Integrating Cognitive Models

Carruth, D. W. (2006). Integrating Cognitive Models. Digital Human Modeling and Virtual Reality for FCS.

Abstract

Cognitive architectures are software tools that allow researchers to computationally simulate human thought processes and task performance to develop psychological theories, to construct and evaluate training tools and methods, and to evaluate product designs. Digital human models such as Santos are capable of predicting realistic human postures and motions, but to date have had little or no cognitive component. There exists a need to develop an integrated model that leverages the capabilities of cognitive architectures as well as the capabilities of digital human models for the evaluation of human task performance and product design. Our current work is focused on integrating the sensation and perception modules of the ACT-R cognitive architecture with the Virtools virtual environment. This will allow Santos to see objects and hear sound events, process the perceptual information, and plan and execute actions in the virtual environment. To implement this capability, methods for extracting sensory information from the virtual environment were developed, extensions to the cognitive architecture’s perceptual modules were implemented, and a network interface between the virtual environment and the cognitive architecture was put in place. In order to validate the current perceptual and future motor capabilities, a study was designed to collect human data for four simple tasks: model assembly, sorting objects, navigating hallways, and using a vending machine. Eyetracking and motion capture data for these tasks are currently being collected. Virtual versions of the physical environments are being constructed to allow models of the human task performance to be developed and the resulting data will be compared to human data to assess the integrated digital human model’s capabilities.