Concept is a container
This thesis puts forward a theory which I call container theory for how a single notion of concept can satisfy the two desiderata that Machery (2009) sets out for concepts: (a) the Judgement Desideratum - a concept must permit categorisation, typicality and inferential judgements; and (b) the Propositional Desideratum: a concept must be capable of being used as a unit or constituent in compositions of propositional thoughts. Achieving this depends on two core claims: (a) that a concept is a container of stored knowledge that pertains to a single category and (b) that double-armed finsts (DA-finsts), which keep track of both objects and the categories they have been recognised as, are composed into propositional thoughts. In addition to satisfying the desiderata, the framework allows a full explanation of what a concept is, how one is acquired, how it comes to have meaning, and how they are involved in matters of sense and reference and wide and narrow content. The first core idea is that a concept is a container which gathers together information pertaining to a single category. Recognitional information (e.g. exemplars and prototypes) in the container allows categorisation and typicality judgements and functional information facilitates inferential judgements. Together these satisfy Machery's judgement desideratum. I stress that a concept is the container itself and not the information stored in the container and not the label or symbol of the container. The second core idea is that DA-finsts, which are an extension of Pylyshyn's (1988) notion of a finst, are what are composed into propositional thoughts, thus satisfying Machery's propositional desideratum. These finsts have no intrinsic meaning; they inherit their meaning from what concept/container they are currently keeping track of. DA-finsts are disposable symbols in that they dispose of their meaning once they stop pointing to a container or point to a new container. In addition to satisfying Machery's desiderata this framework has three major benefits. First, because DA-finsts can be substituted for Mentalese symbols in Fodor's (2008) LOT account and for perceptual symbols in Barsalou's (1999) account an attractive fusion of these two accounts can be achieved. Second, a full account can be given of how new concepts/containers are acquired and come to have the meanings they do. In container theory concepts are individuated by their container structure rather than by their semantic structure. Information that is not recognised as belonging to any existing container causes a new container to be opened. What a container/concept pertains to is determined by what information is in the container, but the concept itself is determined by the container structure. Error (misrepresentation) occurs when the expected outcomes of inferences and interactions based on the stored information in a container do not match the actual outcomes experienced. Finally, the framework is used to make sense of puzzles that give rise to distinctions between wide and narrow content and between sense and reference.