Back to search

Clemson University

Impacts of Cognitive Workload on Veteran Driver Reaction Time: Predictive Modeling Using Bayesian and Machine Learning Methods

Abstract

dc:description.abstract

In high-demand environments, the ability to manage cognitive workload can mean the difference between optimal or safe performance and critical failure or accidents. Veterans face an elevated risk of fatal motor vehicle accidents due to post-deployment stress, combat-related injuries, and challenges readjusting to civilian driving. This study explores how cognitive workload affects reaction time performance using a driving simulator by collecting and analyzing subjective workload ratings (using the NASA-TLX survey), physiological signals from eye-tracking and performance data from a sample of 28 Veterans. We examined how task difficulty, cognitive indicators and personal attributes influence reaction times across an interactive driving exercise’s central and peripheral tasks. Reaction times were classified using Z-score-based and K-means clustering methods to explore how classification strategy influences interpretability and prediction. Using Bayesian Multinomial Logistic Regression models, we found that increased task difficulty significantly reduced the likelihood of extreme (Fast or Slow) reaction times, indicating performance regulation under cognitive pressure. The average pupil diameter was a key physiological marker, particularly during peripheral and central tasks. Here Veterans exhibited strategic allocation of cognitive resources by shifting engagement depending on task importance and difficulty. The models also showed minimal effects of self-reported workload and demographic variables, aligning with literature that questions the sensitivity of subjective workload ratings in multitasking exercises. Complementary machine learning models confirmed these findings, with support vector machines and neural networks demonstrating high predictive performance – especially under Z-score-based classification methods. By integrating subjective, physiological and performance data with advanced modeling, this research contributes a novel comparative modeling framework and emphasizes the value of integrating classification methodology for cognitive workload modeling in safety-critical domains such as driving.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MS)
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Civil Engineering

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Ofosu-Kwabe, Kenneth
Contributors dc:contributor
  • Pamela Murray-Tuite
  • Wayne A. Sarasua
  • Johnell Brooks

Subjects

dc:subject × 8

Identifiers

dc:identifier.*
Repository record dc:identifier
https://open.clemson.edu/all_theses/4501

Chain of custody

source
Harvested from
Clemson University
Base URL
open.clemson.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Ofosu-Kwabe, Kenneth. Impacts of Cognitive Workload on Veteran Driver Reaction Time: Predictive Modeling Using Bayesian and Machine Learning Methods. Thesis thesis, https://open.clemson.edu/all_theses/4501