Aesthetic Complexity: Practice and Perception in Art & Design
If you look around now, you will probably see mostly plain and simple areas (such as sky or walls). There might be regular patterns in smaller areas (clothing and furnishings) and also some irregular patterns (grass or clouds). Besides anything living (people, pets and plants), the most complex visual things around are probably objects of aesthetic value. Why artworks tend to be complex and how visual complexity relates to aesthetic value are the questions that drive this research.
This PhD was undertaken in the School of Art & Design at Nottingham Trent University, UK, 2006—2010. The examiner was Linda Candy of the University of Technology, Sydney. The supervisory team at NTU comprised:
- Prof. Judith Mottram, Dean of the School of Art & Design; Professor of Visual Art
- Prof. Terry Shave, Academic Team Leader, Visual Arts
- Dr. Andrew Dunn, Senior Lecturer in Psychology
This research investigates the aesthetics of visual complexity in the practice and perception of visual art and design. The aim is to understand visual complexity in terms of the relationship between the objective properties of images and subjective properties of perception. I take a computational and empirical approach to this subject, incorporating methods from information theory, computer graphics, complexity theory and experimental psychology. For testing, I create cellular automata programs to generate sets of stimulus images, and borrow a variety of visual material from students and professional artists, designers and craftspeople. Visual complexity is measured in two ways: Firstly, an objective measure of complexity is based on the data compression of digital image files, which provides an information-based scale of order to randomness. Secondly, psychophysical techniques are employed to measure the perceived complexity of the images and judgements of preference and artistic quality. Research in complex systems theory and empirical aesthetics suggests that we can expect to find an inverted ‘U’ correlation between the objective and subjective measures of visual complexity.
The project makes an original contribution to knowledge with empirical evidence for the inverted ‘U’ correlation of image file compression and perceived complexity. With cellular automata images from simple to complex, the two measures show a correlation which diverges as images approach randomness. With art and design images the correlation is weaker, probably due to the wider variety of visual material, which made it harder to compare the images and judge complexity. Image file data compression can be a useful measure, and is better with similar images. The correlation between data compression and perceived visual complexity supports the idea that we understand complexity as a mixture of order and chaos. Artworks with a balance of order and chaos are visually appealing because they allow for visual exploration and pattern-finding, which contributes to their aesthetic value. The statistical analysis shows that people’s judgements of perceived complexity and artistic quality are both surprisingly consistent, whereas preference is much more varied. As a result, perceived complexity correlates more strongly with judgements of quality than with preference for artworks. A qualitative analysis of participant interviews reveals that this is due to shared evaluative criteria between complexity and quality. Both visual complexity and aesthetic value are judged in terms of the amount of time, effort and skill that appears to have gone into its making or that is required to fully appreciate it. In other words, the things that make an image look complex are the same things that make it a good work of art.