Integrating new and prior knowledge in semantic memory

Consider the words castanet and Casablanca. Most speakers of English experience little difficulty when processing these items: Castanets are quickly recognized as musical instruments associated with the Spanish dance, flamenco; and Casablanca might call to mind the classic movie, complete with snippets of dialogue or images of Humphrey Bogart and Ingrid Bergman. The ease and speed with which prior knowledge is accessed is impressive, considering the vast amount of knowledge acquired over the course of a lifetime. Merely considering words, one estimate is that the average adult has about 50,000 lexical entries stored in semantic memory. For each item, in addition to the knowledge about the word itself – spelling, pronunciation, grammatical class – an individual can access rich conceptual knowledge, such as prior experiences with the word’s referent, information related to or associated with the word, and so on. In addition, semantic memory includes a large database of people both famous and intimately known, and the extensive list of factual or fictional knowledge acquired through formal education, media outlets, and personal experience. The semantic system supports, stores, and organizes this information in a powerful knowledge base, which is critical for supporting how individuals navigate their world on a daily basis: Without it, we would have difficulty processing language, understanding pop culture referents, and, more generally, understanding the world we live in.

A key characteristic of the knowledge stored in semantic systems is that these memories are de-contextualized. In other words, the specifics of when and how this knowledge was acquired are not associated with the content of the memory itself. The context-independent nature of semantic memory distinguishes this type of knowledge from episodic memories, which are situated in time and space and often context-dependent. Because, ultimately, one goal of educational systems is to create persistent and accessible knowledge that will support reasoning and problem-solving across contexts and disciplines, my research program examines which types of learning and encoding techniques promote durable knowledge that can be retrieved in a variety of situations with multiple cues. I hope to determine whether learning techniques that have been empirically validated to promote long-term contextually dependent episodic retention (e.g., distributed practice, retrieval practice, meaningful processing) support the integration of recently acquired information into semantic systems.

My research focuses on three inter-related aspects of semantic memory: How knowledge is acquired and integrated into semantic networks, how it is organized in a way that facilitates efficient retrieval and use, and how it can be called upon to support performance in a variety of tasks and situations. An important assumption about semantic knowledge is that this information can be accessed and retrieved automatically. Thus, I propose a program of research aimed at identifying the contributions of controlled and automatic processes and measures of performance that can dissociate the two, such as response times or error rates. To examine these questions, I rely on a variety of behavioral paradigms and incorporate measures of individual differences, such as age and differences in cognitive abilities (e.g., working memory, attention).