Duration: 3h

An influential view of concepts distinguishes between stable concepts (‘mental files') and unstable conceptions (entries ‘inside' files) (Rey, 1983; Recanati, 2012, 2016; Murez & Recanati 2016). Concepts combine systematically into complex thoughts, and ground inferential dispositions by encoding mental coreference. They also provide access to conceptions, changing bodies of information which subserve categorization, induction, etc.

This symposium reexamines the distinction between concepts/files and their entries in light of recent theoretical and empirical advances, highlighting its relevance in several areas: (1) rational relations between attitudes (2) the language/thought interface (3) the interface between conceptual processes and nonconceptual systems and (4) pre-conceptual object tracking.

(1) François Recanati (Collège de France) Mental files, concepts, and modes of presentation

Many philosophers have emphasized the need to distinguish between concepts and conceptions. The distinction can be formulated by appealing to the mental file idea: concepts are mental files, while conceptions correspond to their contents (the information, or misinformation, stored in the file). Mental files persist through changes in conception. They are continuants, and, like other continuants, should be individuated by their origin (Sainsbury and Tye 2012) rather than by the contents they carry at a given time.

In Frege cases, one and the same entity is represented under two distinct ‘modes of presentation' (Frege 1892). Thus a rational subject may accept that Hesperus is F while not accepting that Phosphorus is F. In such a case, the relevant concepts (of Hesperus and Phosphorus) must be distinct because, if they were the same, the thoughts towards which the subject holds conflicting attitudes would be the same, and that would impugn the subject's rationality. So we need to posit two distinct mental files in the subject's mind. On the other hand, if the files in question were associated with exactly the same conception and differed only numerically, no Frege case could arise (the subject would have no reason to hold different attitudes towards ‘Hesperus' and towards ‘Phosphorus'). This suggests that modes of presentation should be equated with mental files qua carrier of certain contents – ‘thick files' rather than ‘thin files'. A thick file is a file-stage, a file-at-a-time, together with the contents (the conception) it carries at that time. Conceptions are thus integral to modes of presentation, while concepts (‘thin files') are independent of conceptions.

(2) Michael Murez (CAPHI, Nantes & Collège de France) Polysemous mental files

According to a broadly Fregean approach, concepts/files are transparent, i.e., subjects know immediately whether they are deploying the same or different concepts. Different concepts may corefer, unbeknownst to the subject. But occurrences of the same concept corefer de jure, from the subject's perspective.

Recently, this last Fregean thesis has been contested. Quilty-Dunn (2020) argues that concepts, like meanings, can be polysemous. Just as speakers know that the reference of a polysemous expression – like “chicken” – may vary, occurrences of the same concept can refer to different things from the subject's own perspective.

I will grant the existence of conceptual polysemy and argue that it additionally threatens the transparency of concepts. Nevertheless, I will ultimately defend the Fregean approach, by distinguishing between different varieties of polysemy, only some of which involve polysemous concepts, and between different degrees of transparency for concepts (files) and for ‘modes of presentation' (file-labels deployed in working memory with activated subsets of their entries). I will also draw on another Fregean idea: the force/content distinction. Just as predication can occur without assertion in language, mental files enable coreference to be temporarily ‘simulated' in thought, without doxastic commitment.

(3) Nicholas Shea (Institute of Philosophy, University of London) Concepts as an Interface

The approach to concepts I advance in Concepts at the Interface (2024, OUP; esp. ch. 5) makes a sharp separation between concept qua mental representation used in active thinking and reasoning, on the one hand, and concept qua store of information about a subject matter. As a terminological choice, I reserve ‘concept' for the former. We reason with concepts in working memory, for example in performing logical inference, but conceptual thought can also draw on a much wider suite of informational models and rely on the inferences they support. When a concept is tokened in thought, the store of information to which it affords access is thus not limited to further conceptual representations (psychologists' semantic memories). It includes representations in sensory, motor, affective and evaluative systems. A conceptual thought in working memory can serve to sustain and manipulate representations in special-purpose systems. Concepts thus act as an interface between general-purpose conceptual thought and special-purpose informational models. A concept is, then, a ‘plug-and-play device' that connects deliberate thinking to simulations and inferences in sensory, motor, affective and evaluative systems, while also plugging into compositional structures in working memory that support general-purpose composition and content-general inferences.

(4) Santiago Echeverri (Institute for Philosophical Research, UNAM) Iconic Object Tracking

Many philosophers and cognitive scientists think that the visual system uses representations analogous to demonstratives such as ‘this' or ‘that' to keep track of moving objects. I sketch an alternative theory according to which existing findings are best explained with iconic representations. My argument has four steps. First, I argue that visual segmentation—which enables us to distinguish an object from its surroundings—is only possible if the visual system represents pre-objective elements as exemplifying spatial properties and suggest that iconic representations offer a more efficient means than discursive representations to represent those properties. Second, I suggest that tracking can be understood as a dynamic segmentation process because, when objects move, the spatial properties exemplified by pre-objective elements change, producing segmentation ambiguities. The visual system must resolve those ambiguities during tracking. Third, I present empirical findings that support this account. Those findings suggest that grouping principles can produce segmentation ambiguities that are resolved during tracking, so segmentation and tracking are at least concurrent operations. Fourth, I present a theory of iconic representations that explains available empirical findings and offer an error theory explaining why many theorists—including those in the iconicity camp (like Block and Burge)—still posit perceptual demonstratives.