Francois Recanati

It is commonly held that all truth-conditional effects of context result from a pragmatic process of value-assignment that is triggered (and made obligatory) by something in the sentence itself, namely a lexically context-sensitive expression (e.g. an indexical) or a free variable in logical form. Such a process has been dubbed ‘saturation'. It stands in contrast to so called ‘free' pragmatic processes, which are supposed to take place for purely pragmatic reasons — in order to make sense of what the speaker is saying. For example, the pragmatic process through which an expression is given a nonliteral (e.g. a metaphorical or metonymical) interpretation is context-driven: we interpret an expression nonliterally in order to make sense of the speech act, not because this is dictated by the linguistic materials in virtue of the rules of the language. The dominant view, then, is that no free pragmatic process can affect truth-conditions — such processes can only affect what the speaker means (but not what she says). But there is a dissenting position, according to which free pragmatic processes can take place locally and interact with semantic composition, thereby affecting truth-conditions. Three such processes have been discussed in the literature : pragmatic enrichment, predicate transfer, and loosening/broadening. Sometimes ‘pragmatic enrichment' is used as a cover term for these modulation processes because, on a certain understanding, they all result from ‘enriching' the logical representation which is the output of the translational phase of semantic interpretation. The paper presents an overview of this research area.

I argue that immunity to error through misidentification primarily characterizes thoughts that are 'implicitly' de se, as opposed to thoughts that involve an explicit self-identification. Thoughts that are implicitly de se involve no reference to the self at the level of content: what makes them de se is simply the fact that the content of the thought is evaluated with respect to the thinking subject. Or, to put it in familiar terms : the content of the thought is a property which the thinking subject self-ascribes (as in the Loar/Lewis/Chisholm analysis). After answering an objection (to the effect that immunity can affect explicit de se thoughts), I extend the analysis to demonstrative thoughts, which also exhibit the property of immunity to error through misidentification.

Mental files serve as individual or singular concepts. Like singular terms in the language, they refer, or are supposed to refer. What they refer to is not determined by properties which the subject takes the referent to have (i.e. by the information stored in the file), but through relations to various entities in the environment in which the file fulfills its function. Files are based on acquaintance relations, and the function of the file is to store whatever information is made available through the relations in question. I offer a typology of files. The most important distinction is between proto-files and conceptual files. In contrast to proto-files, conceptual files can host not only information derived through the specific relation on which the file is based, but also information about the same object gained in some other way. In this framework identity comes into the picture twice. (i) Identity is presupposed when two pieces of information occur in the same file. Such 'presumptions of identity' ground the linguistic phenomenon of de jure coreference, which takes place when two singular terms, or two occurrences of a singular term, are associated with the same file. (ii) Judgments of identity work by linking two distinct files, thereby enabling information to flow freely between them. This corresponds to de facto coreference. (Linking is not merging ; identity judgments have the effect of merging files only when the files belong to a very specific category, that of 'encyclopedia entries' -- a type of conceptual file based on a higher-order relation rather than on a specific acquaintance relation.) In the last part of the paper I will discuss, and attempt to rebut, two objections to the mental-file account. According to the first objection, the account is circular ; according to the second objection, de jure coreference cannot be accounted for it in terms of identity of the associated mental files because de jure coreference is not a transitive relation.

Mental files have a referential role—they serve to think about objects in the world—but they also have a meta-representational role: when ‘indexed’, they serve to represent how other subjects think about objects in the world. This additional, meta-representational function of files is invoked to shed light on the uses of empty singular terms in negative existentials and pseudo-singular attitude ascriptions. -/- For a longer version see "Empty Singular Terms in the Mental-File Framework" In Manuel Garcia-Carpintero & Genoveva Marti (eds.), Empty Representations. Oxford 162-185.

In this paper, I argue that singular thought about an object involves nondescriptive or de re ways of thinking of that object, that is, modes of presentation resting on contextual relations of ‘acquaintance’ to the object. Such modes of presentation I analyse as mental files in which the subject can store information gained through the acquaintance relations in question. I show that the mental -file approach provides a solution to a vexing problem regarding the communication of singular thoughts: If singular thoughts depend upon contextual relations to the objects of thought, how can they be communicated across contexts? What makes communication possible when the speaker and the addressee do not stand in the same contextual relations to the objects the speaker’s thought is about?

Francois Recanati presents the basic features of the *indexical model* of mental files, and defends it against several interrelated objections. According to this model, mental files refer to objects in a way that is analogous to that of indexicals in language: a file refers to an object in virtue of a contextual relation between them. For instance, perception and attention provide the basis for demonstrative files. Several objections, some of them from David Papineau, concern the possibility of files to preserve and add information about objects across contexts. How is it possible to think about the same object when the subject no longer is in the original context? How is it possible to think of a perceived object as already known? Can this be done without an explicit identity judgment? Recanati answers these questions by invoking mental files of non-basic kinds and by describing the cognitive dynamics in which they take part.