Back to search

Boston University

Inferring rules from evidence: behavioral, computational, and neural mechanisms of flexible decision making

Abstract

dc:description.abstract

Humans translate continuous evidence into discrete choices by inferring decision rules from environmental feedback. This process requires transforming perceived information into coherent internal representations, extracting predictable statistical patterns from uncertainty, and generalizing them to novel instances. Although these capacities are central to everyday behavior, the mechanisms underlying adaptive decision making remain elusive. The studies in this dissertation examined: 1) converting dynamic evidence to stable internal representations, 2) feedback-driven learning of context-dependent decision rules, and 3) the neural patterns implicated in context-dependent valuation. Chapter 2 investigated sequential evidence integration and feedback-guided rule inference. The hypothesis was that adolescents with neurodevelopmental disorders exhibit noisier evidence integration, resulting in less precise decisions than typically developing (TD) peers. Across two experiments differing in sensory modality (visual-only and multisensory), adolescent participants with (N=269) and without (N=141) autism spectrum disorder (ASD) received streams of leftward and rightward stimulus pulses and selected the side associated with the greater number of pulses. Computational modeling revealed that ASD adolescents showed greater suboptimality in evidence integration than TD peers, with differences most pronounced in the multisensory experiment. Chapter 3 examined feedback-driven learning of temporally structured decision rules in environments where rewards depend on internally maintained memory representations. Participants (N=150) were randomly assigned to one of three conditions governing how item value changed as a function of elapsed trials since the last encounter: systematically increasing, decreasing, or varying non-monotonically. The resulting acceptance probability varied parametrically with each item's expected value, indicating that humans can internalize abstract decision rules based on experience. Chapter 4 extended the Chapter 3 paradigm to a multi-session neuroimaging experiment (N=32), aimed at identifying neural mechanisms underlying flexible context-dependent valuation. Prior work has shown that signals in ventromedial prefrontal cortex (vmPFC) and striatum track value, whereas medial temporal lobes (MTL) are theorized to track temporal recency. The present experimental design orthogonalized the parametric effects of value, recency, and task difficulty, enabling examination of neural activity associated with each factor across contexts governed by opposing decision rules. Univariate and multivariate analyses revealed that canonical value-encoding regions tracked parametric value effects, whereas MTL and vmPFC tracked temporal recency across contexts.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Psychological & Brain Sciences
Grantor
Boston University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Li, Yutong
Advisor dc:contributor.advisor
  • McGuire, Joseph T.
ORCID dc:identifier.orcid

Subjects

dc:subject × 1

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/2144/53587

Chain of custody

source
Harvested from
Boston University
Base URL
open.bu.edu/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Li, Yutong. Inferring rules from evidence: behavioral, computational, and neural mechanisms of flexible decision making. doctoral thesis, Boston University, 2026. https://hdl.handle.net/2144/53587