Robin Jeshion, New Essays on Singular Thought. Oxford. Oxford University Press. 2010. xii + 324 pp. The contributors to this thought-provoking and timely collection of essays all begin from a familiar and intuitive distinction between singular/de re/purely referential thoughts, on the one hand, and general/de dicto/descriptive thoughts, on the other. In broad outline the distinction goes back to Russell. Thoughts of the first kind are directly about individuals in the world, in the sense that they directly contain those individuals. Thoughts of the second kind are about individuals indirectly, in virtue of those individuals satisfying some general characterization. The content of the thought is given by the general characterization, rather than by the individuals (if any) whom the characterizations picks out. Relatively little of the book is concerned with the question of whether there actually are such things as singular thoughts. The contributors largely take that issue to have been resolved by the work of Kripke and others on semantic direct reference, in addition to related work in the philosophy of mind and philosophy of perception. Recanati and Bach are exceptions here, with Recanati in particular giving a thorough analysis of some strategies that a “descriptivist’ must use to accommodate the phenomena that have led others to accommodate singular thoughts. But the other individual papers typically take the existence of singular thoughts for granted, focusing instead on the analysis of those thoughts, and in particular on what conditions there should be upon grasping them. There is a continuum of views here. At one extreme lies the position that Robin Jeshion terms semantic instrumentalism, according to which successfully deploying the linguistic apparatus of direct reference suffices for thinking singular thoughts. On this view, using a directly referential proper name is sufficient for having a singular thought about the referent of that name. It may even be (as discussed in Sainsbury’s paper) that the referent of that proper name is a fictional character, so that thinker can have singular thoughts about a non-existent entity. At the other extreme lies Russell’s own view, which ties the capacity to think a singular thought about a particular object to the thinker being acquainted with that object. As is well-known, at one stage of his career Russell held that the acquaintance requirement permitted us to have singular thoughts only about our own sense data, universals, and (possibly) ourselves. Many of the authors stake out positions on this continuum. The question of whether there is an acquaintance requirement upon singular thought is explored in the papers by Robin Jeshion (‘Singular Thought: Acquaintance, Semantic Instrumentalism, and Cognitivism’) and Francois Recanati (‘Singular Thought; In Defence of Acquaintance’). Recanati is much more sympathetic than Jeshion to the acquaintance requirement. He distinguishes singular thought-contents from singular thought-vehicles (which he thinks of in terms of the tokening of a mental particular involving a mental singular term). There is no requirement of acquaintance upon thinking a singular thought-vehicle (any more than there is on successfully using a singular term), but successfully thinking singular thought-contents does, he claims, require at least the expectation of acquaintance with the relevant object. This is because he thinks of singular thoughts in terms of non-descriptive modes of presentation, which in turn he thinks of in terms of mental files. Mental files store information about objects – information gained in virtue of the thinker standing in an acquaintance relation to the relevant object. A mental file can be opened, as he puts it, when acquaintance is imagined, or expected – but not when acquaintance is completely off the table. Jeshion also thinks of singular thoughts in terms of mental files, which she thinks of by analogy with the visual indices featuring in Zenon Pylyshyn’s theory of object perception. Pylyshyn’s visual indices have a deictic component that allows us visually to reference objects in the distal environment. Likewise object files contain mental demonstratives and/or mental names that allow us to think singularly about individuals. Opening a mental file upon a particular individual does not require acquaintance with that individual, but it does require that individual being in some sense significant to the thinker’s plans, projects, or affective states. But significance does not require acquaintance (and nor does acquaintance imply significance). I must admit to being rather troubled by appeal to mental files in this context. Mental files are repositories of information. But in what form does that information come? Both Jeshion and Recanati talk freely about object files containing mental terms. These mental terms include mental names and mental demonstratives which, one imagines, refer directly to their objects in precisely the manner of linguistic singular terms. But how exactly is this supposed to work? Do mental demonstratives, for example, have their own object files, in virtue of which they latch on to particular objects? Surely not, on pain of an obvious regress. But then how do they latch on to particular objects? But this is what a theory of singular thought is supposed to be explaining. In which case bringing mental files into the picture does not explain singular thought. It seems simply to push the problem a step further back. A similar problem threatens Kent Bach’s attempt in ‘Getting a Thing into a Thought’ to steer between semantic instrumentalism and the acquaintance requirement, but without appeal to mental files. For Kent Bach, singular thoughts are mediated by de re representations (as opposed to representations grounded in definite descriptions). These de re representations function as mental indexicals. They are grounded, he claims, on the thinker’s representational connection with the object being thought about – a connection that can hold in virtue of perception, memory, or communication. But what is it for there to be a representational connection holding in virtue, say, of memory? Talk of representation suggests that there is a connection at the level of thought. These cannot be singular thoughts, however, if the analysis is to be informative. They cannot be descriptive, because that would be to accept a form of reductionism/eliminativism about singular thought that Bach explicitly rejects. Singular thought theorists are not without potential resources at this point. There is a relatively developed literature on nonconceptual content that could be brought into play – particularly, discussions of nonconceptual content at the subpersonal level (as opposed to discussions of the nonconceptual content of perception). Also potentially relevant are the causal, covariational, or teleological theories of content that have been used to illuminate the intentionality of the (so-called) mental lexicon. There are some interesting and complex issues to be explored about the extent to which these theories can help make the appeal to mental files truly explanatory. It is a shame that they will have to wait for future work. The closer one moves towards semantic instrumentalism the less severe this problem is. Nathan Salmon’s ‘Three Perspective on Quantifying In’ offers a much more limited conception of what he terms de re connectedness with the objects of singular thought. His intricate paper analyzes Kaplan’s two commentaries on Quine’s famous paper ‘Quantifiers and Propositional Attitudes’. Kaplan originally proposed the notion of representingB in order to provide a more de dicto analysis of Quine’s de re notion of relational belief – a proposal that he [Kaplan] subsequently retracted on the grounds that the problem Quine set out to solve was misconceived. Salmon thinks that representationB can be put to work to explain de re connectedness. A term t representsB an individual x to a thinker y iff t designates x and is a vivid name of x for y. The idea that Salmon ends up with is that this notion, without the vividness requirement, can serve as a surrogate for Russell’s notion of direct acquaintance. The things that we can think about through singular thoughts are those things for which we have a name, in the sense of “name” on which “the author of Waverly” is a name, while “the shortest spy” is not. Salmon’s claim is that it is because “the author of Waverly” names Sir Walter Scott, while “the shortest spy” does not name the fabled Ortcutt, that we cannot cognitively access singular propositions about the shortest spy in the same way that we can cognitively access singular propositions about the author of Waverly. I wonder about this, though. It is true that we can think (singularly) of the author of Waverly that he is F, but we cannot think (singularly) of the shortest spy that he is F. But surely the direction of explanation goes the other way. It is because we can think singularly about the author of Waverly that “the author of Waverly” is a name – not vice versa. But then we are no closer, I suspect, to explaining in what de re connectedness consists. Despite these recurrent problems in getting a clear fix on de re connectnedness, none of the authors in this volume are prepared fully to adopt semantic instrumentalism. Some of the reasons for this are explored in Arthur Sullivan’s ‘Millian Externalism’, which persuasively argues that semantic instrumentalism (a.k.a. Millian externalism) is not entailed by the rejection of individualism of reference. He is very sympathetic to a broadly Fregean position combining anti-individualism and externalism about reference with an intensional approach to content. The most obvious way of developing this (partially) neglected alternative would be through the notion of object-dependent senses proposed by McDowell and Evans, among others. Apart from some discussion in Recanati’s paper, the general topic of object-dependent senses (and hence of a broadly Fregean approache to singular thoughts) is not explored in the collection as fully as it deserves. The closest that we come is in John Campbell’s paper ‘Demonstrative Reference, the Relational View of Experience, and the Proximality Principle’. Campbell defends the relational view of experience (according to which experience is a relation between the perceiver and an element of the distal environment, so that the qualitative character of experience is constituted by objects and properties about which we think demonstratively). His foil is Burge’s Proximality Principle, according to which perceptual states can be typed in terms of proximal stimulation (together with associated afferent and efferent input into the perceptual system). Against Burge he argues that, although the Proximality Principle may be a basic claim of vision science, the relational view is an account of perceptual consciousness. There is no conflict between the two because they operate at different levels of explanation. There are hints here of a strategy that may be effective against the circularity problem identified earlier for accounts of accounts of de re connectedness. This theme of perception as involving direct acquaintance with objects is continued in Imogen Dickie’s ‘We Are Acquainted with Ordinary Objects’. Dickie draws on experimental evidence to argue that an account of acquaintance-based singular thoughts can be constructed from empirical results on attention. Experiments show that attention is sometimes directed, at the preconceptual level, to objects in the psychologists’ sense. Even though these objects are not objects in the full-blooded philosophers’ sense – they could be geometric shapes, for example – Dickie emphasizes empirical evidence that the attentional channel is closed down by apparent violations of ordinary objecthood (i.e. geometrical shapes that are not real objects). She uses this to distinguish between acquaintance-based singular thoughts and illusory singular thoughts. The latter occur when the thinker attends to something that is not a real object, but the attention channel remains open. Kenneth Taylor’s ‘On Singularity’ gives us another way of thinking about illusory singular thoughts. He distinguishes singularity of content, which is what we have been primarily discussing up to now, from singularity of form. A mental item has singularity of form to the extent that it is apt to represent an individual item – to the extent, that is, that it is apt to be de re connected to an object. But of course that connection can fail to hold, in which case we have singularity of form without singularity of content. A similar distinction is made in Mark Sainsbury’s ‘Intentionality without Exotica’ A thought is externally singular when there is an object the thought is about. It is internally singular when there is no such object, but the thought recruits the resources appropriate to external singularity. Both Sainsbury and Manual Garcia-Carpintero in ‘Fictional Singular Imaginings’ attempt to account for illusory singular thoughts that might be inspired, for example, by fiction. Garcia-Carpintero thinks that fictional discourse is, strictly speaking, “gappy” (neither true nor false). But nonetheless we need to distinguish between the thoughts expressed by, on the one hand, the sentence “Leopold Bloom ate the inner organs of beast and fowl” and, on the other, “Ulysses ate the inner organs of beast and fowl”. Garcia-Carpintero’s solution is that singular thoughts have both singular and descriptive contents. In fictional discourse the descriptive contents are the ones that we are supposed to imagine. Sainsbury focuses in particular on what he terms V-sentences – sentences formed from intentional transitives (e.g. ‘John is thinking about sentences’) – as opposed to O-sentences, in which propositional operators are applied to a sentence (e.g. ‘John is thinking about Pegasus’). Internal singularity of form in O-sentences can, he maintains, be dealt with using the free logic and other tools developed in his book Realism without Reference. The contribution of this paper is to show how this account can be extended to V-sentences. He appeals to principle (*): If p entails q then the ontology of q does not exceed that of p. Because V-sentences are typically entailed by O-sentences, even if not paraphrasable in terms of them, principle (*) ensures that V-sentences incur no new ontological commitments. The contributors to this volume cover a wide range of issues about singular thoughts. As I have noted, some important issues remain unexplored. More detailed discussion of object-dependent senses would have been welcome. So too would further exploration of how exactly de re connectedness is supposed to work. But the editor deserves our thanks for drawing together these thought-provoking papers and making it easier to see some of the profitable areas for future work on singular thought.