Modeling Feeling of Knowing
Faculty Mentor: Justin Li, Computer Science & Cognitive Science Department
Major: Computer Science & Economics
Funding: Ford Research Mentor's Endowment
Humans are able to accurately estimate what they know and do not know. This is a metamemory judgement called feeling of knowing (FOK). FOK signals the probability of a person being able to procure a piece of information from memory. Researchers have yet to establish a concrete basis for FOK judgements, but have put forward many hypotheses, including cue-familiarity, accessibility, competition, and a combination of accessibility and competition. I built a computational model of each of these hypotheses. My models of FOK calculations were applied to two types of scenarios: complex mulit-step trivia questions and a computational replica of a psychology experiment focused on feeling of knowing. My goal was to see which models matched human behavior. My results show that no single hypothesis can produce computational results that match human studies, and that FOK is complex and likely based on more than one factor. My hypotheses could be more rigorously explored if tested on a larger/more standardized knowledge base that is more reflective of the human mind.
Watch my research presentation below.
Questions or comments? Contact me at: firstname.lastname@example.org