Back to search

Cornell University

Quantitative modeling of collective behavior

Abstract

dc:description.abstract

We show how ideas, models, and techniques from statistical physics prove useful for building quantitative theories of collective behavior with a focus on social systems. In the first chapter, we develop a statistical mechanics of political voting on the US Supreme Court. By building minimal models of voting behavior, we find that signatures of strong consensus and partisanship are captured by a maximum entropy model only relying on the pairwise correlations between voters. We extend this maximum entropy approach to collective voting outcomes on a Super Supreme Court, inferring how the set of historically disjoint justices from 1946-2015 might have voted with one another. When we measure how correlations decay between justices across time, we find a long, institutional timescale approaching a century, a quantitative signature of historical precedent. Beyond consensus, we find that voting blocs fracture in many possible ways, belying a common assumption that partisan intuition generalizes to the history of the court; actually, Supreme Court voting over time is immensely more complex. Then, we use minimal models to measure how sensitive collective outcomes are to perturbations to "pivotal components," akin to how majority outcomes are sensitive to swing voters. We demonstrate how to extract these pivotal components from an information-geometric analysis of example social systems including Twitter, financial markets, legislatures, and judicial courts. Our approach presents a principled, quantitative step towards characterizing the robustness of social institutions to changes in component-level behavior. The second topic we address is about conflict dynamics. In a society of pigtailed macaques and in armed conflict in human society, we find remarkable, emergent regularities suggesting that conflict is dominated by a low-dimensional process that scales with physical dimensions in a surprisingly unified and predictable way. For macaque conflict, we discover a temporal scaling collapse for conflict duration distributions. This collapse indicates the presence of long-time correlations that connect early conflict events with later ones. We propose a model that explains this collapse and consider how we might predict conflict evolution. For armed conflict, we find that social and spatiotemporal properties of conflict can be unified by a reduced scaling framework, and we make initial steps towards a model that captures the dynamics of observed properties of conflict. The final topic we address is that of interpersonal coordination of motion. We describe an experimental apparatus that combines a commercial virtual reality platform, a human motion-capture suit, and a mirroring game with an avatar. We use this apparatus to show how frequency-based auditory cues enhance the ability to mirror motion. Our work lays the groundwork for future experiments: better understanding of how information is encoded in visual or auditory cues could facilitate joint coordination when navigating visually occluded environments, improve reaction speed in human-computer interfaces or measure altered physiological states and disease.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Physics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Physics
Year dc:date.issued
2019

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Lee, Edward D
Chair dc:contributor.chair
  • Ginsparg, Paul
Committee members dc:contributor.committeemember
  • Elser, Veit
  • Cohen, Itai

Subjects

dc:subject × 6

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
13119707
Dc Identifier Other
ProQuest Submission ID: 11806
ProQuest Publication ID: 27663595

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Lee, Edward D. Quantitative modeling of collective behavior. Doctor of Philosophy thesis, 2019. https://hdl.handle.net/1813/70071