, Volume 23, Issue 3, pp 543–550

Ontology, Reference, and the Qua Problem: Amie Thomasson on Existence


    • Department of PhilosophyHankuk University of Foreign Studies
Original Paper

DOI: 10.1007/s10516-011-9172-x

Cite this article as:
Sauchelli, A. Axiomathes (2013) 23: 543. doi:10.1007/s10516-011-9172-x


I argue that Amie Thomasson’s recent theory of the methodology to be applied to find the truth-conditions for claims of existence faces serious objections. Her account is based on Devitt and Sterelny’s solution to the qua problem for theories of reference fixing; however, such a solution cannot be also applied to analyze existential claims.


OntologyExistenceReferenceAmie ThomassonMetaphysics

1 Introduction

What are the truth-conditions for claims of existence? In this paper, I discuss Amie Thomasson’s recent answer to this question. According to her, claims of existence about an entity, e, are answerable, that is, they can be judged as true or false, if we can recognise and elaborate the conditions necessary for grounding the terms that are used to designate e. Thomasson, in fact, thinks that the truth-conditions for existence claims are to be found in those conditions that are necessary to disambiguate the grounding of the reference of the terms used in the sentences in question. This solution, even if elegant, is ultimately found unconvincing.

In the first section, I introduce the qua problem for theories of reference fixing. In the second and third sections, I explain how Thomasson solves this problem and the connection between this solution and her account of the truth-conditions of sentences involving existence. In the last section, I advance two objections against this account.

2 Reference, Names and the Qua Problem

The basic idea of causal-historical theories of names is that a term “refers to whatever it is causally linked to in a certain way” (Devitt and Sterelny 1999: 66). This means that, for example, a name such as ‘Hokusai’ refers to an object because the referring of the name to the object has been fixed via a causal grounding that consists of two stages: a certain dubbing ceremony in which agents name an object and the repeated use of the name in question by various other agents to refer to the same object. Further reference is possible due to the causal connection of those present at the baptismal ceremony with the rest of the linguistic community. According to Devitt and Sterelny, there is not one single grounding event, but a plurality of them which secure the reference: it is not in virtue of only one use of a name that the reference is grounded, but also in virtue of a subsequent relatively fixed causal chain starting from the initial dubbing.

One argument against descriptive theories of names is that an agent can refer successfully when using a name even in the absence of identifying beliefs about the object of reference.1 A further issue is whether the agent needs to have any true belief about the object at all the difference stages of the process of reference. According to Devitt and Sterelny, a complete rejection of any descriptive element from the grounding ceremony is too extreme and leads to the so called qua problem. They note that during a dubbing ceremony, we can be perceptually acquainted not with the whole object we want to name but only with a proper part of it and still ground the term successfully. For example, it may be that we perceive only the head of a cat we want to name because the rest of its body is somehow concealed. How is it possible that, despite our lack of a complete causal connection with the whole animal, we successfully name the cat as a spatial and temporal whole? Devitt and Sterelny’s solution is that the agent who is performing the dubbing must have the intention of naming the cat as if it were a member of the category ‘cats’, thus as a spatial and\or temporal whole. In other words, the agent responsible for the grounding of the name seems to think of the object as being part of a “categorial term like ‘animal’ or ‘material object’” (Devitt and Sterelny 1999: 80) while fixing the reference. According to them, the reference of a term is grounded only to the extent that the object of experience fits the categorical term under which the agent has conceptualized it. Of course, the categorial term does not actually identify the object; in particular, it does not distinguish it from other members of the same sort.

3 Thomasson’s Solution to the Qua Problem

From the previous discussion, it seems that to successfully disambiguate the reference of a name (or other general terms), the grounding process should involve the association of a sortal or categorical concept to the term in question. These ‘frame-level application conditions’, as Thomasson calls them, allow us to determine whether a reference succeeds or not and, at the same time, when we are justified in applying the same term again.

The conditions of application Thomasson has in mind should not be taken as strictly determined in all possible cases. Some of these conditions, in fact, can be more or less vague or even not completely specifiable in linguistic form. These categorical terms, as Thomasson suggests, can be thought of as “highly general sortal terms” (Thomasson 2007: 41). In addition, these categorical application conditions are interrelated: from the fact that ‘Hokusai’ successfully refers to a man, we can also infer that other sortals, such as ‘human being’, are instantiated.2 The main reason for introducing these conditions of application, is to solve the qua problem: the name ‘Hokusai’ is grounded and refers successfully because, in the grounding process, the agents responsible for the naming of Hokusai implicitly intended to apply the name to an entity qua human being. Given that the entity in question existed (and was a human being), the reference was successfully grounded. Yet there is a further issue as to whether other possible language users with different categorical intentions refer successfully when using the term grounded by others. Thomasson is neutral on this issue. What is important for Thomasson’s meta-ontology is that those agents or speakers who are competent users of the term can grasp its analytic entailments: only those who correctly grasp the framework application conditions of the term at the moment of grounding, that is, those at the end of a reliable causal chain of reference borrowing, are justified in drawing philosophically interesting consequences from them, such as existence conditions for objects. In the next section, I discuss how Thomasson’s solution to the qua problem is related to questions of existence.

4 From Reference to Existence

According to Thomasson, the conditions of application that are necessary to ground reference play another important role, in addition to helping with the qua problem. Starting from the idea that reference and existence are strictly connected, she maintains that these conditions also specify the truth-conditions of claims of existence. She argues that, for things of a general kind K, Ks exist iff ‘K’ refers (provided that ‘K’ does not change its meaning).3 The conditions of application that are necessary to ground the reference of a term determine the existence conditions for the object (or kinds) of reference. In other terms, the conditions of application of ‘Hokusai’ are principles that determine when the term can be applied. But given that Hokusai exists iff ‘Hokusai’ refers, the existence conditions of Hokusai are determined correspondingly by the conditions of application of ‘Hokusai’. Thomasson further claims that a sound methodological principle to establish whether an entity exists is to clarify the conditions of application of the term used to refer to the entity and, of course, verify whether these conditions are actually satisfied.4 This approach is elegant and has some advantages. For example, in addition to providing a methodology for dealing with existence claims, we are not required to think that ‘exist’ is ambiguous, or that ‘Kx exists’ and ‘∃x(Kx)’ have different meanings.5

Our original question was about how we should understand claims of existence. Thomasson’s answer is that, if N is a proper name (or K a kind term) that has been used in predicative statements with the intention to refer to some individual (or kind) of category C, then ‘N exists’ (or ‘Ks exist’) is true if and only if the history of those uses lead us back to a grounding in which the application conditions are satisfied.6 For example, the claim ‘Barak Obama exists’ is true iff the conditions of application of ‘Barak Obama’ are satisfied in the grounding of the name. According to Thomasson, metaphysical enquiries about conditions of existence should begin with a conceptual analysis of the rules of applications of referring terms. The role of the metaphysician is to make explicit the rules we already master in learning to use referring expressions. Philosophical disputes about questions of existence are answerable only to the extent that application conditions related to the debated terms are found.7

5 Grounding, Reference Fixing and Application Conditions

Thomasson’s characterization of the grounding process is still not completely satisfactory. For instance, Devitt and Sterelny report another problem for simple causal theories of reference fixing that are based on the idea of a dubbing ceremony. This problem is based on some remarks put forward by Gareth Evans in which reference does not seem to be always due to a single ceremony or baptism.8 For example, the name ‘Madagascar’ now refers to a large African island. However, the object of reference seems to have shifted from the original baptism (assuming that something of the sort ever occurred): originally it was the name of just one portion of the mainland and our use derives from a misunderstanding on the part of Marco Polo. This mistake led to a different causal chain that resulted in ‘Madagascar’ referring to an island instead of a portion of the African mainland. One of Evans’s main complaints against the original formulation of the causal theory of reference is that it does not take into account the “general principles of contextual disambiguation” (Evans 1973: 195) necessary to ground the reference. According to him, it is not just the qua problem that is involved here. It seems that, in at least some cases, such as in the Madagascar example, the reference is simply not fixed by a single ceremony. Another example reported by Evans is particularly interesting for our discussion. We are told by Chambers’ Arthur of Britain that Arthur had a son, Anir. However, the legend seems to have confused the name of the son’s burial place with the name of the son. Thus we had a reference shift involving entities with presumably different existence conditions (a person and a place). These examples show that we should improve Thomasson’s formulation of the grounding process. Devitt and Sterelny, drawing from Devitt (1974) and (1981), propose the notion of multiple grounding. The idea is that the burden of linking a name to its referent is not carried entirely by the initial ceremony: in addition, the subsequent uses of the name in the linguistic community play a relevant and decisive role in fixing the reference, not only in transmitting it. However, this can also result in shifts in reference, as in the Madagascar example.

The possibility of ontological shifts has disastrous consequences for the present form of Thomasson’s account. Suppose that a supporter of Obama thinks that the current president of the USA is quite literally an angelic being. If we accept Thomasson’s theory, then it seems that we should take the claim ‘Obama exists’ uttered by this agent as false: the application conditions of the term ‘Obama’ are not satisfied for Obama, the real spatio-temporal entity, does not satisfy the conditions of existence of angels. Therefore, it follows that we should maintain that Obama does not exist. One possible answer to this objection, which has been originally raised by Sainsbury (2010), is that the conditions of reference (and thus of existence) are fixed at the dubbing ceremony and that these conditions are what is relevant for reference or, at least, for the truth-conditions of claims of existence: Obama exists iff the term ‘Obama’ satisfies those conditions fixed at the dubbing ceremony in which, presumably, his name was intended to refer to a human being. In this way the subsequent intentions of those making mistakes in the metaphysical classification of an entity do not imply mistakes in the truth values and truth-conditions of a claim of existence. However, this solution does not take into account the aforementioned phenomenon of reference change: which grounding should count as relevant for the attribution of truth-conditions for an existence claim? It seems that there are cases in which the grounding of a term is more similar to a continuous process, as in the case of ‘Madagascar’. Suppose that at some point the portion of mainland which was originally referred to by the name ‘Madagascar’ disappears, leaving intact Madagascar, the island. The claim ‘Madagascar exists’ would not presumably change its truth value, despite the fact that the original grounding object no longer exists.9 This suggests that the application conditions to be taken into account for determining the truth-conditions of claims of existence cannot be limited to those intended at the dubbing ceremony. Thomasson’s conditions for treating claims of existence should be thus modified: If N is a proper name that has been used in predicative statements with the intention to refer to some individual of category C, then ‘N does not exist’ is true if and only if the history of those uses does not lead back to a multiple grounding in which the application conditions associated with C are met.

There is still room for being sceptical about this solution: it is not clear when we are entitled to assume that the multiple grounding of a term is stable enough to provide conditions of application that are relevant to existence claims. After all, as in cases reported by Stanford and Kitcher (2000), it seems that the application conditions of natural kind terms are also open to revision and continuous assessment. These doubts stem from the fact that the conditions of application of terms do not seem to have the stability that is usually required for certain metaphysical claims, such as those regarding the existence conditions of objects. As a response, it may be argued that issues pertaining to the existence of objects are inherently and radically bounded to the language in which they are expressed. However, this solution seems to lead to a sort of idealism grounded in language that does not seem appealing. The problem of finding criteria for determining whether a reference is constant enough to justify existence claims remains to be solved.

Another worry for Thomasson’s account, even emended with the inclusion of the notion of multiple grounding and supposing that somehow we can determine whether reference is stable, is that the conditions of application are not strict enough to allow, via analytic entailments or conceptual analysis, interesting conclusions to be drawn about existence conditions.10 For example, Japanese emperors were considered as divinities before the so-called Humanity Declaration (人間宣言 Ningen-sengen), in which the Emperor of Japan, Hirohito denied that he was a living god. Up to January 1, 1946, the reference conditions for the ‘Emperor Shōwa (Hirohito)’ in the linguistic community in which the term was grounded allowed the successful reference of ‘Hirohito’ to the Emperor. After the declaration, the conditions of application still allowed speakers to refer to the entity in question. In one case, the linguistic community was supposed to refer to a divinity; in the subsequent case, to a human being. Even though the recognised ontological category of the entity has changed, we should not take claims that Hirohito did not exist before the 1946 Declaration true: it would simply be false that, in 1945, Hirohito did not exist. However, this is what Thomasson’s account seems to imply: if we take it to be true that the multiple grounding of a term has fixed the conditions of application of ‘Hirohito’ at January 1, 1946, then before this date, the conditions of application of the term were not satisfied at that time. It seems that the best account of what happened in the linguistic community is that the initial grounding conditions of the term ‘Hirohito’ were loose enough to allow substantial shifts in the perceived ontological classification of the designated object. In contrast to the case of ‘Madagascar’, the shift has not involved any change in the object(s) of reference. The point here is not related to the possibility of a shift in reference or the phenomenon of multiple grounding, but to the fact that the conditions of application seem to be loose enough to allow reference to one and the same entity even though the recognised ontological category of the entity in question shifts. This seems to imply that the conditions of application of terms are not strict enough to allow interesting conclusions about metaphysical issues, in particular regarding existence conditions. We must take it for granted that even though the conditions of application for ‘Hirohito’ were mistakenly taken as those of a divinity, the name ‘Hirohito’ was used successfully to refer to the Emperor of Japan, before and after the Declaration. It seems that Thomasson cannot just remain neutral as to whether or not references of names grounded in wrong metaphysical categories or subsequent uses of a term performed by agents with a mistaken ontological classification are successful. It seems highly counter-intuitive to maintain that the name ‘Hirohito’, when used by Japanese before the 1946 Declaration, did not refer to the Emperor of Japan. The most natural solution seems to be that Japanese used the name successfully to refer to their Emperor, despite the ontological misclassification. If the conditions of application can be loose enough to allow reference, they cannot be, at the same time, strict enough to allow us to draw interesting conclusions about the sortal or the category instantiated by the entity in question and/or its existence conditions. Thomasson seems to face a dilemma: either she must abandon the idea that claims about Hirohito’s existence were true even in 1945, which is absurd, or she must concede that the conditions of application for terms are not strict enough to support interesting metaphysical conclusions about the nature of the named entities.

6 Conclusions

This paper has presented two problems for the otherwise appealing theory about the methodology of investigating existence claims put forward by Amie Thomasson. According to her, the conditions of application that are necessary to ground the reference of a term on an object also specify, after conceptual analysis, the truth-conditions for claims of existence involving the object in question. Against this, I have argued that the theory should introduce the notion of multiple grounding and that, even after this modification, the conditions of reference are still too loose to draw interesting consequences about metaphysical issues. The term ‘Hirohito’ was supposed to refer to a divinity, thus not to an entity of the same kind as the rest of us. The reference was successful, so we must allow that the conditions of application were loose enough to allow this kind of ontological mistake in the grounding ceremony of the term. Certain loose conditions of application were applied to the multiple grounding of ‘Hirohito’; for example, the term was intended to name an entity through all its spatial and temporal parts (and thus the qua problem does not arise). However, it seems that these conditions successfully dealing with the qua problem were loose enough to guarantee reference in cases of ontological shifts of the referent. Yet if the conditions of application are so loose, it seems that we are no longer justified in inferring from them conclusions about existence conditions.11


Not everybody would agree with this formulation. See Sainsbury (2005: 1–47) for a short history of theories of names.


This entailment is analytic. Thomasson maintains that analytic interrelations can hold in virtue of various factors, including rules “explicitly stipulated or just implicitly established.” (Thomasson 2007: 37).


Thomasson (2008: 65–66). Among others, Colin McGinn defends precisely the opposite view: “To exist is to have a property that only some of things we refer have” (McGinn 2000: 16). Similarly, Sainsbury (2005) defends the idea that there can be reference without referents.


Even though I have used names to introduce Thomasson’s theory, the same account can be applied to nouns like ‘tables’ and ‘chairs’. For example, the existence conditions of tables and chairs are those conditions of application necessary to meet the qua problem for expressions like ‘this table’ or ‘that chair’.


According to Lewis, for example, there is no difference in meaning between a claim such that some things are of a certain kind, e.g. “Some things are donkeys” and a claim such that things of that kind exist, e.g. “Donkeys exist”. See Lewis Lewis (1990) for discussion.


Paraphrase of Thomasson (2007: 48). See also Thomasson (2003) for a discussion of the connection between realism and human kinds.


This view is extensively defended in Thomasson (2009a) and (b).


See Evans (1973) and (1982).


One hidden premise of this argument is that we should not change the current use of the name ‘Madagascar’.


A set of conditions of application is strict, in the sense employed here, in case these conditions determine stable existence conditions.


See Schaffer (2009) for other criticisms of Thomasson’s theory.



This work was supported by Hankuk University of Foreign Studies Research Fund of 2011. Many thanks to Paisley Livingston and Amie Thomasson for comments on previous versions of this paper.

Copyright information

© Springer Science+Business Media B.V. 2011