Understanding Human Spatial Navigation Behaviors: A Cognitive Modeling Approach

Open Access
Zhao, Changkun
Graduate Program:
Information Sciences and Technology
Doctor of Philosophy
Document Type:
Date of Defense:
March 03, 2016
Committee Members:
  • Frank Edward Ritter, Committee Chair
  • Xiaolong Zhang, Committee Member
  • John Yen, Committee Member
  • Richard Alan Carlson, Special Member
  • Cognitive Modeling
  • ACT-R
  • Navigation
  • Spatial
  • Cognitive Model.
Spatial navigation behavior is a basic ability for humans and animals to survive on the earth, as it allows us to seek food, return home, and localize friends. It is widely accepted that human navigation relies on some solid representations of space. Previous studies also show that there are two basic spatial representations: 1) the configurational (map) representation that consists of distances, Cartesian (absolute) directions, and geometric relations; 2) the sequential (route) presentation that involves landmarks and orientation sequences. However, how humans apply the two representations in their daily activities and the key factor of navigation process is still under debate. Two contradictory understandings are: 1) that configurational representation is a solid representation that could lead to accurate navigation behavior; 2) humans basically rely on the sequential representation, and the configurational representation can only lead to inaccurate navigation results. This dissertation explored these issues with a new empirical scenario and a novel cognitive modeling approach. First, I conducted an empirical study using NavMaze to examine three new influence factors of navigation process: spatial retention, navigation preference, and mental rotation ability. The empirical results suggest that spatial retention is not a key factor or human navigation process, and the navigation performance is more correlated to navigation preference and mental rotation ability. This result reveals that human navigation process more replies on procedural skills. Second, I implemented a comprehensive cognitive model named NavModel in ACT-R to replicate empirical data. NavModel consists of a text-based testing platform for ACT-R, a mental rotation model based on an extended imaginal module of ACT-R, and an implementation of spatial representations and navigation strategies in ACT-R. The model fits the empirical data well; the mental rotation model, especially, can generate a very accurate prediction. In the modeling and data fitting process there are three new understandings of human navigation process: 1) humans might rely on the sequential representation during navigation; 2) mental rotation ability is a key procedural skill in navigation; 3) humans use object separation and visual matching in the mental rotation process rather than rotating the entire object in their imagination.