Abstract
Whilst it is common for artists asking questions about the very nature of beauty, and universal aspects of aesthetic experiences, such questions are rarely picked up by scientists. However, psychophysics can make this field of enquiry accessible as demonstrated by Fechner (1876) in his rather programmatic experimental aesthetics. A key issue to overcome in such attempts in the visual domain is the apparently huge range and irregularity of possible designs in paintings, and the need to use systematic and well-defined stimuli in psychophysics. On the way towards making actual paintings accessible to psychophysical experimentation, we analysed the 'canonical' paintings of one of the most iconic artists of the 20th century - Piet Mondrian. Mondrian coined the name 'Neoplastic Abstraction' for his works between 1921 and 1939, which are characteristically composed of a grid of horizontal and vertical lines with rectangular colour patches embedded in the grid. We determined sizes, numbers, positions, and colours of paintings, lines and patches for all 55 canonical paintings listed in the catalogue of Mondrian's complete works, generating a database of numerical descriptions for each of these designs. From the database we calculated descriptive statistics of image elements, such as aspect ratios, line positions, patch sizes, colour areas, etc., which not only are characteristic of Mondrian's oeuvre as such, but also reflect his change of style over the 18 years. At an analytical level the statistics are further developed into compound measures such as probability density functions for lines, which can then be used to generate 'synthetic' patterns ('Mondroids') from a small set of parameters that look like paintings from a particular phase. Such Mondroids can now be the basis for systematic experiments, involving psychophysical methods as well as eye tracking, or evolutionary algorithms to study aesthetic preference (see Holmes & Zanker, i-Perception 3, 2012).
Meeting abstract presented at VSS 2014