Files and Singular Thoughts Without Objects or Acquaintance: The Prospects of Recanati’s (and Others’) “Actualism”

Abstract

We argue that Recanati (2012) burdens his otherwise salutary “Mental File” account of singular thought with an “Actualist” assumption that he has inherited from the discussion of singular thought since at least Evans (1982), according to which singular thoughts can only be about actual objects: apparent singular thoughts involving “empty” (referenceless) terms lack truth-valuable content. This assumption flies in the face of manifestly singular thoughts involving not only fictional and mistakenly postulated entities, such as Zeus and the planet Vulcan, but also “perceptual inexistents,” e.g., Kanizsa figures, rainbows, words and phonemes, as well as hosts of at best metaphysically problematic “objects,” such as properties, numbers, ceremonies, contracts, symphonies, “the sky,” “the rain.” Indeed, reflection on what seems to be the boundless diversity of “things” about which we seem to be able to have singular thoughts strongly suggests that there may be no general metaphysics of objects, much less (what Recanati calls) “acquaintance” and “epistemically rewarding” relations that would distinguish singular from non-singular thought. We recommend that Recanati and other mental file theorists confine the theory to a metaphysically neutral account of singular thought as specific kind of internally “focused” computational state, and not seek any general account of the relation of thought to reality.

François Recanati’s (2012) Mental Files is a recent, quite engaging contribution to a now substantial discussion, extending back many decades, on singular thought, or those mental structures that are responsible for agents’ abilities to think about and keep track of particular things. Although we are immensely sympathetic to his discussion of mental files as a way to account for such thoughts, in this paper we want to focus on one very widely shared assumption of that discussion that plays a central role in Recanati’s book: ¹

Actualism: “thought about actual individual objects” is thought about actual objects virtually all² external to the cognitive system.

This might sound to many ears as an innocuous, almost tautological assumption. However, it’s important to see that, far from being verbal, it’s a quite substantive claim, flying in the face of what would seem to some of us to be manifest cases in which people have singular thoughts about things that don’t exist, e.g., Zeus.³ We want to argue that many of the insights of Recanati’s conception of mental files can be not only preserved, but improved without this, we think, deeply problematic assumption.

1 Singularism

Recanati advances his views as an alternative to “Descriptivism” according to which:

we live in a qualitative world of properties -a world where objects only have secondary or derivative status, from an epistemic point of view. ... Statements allegedly about individual objects turn out to express general propositions: “a is G” translates as “The F is G, [which] as Russell pointed out... expresses a general proposition. -(1912:4)

By contrast, “Singularism”:

holds that our thought is about individual objects as much as it is about properties. Objects are given to us directly, in experience, and we do not necessarily think of them as the bearers of such and such properties. -(2012:4)

This latter thesis is actually the conjunction of two claims that we would like to distinguish, what in view of our reservations we will call:

Actualist Singularism: Our thought is about actual individual objects as much as it is about properties.

Acquaintance: Objects are given to us directly in experience, and we do not necessarily think of them as the bearers of such and such properties -(2012:p4)

One of the main claims Recanati defends is an explanation of actualist singular thoughts by his:

Actualist Mental Files: “To entertain a singular thought about an object a is to activate a mental file based upon some acquaintance relation with a. The “mode of presentation” under which a is thought of is not constituted by the properties which the thinker takes the referent to have (i.e., the properties represented in the file), but, rather, by the file itself. The file is what plays the role which Fregean theory assigns to modes of presentation.” -(2012:221)

Recanati claims that his theses are “semantic/epistemological…not metaphysical theses” (2012:5fn3). However, we think he fails to notice that his theses do in fact involve quite substantial metaphysical commitments, viz., to there actually being objects of singular thoughts for us to be acquainted with! But, for reasons we will set out below, we think it is doubtful that we can count on the world really to afford all the varieties of “objects” about which we can have singular thoughts, much less that there is some special acquaintance relation a thinker must bear to them. Consequently, we would like to urge more modest “neutralist” versions of his (and others’) Actualist theses regarding Singularism and Mental Files: ⁴

Neutralist Singularism: We can think of objects with singular thoughts involving inidividual concepts, which may be expressed by a proper name or some deictic or indexical representation and which (i) need not be equivalent to any description deployed by the thinker, and (ii) need not refer to some actual object.

Neutralist Mental Files: To entertain a singular thought is to activate a mental file based upon an individual concept, which need not refer to an actual object. Some portion of that mental file may correspond to a Fregean “sense.”

That is, we want to endorse a view of Singularism and Mental Files that is neutral with respect to whether the objects of singular thoughts are actual things, as well as to whether we are “acquainted” with “them.”⁵

We should stress that we are sympathetic with the, by now, quite familiar considerations of, e.g., Kripke (1972/80) that have led Recanati to his views. There seems to us no question that the kind of simple Descriptivist theories of reference that have been proposed are inadequate to capture the singular thoughts that play an important role in our mental lives. And we find Recanati’s hypothesis of Mental Files, at least as a specific kind of internal representation, a quite promising account of their possibility, in contrast to Quine’s Descriptivist proposal of eliminating singular terms altogether (2012:4).⁶ But we’re puzzled by his burdening the existence and identity of these internal files with a relation of acquaintance with actual objects in the external world. It seems to us that Descriptivism, at least as a psychological (as opposed to a purely metaphysical, cf. fn5) hypothesis, can be resisted without such a burden.

A crucial caveat from the start: one might worry that our neutrality about actual objects would invite an extreme internalism, according to which, for all that psychology cares, people could be brains in vats. But nothing of the sort is implied by the internalism we’ll be urging. At least for the nonce, we are happy to embrace:

Weak Externalism: some attitude contents depend for their identity in some way or other upon some external facts.

As one of us has emphasized elsewhere,⁷ this is pretty much the only reasonable conclusion to draw directly from the numerous “twin” thought experiments (where two physically identical brains are thinking of different things as a result of being situated in different environments). Any “stronger” externalisms, which try to cash out any of the three existential quantifiers, are, like Recanati’s present proposal, speculative theories of the specific relations that must obtain. It is these specific theories, not the weak “triply existential” requirement that we doubt.

Although we endorse Weak Externalism, we don’t want to do so at the risk of denying what we take to be the largely internalistic focus of explanatory psychology. We reconcile the two by endorsing what might be called a correspondingly

Weak Internalism: Within the constraints of Weak Externalism, psychology is largely concerned with causal and computational processes over internal intentional states individuated largely independently of their relations to any specific actual objects.

Thus, although, say, vision theory may well be constrained by operating in the natural environments in which animals have evolved, and so may be committed to internal representations of the edges and surfaces that those environments afford, we doubt that it’s committed to internal representations of specific edges and surfaces (we’ll consider some exceptions in §6 below).

In what follows, we’ll discuss in §2 the relation of Recanati’s views to views of Frege and Russell, and in §3, his views of acquaintance. In §4 we’ll consider a wide diversity of cases that we think present serious prima facie problems for his Actualist views, the most difficult of such cases being those involving “empty” (referenceless) thoughts. We’ll discuss Recanati’s proposals about these latter cases in §5, concluding in §6 that these and our other cases cast doubt on the kind of general theory of the relation of singular thoughts to reality that Recanati and others have been seeking.

2 Russell, Actualism, and the Notion of a Singular Thought

Russell (1912) famously insisted that our knowledge of the world was based upon acquaintance with “sense data.” Whether or not there are such “sense data,” it was fairly clear what sort of things they were supposed to be. Unlike Russell, Recanati doesn’t provide an explicit discussion of what he takes the relevant class of “objects” of acquaintance to be, though, as we’ll see, particularly from his incorporation of the views of Lewis (1999), it may be a very large class indeed. But modulo their different understanding of the relevant acquaintance relation(s), Recanati’s view is in essential respects Russellian.

Recanati presents his view as one that rejects Russell’s “one-level semantics” in favor of a Fregean “two-level framework.” However, for Recanati this doesn’t entail the introduction of a second semantic notion, because what explains cognitive significance are mere mental files. And so, this doesn’t amount to an abandonment of Russell’s views about meaning. It’s better seen as showing how a Russellian view can be augmented by deploying the notion of a mental file to account for phenomena of cognitive significance. (Recanati also operates with a “character/content” distinction, but that too amounts to an augmentation of a Russellian framework rather than a revision.)

According to Recanati, Russell made a “major mistake” in departing from Frege on whether the notion of reference alone suffices for an account of the content of a representation (2012:7). In our view, Recanati makes a similarly large mistake in not following Frege in allowing that there can be a sense without a referent. That, too, is arguably an inheritance from Russell. Two semantic levels may in fact be needed.

What motivates Recanati’s actualism? It doesn’t seem to be (anyway, we hope it’s not) Russell’s foundational epistemic concerns. He just wants an alternative to Descriptivism, and, with Bach (1987), wants therefore a distinction between “satisfactional” (or “ascriptive”) modes of referring to an object, and “relational” ones (2012:19–20). He thinks a relation of acquaintance will do this work. But Bach’s distinction is meant merely to capture the distinction between modes of reference determination that do and modes that don’t (in Recanati’s own words) “exploit the contextual relations in which we stand to what we think about” (2012:21). And there may be other ways to exploit “contextual relations” other than by appealing to an actual object of thought.⁸ This is at least the possibility to which we think our discussion will point (see §§5-6 below).

3 Acquaintance

For all of Recanati’s appeals to a relation of “acquaintance,” we find it difficult to determine precisely what he has in mind. His initial characterization that we quoted above stresses the lack of (property) concepts, as do later passages (2012:29), where he also cites with approval Pylyshyn’s (2003, 2007) view, according to which there’s no conceptualization whatever for basic visual “indices.”⁹ However, in other passages he does enrich the idea, writing:

In general, there is acquaintance with an object whenever we are so related to that object that we can gain information from it, on the basis of that relation. Acquaintance relations are epistemically rewarding (ER) relations… -(2012:20)

ER relations in turn are relations that:

enable the subject to gain information from the objects to which he stands in these relations....Relations of perceptual acquaintance are ER relations: they are the sort of relation to objects which makes the perceptual flow of information possible. -(2012:35)

But he then adds a crucial footnote, relying on a suggestion of David Lewis (1999), which extends ERs far beyond immediate perception, indicating how broadly he understands how we can be “so related to an object that we can gain information from it”:

The paradigm [of acquaintance relations] is, of course, perceptual acquaintance, but the notion can be generalized “in virtue of the analogy between relations of perceptual acquaintance and other, more tenuous, relations of epistemic rapport. There are relations that someone bears to me when I get a letter from him, or I watch the swerving of a car he is driving, or I read his biography, or I hear him mentioned by name, or I investigate the clues he has left at the scene of the crime. In each case there are causal chains from him to me of a sort which would permit a flow of information: perhaps I get misinformation, but still the channel is there. I call such relations as these relations of acquiantance (Lewis 1999:380–1).” [returning to main text:] According to the account I develop in this book, different types of file correspond to different types of relation. -(2012:34–5,fn5)

Now, it’s a shame that Recanati relegated this expansion of (ER) relations to a footnote.¹⁰ For it’s hard not to wonder what’s left of the notion of acquaintance, once it’s allowed that it may involve the kinds of cognitively rich causal chains that mediate a person’s understanding of letters, cars, biographies, uses of names, or evidence of a crime.¹¹ To be sure, in the cases being imagined, there are causal chains between a thinker and the object being thought about; but there are causal chains connecting most pairs of things in the world: any normally educated person has a pretty good idea about how in principle to “gain information” about most anything.¹² After all, everything is potential evidence of practically everything else, if you can only figure out how to mine it, since most everything (at least in the relevant past light cones) is causally related to most everything else.

Thus, take Kaplan’s (1969) parade example of what would appear to be a non-singular case: at least ordinarily, Ralph doesn’t really suspect someone of being a spy merely by virtue of believing the shortest spy is a spy, and the fact that very spy does in fact exert a causal, e.g., merely gravitational, influence on him (and he’d find him, if only he’d really exploit the available information channels!). Recanati needs some way of specifying the right, distinctive (ER) relations for what he regards as genuine singular thoughts. We suspect he has in mind the kind of “causal chains” that Lewis mentions, and which Kripke (1972/80) and Devitt (1981) discuss, but we haven’t been able to find any discussion of constraints on these chains that would suitably restrict the (ER) relations.¹³ However, we think the wide variety of examples of “objects” about which people can have singular thought that we will consider argues against there being any generally specifiable distinctive (ER) relations to which Recanati can appeal. Indeed, it seems to us that there are plenty of singular thoughts that would seem to involve no real relations at all. ¹⁴

4 Problematic Cases

There are both metaphysical and epistemic problems with Recanati’s notion of acquaintance, his epistemic claims depending upon problematic metaphysical presumptions about the kinds of “objects” there are with which someone could plausibly be acquainted.

So far as we can find, Recanati nowhere discusses what counts as an “object.” We will assume he is simply taking for granted the kinds of examples standard in the literature that he discusses: people, planets, animals, concrete artifacts, and concrete natural kinds (where “concrete” means “having fairly definite space time location”). We’re not sure his theory will in the end work even for them, but we leave that for the reader to judge in view of the problems raised by more obviously difficult cases (of which we think we are really only providing an almost random sample).

4.1 The General Diversity of “Objects”

On the face of it, it would certainly appear that one can and does have singular thoughts about anything: not only Recanati’s (excessively?) familiar examples of people and planets, but: species, kinds; (types and tokens of) books, plays and symphonies; performances, ceremonies, (annulled) marriages, (forged) contracts; the stock market, stock market crashes, companies, stores, clubs; galaxies, Black Holes; and (to take some now standard examples from Chomsky 2000:135) flaws in arguments, “Joe Six Pack,” a person’s health, and “the inner track that Raytheon has on the latest missile contract.” It is at least controversial whether all these “things” (as opposed to our representations of them) enter into any serious causal relations at all, much less the special (ER) ones that Recanati needs to sustain “acquaintance.”

Some particularly troublesome cases:

4.2 Abstract Objects

Abstracta, e.g., numbers, sets, properties, categories, are, of course, familiar enough to philosophers, and it’s more than a little surprising that Recanati nowhere discusses them. It’s hard to believe people don’t sometimes have singular thoughts about them (“the winning number,” “His favorite category,” “that property that makes her face so beautiful,” cp. Loar 1997, on “recognitional concepts”). Whatever these various abstracta are, it would require considerable discussion to show how we have any serious acquaintance with them (Russell 1912, flirts with acquaintance with some universals, but provides no details, and soon gives up on such a view). But, in any case, can we have “acquaintance” with a number or property without having a concept of it? If, as seems likely, we can’t, it’s unclear what would be left of the distinction between acquaintance and description, at least as Recanati seems to want to draw it.

Indeed, notice that here the implicit verificationism that we fear Recanati’s proposals presume begins to be problematic. Speaking at least for ourselves (but we expect virtually all those unequipped with a happy epistemology to cover such cases), we are not at all sure how or even whether we anticipate becoming the least acquainted, or even really seriously knowledgeable, about many abstract entities about which we seem to have singular thoughts: goodness, justice, the American Way (“I’m not sure exactly what justice is, or whether anyone will ever know, but it is a cardinal virtue about which Rawls wrote an interesting book”; “There’s this thing, ‘the American Way,’ that we fear too many people may regard as sacrosanct”).

4.3 Weird “Objects”

An instructive exercise for anyone concerned with reference and ontology is to consider what the referents might be of the common singular thoughts we seem to have about “the sky,” “the heavens,” “the wind,” “the rain,” “the tide,” “ocean waves,” “shadows,” “reflections,” and, again, rainbows and “halos” around the moon -and even (especially?) the “present time” and “present place” needed by Kaplanian “character”: are these really serious things with which anyone stands in any special causal or (ER) relations? Perhaps some metaphysicians can provide suitable paraphrases or constructions for such cases, but it's hard to believe they, much less (ER) relations to them, are really required in order to vindicate our frequent singular thoughts about such “things.” Indeed, we strongly suspect the right view about many of these “things” is that they don’t exist at all.

This last possibility, of course, raises what we think is an immense category that obviously presents a problem for any Actualist:

4.4 Empty terms/Intentional Inexistents

This is a category that Franz Brentano (1874/1973) famously resurrected from medieval philosophy, whose importance to psychology we think has been greatly underestimated by much contemporary philosophy. For our purposes they may be regarded as “objects” of thought or percpetion that don’t actually exist, indeed, whose existence a thinker might readily deny.¹⁵ They are usually associated with merely deliberate “fictional” characters (Sherlock Holmes, Santa Claus), or objects posited by elaborate theories (the ether, phlogiston, either the god or the planet Vulcan). Of course, as Recanati notes, these also present prima facie problems for his views, and he considers several strategies to deal with them -e.g., pretense, meta-representation (2012:204)- that we will consider in due course (§5). Before we do, however, we want to call attention to the full range of cases that needs to be considered.

A particular class of cases that we think has not received sufficient attention is what we call “perceptual inexistents”: as one of us has stressed elsewhere (Rey 2006, 2012), such “entities” seem to be posited quite regularly in psychology in cases of early perceptual content, as when one “sees” Kanizsa figures with illusory contours (examples are easily available via google), or rainbows, mirages, cartoon figures, or–for many linguists going back at least to Sapir (1933)/49)–standard linguistic entities such as phonemes, words and sentences.¹⁶ What’s challenging about such cases is that some of the usual strategies for dealing with empty terms don’t seem quite so plausible for them (people don’t normally regard words and rainbows as involving deliberate pretense or meta-representation). Worse, many philosophers and psychologists have reasonably argued that there are no real triangles (or none that anyone could causally interact with), nor any real “secondary properties/objects,” such as colors, tastes, sounds, despite such “things” and “properties” being routinely represented, often singularly, in perception (“that shade of blue,” “that sublime taste”). Any doctrine of singular properties and thoughts that demands acquaintance as a real relation to such at least highly controversial “things” would have a lot of explaining to do.

Or so we would like to insist. We’re aware, however, that there is the authority of Gareth Evans (1982) and his disciples against us, according to which singular thought is akin to thoughts expressed by demonstratives, and that a sentence with a demonstrative that doesn’t succeed in picking out anything real in the world does not have truth-valuable content (at best, people in such cases have “nearby thoughts”). We seriously doubt this latter view, but the topic deserves extended discussion.

5 Can there be Referenceless Singular Thoughts?

Pace Evans, it seems to us that there is an important place in psychology for taking “empty” demonstratives that fail to refer to actual objects to contribute crucial content to both perceptions and thought, particularly in the perceptual cases we’ve mentioned. After all, people can reason quite elaborately about the apparent specific objects (“If that triangle were removed, there’d be three disks,” “That word was either foreign or mispronounced.”). Indeed, they –especially philosophers!- sometimes will press arguments that the “things” are really there after all. But sometimes –for example, ourselves in the case of rainbows– people know full well the things don’t exist, but they acquiesce in ordinary reasoning about “them” nonetheless (“Somewhere over the rainbow…”) without any serious expectation that there are (ER) relations that will allow them to become better acquainted with them (n.b., the rainbows themselves). Or consider those “inner voices” (e.g., Socrates’ “daemon”) the likes of which many of us seem to hear urging us on: unless we’re schizophrenic, we don’t believe they’re actual, and certainly haven’t the slightest idea of how we’d find out more about “them.” They seem to come from “nowhere.”

As we mentioned (fn17), there are a number of points at which Recanati appears to allow for acquaintanceless thoughts, but we think these appearances are misleading. Thus, replying to an objection similar to ours raised by Crane (2011), Recanati writes:

In contrast to Crane, I hold that there are two distinct notions: what someone is thinking in the sense of the mental representation that is tokened in one’s mind (which representation is endowed with a primary content akin to a Kaplanian character), and the semantic (truth-conditional, secondary) content of that representation. –(2012:247)

But such a concession misses the fact we’ve been stressing that, unlike Kaplanian character, referenceless thought enters into the rich rational, truth-evaluable reasoning about perceptual and other intentional inexistents.

A more complex possibility, which we confess we find a bit Pickwickian, is raised by Recanati’s appeal to a “normative” condition. Since he thinks of a mental file as an item that serves a certain “function” in the mind, it involves a “normative requirement corresponding to the function of the file” (2012:63). The relevant normative requirement is not a “de facto” requirement that the agent actually be acquainted with an object at the time he is perceiving or thinking about it, but a de jure requirement that a subject “should stand in a suitable (ER) relation to some entity (the referent of the file)” (2012:63,156). Since, “there is no function without a possibility of malfunction…, there is no reason why a file could not be tokened even though the normative requirement is not met” (2012:63). Thus, there can be empty singular thoughts, since it’s enough that a singular thought be subject to the de jure condition, even if the condition is not satisfied. But the de jure condition has to be satisfied in a specific way.

In particular, for Recanati, this de jure condition is satisfied if(f?) the thinker correctly anticipates acquaintance with the object: ¹⁷

What are the conditions on successful singular thought contents? … [O]ne can express a singular thought only in virtue of some relation to the referent. But, I have tentatively argued, the relation need not necessarily hold at the time of tokening the singular thought.... One can think a singular thought (content) for the time being, one has only the description to rely on, provided one is right in anticipating that one will come into relation to the denotation of the description, and be in a position to gain information from it. -(p169)

Put aside whether “anticipation” actually provides any serious constraint.¹⁸ The reason we find all this a bit Pickwickian is that it (to our mind) drastically limits acquaintanceless thought to merely thoughts that are acquaintanceless at a particular time: the requirement of acquaintance with an actual object is still a de facto requirement across time, i.e., from a timeless perspective. Thus, an astronomer can be acquainted with Neptune at a particular time if he correctly anticipates being in (ER) relations with it.¹⁹ But, of course, he still couldn’t be acquainted with (the forever non-existent) Vulcan (p164). Again, given the richness of thoughts that can occur in such cases –not only the perceptual cases we discussed, but, one would have thought, especially the careful astronomical calculations about just what properties Vulcan should have– this just seems inconsistent with serious psychological explanation.

Recanati does recognize he should say a little more about such cases, but spends only a few pages on them, considering options for which it’s hard to imagine any independent psychological evidence: e.g., that such thoughts are “mock” or “pretend” thoughts,” or that they are “meta-representational,” thoughts about the empty words or thoughts themselves (2012:202–4). Although these latter options might sometimes be plausible in the case of fairy-tales, it’s hard to see how they could be seriously sustained in the case of perceptual inexistents, or in the cases of astronomers focused intently on the causes of Mercury’s perihelion. ²⁰

Instead of struggling to defend such tenuous accounts, we suggest re-thinking Evans’ considerations. For starters, why not suppose that all singular thoughts come with a demonstrative element that is also bound by a uniqueness existential quantifier; or perhaps that such quantification is a presupposition of such thoughts that renders those thoughts false when the quantification is? Or perhaps the “character” of the element requires not only a function mapping a context to a thinker, a time, a place and a world, but also to a contextually determined, internal “focus” of the thought, the output of material that is the focus of the thinker’s attention –e.g., the patterns created by refracted light in the sky, the representations of Grecian talk of their gods– which, perhaps most importantly, enter into a psychologically plausible explanation of how these stimuli gave rise to an internal singular representation (e.g., it sure looks like there’s a rainbow in the sky; ancient Greeks did seem to posit a god named “Zeus” etc.).²¹ N.b., the “focus” need not refer to anything outside the nervous system, but simply to the internal material involved in the process of attention, e.g., relatively proximal perceptual output. It’s the psychological explanation of this material (if there is one) that would provide the grounds for the ascription of one singular content rather than another to the mental file. Particularly given the problems raised by many of the other kinds of cases we mentioned, such strategies deserve much more discussion than Recanati (or, as far as we know, anyone else) has provided, before such a large range of apparent singular thoughts is treated as spurious.²² In fact, we see Recanati’s very own postulation of mental files as the “senses” of tokens of terms in contexts as an excellent suggestion for the content of empty terms²³ -provided of course it’s freed of the commitment to acquaintance with real objects!

6 General Prospects of Actualist Singularism

Recanati is seeking a quite general theory of the nature of singular thought (2102:153), presumably comparable to the kind of theories sought in linguistics and theories of, e.g., perception, attention, memory, cognitive dissonance, framing effects, etc. However, it is worth noticing that almost all of his examples are of occasional folk, singular explanations and intuitive judgments that have been the staple of the field for the last 50 years or so: who was the speaker referring to at the party? How could John think that Cicero but not Tully is bald? Given the diversity of kinds of “objects” we’ve mentioned, is there really any prospect of a general theory either of the metaphysics of all these "objects" and their causal powers, or of any sufficiently specific relation we might bear to them when we think about them with singular thoughts? Why think so?

To be sure, there’s a puzzle about when a singular thought can be regarded as genuinely about some specific real thing, and, correspondingly, when we can significantly “quantify into” the content of the thought: when can someone think the shortest spy is a spy having a particular spy “in mind” (Kaplan 1969)? Specific causal relations are no doubt relevant in many cases; but why think they are present in all?²⁴ Why think there’s any general solution to puzzles of this sort beyond occasion-relative pragmatics and forensics? Indeed, pace the recent resurgence of interest in traditional “metaphysics,” why think that there’s a general satisfactory account of all the multitude of “things” that we are able to think about? It’s hard not to suspect that the majority of such issues are really just matters of pragmatics and forensics.²⁵ We submit that Recanati (inadvertently) expressed exactly the right idea when, explicating Bach, he wrote what we quoted earlier: “indexicals systematically exploit any of various contextual relations in which we stand to what we talk about” (2012:21, emph ours), which might, again, be understood as “focusing” singular thought, without, however, necessarily involving any genuinely actual object we’re talking about.

What we think Recanati is right about is that there are categories of singular thought that will require the kind of computational operations he nicely describes: proto and other temporary files that undergo “incremental conversion” into more stable ones (2012:chps 6–7), to which various “objects” of thought and perception thereby get attached -even when (for us) the “objects” are unreal. That is, all of this can be subsumed under our neutralism and weak internalism: there is, so far as we can see, no need to burden the account with Recanati’s claims of Acquaintance or his Actualist versions of Singularism and Mental Files; and there are all the reasons we have provided to embrace the more modest Neutralist versions of these views that we have recommended.

It’s worth noting some interesting apparent exceptions to our sceptical line, which, however, we see as merely ones that “prove the rule” -or, anyway, provide an instructive contrast to the usual cases. As Recanati notes (2012:61), each person is in a quite systematic way acquainted with themselves, and his positing of SELF files provides a nice account of this. Moreover, the role of such a file with the “self” as its referent is systematic and arguably essential to the coordinating system of thought, perception and action (“I’m late, so I’d better run now” –although, again, it’s actually pretty obscure what sort of “object” now is).

Other possible examples are afforded by the remarkable navigational abilities of animals, e.g., ants and bees, where there does seem to be a quite systematic relation between at least the local geometry of space and time and the vector algebra by which they appear to compute something that serves to guide them to burrows, prey or sources of nectar they have discovered.²⁶ And in some cases it does appear as though something at least very like singular representations of specific objects may be involved, as when birds represent various stars, and bees the azimuth of the sun (Gallistel 1990).

But, again, these cases are striking for being exceptional. As creatures become more intelligent, flexible, ingenious and creative, their relations to their environments become almost boundlessly varied, making them, we submit, capable of singular thought about “objects” to which they stand in any number of different relations -or none, as in the case of inexistents.²⁷

Are we endorsing the kind of scepticism about the prospects of any determinate intentional psychology that one finds in the later Wittgenstein and Quine and their followers? No. As we were at pains to stress at the outset, none of the examples we have considered argue against at least weakly internalist theories of mental processes, along the lines of the on-going psychological research into the areas we mentioned.  Recanati’s discussion of mental files is a welcome contribution to that research, but only if it is shorn of its, we think, forlorn commitment to acquaintance, special (ER) or causal relations, or, really, any single, general  relation, between singular thoughts and their objects.