The visual complexity of instructional animations in training simulations to promote learning

There is a trend in systems design in creating complex visuals for Virtual Reality (VR) simulations. Given "generous" polygon and hardware allowance in developing digital contents, we may reach the hardware limitations easily to create aesthetically pleasing visuals. With affordable and free 3D engines, and modelling and animation tools, more educational institutions are expected to use the technology for teaching and learning. One concern is that the latest VR technology often surpasses the optimum requirements for learning, especially in visual complexity. There are two problems of learning with complex visuals: Simulator Sickness (SS), due to greater optic flow from greater visual details and cognitive load due to extraneous visual information. At the same time, complex visuals may add motivational factors such as presence and perceived affective quality, to improve learning. In assessing the effects of visual complexity in VR-based instructional animation, this thesis draws findings from studies in SS, Cognitive Theory of Multimedia Learning (CTML), presence and perceived affective quality within the flow theory. -- The thesis contains three main parts consisting of smaller chapters. The first part contains two chapters including the introduction and the making of the VR simulations. The second part contains the chapters of published journals and papers, a book chapter as well as manuscripts submitted to journals. The last part of this thesis contains the conclusion that maps all the research questions to the empirical results, and gives an overview for the uture directions of research. -- This thesis started with the completion of two research projects at the beginning of the candidature: SS in VR simulations and the development of a risk assessment simulator. The author merged the findings of these two projects while exploring visual complexity, which became the focus of this thesis. Since the existing simulations could not be used to assess SS and learning outcomes in a unified approach, the candidate developed two non-interactive VR simulations: (1) ANTECATALYST (ANimaTEd CATAgLYphiS anTs) and (2) VEAR (VEhicle Accident and fiRes). ANTECATALYST examined the effects of Animated-Virtual Actors' (AVAs) visual complexity and learning, and showed Cataglyphis ants' navigational behaviour on a flat and featureless desert with three learning conditions: (1) flat, (2) cartoon, or (3) lifelike AVAs. VEAR confirmed and extended the findings by taking into accounts the visual complexity of the AVAs and the Virtual Environments (VEs). VEAR is a firefighter training simulation and shows the hazards and the best strategy to manage car accidents safely. VEAR contains four visual conditions for learning: (1) Simple - visually simple AVAs and VE, (2) Simple AVAs - simple AVAs and a visually complex VE, (3) Simple world - visually complex AVAs and a simple VE, and (4) Lifelike - visually complex AVAs and VE. In assessing SS and learning outcomes, the thesis employed standardised and custom-made questionnaires: Simulator Sickness Questionnaire (SSQ), Presence Questionnaire (PQ), Perceived Affective Quality (PAQ) ratings, Program Ratings (PR), and custom-made retention and transfer tests. -- The results from two principal experiments encourage the application of Coherence principle, derived from CTML, to promote retention and transfer by removing extraneous visual information, while keeping the production cost and each project's time-frame down. Due to the slow-moving camera in both VR simulations, none of them generate sufficient level of optic flow to induce high level of SS. The results also show that motivational factors in VR simulations, presence and perceived affective quality do not correlate with retention and transfer. In closing, the thesis leads to future research in cinematography and quantifying visual complexity in VR-based instructional animations.