Semester Award Granted

Spring 2025

Submission Date

May 2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Thesis/Dissertation Advisor [Chair]

William Alexander

Abstract

The prefrontal cortex (PFC) plays a crucial role in complex behaviors such as planning, decision-making, and adaptive problem-solving, which require integrating new information over time. Research has shown that information related to these behaviors is organized along a hierarchical rostro-caudal abstraction gradient in PFC, with abstract information represented in rostral (anterior) regions and concrete relationships in caudal (posterior) regions. Abstraction involves generalizing information by emphasizing broader concepts over specific details and can be understood more specifically by the context its applied to. However, past studies often confound different forms of abstraction, making it unclear how these different types of abstraction may uniquely shape internal representations. In this dissertation, I examine how structural and temporal abstraction interact in shaping behavior by manipulating the presentation order of informational cues. In agreement with computational models of PFC, I find that behavior is governed by temporal abstraction, with behavioral differences further influenced by the order in which structurally abstract information is presented. Furthermore, by selectively manipulating the salience of information presented immediately prior to a response deadline relative to preceding information, I demonstrate how learning to integrate information sources more efficiently proceeds in a bottom-up fashion. Finally, I show how the insights derived from laboratory experiments can be applied in the real-world context of health decision-making. Consistent with prior findings, participants’ decisions are influenced by information encountered immediately before the choice, while changes in the salience of health outcome information immediately prior to a response suggested differences in how subjects processed quality of life vs lifespan information. Overall, this work suggests that internal representations in learning and decision-making are governed by temporal abstraction, with improved learning and behavioral outcomes observed when different forms of abstraction align. The overall approach of manipulating temporal and structural abstraction is a valuable approach for assessing how information is integrated and decisions are made under time-restricted conditions, with applications for understanding how individuals make decisions in real-world contexts with time pressure.

Download

Share

COinS