Date of Award

12-2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Department of Electrical and Computer Engineering

First Advisor

Gilbert L. Peterson, PhD

Abstract

High performing human teams transcend complex domain uncertainty by achieving an emergent state of shared cognition, in which knowledge is organized, represented, and distributed to team members for rapid execution. However, this requires that individuals emit perceivable qualities upon which other members can make inferences about intent. In pursuit of future human and machine team studies, this research presents a hybrid cognitive model for machine agents in fully cooperative and semi-cooperative action and project teams. The hybrid cognitive model unifies the characteristics of the shared mental model and transactive memory system. The resultant model facilitates anytime selection over the two cognitive representations with the computational complexity of a single model. Evaluation of the hybrid cognitive model occurs in multi-agent domains with increasing complexity and levels of cooperation. Agent performance is assessed according to four cognitive characteristics that capture aspects of the natures and forms of cognition found in project and action teams. The studies utilize a mixed methods approach in the analysis of four established characteristics and measures. The results demonstrate that agents using the cognitive model form aligned representations that encode structural, perceptual, and interpretive cognitive forms. Additionally, the results suggest that agents employing the hybrid cognitive model can switch between compositional and compilational natures of emergence as necessary to integrate behaviors or knowledge.

AFIT Designator

AFIT-ENG-DS-23-D-031

Comments

A 12-month embargo was observed in posting this dissertation.

Approved for public release, case number on file).

Download

Share

COinS