1.INTRODUCTION
The aim here is to vindicate a particular account of propositions that can be traced in the early writings of Russell (1903) and Moore (1899; 1902). According to it, propositions are structured entities of a very special sort: they are non-representational chunks of reality about which we directly think. This account accommodates standard desiderata about propositions elegantly and straightforwardly. However, as we will see later, it is exposed to at least one serious objection: it seems unable to deal with falsehoods. As the objection goes, it seems that this neglected account, by identifying the proposition that Socrates is wise with the state of affairs of Socrates being wise, is confusing propositions with what makes propositions true, the representation of things being thus and so with the very things being thus and so. So, it is unable to accept false propositions. This inability is the main reason why both Russell and Moore abandoned their early accounts. Furthermore, this is the main reason why, nowadays, the account is neglected or quickly dismissed. In its place, some representational account of propositions is given. We can make sense of falsehoods only if propositions are representational entities of some sort. These representational accounts could preserve the structured character of propositions or favor the idea that propositions are simple entities. But I am not convinced by these alternatives. Let the unattended account be given another chance and show that it is still a powerful alternative.
From the very start, I will assume that there are such entities as propositions. And I will assume that these entities are called for to fulfill the following desiderata in an optimum degree^1^:
(i) Propositions are the premises and conclusions of modally valid arguments, like the ones expressed by the following sentences: "All men are mortal; Socrates is a man; therefore, Socrates is mortal"; "There are prime numbers greater than 2; therefore, there are numbers". The necessity exhibited by arguments like these ---the fact that the conclusion must be true, or cannot be false, if the premises are true--- is not grounded in the words used to express those premises and conclusions but in the nature of what is said by those words.
(ii) Propositions in themselves are not mental states. They are not private or subjective entities of some sort. Instead, they are the objects of certain intentional mental states and they are common or shareable content. A proposition is what is believed (known, doubted) when someone believes (knows, doubts) that something is the case, in such a way that when you and I believe that all human beings are mortal, there is a numerically identical content that is common to our numerically distinct beliefs: we believe the very same thing, namely that all human beings are mortal.
(iii) Propositions are not declarative sentences. They are the content expressed by declarative sentences in their contexts of utterance. The sentence "Abelard loves Heloise," uttered by me today here, expresses the same proposition as the sentence "He loves her," uttered by someone else centuries ago over there. The sentence "He loves her," uttered by someone centuries ago over there while pointing to Abelard and Heloise, expresses a distinct proposition from the sentence "He loves her," uttered by me today while pointing to John and Mary. We can say the same thing uttering distinct sentences; we can say distinct things uttering the same sentence. What we say when we say that something is the case, that is, what we express when we utter a declarative sentence in a context of utterance, is a proposition.
(iv) Propositions are the fundamental or primary bearers of truth values; that is, a true (false) belief is true (false) in virtue of the fact that its object is a true (false) proposition. A declarative sentence is true (false) in virtue of the fact that what is expressed by it is a true (false) proposition--but not the other way around.
(v) Propositions essentially represent things as being a certain way, they essentially have the truth conditions they have, and they essentially have the truth conditions they have in virtue of essentially representing things as being a certain way.
(vi) Propositions are necessary existents; in fact, at least to perform the roles assigned by desiderata (i)-(v), propositions must exist and be selfidentical regardless of the truth value they happen to have.
I do not claim that the former list of desiderata is uncontested; far from that. There may be more, which I have not included, or maybe we are facing "a jumble of conflicting desiderata," which no unique type of entity can fulfill (Lewis 1986, p. 54). However, these disputes go beyond the scope of the present paper. Even if the list of them is incomplete, even if no unique entity satisfies them all, we can still take it as a valuable pointer of what a proposition is meant to do; so, a pointer of what a proposition is meant to be. The focus here will be on how some of the most salient accounts of propositions perform under these commonly accepted demands. Furthermore, it is assumed that the connection between these desiderata and the nature of propositions will be an Eleatic one: what a proposition is cannot be divorced from what a proposition does; if a proposition exists, then it thereby does something; to be is to make a difference.
2. STRUCTURED PROPOSITIONS: THE STANDARD ACCOUNT
Most philosophers think that propositions are complex representational entities. This general view adopts two main alternatives that seem to dominate the landscape: on the one hand, some take propositions to be sets of possible worlds (e.g., Lewis 1986; Stalnaker 1976); on the other hand, some take propositions to be structured entities of some sort, that is, complex entities that require a more demanding form or arrangement than the one exhibited by sets. The standard account of structured propositions understands that just like sentences are not mere collections, sets, or lists of words, propositions are not mere collections, sets, or lists of constituents. Propositions have constituents and a structure, form, or arrangement that binds them together into a special kind of unity, a single proposition. A pretty standard characterization of structured propositions could be the following:
Roughly, to say that propositions are structured is to say that they are complex entities, entities having parts or constituents, where the constituents are bound together in a certain way. Thus, particular accounts of structured propositions can (and do) differ in at least two ways: 1) they can differ as to what sorts of things are the constituents of structured propositions; and 2) they can differ as to what binds these constituents together in a proposition. (King 2019, Intro.)^2^
One main motivation for understanding propositions as structured entities is that propositions are meant to be more finegrained than sets of possible worlds. It seems intuitively clear that there might be necessarily equivalent propositions that are numerically distinct. If Abelard believes that whales are mammals and Heloise believes that whales are mammals and 2 is a prime number, then Abelard and Heloise believe necessarily equivalent but distinct propositions. However, that possibility is not allowed if propositions are understood as sets of possible worlds. If propositions are sets of possible worlds, all necessarily equivalent propositions are the same proposition, the same set of possible worlds. The structured proposition theorist claims that such a result cannot be right: even if they somehow map the very same set of possible worlds, the proposition that whales are mammals is not identical to the proposition that whales are mammals and 2 is a prime number. A correct theory of propositions should make room for this difference.
The claim that propositions are more finegrained than sets of possible worlds does not entail the claim that propositions are structured (King 2007, p. 6; Richard 1990, p. 34). After all, one could say that propositions are simple entities that just happen to be more fine-grained than sets of possible worlds; that is, the fact that propositions are more fine-grained entities than sets of possible worlds could just be a brute fact. The structured proposition theorist claims that this fact about propositions is not a brute fact. The proposal of the structured proposition theorist attempts to satisfy an additional explanatory demand: the idea that there must be something in virtue of which propositions are more fine-grained than sets of possible worlds. The structured proposition theorist says that it is precisely in virtue of the structured nature of propositions that propositions are more fine-grained than sets of possible worlds; that is, the fine-grained character of propositions is not a brute fact. As King puts it:
I think that propositions do have constituents. This is mainly because I find the idea of ''simple fine grained propositions'', fine grained propositions without constituents or parts, mysterious. What would make such a simple proposition be about, say, Paris as opposed to Santa Monica? In virtue of what would it have the truth conditions it in fact enjoys? I cannot see that these questions have answers if propositions are held to be simple and fine grained. But it seems to me they should have answers. (2007, p. 6)
Notice that mere complexity is not sufficient. Sets also may have internal complexity. Something additional is required: the structure that binds the constituents together into a special kind of unity, a unity akin to the one exhibited by a declarative sentence. Allegedly, there is some mirroring between the structure and constituents of propositions and the structure and constituents of the sentences that express them. The structured proposition theorist would typically claim that this isomorphism is not a sheer coincidence and that her account of propositions naturally matches the language's compositional character.
*We can say that structured proposition theorists hold that sentences express propositions that are complex entities (most of) whose constituents are the semantic values of expressions occurring in the sentence, where these constituents are bound together by some structure inducing bond that renders the structure of the proposition similar to the structure of the sentence expressing it.**This highlights an important feature of structured proposition accounts that distinguishes them from the other main competing account of propositions, namely the account of propositions as sets of possible worlds... Because structured propositions have as parts the semantic values of expressions in the sentences expressing them, the semantic values of those expressions are recoverable from the semantic values of the sentences (i.e. the propositions). (King 2019, §1)*
Nevertheless, we cannot say there are structured propositions in virtue of the compositional character of language. Instead, what we can safely say is this: considerations in favor of structured propositions based on the compositional character of language already presuppose the very idea that it is in virtue of its constituents and structure that a proposition essentially represents things as being a certain way (Merricks 2015, pp. 130-133). The main motivation for structured propositions is that they, allegedly, are capable of explaining how propositions manage to satisfy desideratum (v). As the structured proposition theorist claims, it is in virtue of its components and structure that a proposition essentially represents things as being a certain way and essentially has the truth conditions it has.
According to the standard structured proposition theorist, desideratum (v) is only the surface of the following deeper explanation:
A. Propositions essentially represent things as being a certain way, they essentially have the truth conditions they have, and they essentially have the truth conditions they have in virtue of essentially representing things as being a certain way. [desideratum (v)]
B. Propositions are structured entities.
C. The constituents and structure of the proposition fully explain, fully ground, the fact that the proposition essentially represents things as being a certain way and, therefore, the fact that it essentially has the truth conditions it has.^3^
Thus, for example, someone who embraces theses (A)-(C) typically understands that: (i) the proposition that Abelard loves Heloise is a particular structured whole (it might be a mereological fusion, a settheoretical structure, a state of affairs, or other; the structured whole in question varies according to how one accounts for the unity of the proposition); (ii) that such structured whole is constituted by Abelard, Heloise and the relation of loving, arranged in a certain manner due to some structure, say R*; (iii) that the constituents and structure of such whole are what fully explains how is it that the proposition essentially represents things as being a certain way and, therefore, why it essentially has the truth conditions it has; and (iv) that the proposition that Abelard loves Heloise represents Abelard loving Heloise even if the sentence "Abelard loves Heloise" is false, so the proposition must exist even if it is false. The standard structured proposition theorist---the one who is the target of Merricks' criticisms too---unders‐tands that distinct sentences, such as "Abelardo ama a Eloísa" and "Abélard aime Héloïse," may express the same proposition, that Abelard loves Heloise. That they do so is a contingent fact about language: after all, the name "Abelardo" could have referred to Plato. Neverthe-less, how the proposition itself represents things as being is not a sheer coincidence: that Abelard loves Heloise, the very proposition expressed by those sentences, is a structured whole that essentially represents Abelard loving Heloise---and not, say, Heloise loving Abelard, or William hating Heloise---is something fully explained by its constituents and structure, or so it is argued.
3. AGAINST THE STANDARD ACCOUNT: SIMPLE PROPOSITIONS
Defenders of simple propositions claim that the components and structure of the proposition do not fully explain, do not fully ground, the fact that the proposition essentially represents things as being a certain way. Furthermore, they consider this insufficiency a good reason for understanding that propositions are simple entities that fundamentally and primitively represent things as being a certain way. Thus, they embrace thesis (A), reject thesis (C), and, therefore, see no reason for accepting thesis (B). Merricks (2015) and Keller (2013) are among those who take propositions to be simple entities. Here the focus will be on Merricks' proposal, which is the most recent and developed.
Merricks' main objection against the position defended by the standard structured proposition theorist undermines thesis (C). According to Merricks, the standard structured proposition theorist is unable to explain how the constituents and structure of the proposition manage to fully explain, fully ground, the fact that the proposition essentially represents things as being a certain way and, therefore, essentially has the truth conditions it has. The problem raised by Merricks is, ultimately, this one: What unites constituents and structures into a proposition? What brings the constituents and structure together into the proposition itself, the necessary existent entity that manages to essentially represent things as being a certain way? Merricks' attack is directed against various accounts of the unity of the proposition (2015, ch. 4). I cannot go through all of them here, but this is his general objection: From the fact that some constituents are arranged in a certain way it does not follow that the resulting complex essentially represents things as being a certain way at all. Therefore, it simply does not follow that such a complex entity essentially represents things as being a certain way solely in virtue of its constituents and structure. These complexes---i.e., mereological fusions, settheoretical structures, states of affairs, or others---do not have a truth value, as propositions do. Furthermore, they do not have a truth value precisely because they do not essentially represent things as being a certain way at all. Likewise, if they do not essentially represent things as being a certain way, a fortiori they do not essentially represent things as being a certain way solely in virtue of their constituents and structure. One could bestow the structure itself, say R , with the special power, say P, of making the structured whole
The main point here is that the accounts of the unity of the proposition offered by structured proposition theorists do not bring us the right kind of unity, namely, a proposition. This point is so even if one assumes, as I am doing here, that R is capable of uniting many things into one thing or another, that is, even assuming that R is, as Baxter (1996) would put it, a one-making relation that makes one entity out of various entities. For R* must be more: it must be a one-proposition-making relation. In Merricks' words:
...a successful account of the unity of the proposition tells us what unites the relevant constituents not merely into some entity or other, but into a proposition. That is, it tells us what unites the relevant constituents into a necessarily existing entity that essentially represent things as being a certain way... Moreover, given a successful account, those constituents together with the way in which they are united fully explain how the resulting entity manages to represent things as being that way... (2015, p. 155)
Merricks' main objection against structured propositions is guided by the idea that something must represent in fundamental terms, that is, something must represent not in virtue of something else. Nevertheless, the constituents and structure of a structured proposition do not essentially represent things as being a certain way in fundamental terms. If they do manage to essentially represent things as being a certain way, it is in virtue of something else, something that is bestowed with a mysterious representational power.
Merricks takes this failure to be a reason for simple propositions. Propositions, he claims, not only represent fundamentally (that is, not in virtue of something else) but also primitively (that is, brutely, wit‐hout further explanation). The structured proposition theorist cannot afford the luxury of saying that propositions primitively represent things as being a certain way. Because if propositions are structured, then at least some of their constituents would appear magically or mysteriously correlated with how the proposition represents things as being (Merricks 2015, pp. 200-205). Thus, those who defend structured propositions should reject that a proposition primitively represents things as being a certain way. The whole point of bringing propositional constituents and structure to the scene was to give a full explanation of the representational power of propositions. Regardless, now we have seen that the move was superfluous. The mystery of the proposition's representational power has not been explained away; it has only been transferred to some constituent(s) of its complex nature or some other supplementary ingredient, which is then bestowed with some mysterious representational power.
Some principle of ontological parsimony seems to be ruling the situation here. If the complex nature of structured propositions is posited to explain the mysterious representational power of propositions, but such complex nature cannot remove the mystery, then we better posit simple entities with that mysterious representational power. Merricks' main argument for denying that propositions have constituents and structure is precisely that, under the standard theory of structured propositions, the constituents and structure do not fully explain, do not fully ground, the representational power of propositions. For all we know, it is still a mystery, a magical trick, a brute or arbitrary fact, how does the structured whole <Abelard, lo‐ving, Heloise>represents things as being a certain way, that is, as Abelard loving Heloise, and not, say, as Heloise loving Abelard, or as William hating Heloise. Thus, Merricks concludes that propositions lack constituents, they lack internal complexity, and they are simple abstract necessary existents that primitively and fundamentally represent things as being a certain way.
It is based upon these reasons that Merricks ends up defending the simplicity of propositions:
- If propositions have constituents, then each proposition has constituents that are intuitively correlated with how that proposition represents things as being. [...]
- It is not the case that each proposition has constituents that are intuitively correlated with how that proposition represents things as being. [...]
- It is not the case that propositions have constituents. (2015, p. 206)
If propositions lack constituents, they lack internal complexity and are simple entities. In this view, propositions are fine-grained sui generis entities, and their representational power is both primitive and fundamental. Keller (2013), who endorses a similar view, calls it "propositional primitivism."
Here is a second objection against structured propositions, directed against the standard account of singular propositions (Merricks 2015, ch. 5; Fitch and Nelson 2018). We can put it thus: There are singular propositions, that is, propositions that are directly about a particular entity, like the proposition that Socrates is wise, which is directly about Socrates. According to the standard account, this is another reason for understanding that propositions are structured entities: singular propositions represent the entities they are about by having them as constituents. If a proposition has constituents, then it exists only if its constituents do. Nonetheless, some singular propositions are directly about particular things that no longer exist, that have never existed, or that possibly do not exist. Presumably, Socrates is one of those things. If so, then some propositions do not necessarily exist. However, propositions do exist necessarily (desideratum (vi)). Therefore, the standard account is false. So, this reason counts in favor of the simple account. Since a singular proposition exists necessarily and is directly about an entity, no matter whether this entity happens to exist or not, then it should not need to have this entity or any other entity as a constituent. Such a proposition could be a simple entity that is directly about some entity without having any constituents or internal structure.
Merricks' account faces the following problems. Firstly, it involves too many brute, primitive, or arbitrary facts. In general terms, it posits a realm of propositions as a mysterious veil between thinkers and the world. Then, it bestows propositions with a representational power that is as mysterious and magical as the one that Merricks himself attributes to structured propositions. Moreover, it entails somewhat mysterious brute facts concerning aboutness. Let me explain this charge. Following Plantinga (1983), we can distinguish three individually plausible but jointly incompatible theses concerning propositions:
Serious Actualism : Necessarily, no object can instantiate a property without existing; therefore, necessarily, if a proposition is true, then that proposition exists. Existentialism : Necessarily, singular propositions ontologically depend upon the entities they are about. Necessarily, if a proposition is about an object, then that object exists. So, the proposition that Socrates is wise, which is about Socrates, cannot exist unless Socrates exists. Contingency : Possibly, there are objects that do not exist. Typically, we treat concrete particulars like Socrates in this way: Socrates could have failed to exist.
Following Plantinga's solution, Merricks endorses Serious Actualism and Contingency, and rejects Existentialism. He has independent reasons for believing that all propositions are necessary existents and that there are singular propositions. He also holds the commonsensical belief that some individuals possibly do not exist. He believes that there are possibly true negative existential singular propositions, such as that, possibly, Socrates does not exist. In other words, Merricks accepts that it is possible that Socrates does not exist; that the singular proposition that Socrates does not exist must exist to be true; and, therefore, that the singular proposition that Socrates does not exist can exist and be true without Socrates existing.
Nevertheless, here is the problem that Merricks must face: How can he account for the fact that the proposition that Socrates does not exist is a necessary existent which is directly about Socrates, the individual who, as Merricks himself claims, does no longer exist or possibly does not exist? Merricks strategy is to reject what he calls The Aboutness Assumption, the idea that "necessarily, if a proposition is directly about an entity, then that proposition stands in a relation to that entity" (2015, p. 186), which is just another version of the thesis of Existentialism. To get rid of The Aboutness Assumption, Merricks treats aboutness not as a relation but as a monadic extrinsic property of propositions, in such terms that "a claim regarding what a proposition is about is really a partial description of that proposition's property of representing things as being a certain way" (2015, p. 189).
I do not find Merrick's strategy persuasive enough.
Firstly, the idea that aboutness is a monadic extrinsic property seems difficult to understand. Merricks attempts to illustrate it with the example of the property of being the only object in the universe (2015, p. 189). Nevertheless, it is not clear how this example helps. How is being the only object in the universe not a monadic intrinsic property of the universe itself? After all, if O is the only object in the universe, O is the universe, and that is all one can safely assert in such a scenario. Regardless, even if we can make sense of monadic extrinsic properties, how understanding aboutness in that way can help us to deal with how singular propositions are supposed to work needs further clarification. Granted that the proposition that Socrates does not exist is true; granted that we do not count Socrates himself as a relatum of aboutness since, allegedly, Socrates no longer exists and aboutness is not a relation; and granted that we can partially describe the property of representing things as being a certain way that the singular proposition has: In what does this partial description consist? What are we using to partially describe the property of representing things as being a certain way that the singular proposition has if not Socrates himself? One could follow Plantinga (1983) and appeal to individual essences: the proposition that Socrates does not exist is about Socrates' individual essence, the property of being identical to Socrates, a property that only Socrates himself can instantiate, but that is not necessarily instantiated. Yet this sounds implausible. Individual essences are somewhat mysterious entities, at least how Plantinga understands them. Is not Socrates himself a constituent of the property of being identical to Socrates? If so, how can the latter exist without the former existing? Besides, if we appeal to individual essences, in what sense the proposition that Socrates does not exist keeps being a singular proposition about Socrates and not about a proxy of him, such as his individual essence? Are the propositions that Socrates is wise and that Socrates does not exist about the very same object or not? If they are about the same object, it is difficult to resist the idea that they are about Socrates himself. If they are not about the same object, it seems difficult to understand why the first one is about Socrates himself and the second one about Socrates' individual essence but not about Socrates himself.
Secondly, taken at face value, aboutness does seem to be a relation. Moreover, if this relation is meant to be instantiated, it must have relata. Nevertheless, if one relatum is missing, how can the relation __ being about __ be instantiated? The objection holds whether we understand propositions as simple or as structured entities. It does not even require to endorse a theory of direct reference. All that is required for it to work is to accept, prima facie, the existence of a proposition about a particular object and understand aboutness as a relation. If so, the nonexistence of the object which the proposition is about cannot be accepted. In Williamson's words: "Necessarily, if o does not exist then there is no such item as o, so there is no such item as the proposition that P(o), so the proposition that P(o) does not exist" (2002, p. 241). If a proposition is (directly) about a particular entity, then it better be the case that this particular entity exists.
A different solution would be to accept that true singular propositions about contingent existents are contingent existents, too, as Salmon (1998) does. Salmon accepts both Existentialism and Contingency but is forced to reject Serious Actualism. That is, he is willing to accept that, possibly, there are things, such as Socrates, that do not exist; that there are singular propositions about Socrates that no longer exist since Socrates does no longer exist; and that the present lack of existence of those singular propositions about Socrates does not prevent them from being true. So, according to Salmon, there can be true propositions that do not exist. Nonetheless, how can something be true without existing? How can something instantiate any property without existing? It sounds very implausible, too.
Merricks chooses the least plausible path because he embraces an account that is internally divided by components that push in opposite directions: on the one hand, the idea that propositions are necessary existents, in particular, that a proposition cannot be true without existing; on the other hand, the idea that there are possibly true negative singular propositions.
Finally, to summarize this section, let us add something more general about Merricks' proposal, particularly his critical stance against structured propositions. One could say that Merricks is demanding from a metaphysical explanation something impossible: to fully account for the nature of propositions solely in terms of their constituents and structure is to demand a full account of the identity of propositions in terms of how they are composed or grounded. However, it can be argued that neither composition nor grounding are identity. Both constitution and grounding are relations between distinct relata, whereas identity only holds between an object and itself. Granted that the proposition is a complex entity, if someone attempts to explain its nature and function in terms of its constituents and structure or full grounds, she can only offer a partialaccount of it insofar as the proposition itself is not identical with its constituents and structure or full grounds, just like a table is not identical with many things being related. One entity is not and cannot be identical with some other entity or with some other entities-in-whatever-relation. If a table exists, it is one thing, distinct from all other things, and not many things in some relation (Russell 1903, pp. 133, 140-141, 473). In fact, in virtue of the Principle of the Indiscernibility of Identicals, the whole and its parts, the composed and its components, the grounded and its many partial grounds, no matter how related, are discernible and therefore distinct: one thing is not and cannot be identical with many things. Strictly speaking, if one attempts to explain the identity of a thing, one can only offer a metaphysically trivial (typically circular) or false (typically insufficient) account. However, from the fact that one cannot offer a metaphysically informative and sufficient account of the identity of a thing in terms of other things it simply does not follow that the former entity is simple.
In general terms, I have sympathy for this objection, but not in this particular case. What we say about tables cannot be extended to propositions without further argument, even if we understand tables to be merely apparent entities. We can see that a table has parts; we can see the partial role that the legs play when we attempt to unravel the identity of a table; we can see that without the legs we could not have a table; we can see that the table and its legs stand in various ontological relations; and so on. However, we cannot say the same about propositions. Firstly, it is far from evident that propositions have parts. Assuming that a proposition that represents a table being thus and so has parts, just like the table represented by it, would beg the question about the complexity of propositions. Secondly, even if we grant the complexity of propositions, we need to see in an obvious way the role that the constituents and structure of propositions may have when accounting for their nature and function. In this respect, Merricks has a point. Suppose the only reason for positing constituents and structure of propositions is that they are allegedly capable of fully explaining the representational powers of these. In that case, their positing is insufficiently motivated. One could perfectly make sense of the repre‐sentation of complex entities through simple means, and so does Merricks' proposal.
Still, against Merricks, we could ask the following: Why should we posit representational entities at all, regardless of their simplicity or complexity, if we can think directly about the presumably complex entities allegedly represented by them? That we directly think about the complex entities that constitute the world is precisely the route offered by the neglected account that I examine in the final section.
4.STRUCTURED PROPOSITIONS VINDICATED: THE NEGLECTED ACCOUNT
As I said, there is a non-standard account of structured propositions. This account was embraced by the young Russell (1903) and the young Moore (1899; 1902), but they abandoned it quite soon in their respective careers. Apparently, the reasons for abandoning it were pretty much the same ones that others now use for rejecting it. Despite this current lack of favor, I think the account deserves another opportunity.^4^
Both the young Russell and the young Moore understood propositions to be structured entities. Russell un derstood them as states of affairs, that is, as some object or objects having some property or relation. Moore understood them as complex concepts, that is, as concepts standing in some relation or relations---under the peculiar understanding of concept embraced by Moore back then.^5^
Neither Russell nor Moore understood propositions as representational entities. Propositions, according to them, do not point to something beyond themselves, like Fregean thoughts or senses are supposed to do, or as propositions as abstract representational states of affairs are typically conceived (Chisholm 1989, ch. 15; Plantinga 1974). In particular, propositions are not made true by something distinct from themselves. They are, in themselves, chunks of reality, in such a way that a true proposition is identical to its truthmaker, the chunk of reality that makes it true. In other words, according to this account, the truth-bearer is the truth-maker.^6^
In what follows, I will focus mainly on Moore's version, but everything I say here can be applied to Russell's version with slight amendments. Here is a representative passage of Moore's account:
Once it is definitely recognized that the proposition is to denote, not a belief or form of words, but an object of belief, it seems plain that a truth differs in no respect from the reality to which it was supposed merely to correspond: e.g., the truth that I exist differs in no respect from the corresponding reality---my existence. (1902, p. 21)
Later, after having abandoned his early theory, Moore retrospectively describes it thus:
It is a theory which I myself formerly held, and which certainly has the advantage that it is very simple. It is simply this. It adopts the supposition that in the case of every belief, true or false, there is a proposition which is what is believed, and which certainly is. But the difference between a true and a false belief it says, consists simply in this, that where the belief is true, the proposition, which is believed, besides the fact that it is or 'has being' also has another simple unanalysable property which may be called 'truth'. 'Truth', therefore, would, on this view, be a simple unanalysable property which is possessed by some propositions and not by others. The propositions which don't possess it, and which therefore we call false, are or 'have being'---just as much as those which do; only they just have not got this additional property of being 'true'. And the explanation of those two different facts having the same name which are in the Universe if a belief is true, and one of which is absent if it is false, and of their relation to one another, would be simply as follows. One of these two facts, the one that is equally whether the belief be true or false, is of course, the proposition. And the other one, the one which is only if the belief be true, consists simply in the possession by the proposition of the simple property 'truth'. (1953, p. 261)
This understanding of propositions nicely accommodates our desiderata. Recall that propositions are meant to be the premises and conclusions of modally valid arguments; the objects of belief and other mental states; the contents expressed by declarative sentences; and the primary bearers of truth values---desiderata (i), (ii), (iii) and (iv), respectively. Well, as it happens, for the young Moore, propositions (or "judgments," as he called them in 1899) are nothing but complex concepts, and complex concepts are ultimately nothing but simple concepts standing in some relation or relations, and concepts exhaust reality. A Moorean concept is a "universal meaning," a "content," "the symbolized" (1899, p. 177), "something which we mean" (1899, p. 180), a "logical idea" (1899, p. 193). Concepts are meant to be the "possible objects of thought" (1899, p. 179). Thus, the young Moore offers us a world which is readymade for thought and language, readymade for logical inferences, and readymade for bearing truth values. It is a world where there is no gap between thought, as such, and reality. In fact, everything socalled concrete that exists, our minds and the physical or sensible world included, is ultimately made of proposi‐tions and their ultimate constituents, simple concepts:
From our description of a judgment, there must, then, disappear all reference either to our mind or to the world. Neither of these can furnish "ground" for anything, save in so far as they are complex judgments. The nature of the judgment is more ultimate than either, and less ultimate only than the nature of its constituents---the nature of the concept or logical idea. (Moore 1899, p. 193)
Unsurprisingly, Moore called his own account "the most Platonic system of modern times" (quoted in Preti, 2013, p. 187). It is a world made of nothing but concepts, and Moorean concepts are just like Platonic forms or ideas: their identity and existence do not depend on their being instantiated by concreta because they enjoy epistemic, metaphysical, semantical, and explanatory priority over any other possible beings (Harte 2019; Irwin 1999).
From this world image follow at least three pleasant consequences, which Merricks (2015, p. 123-130) himself admits as virtues or relative advantages of the neglected theory:
Firstly, we obtain direct realism about belief. When you truly believe that p , you directly believe some chunk of reality, namely that p .
Secondly, we obtain a straightforward and neat account of truth. Truthmaking is not really a relational affair between two distinct entities, one that represents and another one that is represented. Those who think of propositions as a medium between ourselves and reality suffer from double-vision: reality is propositional/conceptual; propositions are not a sort of veil between us and the world; facts are (identical with) true propositions. Thus, the way in which Moore's account deals with desiderata (v) is indirect: it denies its very presuppositions. According to this desideratum, propositions essentially represent things as being a certain way, essentially have the truth conditions they have, and essentially have the truth conditions they have in virtue of essentially representing things as being a certain way.
Nevertheless, the young Moore can block this chain of thought from its very start. According to him, propositions are not representational entities at all. A fortiori they do not have their truth conditions in virtue of representing things as being a certain way and they are not true in virtue of representing things being in a certain way. Truth is a primitive, ultimate, undefinable property that some propositions happen to possess. Propositions are the direct objects of thought, and objects of thought are chunks of reality in itself, not entities that represent or point towards something beyond themselves.^7^
Thirdly, the problems concerning the unity of the proposition get dissolved or reduced to a more fundamental level. Since reality is conceptual and propositional through and through, problems of the unity of the proposition are now problems of the unity of reality, or, better said, of the unity of Moorean complex concepts or Russellian states of affairs in general. Either complex concepts have the property of being true or they lack it. Either states of affairs obtain or do not; that is, either states of affairs are true propositions (i.e., facts) or they are false propositions. However, either way, they must be united and cannot be mere aggregates of their constituents. The problem now is not to obtain a unity of the right sort, a unity that exhibits representational powers, but a more fundamental one: the problem of the unity of reality, that is, the problem of how things hang together to make up one thing and ultimately one reality, or, in general, how can some entity be related with other entity at all. Of course, questions concerning the unity of reality may still be troublesome. As Bradley would claim, pluralistic ontologies are unintelligible insofar as they need to explain how relations manage to relate their relata (1930, chs. II-III). Regardless, this ancient problem is a problem for everyone who admits that reality contains at least two entities of whatever sort standing in some relation of whatever sort, that is, for everyone in town except perhaps the radical Parmenidean (Della Rocca 2020, ch. 1).
What about desideratum (vi)? Young Moore would say that neither propositions nor their constituents change across times or worlds. Both propositions and their constituents are in timeless and necessary terms. Each proposition is a whole understood as the mereological essentialist understands wholes: something that cannot be what it is if its parts change (Chisholm 1989, ch. 7). Nevertheless, propositions are special wholes. They are not material aggregates or bundles of properties. They are structured in themselves, and together they form a vast network or graph, the nodes of which are the concepts that constitute them and are shared by them, the concepts that make them belong to a single Figura. Such a network or graph, on this account, is reality itself.
*At this point, the main objection against the theory is raised. The account cannot deal with false or contingently true propositions (possibly false propositions). Merricks, who has Russell (1903) as his target, puts this main objection thus: Suppose that A loves B. Then there is the state of affairs of A's standing in the loving relation to B. But had A not loved B, that state of affairs would not have existed. So that state of affairs exists contingently. Russell 1903 implies that that state of affairs is the proposition that A loves B. So Russell 1903 implies that A loves B exists contingently. But all propositions exist necessarily. [...] So Russell 1903 is false. And suppose that A does not love B. Then there is no state of affairs of A's standing in the loving relation to B. Then, given Russell 1903, the proposition that A loves B does not exist. More generally, Russell 1903 rules out the existence of false propositions.[...] Russell--even in 1903--seems to recognize that Russell 1903 ill accommodates false propositions. (Merricks 2015, pp. 126-127)*
Two distinct objections seem entangled here but point to the same problem. In the first place, prima facie, at least some states of affairs exist contingently. For instance, it seems contingent that Abelard loves Heloise. However, if propositions are both necessary existents and identical to states of affairs, then there is no room for contingently true propositions or contingent states of affairs. In the second place, it is claimed that when something is not the case, say when it is false that Desdemona loves Cassio, then there is no fact like Desdemona standing in the relation of loving to Cassio. Thus, the (false) proposition that Desdemona loves Cassio does not exist either. Either the propositions are true and necessarily exist, or they are fal‐se and necessarily do not exist. Either way, there is no room for propositions that exist and are false or possibly false; but certainly, there are propositions that exist which are false or possibly false; therefore, the neglected account cannot be true.
I see at least two ways of dealing with these entangled objections.
An initial way is to bite the bullet and accept that only true propositions exist (necessarily) and that false propositions do not exist (necessarily). When we think falsely, we are not in contact with reality since reality is the totality of what is the case and nothing more. When you and I falsely think that there are polar bears in Egypt , we are not thinking about the same thing because, in this case, there is simply no shareable content, no common object of thought. Of course, we may say false sentences and have false thoughts. These, as utterances and mental acts, respectively, exist and have properties of their own, but they are not about and do not express propositions. When we think falsely, we simply are not engaging with propositions. If you and I think, falsely, that there are polar bears in Egypt, each of us has a numerically distinct mental state, yet these mental states are not about something common to them. Each of us may be in contact with a private illusion or fictional entity of some sort, with some minddependent, subjective, purely intentional entity, but not with propositions, and so goes for contingency. If the proposition that Abelard loves Heloise is true, then it is a fact that Abelard loves Heloise. We may think this is a contingent arrangement because Abelard could have hated Heloise, but then again--this position will hold--we are not thinking about reality. We are experiencing an "illusion of contingency" (Kripke 1980, p. 150).
Since we cannot think of anything but only of what is, false thoughts must refer to mind-dependent, private, subjective, purely intentional entities but not to propositions as understood by the neglected account. Similar treatment should be applied to contradictions or impossibilities: they do not exist, and necessarily so. Things are the way they are, and there is absolutely no other way in which they could have been. Of course, this first strategy implies embracing a radical form of necessitarianism, a highly controversial thesis. However, as Karofsky (2022) has recently argued, the idea that some things are contingent is an extended prejudice that does not resist philosophical scrutiny. Her main argument (2022, ch. 3) can be summarized thus: Every entity is such that it could have been otherwise (contingent) or could not have been otherwise (necessary). Contingent entities call for an explanation, an explanation that "would indicate that in virtue of which a contingency is such as it is rather than not" (2022, p. 91). Yet this explanation cannot be contingent; otherwise, we would face either an infinite series of contingent entities or some brute, unexplained, contingency. Furthermore, the explanation cannot be necessary since whatever is sufficiently explained by a necessity is itself necessary and not contingent. Thus since nothing can explain contingent entities, there are no contingent entities. Everything is necessary. In Parmenidean style, she claims that once we accept that something is, then we cannot accept that it could have been otherwise. A thing is all the ways it is, and necessarily so.^8^ This necessitarian view is undoubtedly counterintuitive, but it is philosophically powerful and neat. As it can be noticed, this strategy implies that propositions satisfy desiderata (i)-(vi) only when they are true, which is the only way in which they exist, and necessarily so.
A second way to deal with the entangled objections demands a less radical form of necessitarianism. It also demands a sharp distinction between, on the one hand, states of affairs, propositions, or possible arrangement of objects and, on the other hand, facts, true propositions, or actual arrangement of objects. That propositions are necessary existents is wholly consistent with true propositions being contingent. Moore's theory does not equate propositions, as such, with true propositions. True propositions are a subset of propositions whose members are those propositions that have the additional and primitive property of being true. Moore (1899) said that all concepts (propositions being nothing but complex concepts) have being but only "empirical" propositions involve the concept existence as a constituent. However, we do not need to buy the distinction between being and existence to make sense of the young Moore's account of true propositions. We can make sense of this theory in simple words: we can take existence to be the same as being, and we can take propositions and their constituents as necessary existents, as the objector does. Propositions are the fundamental bearers of truth and falsity, and nothing can be true or false without existing because no entity can instantiate a property or relation without existing. In this theory, the property of being contingent does not qualify the existence of propositions but only their truth value .
Furthermore, the truth value of a proposition is not to be explained in terms of a relation between the proposition and another entity represented by it. It is a primitive property that some propositions have, and others lack. Thus, the proposition that Abelard loves Heloise and the proposition that Desdemona loves Cassio are both necessary existents, since they both must exist for them to have a truth value, and so goes for their constituents: Abelard, Heloise, Desdemona, Cassio, and loving must all exist if the propositions they constitute do. As it happens, the proposition that Abelard loves Heloise exists and is true; however, the proposition that Desdemona loves Cassio also exists but is false. Does this mean that Desdemona does not stand in the relation of loving Cassio? No. The proposition that Desdemona loves Cassio is a perfectly respectable and united existing proposition. It just happens that it lacks the property of being true. We can say that it is a contingently false proposition, that is, a possibly true proposition. But, again, to be possibly true, the proposition itself and its constituents must exist.
If this interpretation of the theory is right, then there are no true negative existential singular propositions. A proposition like that Socrates does not exist cannot be true since it is directly about Socrates, and if it is directly about Socrates, then Socrates must exist (Williamson 2002, pp. 240-242). We could also say that all contingently false propositions, qua propositions, exist precisely because a contingently false proposition is a possibly true proposition, and a possibly true proposition is still a possible arrangement of constituents, a possible object of belief, and, as such, a part of reality. Thus, the bearer of a truth value, the proposition itself, must exist to be true or possibly true. It is not existence what distinguishes an actually true proposition from a possibly true proposition.
The neglected proposal, under this second interpretation, has a Tractarian flavor. We can rephrase it thus: The world is everything that is the case: the totality of facts. Facts are states of affairs that happen to obtain. States of affairs exhaust logical space, the space of possibilities, reality as a whole. What we call the world, which is the totality of facts, is just a tiny subregion of reality as a whole. States of affairs, propositions, are possible arrangements of objects. As such, they exist necessarily, regardless of which of them happen to obtain.
Moreover, objects exist across all possible arrangements of them, and they always come in states of affairs. An object always exists and is this way or another. Objects are the world's substance, which remains invariant across logical space. Ontology is necessary: "It is neces‐sary that everything is such that it is necessary that something is identical with it" (Williamson 2013, p. 2). To this basic picture we just need to add the idea that states of affairs are identical with propositions and the idea that a true proposition is identical with a fact. Thus, the obtaining of a state of affairs is simply its truth. All possible propositions are state of affairs because a proposition must already be an arrangement of objects to be true (i.e., possibly false) or false (i.e., possibly true). If a proposition is true, it is a fact, a state of affairs that obtains; if it is false, it is a state of affairs that does not obtain, a possible arrangement of objects, a possibly true proposition.^9^ If what we think happens to be true, then what we think is a fact; we think what is the case. There is no further explanation of the truth of a proposition. Truth is a primitive, ultimate, undefinable property that some propositions have, and others lack. Likewise, there is no distinction between what is thinkable and reality as a whole.
As it can be appreciated, this second way of dealing with the related objections to the neglected account still carries an important component of necessity. In particular, it carries with it a commitment to the idea that ontology is necessary, which has been supported by Williamson (2002; 2013). For this second way to work, we must reject the Contingency thesis, and this fact could be considered a deficiency of the story. After all, our common intuitions tell us that there are plenty of things that could have failed to exist, such as Socrates, and plenty of things that do not exist but could have existed, such as the fourth sibling that I never had. However, recall that the other relevant alternatives we have when dealing with singular propositions about things that we intuitively take to be contingent existents are not very promising. If we want to accept Contingency, then we must reject either Existentialism, as Plantinga (1983) and Merricks (2015) do, or Serious Actualism, as Salmon (1998) does. Rejecting Contingency, as Williamson does, is the price to pay if we want to preserve Serious Actualism and Existentialism. This price seems fair. After all, how can we possibly think truly about something that does not exist, not even as a possibilium? It seems impossible! For those who worry about being loyal to our commonsensical intuitions, the counterintuitive character of Williamson's necessitarian ontology is compensated by two factors. Firstly, it is philosophically powerful and neat--not as much as Karofsky's view, but more than the other alternatives discussed here. Secondly, objects are necessary existents only insofar as they are logical objects. Necessarily, all things exist and are what they are and have the potential they have.
However, on the other hand, and in contrast with Karofky's radical necessitarianism, it is contingent whether things fully realize their potential. For instance, it is contingent that Socrates is concrete; he could have never been born. So, we can accept that Socrates is a necessary existent while rejecting the counterintuitive idea that Socrates necessarily exists concretely.
As we can see, the two ways in which the neglected account can deal with falsehoods involve abandoning commonsensical intuitions: either we abandon the intuition that things could have been otherwise, or we abandon the intuition that at least some things could have failed to exist. Once more, this might be a fair price to pay. After all, intuitions are just psychological facts about ourselves. They are not sacrosanct principles and are not reasons for or against any philosophical theory.
5. CONCLUSION
I have discussed the standard account of structured propositions and the main objections against it. I have also discussed and criticized the alternative account that takes propositions to be simple representational entities. Then I have tried to restore the confidence on a neglected account of structured propositions, once defended by the early Moore and the early Russell, particularly by offering two alternative ways of dealing with falsehoods. These two alternative ways appear effective, but they demand that we give away commonsensical intuitions about the modal status of reality as a whole or of relevant parts of it. As I hope to have shown, the neglected account is still viable and powerful.
ACKNOWLEDGEMENTS
I am grateful to José Tomás Alvarado, Javier Vidal, Carlo Rossi and two anonymous referees for helpful criticism.
CONFLICT OF INTEREST STATEMENT
The author declares that he has no conflict of interest.
FUNDING
This paper has been written as part of the research project Fondecyt Regular 1200002, ANID, Chile.
NOTES
- See, e.g., Bealer (1998) , Cartwright (1987b) , Merricks (2015: ch. 1), and McGrath & Frank (2020). There are arguments for each of these desiderata, but here, for brevity, I will assume them as given.
- Thus, what I call here "the standard account" admits internal variety. Structured propositions have been defended in different ways. Among these, we should highlight the neo-Russellian approach (e.g., Braun 1993; King 2007; Salmon 1986; Soames 1987) and the neo-Fregean approach (e.g., Evans 1982; McDowell 1994). Although the early Russell (1903) and the early Moore (1899; 1902) also believed in structured propositions, they understood them as non-representational chunks of reality, so I am not including them in this family. I consider theirs as a nonstandard, neglected account of structured propositions.
- There is an intuitive contrast between full grounding and partial grounding. For example, take the atomic facts
and
, the disjunctive fact q>, and the conjunctive fact q>. It can be said that
is full ground of
q> and only partial ground of q>. Full grounds of X need nothing else to metaphysically explain X; they provide a sufficient explanation of X. Full grounds have the interesting consequence that, if they are posited, then it is metaphysically necessary that what they ground obtains. See Bliss Trogdon (2021, §§1.3-1.4) .For historical revisions of the accounts of the early Moore and the early Russell, see Briceño (2021) and Candlish (2007, ch. 3) , respectively.
According to the young Moore, concepts were like Platonic forms, and the whole of reality was made of them. Of the relation (or relations) that ultimately holds between concepts, Moore says nothing. But if that relation (or relations) is what glues simple concepts together into complex concepts, then it seems to be like Tractarian logical form: something ineffable, something that can only be shown, not said. See Briceño (2021, p. 110) .
For the idea that Russell and Moore, in their early writings, were embracing versions of the identity theory of truth, see Baldwin (1991) , Candlish (2006), Cartwright (1987a) , Dodd (2008) , and Hornsby (1997) . There is room for variety, of course. Some identity theorists take notions that play the truthbearer role (e.g., truth, proposition) as explananda and try to explain notions that play the truthmaker role (e.g., fact, Reality) in terms of them. Others do it the other way around. Regardless, the statement held by all of them is a statement of identity, which obviously, expresses a symmetric and reflexive relation. Thus, the identity theory of truth (or reality) does not count as a theory or account of truth (or reality) if, by theory or account, you understand some asymmetric and irreflexive account or explanation in which something obtains in virtue of something else, such as the relations of grounding, supervenience or ontological dependence are commonly understood.
These two closely related ideas--direct realism and truth as identity--can be traced in the work of McDowell (1994) and Hornsby (1997) . As McDowell puts it: There is no ontological gap between the sort of thing one can mean, or generally the sort of thing one can think and the sort of thing that can be the case. When one thinks truly, what one thinks is what is the case. So since the world is everything that is the case..., there is no gap between thought, as such, and the world. (1994, p. 27; emphasis in the original)
I cannot present Karofskys argument with all its details for lack of space, so I leave this task to the open-minded reader.
If we follow the Tractarian doctrine to the full, we should leave aside necessary truths and necessary falsehoods since they are not considered to be genuine propositions (they lack a proper sense, they do not say that things are thus and so). Nevertheless, we can remain neutral on this matter, for we could also claim that necessary truths are states of affairs that necessarily obtain, while necessary falsehoods are states of affairs that cannot obtain. I thank one of the referees for this suggestion.