A conceptual model for designing adaptive human-computer interfaces using the living systems theory.(Research Paper)(Report)

INTRODUCTION

Advances in computer technology have resulted in modern work systems that are complex. This complexity arises from a large number of interacting components whose behaviours and variables constantly change over time. System responses are also generally slower, making control more discrete and open loop rather than closed loop with instant feedback. Due to their slower control nature, they are less constrained by human motor limitations but rather make issues of decision-making, attention allocation, perception and memory paramount to their design (Wickens and Hollands, 2000). This makes decision-making and information management essential components of complex systems tasks. Complexity calls for human-computer interfaces (HCIs) that change over time with changing user characteristics and domain features.

To address the changing HCI demands, it is possible to adapt the interface components: content, structure and form to address situational and dynamic needs. This necessitates an HCI system that has the capability to alter certain aspects of its information structure or functionality to cope with the complex changing environmental inputs, including the cognitive needs of the operator over time. For this to happen, the HCI system must have self-learning capabilities to support adaptive information behaviours in the system. Designing such interfaces is a major challenge since it requires the management of complex information elements within the HCI system.

The notion of changing aspects of an interface with time to meet requirements within the immediate context gives rise to the concept of adaptive interfaces. Such an adaptive interface is conceptualized to provide the operator with the most salient information in any particular context, in the best form, and at the most appropriate time (Viano et al., 2000). Interfaces of this kind are more relevant in today's complex environments such as information management in battlefield systems, medical systems or power systems. The price for error in such environments is very high making the role of interfaces crucial.

As mentioned earlier, the requirements for an interface to adapt introduces a unique challenge for HCI design. In their review of the literature on the subject matter, Rothrock et al. (2002) identified gaps in design methodologies for interfaces with adaptive capabilities. Two reasons are responsible for this: (a) lack of rigorous biologically inspired models in the current HCI design corpus; and (b) insufficient universal principles and paradigms that support adaptive interface designs. Rothrock et al. (2002) advocate decentralized architectures in which adaptation is achieved through biologically inspired models. The focus of adaptation in biological systems is on the emergence of complex behaviours from interactions between relatively simplistic organisms. A biologically inspired approach to HCI design represents one of the several innovative methods explored to model the subtle and difficult topic of adaptation. These methods can be (and have been) captured in terms of frameworks, paradigms, or rules. As an example, Rothrock et al. (2002) cited complex adaptive systems (Holland, 1998) as a viable biologically inspired framework. In this paper, we show that living systems theory (LST) (Miller, 1978) is also a biologically inspired framework suitable for an application to adaptive HCI designs.

The LST belongs to a class of biologically inspired models and has certain properties that allow designers to mimic adaptive behaviours reminiscent of "intelligent" systems (Holland, 1998). Metaphorically, an adaptive interface is construed to be similar to biological species that can adapt to environmental situations and can self-organize their behaviours to achieve intended context-based goals. For example, biological systems have the capability to achieve adaptation through homeostasis--a process in which an organism maintains a constant internal environment (e.g. body temperature and fluid content) through regulatory mechanisms that compensate for a changing external environment. This adjustment process can be triggered by either internal or external events or signals. In the same context, HCI systems are subject to information perturbations from the users, tasks variations, and the environmental noise. There are few HCI models that are designed to recognize and adjust their behaviours to these information state changes. The properties of LST enumerated by Miller (1978) provide immense opportunities for designing HCI with these adaptive capabilities.

ADAPTIVE INTERFACES