Exploratory Data Analysis of Autobiographical Memory Trends

Abstract

A vital component of human cognition is the ability to recall past episodic events, otherwise known as autobiographical memories. These memories serve as references, constructed from experiences, to guide problem-solving and decision making. Despite their daily utilization, our quantitative knowledge of autobiographical memories is surprisingly limited. This research furthers the understanding of human memory retrieval through big data analytics on a dataset of 13,705 autobiographical memories scored for 8 distinct features such as the number and type of recalled details. Two-step cluster analysis identified two distinct clusters mainly classified by overall total content and not by specific memory features. After data normalization to discount for the effect of total content, K-means clustering revealed three recollection types with distinct feature patterns: memories with uniform feature distribution, memories with a clear majority of “people” details, and memories with a clear majority of “places” details. Recollections of the uniform type had the highest mean total content, indicating that when individuals remember one feature in greater detail, they lose other details of autobiographical memory. The “people” cluster had the lowest subject age, suggesting a tendency of younger individuals to focus on this aspect. The “places” cluster corresponded to the most remote memories. Moreover, the individual standard deviation of memory content increased drastically when memories were randomly reassigned to subjects, showing that some subjects have systematically richer recall than others. These findings serve as building blocks for further exploration to gain an in-depth understanding of the underpinnings of autobiographical memories.