Mathematical modeling and experimental study of interface aesthetics.

Abstract

A strong need is emerging that requires designers and engineers to understand and address the influence of aesthetics and aesthetic judgment in the analysis and design of modern interfaces and products. The capability to objectively and quantitatively include user feedback about the aesthetic components in a designed system would pay dividends in increased user satisfaction and performance over the lifetime of the system. The work presented in this dissertation examines the field of engineering aesthetics and attempts to use quantitative-based methodologies to explore design in a structured manner. This work addresses three major goals: to develop and validate quantitative metrics that represent common compositional attributes; to determine how those attributes relate to appraisals of aesthetic appeal; and to develop and experimentally test an algorithm that can use iterative feedback from end users to tailor designs to their aesthetic preference. The attributes of symmetry and balance are both quantified and validated by subject ratings. These attributes are then used, along with the number of compositional elements, to illustrate that the aesthetic appeal of visual stimuli is partially dependent on the quantitative representation of these attributes. The relationships established between aesthetic appeal and the attributes shows that subjects prefer symmetry to non-symmetric stimuli and that the fewer the number of compositional elements, the more aesthetically appealing the visual stimuli. The latter portion of the dissertation is dedicated to establishing support and gathering human-subject data for an interactive genetic algorithm (IGA) approach to optimizing designs based on interactive feedback from the designer or user. These results show that the IGA is moderately effective in presenting good designs when using abstract imagery and that it is even more effective when presenting weblog templates. The resulting research presented in this dissertation is theoretically useful in creating the first steps towards a cognitive model of aesthetic preference and it is also of use to practitioners in that the IGA methodology is validated as being an effective tool for improving aesthetic preferences in design.