Advancing models of semantic representation: empirical study designs, network analysis methods, and computational tools

People learn an astonishing amount of things about the world throughout life. This knowledge is retained in semantic representations, the cognitive manifestation of factual information in memory. Modeling semantic representations is an important ingredient to understanding how people acquire knowledge from their environment, how this knowledge is organized in memory, and how it affects behavior. Given the wide-ranging implications of semantic representations, research questions address multiple aspects, including their structure and content. Various methodological approaches have been used to successfully model semantic representations from text and behavioral data. However, much is still unknown about the optimal ways to model semantic representations, particularly regarding structural differences of representations among people. In addition, further investigation is needed to connect the structure of semantic representations to environments and behavior, as well as to map the content of individual concepts’ semantic representations. This dissertation aims to improve the methods and tools used for modeling semantic representations across various applications while also establishing links between semantic representation and cognitive performance. The first two manuscripts explore the relationships between environment and semantic representation as well as between semantic representation and cognitive performance of younger and older adults in a proof-of-concept study. The third manuscript presents a comprehensive recovery simulation study that evaluates which empirical study designs yield accurate individual-level semantic networks based on free associations and relatedness judgments. The fourth manuscript introduces the R package associatoR and includes a tutorial on how to use it to map the content of semantic representation. Together, this dissertation contributes to the understanding of individual differences in semantic representation by linking it to cognitive performance, guiding study design for assessing individual semantic networks, and facilitating the analysis of free association based models of semantic representation.