Modal Meinongianism and Actuality

Modal Meinongianism is the most recent neo-Meinongian theory. Its main innovation consists in a Comprehension Principle which, unlike other neo-Meinongian approaches, seemingly avoids limitations on the properties that can characterize objects. However, in a recent paper A. Sauchelli has raised an objection against modal Meinongianism, to the effect that properties and relations involving reference to worlds at which they are instantiated, and specifically to the actual world or parts thereof, force a limitation of its Comprehension Principle. The theory, thus, is no better off than other neo-Meinongian views in this respect. This article shows that the notion part of actuality in Sauchelli’s paper is ambiguous from the modal Meinongian viewpoint. Accordingly, his objection splits into two, depending on its disambiguation. It is then explained how neither interpretation forces modal Meinongianism to limit its Comprehension Principle. A third problem connected to Sauchelli’s objection(s) is addressed: how to account for our felicitously referring to nonexistent objects via descriptions that embed reference to properties not actually instantiated by the objects. Overall, the replies to these difficulties provide good insights into the workings of the new Meinongian theory. 1. Modal Meinongianism and Comprehension. Meinongianism is the view that some objects do not exist, but we can generally refer to them, quantify on them, and state true things about them. Any Meinongian theory needs some principle of comprehension for its objects: a † Northern Institute of Philosophy, University of Aberdeen, Scotland, UK. 1 This is the characterization provided in Sainsbury (2010), p. 45. 156 Humana.Mente – Issue 25 – December 2013 principle explaining which (nonexistent) objects there are and which properties they can bear. So-called naïve Meinongianism endorses what Parsons (1979), (1980) has called, by analogy with naïve set theory, an “Unrestricted Comprehension Principle” for objects: (UCP) For any condition A[x] with free variable x, some object satisfies A[x]. Take Joseph Conrad’s (allegedly) nonexistent fictional character Charles Marlow. The naïve view has it that the object is specified via a package of properties, like x is a sailor of the British Empire, x is from London, x transports ivory on a boat through the river Congo, etc. If A[x] stands for the conjunction of the corresponding predicates or open formulas in the appropriate language, then, according to the UCP, an object is characterized by A[x] and, calling it “Charles Marlow”, m, Marlow really has the relevant properties, A[m]. The intuition is that nonexistent objects should in some sense have the properties they are characterized as having – otherwise, how could we know what we are thinking and talking about when we refer to them? We can in principle causally interact with ordinary, concrete, existent objects, thus being perceptually acquainted with many of their features. But when the thing does not exist, we need something like a Comprehension Principle. The UCP does not last long. As Russell (1905a-b) famously showed, when A[x] = Px  Px for some predicate P, the Principle delivers inconsistent objects violating the Law of Non-Contradiction. Additionally, one can prove the existence of anything one wills. For A[x] = “x is made of gold  x is a mountain  x exists”, the UCP allows a priori an object actually having the features of being a golden mountain and existing; so there actually exists a golden mountain – which will not do (as Kant remarked: if the existence predicate could legitimately enter into definitions or characterizations, we could define things into existence). Worse, as pointed out by Graham Priest (2005, p. 83), one can prove anything. If A[x] is x = x  B, with B standing for any sentence, by the UCP for some object, say b, it is actually true that b = b ∧ B; from which B follows by Conjunction Elimination. Neo-Meinongians have traditionally tried to fix the Comprehension Principle by limiting the range of properties that can figure in characterizing conditions. So-called nuclear Meinongianism (Parsons (1980), Routley (1980), (1982), Jacquette (1996)) distinguishes between two families of Modal Meinongianism and Actuality 157 properties, the nuclear and extranuclear ones. Only conditions including just predicates standing for nuclear properties can characterize objects (and a crucial move consists in denying that existence is nuclear). The strategy faces various problems, one of which consists in providing a principled criterion to distinguish nuclear from extranuclear properties. In his book Towards Non-Being (2005), Priest has claimed that the Meinongian can do better. Drawing on insights by Daniel Nolan (1998) and Nick Griffin (1998), he has proposed a Qualified Comprehension Principle for objects: (QCP) For any condition A[x] with free x, some object satisfies A[x] at some world. Because of its QCP’s explicitly referring to worlds, this new kind of Meinongianism has been called “the other worlds strategy” (Reicher (2010)), or “modal Meinongianism” (Berto (2011), (2012)), and I will stick to the latter label. Objects characterized by a condition should have their characterizing features, not automatically at the actual world, but at others: those that make the characterization true. The justification for the QCP bears on the fact that nonexistent objects typically are the targets of intentional, representational states: Cognitive agents represent the world to themselves in certain ways. These may not, in fact, be accurate representations of this world, but they may, none the less, be accurate representations of a different world. For example, if I imagine Sherlock Holmes, I represent the situation much as Victorian London (so, in particular, for example, there are no airplanes); but where there is a detective that lives in Baker St, and so on. The way I represent the world to be is not an accurate representation of our world. But our world could have been like that; there is a world that is like that. (Priest 2005, p. 84) By parameterising to worlds the having of properties, modal Meinongianism promises to avoid restrictions on the range of properties that can characterize objects. Given any property whatsoever, the represented object does exemplify it – at the worlds where things are as they are represented. 158 Humana.Mente – Issue 25 – December 2013 Besides the QCP, modal Meinongianism rests on two other pillars: (1) a modal framework including so-called non-normal or impossible worlds, broadly taken as worlds that are not possible with respect to an unrestricted (logical, perhaps metaphysical and/or mathematical) notion of possibility; and (2) a natural distinction between properties the having of which entails existence, and properties the having of which does not. As for (1), non-normal worlds help with inconsistent characterizations: A[x] = Px  Px characterizes something which is and is not P – but only at the worlds where the characterization holds; and these are no possible worlds for sure. The theory avoids commitment to actually, or even possibly inconsistent objects. As for (2), for instance, the actually nonexistent Marlow cannot actually have such properties as being a sailor, or transporting ivory on a boat, or thinking about Kurtz. To have such features one must be endowed with a physical location and causal powers, which Marlow as a fictional object actually lacks – in short: one must exist. But Marlow has those properties, at the worlds described by Conrad’s story (the ascription of such properties to Marlow is “intra-fictional”, as those working on the philosophy of fiction often say). At those worlds, Marlow is very much existent. His lacking existence at the actual world does not preclude Marlow from actually instantiating several other properties that do not entail existence, for instance: being a fictional character due to Conrad; being Marlow; being a nonexistent object; or being thought about by the Conrad readers (these count as “extra-fictional” ascriptions of features: Marlow does not have such properties within the Conrad fiction). It is easily seen how this may seem to help with the existent golden mountain: A[x] = “x is made of gold  x is a mountain  x exists” characterizes an object represented as an existent golden mountain, and which is a golden mountain at the worlds where the representational characterization is realized, which need not include the actual one. So the QCP does not allow one to prove the existence of whatever one wills. An antecedent of this modal Meinongian setting may be due to Kit Fine, who proposed it in his critical discussion of nuclear Meinongianism back in the Eighties. Fine’s early version of the QCP says: “For any class of properties, there is an object and a context such that the object [...] has in that context exactly the properties of the class” (Fine (1984), p. 138). Now Fine’s contexts 2 See Yagisawa (1988), Restall (1997), Mares (1997). Modal Meinongianism and Actuality 159 play a role similar to the one of the modal Meinongian’s non-normal or impossible worlds: they are fictional or represented situations which can be locally inconsistent or incomplete. Also according to Fine, by parameterizing to contexts the having of properties by objects, one needs no restriction on the properties that can appear in the characterizing conditions, and “the whole apparatus of nuclear properties can drop out as so much idle machinery” (p. 139; see also Fine (1982), pp. 108-9). In an interesting and thoroughly argued recent paper, Andrea Sauchelli (2012) disagrees. According to Sauchelli, certain characterizations of nonexistent objects spell trouble for the modal Meinongian, in such a way that she is forced to introduce restrictions to the QCP. Once the restrictions are in play, modal Meinongianism is no better off than nuclear Meinongianism or other neo-Meinongian theories. The supposedly unmanageable characterizations involve properties encompassing reference to the actual world or parts thereof, or entailing that the characterized object has relations to things that are part of actuality. Sauchelli’s point is introduced in Section 2. As we will see, the notion part of actuality in play in the objection can be read in two different ways from the modal Meinongian viewpoint, thus giving rise to two distinct problems; and the theory has different replies to them. In view of such replies, Section 3 makes things precise by providing a compressed formal presentation of modal Meinongianism, in the shape of a modal semantics including non-normal worlds. Section 4 shows how the theory can effectively address Sauchelli’s (twofold) concern. In the closing Section 5 a third problem is addressed, not directly raised by Sauchelli but connected to his remarks, and having to do with the way definite descriptions referring to nonexistents work according to modal Meinongianism. Taken together the three issues, and their being dealt with by the theory, provide a deeper understanding of the workings of this new kind of Meinongianism. 2. The Objection from Actuality. Sauchelli’s objection to the QCP in unrestricted form is based on the idea that nonexistent and, in particular, fictional objects are often represented by cognitive agents as existing at our world and as having relations to objects that are part of our world. This means that the content of their representations contains attributions that are meant to relate them to parts of our world. [...] These 160 Humana.Mente – Issue 25 – December 2013 representational properties are indexed at our world, in the sense that they are supposed to hold at our world. For example, Joseph Conrad, a member of our world, characterised Marlow, and thus attributed to him certain representational properties, that, among other things, contain reference to our world and that were meant to relate them to objects of our world. In particular, Marlow is represented as being in London, the London that is part of our world. (p. 3) For another example: Travis Bickle, the main character of Taxi Driver, is represented in the movie as such that he drives a taxi in New York, “the New York that is part of @” (“@” standing for the actual world), and as “talking to the mirror (which is in the New York that is part of our world)” (p. 4). Realistic fictional works like Heart of Darkness or Taxi Driver, according to Sauchelli, prescribe us to imagine certain things as happening at the actual world. For example, the movie prescribes us to imagine “that our world contains a taxi driver who turned into a vigilante” (ibid.). The features Marlow or Travis are characterized as having are properties they are “represented as possessing as a part of our world; [they are] not represented as having those properties in other worlds” (p. 6). It is thus in the content of the respective representations that Marlow, or Travis, be part of the actual world @, and thus exist at @: one who is represented as driving a taxi at @, since driving a taxi is an existence-entailing property, must exist at @. One who is represented as being from London – the London which is part of @ – must exist at @. But this flatly contradicts the modal Meinongian view that that Marlow or Travis are nonexistent objects. We should therefore conclude that the theory “is either inconsistent (if it embraces an unrestricted principle of characterization) or incomplete (for it cannot accommodate certain properties attributed to fictional characters)” (p. 6). So formulated, the objection adopts a notion, being part of actuality, or being part of @, or being part of our world, crucially ambiguous from the modal Meinongian viewpoint. Which does not mean that the objection is flawed because of this. Rather, as we will see, it amounts to two different points, depending on how one disambiguates it. The ambiguity becomes apparent once modal Meinongianism has been phrased in formally precise terms; and to this we now turn. Modal Meinongianism and Actuality 161 3. Modal Meinongian Semantics In this Section we use standard tools of world semantics to describe a simple model for the modal Meinongian theory. Take a standard first-order language, L, having individual variables: x, y, z (x1, x2, ... , xn); individual constants: m, n, o (o1, o2, ... , on); n-place predicates: F, G, H (F1, F2, ... , Fn); a designated oneplace predicate, E; the usual logical connectives: negation ¬, conjunction , disjunction , the conditional ; the two Meinongian quantifiers Λ and Σ (written thus, for reasons to be explained soon); a sentential operator, ®; and round brackets as auxiliary symbols. Individual constants and variables are singular terms. If t1, ..., tn are singular terms and P is any n-place predicate, then Pt1 ... tn is an atomic formula. If A and B are formulas, then A, (A  B), (A  B), (A  B), and ®A are formulas; outermost brackets are omitted in formulas; if A is a formula and x is a variable, xA and xA are formulas, closed and open formulas being defined as usual. The only notational novelty is ®, called the representation operator. The intuitive reading of “®A” will be “It is represented that A”, representation being understood as a generic for the intentional activities relevant for the characterization of our nonexistent objects – from imagining, to picturing, to envisaging, to describing in a fiction, etc. An interpretation of L is an ordered sextuple . P is the set of possible worlds; I is the set of non-normal or impossible worlds; P, I are disjoint, W = P  I is the totality of worlds; @  P is the actual world, a possible one. R  W x W is a binary relation on the whole set of worlds. If  R (w1, w2  W), we write this as w1Rw2 and say that w2 is representationally accessible or, quickly, R-accessible, from w1. D is the set of objects of the theory; v assigns denotations to the descriptive constant symbols of L: If c is an individual constant, v(c) ∈ D; If P is a n-place predicate and w ∈ W, v (P, w) is a pair, , with v+(P, w)  D, v-(P, w)  D. D = { d1, ..., dn  D}, the set of n-tuples of members of D ( is stipulated to be just d, so D is D). To each pair of n-place atomic predicate P and world w, v assigns an extension v+(P, w), and an anti-extension, v-(P, w). 162 Humana.Mente – Issue 25 – December 2013 The (anti-)extension of P at w is the set of (n-tuples of) things of which P is true (false) there. The following twofold clause, called the Classicality Condition, ensures that at possible worlds the extension and anti-extension of each predicate be mutually exclusive and jointly exhaustive: (CC) If w ∈ P, for any n-ary predicate P: v +(P, w) ∩ v-(P, w) = ∅ v +(P, w)  v-(P, w) = D If a is an assignment for the variables of L (a map from the variables to D), then va is the denotation function indexed in the usual way: If c is an individual constant, then va(c) = v(c); If x is a variable, then va(x) = a(x). Then, “w ⊩a A” means that A is true at world w, with respect to assignment a, “w ⊩a A”, that A is false at w, etc. (we will omit the assignment subscript when we deal with closed formulas). Atomic formulas have truth and falsity conditions phrased as follows: w ⊩a Pt1... tn iff ∈ v+(P, w) w ⊩a Pt1... tn iff ∈ v-(P, w). The extensional logical words have familiar clauses at all w ∈ P: