Item

Abstract

Complexity has emerged as an important factor driving both aesthetic evaluation and curiosity. In this thesis we present one of the first behavioural studies linking the fields of aesthetics and curiosity through the lens of visual complexity. We generated diverse, controlled, black-and-white pixel-grid patterns using cellular-automata (CA) and recorded subjective complexity evaluations on them. We found that computable objective complexity measures of spatial complexity (SC) and a novel intricacy measure were effective predictors of subjective complexity. These results generalised to larger and semi-revealed CA patterns where SC and intricacy of both the "visible pattern" (the semi-revealed pattern as it is seen) and the "underlying pattern" (the actual complete pattern underlying the semi-revealed version) together predicted subjective complexity. This complexity metric operationalised subjective complexity which was then used to model curiosity and exploratory behaviour. Participants freely explored grids hiding underlying patterns where they could click on tiles to reveal parts of the underlying pattern. Visible complexity measures (visible SC and visible intricacy) were evaluated at each click. We performed two levels of analysis - one at click level, modelling participants' probabilities of moving on from a grid, and another at grid level, modelling the number of clicks a participant would make on a grid. Across both forms of analysis, we consistently found two two-way interactions explaining behaviour, namely, visible SC interacting with underlying SC, and visible intricacy interacting with underlying intricacy. While SC had a positive influence on exploratory behaviour, intricacy had a negative influence. By contrast, both SC and intricacy correlated positively with subjective complexity. Inferences from these results are discussed. This work paves the way for further research at the intersection of curiosity and aesthetics, including creativity and pattern production behaviour.