Formalising Human Mental Workload as a Defeasible Computational Concept

Human mental workload has gained importance, in the last few decades, as a fundamental design concept in human-computer interaction. It can be intuitively defined as the amount of mental work necessary for a person to complete a task over a given period of time. For people interacting with interfaces, computers and technological devices in general, the construct plays an important role. At a low level, while processing information, often people feel annoyed and frustrated; at higher level, mental workload is critical and dangerous as it leads to confusion, it decreases the performance of information processing and it increases the chances of errors and mistakes. It is extensively documented that either mental overload or underload negatively affect performance. Hence, designers and practitioners who are ultimately interested in system or human performance need answers about operator workload at all stages of system design and operation. At an early system design phase, designers require some explicit model to predict the mental workload imposed by their technologies on end-users so that alternative system designs can be evaluated. However, human mental workload is a multifaceted and complex construct mainly applied in cognitive sciences. A plethora of ad-hoc definitions can be found in the literature. Generally, it is not an elementary property, rather it emerges from the interaction between the requirements of a task, the circumstances under which it is performed and the skills, behaviours and perceptions of the operator. Although measuring mental workload has advantages in interaction and interface design, its formalisation as an operational and computational construct has not sufficiently been addressed. Many researchers agree that too many ad-hoc models are present in the literature and that they are applied subjectively by mental workload designers thereby limiting their application in different contexts and making comparison across different models difficult. This thesis introduces a novel computational framework for representing and assessing human mental workload based on defeasible reasoning. The starting point is the investigation of the nature of human mental workload that appears to be a defeasible phenomenon. A defeasible concept is a concept built upon a set of arguments that can be defeated by adding additional arguments. The word ‘defeasible’ is inherited from defeasible reasoning, a form of reasoning built upon reasons that can be defeated. It is also known as non-monotonic reasoning because of the technical property (non-monotonicity) of the logical formalisms that are aimed at modelling defeasible reasoning activity. Here, a conclusion or claim, derived from the application of previous knowledge, can be retracted in the light of new evidence. Formally, state-of-the-art defeasible reasoning models are implemented employing argumentation theory, a multi-disciplinary paradigm that incorporates elements of philosophy, psychology and sociology. It systematically studies how arguments can be built, sustained or discarded in a reasoning process, and it investigates the validity of their conclusions.