Control States and Complete Agent Architectures

Abstract

This paper presents a developing concept of mind defined in terms of external and internal niches. This perspective on mind is described primarily in terms of the niche space of control states and the design space of processes that may support such phenomena. A developing agent architecture, that can support motivation and other control states associated with mind, is presented. Different aspects of agent research are discussed in terms of three categories of agents. Each agent category is characterized primarily in terms of their task-related competencies and internal behaviors and discussed in terms of our taxonomy of control states. The concept of complete agents is then introduced. Goals are described in terms of their generation across a number of computational layers. Experimental analysis is provided on how these differing forms of behaviors can be cleanly integrated. This leads into a discussion on the nature of motivational states and the mechanisms used for making decisions and managing the sometimes-competitive nature of processes internal to a complex agent. The difficulty of evaluating complete agents is discussed from a number of perspectives. The paper concludes by considering future directions related to the computational modeling of emotions and the concept of synthetic mind.

Linking perception and action through motivation and affect

“…We are developing a theory of affect that draws on themes such as control states and motivators (Simon 1967;Sloman 1987Sloman , 1993Davis 2001) and affordances (Gibson 1979). …”

“…Previous research has attempted to organize motivational control states in an initial taxonomy as a starting point for future research (Davis 2001). Figure 1 provides one taxonomy of five major types of motivational control states with subcategories.…”

Computational architectures for intelligence and motivation

“…This theory of mind [3] considers that cognition is driven by and consists of control states. Relating this to more epistemological approaches to cognition (eg [4]) is a long standing goal of the current work.…”

“…Hence in the soapy knife scenario the catching reflex is over-ridden by a self-preservation alarm. In some computational experiments [3] this layer has been made separate for design and domain reasons, giving rise to a four-layer architecture. The reactive layer typically models transient control states and continually adjusts to the external world in the same way as behaviour-based systems [7].…”

“…The use of propositional calculus and finite state graphs allows the specification of the processes that manage incoming information; select and activate external behaviours; generate and manage motivators; plan; make decisions; and manage memories. An explicit motivator structure (see [3] for a full description) draws together many aspects of the control state taxonomy. The same computational structure are used for motivators at the reactive level as at the deliberative layer even though many components, for example plans, motivator rationale and resource management, are only relevant to the deliberative-layer.…”