，理由是这一界定会要求一份描述性诸事实无从供出的担保。到 $\hat{G}$ 的动作或许看似回答那一困难，因为一条被投射的轨迹不引起任何关于本应为何的问题。它确实回答它，而它回答它的方式，是放弃任何轨迹能凭之被批评的那块地基：一个模型可能任意地弄错，而框架不保留可用来这样说的任何资源。该困难被规避而非被解决，而§6明确地拒绝那个更强的主张。

其余部分如下推进。§2把 $\hat{G}$ 同 $G$ 区分开来，并为它的因果效力辩护。§3铺陈 $\hat{G}$ 的时间构成与评估所采取的函数形式，并确立什么是无时态偏向的、什么不是。§4处理三个实例：一个被重新诠释的过去、一个被预期的未来，以及怨愤。§5把前一篇论文的结果扩展到时间场，并供出没有它这一扩展便会够得太远的诸条件。§6陈述这一扩展耗费什么，而§7陈述它留作悬而未决的东西。

§2 被估计的生成性空间

本节引入本文其余部分在其之上运作的对象，把它同前一篇论文所采用的对象区分开来，并论证它以一种前一个对象所不具的方式关乎行为。

2.1 把被估计之空间同可用之空间区分开来

前一篇论文以 $G$ 记一个系统可用的延续之空间：它从其所立之处、鉴于其历史与其约束能够继续生成什么。阻碍由这个空间上的真包含来定义，$G’ \prec G$，而定义关乎事实上可用的诸延续。

以 $\hat{G}$ 记一方的工作模型表示为可用的延续之空间。那个尖号携带前一篇论文赋予它的意涵：这是一个为协调而持有的临时构造物，受制于修正，不携带任何本体论地位。

这两个空间是松散地相关的，且在两个方向上分离。一方可能把事实上被封闭的诸延续表示为可用。一个在其诸前提已失败时仍把一条工作路线当作敞开的研究者，在那一方面持有一个大于 $G$ 的 $\hat{G}$。一方同样可能把仍然可用的诸延续表示为被封闭。一个在一个职位仍空缺时把它当作已被填补的候选人，持有一个小于 $G$ 的 $\hat{G}$。

这一分歧是常态而非一个缺陷。§3论证 $\hat{G}$ 由被诠释的诸材料装配而成，而对一个系统自身前景的诠释，在诸保证不完美的条件下进行：该方的信息是局部的，它的诸约束对它是不完美地被知的，而它的未来被它所能观察到的东西所欠决定。一个与 $G$ 重合的 $\hat{G}$ 会是一项要求解释的成就。

前一篇论文对参照模型的处理在此无需修正地适用。一个偏离它所建模者的模型并不因此有缺陷，因为它从未作为一个对对应负责的表示而被持有。它是为协调而被持有，且对它是否服务于协调负责。前一篇论文添加了一个关于这类模型如何被持有的条件，而§5把那个条件扩展到当下的情形。

2.2 行为对之负责的那个空间

本文所转于其上的主张现在可被陈述。

主张 1。 一方的行为被 $\hat{G}$ 所条件化。当 $\hat{G}$ 与 $G$ 分歧时，行为对前者负责。

支持它的论证是简短的。一方依据其前景行动，是依据它当作前景来看待的前景行动。事实上对它可用的诸延续，关乎它的诸行动成就什么；它们只通过它对它们的表示，才关乎它选择哪些行动。一方在未觉察一个延续被封闭时追求它；一方在被说服一个可用的延续被封闭时放弃它。这是在一个表示之下行动的一个一般特征，且无需任何特殊装置。

两条限定使该主张保持在界限之内。它关乎行为的选择而非结果。对着一个弄错的模型被选择的行为，遭遇如其所是的世界，而所得由 $G$ 以通常的方式决定。一方追求一个被封闭的延续，达不到它。该主张断言那一追求的发生，并把那一发生归于 $\hat{G}$。

该主张把两个空间之间的关系留作悬而未决。这里没有任何东西主张 $\hat{G}$ 不受 $G$ 约束，而通常有大量东西约束它：一方遭遇它的诸约束、接收矫正并修正。该主张仅就分歧的情形断言，行动对模型负责。

2.3 表示的因果效力

主张 1 有一个携带本文主要扩展的后果。倘若行为对 $\hat{G}$ 负责，且倘若行为进入系统生成于其下的诸条件，那么一个表示便关乎生成性动态，无论它是否准确。

这一结构被默顿描述为自我实现的预言 [11]，而有一个实例值得陈述，因为它的机制是透明的。当一个市场中的参与者当作一场下跌正在到来时，它们借抛售依据那个预期行动，而抛售产生那场下跌。事先无需有任何关于该市场前景的事实成立。那个预期在它被形成的那一刻是弄错的，而到它被评估之时是准确的，而弥合那一裂隙的，是那个预期自身所条件化的行为。

同一结构在本文所处理的诸情形中运作。一方当作一位同事的晋升封闭其自身前景。那一晋升尚未发生，而那一封闭是一个投射。该方相应地举止：它保留合作，它在本会协作之处竞争，它把该同事当作一个障碍。该同事，遇到这一行为，以同样方式回应。两者的前景如今在事实上被削减，而这一削减是由一个关于一场尚未发生之削减的表示所产生的。

§5的论证依赖于这一结构的三个特征。表示在被表示的事件之前做它的工作。行为在表示被持有的那一刻被条件化，而被表示的事件可能在此后发生、或以别样方式发生、或从不发生。这正是把被预期的封闭置于框架触及范围之内的东西，且是前一篇论文所推延之物的实质。

表示可能是自我确证的。被一个表示所条件化的行为，可能产生该表示所断言的那个状态，而当它产生时，该方接收到它自己的行为所安排的确证。这正是前一篇论文在其对象征性机制的处理中所描述的闭合结构，此处在一个关于自身前景之模型的特定情形中出现。

该机制在两侧都不带误差地运作。上面情形中的任何一方，除了那个最初的投射之外，都无需对任何东西弄错，而那个最初的投射无需多于对一个不确定情境的一个合乎情理的读解。这与它在前一篇论文中要紧出于相同的缘由：一个要求恶意或严重错误的机制，会解释一类比诸现象所在场者更狭窄的情形。

2.4 $\hat{G}$ 在本系列中的地位

一套提议一个极简原初基底的框架，每当一个新对象显著地出现时，便欠一个关于其清单的说明，而此处该给这个说明。

前一篇论文关于参照模型确立了四样东西：对阻碍的诸归属要求它们、各方持有它们、它们是临时的，以及它们充分地条件化行为，以致它们的刚性关乎阻碍。当下这一节命名这样一个模型所表示的空间，并断言行为对它负责。两者都随前面的诸论证而来，且对它们什么也没添加、无一是前一篇论文所缺的。

这一改变在于显著度。前一篇论文引入参照模型来处理其定义中的一份困难，并此后把它们当作装置来对待，从属于其原初对象在其上被定义的空间 $G$。本文把它们当作评估在其之上被进行的对象，从而 $\hat{G}$ 在当下的处理中占据 $G$ 在前一篇处理中所占据的位置。这一转移是可观的，而它是一个侧重的转移，使清单保持不变。

一个这种量级的侧重是否保存前一篇论文所声称的极简性，是一个公道的问题。§6接手它，连同这一扩展所携带的其他诸代价。

§3 被估计之空间的时间构成

§2确立了行为对 $\hat{G}$ 负责，而把那个空间的构成留作未考察。本节供出该构成，陈述评估所采取的函数形式，并固定该说明在其中是时间上有序的两个方面，与它不是的那一个方面。

3.1 三个视域

一方对它可用之诸延续的模型，汲取三个来源，由它们的时间位置来区分。

第一个是过去，如该方重构它。一段封闭、成就与约束的历史，作为关乎什么仍然可用的证据进入该模型，而它作为被重构者、而非作为它发生的样子而进入。回忆是重构性的、而非再现性的，这在关于记忆的实验文献中是一个由来已久的发现 [2, 10]，而当下的主张是那个较弱的主张，即一方对其自身历史的读解作为被读解者进入 $\hat{G}$。它也随前一篇论文关于生成的说明而来：一个系统所生产之物进入它后续生产的诸条件，而一方对其自身历史的读解正是一个这样的生产，随后来的诸生产条件化它而被修正。

第二个是当下，如该方构想它。当前诸约束直接关乎可用延续，而它们通过一个构想而关乎它们：一方对当前约束它之物的读解是一个读解，受制于它对任何别的东西之读解所受的同样局部性。当下作为被诠释的材料同另外两个并列进入该模型，其即时性无论如何。

第三个是未来，如该方投射它。对将临诸状态的诸投射，以不亚于另外两者的力关乎该模型，而§2论证它们在被投射的诸状态成立或不成立之前条件化行为。一个投射自始至终是诠释性的，由该方对其情境的读解及其关于何者随之而来的诸预期装配而成。

每一个视域都经由诠释而进入，而这一齐一性是本节的第一个主张。记忆文献中的工作报告了支持对过去之回忆的诸过程，与支持对未来之模拟的诸过程之间的可观重叠 [14]。那一发现关乎认知架构，而非此处所引入的构成性主张，且它作为与该齐一性相一致者、而非作为支持它的证据而被记下。

主张 2。 被估计的生成性空间 $\hat{G}$ 由三个经诠释中介的来源构成：一个被重构的过去、一个被构想的当下与一个被投射的未来。没有来源不经中介地进入该构成。

该主张的内容在于中介的齐一性。一套承认诸投射之诠释性特征、同时把过去当作固定材料的说明，会按诸视域的认识论地位给它们分类，理由是过去已经发生而未来尚未。主张 2 拒绝这一分类。发生固定一个事件；它把该事件对一个可用延续之模型贡献什么留作悬而未决，而那一贡献正是进入 $\hat{G}$ 的东西。

3.2 函数形式

主张 2 许可对一方所进行之评估的一个紧凑陈述。以 $\Pi$ 记被重构的过去，以 $\Sigma$ 记被构想的当下，以 $\Phi$ 记被投射的未来，以 $\hat{G}$ 记由它们装配而成的被估计之空间：

$$\hat{G} ;=; F(\Pi, \Sigma, \Phi).$$

该表达式是一个纲要、而非一个模型，而这一区分应在它被使用之前引入。它固定一次评估的诸自变量是什么，以及它们以何种方式相对于该评估而立；它不界定它们之间的任何关系。这里没有任何东西供出 $F$，而诸视域是否可加地构成、它们是否交互，以及一个是否在可界定的诸条件下占主导，都是被引入§7的悬而未决的问题。本文没有任何主张从该表达式推导出来。下面的每一个论证都以散文进行，且会经受住那记号的移除，那记号压缩对一个依赖的陈述、而非确立一个依赖。

诸自变量是该方的，而评估是该方的。$\Pi$、$\Sigma$ 与 $\Phi$ 是该方的重构、构想与投射，而 $\hat{G}$ 是它的模型所表示的空间。前一篇论文对诸归属限制于关系之内各方的做法，适用于整个表达式，而§6陈述这留下一个观察者能说什么。

该形式关乎评估、而非能力。事实上可用的延续之空间 $G$，由该系统的历史与约束以前一篇论文所描述的方式决定，而 $F$ 对它什么也没说。当两者分歧时，主张 1 主张行为对 $\hat{G}$ 负责，而 $G$ 决定那一行为成就什么。

3.3 构成与演化

一个上面所给形状的论证，可被推向一个本节所抵制的结论，而这一抵制要求陈述，因为该结论是有吸引力的，且它的代价容易被漏看。

那个有吸引力的结论是，生成性评估在时间上是无结构的：既然所有三个视域都以同等地位进入同一个评价，时间位置便对框架无关紧要，而当下不享有任何享有特权的地位。这一点的前一半在下面被承认。后一半会移除一个前一篇论文所推导的结果，而它被拒绝。

本节所提议的区分是在构成与演化之间。

构成是无时态偏向的。 三个视域以同等地位进入 $F$，而它们的时间位置不给它们的诸贡献排序。一个从二十年前重构的封闭，可能比一个如今成立的约束更沉重地关乎 $\hat{G}$，而一个关于一场可能从不发生之封闭的投射，可能比两者中任何一个都更沉重地关乎。诸视域由它们的来源区分，而不由它们排序。那个有吸引力之结论的这一部分被承认，而§4展现它。

演化是时间上有序的。 评估 $F(\Pi, \Sigma, \Phi)$ 发生在一个时刻。它的诸自变量是该方在那个时刻所立的重构、构想与投射；它的输出在那个时刻条件化行为；而那一行为进入该方在随后诸时刻重构、构想与投射于其下的诸条件。以 $\hat{G}_t$ 记在 $t$ 处的被估计之空间：序列 ${\hat{G}_t}$ 是一条轨迹，有序的，而每一项都被它的诸前身所条件化。

主张 3。 $\hat{G}$ 的构成是无时态偏向的：三个视域进入评估而不按它们的时间位置排序。$\hat{G}$ 的演化是时间上有序的：评估发生在一个时刻，在那个时刻条件化行为，并通过那一行为条件化后续的评估。

这一对携带一个前一篇论文的核心推导所要求的后果。

那篇论文从生成能力的历史条件化推导出阻碍的路径依赖：一个系统接下来能生产什么，取决于它已经生产了什么，从而一个互动是否阻碍，取决于它所落入的历史。该推导要求一个有序的状态序列。一套在其中时间位置通篇无关紧要的说明会供不出任何这样的序列，而路径依赖会丧失。

主张 3 借把无时态偏向定位在它所属之处，保存了该推导。一次评估的构成汲取三个视域而不给它们排序；诸评估的序列是有序的，而前一篇论文的论证无变更地适用于那个序列。被估计之空间在一个诠释场、而非一个物理场之上路径依赖地演化，而该论证的结构经受住这一替换。

3.4 当下作为所在

有一个进一步的区分随主张 3 而来，且值得明确地划出，因为享有特权的语汇在上面已在两个意义上被使用。

当下丧失它作为内容的特权。一方的当前诸境况没有特别的凭据去决定它对可用延续的评估，而它们在主张 2 所使之齐一的诸条件上，同诸重构与诸投射竞争。这是一个实质的结果。把一方前景定位在其当前情境中的诸说明，在这一读法上，是在留意 $F$ 的一个自变量而把它当作全部。

当下保留它作为所在的特权。评估发生在此刻，在一个此刻被构成的场之上，发为此刻的行为。被重构的过去是一个当下的重构；被投射的未来是一个当下的投射；而进行该评估的一方，在一个它所汲取之诸视域所不具的意义上占据当下。移除这一点会使框架无法把一次评估定位在时间中，连同主张 3 所要求的那个有序性。

这两个意义是相容的，而它们之间的区分正是容许本节论证进行的东西。一套框架可以主张当下没有特别的凭据去决定一方当作可用的东西，同时主张那一当作发生在当下。第一个是一个关于 $F$ 之诸自变量的主张；第二个是一个关于 $F$ 在何处被评价的主张。

§4 时间构成的三个实例

§3的诸主张在此于三个案例上被展现，选取它们以孤立每一个视域的贡献。第一个让被重构的过去变化而把另外两个固定不变；第二个同样地让被投射的未来变化；第三个展现一个案例，其中一个久远的重构主导一次评估。这些案例是构造的，而构造案例在本系列中的定位一如前一篇论文所描述：它们在一个特定情境中展现一个论证，为它们所供出的控制而放弃证据分量。

4.1 一个被重构的过去

两方处在一个紧密的耦合之中。一方未能标记另一方当作有意义之物的一个场合，而这一未能落在若干月前。该事件已定：它发生了，它以它发生的样子发生，而两方都不争议该描述。

对第二方可用的诸重构有别，而有两个值得并置。

在第一个重构上，那一未能指示一种性情：未能的那一方微弱地留意另一方的诸关切，通篇如此，且将继续。该重构直接关乎 $\hat{G}$。要求那一方持续留意的诸延续被表示为不可用，因为该重构所归属的那种性情封闭它们。被估计之空间收缩。

在第二个上，那一未能指示一种境况：未能的那一方处于负荷之下，糟糕地应对了那个场合，并在该境况所容许的范围内尽力留意了。这一重构使同样的诸延续保持被表示为可用，因为一种境况会过去，而一种性情会持续。

该事件在两者中是相同的。它对 $\hat{G}$ 的贡献有别，而它有别是因为重构有别。这是主张 2 在第一个视域上被展现：过去作为被重构者进入被估计之空间，而重构正是该空间由之构成的东西。

诸重构不由准确性来区分，而该案例被构造得使任何一方都不处在敲定哪一个正确的位置。一种性情与一种境况在单一场合上与同样的行为相容，而会把它们分开的证据在于尚未发生的行为。进行重构的一方在欠决定之下选择，而该选择在会告知它的证据到来之前条件化行为。

如此被条件化的行为关乎那证据。一方把诸延续表示为被封闭，便相应地举止：它撤回它当作未被回报的留意，而这一撤回改变另一方所做的事。该重构因而参与产生本会用来评估它的证据，而这正是§2所描述、§5所接手的结构。

4.2 一个被投射的未来

一方在一个组织之内工作，其中一个职位将被填补。一位同事是一个候选人。该职位仍然空缺，该决定在前方，而没有分配发生。

该方投射。在该投射上，该同事被任命，而这一任命把该方自身的晋升封闭一段它估计为可观的时期。该投射进入 $\hat{G}$，而被估计之空间收缩：要求成问题之晋升的诸延续被表示为不可用。

行为对被收缩的空间负责。该方保留它本会施予的合作，在本会协作之处竞争，并通过它所持的投射构想该同事的行为。请注意，事实上对该方可用的延续没有被封闭。空间 $G$ 一如它所立；该方依据 $\hat{G}$ 行动。

该同事，遇到这一行为，回应它。被撤回的合作是不可用的合作，而该同事相应地调整它自己的模型，这又条件化它的行为。两方的前景都被这一交换在事实上削减，而这一削减是由一个关于一场尚未发生之削减的投射所产生的。

这是前一篇论文所推延的情形，而当下的诸资源够到它。前一套框架的范围判据问满足一个需求是否会封闭诸延续，以及受挫是否可描述地败坏能力。被预期的封闭两者都不满足，因为没有封闭发生。在 $\hat{G}$ 之上读来，该案例是直截的：该方的被估计之空间已收缩，它的行为对那一收缩负责，而它所归属的阻碍是它的模型所表示的一个。

该案例所确立的少于它看似确立的。该投射可能是准确的。该同事可能被任命，而该晋升可能被封闭，在那种情况下该方的行为是在对着一个变得与可用空间重合的空间负责。此处所描述的机制在两种情况下都运作，而它的运作不取决于该投射弄错。该案例展现在被投射的事件解决之前对行为的条件化，无论它以哪种方式解决。

4.3 一个久远的重构

第三个案例展现§3.3所声称的构成之无时态偏向，而它处理一个前一篇论文列在它所略去之诸机制中的现象，一个有它自己可观文献的现象 [7, 3]。

一方在若干年前承受了一场封闭：一个机会被另一方的行动对它关闭，而那一关闭是真实的。产生它的诸条件此后已过去。另一方的位置已改变，容许那一关闭的安排已被消解，而当前成立的东西无一会产生一场相当的封闭。

该重构仍在。在它上，另一方是一个进行封闭的一方，该安排是一个容许封闭的安排，而要求另一方合作的诸延续被表示为不可用。被估计之空间被一个关于不再成立之诸条件的重构所收缩，而行为对那一收缩负责。

该案例转在三个特征上。该重构压过被构想的当下。该方的当前诸境况会支持一个评估，在其上成问题的诸延续是可用的，而该重构主导它们。这正是构成的无时态偏向：诸视域进入 $F$ 而不按时间位置排序，而一个多年前的重构可能比一个如今成立的约束更沉重地关乎。

该重构就所发生之事而言是准确的。那一封闭发生了，而该方对它的说明是正确的。该案例展现一个关于过去准确、关于可用延续不准确的重构，而两者分离，因为产生该封闭的诸条件已过去，而该重构未被修正。

怨愤，如此描述，无需一个新机制。前一篇论文把怨愤列在它所略去的诸机制中，且没有提供任何论证说它化约为它所处理的那些。当下的说明够到它，作为时间构成的一个情形：一个未被修正的重构收缩一个被估计之空间，而行为对那一收缩负责。这提供一种关于前一套框架所缺之机制的读法，通过§2论证业已在场的诸资源。这一读法是否捕捉了怨愤文献所描述的东西，是一个由那一文献去敲定的问题。

4.4 三个案例所共有的结构

这些案例在哪一个视域承载评估上有别，而在三个方面一致。在每一个中，被估计之空间同可用的空间分歧，而行为对被估计之空间负责。在每一个中，这一分歧经由对该方所拥有之材料的诠释而生起，而非经由对它所缺之材料的无知。且在每一个中，如此被条件化的行为关乎本会用来评估该诠释的证据，从而该诠释参与产生它自己的确证。

第三个方面正是本文对前一结果之扩展所运作之处，而§5接手它。缺那一处理，这些案例便只会确立诠释要紧，而这是共同的地基，且会以无物扩展框架。

§5 时间场之上的刚性

§4的诸案例展现诠释进入评估并条件化行为。单独取来，它们确立诠释要紧，而框架无需任何装置去说这一点。本节供出那装置，把前一篇论文的结果扩展到被估计之空间，并固定没有它这一扩展便会够得比其论证所支撑者更远的诸条件。

5.1 这一扩展及其必要性

前一篇论文论证一个持有于另一方的工作模型，败坏本可用来矫正它的通道，当该模型被当作准确、而非临时地持有时。该论证分三步进行。一个指示该模型出错的矫正，在一个被当作准确的模型上是反常的，这使一个把该反常归于被矫正的一方、而非归于该模型的构想成为可用。如此被构想的诸矫正，所成就的少于该矫正所意图的，而它们使进行矫正的一方付出的多于它们所回报的，因而它们的速率下降。此后该模型被对着一个更薄的矫正通道持有，而容许该构想的诸条件被加强。

被估计之空间受制于同一论证，而这一扩展要求一个观察。构成 $\hat{G}$ 的诸项是诠释：一个关于过去的重构、一个关于当下的构想、一个关于未来的投射。每一个在原则上都是可矫正的，而每一个都以同一方式、由该诠释未能预料到的材料被矫正。一个重构被它所归属之性情本会未能产生的行为矫正；一个投射被以别样方式解决的被投射事件矫正；一个构想被被证明异于所构想的诸约束矫正。

当一个项被当作准确地持有时，该论证无变更地适用于它，而那个项的矫正通道败坏。

主张 4。 构成 $\hat{G}$ 的诸项受制于前一篇论文所描述的败坏。当一个重构、构想或投射被当作准确、而非临时地持有时，本可用来矫正它的通道败坏，而该项趋向不可修正。

主张 4 或许会被读作对一个业已完整之论证的一个附加。没有它，§4的诸案例会许可一个框架应当拒绝的结论：即诸诠释阻碍。每一方都诠释，每一个诠释都可能出错，而一套主张错误的诠释阻碍的框架，会在它所看之处处处发现阻碍。这正是前一篇论文在未获满足之需求的情形中所遇、并借同一路径所拒绝的困难：与阻碍相关的性质，是一个项被持有的刚性，而非该项所断言的内容。

5.2 内容条件

第一个条件随主张 4 而来。一个重构、构想或投射，不凭其内容携带任何阻碍。一方可能不利地重构一个过去、悲观地构想当前诸约束，并投射一个其前景被封闭的未来，而临时地持有这些中的每一个，并随材料到来而修正。这样一方是在协调，而该说明在它的诸持有中发现不出阻碍，无论它们多么不利。

对应的情形是一方刚性地持有一个有利的诠释。一个归属善意的重构，被当作准确并对矫正封闭，凭§5.1的论证败坏它自己的通道，而这一败坏进行下去，无论该重构断言什么。有利的内容供不出任何保护。

该条件齐一地跨三个视域适用，而这一齐一性是本文对前一结果的第二个扩展。时态供不出比内容更多的保护：一个刚性的投射与一个刚性的重构一样地败坏它们的通道，而§4.3展现了一个多年前的重构以§4.2的投射所运作的方式运作。

主张 5。 阻碍附着于一个 $\hat{G}$ 之项被持有的刚性，把该项的内容及其时间位置搁在一边。临时地持有的不利诠释不阻碍任何东西；刚性地持有的有利诠释以主张 4 所描述的方式阻碍。

5.3 行为条件

第二个条件关乎哪些持有根本进入该说明。一个 $\hat{G}$ 之项通过它所条件化的行为关乎阻碍。一个一方所持有并据之行动的投射，经由那行动进入关系动态；一个私下持有、没有行为对之负责的投射，留在它们之外。该说明关乎被估计之空间在它条件化行为时的样子，而一个没有行为对之负责的诠释落在它的触及范围之外。

一方的私下评估是它自己的。一个不改变一方与另一方交往中任何东西的封闭投射，产生不出对那另一方的任何阻碍，无论它给持有它的一方招致何种苦恼。该说明描述关系动态，并在该动态登记一个诠释之处把它纳入考虑。

该条件容许程度，且抗拒锐利的适用。行为被以从明确到勉强可辨的诸方式条件化，而一个被条件化的行为低于其便未能登记的那个阈值，在此未被界定。这是一种与前一篇论文在其对关系意义的处理中所承认者同类的含糊，而它按同样的条款被承认：这一含糊附着于该说明的适用、而非它的核心主张。

5.4 象征性机制的位置

前一篇论文把象征性机制当作那些凭之阻碍变得自我再生产的机制，理由是固定一个诠释规约本会修正它的诸互动。当下的说明把那一处理置于一个更宽的情境中。

$\hat{G}$ 的诸项是诠释，而固定诠释的诸机制因而直接作用于它们。诸叙事固定对过去的诸重构，一个在关于集体记忆的文献中被发展的要点 [8]；诸制度范畴固定对当前诸约束的诸构想；而经由一个共同体传递的诸预期固定对何者随之而来的诸投射。每一个都是一个 $\hat{G}$ 之项变得被当作准确地持有所经由的路径，而主张 4 随后适用。

这提示一种关于前一篇论文之处理的读法。象征性闭合在那里被描述为一个在阻碍中产生持续的机制；它在此处可被描述为一个在一个被估计之空间的诸项中产生刚性的机制。这两个描述挑出同一个结构，第二个把它相对于本文当作核心的对象来定位。

第二个描述是否取代第一个，或这两者是否是对一个现象在不同层次的相互补充的处理，被留作悬而未决。该问题关乎本系列如何被组织、而非关乎任何一篇论文所主张的东西，而§7引入它。

§6 这一扩展的诸代价

这一扩展携带四个代价。两个锐化前一篇论文业已记下的诸困难；一个关乎框架作为一个动态说明的地位；而一个关乎一个该论证可能被当作确立、而它并不确立的主张。

6.1 进入诠释场的通路

前一篇论文发现对阻碍的诸归属对关系之内的各方获许可，而对关系之外的观察者未获许可。理由是参照模型由它们在协调中的角色来辩护，而一个不与任何一方协调的观察者，没有这样一个角色可诉诸。该限制是一个许可限制：它关乎一个观察者被有资格断言什么。

当下的扩展在它旁边添加一个认识论限制，而这两者是有别的。

当阻碍在 $\hat{G}$ 之上被评估时，一个阻碍的归属报告一方的被估计之空间。那个空间由该方的重构、构想与投射构成，而一个观察者需要进入这些的通路才能报告它。这样的通路比进入一方之诸境况的通路更难获得，而它更难，以一种框架自身诸承诺所解释的方式。一方的诸境况，在限度之内，是公开可用的。它对其历史的重构经由报告而可用，而报告是行为，§5.3主张它被那个报告所关乎的空间本身所条件化。

其后果是，一个观察者进入 $\hat{G}$ 的通路，经由框架主张被 $\hat{G}$ 所条件化的材料。这是证据处境、而非论证中的一个循环，而它不容许当下诸资源所供出的任何解决。

有两种缓解可用，而这里都不予发展。报告之外的行为关乎 $\hat{G}$ 且是可观察的，从而一个被估计之空间可从一方所做之事、而非它所说之事被推断。而 $\hat{G}$ 与 $G$ 之间的分歧在其诸后果中变得可观察，因为对着一个被收缩之空间负责的行为，产生对着可用空间负责的行为所不会产生的诸结果。这两者中是否有一个支撑一个经验纲领会要求的诸推断，被引入§7。

因而代价是这样。框架的因果触及已被扩展到前一篇论文无从处理的诸情形，而它的证据处境在同一动作中被削弱。一个因前一套说明的相对可检验性而看重它的读者，应把当下的扩展视作一个权衡、而非一个改进。

6.2 构造案例的定位

前一篇论文采用了一个构造案例并陈述了它的定位：这类案例在一个情境中展现一个论证，为控制而放弃证据分量。当下这篇论文采用三个，而它们的定位是一样的。这一点要求重述，因为这一扩展增大了把它们当作更多的诱惑。

§4的诸案例是可辨认的。一个读者会从经验中供出诸实例，而这一供出正是使这些案例可读的东西。辨认不是证据。一个被描述的机制是熟悉的，这确立该描述是可理解的，而把该机制是否如所描述地运作、多么频繁地、以及在什么条件下留作悬而未决。这些案例在其运作可见的诸情境中展现主张 2 至 5，而它们对它们无一确立。

6.3 框架作为一个动态说明

一个这种形状的论证可被当作把框架从动态理论那一类中移除，理由是它的诸对象是诠释、而非物理配置。这一读法被拒绝。

此处所实施的替换是状态空间的替换。前一篇论文的诸评估遍历 $G$，由一个系统的历史与约束决定，而当下的诸评估遍历 $\hat{G}$，如§3所描述地构成。一个其诸状态是被估计之空间、如§3.3主张它们演化那样演化的系统，具有一个动态说明所要求的形状：诸状态、它们之上的一个有序性，以及后来诸状态被较早者的条件化。

这篇论文与前一篇都不供出这样一个说明，而§3.2明确地表示它的记号什么也没确立。这一替换更精确地界定一次形式化会形式化什么，使前一篇论文的形式纲领保持完好，而其状态空间被替换。上面所拒绝的读法会移除那个纲领；此处所取的读法使它立着且未被执行。

6.4 休谟式困难

前一篇论文拒绝去界定一个系统本应遵循的一条轨迹，理由是这样一个界定要求一份描述性诸事实无从供出的担保 [9]。当下的扩展可能被当作回答那一困难。一条被投射的轨迹不引起任何关于本应为何的问题，因为一个投射声称表示正在到来之物、而非被亏欠之物，而一套在诸投射之上运作的框架，可能看似已逃脱那问题。

它已在不如它们看似有利的条款上逃脱那问题。该困难从来不是轨迹不可用。它是一条携带规范分量的轨迹要求一份担保，而描述性诸资源供不出任何担保。以 $\hat{\gamma}$ 替换 $\gamma^*$ 借移除那规范主张而溶解该困难：一个投射不作出任何规范主张，而无担保被亏欠。这一溶解是完全的，而它使框架付出任何投射能凭之被批评的那块地基。一方的重构可能是自利的，它的投射可能是极其弄错的，而该说明不保留可用来责难任一者的任何资源。它能说一个项被刚性地持有，§5确立这关乎阻碍。它不能说一个项在任何携带分量的意义上是错的。

因而休谟式困难被规避而非被解决，而这一规避是以前一篇论文对规范性问题的拒绝业已蕴含的价钱购得的。一套希望说某些诠释优于另一些的框架，会要求两篇论文都推延的规范性层次，并会在那里以其原初形式面对该困难。

§7 留作悬而未决的问题

下面的诸项按它们对该说明地位的关涉排序，从该说明最为暴露的那些开始。

7.1 构成的结构

§3陈述 $\hat{G}$ 取决于三个经诠释中介的视域，而供不出关于该依赖的任何说明。供出一个，是本文留作未做的主要形式任务，而三个问题组织它。诸视域是可加地构成还是交互。一个可加的构成会容许每一个视域的贡献被独立地评估；一个交互的会使一个重构的贡献取决于伴随它的投射，而§4的诸案例提示这一点而未加确立。

一个视域是否在可界定的诸条件下占主导。§4.3展现一个重构主导一个被构想的当下，而这样的主导在其下发生的诸条件在此未被考察。

什么结构表示一个被估计之空间。前一篇论文为 $G$ 提出了对应的问题，并指出一个延续空间容许收缩、扩张与变形。一个被估计之空间容许这些，且还容许本文所处理的、同 $G$ 的分歧，而一个足以对付两者的表示是所想要的。

7.2 检验一套关于模型的说明

主张 4 与 5 以前一篇论文的核心主张可证伪的方式可证伪，而检验它们的困难更大。

预测是，一个被当作准确地持有的 $\hat{G}$ 之项败坏它的矫正通道，而阻碍伴随那一败坏。检验它要求观察各方被估计之空间的诸项，以及它们被持有的刚性。§6.1记下进入这些的通路经由那些项本身所条件化的行为。

那里提到两条路径，而两者都不予发展。从报告之外的行为推断，可能支撑关于一方被估计之空间的诸主张，而该推断会要求一个关于哪些行为关乎哪些项的说明。$\hat{G}$ 与 $G$ 之间的分歧在其诸后果中是可观察的，而一个关于哪些后果把一个被收缩的被估计之空间同一个被收缩的可用空间区分开来的说明会被要求。这两者中是否有一个支撑一个经验纲领，是悬而未决的。

7.3 本系列的组织

§5.4观察到象征性闭合容许两个描述：作为一个在阻碍中产生持续的机制，前一篇论文给出了它，以及作为一个在一个被估计之空间的诸项中产生刚性的机制，当下的说明容许它。这两者是被在两个层次描述的一个现象，还是以一种要求陈述的方式相关的两个现象，关乎本系列如何被组织。

一个相关的问题关乎那个原初对象。前一篇论文在 $G$ 之上定义阻碍；当下这篇论文在 $\hat{G}$ 之上评估它；而这两个定义之间的关系要求多于当下处理所供出者。§2.4主张清单不变而侧重已移动。一篇第三论文是否应直接在 $\hat{G}$ 之上定义那个原初对象，把前一个定义当作模型与可用空间重合的那个特殊情形，是悬而未决的。

7.4 与认知科学中诸预测性说明的关系

此处所集结的装置，同认知科学中在预测编码、预测处理与主动推断名目下所发展的装置，有一种结构上的相似 [13, 5, 4, 6, 12]，而这一相似在前一篇论文对参照模型的处理中被指出而未被追寻。没有派生的主张被引入，而这些说明在下面所铺陈的诸方面有别。

这一相似是结构上的：一个依据其情境的一个生成模型行动的系统，对着该模型未能预料到的材料修正该模型，并如此举止以使其情境与该模型所表示者相合。这一相似是否支撑任一方向上的迁移，是悬而未决的，而两个障碍是可见的。诸认知说明关乎一个建模一个不回返模型之环境的系统；当下的说明关乎相互建模的各方。而诸认知说明界定一个目标，即预测误差，当下的说明出于§6.4所给的诸缘由拒绝供出它。

7.5 规范性层次

§6.4发现该说明不保留可用来责难一个诠释的任何资源。一套希望供出它们的框架，会以其原初形式面对休谟式困难，而前一篇论文所辨认的那个候选，在此处仍是那个候选：不可修正性，作为一个关系的一条性质，是在诸轨迹被剥夺规范地位之后所存留下来的东西。为什么一个关系对其各方之修正的开放应携带规范分量，被两篇论文都未加回答。

§8 结语

本系列前一篇论文在一个系统事实上可用的诸延续之上定义了生成性阻碍，并推延了一类它的资源无从够到的情形：起于一方所预期之封闭、而非它所承受之封闭的冲突。

本文论证前一篇论文所要求的诸参照模型，由三个经诠释中介的视域构成，一方对其自身前景的评估是这三者的一个函数，而行为对如此构成的空间负责。这一扩展够到被预期的封闭，并提供一种关于怨愤的读法，两者都通过前一套说明业已包含的诸资源。它的构成是无时态偏向的，而它的演化是时间上有序的，这保存了前一篇论文所推导的路径依赖。而与阻碍相关的那条性质，仍是前一篇论文所辨认的那一条：一个项被持有的刚性，把该项的内容及其时间位置搁在一边。

诸代价已被充分陈述。框架的证据处境弱于从前，因为在一个诠释场之上的评估要求进入那个场的通路，而那通路经由那个场所条件化的行为。休谟式困难被规避而非被解决，其价钱是任何一个诠释能凭之被责难的地基。而一套围绕被估计之空间组织的框架，是否保留本系列开始时的极简性，是一个本系列尚未敲定的问题。

关于 AI 协助的声明

作者使用了 Claude（Anthropic）进行讨论、起草协助与编辑修订。所有理论主张、论证与结论均为作者本人所有，作者对内容承担全部责任。

参考文献