Towards a Sensible Bifurcationism (concerning what grounds thought about particulars)

In virtue of what are particular individuals or objects thought about ? I call this "The Grounding Question." A consensus answer is bifurcationism : objects can be thought about in virtue of both satisfactional grounds--roughly, in virtue of their unique satisfaction of conditions that figure in a subject's thought--and non-satisfactional grounds. Bifurcationism is a consensus view, but it comes in different flavors, which correspond to different approaches to answering the Grounding Question. This paper draws on Saul Kripke's approach to linguistic reference in order to make recommendations about how to move toward a sensible bifurcationism concerning what grounds thought about particulars

(1) Introduction 1 Acknowledgments: Thanks to the editors for the opportunity to explore the importance of Saul Kripke's field-changing work for the question of how we think about particular objects, in addition to the question of how we refer to them. Joseph Almog especially deserves credit for turning my attention to these matters long ago. Earlier versions of this paper were presented at the Particularity and Generality in Mind Workshop at Umeå University and the Uppsala University Higher Seminar in Theoretical Philosophy. I am grateful to all the participants for their feedback, and especially thank Martín Abreu Zavaleta, Tobias Alexius, Marvin Backes, Karl Bergman, Matti Eklund, Nils Franzén, Rachel Goodman, Torfinn Huvenes, Carl Montan, James Openshaw, Traugott Schiebe and Andreas Stokke for helpful discussion. Thanks also to Indrek Reiland for comments and discussion. I am indebted to conversations and correspondence on the topics of this paper with Frank Jackson, Alex Sandgren and Pär Sundström. What makes a particular object be, as we might put it, in someone's mind, or thought about by someone? What sorts of relations between a thinking subject and a particular object ground or underlie this relation of thinking about? I will call this "The Grounding Question." The Grounding Question, and proposed answers to it, are not always clearly distinguished from other questions about the contents of thought and mental representation. Still, it seems to me that a partial consensus has emerged about how to answer it. This partial consensus is a commitment to what I call (following Pär Sundström, p.c.) "bifurcationism". Bifurcationism is the view that the Grounding Question has two different kinds of answer, depending on the case. In some cases, particular individuals are thought about in virtue of uniquely satisfying certain sorts of conditions. In other cases, particular individuals are thought about not in virtue of this kind of condition satisfaction, but in virtue of some other grounds. I will say more about bifurcationism in section 3, but perhaps this is enough to give a sense of the view.
Bifurcationism is a consensus view, but it comes in different flavors, which correspond to different approaches to answering the Grounding Question.
In section 4, I will propose an interpretation of some of Saul Kripke's arguments in Naming and Necessity according to which Kripke advocates a parallel bifurcationist view about what grounds linguistic reference to particulars. On this interpretation, Kripke's method for moving towards a sensible linguistic bifurcationism is a method of counterexample and extrapolation. First, one describes compelling cases of linguistic reference to particular objects in the absence of (a linguistic analog of) satisfactional grounds. Then, one extrapolates from these cases to other cases, inferring that a non-satisfactional ground is or is not also present in those cases, irrespective of whether satisfactional grounds obtain in those cases.
Interpreting Kripke in this way suggests that one influential line of reply to his arguments does not (immediately) hit its target. But, as I discuss in section 5, it also illuminates obstacles this reply presents for those who would apply Kripke's method of counterexample and extrapolation to the investigation of thought about, rather than linguistic reference to, particular objects.
To overcome these obstacles, I propose a new type of case to be employed in the counterexample stage of the method: cases of false belief in the non-existence of particular objects. Then I round off the discussion, in section 6, by emphasizing the importance of the extrapolation part of the method. The upshot of the section is that compelling counterexamples must be complemented by positive accounts of non-satisfactional grounds for thought about particulars, in order to move toward a sensible bifurcationism. Moreover, it is not only those, like myself, who argue for more cases of non-satisfactionally grounded thought about particulars who need these accounts, but also those who argue that the range of such cases is quite limited.
(2) The Grounding Question First, as I am using the phrase, "in virtue of what" asks for what might be called a "metaphysical explanation" or a "grounding explanation" of the fact, or perhaps state of affairs, of a particular individual's being thought about. The question might also be phrased as, "What makes it the case that a particular individual is thought about?" or "What underlies a particular individual's being thought about?" or "What explains a particular individual's being thought about?" The Grounding Question, in all these forms, asks for a description of some property P that a particular individual i may have, such that i is thought about because i has P, but not vice versa. That is, it is not the case that i has P because i is thought about: the having of P is explanatorily prior to being thought about. (Presumably, P is a relational property of bearing a certain relation to a thinking subject.) In this paper, I am assuming that grounding is asymmetric, transitive, irreflexive and well-founded. This assumption is common, but by no means beyond challenge. (See, e.g., Bliss 2018.) Second, notice the difference between the Grounding Question as I have phrased it and another, related question: In virtue of what is a particular thought about a particular individual? This second question takes for granted two entities, a particular thought and a particular individual, and asks what it is for a certain relation to hold between them. It thus highlights thoughts as the primary things that we think, as opposed to worldly objects and properties. It might seem that this is a better way to put the question, since this is how we talk.
We say that we think thoughts, not that we think birds, or think John, or think the number two.
Rather, we think about birds, or John, or the number two, by thinking thoughts that are about But I would like to avoid presupposing this framework, in part because it requires assuming some way of identifying and individuating thoughts, such as by their content or by (something like) their form. Identifying them by their content threatens to trivialize the question I am interested in, since, at least on some ways of understanding belief-contents, assuming that a belief has a certain content is tantamount to assuming that it is about a certain For one thing, the problem is not that the relevant use of "see" is extensional while "think" is intensional, since intensional verbs like "seek", "love", "visualize", "desire" and so on can all take as direct objects terms referring to particular individuals. In any case, to keep my question natural sounding I will use the "thought about/think about" construction, with the caveat that my aim is to ask in virtue of what a particular individual bears a certain relation to a thinking subject (the relation of being-thought-about-by). To do this might ultimately also require an account of what these things we call "thoughts" are, but I would rather not assume this structure in the very question.
A third remark is related to the second one. It is important to distinguish The Grounding Question from some closely related questions that arise in the literature on "singular thought".
In this literature, the focus is typically on thoughts. The central opposition is between the view that our thoughts about particular objects are (in some sense) "singular" and the view that they are (in some sense) "general". The parenthetical "in some sense"s reflect quite a bit of variation in how different writers distinguish singular from general thoughts. Sometimes these two types of thought are distinguished by features of their contents. For instance, the contents of singular thoughts, but not the contents of general thoughts, might be taken to include or contain the particular objects they are about. Sometimes they are distinguished instead by features of a thought's form or vehicle. For instance, singular thoughts, but not general thoughts, might be taken to use a mental file as a vehicle. Sometimes these two types of thought also seem to be distinguished by there being different answers to a version of the Grounding Question about them. For instance, it is sometimes suggested that general thoughts are about particular objects because those objects fit some description that is also part of the thought, while singular thoughts are not about objects in this way. 2 So the question of whether thoughts are general or singular often looks like a mix of questions concerning features of the content, vehicle and grounds for those thoughts. 3 One goal of the present paper is to highlight the Grounding Question as a distinct question of interest.

(3) Cognitive Bifurcationism
Various answers to the Grounding Question have been offered. Some have been given as explicit answers to the related kind of question mentioned above: what grounds aboutness relations between particular thoughts (individuated in one way or another) and particular individuals? From answers to these questions, one can build back answers to the Grounding Question: a particular individual is thought about in virtue of someone having a thought that stands in the relevant relation to the individual. Other answers are implicit in broader theories of the content of thought and of mental representation. 4 A common commitment across many of these answers is what I will call "bifurcationism". 5 Bifurcationism is the view that the 2 e.g. Recanati (2012: 3-5), Martone (2016: 35-6).

3
A number of writers have clearly distinguished the question of whether a thought content is singular or general from the question of what makes thought be about a particular object. See, for instance, Lewis (1986: 33), Jackson (2015), Pepp (2019), Sandgren (2019).

4
For instance, Recanati (2012) and Dickie (2015) address fairly directly the question of what grounds the fact that certain types of thought are about what they are about. Both argue that (different types of) epistemic relations between speaker, thought, and object serve as grounds. An earlier epistemic account is found in Evans (1982). Devitt (1981Devitt ( , 2015 instead argues that a thought (which he considers to be a representation in the language of thought) is about a particular object in virtue of a certain sort of causal chain, originating in a perceptual "grounding" in the object, leading from the object to the tokening of the thought. (NB, Devitt's notion of grounding is not the same as the metaphysical grounding of the Grounding Question.) Other broadly casual, non-epistemic accounts of what grounds thought about particulars can be found in Almog (2014), Capuano (2015) and Wulfemeyer (2021). Bach (2010) seems to sit somewhere on the fence between an epistemic account and a causal, non-epistemic one. Millikan (2017) suggests that we think about objects by having certain systems ("unicepts") that are causally derived from those objects and serve to focus information about them and mark it as information about the same thing. A very different sort of answer emerges from the broader approaches to content and representation developed across bodies of work by Chalmers and Jackson (see References section for relevant citations), as well as Searle (1958Searle ( , 1983 and Strawson (1959). (This list is not intended to be complete. Also, the writers mentioned generally do not use the term "grounding" in the way that I am using it in the present context, so it is not always obvious that they are addressing a version of the Grounding Question. In any case, partial answers to the Grounding Question can be extrapolated from their work.)

5
This label for the view is due to Pär Sundström. Grounding Question has two different kinds of answer, depending on the case. In some cases, particular individuals are thought about in virtue of what I will call satisfactional grounds. 6 In other cases, particular individuals are thought about in virtue of some other type of grounds, which I will call non-satisfactional grounds.
The basic idea of a satisfactional ground for being thought about is that an individual is thought about in virtue of being "picked out" by, or uniquely satisfying, a condition that figures in a subject's thought. For instance, if a subject thinks that there is exactly one cat currently in her garden and that it is making a lot of noise, the fact that a unique individual satisfies the condition of being a cat in the subject's garden is a satisfactional ground for that cat's being thought about by the subject.
As an initial schematic gloss, let us say that For a particular individual to be thought about in virtue of satisfactional grounds is for it to be thought about in virtue of uniquely satisfying a condition that has a certain special status.
What this special status is can be spelled out in different ways. In the rough statement in the previous paragraph I spelled it out as "figures in a subject's thought." That is imprecise, and I will say more about what the special status might be in a moment. But first I will make a couple of remarks about the division between satisfactional and non-satisfactional grounds. If particular individuals are thought about in virtue of something other than satisfactional grounds, then they are thought about in virtue of non-satisfactional grounds. The division of grounds for an individual's being thought about into satisfactional and non-satisfactional grounds is mutually exclusive and exhaustive, so that every grounds for thought about a particular individual is either satisfactional or non-satisfactional, but not both. Although every ground for being thought about is either satisfactional or non-satisfactional, it is consistent with bifurcationism that a particular individual may be thought about in virtue of both kinds of ground, simultaneously and by the same subject. That is, there may be overdetermination in the grounding of a subject's thinking about a particular object. Now back to the special status. The special status should be spelled out so that it makes sense that an object's uniquely satisfying a condition that has the status would make the object be thought about. One way of doing this is to draw on Bertrand Russell's (1910-11) notion of 6 Again, thanks to Pär Sundström for this label. knowledge by description. Russell claimed that some things we can think about are known to us only by description, in the sense that we know that some description is uniquely satisfied, but have no other "direct cognitive relation" to the thing that uniquely satisfies the description.
Set aside, for now, Russell's further claim that most ordinary things are known to us, and thus thought about by us, only by description. It does at least seem correct that knowing that some description is uniquely satisfied can make it the case that one thinks about the satisfier, at least in some ordinary sense of "thinks about". For instance, if I know that at the present time of day there is always exactly one cat in my garden, this makes it the case that I think--or at least that I am immediately positioned to think 7 --about the particular cat that is now in my garden. I might believe that this cat is hungry, and is a danger to the birds who are also in the garden.
Likewise, if I know that exactly one human being is the most recent common ancestor of all currently living human beings, this makes me (be in a position to) have thought about that long-ago individual. I might believe that they lived in the Upper Paleolithic period and that their life was difficult.
Perhaps one can think about particular things in this way even if one does not know that the relevant condition is uniquely satisfied--belief that it is uniquely satisfied might be sufficient.
Perhaps even belief that a condition is satisfied, uniquely or not, grounds one's thought about its particular satisfier, as long as it is in fact uniquely satisfied. It will not matter for the rest of my discussion which of these notions of satisfactional ground is selected. For the sake of concreteness, I will focus on this one: For a particular individual to be thought about in virtue of satisfactional grounds is for it to be thought about in virtue of uniquely satisfying a condition that a subject believes to be uniquely satisfied.
Allowing satisfactional grounds for thought about particular objects seems to make thought about particular objects "cheap". As long as I believe that there is a unique tallest human in the world, a second tallest, a third tallest, and so on, satisfactional grounding entails that all of these individuals get to be thought about by me. One might wonder whether this is really enough to ground thought about these individuals. The matter is difficult to adjudicate. 7 I am not sure whether this qualification is needed. It seems plausible that a subject's knowledge that there is a unique cat in her garden itself grounds thought about that particular cat. (Here and throughout I intend a broad sense of thinking about or having thought about a particular object, so that it does not require present awareness or entertaining of the thought.) For instance, in virtue of having this knowledge she believes that it is a cat, that it is in her garden, etc. But even if this is questionable, the subject is immediately positioned to think about that particular cat, for instance if she comes to believe that it (the unique cat in her garden) is black, furry, noisy, etc.
On the one hand, satisfactional grounds seem to make thought about particulars come cheap.
On the other hand, it seems that we do have thoughts about the human most recent common ancestor, or a colleague's personal doctor, or the prize behind door number four, and it is not clear what else but satisfactional grounds underlies such thought. At any rate, our ordinary talk of "thinking about" things seems to sanction this. So for present purposes, I am going to assume that an object's unique satisfaction of a condition believed to be uniquely satisfied grounds the object's being thought about. 8 This also makes sense dialectically, since one of my aims in the paper is to offer a new partial argument that a wide range of thought about particulars is grounded non-satisfactionally. Given this aim, I should not insist at the outset that thought about particulars cannot have satisfactional grounds.
Satisfactional grounds for thought about particulars is cheap, but one might wonder whether there are more "expensive" grounds available. On this there is broad agreement. Even theorists who hold that most thought about particulars has satisfactional grounds allow that there is also some thought about particulars that has non-satisfactional grounds. One reason for this is that many conditions subjects believe to be uniquely satisfied are ultimately perspectival conditions, such as the condition of being a cat in my garden, or being a round object I now see before me. To believe that a perspectival condition is uniquely satisfied, one must have thought about particulars--oneself, one's location in time and space, one's current experience--that is itself perspectival. This sort of thought about particulars might be grounded satisfactionally, but only by the subject's having further perspectival beliefs about particulars (for instance, thinking of herself in virtue of believing that something is uniquely the subject of this experience). If these sorts of grounds are to be well-founded, then some perspectival thought about particulars must be grounded non-satisfactionally. 9 One might make do with just one exception to satisfactional grounding of thought about particulars: thought about oneself. This position fits nicely with the view that all thought is de se in nature (Lewis 1979): all thought is "to the effect that one belongs to the class of things that are thus and so." (Jackson 2020) According to this view, a subject's thought that there is a unique cat in her garden might be to the effect that she belongs to the class of things in whose garden there is exactly one cat. Her thought that the tallest human in Sweden is over 2 meters 8 So long as the condition believed to be uniquely satisfied is not an identity condition. If a subject sees an object, she might believe, or know, that something uniquely satisfies the condition of being identical to that object. Belief that an identity condition like this is uniquely satisfied cannot ground the subject's thought about the object, because her belief that the condition is satisfied already requires thought about the object. Thanks to Martín Abreu Zavaleta for pointing this out.

9
As noted above, I am assuming that grounding is asymmetric, transitive, irreflexive and well-founded. tall might be to the effect that she belongs to the class of things whose home country's tallest human is over 2 meters tall. On this view, all of a subject's thought is about herself in virtue of the nature of thought and the fact that she is the subject of that thought. She need not uniquely satisfy a condition with the special status to be thought about by herself. But with this "anchor" for her thought in place, 10 her thinking about any other particular will have only satisfactional grounds.
A different form of bifurcationism is part of Russell's approach to thought about particulars. Although he held that most particulars are thought about in virtue of Russellian satisfactional grounds ("by description"), he also held that some particulars--namely oneself and one's own sense data--are thought about in virtue of something else, the subject's "acquaintance" with them. A modern version of this kind of bifurcationism might posit nonsatisfactional grounds for thought about oneself and one's own subjective experiences, while claiming only satisfactional grounds for all other thought about particulars. (e.g. Chalmers 2018Chalmers , 2012Chalmers , 2001 Other forms of bifurcationism allow that a broader range of particulars are thought about in virtue of non-satisfactional grounds, such as those particulars that subjects perceive, remember, or learn about through testimony. This is arguably the kind of bifurcationism endorsed by many of the writers mentioned in footnote 3. I say "arguably" because there is an important caveat to my claim that bifurcationism is a consensus view. To be sure, some philosophers who think that much thought about particulars is grounded non-satisfactionally are also happy to allow that it may be grounded satisfactionally as well. 12 But others would likely resist bifurcationism as I have articulated it, because they think that real thought about particular objects cannot be satisfactionally grounded. 13 Thus, they do not think there is any 10 I think I am borrowing this figurative use of "anchor" from Jeshion (2012), but I do not know whether it originates there.

11
Another feature of Russell's view (and Chalmers') is that at least some properties--or universals, in Russell's terminology--figure in thought in virtue of non-satisfactional grounds. A more complete exploration of cognitive bifurcationism would address the grounding of thought about properties or universals as well as the grounding of thought about particulars, and the interaction between them. To keep things manageable, I set this aside for now.

12
For instance, Dickie allows that one might think of an object in one's unattended peripheral vision in virtue of its unique satisfaction of a condition one believes to be uniquely satisfied. (2015: 124) And Recanati (2012, ch. 12) allows for satisfactionally grounded thought about particulars.

13
For instance, Bach (2010: 56-7) uses scare quotes when speaking of satisfactionally grounded thought about particulars, and slides between the claim that thought about such particulars is not "singular" and the claim that we cannot think of such particulars at all: "...we cannot form a singular thought about an individual we can 'think of' only under a description. For example, we cannot think of the first child born in the twenty-second century because we are not representationally connected to that individual. And giving it a name doesn't help. Our thought 'about' that child is general in content, not singular." bifurcation between satisfactional and non-satisfactional grounds for real thought about particular objects. I have sympathy for this position. It may be that thought about particulars that is so cheap as to be satisfactionally grounded is not thought about particulars worth having. Certainly, satisfactional grounds and non-satisfactional grounds are metaphysically different sorts of subject-object relations. One might conclude that the obtaining of these different sorts of relations in various cases does not ground the obtaining of any unified relation of thinking-about. Rather, they are simply different relations that ordinary talk incorrectly lumps together. (In this respect, locutions such as "thinking about" or "belief about" would be like "jade.") 14 Nothing I say in this paper tells against such anti-bifurcationist positions. My aim is to explore the question of in which sorts of cases thought about particular objects has non-satisfactional grounds, either instead of or in addition to having satisfactional grounds. Anti-bifurcationists of the sort just described might recast this, instead, as the question of which particulars are really thought about: is it only ourselves and/or our own experiences that we really think about, or do we really think about lots of other particular objects? Nonetheless, in what follows I will speak of the quest for a sensible cognitive bifurcationism. The question of whether our ordinary talk of thought about particular objects marks a unified kind of relation is something I must set aside for future investigation. I aim to move toward a sensible cognitive bifurcationism by reflecting on Kripke's critique of what he called the "description theory" in Naming and Necessity. There is nothing new in seeing Kripke's critique as important for the question of how we think about, as well as refer to, particular objects. But it is illuminating to interpret Kripke as a linguistic bifurcationist, and to consider his critique of description theory as well as responses to that critique in this light. As I will try to show, doing so provides valuable lessons about how to reach a sensible cognitive bifurcationism.

(4) Kripke's Linguistic Bifurcationism
In Lecture II of Naming and Necessity, Kripke delivers his famous critique of a view of proper name reference that he calls the "description theory". One interpretation of the description theory as presented by Kripke is as an answer to a linguistic version of the 14 Along these lines, Almog (2012Almog ( , 2014 insists on a sharp distinction between reference and designation (or denotation). See also Capuano (2012).
Grounding Question: In virtue of what are particular individuals referred to by particular (uses of) linguistic expressions? The answer given by the description theory is roughly as follows: a use of an expression refers to a particular object in virtue of that object, uniquely, coming closest to instantiating a (complex) property that the user believes to be uniquely satisfied and associates with her use of the expression. Accordingly, one interpretation of Kripke's criticism of the description theory is as a rejection of that answer. It is a rejection of that answer in favor of a bifurcationist answer: linguistic reference may be grounded satisfactionally (i.e., it may be grounded in the instantiation of speaker-associated properties), but it may also be grounded non-satisfactionally.

Grounding claims vs identity claims
Before I launch into discussing Kripke's criticism in this light, I will defend the reasonableness of the interpretation. Actually, I am not sure that it is reasonable from an exegetical point of view, but I will explain why it is interesting to take it on board. (It may be that I am taking another cue from Kripke and introducing a separate character, Pepp's Kripke, whose approach to the linguistic version of the Grounding Question interests me.) For, arguably, Kripke viewed the description theory as an answer to the question, what is it for an expression to refer to a particular object?, where the answer being sought is a "philosophical analys[is] of some concept like reference, in completely different terms which make no mention of reference." (Kripke 1980: 94) It is easy to confuse the question of what it is for an expression to refer to a particular object with the question of what grounds the fact that an expression refers to a particular object. But they are different questions. 15 In this sub-section, I will illustrate the difference by expression to refer to that object. People who associate only a property like being a famous physicist with the name 'Richard Feynman', and do not believe this to pick out anyone uniquely, nonetheless refer to Richard Feynman by using that name. So it is not necessary.
And people who associate the property having invented the atom bomb with the name 'Albert Einstein', and do believe this to pick someone out uniquely, nonetheless refer to Albert Einstein by using that name. So it is not sufficient. (82-5) This failure of necessity and sufficiency would not imply that linguistic reference is never satisfactionally grounded, but it would imply that linguistic reference is not always satisfactionally grounded. Kripke's counterexamples thus support linguistic bifurcationism.
A prominent line of defense of the description theory against Kripke's criticism claims that the empirical examples Kripke describes do not show that the description theoretic condition on reference fails of necessity or of sufficiency. For no matter how ignorant speakers may be about the objects they refer to, these objects do always uniquely satisfy a property that the speaker believes to be uniquely satisfied and associates with the expression she uses. This is suggested by Kripke's own methodology of appealing to our intuitions that the speakers in his examples refer to Richard Feynman and to Albert Einstein. For these speakers themselves-being not importantly different from us--would, if given full information about the world, agree that they refer to those individuals, despite the incompleteness or falsity of their beliefs.
This shows that the speakers do associate the names with properties that the referents came closest to uniquely satisfying. They associate the names with complex disjunctive properties given by their intuitions about what their uses of 'Richard Feynman' or 'Albert Einstein' refer to, given that the world is thus and so. Speakers may not be able to express these properties with linguistic descriptions, but they associate them with the names nonetheless. Indeed, Kripke's own "rough statement of a theory" is an attempt to articulate the property that he would associate with these and other names. 17 Just to give it a name, let me call this line of defense the "implicit association reply". The core assumption of the implicit association reply is that there is some general condition under which an object is the referent of a use of an expression, and that speakers implicitly grasp this condition, even if they cannot express it in words. If this is right, then something like the following is true: for a use of an expression to refer to a particular object is for that object to uniquely satisfy C in relation to that expression (where C is the implicitly grasped condition).
Call this the "reference-satisfaction identity claim." Kripke expresses doubt about the 17 See, e.g., Jackson (1998: 212), Chalmers (2012: 19ff). reference-satisfaction identity claim because he does not think that such a condition could be "spelled out in terms independent of the notion of reference." (1980: 88, note 38) But according to the implicit association reply, this does not matter: the reference-satisfaction identity claim holds, whether or not C can be spelled out in terms independent of the notion of reference.
It is important to note that the reference-satisfaction identity claim is just that: an identity claim. It says that what it is for an expression to refer to a particular object is for that object to uniquely satisfy C. Ipso facto, the reference-satisfaction identity claim is not a grounding claim, since (I am assuming) no fact grounds itself. So, even if we assume that the implicit association reply rescues the description theory considered as a reference-satisfaction identity claim, this does not imply that it rescues the description theory considered as a grounding claim. As a grounding claim, the description theory says that the speaker's grasp of the relevant condition, whether implicit or explicit, is part of what makes it the case that her use of an expression refers to a particular object. Thus, to extend the implicit association reply to defend the description theory considered as a grounding claim, one would need to add that speakers' beliefs that the referents of their expressions uniquely satisfy C do grounding work.
Or, at least, that these beliefs are candidates for doing grounding work. This is dubious. To see why, imagine making a similar reply to defend an analogous sort of grounding-oriented description theory, which I will call the "photographic description theory." This theory holds that a photograph is of a particular object in virtue of that object, uniquely, coming closest to instantiating a property that the photographer believes to be uniquely satisfied and associates with the photograph. It is easy to see how one might produce Kripkestyle counterexamples to such a theory. For instance, suppose a photographer absent-mindedly takes a photograph and associates with it only a property like being a man in a red shirt. Or suppose she thinks the photograph is of the tallest person in Sweden, but actually it is of that person's twin, who is a centimeter shorter. Despite these beliefs, the first photograph is of John, and the second is of the twin. This suggests that these photographs' being of these individuals is grounded in some other way than by satisfying the photographer's associated beliefs. Now imagine that a defender of this photographic description theory argues that even in such cases, there is always a satisfactional ground for the photograph being of the object it is in fact of. For we--the photographer included--can identify what a photograph is of given enough information about how things are. This shows that whenever a photograph is of something, the photographer does associate the photograph with a property that she believes to be uniquely satisfied, which the thing the photograph is of comes closest to uniquely satisfying.
As before, one can see how this reply might defend the view that what it is for a photograph to be of an object is for that object to satisfy some condition, which is implicitly grasped by photographers. But to defend the grounding-oriented photographic description theory, one would need to add that the photographer's associating this condition with the photograph is at least a candidate for making it the case that the photograph is of a particular object.
I take it that this is not an attractive option. The grounding-oriented photographic description theory is clearly false. Photographers' attitudes do not play any role in making photographs be of particular objects, even if it is true that photographers uniformly grasp what it is for a photograph to be of a particular object. This raises the question of whether there is reason, when it comes to linguistic expressions, to inject speakers' grasp of what it is for an expression to refer to a particular object (assuming they have such grasp) into the grounds of this referring. For present purposes, I simply wish to point out that the implicit association reply needs to provide such a reason if it is to defend the description theory considered as a grounding claim. On the other hand, if the description theory is taken as a referencesatisfaction identity claim, then the implicit association reply, if correct, provides a direct defense against Kripke's counterexamples.
There is some evidence that Kripke saw the description theory as a reference-satisfaction identity claim, so that something like the implicit association reply might rescue it from his counterexamples directly. He reports that Robert Nozick suggested a reply in this vicinity, and he seems to agree with Nozick that the reply could save (at least the letter of) the description theory. 18 If so, this would be a "trivial fulfillment of the description theory", according to Kripke, and he would be content to have shown that the condition featuring in the description theory "must be one of a completely different sort from that supposed by Frege, Russell, Searle, Strawson and other advocates of the description theory." (1980: 88, n. 38) But if the description theory is understood as a grounding claim, then it is not so easily rescued by Nozick's reply or by the implicit association reply. Such replies would have to be supplemented with considerations in favor of treating speakers' implicit grasp of what it is for 18 Except for the aforementioned concern about not being able to spell out the condition. See Kripke 1980: 88, footnote 38. an expression to refer to an object as part of the grounds of that reference. Given this, whatever Kripke's own view of the description theory, it seems to me that the theory is more interesting when considered as a grounding claim. So considered, it is not made true simply by speakers' having an implicit understanding of linguistic reference. In the remainder of this section, I will interpret the description theory in this way. Doing so casts Kripke's arguments against the theory as a plea for linguistic bifurcationism.

Kripke's method for reaching a sensible linguistic bifurcationism
Kripke does not claim that the description theory is universally false about what grounds linguistic reference. He acknowledges that a speaker's use of an expression may refer to a particular object in virtue of that object uniquely coming closest to instantiating the associated property: ...one determines the reference for himself by saying--'By "Gödel" I shall mean the man, whoever he is, who proved the incompleteness of arithmetic'. Now you can do this if you want to. There's nothing really preventing it. You can just stick to that determination. If that's what you do, then if Schmidt discovered the incompleteness of arithmetic you do refer to him when you say 'Gödel did such and such'. Kripke's method for reaching a sensible bifurcationism is a method of counterexample and extrapolation. The method has three steps. First, describe cases in which it is intuitively compelling that an expression refers to a particular object without uniquely instantiating a property that the speaker associates with the expression and believes to be uniquely instantiated. (These are counterexamples to the view that the description theory is universally correct.) Second, conclude from this that in such cases reference would be grounded in some

19
Cf. Jackson (1998: 205). other way. Third, extrapolate that these alternative grounds would also obtain in a range of other cases (even if description-theoretic grounds also obtain in those cases), but would not obtain in another range of cases. For example, it is intuitively compelling that the speakers in Kripke's 'Feynman' example refer by 'Richard Feynman' to Richard Feynman, even though (according to Kripke) Feynman does not uniquely satisfy any property that they believe to be uniquely satisfied and associate with their use of the name. So, their reference to Feynman must be grounded in some other sort of fact, such as (Kripke suggests) the causal connection between Feynman's being named and the speakers' acquisitions of the name. It seems that this same sort of ground would obtain even in cases where the speakers do associate the name with a property that Feynman uniquely instantiates. For even in such cases, the speakers will have acquired the name in the same sort of way and have the same sort of causal connection to its introduction. Thus, even in such cases reference to Feynman is (at least also) grounded in a way that does not accord with the description theory. On the other hand, such grounds would not obtain in a case in which the speaker introduces a new name for an object by the kind of descriptive stipulation set out in the quoted passage above. This is not to say that Kripke's examples are definitive against the grounding version of description theory. The ignorant users of the name 'Richard Feynman' might well believe that something is uniquely the person referred to by 'Richard Feynman' in most uses of the name that they have encountered. Their association of this property with the name, together with this property being uniquely instantiated by the man Richard Feynman, might ground the fact that their uses of 'Richard Feynman' refer to Richard Feynman. Of course, as Kripke points out, it will not do so if the facts that all these other uses refer to Feynman are themselves partially grounded in such descriptive "borrowing". For this would be a situation in which the reference of at least some uses of 'Feynman' would have to be grounded circularly, and I have assumed there is no reflexive grounding. But Kripke's example as presented is not one in which it is clear that this is the case.
To proceed with the method of counterexample and extrapolation, the example would need to be described in such a way as to rule out the envisioned chain of grounding via descriptive borrowing. I suspect this could be done. But instead of pursuing it further, I now switch onto the parallel track that is my main focus: applying Kripke's method instead to the quest for a sensible cognitive bifurcationism.

(5) Application of Kripke's method to the quest for a sensible cognitive bifurcationism
I argued in the last section that the fact (if it is one) that speakers have an implicit idea of what it is for an expression to refer to a particular object does not defang Kripke-style counterexamples to description theory considered as a grounding claim. Nonetheless, it does present an obstacle in applying Kripke's method to cognitive bifurcationism. To do so, we need to describe clear cases in which a subject thinks about a particular object but there is no satisfactional ground for this thought. (In what follows, I will sometimes refer to such cases simply as "counterexamples". They are counterexamples to the view that all grounds for thought about particulars are satisfactional grounds.) It is tempting to follow Kripke in appealing to cases in which the subject is quite ignorant concerning the object she thinks about. The problem with this approach is that in such cases the subject invariably has what Frank Jackson (1998: 209) calls an "information-carrying trace" with a causal source in the object. The subject may have a name for the object, an episodic memory of the object (however hazy) or an immediate perceptual experience of the object (however indistinct).
And, if the implicit association reply considered in the previous section is on the right track, the object will uniquely satisfy a condition that the subject believes (implicitly) to be uniquely satisfied. Roughly, it will uniquely satisfy the condition of being a certain sort of causal source of that name, memory, or experience. Thus, a satisfactional ground for the subject's thought about the object will obtain. Importantly, the proposed satisfactional ground does not rely on the subject's having an implicit idea of what it is to think about a particular object. The suggestion is not that the subject's thought about the object is grounded in her implicit belief that something is uniquely the thing she thinks about. Instead, on the proposal under consideration, thought about particulars is often grounded in beliefs about other stuff: reference, memory and perception, for instance.
The viability of this proposal depends, among other things, on whether subjects in general have these sorts of beliefs. This is a long-debated matter. The nays argue that the hypothesis that subjects have such beliefs over-intellectualizes subjects, especially children, and the yeas respond that the relevant implicit beliefs do not require as much cognitive sophistication as the nays suppose. 20 21 It seems to me that this back-and-forth is so far inconclusive. This is partly because it turns on empirical questions concerning how subjects of varying levels of cognitive sophistication are equipped to respond, in their thought and in their actions, should they accept as correct various representations of how things are from their point of view. These include questions about subjects' ability to understand and apply various kinds of representations, as well as questions about their grasp of the workings of language, memory and perception.
Answering these questions in order to produce more conclusive counterexamples or more conclusive rebuttals of those counterexamples is a worthwhile endeavor. But it is not one I can take up here. Instead, I will introduce a different sort of counterexample that avoids these thorny empirical matters. One type of case in which there is thought about a particular object without satisfactional grounds for that thought are cases in which a subject thinks about an existing object, but believes that it does not exist.
For instance, suppose that a subject falsely believes that she is suffering from a visual hallucination. In fact, she is seeing normally. She sees a black cat in front of her and believes that she is hallucinating a black cat. In such a case, the subject does not believe that anything uniquely satisfies a condition like being a black cat in front of me, or even being the cause in a certain way of a certain aspect of my current visual experience. To the contrary, she believes that nothing satisfies such conditions. (She may believe that something is causing her hallucinatory experience, but not in the "certain way" that a cat would cause a veridical version of the experience.) Nonetheless, I would submit that she does think about this particular cat (the one she in fact sees). She believes (for instance) that it is black, cute, fuzzy, and so on. She also believes (falsely) that it does not exist.

21
A different version of this sort of proposal, suggested by Chalmers (2002bChalmers ( , 2002cChalmers ( , 2006Chalmers ( , 2012, does not include subjects' beliefs, however implicit, in the satisfactional grounds for thought about particulars. Instead, Chalmers might be taken to suggest that a subject's having thought about a particular object is grounded in the object's being determined by the "epistemic intension" of a concept that the subject employs. The epistemic intension of a concept is a function from centered worlds to objects, and corresponds roughly to the disposition of an idealized reasoner in the subject's position to identify the extension of the concept (Chalmers thinks of concepts as expressions in a language of thought), when accepting as veridical a certain representation of how the world is (from the subject's point of view). But this approach seems to me to address the question of what it is for an object to be thought about, rather than the grounding question. According to Chalmers, what it is for a subject to think about a particular object is for the object to uniquely satisfy the condition that an idealized reasoner would track in selecting the extension of a concept the subject employs across different scenarios. If, in fact, satisfying such a condition is what it is to be thought about, it cannot also be what grounds an object's being thought about.
At least, these claims strike me as correct. Perhaps someone will object that in such a case the subject does not believe or think about the cat at all; that her thought about it is somehow blocked by her belief that it does not exist. I would need to hear more about why belief in an object's non-existence blocks thought about the object, for I do not immediately see a reason to accept this. If I am right that this subject thinks about the particular cat she sees, then it is not clear what satisfactional ground obtains for this thought about the cat. There is no condition that the subject believes to be uniquely satisfied such that the cat uniquely satisfies that condition, since the subject believes that all candidates for such a condition are not satisfied.

Cases of negative existential beliefs about existing objects thus look like compelling
examples of thought about particulars in the absence of a satisfactional ground for that thought. Assuming that thought about objects in such cases is grounded, it must have nonsatisfactional grounds. But if the fact that the cat is thought about by the subject has nonsatisfactional grounds in the case described, then it seems likely that those grounds would also be present in a case that is the same except that the subject lacks the false belief that she is hallucinating. Following Kripke's method, it seems that we can extrapolate from there being non-satisfactional grounds for the thought in the case described to there being nonsatisfactional grounds in run-of-the-mill cases of thought about perceived particulars as well.
(But more on this in the next section.) In the latter kinds of cases, typically there will also be satisfactional grounds for such thought. But this does not crowd out the non-satisfactional grounds.
It is also worth noting that parallel cases might be used to argue against the description theory (as articulated by Kripke, and taken as a grounding claim). For instance, imagine a speaker who learns about Richard Feynman in physics class, knows a good deal about him, but also falsely believes that he did not really exist. This speaker does not associate any property with their use of the name 'Richard Feynman' that they also believe to be uniquely instantiated. Still, they refer to Feynman.

(6) Extrapolation from cases
In section 5, I focused on the first step of Kripke's method of counterexample and extrapolation. I acknowledged the difficulty of finding compelling examples of thought about particulars in the absence of a satisfactional ground for that thought, and I described a type of example that does not depend on difficult empirical questions about cognitive capacities.
This kind of exercise is critical to making a start on the quest for a sensible bifurcationism.
But in this section, I wish to emphasize that even if successful, it is only the first step in Kripke's method of counterexample and extrapolation. The method aims to draw lessons from counterexamples about how to bifurcate thought about particulars into the purely satisfactionally grounded and the (at least also) non-satisfactionally grounded. To do this, one needs to extrapolate from clear examples of thought about particulars without satisfactional grounds to other cases of thought about particulars, including those in which there is (or at least, it is not clear that there is not) a satisfactional ground for such thought. And to do this, one needs to have some idea of what the non-satisfactional grounds in the clear cases are, so that one can assess whether similar non-satisfactional grounds would obtain in other cases.
For instance, suppose I am right that in the case of falsely believing that a cat you are seeing does not exist, you think about that cat in the absence of a satisfactional ground for your thought. This is an unusual kind of case. It is important to the quest for a sensible bifurcationism mostly to the extent that it has implications for more ordinary cases, such as those where I see a cat and do not falsely believe that it does not exist. In supposing that the case has such implications, as I did at the end of the last section, I am supposing that the type of non-satisfactional ground that obtains for the false hallucinator's thoughts being about the cat also obtains in the more ordinary case. The type of ground that I am supposing obtains in both types of case has something to do with the way in which visual perception of cats gives rise to thought. That I am presuming something along these lines also explains why I do not take the case of the false hallucinator to suggest non-satisfactional grounds for my thought about (say) the tallest second cousin of the man in front of me in the supermarket line. I do not take it to suggest this because I do not have the same sort of perceptual link to that individual.
So, in order to use counterexample cases (i.e., cases in which it is clear that there must be non-satisfactional grounds for thought about a particular) to move toward a sensible cognitive bifurcationism, we also need positive accounts of the non-satisfactional grounds that are in place in those cases. Such accounts are not additional or subsequent to the method of counterexample and extrapolation, but an integral part of that method. Likewise, if Kripke's criticism of description theory is a plea for linguistic bifurcationism, then Kripke's "chain of communication" picture of what grounds linguistic reference in typical cases is part of that criticism, not an add-on. 22 23 To be sure, some rigorous attempts to specify non-satisfactional grounds for thought about ordinary, external particulars have been made. 24 These might be applied within Kripke's method of counterexample and extrapolation to reach a sensible bifurcationism. To assess their applicability, a bifurcationist would have to assess whether they are compelling as accounts of the non-satisfactional grounds for thought about particulars in whichever cases she judges to be genuine counterexamples.
Carrying out such assessments is another matter that I must leave for another occasion.
What I do wish to highlight, though, is that extrapolation should be of interest to any cognitive bifurcationist--to anyone who holds that thought about particulars may be grounded satisfactionally but may also be grounded in other ways. Consider a bifurcationist who does not accept any Kripke-style cases as genuine cases of thought about particulars without satisfactional grounds, including the case of false belief in a particular object's non-existence that I presented in section 5. She might nonetheless accept that thought about oneself and/or one's own experiences is grounded non-satisfactionally, for the reasons discussed in section 3.
In order to use these cases to move toward a sensible bifurcationism, she needs to draw out their implications for other sorts of cases. For instance, perhaps the fact that some thought about particulars to which the subject stands in certain special, perspectival relations (e.g., thought about oneself) must be grounded non-satisfactionally suggests that other thought, such as thought about one's own experiences, is also grounded non-satisfactionally. (This is roughly the Russellian view). But to draw this conclusion, one must have some idea in the background about what the non-satisfactional ground in such cases would be. Conversely, suppose that one does not extrapolate from there being non-satisfactional grounds for (say) thought about one's own experiences to there being non-satisfactional grounds for thought about (say) the external objects one immediately perceives. Then one needs to have a background idea about what sort of non-satisfactional ground there is for the former, such that it does not obtain for the latter.

23
So I agree with Dickie (2015: 154) that we cannot rest content with an exercise of coming up with putative cases of thought about particulars in the absence of a satisfactional ground, evading the cases by showing how a satisfactional ground would be present after all, changing the cases to remove this ground, and so on. But this is not because the method of counterexample and extrapolation is at the wrong "level of explanatory depth", as she claims, but because the extrapolation step cannot be ignored.

24
See references in note 3.
One type of background view is the one mentioned in section 3, that what grounds thought about oneself or about one's own experiences is a kind of Russellian acquaintance with these particulars. Perhaps it is compelling that the kind of direct and immediate awareness one has of oneself and one's own experiences makes it the case that one thinks (or is positioned to think) about oneself. On the other hand, it is less clear, and much debated, whether subjects also have this kind of awareness of external objects that they perceive, remember, and hear about. The decision to extrapolate or not to extrapolate from cases of thought about oneself or one's experiences to cases of thought about other particular objects requires some resolution of such issues.
Another type of background view would follow Lewis and Jackson in claiming that all thought is de se. If this is right, then there is a non-satisfactional ground for thought about oneself: you think about yourself by virtue of thinking at all, since it is built into the structure of thought to be about the thinker. However, just the fact of there being thought does not ground thought about anything other than the thinker. So, on this view of what grounds thought about oneself, there is no reason to extrapolate that such non-satisfactional grounds obtain for thought about other particulars. Of course, the claim that all thought is about the thinker is a theoretical claim, not a commonsense one, so this view needs to be independently motivated. But if one accepts it, one can marshal it to explain why one should not extrapolate from the case of thought about oneself to cases of thought about other particulars.
In sum, to use Kripke's method to reach a sensible cognitive bifurcationism, we must continue to seek compelling examples of thought about particulars in the absence of satisfactional grounds. But we must also figure out, at least in broad outline, what makes it the case that particulars are thought about in these cases, so that we know how to extrapolate to other cases. And "we" here does not only refer to those who believe that there are nonsatisfactional grounds for thought about particulars other than oneself and one's experiences. It refers to all bifurcationists, since all bifurcationists posit non-satisfactional grounds in some cases but not others, and all need justification for these choices. 25

(7) Conclusion
The aim of this paper has been to move toward a sensible cognitive bifurcationism by

25
In this section I am grateful for discussion and correspondence with Pär Sundström. reflecting on Kripke's method for reaching a sensible linguistic bifurcationism. I recast Kripke's approach to linguistic reference as parallel to inquiry into what grounds thought about particulars. I showed how this puts Kripke's arguments, and one influential line of reply to them, in a different light, and highlights the implications of that line of reply for the Grounding Question. I then proposed a new resource for the counterexample stage of Kripke's method: cases of false belief in the non-existence of existing objects. Finally, I emphasized the importance of the other part of the approach, extrapolation, not only for those who wish to argue for expansive positions on the range of non-satisfactionally grounded thought about particulars, but also for those who wish to argue for limited ones.