Investigator: Abhilasha Kumar, Ph.D., Bowdoin College
Project Summary
Retrieval of concepts from semantic memory involves interactions between several brain regions and a complex interplay of lexical sources. Recent work with the verbal fluency task (VFT; where participants have to generate as many examples as they can from a given category within a fixed duration) has shown that individuals delicately balance exploiting current clusters (i.e., groups of related items) within semantic memory with exploring new clusters. However, the temporal dynamics and cognitive processes underlying explore-exploit decisions during semantic memory search remain understudied and underspecified. Our long-term goal is to develop a theory of naturalistic memory search guided by behavioral, computational, and neural evidence by investigating the mechanisms underlying memory search in the VFT. The objective of the proposed research is to identify the temporal dynamics of explore-exploit decisions involved in naturalistic memory search during the VFT using a combination of computational modeling and electroencephalography (EEG). Based on prior work, the hypothesis is that predictive processing indexed by theta oscillations mediates active exploration and decisions to exploit new clusters during memory search. To achieve this objective, a computational model will be used to classify individual transitions from the VFT into explore vs. exploit decisions. Oscillatory theta power differences and the relationship between theta power and predictive processing signals obtained from auxiliary tasks will be examined. This research is innovative because it combines advanced computational modeling with temporally robust EEG methods to study semantic search and retrieval processes and moves our understanding of predictive processing beyond standard episodic and reinforcement learning paradigms. This work is significant and relevant to public health because it will help identify the mechanisms that contribute to exploration and exploitation behaviors during memory search, which are often disrupted in mental illnesses including psychotic disorders and dementia. The proposed research has the potential to inform the use and interpretation of fluency measures in clinical settings and its connection to specific diagnoses and ultimately inform the design of evidence-based interventions for search and retrieval-related dysfunction.