Is There Backward Generation in the Institutional Realm?

1. Introduction

In the field of social ontology, it has generally been taken for granted that institutional reality is created synchronically or in a forward-directed manner. According to one influential school of thought (the Searlean), all of institutional reality is created by ‘Status Function Declarations’ (Searle 2010), and Status Function Declarations are supposed to generate institutional states of affairs synchronically (Searle 1989). It can be debated whether all of institutional reality is created by Status Function Declarations (e.g. Epstein 2014), but I think it is quite plausible to hold that at least some quarters of institutional reality are created by Status Function Declarations: think for example of grades (‘N.N. is hereby assigned the grade Y’) and marriages (‘I hereby pronounce you husband and wife’). Now, interestingly, over the last decade it has been pointed out by several philosophers that not all Status Function Declarations are synchronic: some such declarations are directed towards the past (e.g. Barlassina and Del Prete 2015; Silver 2024; Torrengo 2018). More precisely, some Status Function Declarations are directed towards their absolute past – that is, in the jargon of the special theory of relativity, towards their past light-cone. Such Status Function Declarations are perplexing if one is an ontic realist with respect to institutional properties and states of affairs. If successful, such Status Function Declarations seem to change the absolute past; at the very least, they seem to involve some form of absolute backward generation. Both consequences look highly problematic: the notions that the absolute past can change (or be changed) and that there is absolute backward generation are both regularly accused of being contradictory or entailing implausible metaphysics (see e.g. McTaggart 1927 and Black 1956 for some classic arguments).

In this paper, I will argue that neither of the two problematic consequences arises if the ‘results’ of absolutely backward-directed declarations are analysed in terms of mere Cambridge changes realised in B-time. The proposed account extends the mere Cambridge change analysis I developed in my (2021) to account for apparent relative backward generation of institutional properties.¹ The present paper demonstrates how seemingly mundane cases of absolute backward ‘creation’ – but also ‘revocation’ – of putative institutional properties can be straightforwardly accounted for if such ‘properties’ are understood in terms of true predications rather than ontic entities. A key upshot of the account is that ‘institutional properties’ are neither identifiable nor perfectly correlated with enablements and constraints, contrary to the claims of some social ontologists (see e.g. Ásta 2018).

The structure of the paper is as follows. In Section 2, I briefly explain the notion of a Status Function Declaration, give examples of absolutely backward-directed Status Function Declarations and illustrate how such declarations, on an ontic understanding of institutional properties, seem to change the past by means of (positive or negative) backward generation (causation or grounding). In this section, I also explain that the changing-past issue can be effectively circumvented if the revisionary ‘B-theory’ of time (which has independent support) is adopted, but that the issue of backward generation remains unresolved by such a move. Section 3 consequently tackles the latter, more stubborn, difficulty, local to social ontology. I argue that a mere Cambridge change analysis of putative creations of institutional properties, and that a modal mere Cambridge change analysis of putative revocations of institutional properties, can handle such cases without positive or negative backward generation being postulated. I also clarify how my accounts differ from the sort of analysis that has recently been defended by Richard Corry (2025), which is also phrased in terms of mere Cambridge change, at least nominally. In Section 4, I argue that successful instances of backward-directed Status Function Declarations effectively show that institutional ‘properties’ and ‘enablements and constraints’ are not perfectly correlated. I sum up the main points of the paper in the concluding Section 5.

2. Declarations and Retroactive Assignments and Revocations of Institutional Properties

Searle famously introduced the notion of a declaration in his (1975/1979), where he developed a taxonomy of illocutionary acts. Declarations, he explains

Status Function Declarations constitute a sub-class of declarations (Searle 2010: 13, 96, 114), and they ‘create’ institutional state of affairs such as Keir Starmer’s being Prime Minister, this piece of paper’s being a euro bill and person a’s being a convicted criminal. Status Function Declarations may be implemented in various ways, according to Searle: for example, in the form of general rules (typically, as laws, Searle 2010: 97), as ‘assertive declarations’ (e.g., when a judge or referee declares ‘You’re guilty!’ or ‘That’s an offside!’, 1975/1979) and as ‘ad hoc declarations’ (when we assign an institutional property to some entity without relying on a pre-existing rule, 2010: 19–20, 94–96).

Searle takes it for granted that declarations are synchronic – that is, that they create corresponding worldly institutional states of affairs simultaneously with their being performed or applied (Searle 1989: 556–557; 2010: 99). He does not say much to back up the synchronicity view, but he notes that declarations, when spelled out, typically are present-tensed (Searle 1989). As he sees it, the present tense – or what he sometimes calls the ‘dramatic present’ or the ‘present present’ (1989: 556) – signals that the creation of the represented fact, i.e. the event of creation, is simultaneous with the speech act:

Over the last decade, however, philosophers have come to realise that we often make declarations about institutional reality that are retroactive – that is, directed towards the absolute past (henceforth, simply ‘past’).² Philosophers have here tended to focus on what we might call ‘revoking declarations’ or ‘negative retroactive declarations’ (or, if we want to stick to a more Searlean terminology, Negative Retroactive Status Function Declarations) – that is, declarations that nullify an earlier declaration, whereby the institutional property or state of affairs allegedly created by the first declaration is, apparently, revoked or deleted from reality.

Examples of negative retroactive declarations discussed in the literature are: disqualifications of winners in sports (Barlassina and Del Prete 2015; Corry 2025; Geddes 2025; Iacona 2016; Schipper 2022; Silver 2024); revocations of marriages (Ásta 2018: 100–102; Silver 2024); revocations of racial statuses (Ásta 2018: 98, 102); and revocations of criminal statuses (Corry 2025; Torrengo 2018).

Personally, I am not convinced that all of these are good examples of successful negative retroactive declarations, but some of them certainly are. Consider, for example, retroactive disqualifications in sports. Barlassina and Del Prete (2015) discuss the case of Lance Armstrong: in the year 2000 Lance Armstrong was declared the winner of the Tour de France, but later it was discovered that Armstrong made use of banned substances, and in the year 2012 Armstrong was stripped of all his wins at Tour de France, including the one in 2000. Other examples come to mind, such as the 100 metres dash for men in the Summer Olympics 1988: initially Ben Johnson was declared the winner, but three days later he was disqualified because of a positive doping test.

Arguably, then, there are actual cases of successful negative retroactive declarations. However, it should be noticed that apart from such revoking cases, there are also standard positive declarations – that is, declarations that putatively create institutional properties and states of affairs – that are directed towards the past, even if such declarations have not received as much attention as the negative ones. For example, grades are often assigned retroactively (I will return to this topic soon). And so are job assignments: looking at my own Lund University documents appointing me to various administrative roles, such as Director of Doctoral Studies, it transpires that they invariably involve retroactive appointments: the date of the first day on the job, stated on the relevant document, is before the date of the document itself, signed by a superior official. And verdicts in sports – the paradigm assertive declarations – almost universally involve assignment of an institutional status to some event in the past, although no attention has been paid to this peculiar fact in the literature. When a referee stops the game and signals ‘offside’, the referee is in effect expressing that an event that just occurred is deemed (declared) to be an instance of offside. And, arguably, verdicts in courts sometimes concern the institutional status of a past event or deed: for example, whether a certain deed (which might be known to have happened) was a criminal offence. A final example: enrolments on university courses are sometimes made retroactively. Kenneth Silver (2024) discusses a mere hypothetical case, but at my own home university it is in fact common to register students retroactively (what in Swedish is called ‘sen anmälan’).

Now, both sorts of declarations (negative and positive ones) directed towards the past are highly bewildering given an ontic understanding of institutional properties – that is, if one thinks of such properties as genuine properties really out there, as entities that can be quantified over by referential quantifiers (not just substitutional ones, cf. my 2023 and 2025), and possibly as having causal powers of their own.³ Both sorts of declarations, if successful, seem to alter what properties are instantiated in the past (see e.g. Barlassina and Del Prete 2015; Silver 2024); at the very least, they seem to involve some form of backward grounding or causation (Silver 2024; Torrengo 2018).

To see how retroactive creations and annihilations of institutional properties might be taken to involve the past changing as time goes by (assuming a commonsensical ‘growing-block’ conception of the lapse of time), consider, first, a case of a positive retroactive declaration (see Figure 1): initially an event, object or person (or a certain stage, time slice or temporal part of the relevant object/person) does not have the institutional property in question because the relevant declaration has not been made yet; this situation remains unchanged as the event/object/person recedes into the past with the flow of time; but then, abruptly, when some additional time has passed, that event/object/person, now located further into the past, acquires the relevant institutional property because a positive retroactive declaration to this effect is made at the currently present time – a time which is considerably later than the past time at which the event/object/person is located.

Figure 1

Next, consider a case of a negative retroactive declaration: initially, an object/person/event has the institutional property in question because a positive declaration to this effect has recently been made; this situation remains unchanged as the event/object/person recedes into the past with the flow of time; but then, abruptly, when some additional time has passed, that event/object/person, now located further into the past, loses the institutional property because a negative retroactive declaration to this effect is made at the currently present time – a time which is considerably later than the past time at which the event/object/person is located.

This idea (in either of its positive and negative incarnations) involves the notion that one and the same entity, located at a specific time, can change depending on what happens or is done later, as time progresses. Thus, on this view, entities located at certain times cannot only change so-called A-properties – going from, say, being present to being five minutes into the past, to being a year into the past, etcetera – but also institutional properties.

The conception that the past can change is regularly accused of being contradictory (see e.g. McTaggart 1927; Torrengo 2018: 243–244): such changes involve some entity (or entities) both having and not having a particular property, it is objected. For my part, I am not convinced that this accusation of strict self-contradiction is valid. Concerning successive A-properties it is clear that they – if they at all exist in an ontic sense – are not had simultaneously or timelessly by the relevant entity or entities (see my 2013 for discussion of McTaggart’s paradox). And it seems equally clear, given the growing-block conception of time we have been operating with here, that the having and the not-having of the relevant institutional property do not obtain simultaneously (i.e., relative to the moving Present at some specific moment, as depicted in Figure 1) or timelessly either. However, the metaphysics in question is certainly very weird: apart from backward generation (which only gets realised when a certain time becomes present/comes into being), it entails an infinite hierarchy of higher time-dimensions relative to which the changes of the past of the first time-dimension take place (as I argue and illustrate in detail in my 2013, inspired by classic arguments in Smart 1949, Williams 1951 and others; see also Corry 2025). But I do doubt that this sort of view is strictly contradictory.

Howsoever may be, it might be suggested that the changing-past issue can be completely evaded – which is certainly advisable, not least because of its apparent commitment to higher-order time dimensions – if the defender of institutional properties as ontic entities adopts the revisionary B-theory of time. According to the B-theory, all times – that is, times that from our position in time appear to be either ‘past’, ‘present’ or ‘future’ – are equally real: they and the entities that are located at them can all be quantified over by referential quantifiers. Together, they form a four-dimensional unit: viz., our universe with its total history, ‘past’ and ‘future’, laid out once and for all. Further, there is no flow of time: no Present that moves along the ‘time block’. A static arrow of time is substituted for the passage or flow of time: the various times and the entities located at them are temporally ordered by so-called B-relations – the relations of being earlier/later than and being simultaneous with, relations which are not subject to change, but which hold ‘timelessly’ or ‘permanently’. Thus, what we, from our point view in the universe, regard as future and non-existent, is in fact already real, although we have little or no knowledge about how this future is, about what is realised at those times.⁴

Importantly, if the B-theory of time is adopted, it can be held by defenders of institutional properties as ontic entities that the last declaration (of the relevant ones in question) in the time series takes precedence and overrules earlier ones: this final declaration governs whether or not the entity in question, located at some earlier time, instantiates the relevant institutional property (Iacona 2016 appears to defend a view like this, although the premise about the B-theory of time is left implicit in his discussion). Thus, in a case of a positive retroactive declaration, it is always or timelessly the case that the entity in question, located at some time t which is earlier than the time t’ of the retroactive declaration, instantiates the institutional property, although no one knows about this fact at the earlier time t. And in a case of a negative retroactive declaration, it is always or timelessly the case that the entity in question, located at some time t which is earlier than the time t’ of the retroactive declaration, does not instantiate the institutional property, although people located shortly after t think that the entity instantiates the institutional property (cf. Austin’s void marriages, 1962: 16–17, and Searle’s counterfeit money, 1995: 32–33). Thus, a B-theoretical move along these lines does arguably dissolve the changing past issue.

This kind of analysis is in my view on the right track, but it still involves some form of backward generation: positive backward generation in cases of positive retroactive declarations, and ‘negative’ backward generation in cases of negative retroactive declarations.⁵ Being committed to backward generation is, I take it, a theoretical vice, ceteris paribus. And it is so irrespective of whether the generation is supposed to be of a causal or grounding variety.

Philosophers who discuss backward generation (positive and negative) of institutional properties tend to assume that the generation in question, if it is real, is a case of backward causation (e.g. Corry 2025; Silver 2024). It is not evident why, though. If we go to the ‘source’ and look at how Searle conceptualises generation by declarations (which he thinks is a synchronic phenomenon, as we have seen), we find that over time he has oscillated between causal and constitutive accounts.

In his (1983), Searle says:

But in his (1989), Searle appears to have changed his mind about the generative means, although he does not explain why (presumably, he wants to avoid committing to instantaneous causation across spatial distances). He now says:

Searle here seems to be suggesting that declarations – or the intentions behind them, and the collective recognition of them – and their associated results are related, not by causation, but by a relation of ‘constitution’. Since Searle also denies that institutional phenomena are located in people’s minds (2010: 17 – and I presume he would say corresponding things about absences of institutional phenomena due to declarations), this constitution relation cannot be conceived of in terms of identity. Rather, Searle is arguably alluding here to a non-causal, asymmetric ontological dependence or grounding relation. Such a view appears also to be adopted by Searle in his subsequent work on social ontology.⁶ Thus, if one wants to stick to a conformist Searlean conception of how declarations generate their results, but wants to allow for retroactive declarations, then arguably one should conceptualise the relevant backward-directed generation in terms of ontological dependence or grounding (as done by Torrengo 2018).⁷

No matter: as I said above, I think a commitment to backward generation (positive or negative) is a theoretical vice ceteris paribus, irrespective of whether the generation is supposed to be causal or in the form of grounding. Backward causation (typically in the form of positive backward causation) has been discussed a good deal in the philosophical literature, and though I am unaware of any conclusive proofs to the effect that backward causation is impossible, the onus is surely on defenders of backward-causation-as-actual to provide compelling reasons for thinking that backward causation is actual – certainly so given all the well-known problems that afflict backward causation, such as bilking arguments and various apparent paradoxes (see Faye 2024 for an overview),⁸ and given that orthodox physics, adhering to the special theory of relativity, does not invoke backward causation when explaining events. Thus, a theory which can avoid postulating backward causation is, I submit, ceteris paribus preferable to a theory that is committed to backward causation.

Similarly concerning backward grounding, although that hypothetical phenomenon is much less discussed in the literature. Standardly, grounding is taken to be a synchronic relation, typically between spatially co-located entities: wholes grounded in their proper parts, mental states grounded in brain states, sets grounded in their members, and so on. Recently, Baron, Miller & Tallant (2020), drawing on work of others (such as Koslicki 2015 and Wilson 2020), have provided interesting arguments that might be taken to support forward-directed grounding over time (based on forward-directed causal processes) and instantaneous grounding (mediated via asymmetric relations) between spatially disjoint entities. However, they do not explicitly provide a mechanism for, and do not explicitly argue for, backward grounding.⁹ Absent a convincing general account of backward grounding that shows that such grounding would be metaphysically unproblematic or innocent – and I think it is unlikely that such an account can be put forth given the prima facie similarity between backward causation and backward grounding¹⁰ – I think it is reasonable to hold that a theory that can avoid postulating backward grounding is ceteris paribus preferable to a theory that is committed to backward grounding.

3. Avoiding Backward Generation

I now argue that a (modal) mere Cambridge change account of retroactive declarations can avoid committing to backward generation of ontic institutional properties or absences thereof. First, I address a case involving a positive retroactive declaration, then a case involving a negative retroactive declaration.

As a case potentially involving positive backward generation of an institutional property, consider a retroactive assignment of a grade to a student’s script by a university professor (here I elaborate on brief comments in my 2021). In the Swedish university system (and in the Dutch, I am told¹¹), grades are often assigned retroactively: a student completes an assignment on a certain date t, but a busy professor grades the assignment at a later date t’ and in so doing gives the student an official grade which is valid from the earlier time t onwards. In Sweden, this is executed by the professor explicitly indexing the grade to the earlier date t in the IT-system used (at the time of writing: the ‘Ladok’ system). Now, if something were really created at t by the declaration at t’ (in Ladok, the declaration is implemented by the professor digitally certifying the grade entered), we would have a case of positive backward generation. I do not think, however, that we have to assume that such a generative activity takes place. On the view I advocate, such a scenario involves ‘mere Cambridge change’, and thus does not involve the student acquiring an ontic property.¹²

The term ‘mere Cambridge change’ was coined by Peter Geach (1969). He discussed the following ‘Cambridge criterion’ of change, which is based on the work of the Cambridge philosophers Bertrand Russell and Ellis McTaggart (e.g. McTaggart 1927: 14; Russell 1903: 469):

Geach pointed out that this Cambridge criterion gives the odd result that a person a ‘changes’ when she comes to be shorter than another person b in virtue of the latter’s growth. For a monadic predicate, ‘is shorter than b’, becomes true of a having not been true of a at an earlier time. This is not a genuine or real change in the person a, according to Geach: it is a mere Cambridge change. The predicate ‘is shorter than b’ comes to apply to a solely because of a genuine change that happens to someone else, b, somewhere else.

An analogous analysis applies, I submit, in the case of a retroactive ascription of a grade. In such a case, a specific predicate of the form ‘has grade Y on the course C at university U’ applies to a student at some time t because of a genuine change or event that happens elsewhere and later in time, namely the professor’s official grading or declaration of the form: ‘The student N.N. is hereby assigned the grade Y on the course C at the university U, indexed to time t’. No ontic property comes to be instantiated at t by the student because of the later declaration, it is just that the predicate is applicable to him or her at t because of the professor’s declaration. At times before t, the present-tensed predicate does not apply to the student, but from time t onwards the predicate does apply to student. (A tenseless version of the predicate – in which the time clause features as a built-in component – parsed either as ‘has-at-t-the-grade-Y’ or as ‘has-the-grade-Y-at-t’, is true of the student at all times. It is also always true to say, using the time clause as a sentence operator prefixed to the predication construed as present-tensed: ‘at t (the student has the grade Y)’.¹³)

I want to clarify that the idea is not that the past has changed with respect to whether the present-tensed predicate applies at t.¹⁴ I agree that it is desirable to avoid, if possible, the notion of a changing past, even if the change concerns merely what predicates apply at specific times (and not what ontic properties are instantiated). Thus, the B-theory of time is still assumed at this stage of the argument. (This is not just to avoid the changing past problem, however. There are many reasons for endorsing the B-theory apart from avoiding this issue: see e.g. Sider 2001 and my 2009 for detailed discussion. Thus, the endorsement of the B-theory here is not an ad hoc move.) Given the B-theory of time, the present-tensed predicate applies to the student already at t. This is because the professor’s declaration, located at the later time tʹ, already exists at t in the sense that it can be quantified over, at t, by ‘∃’. Thus, the mere Cambridge change that I claim is involved in this kind of case is of a somewhat unorthodox sort. The standard examples considered by Geach and others are synchronic: the mere Cambridge change and the genuine change happen at the same time – they are simultaneous. In the case of a retroactive assignment of a grade, the mere Cambridge change (the initial applicability of the present-tensed grade predicate) ‘happens’ or is located earlier in time than the genuine change (the professor making the relevant declaration).

Importantly, the backward-directed generative relation (in the form of causation or grounding) drops out of the picture (cf. Figure 1). The applicability of the relevant predicate is not, I take it, caused or grounded: the existence of the student (located at the relevant time) and the valid declaration are by themselves sufficient for the applicability of the predicate – no external generative relation has to exist (cf. note 9).¹⁵ Thus, the mere Cambridge change account of the retroactive grading example can be illustrated as in Figure 2.

Figure 2

Of course, even if the present-tensed predicate applies already at t (because of the later declaration at t’), no one knows about this applicability already at t: its applicability is not disclosed at t by some detectable change, and what happens in the future is in general epistemologically inaccessible to us (arguably partly because there is no backward causation or grounding). Thus, on the view I advocate, an entity can have an institutional ‘status’ or ‘property’ (in the sense that an institutional predicate applies to it) although no one is aware of this fact at the time in question (cf. Searle’s case of a dollar bill that has slipped through a crack in the floor, 1995: 32). In a case of a retroactively assigned grade, no one will act at the earlier time at issue on the fact that the student has the grade in question. But later, when the retroactive declaration has become official so to speak, people might very well act on the fact that the grade applied already at the earlier time. For example, in Sweden, whether a student is entitled to further financial support from the government (specifically, the ‘CSN agency’) depends on when the student took and passed his or her courses.

What I have said about retroactively assigned grades can be easily generalised to other types of positive retroactive assignments, such as retroactive job appointments, retroactive course enrolments, retroactive verdicts in sports and retroactive legal verdicts. I leave the details as exercises to the zealous reader.

Before I go on to consider negative retroactive declarations, I should comment on Richard Corry’s account of retroactive declarations which is also formulated in terms of mere Cambridge change (Corry 2025). It may seem that we are essentially in agreement (Corry refers to my 2021 approvingly), but one important difference between our accounts (which is not explicitly noted and commented upon by Corry) is that Corry conceptualises mere Cambridge change in terms of ontic extrinsic properties and relations, something I do not do.

Corry writes:

And when discussing Kenneth Silver’s brief critical remarks about mere Cambridge change accounts of retroactive declarations, Corry says:

Corry here apparently agrees with Silver that the monadic institutional ‘properties’ in question are instantiated and not merely ascribed, and that they hence are, in my terminology, ontic properties (although he disagrees – rightly so, in my view – with Silver’s claim that they are not extrinsic or relational). Corry goes on to argue that mere Cambridge changes typically involve ontic relations (to put it in my terminology), specifically in institutional occurrences:

In response, I want to highlight that mere Cambridge change – as Geach specified it – need not involve any extrinsic properties, nor any relations, ontologically understood. Quite the contrary, as I will now argue, beginning with extrinsic properties.

As we saw above, mere Cambridge change is explained by Geach in terms of monadic, extrinsic predicates that ‘change’ (that have and do not have application at distinct times) because of genuine changes or events that occur elsewhere; and he explained mere Cambridge change in such terms because people like Bertrand Russell defined change in such linguistic terms. The reason they are mere Cambridge changes, and not genuine changes, is that they do not correspond to any varying ontic properties of the subject in question. If they did correspond to varying ontic properties of the subject, the change would presumably be a case of genuine or real change, even if the ontic properties would not be intrinsic but extrinsic. And in such a case, the relevant change would arguably be ontologically dependent on the intrinsic change that happens elsewhere (cf. Kim 1974; Schaffer 2019: 763–766). But as I see it, the main virtue of a mere Cambridge change account (in the proper Geachean sense) is that it avoids reference to grounding and causation, a fact which frees us (in the retroactive cases under discussion in this section) from backward generation.

Moreover, the relevant relations that are ostensibly involved in mere Cambridge changes need not, as far as I can see, be understood as ontic either. For example, the relation being taller than (Geach’s paradigm example) is often understood as an internal relation, and internal relations are standardly understood as ‘ontological free lunches’ (Armstrong 1997: 12–13) – a metaphorical expression which I take to indicate that they are not ontic. I see no pertinent reason to think that things are different in the institutional cases. The relevant relations are in such cases relations like being assigned the status Y by, and such relations are arguably not external, ontic ones (I discuss this point in detail in Hansson Wahlberg (n.d.), partly by way of criticising Baron, Miller & Tallant 2020). Thus, while Corry, seemingly, wants to commit to some form of ontic realism in relation to extrinsic institutional properties and relations (despite nominally adopting a mere Cambridge change approach), I do not. Finally, I read Corry as holding that cases of negative retroactive declarations should be analysed in terms of mere Cambridge change too (although he focuses on a positive case in the latter half of his paper) – but again, I do not (at least: not in terms of actual mere Cambridge change), as I will now explain.

Time, then, to account for negative retroactive declarations. Disqualifications of winners in sports, such as the Lance Armstrong case, are, I think, the clearest instances of such declarations (at least of those hitherto discussed in the literature, but I think revoked sentences might also be good examples). A typical case of a disqualification of a winner in sports looks, schematically, like this: An athlete apparently meets the requirements – as specified by the relevant rules of the game/sport/tournament – for winning the competition in question. An assertive declaration is made by some authorised person or committee to the effect that the athlete is indeed the winner. However, unbeknownst to the authorised person, the athlete does not actually fulfil the requirements for winning (according to the rules) because the athlete has cheated in some hitherto undetected way. Later, this act of cheating is disclosed, and the athlete’s official winner status is retroactively revoked by some authorised person or committee.

I think that such scenarios can be accounted for by invoking neither annihilations of ontic institutional properties nor backward negative generation. Again, I assume that the B-theory of time is correct. On that assumption, the revoking declaration is real already at the time of the initial assertive declaration, although no one knows about this fact at that time. Because of the later revoking declaration, the initial assertive declaration is overruled: it is cancelled out and hence void. Thus, it is not true at the earlier time that the athlete is the official winner (i.e., winner in virtue of a valid positive assertive declaration),¹⁶ although everyone thinks and behaves at that time as if the athlete is the winner (according to both the rules and the assertive declaration). Thus, had anyone said at that time that the athlete is not the official winner, then that person would have said something true (though everyone would have denied this at that time) – and the key truthmaker for that claim, had it been made, would have been the later ‘revoking’ declaration.

In other words, no ontic property is deleted from reality. If the term ‘revoke’ suggests such a thing, then the term is misleading. What is the case is that, had not the negative retroactive declaration been made, then the athlete would have been the official winner, because of the initial assertive declaration. But as we can see from the discussion of the positive cases above, the athlete’s being the official winner would not in such a case have involved the athlete instantiating an ontic property. A mere Cambridge change account of this status could have been provided, along the lines outlined above.

Let me stress here that because the athlete in question on this account does not actually change winning-status (not even in a B-theoretical sense in which the present-tensed predicate ‘is the official winner’ applies to the athlete at some initial date and does not apply at some later date), a mere Cambridge change analysis is not applicable to this scenario as it is taken to actually unfold. But something in the vicinity is applicable, namely what might be called ‘modal mere Cambridge change’, involving mere predicative change across possible situations: in the actual world, the predicate ‘is the official winner’ is not true of the athlete in question, but in the counterfactual situation (where the revoking declaration is not made) the predicate is true of the athlete. This ‘change’ of the athlete across possible situations does not however involve genuine change: the variation does not consist in the athlete instantiating an ontic property in one situation (the counterfactual one) and failing to instantiate it in the other situation (the actual one). As already pointed out, in the counterfactual situation the predicate ‘is the official winner’ applies to the athlete (from a certain time and onwards) merely because of an assertive declaration – its application is thus an instance of ordinary, non-modal, mere Cambridge change realised within that counterfactual situation.

I admit that it looks somewhat awkward, initially, to hold both that it is never true to say that the athlete won and that it is true to say that the athlete’s win was revoked: the latter proposition seems, after all, to suggest that there was a win, that the athlete did initially win. But again, I think this awkwardness arises from the relevant propositions, such as <the athlete’s win was revoked>, conveying misleading and false ideas about what actually went on (as discussed above). Once attention is paid to what should be taken to be really going on – namely, that an initial assertive declaration is rendered void by a revoking declaration located later in B-time – the initial feeling of awkwardness should recede.

To distil the take-home messages of the long-winding discussion of this section: On a mere Cambridge change account of positive retroactive declarations in B-time, no positive backward generation occurs: no instantiations of ontic institutional properties are brought into being in the past by present declarations. On a modal mere Cambridge change account of negative retroactive declarations in B-time, no negative backward generation occurs: no instantiations of ontic institutional properties are hindered from obtaining in the past by present declarations. Ontic realists with respect to institutional properties, on the other hand, are arguably committed to thinking of such cases in terms of positive or negative backward generation (causation or grounding): that is, in terms of present declarations making instantiations of ontic properties obtain in the past or in terms of present declarations hindering instantiations of ontic properties from obtaining in the past. The account I advance, then, is not only ontologically simpler than ontic realist ones, in line with Ockham’s Razor, it also avoids indulging in backward generation, positive and negative, which is a further virtue of the account, ceteris paribus.

4. Divorcing Institutional Statuses from Enablements and Constraints

Before concluding this paper, let me say a few words about the (putative) connection between institutional properties or statuses and ‘enablements and constraints’ in the social realm. These are sometimes held to be intimately connected: if not identical then at least perfectly correlated – if you have an institutional property, then you are enabled to do (or hindered from doing) certain things; and if you are enabled to do (or hindered from doing) certain things in the social realm, then you have an institutional property. Ásta, for example, says:

Now, an identity view is hard to square with the sort of analysis of institutional properties defended in this paper, according to which the key truthmakers for truths about the having of institutional properties are declarations. If institutional properties were identical with enablements and constraints, then the truthmakers for such truths would arguably be enablements and constraints (cf. Asay 2023: 55) – or whatever serves as truthmakers for claims about enablements and constraints, and I do not think that declarations are suitable for the latter role. In fact, I think that cases of retroactive declarations effectively show that institutional ‘properties’ or statuses are not even perfectly correlated with enablements and constraints – certainly not with concurrent enablements and constraints. Let me briefly motivate this claim.

In cases of positive retroactive declarations, an entity can acquire an institutional ‘property’ or status (in the sense that a relevant predicate gets applicable to it) although this has no impact on, makes no difference to, what that entity can do at the time in question. This should be contrasted with what the entity is entitled to in virtue of acquiring the ‘property’ – that is, what deontic powers apply. Of course, later when the past having of this property or status becomes widely known, then this past having of the property (or rather the current representations thereof) may result in ‘enablements and constraints’. Again, a student who has been retroactively assigned the grade pass on a university course may thereby (in Sweden, at least) be legally entitled to economic support the upcoming semester (a positive deontic power, in Searle’s terminology), a fact that – once it has been noted and acted upon by the governmental agency that decides on the matter (in Sweden, the CSN agency) – enables the student to go on with his or her university studies. The grade may also entail that the student is no longer entitled to re-register on the relevant course (a negative deontic power, in Searle’s terminology), a fact that – once it has been noted by the relevant university administration – constrains the student’s choices about what courses to take the next semester. But at the time when the grade was initially ‘had’ (by retroactive assignment), no enablements and constraints were realised (although the legal deontic powers already applied because of the regulation), because no one was aware of the student having the grade at the time in question. Relatedly, since the acquisition of the grade did not result in the student starting to ‘function’ in a new way at the earlier time (even granting that ‘function’ is to be understood in a loose sense only, as pleaded by Searle 2010: 95, n. 2), I also think that it is misleading to use the term ‘status function’ for the grade at the earlier time. At that time, the having of the grade or institutional status merely involved the applicability of a predicate (entailing certain legal deontic powers): the having of the grade had no functionality or causal consequences at the relevant time, because no one knew about it at that time.¹⁷

In cases involving negative retroactive declarations, it may be that some person or other entity never had the institutional status or ‘property’ in question (say, being winner of some prestigious competition) because of a later revoking declaration; but since people thought at earlier times that the person was the winner, they may also have thought that the person was entitled, for example, to a certain amount of money (a positive deontic power), which was in fact transferred to the person, a transaction which enabled him or her, at a time before the revoking declaration, to buy a fancy apartment. All the same, the person did not have the institutional winner status at these times. And so, an institutional status and ‘enablements and constraints’ can come apart in this way too – that is, by enablements and constraints being realised without the relevant status applying.

5. Conclusion

I have put forward a theory of positive and negative retroactive declarations according to which they – even if valid and ‘effective’ – do not change the past and do not take part in positive or negative backward generation. The changing-past issue is avoided by incorporation of the B-theory of time. The backward-generation issues are avoided by analysing the ‘results’ of successful backward-directed declarations in terms of (modal) mere Cambridge changes. A key outcome of the account is that ‘institutional properties’ are neither identifiable nor perfectly correlated with enablements and constraints, challenging the views of prominent social ontologists.

Competing Interests

The author has no competing interests to declare.