This paper describes a project to investigate methods and techniques to represent physical and perceptual activity in the context of cognitive models of human performance. The application of principal interest is that of constructing computer-generated forces (CGFs) for use in large-scale distributed simulations of military forces. Of the several frameworks currently in existence for simulating human behavior at both a cognitive and behavioral level, CHI Systems’ COGNET toolset represents a particularly attractive foundation for creating a tool to develop and execute CGFs. COGNET already possesses features that meet many of the human behavioral requirements of computer generated force models, including particularly its orientation toward representation of realtime, multi-tasking work environments. In this paper we describe extensions in the capabilities of the COGNET system for construction of CGF applications primarily through enhancements in the areas of sensory and motor abilities and constraints, individual differences, and situational effects. Enhancements in these areas have been accomplished through a number of extensions in which new model features have been developed using recent research results and theories. Although COGNET does not currently incorporate a detailed graphic body model, it offers a versatile tool for integrating with such models with anticipated benefits both in improving the representation of visual activities for COGNET and in providing intelligent guidance for graphic body model behavior. Demonstrations of COGNET models emphasizing perceptual and motor activities are presented.