Remembering that Neil Armstrong was the first person to walk on the Moon feels like pulling a file stored somewhere in the mind. But is this really how we know facts about the world, by retrieving them from storage in the mind? This capacity to draw on general facts and knowledge about the world is semantic memory. Yet despite its central role in nearly all forms of cognitive endeavor, it remains largely neglected within the philosophy of memory, and has not been given a direct philosophical treatment in the literature despite its importance to a wide range of topics. 

The relative lack of research on semantic memory in the philosophy of memory may be due to the influence of a widely accepted, yet seldom thoroughly analyzed orthodox view of semantic memory that holds that semantic memory is a dedicated storage space of facts about the world, as well as personal information. It is furthermore, assumed to be the product of a chronic, unconscious mechanism of abstraction and generalization. 

But acceptance of the orthodox view of semantic memory belies fundamental issues with its conception. First, the apparent mechanisms of abstraction and generalization have yet to be demonstrated, despite decades of empirical research. Moreover, the most widely accepted view of memory traces as distributed and superpositional are in tension with the idea that discrete packets of information are stored and retrieved, thus undermining the orthodox view of semantic memory as a storage space of explicit facts. Finally, despite decades of research, attempts to identify the neuroanatomical structure, location, and mechanisms of memory storage have consistently fallen short, leaving the very notion of a dedicated storehouse of facts without empirical support. Reassessment of standard notions of semantic memory are thus required to accommodate the hostility of the radical enactive approach to stored content. Taken together, these pressures reveal the orthodox view to be conceptually unstable, and empirically unsupported. 

My dissertation challenges the orthodox framework and develops a constructive, procedural theory of semantic memory. On this view, remembering isn’t like pulling a finished memory out of a filing cabinet. Instead, it is an active skill, where practiced abilities allow us to rebuild a memory anew each time we need it. In other words, what's stored in memory isn't the fact itself, but the know-how needed to reconstruct and bring that fact to mind. Semantic memory is not a repository of information but a dynamic process, expressed in different modes depending on present demands. Rather than appealing to storage and retrieval, I argue that semantic memory is best understood in terms of what I call, ‘modes of expression’. At one end of the spectrum are deliberate, effortful constructions that mobilize prior experience in order to solve current problems. At the other are fluent, embodied certainties that operate automatically, the result of skills and dispositions that have become integrated into our patterns of thought and behavior. These modes of expression are not separate systems but different ways in which constructive processes are retained and brought to bear on new situations.

By reconceiving semantic memory in procedural and constructive terms, this dissertation resolves long-standing conceptual puzzles surrounding abstraction, generalization, and trace content. It also reframes semantic memory, not as a passive storehouse of facts, but an active, generative capacity underpinning nearly all human cognition, from imagination and counterfactual thinking, to future planning, and socially embedded knowledge practices.