Does it Make Sense to Ascribe Object-Dependent Senses to Names?

Introduction

Some philosophers of language1 [1, 2, 3] believe that the neo-Fregean conception of object-dependent senses2 [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17] cannot and should not be attributed to Frege3 [18, 19, 20, 21]. More importantly, that it does not make sense and it is not of any use in developing a systematic theory of meaning. More precisely, that it is not of any use in providing a theory of meaning for singular terms4. More conclusive, that without any doubt Conceptual Paper it does not make sense to ascribe object-dependent senses to names. To resist the first contention requires providing substantive textual evidence. A difficult and valuable work I am neither in the position to do nor interested in. Here I will resist the subsequent contentions. More precisely, I would like, firstly, to review how the neo-Fregean characterization of senses is available for us to steer between a descriptive and a purely referential conception of meanings. Secondly, to revise how it is possible to ascribe object-dependent senses to singular terms, specifically, to indexicals like ‘here’ and ‘now’. Thirdly, to show that there is nothing in principle to prevent us from ascribing object-dependent senses to names. That ascription depends upon a theoretical revision about how to set the limits of cases of singular reference. Only under a theoretical carved framework would it make sense to enquire if names and descriptions belong to the category of genuine devices used to refer to particulars [16, 14, 22, 23]5.

category.

5 Notice that the notion of singular term is different from the notion of singular reference. Whereas the first one corresponds to a syntactical My main aim is to show under what characterization of senses would it be possible to ascribe senses to names, and to show that that characterization is compatible with the conception of names as rigid designators, and as devices of direct reference.

In order to state clearly what is at stake here, it is useful to recast the previous issues into the following questions: Where does the difficulty in ascribing object-dependent senses to names reside? Does it reside in the idea of considering singular terms as devices whose use depends on the assignment of a semantic value? Does it reside in ascribing a dependency on referents or on the existence of the object referred? Does it reside in the idea that referring is not a relation mediated by senses? Does it reside in a cautious attitude against intensional notions and a preference for extensional reductionist accounts of meanings, an attitude aligned with Quine’s [22] famous dictum “intentions are creatures of darkness, and I shall rejoice with the reader when they are exorcised”? Or does it reside in the idea of extending the boundaries of object-dependent senses to provide an analysis not just for the meaning of essential indexicals and demonstratives [1, 2]6 but also for the meaning of names, and maybe also for the meaning of descriptions?

If the difficulty is with the notion of referent-dependency or reference-dependency [4, 21]7, we are moving away from Russell’s [24, 25] and Kripke’s [26, 27, 28] conception. Since Kripke thought about names as rigid designators and Russell thought about logically proper singular propositions as constituted by bare objects. According to Russell, a singular proposition is more than referent-dependent, it is completely (grammatical) category the second one corresponds to a semantical one. The delimitation of the grammatical category depends on the delimitation of the semantic one, and not the other way round. We could use definite descriptions to refer, as in Donnellan’s (1966) famous distinction between attributive and referential uses of definite description. To see also how it is possible to refer to particulars using definite descriptions see Cussins (1999). We could even use names to describe things as in the following case “you know, Trump is Trump”. “The Age of Enlightenment” can be used as a description and also as a name. Linguistic devices can be used in a very liberal way, and the way they are used depend on our standards and our more entrenched practices and institutions. For what follows it is better to bear in mind McDowell’s following remark: “the taxonomy for varieties of thought that Evans offers need not correlate straightforwardly with any taxonomy of their linguistic expressions.” (McDowell, 1990, p. 259).

6 See Perry (1977, 1993).

7 It worths noting that Evans (1982, pp. 8-10), following Dummett (1978, pp. 116-144), makes a distinction between the ascription of semantic values or references -that is, the ascription of a role to an expression relative to its contribution to the truth-values of the sentences in which it appears-, and the assignment of referents -that is, to assign an object or a feature in the world (or in a possible worlds) as the meaning of the expression-. The second notion as opposed to the first necessarily entails some ontological commitments. Specifically, the postulation of objects or existing entities as the things that determine the truth-values of the sentences in which the term occurs.

constituted by the objects we talk about, not by conceptual intermediaries. It does not seem a good idea to step back from the model of referent-dependency when we are talking about singular reference. If referent-independency is an option here, then surely their advocates face a deeper problem: how it is possible that our thoughts touch reality, so to speak, if at the bottom nodes of singular thinking we only find conceptual intermediaries and object-independency. The notion of referent-independent identification is so contrary to the notion of singular reference that it does not deserve to be explored in this context.

If the difficulty is with the ascription of senses to singular terms then, we move backwards from Frege’s conception: since Frege ascribed senses to every non-defective use of an expression at the indicative mode8. Now, although it seems useful to retain the notion of referent-dependency in a research about the limits of singular thinking, it does not seem to me a good idea to retain the Russellian model of referent-constituency. If we excise the notion of sense and pursue the idea of extensional reductions, some other problems emerge. How do we deal with the normative dimension of our thinking? Do we just eliminate that dimension? How is it possible to derive the normative (and intensional) dimension of our thinking from a natural (and extensional) reconstruction of the notion of thought?.

Frege developed the notion of sense to deal with differences in cognitive meaning when expressions co-refer. The postulation of senses provided for him a way to explain how speakers’ rationality is preserved in cases where the purely referential or purely extensional specification of beliefs would entail the attribution of an inconsistency. If ‘Ch’aska Qoyllur’ and ‘Choque Chinchay’9 are co-referential terms, then ‘Ch’aska Qoyllur is Ch’aska Qoyllur’ and ‘Ch’aska Qoyllur is Choque Chinchay’ are true sentences. If meanings are exhausted by the referents we talk about, then to understand ‘Ch’aska Qoyllur is Choque Chinchay’ implies to know that ‘Ch’aska Qoyllur is Choque Chinchay’ is true. Some people, however, can understand ‘Ch’aska Qoyllur is Choque Chincha’ without knowing that it is true, and develop a conception of two opposite ways to conceive life10.

If there is no room for a notion that explain differences in cognitive significance and how it is possible not to be

8 See Frege (1980, p.63).

9  ‘Ch’aska Qoyllur’ and ‘Choque Chinchay’ are the names for Hesperus and Phosphorus given by the Incas.

10 As the Incas seemed to do. The example is useful also because a semantic theorist cannot know in advance if the expressions correspond to names or to descriptions. He should then remain neutral about the possibilities for the expressions to express singular senses. That is, the issue is not decided by a grammatical distinction between proper names and definite descriptions.

inconsistent when having different attitudes about the same facts, then how it would be possible to understand sentences like “Ch’aska Qoyllur is Choque Chinchay” without knowing that the sentence is true? Although our thoughts touch reality, there is a certain distance too, so to speak. Contrary to the purely Russellian and to the Quinean radical extensional attitude I will not presume that all intensional dimensions of meaning should be excised. To argue in favor of the necessity to include intensional notions in a theory of meaning, and against the possibilities to eliminate or to reduce the normative dimension of singular thoughts, will be too much for this paper. Hereinafter then, I will not consider deeper arguments against the postulation of modes of presentation, intensions or Fregean senses. What I would like to consider is the neo-Fregean promise of providing a unifying notion whose purpose is to deal at the same time with the normative and realist dimensions intrinsic to the expression ‘thinking about things’11 [10].

The Descriptivist Canonical Review

Kepa Korta and John Perry (p. 15) [3] claim that there are only two models -exhaustive and mutually exclusive- for an account of the meaning of names: the model of direct reference, and the model of reference by description. They also claim that Frege and Russell did not conceive names as linguistic devices of direct reference. According to them, Frege and Russell associated wrongly definite descriptions to names. They tell us that ‘[Frege and Russell] thought that ordinary proper names are not, as Mill thought, simply tags for objects, but instead or in addition incorporated something like abbreviations for descriptions.’ (p. 17 [3], square brackets are mine).

From this point of view, Fregean accounts of singular thinking are descriptivists and on the lines of object- independent identification. Russellian accounts are referentialist and on the lines of an object-constituent view. Korta and Perry consider that Frege’s assignments of senses to proper names are equivalent or incorporate the ascription of definite descriptions (or a cluster of definite descriptions). So that, it would be perfectly possible to admit cases of appropriate uses of a name that are not tied to the existence of the referent, since the use of a definite description is not tied to the existence of a referent. They also consider that Russellian propositions -general or singular- are not constituted by senses, and that sentences involving proper names are not appropriate to express singular propositions since proper names mask definite descriptions.

To be fair, Russell advanced two theories of names: a descriptivist conception about ordinary names and a

11 See McDowell (2005, pp. 64-65).

Millian one about logically proper singular terms. For Russell, ordinary names do not refer to singulars, they are independent of the existence of the object and they do not have a sense. Sentences containing names are equivalent to existential quantified sentences. Only logical proper singular terms -expressions delimited by the cognitive relation of acquaintance- refer to singulars and depend on the existence of the objects referred. Logical singular terms, however, do not have a sense.

The previous lines correspond to a canonical reconstruction of Frege and Russell. Recanati’s reconstruction follows the same ideas. He tell us that ‘Russell’s insistence on acquaintance and direct reference led him to reject Frege’s sense/reference distinction, on the grounds that, if reference is mediated by sense, we lose the idea of direct acquaintance and succumb to Descriptivism (Hylton 2005)’ (p. 143) [29]. In other words, for Recanati, Russell discarded the notion of Fregean senses as mediating entities because there were available only two alternatives: the model of direct acquaintance and the descriptivist model.

Within that exclusive and exhaustive framework Kripke [26, 27] is canonically understood as arguing against the descriptivist theory, and in favor of a conception of names as devices of direct reference. Whereas names behave rigidly under modal operators, descriptions behave differently. Whereas names denote objects, definite descriptions are satisfied by objects. Sentences containing descriptions about an object can be substituted by sentences containing the name of the object in opaque contexts without truth- preservation. Therefore sentences containing descriptions and sentences containing names can be co-referential without being logically equivalent. Thus, senses of names are not equivalent to senses of definite descriptions.

Since under the previous reconstruction we have available only two models of singular reference (direct reference or reference by description), it would then better to abandon the ascription of senses to names. As I said before I will not consider deeper arguments against the postulation of intensional notions. What I will consider is how to counteract the previous contention.

The Neo-Fregean Review

Let us contrast the canonical descriptivist review with Evans’s and McDowell’s reconstruction of Frege and Russell. Evans’s main point was that Frege can be better understood as making room for the notion of Russellian Singular Terms, or at least that a Fregean can accommodate his notion of sense to the conception of Russellian Singular Terms. Evans’s main point was not that Russell or Kripke are better understood as making room for the notion of sense, nor that a Russellian can or would accommodate his notion of object- constituency to the Fregean conception of senses. McDowell and Evans conceived how it is possible to throw out the descriptivist conception of senses and persevere into the equation between the three relevant roles that Frege ascribed to his unified notion of thought: (1) what is said by uttering a sentence, (2) the appropriate vehicle of truth-values, and (3) the content of propositional attitudes [1]12. They restated two Fregean conditions that disappeared in the canonical review: Firstly, every ascription of senses depends upon the expression having a reference (a semantic value). Secondly, for the case of singular thoughts, every ascription of singular senses depends upon having a referent (a object). Those two statements are equivalent to the following claim: although singular thoughts are not object-constituted, they are still object-dependent.

The previous conception can be unpacked in three sub- thesis. Firstly, assignments of senses to terms are dependent on assignments of references or semantic values. Secondly, singular senses are not independent of the existence of the referent. Thirdly, Fregean semantics assign singular senses to (genuine) singular terms and not bare referents. It does not seem too problematic to accept the first two claims. That there is at least nothing to prevent Frege to accept Russellian Singular Terms understood as object-dependent -and not as object-constituted- expressions is plainly obvious by recognizing, firstly, that Frege’s developed his Conceptual Notation in order to apply it to mathematical claims, and strongly demanded that every name introduced ought to be mapped to one and only one Bedeutung. Secondly, that the Fregean notion of Bedeutung is a realist one: Frege demanded that the references of singular thoughts were actual objects -the things we talk about-, and not merely semantic values -items defined exclusively by their contribution to the truth- value of the thought expressed. Consequently, Frege assigned ‘the things we talk about’ as the referents of singular terms, and truth-values -objects in his conception- as the referents of singular sentences. Even if Evans’s conception of Fregean thoughts in contexts of fiction as mock-thoughts -and correspondingly about Fregean senses ascribed to names in contexts of fiction as defective- were not accurate and were not supported by textual evidence [4]13, it is plainly obvious that Frege’s theory is operative -and is designed to be operative- only in cases in which every singular term is mapped onto one (and only one) referent.

The controversy is more about the third claim. The basis to claim that object-dependent senses do not make sense

12 For Perry (1977) it is better to keep the three roles apart. He argues that there is no unified notion that can satisfy at the same time those three roles.

13 See Evans (1982, pp. 22-30).

consists mainly in a difficulty to accept a synthesis between a Kripkean (or Russellian) model of direct reference and rigid designation, and a putative Fregean model of senses as descriptive or conceptual intermediaries. Whereas Kripke denied systematically the necessity to ascribe senses to names supporting a direct theory of reference, Frege admitted -at least in some passages- that a name can have a sense without a referent. Thus, it seems inconsistent to elaborate a synthesis between Kripke’s conception of names as devices of direct reference and Frege’s ascription of senses as conceptual intermediaries that allow us to explain differences in cognitive significance. However, since what Kripke denied systematically was -strictly speaking- a logical equivalence between (senses of) names and (senses of) descriptions, and Frege -strictly speaking- denied the ascription of referents to defective uses of expressions at the indicative mode; then, if we were able to provide a non- descriptive account of senses, and to explain why defective uses of expressions do not count as cases of singular reference, the alleged incompatibility would disappear [4, 5, 7, 8, 9, 11, 12, 13, 14, 15, 16, 17, 30, 31, 32, 33, 34, 35]14. Evans’s [4, 5] (pp. 7-79; pp. 291-321) proposal can be minimally presented in the following terms: Kripke’s arguments are sufficient to deny the possibility of reducing names to descriptions and therefore to deny the possibility of ascribing descriptive senses to names, but not to deny the possibility of ascribing non-descriptive senses to singular terms. The main issue is then how to provide a characterization of the notion of sense that steers between descriptivism and referentialism. Once this issue is solved we could enquire if names are appropriate members of the category of singular terms, and consequently see in what conditions to ascribe singular senses to names.

Here-Thoughts and Now-Thoughts.

Can we provide a non-descriptive characterization of the notion of sense and extend the boundaries of singular reference to cover naming cases?.

Whereas McDowell [8] (pp. 171-198) didn’t hesitate in conceiving names as members of the category of Russellian Terms, Evans [4, 5] (pp. 64-79, 371-404; pp. 1-24) was more careful in asking about the limits of such category. He conceived here-thoughts, now-thoughts, this-thoughts, and I-thoughts as paradigmatic cases of Russellian Singular Reference [4, 5] (pp. 151-170; pp. 291-321). Since, it seems

14 That is a work accomplished mainly in McDowell (1998, pp. 171-294; 2002, 211-262; [1994/2000], 24-65; ), Evans (1982, pp. 7-85, pp. 141-91; 1985, pp. 1-24,) and reviewed in Luntley (1996); Cussins (1990, 1992, 1999, 2003, 2012). See also Kelly (2001), Campbell (2002), Soutif (2012), Thornton (2019), Mertel (2017), Grush (2018). For critical revisions see Hawthorne, J., & Manley, D. (2012 pp. 71- 92) and Burge, (pp. 154-210). The literature on the neo-Fregean conception of object-dependent sense is extensive, but the previous core is enough to get an adequate idea.

not possible -Evans argued it is not- to have a thought about -for instance- a position as here, if that position does not exist (consider if it makes sense to ask “where I am?” if the position where I am does not exist [4, 36]15). But it seems possible to refer to an object by a name although the object does not actually exist (consider canonical cases about fictional characters, or about persons of the past and persons that will be born in the future). It is not a nonsense to think about Winston Smith (the character of Orwell’s novel 1984). It is meaningful for a mother to think about the child she will give birth to by his name. It is meaningful for me to think about the sailboat I will buy by the name I will give to it [35, 37]16. In the same spirit, I believe it is better not to go so quickly. It may be better to begin asking how we can provide a non- descriptive and object-dependent notion of sense to the case of here-thoughts and now-thoughts. Later we can enquire if the notion can be extended to name (and describe) things in our perceptual domain, and to name (and describe) things outside our perceptual domain [38]17.

According to the orthodox view, the meaning of indexicals is exhausted by a two-dimensional explanation. Briefly: the meaning of indexicals is explained by providing a grammatical rule that determines how to apply the indexical in all contexts of use, and a contextual rule that determines how the grammatical rule determines a referent in a particular circumstance of evaluation. Grammatical rules like the following: a token of the expression type ‘here’ denotes the place in which the utterer is located when he is making the claim. Similarly, a token of the expression type ‘now’ denotes the time at which the utterer makes the claim. In a more demanding token-reflexive style: a token of the expression type ‘here’ denotes the place in which the utterance in which the expression occurs is made. Similarly, a token of the expression type ‘now’ denotes the time at which the utterance in which the expression occurs is made. When I utter the sentence: ‘It is raining here and now’ the rules of the language demand to complete the sentence with a particular place and time in order to determine the proposition expressed. In the case in which I utter the sentence ‘It is raining here and now’ the rules of the language determine how to saturate semantically -or how to modulate pragmatically [39]18- the sentence in the

15 To provide a substantive answer to this question will require to block Putnam’s (1981, pp. 1-21) solipsistic fiction of a brain in a vat. For Evans’s answer see (1982, pp. 151-170).

16 For exploitation of cases like these to counteract a purely causal theory of singular reference, and a theory on the model of the reduction of cases of singular reference to cases of knowledge by acquaintance, see Jeshion (2010 pp. 105-140), and Hawthorne and Manley (2012, pp. 37-70).

17 On the extension from a core conception of singular reference in the perceptual domain to subsidiary cases of singular reference in the non- perceptual domain see Fodor and Pylyshyn (2015, pp.83-155).

18 On the difference between semantic processes of saturation and primary pragmatic processes of modulation see Recanati, 2010a, pp. 1-47.

context of utterance and how to determine -semantically or pragmatically- the truth-value according to the circumstance of evaluation. So that, at the end, we get a proposition like: it is raining in the place and the time where I am writing the sentence (or something close to that proposition).

The previous account leaves open the possibility of having here-thoughts and now-thought when the position and time of utterance does not exist. It is possible to determine the meaning of sentences type ‘it is raining here and now’ when locations and times are frame-dependent. That is, when places and times are determined relative to a set of coordinates independent of the places and times where the utterer is, and dependent on a frame of reference created by conventions. For instance: when George Orwell writes a token of a here-and-now-sentence in his novel 1984 -by the rules previously stated- he will be giving expression to the thought it is raining in the place and time where and when he is writing the sentence. But he is clearly not giving expression to that proposition, since it is irrelevant for understanding the sentence (and the story) what is happening when and where Orwell is. The place in, and the time at, which the utterance is specified is relative to the frame provided by the novel. If that is so, the thought expressed would be -according to the rules- let’s say: it is raining in the place and time where and when Winston Smith is uttering the sentence. Lets say, it is raining in Oceania on May 23 of 1984. Characters of fiction would then be able to entertain here-and-now-thoughts. But fictional characters are fictional, they do not have thoughts about the places and times they are in, since they are not in places and times. They are in places and times only in a derivative sense relative to how we are situated in our spatiotemporal surroundings. I can understand what is for Winston Smith to be placed there (in Oceania), and at that time (May 23 of 1984) because I know what is for me to occupy a place and to live upon a period of time. I do not even need to know how places and times are named according to geopolitical and chronological coordinates in order to know where and when I am here and now. I do not need to know that I am in a place named São Paulo and at a time named wednesday, May 23, 18:24:19 GMT-3, to understand a here-and-now- sentence. I just need to vaguely (and adjustable) know that I am occupying a patch of space and living through a stretch of time where my body with its complex sensorimotor connections and its possibilities of action and perceptions is.

In order to have a thought about a location as here, and about a time as now, Evans demanded an intertwined set of conditions on the lines of the previously introduced ideas. This set of intertwined conditions do not leave room to frame-dependent and referent-independent identification, and allow us to state some differences on modes of presentation in cases of singular reference. Those conditions open the possibility of considering different varieties of singular identification that orthogonally cut grammatical distinctions, so that it would be possible to refer to particulars using pure indexicals but also using definite descriptions. I will minimally state Evans’s intertwined conditions in the following three formulas: Firstly, the Dynamic Fregean Thoughts condition: our egocentric representations of the spatial content of here-thoughts form part of a dynamic system: [… here will be F in a few moment, here is F now, here was F a moment ago…]19 (hereinafter, for simplicity, I will avoid the qualifications ‘in a few moment’ and ‘a moment ago’). The significance of here- thoughts is dependent on being part of that system, so that ‘here’ and ‘there’ items cannot be treated in an isolated way. The meaning of these series are clarified when Evans introduces the notion of a Dynamic Fregean Thought (pp. 306-311) [5].

Secondly, the non-conceptual information-based states’s condition: the significance of egocentric spatial terms is dependent on the subject’s dispositions to act in some way or other, and direct his rational plans based on sensorimotor and information-based states. The conceptual simplicity of our here-ideas is based on unmediated dispositions (non- mediated by a concept or by a calculation at a personal level) to locate our egocentric representations of the spatial content of our perceptions in objective positions independent of our actual and dispositional states. To put it in a terminology that remains neutral to the introduction of the notion of non- conceptual contents: Egocentric spatial terms are the terms in which the content of our spatial experiences would be formulated, and those in which our immediate behavioural plans would be expressed. This duality is no coincidence: an egocentric space can exist only for an animal in which a complex network of connections exists between perceptual input and behavioural output (p. 154) [4].

Thirdly, an objectivity condition: the ability to entertain series: [… here was F, here is F, here will be F …]20 is dependent on the ability to locate those series in a public conception of how the referents are disposed relative to each other. That corresponds briefly to the following point: consider that I think here is F and there if G. If my conception of the F’s here and the G’s there is public and objective, then if I move from here to there, I would consider that here is G and there is F. The heres and theres have public significance, and not just an egocentric significance. They are disposed simultaneously in

19 I use the square brackets to mark the sequence of sentence/thoughts between the brackets as a unified system: a persistent thought.

20 As I said before, for simplicity, I will drop the qualifications ‘in a few moment’ and ‘a moment ago’.

some relations satisfied between them. I am able to locate the same place at which …is F as here, and as there and I am able to locate the same place at which …is G as here and as there, because there is some relation between the heres and theres where F and G are that stands independently of its location relative to me.

The objectivity condition can be traced back to Strawsons’s (pp. 31-38) [40] argument in favor of a single-unified spatio- temporal system required as a precondition to identify and re-identify particulars, and Strawsons’s (pp. 59-86) [40] argument in favor of the postulation of a pseudo-space as a precondition to make sense of a non-solipsistic mind -i.e., a mind that is able to distinguish between the objects at which his states are directed, and his states and dispositions. The postulation of a pseudo-space is criticized by Evans (pp. 249-290) [5] but on the ground of being an insufficient conception. Not on the basis of being a completely deviated conception. Since the Strawsonian notion of pseudo-space is susceptible to be prey of a phenomenological reduction on the line of states that are dispositionally accessible to the subject, Evans (pp. 249-290) [5] argued that a more substantial and categorical notion of space is required in order to make sense of mind-independency and objectivity. The significance of this condition is clarified when Evans introduces what he calls the Generality Constraint (pp. 101-105) [4], and his proposal of an account that provides substance for Russell’s Principle and satisfy the constraint -what he calls, the Fundamental Level of Thought (pp. 105- 112) [4].

To sum up: here-thoughts and now-thoughts form part of a situated and dynamic system of objective identification. Whereas the sense ascribed to ‘now’ depends on our ability to keep track of a time at an egocentric system, the sense ascribed to ‘here’ depends on our ability to keep track of a position at an egocentric system (p. 153) [4]. At the same time the significance of ‘now’ and ‘here’ is dependent on being able to conceive that those terms can refer to locations that are independent of the subject’s dispositions. The possibility of thinking it is hot here is intrinsically related with the possibility of thinking it is hot there, in that place I was thinking some instants ago as here, and also to the possibility of thinking it will be hot there in that place where I will be thinking as here some instants later. The possibility of thinking it is raining now is intrinsically related with the possibility of thinking It was raining, sometimes ago in that time I was thinking as now; and also is intrinsically related with the possibility of thinking it will be raining there in that place where I was thinking it is raining a moment ago. The meaning of the thoughts here and now is F, there and some instants ago were G, over-there and some instants later will be H is sustained by a dynamical and situated field of our bodily spatiotemporal sensorimotor possibilities: [… here was some instants ago F, here is F, here will be F some instants later ...but if I move over there where it is G: there (was/is/will be) F, here (was/is/will be) G]. So that, the meaning of a thought grasped in one occasion is dependent on the possibility to contrast it to thoughts graspable at other occasions in a system of thoughts. That is, the possibility to have a thought in a particular occasion makes sense only as part of a logical space of related possible thoughts entertained in other occasions.

Generally speaking, our ability to entertain here- thoughts and now-thoughts is sustained by an unmediated and dynamic ability to locate positions whose significance is derived because they are part of a serial and egocentric system at the same positions that derive its significance because they are part of a simultaneous and allocentric system [4, 5]21. By egocentric identification, I mean a way to identify the referent centered on a bodily agent. By a bodily agent, I mean a subject that has a body and is disposed to act in some way while he receives information from his perceptual states. By dynamic, I mean that we are able to maintain the same way we refer to an object under changing circumstances because the stability of the sense is sustained by keeping track the same referent under changing conditions (keep moving to keep stability). By unmediated, I mean that reference to the object does not need to be mediated by a concept or an inference or a calculation at the personal level. By allocentric, I mean a system of coordinates that do not depend to (or is indifferent to) the position of the subject. That is, although we can give a special role to one position in that system by including a here-pointer, the significance of the here-pointer is not reduced to a location determined by a system of axes based on the natural orientations of the agent’s body. Even if the special role given to the here-pointer is useful to align the heres whose significance are egocentrically determined to positions that are allocentric determined, the here-pointers gain significance in virtue of being part of a field of the agent’s possibilities of action and perception.

Once we appreciate that Evans’s conditions are intertwined the point about object-dependency follows easily: it is not possible to have a here-thought, if the subject is not able to locate the position he is thinking about in a location that is not reducible to conception of space explained in terms of the subject’s states and dispositions. The location of places in a public conception of space is a precondition to have a here-thought, so that to presuppose the independency of those places relative to the mind -therefore to assume its existence- is required to make reference to a position by saying ‘here is F’.

The point about non-descriptivist senses can be minimally

21 See Evans, 1982, pp. 151-170; 1985, pp. 249-290.

stated now as follows: it is neither necessary nor possible to reduce the content of ‘here is hot’ to the content of ‘the place where I am is hot’, we can and should explain the significance of those terms without providing an account of the significance of here-senses in terms of a description. Because, between other reasons, the significance of here and now senses in terms of a description simply passes by the intrinsic motivational, bodily and active character of those ways of cognition.

Aphla and Ateb-Thoughts

Let me now reconsider the following quote from Frege in order to return to the cases of naming …let us suppose an explorer traveling in an unexplored country sees a high snow-capped mountain on the northern horizon. By making inquiries among the natives he learns that its name is ‘Aphla’. By sighting it from different points he determines its position as exactly possible, enters it in a map, and writes in his diary: ‘Aphla is at least 5000 metres high’. Another explorer sees a snow-capped mountain on the southern horizon and learns that it is called Ateb. He enters it in his map under this name. Later comparison shows that both explorers saw the same mountain. Now the content of the proposition ‘Ateb is Aphla’ is far from being a mere consequence of the principle of identity, but contains a valuable piece of geographical knowledge. What is stated in the proposition ‘Ateb is Aphla’ is certainly not the same thing as the content of the proposition ‘Ateb is Ateb’. Now if what corresponded to the name ‘Aphla’ as part of the thought was the meaning of the name and hence the mountain itself, then this would be the same in both thoughts. The thought expressed in the proposition ‘Ateb is Aphla’ would have to coincide with the one in ‘Ateb is Ateb’, which is far from being the case. What corresponds to the name ‘Ateb’ as part of the thought must therefore be different from what corresponds to the name ‘Aphla’ as part of the thought. This cannot therefore be the meaning which is the same for both names, but must be something which is different in the two cases, and I say accordingly that the sense of the name ‘Ateb’ is different from the sense of the name ‘Aphla’. Accordingly, the sense of the proposition ‘Ateb is at least 5000 metres high’ is also different from the sense of the proposition ‘Aphla is at least 5000 metres high’. Someone who takes the latter to be true need not therefore take the former to be true. An object can be determined in different ways, and every one of these ways of determining it can give rise to a special name, and these different names then have different senses: for it is not self-evident that it is the same object which is being determined in different ways (p. 80) [21].

According to the canonical view we have available only two models: descriptivism and referentialism. The referentialist would say: Frege’s example can be explained in a simplified way. The explorer refers to the mountain by putting the labels ‘Aphla’ and ‘Ateb’, in virtue of just seeing the object from a specific perspective and hearing the appropriate labels from natives in the communities that surround the mountain. In other words, our explorer is able to refer to the mountain just by having the appropriate causal-informational connections with it and with the native’s tags. The conventions created don’t mean anything more than being a tag attached to the object. Those tags can be interchanged for the object itself without any gain or loss of information. They are proxies of the object. It is indifferent to the explorer to use ‘Aphla’ or ‘Ateb’, since those expressions don’t mean anything by itself. So our explorer does not understand ‘Aplha is Ateb’ without knowing that it is true, and he does not learn anything by understanding ‘Aplha is Ateb’. Although the referentialist captures the senses proposed by Frege -i.e. singular, not general senses-, he does not explain the puzzling asymmetry between understanding and knowledge.

The descriptivist advocate would say something in the following lines: the sense of ‘Aphla’ is the high snow-capped mountain (seen) on the northern horizon, whereas the sense of ‘Ateb’ is the high snow-capped mountain on the (seen) southern horizon. The explorer picks out the mountain under those descriptions, so that we can express the senses of names with definite descriptions and forget about the tags created (‘Aphla’ and ‘Ateb’). Sentences involving ‘the high snow-capped mountain on the (seen) southern horizon’ and ‘the high snow-capped mountain on the (seen) northern horizon’ have different literal meanings, although they are used to refer to the same mountain. So then we can explain, on the one hand, why a speaker can understand the sentence ‘Aphla is Ateb’ without knowing its truth-value, and why he acquires knowledge by knowing the truth of the sentence. On the other hand, we can admit uses of ‘Aplha’ in meaningful sentences, even if ‘Aphla’ does not refer to an existing or a unique object. The descriptivist however would not capture the senses proposed in the quote, since the definite description the high snow-capped mountain (seen) on the northern horizon, on one hand, may not allow the explorer to refer to the mountain, and on the other, may not allow presenting the mountain by the same way in which it is presented by names. On one hand, the explorer can be able to entertain the thought that the high snow-capped mountain (seen) on the northern horizon is F without being able to pick up a unique object that satisfies the condition. On the other hand, he can believe that the high snow-capped mountain (seen) on the northern horizon is F and do not believe that Aphla is F, because he does not know that the object labelled ‘Aphla’ satisfies the description. Although the descriptivist has an explanation for the difference between understanding the sentence and knowing that it is true, he is not capturing the senses proposed by Frege.

Descriptivist Referentialism

Nowadays referentialists have proposed that although the expressions do not mean anything more than the object referred, different explorers associate to the expressions different mental information (or different clusters of information mentally stored). As they say different singular senses are explained by differences in prompted mental files [29, 37, 39, 41, 42, 43, 44, 45, 46, 47, 48]22. In this line, Recanti proposes the following account for indexical senses23: Indexical concepts, I suggest, are mental files which one opens when one stands in a certain sort of relation with some object (the referent of the file): an epistemically rewarding (ER) relation, i.e. a relation such that, when one stands in that relation to some object, one can gain information from the object through the relation. The relation establishes a channel of information between the subject and the object. The suggestion, then, is that there are mental files which are based on such relations and whose role is to store the information one gains in virtue of standing in that relation to the object.

On this view, the type/token distinction applies to mental files. Mental files are typed according to the type of ER relation they exploit. Thus the SELF file exploits the relation to oneself (viz. identity) in virtue of which one can gain information about oneself in a special way, ‘from inside’ - a way in which one can gain information about no one else (as Frege puts it). My SELF file is not the same as yours, and they refer to different persons, of course, but they belong to the same type: they are both SELF files, unified by the common ER relation it is their function to exploit. We see that the function of files -namely, informational exploitation of the relevant ER relation- plays the same role as the conventional meaning of indexicals: through their functional role, mental file types map to types of ER relations, just as, through their linguistic meaning (their character), indexical types map to types of contextual relation between token and referent (p. 1843) [43].

It seems to me that against the previous suggestions we should take into account Frege’s anti-psychologism attitude.

22 See Recanati (2009, 2010. 2012, 2013, 2016), Murez, M., & Recanati, F. (2016), Murez, et al. (2020). Ninan (2015), Onofri (2015), Jeshion (2010, pp. 105-140). According to Robin Jeshion (2010 p. 130) “thinking of individuals from mental files is constitutive of singular thought”. She maintains also that mental files are ontogenetically grounded on Object Files and FINST or visual indexes (Jeshion, 2010, pp. 130-135). Murez and Recanati (2016) argue in favor of a correlation between the Kaplanian approach to direct reference, the Russellian notion of acquaintance and the metaphorical notion of mental files, relative to the psychological notions of Object Files and FINST.

23 The literature on the notion of mental files is extensive and part of an ongoing debate. I will present here the core idea and a simple critic.

Mental files are basically clusters of content or information notionally or individually encoded, and (normatively or causally) [29, 37, 38, 39, 41, 42, 43, 44]24 related to the referent. In the explanation proposed by Recanati, mental files are related to the referent by the normative relation of knowledge by acquaintance. In other explanations, like Jeshion [37] (pp. 105-140) and Fodor and Pylyshyn [38] object files (the empirical counterpart of mental files) are linked to the object by means of perceptual indexes (or FINST25) [38, 49, 50, 51, 52, 53].

Now, consider the following: or the clusters of information/content mentally stored remain within the limits of the subjective sphere or the information is not entirely so confined. On the one hand, if the mental information/content associated with the object remains within the subjective sphere, then how can an explorer be able to understands ‘Aphla is F and Ateb is not F’ without believing by his understanding of the sentence that he is contradicting himself? By hypothesis he should be able to be transparently confronted with the two mental files. But if he has at view the two mental files then he is not able to believe ‘Aphla is F and Ateb is not F’ without knowing that it is false, noticing -by mere introspection- that these two mental files are -although different- of the same object. That explanation simply passes by Frege’s puzzle: a subject in fact at the same time does not contradict himself when he understands the sentence ‘Aphla is F and Ateb is not F’, because from his understanding of the sentence he is not able to know transparently that the sentence is false. An alternative here is to admit that the explorer is able to have in mind the two files, but still argue that he is not able to merge the two files into a unified one. That proposal looks like trying to compare two ideas -in the Fregean sense of ideas-. Two subjective notions are not comparable because, by hypothesis, they are clusters of information subjectively encoded, so there is no basis to compare them, and there is not a basis to merge them.

On the other hand, if the notion of mental files is not entirely confined to the subjective sphere, and can be extended to the normative sphere of cognitive attitudes, it

24 The qualification normatively or causally depends on the framework. For instance, for Fodor and Pylyshyn (2015) the relation is causal, for Recanati (2009, 2010. 2012, 2013, 2016) the relation is normative (or as he says, Epistemic Rewarding). Jeshion (2010 p. 105-140) remarks that mental files should be causally linked to the referent by means of visual indexes, but she also apply a normative restriction that she calls “the significance condition” (see Jeshion, 2010, p.136).

25 The notions of FINST or visual indexes and Object Files are now precise, technical and widely known terms in the area of visual attention psychology. The first two terms were developed by Pylyshyn, et al. (1978) work on Multiple Object Tracking tasks -the MOT experimental paradigm- (see Pylyshyn 2001, 2003, 2007). The second term comes from Khaneman, et al. (1992) work on Object-Specific Preview Benefit effects (OSPB). For an introduction to the notions of Object Files and FINST see Fodor (2015), Green (2017), and Pylyshyn (2001).

will be something close to the Fregean notion of sense but with different unrealistic commitments. Commitments that demand a way to close the gap between the subjective and the objective sphere. The usual way to satisfy this demand is to postulate a relation of knowledge by acquaintance. If the explanation ascribes to the cognitive attitude what Frege ascribed to the content of the attitude, then the explanation will follow the following lines: I am able to believe that Aphla is F and not believe that Ateb is F without contradiction not because my attitudes are related to different thoughts, but because I believe in some way (in the way of an Alpha- believe), and do not believe in other way (in the way of an Ateb-believe) the same proposition Venus is F. But then, in which of these ways I believe that Aphla is F and Ateb is not F? We should say that I believe in both ways the same proposition. Thus, I believe under the mental file Alpha and under the mental file Ateb: Venus is F and Venus is not F. Since not by itself an absurd proposition, that position faces the following dilemma: either by being able to access both mental files I am able to access to the referent; in which case, contrary to the hypothesis, I am contradicting myself. Or by being able to access both mental files I am not able to access to the referent; in which case, contrary to the hypothesis, that is not a case of singular reference. This last route would be available if we had no any problem in accepting that the mental file’s model of singular reference is committed to an object-independent and non-realist specification of senses. But this alternative should be considered as problematic, by mental file’s theorists own lights, because mental files are directly related to the referents by epistemic relations of acquaintance.

The trouble with the notion is that it is supposed to account for two inconsistent characteristics: mental files should be directly related to the object and encode descriptive information about the object. Briefly, they should be descriptive and relational at the same time. This paradoxical position may be the consequence of applying a varnish of Fregean theory to a core of Russellian theory. In the Kaplanian account what a competent speaker knows when he understands an expression is its conventional linguistic meaning, i.e., its character. Characters determine the reference of a expression once a context of use is provided. Names behave as rigid designators, so that their referent is the same through different contexts of use and circumstances of evaluation. Thus, the characters of ‘Aphla’ and ‘Ateb’ return as output the same referent. The object tagged ‘Aphla’ in a context of use is the same object tagged ‘Ateb’. Names as rigid designators are context insensitive. The characters return the same proposition. If we admit only the Kaplanian framework, what a competent speaker knows when he knows the meaning of the sentence ‘Aphla is F and Ateb is not F’ is the proposition Venus is and is not F. Therefore to understand the sentence ‘Aphla is F and Ateb is not F’

would demand for him to have knowledge of an inconsistent proposition. But Frege’s puzzle was that competent speakers can understand (know the meaning) of the sentence ‘Aphla is F and Ateb is not F’ without knowing that the sentence expresses a contradiction.

The proposition should be presented in different ways because the speaker can be in a different cognitive state when he is related to the sentences ‘Aphla is F’ and ‘Ateb is F’. ‘Aphla’ and ‘Ateb’ are neutral characters and return the same object in all contexts of use and circumstances of evaluation, but a competent speaker that understands the sentence ‘Aphla is F and Ateb is not F’ can be differently related with the sentence ‘Aphla is F’ and the sentence ‘Ateb is F’, so that he may believe that Aplha is F and not believe that Ateb is F. An specification of the content of the sentence ‘Aphla is F and Ateb is not F’ as venus is F under the mental file Aplha and venus is not F under the mental file Ateb will imply also that the knowledge that the speaker has in order to understand the sentence involves the knowledge of a contradiction. Since by having the two pieces of knowledge Venus is F under the mental file Aplha and venus is not F under the mental file Ateb he is referentially related to the same object and he is ascribing to it the property of being able to satisfy F and not being able to satisfy F, no matter under which descriptive information (under the mental file Aplha or under the mental file Ateb, he conceives Venus.

The mixed version of a descriptivist referentialism proposed under the framework of the mental files’s theory is just a consequence of setting the stage under the view of the canonical review. In opposition to the canonical review of Frege and Russell, however, there are not just two models of singular reference available for us: descriptivism and referentialism. Fortunately, an alternative can be read off by taking at face value Frege’s words: the explorer is able to think the mountain under some mode of presentation if he is able to locate the object from his perspective, as part of an egocentric system of perceptions that mountain at the north horizon from here and there and over there; and if he is able to locate those perspectival positions in the publicly objective space that mountain at the north horizon from here and there and over there occupy that more or least stable position in a the public space.

How to Ascribe Object-Dependent Senses to Names?

Let us remember that the generality constraint is a regulative ideal that Evans applied to the idea of ​what is thinkable [4]26. Such an ideal could not be satisfied in some

26 For the ideas expressed in this paragraph compare with Evans 1982, pp. 89-120.

concrete cases in which an individual defectively exercises his ability to formulate a thought. For instance, Evans proposed the case of a subject who has the thought of a sphere rotating on the meager basis of an occasion in which he is causally related to one or two spheres and is not able to distinguish between one sphere and the other. According to him, this case of thinking about that rotating sphere is defective. The subject tries to formulate a thought by uttering the sentence but such a formulation does not have the possibility of satisfying - not even approximately - the demand of generality, so that it constitutes only an attempt to think about a particular sphere. A failed attempt and not a case of expressing a genuine thought about that rotating sphere. This is aligned with a main point that Evans makes against a verificationist theory of meaning and against a theory that specify references exclusively in terms of appropriate causal relations. The specification in terms of verification conditions -ultimately, being at appropriate conditions to be in contact with the objects of the thought- as well as the specification exclusively in terms of appropriate causal relations do not make room to account for the possibility of each concept to be recombined with other concepts in the repertoire of concepts that the subject has mastered.

For this reason -because Evans advances this point against the need to postulate a possibility of being in contact with objects of thought in order to be able to have thoughts about objects- and contrary to Hawthorne and Manley (pp. 37-70) [35], Jeshion [37] (pp. 105, 140) and Recanati [29, 39, 41, 43], it seems to me that Evans is not a theorist -let alone a paradigmatic theorist- of the notion of acquaintance. For a theorist of knowledge by acquaintance, it is at least curious that Evans makes only five mentions of the word. The main one (pp. 64-65) [4], to criticize Strawson’s adoption of the Russellian distinction between knowledge by acquaintance and knowledge by description, and to show that the relevance of the Russellian distinction respect to the possibility of extending the limits of singular thought corresponds to the idea that at the basis of thinking about objects -at the basis of singular thinking- is the possibility of having a specific type of knowledge -discriminative knowledge- of such objects. Evans calls Russell’s principle the idea according to which the understanding of singular propositions requires to be able to discriminate the objects involved in the thought expressed. It is important to be aware that the distinction between knowledge by acquaintance and knowledge by description is not equivalent to Russell’s principle27.

27 In fact, Russell’s epistemological principle does not consist in establishing that for every proposition that is understood by a subject (general or singular), it must be possible for him to have knowledge by acquaintance of the objects about which the proposition is. What Russell states is that the speaker should possess knowledge by acquaintance of the Evans develops the previous idea -the relation between the possibility of having thoughts about objects and the requirement of cognitively discriminate one object from another- by establishing a link between the objectivity of a mental state -in the case presented by Evans, the objectivity of a mental state as specific, occasional and subjective as “a pain”- and the possibility for that state to satisfy -in at least some degree- the generality constraint [13]28. That is, the objectivity of the state depends on the possibility of recombining it with other states of the same type. He tell us the following: We should surely be reluctant to assign the content ‘I am in pain’ to an internal state of a subject unless we are persuaded that the subject possessed an idea of what is for someone -not necessarily himself- to be in pain, and unless we are persuaded that the internal state in question involved the exercise of this idea.

What we have from Strawson’s observation, then, is that any thought which we can interpret as having the content that a is F involves the exercise of an ability -knowledge of what it is for something to be F- which can be exercised in indefinitely many distinct thoughts, and would be exercised in, for instance, the thought that b is F (p.103) [4]29.

It is clear that, to the extent that they are concepts and allow the formulation of propositions, it must be possible to articulate [... this red], [... that cube], [... that hue], [ ... this pain] with other concepts. Perceptual states insofar as they are directed to independent items should be able to obey a grammar of combination -perhaps not strictly compositional- relative to other contents of the same type. In this sense, it is important to note the following: a perceptual state of the type [... appears red] -under appropriate conditions-, a neurophysiological state of the type [... irritation x in the retina] or [... activation x in the brain] can be conceived as internal states. They are states whose conditions of individuation are relative to the subject or the representational apparatus of the subject and not exclusively to what is represented. However unlike the latter, the former ones are contents available in the experience insofar as they can be articulated with other contents of the subject’s experience and judgment.

objects denoted by the constituents of the thought (e.g., the objects denoted by the constituents of a definite description), not that the speaker should be able to be in contact with the things about which is the proposition (i.e., the object that satisfies the description).

28 On degrees of satisfaction of the generality (and other extensions) see Cussins, 1992, pp. 33-45.

29 See also the context in which these claims are made and the quote from Strawson The generality constraint -i.e., the possibility of recombining a state with others of the same type- is not a reformulation of the principle of compositionality. The point is not that it is possible to compose the proposition [that cube is red] from its constituents [that cube], [... is red] and its form of composition. The point is that the specification of contents of the states [that cube ...] and [... is red] depends on those states to belong into a logical space. The sense of the thought [that cube is red] is determined by the position it occupies in that logical space as the point of intersection between the two logical series or dimensions thoughts: (1) ... [this is red], [that is red], [ that is red] ..., and (2) ... [that is red], [that is green], [that is blue] ... (p. 104) [54]30. The complete thought takes priority over the form of decomposition. Only in the context of a complete thought or a unit of complete minimal significance does it make sense to ask about the contribution or value of each constituent and the form of composition. In other words, the complete thought and its form of decomposition is determined by the dimensions of the logical space in which it appears. The contribution, meaning or value of a constituent depends on the complete thought (or the occasion of formulation of the complete thought) to which it belongs, but also on the way it appears (and the possibility of appearing) in other thoughts (or on other occasions of formulation of the complete thought). A non-descriptive object-dependent sense ascribed to the names would be more or least something like the following: being able to develop a logical space -maybe a cognitive map or mediational field- that allows the subject to locate the object as independent of his actual and dispositional sensorimotor states.

Let us return now to our mountain explorer. He may be able to compare the stable position of that mountain at the northern horizon from here and there and over there with the stable position of that mountain at the southern horizon from here and there and over there, and gain some knowledge from, say, ‘Aphla is Ateb’ even though both names refer to the same mountain. His having in mind some or other mental file of the object -if there are mental files [46]31- is irrelevant to his acquisition of knowledge in this situation. Our explorer would be able to understand ‘Aphla is F and Ateb is not F’ without knowing that it is false, because although he is able to locate ‘Aplha’-patches of the mountain and ‘Ateb’-patches of the mountain at public locations, he has not located them at the same location. The error can be corrected by revising continuously the evidence provided by his excursions around the communities that surround the mountain, but it

30 About this point is also useful to review Sellars, Science, Perception and Reality, pp. 93-94. For a critical revision see Travis, 1994.

31 Murez, et al. (2020) argue that the psychological evidence that sustains the theoretical notions of FINST and Object Files is not enough to claim that mental files are a natural class grounded on the class of Object Files.

is not necessarily represented in principle as an error. Thus, when he locates ‘Aplha’-patches of the mountain and ‘Ateb’- patches of the mountain at different public locations, but as parts of different mountains that would occupy different positions in a map, he understands the sentence ‘Aphla is F and Ateb is not F’ without knowing that it is false. Thus, our explorer may understand (know the meaning of) a sentence that relates the same object with different names under different forms of identification without knowing that it is inconsistent.

Having an alternative view that steers between the referentialist and descriptivist conception (a view that does not mixes referentialism and descriptivism), we can ask now to what extent is it possible to apply the notion of non- descriptive senses to define the boundaries of a grammatical category of singular terms? Should we remain within the limits of indexicals and demonstratives? Can we try to extend the limits of singular reference to uses of names at core scenarios of (information-based) perceptual objects? Can we extend this notion to scenarios of names applied to non- perceptually accessible objects and relative identifications (i.e., identifications of objects in virtue of descriptions that relate unperceived objects with perceived objects or even un-perceptible objects with perceptible ones)? How much can we extend the category of Russellian Singular Reference? How can we draw the limits between story-relative identifications -uses of terms whose understanding does not depend on the existence of an object but depend on frames whose meaning is derivative from our spatiotemporal frame- and object-relative identifications -uses of terms that are not frame-dependent? How do we distinguish between cases of genuine identifications of objects and cases of unwittingly lapsing into the sphere of fiction?32 Those are indeed very difficult questions whose answers demand more than what was presented here. What I can conclude, having at hand our non-descriptive notion of senses and thoughts, is that it indeed does make sense to ask which are the appropriate scenarios and conditions to ascribe object-dependent senses to names. Even more, it does make sense to ask which are the scenarios and conditions to ascribe object-dependent senses to definite descriptions [55, 56, 57, 58, 59, 60, 61, 62].

Funding

Research for this paper has benefited from the financial support of the Sao Paulo Research Foundation -FAPESP- Post-Doctoral Fellowship process 2018/04058-7.